ICES Journal of Marine Science: Journal du Conseil

Ortuño Crespo,, Guillermo;Dunn, Daniel, C

doi: 10.1093/icesjms/fsx084pmid: N/A

Abstract Open‐ocean fisheries expanded rapidly from the 1960s through the 1980s, when global fish catches peaked, plateaued and possibly began to decline. While catches remain at best stagnant, fishing effort globally continues to increase. The likelihood of ecosystem impacts occurring due to fishing is related to fishing effort and is thus also expected to be increasing. Despite this rapid growth, ecological research into the impacts of fisheries on open‐ocean environments has lagged behind coastal and deep-sea environments. This review addresses this knowledge gap by considering the roles fisheries play in controlling the open-ocean at three ecological scales: (i) species (population or stock); (ii) biological community; and (iii) ecosystem. We find significant evidence for top-down control at the species and community scales. While evidence of ecosystem-level impacts in the open-ocean were not explicit in the literature, we provide examples of these impacts in several marine pelagic systems and encourage further research at this ecological scale. At the species level, fishing can reduce abundance, and alter physiology and life history traits, which, in turn, affect the functional role of the species within the biological community. Fishing may also induce changes to open-ocean community trophodynamics, and reduce biodiversity and resilience in open-ocean ecosystems. Our ability to manage open-ocean ecosystems has significant implications for provisioning of ecosystem services and food security. However, we posit that the monitoring required to assure the sustainability of open-ocean ecosystems is not being undertaken, and will require coordination with the Global Ocean Observing System, industry, and academia. Introduction The world’s oceans are experiencing an unprecedented level of biotic exploitation, which is altering the abundance and population structure of many species, transforming the composition of biological communities, and threatening the integrity and resilience of entire marine ecosystems (Pauly et al., 1998; Jackson et al., 2001; Bellwood et al., 2004; Daskalov et al., 2007). Over the last few decades, a growing body of evidence has accumulated, demonstrating these impacts at different trophic levels and across a wide range of taxa and marine ecosystems. Most fisheries impacts in coastal zones were well described by the turn of the century (Dayton et al., 1995; Jennings and Kaiser, 1998), and our understanding of impacts on similarly static, deep-sea habitats have also been well documented (Koslow et al., 2000; Clark, 2001; Roberts, 2002) and have been reviewed recently (Clark et al., 2016). However, there remain knowledge gaps regarding the potential impacts of biotic exploitation on open-ocean ecosystems. The dynamism, distance from land and sheer scale of these ecosystems have limited the capacity of researchers to study their ecology and the species that comprise them, let alone monitor changes induced by anthropogenic stressors. These gaps in understanding limit our ability to manage and conserve these ecosystems and, if not addressed promptly, may result in permanent structural or compositional changes to these ecosystems, which in turn jeopardize their functionality and thus their ability to provide ecosystems services. Historically, marine fisheries have operated near coastal areas, mainly because of the elevated biological productivity of coastal systems and our reduced ability to store and transport fish from distant waters (Pauly et al., 2005; Swartz et al., 2010). However, this spatial pattern of fishing changed dramatically after onset of the industrial revolution (Swartz et al., 2010). Improvements in locomotion and refrigeration, among other factors, allowed for the expansion of fisheries in terms of fishing capacity, fishing effort and spatial extent (Swartz et al., 2010). Prior to many of these technological advancements, many open-ocean ecosystems had been sheltered from growing commercial fisheries exploitation. However, between 1950 and 1990, landings from areas beyond national jurisdiction (ABNJ) (i.e. the majority of the open-ocean) more than quadrupled to over 80 million tonnes (Merrie et al., 2014). New global fishing catch reconstruction estimates suggest that historical catches have been even higher and the declines since the peak have been even greater (Pauly and Zeller, 2016). While catches have stagnated since 1990, fishing effort has continued to increase, doubling between 1990 and 2010 (Anticamara et al., 2011; Merrie et al., 2014). According to FAO estimates, 6366 fishing vessels from 40 flag states, fish for open-ocean species in the high seas (HSVAR, 2016). The average catch of these fisheries in the first decade of the century was 10 million tonnes, which is equivalent to ∼12% of the total average marine fisheries catch (Sumaila et al., 2015). Catches in pelagic fisheries in ABNJ are dominated by large and medium pelagics which account for 82.69% of total pelagic catches; where tuna species dominate the catches for large pelagics and menhaden lead medium pelagic catches (Figure 1). The majority of the taxa in those functional groups are managed by tuna RFMOs, and the vast majority of those catches (88% of the tuna species) come from purse seine, longline, and pole and line fisheries. We do not directly address potential impacts from trolling (6% of large pelagic catches) or trawling (<6% of large pelagic catches) activities by tuna RFMOs. Figure 1 Open in new tabDownload slide Total catch of the nine pelagic functional groups caught in areas beyond national jurisdiction from 1950 to 2010. The data was aggregated across the 17 high seas regions in the Sea Around Us catch reconstruction database (Pauly and Zeller, 2015). LR(≥90 cm) = Large Rays (≥90 cm); S/MS(<90 cm) = Small to medium sharks (<90 cm); LS(≥90 cm) = Large Sharks (≥90 cm); S/MR(<90 cm) = Small to medium rays (<90 cm); KRI = Krill; CEP = Cephalopods; SP(<30 cm) = Small Pelagics (<30 cm); MP (30–89 cm); LP(≥90 cm) = Large Pelagics ( > =90 cm). The Medium and Large Pelagics are further broken down into the taxonomic groups which account for <95% of their biomass. Large Pelagics: SKJ = Katsuwonus pelamis; YFT = Thunnus albacares; BET = Thunnus obesus; ALB = Thunnus alalunga; SBT = Thunnus maccoyii; SWO = Xiphias gladius; M/T/B = Mackerels/tunas/bonitos; KAW = Euthynnus affinis; T/B/B = Tunas/bonitos/billfishes. The 37 taxonomic groups which individually represented less than 1% of the biomass caught in this functional group were aggregated in “Other”. Medium Pelagics: MHA = Brevoortia tyrannus; CJM = Trachurus murphyi; HER = Clupea harengus; CHM = Scomber japonicus; JPO = Jacks/pompanos; SAP = Cololabis saira; FRI = Auxis thazard; MAC = Scomber scombrus; JHM = Jack/horse mackerels. The 42 taxonomic groups which individually represented less than 1% of the biomass caught in this functional group were aggregated in “Other”. Figure 1 Open in new tabDownload slide Total catch of the nine pelagic functional groups caught in areas beyond national jurisdiction from 1950 to 2010. The data was aggregated across the 17 high seas regions in the Sea Around Us catch reconstruction database (Pauly and Zeller, 2015). LR(≥90 cm) = Large Rays (≥90 cm); S/MS(<90 cm) = Small to medium sharks (<90 cm); LS(≥90 cm) = Large Sharks (≥90 cm); S/MR(<90 cm) = Small to medium rays (<90 cm); KRI = Krill; CEP = Cephalopods; SP(<30 cm) = Small Pelagics (<30 cm); MP (30–89 cm); LP(≥90 cm) = Large Pelagics ( > =90 cm). The Medium and Large Pelagics are further broken down into the taxonomic groups which account for <95% of their biomass. Large Pelagics: SKJ = Katsuwonus pelamis; YFT = Thunnus albacares; BET = Thunnus obesus; ALB = Thunnus alalunga; SBT = Thunnus maccoyii; SWO = Xiphias gladius; M/T/B = Mackerels/tunas/bonitos; KAW = Euthynnus affinis; T/B/B = Tunas/bonitos/billfishes. The 37 taxonomic groups which individually represented less than 1% of the biomass caught in this functional group were aggregated in “Other”. Medium Pelagics: MHA = Brevoortia tyrannus; CJM = Trachurus murphyi; HER = Clupea harengus; CHM = Scomber japonicus; JPO = Jacks/pompanos; SAP = Cololabis saira; FRI = Auxis thazard; MAC = Scomber scombrus; JHM = Jack/horse mackerels. The 42 taxonomic groups which individually represented less than 1% of the biomass caught in this functional group were aggregated in “Other”. The stagnant catch trend and decreasing catch per unit effort trend in ABNJ over the last 20 years have been caused, at least in significant part, by overfishing (Merrie et al., 2014). Migratory and straddling stocks1 spend a significant proportion of their life cycle in ABNJ (Harrison, 2012) and are particularly vulnerable to overfishing, mismanagement and illegal, unreported, and unregulated (IUU) fishing, due to the difficulty of managing their entire range and ensuring the compliance of all parties harvesting such stocks (Maguire, 2006). On-going difficulties in managing these stocks calls into question the once accepted notion of marine fish stock inexhaustibility, particularly that of wide-ranging pelagic species. A recent report by the United Nations Environment Programme (UNEP) and the Convention on Migratory Species (CMS) reinforces this notion of the vulnerability of highly mobile species, reporting that 36% of the 153 migratory or potentially migratory chondrichthyan fishes are threatened with extinction; though this proportion could be larger, as a further 27% of the taxa are data deficient (Fowler, 2014). Implementing measures to abate the negative impacts of fisheries on wide-ranging oceanic species will require advancements in the international management frameworks for these species as well as improvements in the understanding of their ecological function within oceanic systems. We are currently at a point of convergence between growing availability of long-term multispecies catch datasets for open-ocean systems and a more robust framework for ecosystem-level mass balance models, which together allow for a comprehensive assessment of the knowledge gaps regarding the ecological impacts of fisheries on open-ocean ecosystems. These advancements have come together at a time when the international policy arena is orienting its attention towards improving the governance of ABNJ2,3. A review of the impacts and efficacy of current fisheries management regimes are underway at the United Nations, as are negotiations over a new legally-binding, international instrument for the conservation and sustainable use of biodiversity beyond national jurisdiction (BBNJ). Here, our objective is to address a major knowledge gap in our understanding of anthropogenic impacts of fishing on the open-ocean, which we hope will help inform the review of the United Nations Fish Stocks Agreement (UNFSA) and the BBNJ negotiations. Toward that end, we synthesize the current state of knowledge on the effects that open-ocean fisheries have at three distinct ecological scales: (i) species (population or stock); (ii) biological community, and (iii) ecosystem. Managing marine ecosystems across multiple scales—from stock to ecosystem—is essential for their long-term health and resilience (Hunt and McKinnell, 2006), and is a basic tenant of ecosystem-based fisheries management (Francis et al., 2007). Before doing that, we broadly review differences in the control mechanisms of open-ocean ecosystems and define open-ocean species, communities and ecosystems. Top-down versus bottom-up control: complementary mechanisms The notion that changes in the upper trophic levels of an oceanic system can lead to ecosystem-wide changes differs from the traditional view that bottom-up control (in the form of resource dependence) is the main factor shaping the structure and composition of these ecosystems (Cushing, 1975; Aebischer et al., 1990; Verity and Smetacek, 1996; Strong and Frank, 2010; Mulder et al., 2012). This view is strongly reinforced by a body of literature which demonstrates bottom-up control of the biological community in various marine systems around the world; among the most relevant of these examples are empirical demonstrations of positive relationships between fisheries captures and levels of primary productivity (Chavez et al., 2003; Ware and Thomson, 2005; Chassot et al., 2007). A subset of these examples argue against the potential for top-down control of open-ocean pelagic systems, due to a lack of conclusive examples (Steele, 1998; Steele et al., 1998). Among the main arguments are: (i) high species diversity can buffer changes in trophodynamics of open-ocean systems; (ii) the opportunistic character and dietary plasticity of most pelagic predatory species may result in prey switching that dampens the trophic cascades; (iii) the high level of connectivity in these systems may buffer against local depletions; (iv) the dynamism, heterogeneity and patchiness of primary productivity reduces the likelihood of disrupting trophic linkages at any particular location, as feeding is more opportunistic; and (v) the potential dominant role of gelatinous carnivores in these systems also reduce the top-down pressure of fishing on the biological community as these organisms are not harvested (Larkin, 1979; Mills, 1995; Steele, 1998; Jennings and Kaiser, 1998; Link, 2002). This perspective has fostered the belief that fishing cannot exert top-down trophic control in these systems. Here, we review evidence for top-down control on open-ocean ecosystems and encourage a more holistic view where both bottom-up and top-down controls are accepted. We consider this approach to be necessary given the growing evidence of the impacts that climate change may be having on pelagic communities, through changes in a system’s primary productivity (Ware and Thomson, 2005; Frederiksen et al., 2006) or disruption of the timing of ecological events, which may alter the phenological patterns of marine species (Edwards and Richardson, 2004). This more holistic approach is not novel; work by Brander (2007) or Lam et al., (2016) highlights how the interaction between climate change and fishing can negatively affect the maintenance of global seafood production. Corroborating this approach, reviews of regime shifts recorded in Northern Hemisphere pelagic ecosystems by Möllmann and Diekmann (2012) and Beaugrand et al. (2015), identified multiple drivers (including fishing and climate change) as the potential cause of shifts in ecosystem state. Open-ocean species, communities, and ecosystems In setting the stage for this review, we broadly define open-ocean species in terms of their ecology and life history traits, open-ocean communities in terms of their composition and trophodynamics and open-ocean ecosystems according to their spatiotemporal distribution, dynamics, and biophysical characteristics. Different approaches can be taken when defining what constitutes an open-ocean species. The Russian ichthyologist Nikolai Parin established a three category ecological classification for open-ocean fish based on the proportion of the species’ life history that is spent in the open-ocean zone: (i) those species which spend the entirety of their life cycles in the open-ocean (permanent residents); (ii) those species which spend only a part of their life cycle in the open-ocean; and (iii) those species which occasionally spend time in the open-ocean, spending most of their time near coastal waters but occasionally moving offshore to feed (Pepperell and Harvey, 2010). This same approach can be applied to other taxonomic groups such as sea turtles, where certain species (e.g. Loggerhead sea turtles) have oceanic developmental stages in their life cycles (Zug et al., 1995; Bolten, 2003), while others (e.g. flatback sea turtles) have more coastal distributions and lack an oceanic stage, thus venturing into the open-ocean environment much less frequently (Walker and Parmenter, 1990; Limpus et al., 1995). For the purpose of this review, we define open-ocean species as the combination of all three of Parin’s categories. Given the scope of this paper, our definition of an open-ocean community is limited to open-ocean nekton, which are the species most directly affected by open-ocean fishing. The composition of this community is highly dynamic and heterogeneous across space and time, as many open-ocean species are migratory and shift their range throughout the year. A 2006 FAO report on the state of migratory and straddling stocks identified up to 226 highly mobile open-ocean species (Chondrichthyes and Osteichthyes), while the aforementioned CMS and UNEP report identified 153 migratory or potentially migratory chondrichthyan fishes (Maguire, 2006; Fowler, 2014). It is important to note that scientific information on the composition of open-ocean marine food webs is scarce and is largely based on fisheries catch records and fisheries observer programs, which are not homogenous across space, time or fisheries. In recent years, however, there have been improvements in the collection of this type of ecosystem-level data, as in the case of observer programs in the Pacific Ocean basin (Colléter et al., 2015). There is currently no widely accepted, official definition of what constitutes the open-ocean environment. The spatiotemporal variability of oceanographic boundaries in all three spatial dimensions (particularly the vertical dimension) and the lack of structural features to delineate habitats within the pelagic open-ocean, complicate the delineation of this definition. Moreover, in the vertical dimension, the structuring of the physical and chemical properties of the open-ocean water column are dynamic across space and time and have been shown to be different between ocean basins. For these reasons, we opted for a clear-cut definition of the open-ocean environment based on bathymetric and oceanographic principles. The continental shelf break provides a horizontal boundary for an oceanic system which, although it can interact with continental shelf ecosystems, has distinct communities. Thus, we define the open-ocean as extending beyond the continental shelf break (generally delineated as 200 m in depth), and encompassing the entire water column. The biological productivity and connectivity of the upper kilometer of the water column are key elements for setting the vertical boundaries of the open-ocean environment (Angel, 2003), and this zone encompasses most oceanic diel vertical migrations and the community. Below the mesopelagic zone, the biomass of pelagic organisms generally decreases by a factor of ten (Angel and Baker, 1982). Considering these factors together with the near total lack of studies on the impact of fisheries on the bathypelagic and abyssopelagic zones, we loosely draw the vertical boundary for this study at 1000 m. Below we use these three ecological scales (species, communities, and ecosystems) to enumerate evidence of top-down control over open-ocean ecosystems by fisheries. These impacts can result from direct stressors such as mortality derived from target and non-target catch, or indirect stressors, such as changes in trophodynamics, life history traits or biodiversity. Species-level impacts (direct) Declines in abundance While the improvements in the health of stocks within the Exclusive Economic Zones (EEZs) of coastal states such as the United States, Iceland or Australia are positive trends that should be acknowledged, open-ocean species are never found within only one EEZ (Murawski et al., 2007). Rates of overfishing and the per cent of overfished stocks are greater for straddling stocks (64%) under multinational management than those under national management (28.8%) (FAO, 2014). A 2010 assessment of the 48 fish stocks managed by the world’s 18 Regional Fisheries Management Organizations (RFMOs) concluded that 67% of these were either overfished or depleted, all of which are open-ocean species (Cullis-Suzuki and Pauly, 2010); these are consistent with the FAO (2014) estimates for straddling stocks mentioned above. Specific examples abound: according to the International Scientific Committee of Tuna and Tuna like Species in the North Pacific Ocean, Pacific bluefin tuna population (Thunnus orientalis) has declined by 97.4% (ISC Pacific Bluefin Tuna Working Group, 2016). The latest stock assessments for the West and Central Pacific stock of big eye tuna (Thunnus obesus) and southern bluefin tuna (Thunnus maccoyii) indicate spawning stock biomass declines of greater than 80 and 90%, respectively (CCSBT, 2014; Harley et al., 2014). While a number of reviews have shown very strong declines across top predators in pelagic systems (Baum et al., 2003; Myers and Worm, 2003; Baum and Myers, 2004; Ward and Myers, 2005), these reports have been strongly rebutted for assumptions regarding their use of data, small sample sizes, or the reliability of spatial catch-per-unit-effort to infer trends in biomass (Walters, 2003; Burgess et al., 2005; Maunder et al., 2006). However, other methods, including ecosystem models and analyses of trade data, have also identified declines of >2 orders of magnitude in top predators due to fishing pressure (Clarke et al., 2006). At the species level, the greatest exploitation-induced threat that any given species can face is extinction: local, ecological, or commercial (McCauley et al., 2015). Although there are no known examples of ecological extinctions in the open-ocean, there is strong evidence of very high depletion of oceanic predatory fish species, such as sharks, tunas, and billfishes (Cox et al., 2002; Hutchings et al., 2010). A frequent explanation for the lack of examples of ecological extinctions of open-ocean target species is that the economical extinction of a stock precedes its ecological extinction, which leads to a decrease in pressure on the stock. However, this assumption fails to account for dynamics in multispecies fisheries, such as pelagic longline fisheries. Multispecies fisheries may target more abundant, lower-value species to generate the majority of the income within the fishery, but continue to take the more depleted, high-value species opportunistically. Thus, such fisheries allow for the parallel exploitation of rarer species, which are at densities below their bioeconomic equilibrium (Gordon, 1954); below which point single-species fisheries would normally cease to operate (Branch et al., 2013). Unlike generalist species, which will shift the composition of their diet based on the relative abundances of prey (Smout et al., 2010), humans see value in rarity and will continue exploiting depleted species as long as they are economically profitable (Courchamp et al., 2006; Branch et al., 2013); this is the case of the United States pelagic longline fishery where the main target species are swordfish (Xiphias gladius), big eye tuna (Thunnus obesus) and yellowfin tuna (Thunnus albacares), but which still catches Atlantic bluefin tuna (Thunnus thynnus). Impacts of size-based targeting: stock structure and recruitment Selectivity in many marine fisheries extends past species preferences to the population level, leading to the asymmetric exploitation of stocks by age class, maturity status, behaviour, or morphology; all of which may act as selection pressures towards certain life history traits (Heino and Godø, 2002; Sharpe and Hendry, 2009). Any changes to these life history traits will affect the population’s dynamics and structure, which in turn control factors such as abundance, growth rate or demography (Law and Grey, 1989; Conover and Munch, 2002; Jorgensen et al., 2007). Older age classes in fish populations are much more susceptible to fishing pressure, even at moderate levels than those of younger age-classes (Garcia et al., 2012). Sibert et al. (2006) used long-term tuna fisheries catch data from the Pacific to determine that, from 1950 to 2004, fish larger than 175 cm decreased from 5% to 1% of the total population. The uneven exploitation of the larger individuals within fish populations has changed the age-class structure of many open-ocean populations, making them more vulnerable to fluctuations in inter-annual recruitment rates (Hsieh et al., 2006). The age-structure of catches of Thunnus orientalis illustrate how biased fisheries harvest towards adult individuals has resulted in the age-truncation of the stock, where most of the catch (∼90%) belongs to sexually immature age classes of 0–2 years (ISC Pacific Bluefin Tuna Working Group, 2016). These age-truncated stocks are more susceptible to experiencing booms and busts in recruitment, which makes them more vulnerable to stock collapses (Rochet and Benoît, 2011). The targeted exploitation of older age classes in open-ocean species may also lead to the loss of geographic substructure of populations, making them more vulnerable to environmental variability (Berkeley et al., 2004; Ottersen et al., 2006). Further, it can reduce the average reproductive potential of the population (Birkeland and Dayton, 2005; Anderson et al., 2008)—as older age classes are more fecund (Denney et al., 2002). Unfortunately, it is uncommon for fisheries to record data on the size distribution of their catch since the initiation of fishing, particularly for species that are not of high commercial interest (Jackson et al., 2001). The importance of recording these parameters is reflected in studies such as Ward and Myers (2005), which shows that continuous fishing of open-ocean predatory species can lead to reductions in their average body mass, which has implications to their life histories and ecological roles. They demonstrated that 11 of the 12 predatory species assessed experienced reductions in body mass between 29 and 73%. Changes in average body mass is one of the many alterations in life history traits or phenotypic characteristics attributed to fisheries exploitation (Walsh et al., 2006; Sharpe and Hendry, 2009). Fishing may also reduce the age and size at maturation of exploited stocks (Rochet, 1998; Law, 2000; Jorgensen et al., 2007). Although the causality of these changes is still contested—whether triggered by fisheries-induced genetic changes or environmental changes (Kuparinen and Merilä, 2007; Garcia et al., 2012)—a study which analysed these trends in 143 fishing time series (from 37 separate stocks) asserts that the changes in maturation are highly correlated, and can be attributed to increases in fishing pressure (Sharpe and Hendry, 2009). Impacts of size-based targeting: demographic changes The traditional management approach of highly mobile oceanic species through single-stock assessments rarely consider the spatiotemporal information about stock connectivity and population structure and may thus obscure some of the ecological impacts of their exploitation. While catch metrics of a species may be steady across time in a fishery, not accounting for the spatial location of the catch may be masking local extinctions, range contractions, or structure-level effects in the stocks (Taylor et al., 2011; Worm and Tittensor, 2011; Goethel et al., 2012). Worm and Tittensor (2011) used multidecadal catch datasets to address the range-abundance relationship in stocks of 13 exploited marine predators and demonstrated range contractions in 9 of the 13 species of tuna and billfish assessed, mostly along the edge of the ranges. Interestingly, they also quantified range expansions in two of the species (skipjack tuna (Katsuwonus pelamis) and sailfish (Istiophorus platypterus)), which may be a result of changes in the trophodynamics in their communities—such as predatory release, which we discuss later in this review. However, the range expansions of these two species were not replicated across ocean basins, highlighting the necessity of spatially discrete assessments on a regional basis. Changes in range may also result from the asymmetric exploitation of populations. Oceanic species such as Atlantic swordfish (Xiphias gladius), exhibit differences in their ranges at different stages of their life cycle, where adult individuals display larger ranges. Thus, if particular age classes are targeted more heavily, changes in the realized niche of the species could take place (Neilson et al., 2014). This asymmetric exploitation of the older age-classes across open-ocean taxa may also be removing the age-classes that are more physiologically tolerant, which could be leading to range contractions along the latitudinal edges of the range, where temperatures may only be tolerated by those age-classes. Dulvy et al. (2003) suggest that dispersal and geographic range size play a role in regulating the risk of extinction of wide-ranging marine species, where large geographic ranges add an extra layer of ecological resilience by reducing their catchability at low densities. However, certain highly migratory species display dense annual aggregations on feeding and breeding grounds. This is the case of Thunnus thynnus, which congregate in northern part of the Gulf of Mexico and coastal waters in the Mediterranean to spawn; this life history trait can lead to heavy exploitation of even highly vagile stocks (Block et al., 2001; Fromentin and Powers, 2005). Together with the truncation of stock age-structure, the loss of geographic substructure within populations makes them more susceptible to climate-induced alterations (Marshall and Browman, 2007) and genetic changes addressed below. Impacts of size-based targeting: genetic changes Another significant—yet more cryptic—impact on target and non-target species, comes in the form of genetic changes which, in the context of this review, we address as potential impacts given the lack of consolidated evidence of fisheries-induced genetic impacts on open-ocean species. The earliest evidence of induced variations in genetic traits in fish originated in aquaculture programs and experiments; these were induced in a short temporal window of just a few generations (Gjedrem, 1983; Garcia et al., 2012). In wild fisheries, evolutionary changes may be induced by selecting against certain life history traits, through high selectivity towards size and age, and by removing large proportions of the population (Stokes and Law, 2000). For example, in the last half century, 26 harvested stocks of tunas and their relatives have been halved (Juan-Jordá et al., 2011; see “Declines in abundance” section for further details). Fishing-induced genetic changes can increase the risk of extinction and decrease the rate of recovery of overfished stocks (Olsen et al., 2004; Walsh et al., 2006). There are three main types of genetic change: alteration of sub-population structure, decrease in genetic variation and selective genetic changes (Allendorf et al., 2008). Populations may be comprised of spatially discrete breeding groups (sub-populations) that, unless characterized genetically, will be erroneously managed as a single, genetically homogenous population. For example, there is evidence that migratory species such as Thunnus thynnus show sub-population structure, which is not reflected in their management strategies (Fromentin and Lopuszanski, 2013). This lack of consideration of the genetic stock structure may not only be translating into reductions in genetic diversity at the population level, but also the sub-population level (Allendorf et al., 2008). There are two main ways to assess this variation: heterozygosity and allelic diversity. Reductions in heterozygosity of a population can be quantified through its effective population size, which is affected by factors such as demography, sex ratios and fecundity. By reducing the effective population size, selective fishing can thus exacerbate the loss of genetic variation (Allendorf et al., 2008). On the other hand, loss of allelic diversity can be caused through high rates of both targeted or non-selective exploitation (Ryman et al., 1995). Reductions in allelic diversity due to fishing pressure may also reduce the species’ ability to adapt to changing climactic conditions and represents one way in which fisheries and climate (i.e. top-down and bottom-up controls on the system) may act synergistically on populations of marine species (Soule and Wilcox, 1980; Brander, 2007). While species can theoretically maintain levels of genetic heterozygosity during population bottlenecks, allelic diversity can be severely reduced in such events (Allendorf, 1986). Given very limited genetic research on open-ocean species, we can only address this as a potential impact of fishing pressure, likely of increasing relevance as the abundance of a species decreases. Bycatch and other sources of inadvertent mortality The impacts of fisheries on open-ocean species can extend beyond those taxonomic groups targeted commercially, through the unintentional catch of other taxa that is either unused or unmanaged; this catch is defined as bycatch. Pelagic longline fleets primarily targeting billfish and tuna are the most widespread fisheries in open-ocean systems (Worm et al., 2005) and the source of most pelagic discards across ocean basins, together with midwater pelagic trawling and purse seining (Cook, 2003; Kelleher, 2005). Bycatch in open-ocean fisheries can incur high mortality rates and have been implicated in the collapse of many sea turtle (Wallace et al., 2010), seabird (Anderson et al., 2011), marine mammal (Lewison et al., 2014), and shark (Dulvy et al., 2008; Oliver et al., 2015) populations. For example, bycatch of Pacific loggerhead (Caretta caretta) and leatherback (Dermochelys coriacea) turtles in pelagic longline gear have played a key role in the severe (>80% and >95%, respectively) declines in the nesting populations of these species over 20–30 years (Spotila et al., 2000, Limpus and Limpus, 2003; Lewison et al., 2004). Further, all 22 species of albatross and 19 of 21 oceanic elasmobranchs are listed as at least Near Threatened by the IUCN with bycatch cited as the main threat (Robertson and Gales, 1998; Dulvy et al., 2008; Anderson et al., 2011; IUCN, 2015). Quantifying the global estimates of bycatch in the open-ocean remains a challenge due to lack of data (Alverson et al., 1994; Kelleher, 2005). Gilman et al. (2014) estimated that two thirds of RFMO fisheries targeting open-ocean species lack adequate observer coverage, which is a basic requirement to obtain robust bycatch estimates. While global bycatch estimates are useful for starting the discussion on the impacts of bycatch of open-ocean species, taxa-specific studies for the main bycatch taxonomic groups do exist: sea turtles, seabirds, marine mammals, and sharks (Wallace et al., 2010; Anderson et al., 2011; Molina and Cooke, 2012; Lewison et al., 2014; Oliver et al., 2015). Open-ocean species have wide spatiotemporal distributions which can overlap significantly with the range of one or more pelagic fishing fleets (Block et al., 2011). In a recent publication, Queiroz et al. (2016) quantified an 80% spatial overlap between the distributions of six species of oceanic shark and that of two longline fishing fleets, and noted how both sharks and fishermen were tracking similar biophysical cues in the marine environment. Given the high overlap, on-board monitoring of fisheries catch and bycatch must be comprehensive across fishing fleets, gear types and marine regions (Birdlife International, 2004; Queiroz et al., 2016). Just like highly mobile sharks, many of the open-ocean seabird species threatened with extinction, such as albatrosses and petrels, are wide-ranging species whose distributions overlap greatly with those of marine fishing fleets worldwide (Birdlife International, 2004). Mortality due to interaction with longline gear has been cited as a critical threat to these species (Klaer and Polacheck, 1997; Brothers et al., 1999; Tuck et al., 2001). However, seabird bycatch mitigation measures have resulted in strong declines in seabird bycatch rates in many longline fisheries over the last decade (Gilman et al., 2005). Similarly, sea turtle bycatch mitigation has also seen advances over the last two decades, though with more limited success than with seabird bycatch mitigation (Gilman et al., 2006). Quantifying the degree to which these pelagic species interact with fishing gear in the open-ocean is a very challenging issue given the small amount on information available on the distributions of both the animals and fishing fleets at high spatiotemporal resolutions; but as seen in Queiroz et al. (2016), improvements in tracking and vessel monitoring data are allowing for these types of inferences. This issue is catalysed by the low observer coverage and low bycatch reporting rates across open-ocean fisheries (Gilman et al., 2014) and is not unique to seabirds, marine mammals, sea turtles or sharks, as a much wider spectrum of open-ocean species are caught as bycatch and are rarely reported or considered in the management strategies. Other impacts of non-targeted catch Other more cryptic sources of indirect or unaccounted mortality include pre-catch losses, which occur when an organism is caught and killed by fishing gear, yet it is not commercialized for reasons such as depredation by predatory species (Hernandez-Milian et al., 2008), or simply because the catch or bycatch falls from the gear before it is hauled (Gilman et al., 2014). Another source of mortality that is regularly unaccounted for in fisheries management and population models is that of post-release mortality, whereby specimens that are caught in fishing gear are released alive, but because of post-release stress and/or injuries, may later die (Gilman et al., 2005; Campana et al., 2009). Not accounting for these sources of mortality may lead to underestimation of bycatch mortality which can, in turn, compromise the quality of population models for those species (Gilman et al., 2005; Coggins et al., 2007; Molina and Cooke, 2012). Understanding the post-release mortality of specimens that are discarded alive is of major importance, as it may otherwise lead to underestimations of bycatch-induced mortality (Coggins et al., 2007, Molina and Cooke, 2012). A 2009 study used archival satellite pop-up tags to quantify the mortality rate of one of the most frequently discarded fish species in marine open-ocean longline fisheries: blue sharks (Prionace glauca) (Campana et al., 2009). They concluded that while all healthy sharks survived, over a third of those that were injured died within a few days/weeks, which, according to Skomal and Mandelman (2012)m may be the result of disturbances in their behaviour or physiology. These findings on post-release mortality raise fundamental questions about true morality rates of discarded species and how these may affect both stock-level and ecosystem-level models. Another less acknowledged source of indirect mortality is the death by starvation of young individuals if the parent(s) on which they depend for feeding is killed. For example, if an adult albatross from a breeding pair is killed, the chick may starve to death and it may take years before the other adult albatross procreates again (Tasker and Becker, 1992; Brothers, 1995; Gilman et al., 2005). Given their life history strategies, seabirds, and marine mammals may be more vulnerable to this type of indirect impact. Community-level impacts (indirect) Most of the impacts of fishing at the community level concern the trophic relationships and are tightly linked with changes at the species level. Changes in the trophic dynamics of the system are caused by: changes in species abundance, alterations of species size, and behaviour, and changes in the growth and reproductive rates of populations. The community-level impacts come in the form of imbalances in the trophic control mechanisms of the community, where the trophic pressure, feeding rate, or dietary composition of species have changed as a direct or indirect result of fishing pressure on open-ocean species. Top-down trophic control of prey abundance by higher trophic level organisms (Paine, 1980) can propagate across multiple trophic levels and is generally characterized by opposing changes in biomass from one trophic level to the next. This asymmetric trophic imbalance is known as a trophic cascade (Pace et al., 1999). Trophic cascades In the last two decades, there has been a growing body of scientific literature which addresses the role of top-down trophic processes in defining the composition and structure of marine communities and how marine fisheries may be triggering changes in these dynamics (Cury et al., 2000; Worm and Myers, 2003; Ainley et al., 2007; Nicol et al., 2007; Polovina and Woodworth-Jefcoats, 2013). Top-down trophic control of marine community composition, in the form of trophic cascades, has been demonstrated in variety coastal marine systems (Jackson et al., 2001; coral reefs, Bellwood et al., 2004; rocky intertidal ecosystems, Menge, 2000; kelp forests, Estes and Palmisano, 1974; and reviewed across coastal ecosystems, Pinnegar et al., 2000; Steneck and Sala, 2005). However, detecting and characterizing these changes in open-ocean biological communities has proven to be a challenge. Data availability is the main factor limiting any effort to evaluate the impacts of fisheries on the integrity of marine populations, biological communities, and ecosystems (Colléter et al., 2015), and explains why our understanding of open-ocean impacts has lagged behind coastal ecosystems (Webb et al., 2010). Further, open-ocean pelagic food webs are highly dynamic and heterogeneous in composition, making them especially challenging to model in space and time. Despite the paucity of data and obstacles to model development, ecosystem-level models have begun to reveal community-level impacts of marine fisheries on open-ocean communities (Kitchell et al., 2002; Hinke et al., 2004; Kitchell et al., 2006; Polovina and Woodworth-Jefcoats, 2013). We review these models and the evidence for community-level impacts in open-ocean ecosystems below. One of the best-studied regions for the impacts of fisheries on open-ocean communities is the Pacific Ocean basin, where a series of ecosystem mass-balance models have been assembled for this purpose. Hinke et al. (2004) reviewed the impact of commercial tuna fisheries in two published oceanic food-web modelling studies in the Eastern Tropical Pacific (ETP) and Central North Pacific (CNP) ecosystems. Although similar in terms of their biological structure, these systems differ in their fishery histories and in the composition of their target and bycatch species (Cox et al., 2002; Olson and Watters, 2003). Based on the mass balance models, Hinke et al. (2004) concluded that increases in catch by the pelagic tuna fisheries (both purse-seine and longline gears) had similar impacts on the food-web structure in both systems: fishery-induced reductions in the top predators were followed by increases in the biomasses of lower trophic levels (Hinke et al., 2004). The impacts of both fishing gear types were stronger in the upper trophic levels (particularly longline fisheries), while the purse-seine fishery seemed to have a more profound impact on the abundance of intermediate trophic levels. Further empirical and model-based evidence for mesopredator releases in oceanic systems caused by declines in apex predator guilds is becoming plentiful (Carscadden et al., 2001; Cox et al., 2002; Ward and Myers, 2005). However, it is important to note that certain studies have not found such evidence, or only limited evidence of trophic cascades. This is the case of Botsford et al. (1997) and Pace et al. (1999), who made some of first comprehensive assessments of the potential impacts of marine fisheries. Both studies reviewed the potential trophic cascade in the Bering Sea stemming from fisheries-induced fluctuations in the abundance of pink salmon (Oncorhynchus gorbuscha) (Shiomoto et al., 1997). However, the evidence was weak and was only statistically significant between two trophic levels (Shiomoto et al., 1997). A report by the Western and Central Pacific Fisheries Commission recently demonstrated decreasing abundance trends in pelagic Hawaiian waters for five species of trophic level 4.0 or higher and increasing trends for four species of trophic level 3.9 or lower (Allain et al., 2012). In another ecosystem modelling study of the Central North Pacific, Kitchell et al. (2006) assessed changes in community structure as a result of increases in fishing mortality of different predatory species (billfishes, sharks and tunas). They found that the removal of billfishes and sharks led to weak effects on the structure of the Central North Pacific marine community, suggesting that top predators in open-ocean systems may not always be keystone species. Increasing the fishing mortality of yellowfin tuna (Thunnus albacares), however, led to rapid changes in the trophic structure of the system, which was attributed to their role as both predator and prey. Nonetheless, they concluded that none of the predatory taxa were indispensable for the functioning of the ecosystem, as the dietary composition and range of many of the predators overlapped (Kitchell et al., 2006). This question of the “keystoneness” of species in oceanic environments was partly addressed in a recent study, which quantified the keystone role of species in marine communities through three different indices in over 100 Ecopath with Ecosim (EwE) models; 19 of which were oceanic models (Valls et al., 2015). Keystone species were identified in five of the models, however, only one of the models belonged to an open-ocean ecosystem (Kitchell et al., 2002); this model identified blue marlin (Makaira nigricans) as a keystone species in the Central Pacific Ocean. The limited evidence together with the opportunistic nature of feeding behaviour in the open-ocean suggests a limited role for keystone species in this environment. The concurrent exploitation of multiple species, as seen in Kitchell et al. (2006), makes it more difficult for ecosystem modellers to discern the trophic mechanisms shaping the biological community. Perhaps with the exception of the pole and line fishery, most open-ocean pelagic fisheries are multispecies fisheries, which target different trophic levels simultaneously. The interaction between different fisheries targeting different trophic levels in the same ecosystem may offset fishery-induced trophic imbalances in the community (Andersen and Pedersen, 2010). In the CNP, it was demonstrated that the purse-seine fishery reduced the abundance of skipjack tuna, however, parallel declines in big eye tuna (Thunnus obesus), one of its natural predators, resulted in a partial predatory release on skipjack, which reduced the overall impact of fisheries on the community structure (Cox et al., 2002). The simultaneous exploitation of different trophic levels may thus mask the trophic effects associated with declines of upper trophic level predators, which does not imply that there is no impact on the community, but that the depletion across trophic levels is not destabilizing. In the context of a trophic cascade, declines in body size of apex predators can result in body sizes of lower trophic level taxa either being maintained or increased (Ward and Myers, 2005). A reduction in the average body size of a predatory species reduces the size ratio between predator and prey and can thus reduce the magnitude of the top-down trophic control that the predator exerts on the system (Woodward et al., 2005). Animal body size is also positively correlated with parameters such as longevity and trophic status, and negatively correlated with factors such as the rates of growth and turnover of a species. Thus, changes in body size of species may affect the trophic interactions of the species, which in turn affect the stability and rate of propagation of trophic control mechanisms through the system (Emmerson and Raffaelli, 2004; Woodward et al., 2005). Recent studies have highlighted how these fishing-induced trophic imbalances caused by the heavy exploitation of predatory species may increase the abundance of commercially valuable fish species, thus allowing for the creation or expansion of fisheries that target these lower trophic level released prey (i.e. cultivation effects) (Brown and Trebilco, 2014). While the concept of fisheries benefiting from large-scale biomanipulation is not new (Brander, 2010; Lindegren et al., 2010), caution must be exercised, as the food-web impacts of fishing may also lead to the proliferation of commercially unattractive species (Brown and Trebilco, 2014); as shown by Ward and Myers (2005) with the increase in pelagic stingray (Dasyatis violacea) abundance; an elasmobranch species of low commercial value (Báez et al., 2015). Additional research demonstrates the proliferation of species of low economic interest for which no fisheries have been created (Carscadden et al., 2001; Daskalov, 2002; Walters and Kitchell, 2001). Non-consumptive effects Decreases in the abundance of predatory species may also be causing changes in the behavioural dynamics of open-ocean communities, which indirectly affect the trophodynamics. These are known as “non-consumptive”, “trait-mediated” or “risk” effects. Given that prey respond to the presence of predatory species through a series of traits aimed to reduce mortality, the reduction in top-down predator pressure may cause behavioural changes that propagate to other species groups in the community (Peacor and Werner, 2008). In certain cases, non-consumptive effects can also induce changes in prey growth and development (Peckarsky et al., 2008). Better understanding of these dynamics may help explain top-down trophic controls in open-ocean systems (Baum and Worm, 2009). Ecosystem-level impacts Healthy marine ecosystems provide a series of services which maintain the functionality of the system and provide for a variety of societal needs, which include the provision of protein and micronutrients for millions of people worldwide (Holmlund and Hammer, 1999; Postel et al., 2012). Fisheries mismanagement, overfishing, bycatch, and IUU fishing not only threaten the availability of food for millions of people, but may also lead to irreversible changes in the integrity and state of marine ecosystems and the ecosystem services they provide. Fisheries are thus considered a key industry in addressing food security concerns (Godfray et al., 2010; FAO, 2014). Our understanding of the ecosystem-level impacts of biotic exploitation in coastal systems is very developed. Studies have shown that the impacts of fisheries exploitation range from alterations in primary productivity and changes of the physical environment, such as coastal erosion (Estes and Duggins, 1995), to changes in both ecosystem structure and function at large spatial scales (Dulvy et al., 2004). In the case of coastal reef systems, it has been demonstrated that the overharvesting of higher trophic level species has led to profound changes in ecosystem structure and function (Dulvy et al., 2004). Because the ecosystem-level impacts in offshore oceanic fisheries are less well studied, inferences from studies of similar systems must pave the way for new research avenues. Changes in ecosystem-state and biodiversity In the open-ocean, where the water column generally lacks physical habitat, changes at the ecosystem level are mostly expressed as transitions between alternative states of the ecosystem. These “regime shifts” affect both the system’s dynamics and functionality (Scheffer and Carpenter, 2003; Daskalov et al., 2007; de Young et al., 2008; Beaugrand et al., 2015). In the marine realm, this concept was first applied to describe synchronicities between climatological and fish stock indices in coastal ecosystems (Steele, 2004; Wooster and Zhang, 2004) and since then, it has been used to describe general disruptions of ecosystem structure and function (Möllmann and Diekmann, 2012). Although regime shifts can be induced through anthropogenic stressors, most of the studied regime shifts in pelagic marine systems have been triggered by large scale climatological processes, which have led to structural changes in the functioning of the biological community (Hare and Mantua, 2000; Möllmann et al., 2009). This review focuses on the role that a top-down anthropogenic stressor—in this case, open-ocean fisheries—can play in triggering regime shifts in open-ocean ecosystems and how they may interact with other drivers such as climate to reach these tipping points or stability thresholds. A shift between ecosystem states depends on two main factors: the magnitude of the perturbation (whether anthropogenic or natural, biotic or abiotic) that drives the shift, and the current condition of the ecosystem when the perturbation takes place, a concept known as the “size of its attraction basin” (Scheffer et al., 2001; Scheffer and Carpenter, 2003; Möllmann and Diekmann, 2012). Based on this concept, there is an inverse relationship between the integrity of the ecosystem and the magnitude of the stressor which would lead to a regime shift: where a weak stressor may cause a regime shift in a “stressed” system and a much larger stressor would be needed to have the same effect on a “healthy” system (Möllmann and Diekmann, 2012). Stressed systems, where reductions in biodiversity or changes community structure have taken place, will have a smaller attraction basin, which translates to a reduction in the levels of perturbations that they can withstand, i.e. its resilience. Ample evidence from studies in pelagic, non-oceanic systems supports the claim that regime shifts are more likely to occur when the resilience of an ecosystem is diminished by actions such as the reduction of biodiversity, removal of functional groups of species, or trophic levels from a biological community (Folke et al., 2004; Worm et al., 2006). Different studies have concluded that commercial fishing is the main driver of long-term loss of diversity in open-ocean predators across all ocean basins in addition to reductions in oceanic predator abundance and size (Worm et al., 2005; Ward and Myers, 2005). While some of these studies have been criticized (reviewed in Banobi et al., 2011), those critiques question the magnitude of the declines in the abundance of species, not the impact of loss of biodiversity on the system. Given the connection between fisheries and biodiversity loss in open-ocean ecosystems, it is unsurprising that fisheries exploitation has also been implicated in regime shifts in pelagic systems (Daskalov et al., 2007; Möllmann et al., 2009). Worm et al. (2006) alleged that these losses in marine biodiversity could compromise the ability that marine ecosystems have to provide ecosystem services such as seafood provisioning. Over the last 60 years the biodiversity of open-ocean predators across all ocean basins has declined between 10 and 50%; these trends coincide with increases in fishing pressure, while no trend was found between these changes in diversity and major decadal changes in oceanography during the study period (1960s–1990s) (Worm et al., 2005). It is noteworthy that the declines in tuna and billfish diversity were more pronounced in intensely fished tropical areas, where species richness and density had a strong inverse relationship with fisheries catch values from the 1950s until the early 2000s (Worm et al., 2005). Evidence of regime shifts in pelagic systems No fishing-induced regime shifts have been identified in open-ocean ecosystems as defined by this paper. However, a number of very large pelagic ecosystems (e.g. enclosed seas and continental shelves) have encountered regime shifts, and are reviewed here to illustrate the potential for regime shifts to happen in dynamic pelagic systems. Although there is a lack of empirical evidence of abrupt ecosystem-level oceanic changes induced solely by fishing, heavy fisheries exploitation may be gradually corroding the resilience of the system by reducing its biodiversity and restructuring its biological community; making it more vulnerable to regime shifts when exposed to changes in climate. However, the relationship could be reversed if climatological factors push the system towards a tipping point, which is reached by the top-down pressure of fisheries. Subsequent studies have evaluated the community-level trophic changes that have taken place in the Eastern Scotian Shelf ecosystem along with the collapse of the stock of Atlantic cod (Gadus morhua), to assess the ecosystem-level changes in the Eastern Scotian Shelf from 1960 to 2002 (Choi et al., 2004,, 2005). The analysis revealed ecosystem changes of the system during the 1970s and 1990s, and identified that changes in variables related to the abundance of upper trophic level species and conditions of these, such as size and body mass, were the principal explanatory elements of the ecosystem changes in the 1990s (Choi et al., 2005). The author, however, stated that the fishery-induced changes could not explain the ecosystem shift alone. Further multivariate analysis demonstrated that climatological changes between the mid-1970s and late 1980s and between the late 1980s and late 1990s, in the form of changes in water temperature or oceanic front positions, interacted with the fisheries-induced ecological effects and led to the regime shift of the system. Kenny et al. (2009) reached similar conclusions in the North Sea, where the authors interpreted that abiotic changes, in the form of an abrupt water temperature increase by the late 1980s, catalysed the shift in ecosystem state, which had most likely been started by the interaction of intense fishing pressure and gradual sea surface warming (Beaugrand, 2004). The state shifts described by the authors (1983–1993 and 1993–2003) involved a change in the control mechanisms for the pelagic stocks, from top-down (fishery) control prior to the shift, to bottom-up (climatological) control after the regime shift (Kenny et al., 2009). The Black Sea ecosystem has undergone profound ecological changes since the 1970s, and may be a good indicator of how the cumulative impacts of biotic, abiotic, and anthropogenic stressors can lead to several shifts in states in large pelagic systems (Daskalov et al., 2007; Oguz and Gilbert, 2007). Although the Black Sea is almost an entirely land-locked basin, which does not fulfil all the requirements of what we define as an open-ocean ecosystem, we find that its characteristics (e.g. average depth of 1253 m and holopelagic community) are similar enough to that of open-ocean marine ecosystems to be used as a comparative example. From 1960 to 2000, the Black Sea experienced multiple regime shift episodes that were triggered by fisheries exploitation, changes in its biological community, climatological events, and eutrophication (Oguz and Gilbert, 2007). The depletion of pelagic stocks caused a trophic cascade in the ecosystem, which together with abiotic changes in the system (nutrients and temperature), led to a regime shift in the early 1970s. Carnivorous plankton became a dominant taxonomic group until the pelagic fish populations recovered during the late 1990s. Their recovery acted together with a reversal of the climatic state and reductions in nutrient loading, to revert the system to its original “low production” regime state (Oguz and Gilbert, 2007). Restoring the state of a system to that prior to a regime shift is a challenging objective. Not only is it unlikely that an ecosystem is able to return to its original state, but also studies show that adjusting the sources of external pressure (e.g. fishing) to the levels prior to the shift, will be costly and insufficient to restore the biotic balance (Möllmann and Diekmann, 2012). Different studies indicate that restoring the ecosystem balance would require a reduction of external pressures at much more pronounced levels than the original levels that led to the regime shift which, in terms of fisheries, would imply significant socioeconomic impacts (Suding et al., 2004; Möllmann and Diekmann, 2012). Discussion All types of marine systems, including coastal, open-ocean and deep sea ecosystems, can be subject to the three types of fisheries-induced ecological impacts discussed in this this paper. The way these impacts are manifested across ecosystems and the recovery rates of the systems vary. Species-level impacts from fisheries in open-ocean ecosystems are (or are likely to be) the same as those impacts on coastal or deep sea ecosystems. Evidence for community level impacts that mimic coastal and deep sea ecosystems also exists in open-ocean ecosystems, though much longer time-series have been required to identify them. The major difference between the systems comes in the form of ecosystem-levels impacts. In oceanic ecosystems, these take the form of changes in biological community structure, composition, and dynamics and no evidence of impacts on the abiotic environment have been identified to date. On the other hand, the ecosystem level impacts on coastal and deep sea ecosystems can result in the deterioration of physical habitat such as (tropical or cold-water) coral reefs or result in biochemical changes in the fluid environment (Jackson et al., 2001) in addition to population, community, and ecosystem-level impacts. Beyond differences in impacts to coastal, deep sea and oceanic ecosystems, there are difference in governance that directly influences our ability to monitor, understand, and manage impacts. Unlike nationally managed coastal fisheries, deep sea, and oceanic fisheries cross jurisdictional boundaries and are largely managed through international agreements. Among its many mandates, the 1982 UN Convention on the Law of the Sea (UNCLOS) requires Parties to cooperate in the establishment of regional or subregional fisheries management organizations, intended for the conservation and management of living resources within jurisdictional waters and the high seas (Part VII, Section 2, Article 118) (UNCLOS, 1982). UNCLOS entered into force in 1992 and by 1995 had been built on by the UNFSA (1995). The UNFSA promoted the conservation and management of straddling and highly migratory fish stocks through an ecosystem-based approach (UNFSA, 1995: General Principles—Article 5), exercised both within and beyond the 200 nm jurisdictional boundary of coastal states. The components of the ecosystem approach are derived from the mandate in the UNFSA, and laid out specifically in a FAO technical report (Garcia et al., 2003) and the Code of Conduct for Responsible Fisheries (FAO, 1995; UNFSA, 1995). The mandate includes requirements for monitoring and managing impacts not just on target species, but to “species belonging to the same ecosystem or associated with or dependent upon the target stocks” (FAO, 1995; UNFSA, 1995). Although an ecosystem approach to fisheries has been incorporated into the mission of many RFMOs, Gilman et al. (2014) found that up to five of the 17 RFMOs do not work towards mitigating their impacts on non-target fish species and non-fish species, while the rest undertake actions with different success rates. While some tuna-RFMOs have expanded their management efforts to account for impacts on other species groups such as sharks, the current, single-stock assessment approach that dominates RFMO management does not account for impacts on non-target species and the marine biological community as a whole. Gilman et al. (2014) further estimated that only one third of the bycatch problems are addressed through legally binding measures and that over two thirds of RFMO fisheries lack adequate observer coverage. This lack of observer coverage feeds into a more general problem that underlies why it has taken so long to identify impacts of fisheries on open-ocean ecosystems: the limited number of complete and reliable multispecies fisheries catches time series. In this manuscript, we have demonstrated the importance of long-term multispecies catch datasets and stock assessments for understanding not just population-level impacts on target and non-target taxa, but also to parameterize community-level mass-balance models to demonstrate community and ecosystem-level impacts of fishing on the open-ocean. However, such datasets are based on observer monitoring programs which are still absent in many RFMOs. At least one RFMO with competency for pelagic species had no observer coverage as of 2013 (Gilman et al., 2014). As monitoring strategies improve across RFMOs, more pressure should be placed on fishing nations, which have the responsibility of submitting high quality observer data so that RFMOs can do their job effectively. Further attention will have to be placed on the spatial coverage of observer programs of each fishing nation. The biases in taxonomic identification and spatial coverage across fisheries and RFMOs contribute and widen some of the current knowledge gaps about fisheries impacts on certain species. In a recent study on the global trends of shark bycatch, Molina and Cooke (2012) highlighted the regional and taxonomical bias found in 103 papers on shark bycatch, and noted that the South Atlantic, South Pacific, and Indian Oceans and commercially unimportant shark species (such as species of the order Hexanchiformes and Orectolobiformes) were underrepresented in the shark bycatch literature. If long-term multispecies monitoring programs are not established, we will continue to remain blind to the broader ecological impacts of fisheries on open-ocean ecosystems and at risk of failing to recognize early warning signals of trophic cascades or fisheries-induced regime shifts. To ensure the sustainability of open-ocean fisheries, the extent and thematic coverage of observer programs must be increased and include non-target species, as well as other forms of monitoring such as community-level modelling efforts and genetic sampling. Genetic monitoring of harvested wild populations is the most powerful method of tracing genetic changes induced by exploitation (Allendorf et al., 2008). A number of challenges to effective monitoring of open-ocean ecosystems by RFMOs exist. Competency for the management of species in a single ocean basin can be divided among RFMOs, leading to shared management of resources and impacts. Strongly coordinated monitoring by RFMOs of a shared ecosystem is essential. While RFMOs clearly have a duty to monitor ecosystem components beyond target species, even strong coordination among RFMOs is unlikely to be sufficient to monitor species, community, and ecosystem level indicators given current budgets. There is a strong need for enhanced cooperation between organizations with competency for managing open-ocean ecosystems and large-scale biodiversity monitoring programs like the Global Ocean Observing Systems (GOOS). Similarly, the analytical requirements related to monitoring of ecosystem impacts go beyond the capacity of the RFMOs and may require collaborations with industry and academic institutions. The current barriers to such collaborations are largely constructed from lack of funding, poor communication on all sides, and data availability. Only by increasing coordination within RFMOs, cooperation between RFMOs and other competent organizations, and collaboration between RFMOs, industry, and academia will begin to be able to appropriately monitor, and thus, manage open-ocean ecosystems. The cost of mismanaging open-ocean biological resources extends from the ecological dimension into the socioeconomic dimension. Evidence indicates that reversing the ecological impacts of regime shifts would be more costly in socioeconomic and management terms than applying a precautionary approach, which would prevent the shift in ecosystem state by avoiding trophic imbalances and loss of biological diversity (Suding et al., 2004; Möllmann and Diekmann, 2012). Moreover, given that in 2013 fish represented 17% of the global intake of animal protein (FAO, 2016), the social cost of unhealthy open-ocean ecosystems in the terms of food security is too high to ignore. With human population estimates exceeding 8 billion in 2025 and reaching up to 9.7 billion in 2050, it becomes incontestable that the management of marine fisheries and their impacts in the open-ocean over the next few decades will have implications in both ecological and human dimensions worldwide (FAO, 2016). Footnotes 1 " Straddling stocks are stocks of fish such as Pollock, which migrate between, or occur in both, the economic exclusive zone (EEZ) of one or more states and the high seas” (UNAtlas, 2010). 2 " Preparatory Committee established by General Assembly resolution 69/292 “Development of an international legally-binding instrument under the United Nations Convention on the Law of the Sea on the conservation and sustainable use of marine biological diversity of areas beyond national jurisdiction” (28th March to 8th April, 2016). 3 " Review Conference on the Agreement for the Implementation of the Provisions of the United Nations Convention on the Law of the Sea of 10 December 1982 relating to the Conservation and Management of Straddling Fish Stocks and Highly Migratory Fish Stocks (23rd to 27th May, 2016). References Aebischer N. J. , Coulson J. C., Colebrook J. M. 1990 . Parallel long-term trends across four marine trophic levels and weather . Nature , 347 : 753 – 755 . Google Scholar Crossref Search ADS WorldCat Ainley D. , Ballard G., Ackley S., Blight L. K., Eastman J. T., Emslie S. D., Lescroël A., et al. 2007 . Paradigm lost, or is top-down forcing no longer significant in the Antarctic marine ecosystem? Antarctic Science , 19 : 283 – 290 . Google Scholar Crossref Search ADS WorldCat Allain V. , Griffiths S. P., Polovina J., Nicol S. 2012 . WCPO ecosystem indicator trends and results from ecopath simulations. In Eighth Meeting of the Scientific Committee of the Western and Central Pacific Fisheries Commission, WCPFC-SC8, Busan, Republic of Korea, 7–15 August 2012, WCPFC-SC8-2012, pp. 1–29. Allendorf F. W. 1986 . Genetic drift and the loss of alleles versus heterozygosity . Zoo Biology , 5 : 181 – 190 . Google Scholar Crossref Search ADS WorldCat Allendorf F. W. , England P. R., Luikart G., Ritchie P. A., Ryman N. 2008 . Genetic effects of harvest on wild animal populations . Trends in Ecology & Evolution , 23 : 327 – 337 . Google Scholar Crossref Search ADS PubMed WorldCat Alverson D. L. , Freeberg M. H., Murawaski S. A., Pope J. G. 1994 . A Global Assessment of Fisheries Bycatch and Discards. Rome: FAO fisheries technical paper no. 339. Andersen K. H. , Pedersen M. 2010 . Damped trophic cascades driven by fishing in model marine ecosystems . Proceedings of the Royal Society B: Biological Sciences , 277 : 795 – 802 . Google Scholar Crossref Search ADS WorldCat Anderson C. N. , Hsieh C. H., Sandin S. A., Hewitt R., Hollowed A., Beddington J., May R. M., et al. 2008 . Why fishing magnifies fluctuations in fish abundance . Nature , 452 : 835 – 839 . Google Scholar Crossref Search ADS PubMed WorldCat Anderson O. R. , Small C. J., Croxall J. P., Dunn E. K., Sullivan B. J., Yates O., Black A. 2011 . Global seabird bycatch in longline fisheries . Endangered Species Research , 14 : 91 – 106 . Google Scholar Crossref Search ADS WorldCat Angel M. V. , de C. Baker A. 1982 . Vertical distribution of the standing crop of plankton and micronekton at three stations in the northeast Atlantic . Biological Oceanography , 2 : 1 – 30 . OpenURL Placeholder Text WorldCat Angel M. V. 2003 . The pelagic environment of the open-ocean . In: Tyler, P.A., (ed.) Ecosystems of the World, Vol. 28. Ecosystems of the Deep Oceans. Elsevier, Amsterdam, The Netherlands, 39–79, 569 pp. Anticamara J. A. , Watson R., Gelchu A., Pauly D. 2011 . Global fishing effort (1950–2010): trends, gaps, and implications . Fisheries Research , 107 : 131 – 136 . Google Scholar Crossref Search ADS WorldCat Báez J. C. , Crespo G. O., García-Barcelona S., de Urbina J. M. O., José M., Macías D. 2015 . Understanding pelagic stingray (Pteroplatytrygon violacea) by-catch by Spanish longliners from the Mediterranean Sea . Collective Volume of Scientific Papers ICCAT , 71 : 2623 – 2632 . OpenURL Placeholder Text WorldCat Banobi J. A. , Branch T. A., Hilborn R. 2011 . Do rebuttals affect future science? Ecosphere , 2 : 1 – 11 . Google Scholar Crossref Search ADS WorldCat Baum J. K. , Myers R. A. 2004 . Shifting baselines and the decline of pelagic sharks in the Gulf of Mexico . Ecology Letters , 7 : 135 – 145 . Google Scholar Crossref Search ADS WorldCat Baum J. K. , Myers R. A., Kehler D. G., Worm B., Harley S. J., Doherty P. A. 2003 . Collapse and conservation of shark populations in the northwest Atlantic . Science , 299 : 389 – 392 . Google Scholar Crossref Search ADS PubMed WorldCat Baum J. K. , Worm B. 2009 . Cascading top‐down effects of changing oceanic predator abundances . Journal of Animal Ecology , 78 : 699 – 714 . Google Scholar Crossref Search ADS PubMed WorldCat Beaugrand G. 2004 . The North Sea regime shift: evidence, causes, mechanisms and consequences . Progress in Oceanography , 60 : 245 – 262 . Google Scholar Crossref Search ADS WorldCat Beaugrand G. , Conversi A., Chiba S., Edwards M., Fonda-Umani S., Greene C., Mantua N., et al. 2015 . Synchronous marine pelagic regime shifts in the Northern Hemisphere . Philosophical Transactions of the Royal Society of London B: Biological Sciences , 370 : 20130272. Google Scholar Crossref Search ADS WorldCat Bellwood D. R. , Hughes T. P., Folke C., Nyström M. 2004 . Confronting the coral reef crisis . Nature , 429 : 827 – 833 . Google Scholar Crossref Search ADS PubMed WorldCat Berkeley S. A. , Hixon M. A., Larson R. J., Love M. S. 2004 . Fisheries sustainability via protection of age structure and spatial distribution of fish populations . Fisheries , 29 : 23 – 32 . Google Scholar Crossref Search ADS WorldCat Birdlife International . 2004 . Tracking Ocean Wanderers: The Global Distribution of Albatrosses and Petrels. In Global Procellariiform Tracking Workshop. Cambridge: BirdLife International. Birkeland C. , Dayton P. K. 2005 . The importance in fishery management of leaving the big ones . Trends in Ecology & Evolution , 20 : 356 – 358 . Google Scholar Crossref Search ADS PubMed WorldCat Block B. A. , Jonsen I. D., Jorgensen S. J., Winship A. J., Shaffer S. A., Bograd S. J., Hazen E. L., et al. 2011 . Tracking apex marine predator movements in a dynamic ocean . Nature , 475 : 86 – 90 . Google Scholar Crossref Search ADS PubMed WorldCat Block B. A. , Dewar H., Blackwell S. B., Williams T. D., Prince E. D., Farwell C. J., Boustany A., et al. 2001 . Migratory movements, depth preferences, and thermal biology of Atlantic bluefin tuna . Science , 293 : 1310 – 1314 . Google Scholar Crossref Search ADS PubMed WorldCat Bolten A. B. 2003 . Variation in sea turtle life history patterns: neritic vs. oceanic developmental stages . The Biology of Sea Turtles , 2 : 243 – 257 . OpenURL Placeholder Text WorldCat Botsford L. W. , Castilla J. C., Peterson C. H. 1997 . The management of fisheries and marine ecosystems . Science , 277 : 509 – 515 . Google Scholar Crossref Search ADS WorldCat Branch T. A. , Lobo A. S., Purcell S. W. 2013 . Opportunistic exploitation: an overlooked pathway to extinction . Trends in Ecology & Evolution , 28 : 409 – 413 . Google Scholar Crossref Search ADS PubMed WorldCat Brander K. M. 2007 . Global fish production and climate change . Proceedings of the National Academy of Sciences of the United States of America , 104 : 19709 – 19714 . Google Scholar Crossref Search ADS PubMed WorldCat Brander K. 2010 . Reconciling biodiversity conservation and marine capture fisheries production . Current Opinion in Environmental Sustainability , 2 : 416 – 421 . Google Scholar Crossref Search ADS WorldCat Brothers N. P. , Cooper J., Lokkeborg S., Al E. 1999 . The incidental catch of seabirds by longline fisheries: worldwide review and technical guidelines for mitigation. FAO Fisheries Circular No. 937, Rome. Brothers N. P. 1995 . Catching Fish Not Birds: A Guide to Improving Your Longline Fishing Efficiency. Australian Longline Version. Australia Parks and Wildlife Service, Hobart, Australia, 32 p. Brown C. J. , Trebilco R. 2014 . Unintended cultivation, shifting baselines, and conflict between objectives for fisheries and conservation . Conservation Biology , 28 : 677 – 688 . Google Scholar Crossref Search ADS PubMed WorldCat Burgess G. H. , Beerkircher L. R., Cailliet G. M., Carlson J. K., Cortés E., Goldman K. J., Grubbs R. D., et al. 2005 . Is the collapse of shark populations in the Northwest Atlantic Ocean and Gulf of Mexico real? Fisheries , 30 : 19 – 26 . Google Scholar Crossref Search ADS WorldCat Campana S. E. , Joyce W., Manning M. J. 2009 . Bycatch and discard mortality in commercially caught blue sharks Prionace glauca assessed using archival satellite pop-up tags . Marine Ecology Progress Series , 387 : 241 – 253 . Google Scholar Crossref Search ADS WorldCat Carscadden J. E. , Frank K. T., Leggett W. C. 2001 . Ecosystem changes and the effects of capelin (Mallotus villosus), a major forage species . Canadian Journal of Fisheries and Aquatic Sciences , 58 : 73 – 85 . Google Scholar Crossref Search ADS WorldCat CCSBT . 2014 . Report of the Nineteenth Meeting of the Scientific Committee. Canberra, Australia. 115 http://www.ccsbt.org/userfiles/file/docs_english/meetings/meeting_reports/ccsbt_21/report_of_SC19.pdf (last accessed 20 August 2015). Chassot E. , Melin F., Le Pape O., Gascuel D. 2007 . Bottom-up control regulates fisheries production at the scale of eco-regions in European seas . Marine Ecology Progress Series , 343 : 45 – 55 . Google Scholar Crossref Search ADS WorldCat Chavez F. P. , Ryan J., Lluch-Cota S. E., Ñiquen C. M. 2003 . From anchovies to sardines and back: multidecadal change in the Pacific Ocean . Science , 299 : 217 – 221 . Google Scholar Crossref Search ADS PubMed WorldCat Choi J. , Frank K., Petrie B., Leggett W. 2005 . Integrated assessment of a large marine ecosystem: a case study of the devolution of the eastern Scotian Shelf, Canada . Oceanography and Marine Biology , 43 : 67 – 93 . OpenURL Placeholder Text WorldCat Choi J. S. , Frank K. T., Leggett W. C., Drinkwater K. 2004 . Transition to an alternate state in a continental shelf ecosystem . Canadian Journal of Fisheries and Aquatic Sciences , 61 : 505 – 510 . Google Scholar Crossref Search ADS WorldCat Clark M. 2001 . Are deepwater fisheries sustainable? The example of orange roughy (Hoplostethus atlanticus) in New Zealand . Fisheries Research , 51 : 123 – 135 . Google Scholar Crossref Search ADS WorldCat Clark M. R. , Althaus F., Schlacher T. A., Williams A., Bowden D. A., Rowden A. A. 2016 . The impacts of deep-sea fisheries on benthic communities: a review . ICES Journal of Marine Science , 73 : i51 – i69 . Google Scholar Crossref Search ADS WorldCat Clarke S. C. , McAllister M. K., Milner-Gulland E. J., Kirkwood G. P., Michielsens C. G. J., Agnew D. J., Pikitch E. K. 2006 . Global estimates of shark catches using trade records from commercial markets . Ecology Letters , 9 : 1115 – 1126 . Google Scholar Crossref Search ADS PubMed WorldCat Coggins L. G. , Catalano M. J., Allen M. S., Pine W. E., Walters C. J. 2007 . Effects of cryptic mortality and the hidden costs of using length limits in fishery management . Fish and Fisheries , 8 : 196 – 210 . Google Scholar Crossref Search ADS WorldCat Colléter M. , Valls A., Guitton J., Gascuel D., Pauly D., Christensen V. 2015 . Global overview of the applications of the Ecopath with Ecosim modeling approach using the EcoBase models repository . Ecological Modelling , 302 : 42 – 53 . Google Scholar Crossref Search ADS WorldCat Conover D. O. , Munch S. B. 2002 . Sustaining fisheries yields over evolutionary time scales . Science , 297 : 94 – 96 . Google Scholar Crossref Search ADS PubMed WorldCat Cook R. 2003 . The Magnitude and Impact of By-catch Mortality by Fishing Gear. In Responsible Fisheries in the Marine Ecosystem. Ed. by Sinclair, M., and Valdimarsson, G. FAO, Rome. Courchamp F. , Angulo E., Rivalan P., Hall R. J., Signoret L., Bull L., Meinard Y. 2006 . Rarity value and species extinction: the anthropogenic Allee effect . PLoS Biology , 4 : e415. Google Scholar Crossref Search ADS PubMed WorldCat Cox S. P. , Essington T. E., Kitchell J. F., Martell S. J. D., Walters C. J., Boggs C., Kaplan I. 2002 . Reconstructing ecosystem dynamics in the central North Pacific Ocean, 1952-1998. II. A preliminary assessment of the trophic impacts of fishing and effects on tuna dynamics . Canadian Journal of Fisheries and Aquatic Sciences , 59 : 1736 – 1747 . Google Scholar Crossref Search ADS WorldCat Cullis-Suzuki S. , Pauly D. 2010 . Failing the high seas: a global evaluation of regional fisheries management organizations . Marine Policy , 34 : 1036 – 1042 . Google Scholar Crossref Search ADS WorldCat Cury P. , Bakun A., Crawford R. J. M., Jarre A., Quinones R. A., Shannon L. J., Verheye H. M. 2000 . Small pelagics in upwelling systems: patterns of interaction and structural changes in "wasp-waist" ecosystems . ICES Journal of Marine Science , 57 : 603 – 618 . Google Scholar Crossref Search ADS WorldCat Cushing D. H. 1975 . Marine Ecology and Fisheries . Cambridge University Press , London, UK . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Daskalov G. M. 2002 . Overfishing drives a trophic cascade in the Black Sea . Marine Ecology Progress Series , 225 : 53 – 63 . Google Scholar Crossref Search ADS WorldCat Daskalov G. M. , Grishin A. N., Rodionov S., Mihneva V. 2007 . Trophic cascades triggered by overfishing reveal possible mechanisms of ecosystem regime shifts . Proceedings of the National Academy of Sciences of the United States of America , 104 : 10518 – 10523 . Google Scholar Crossref Search ADS PubMed WorldCat Dayton P. K. , Thrush S. F., Agardy M. T., Hofman R. J. 1995 . Environmental effects of marine fishing . Aquatic Conservation: marine and Freshwater Ecosystems , 5 : 205 – 232 . Google Scholar Crossref Search ADS WorldCat de Young B. , Barange M., Beaugrand G., Harris R., Perry R. I., Scheffer M. 2008 . Regime shifts in marine ecosystems: detection, prediction and management . Trends in Ecology & Evolution , 23 : 402 – 409 . Google Scholar Crossref Search ADS PubMed WorldCat Denney N. H. , Jennings S., Reynolds J. D. 2002 . Life–history correlates of maximum population growth rates in marine fishes . Proceedings of the Royal Society of London B: Biological Sciences , 269 : 2229 – 2237 . Google Scholar Crossref Search ADS WorldCat Dulvy N. K. , Freckleton R. P., Polunin N. V. 2004 . Coral reef cascades and the indirect effects of predator removal by exploitation . Ecology Letters , 7 : 410 – 416 . Google Scholar Crossref Search ADS WorldCat Dulvy N. K. , Sadovy Y., Reynolds J. D. 2003 . Extinction vulnerability in marine populations . Fish and Fisheries , 4 : 25 – 64 . Google Scholar Crossref Search ADS WorldCat Dulvy N. K. , Baum J. K., Clarke S., Compagno L. J., Cortes E., Domingo A., Fordham S., et al. 2008 . You can swim but you can't hide: the global status and conservation of oceanic pelagic sharks and rays . Aquatic Conservation: Marine and Freshwater Ecosystems , 18 : 459 – 482 . Google Scholar Crossref Search ADS WorldCat Edwards M. , Richardson A. J. 2004 . Impact of climate change on marine pelagic phenology and trophic mismatch . Nature , 430 : 881 – 884 . Google Scholar Crossref Search ADS PubMed WorldCat Emmerson M. C. , Raffaelli D. 2004 . Predator–prey body size, interaction strength and the stability of a real food web . Journal of Animal Ecology , 73 : 399 – 409 . Google Scholar Crossref Search ADS WorldCat Estes J. A. , Palmisano J. F. 1974 . Sea otters: their role in structuring nearshore communities . Science , 185 : 1058 – 1060 . Google Scholar Crossref Search ADS PubMed WorldCat Estes J. A. , Duggins D. O. 1995 . Sea otters and kelp forests in Alaska: generality and variation in a community ecological paradigm . Ecological Monographs , 65 : 75 – 100 . Google Scholar Crossref Search ADS WorldCat FAO . 1995 . Code of Conduct for Responsible Fisheries. FAO, Rome. FAO . 2014 . United Nations Food and Agriculture Organization. 2014. The state of world fisheries and aquaculture 2014. United Nations Food and Agriculture Organization, Rome. FAO . 2016 . United Nations Food and Agriculture Organization. 2016. The state of world fisheries and aquaculture 2016. United Nations Food and Agriculture Organization, Rome. Folke C. , Carpenter S., Walker B., Scheffer M., Elmqvist T., Gunderson L., Holling C. S. 2004 . Regime shifts, resilience, and biodiversity in ecosystem management . Annual Review of Ecology, Evolution, and Systematics , 557 – 581 . OpenURL Placeholder Text WorldCat Fowler S. 2014 . The Conservation Status of Migratory Sharks. UNEP/CMS Secretariat, Bonn, Germany. Francis R. C. , Hixon M. A., Clarke M. E., Murawski S. A., Ralston S. 2007 . Ten commandments for ecosystem-based fisheries scientists . Fisheries , 32 : 217 – 233 . Google Scholar Crossref Search ADS WorldCat Frederiksen M. , Edwards M., Richardson A. J., Halliday N. C., Wanless S. 2006 . From plankton to top predators: bottom-up control of a marine food web across four trophic levels . Journal of Animal Ecology , 75 : 1259 – 1268 . Google Scholar Crossref Search ADS PubMed WorldCat Fromentin J. M. , Powers J. 2005 . Atlantic bluefin tuna: population dynamics, ecology, fisheries and management . Fish and Fisheries , 6 : 281 – 306 . Google Scholar Crossref Search ADS WorldCat Fromentin J. M. , Lopuszanski D. 2013 . Migration, residency, and homing of bluefin tuna in the western Mediterranean Sea . ICES Journal of Marine Science , 71 : 510 – 518 . Google Scholar Crossref Search ADS WorldCat Garcia S. M. , Zerbi A., Aliaume C., Do Chi T., Lasserre G. 2003 . The ecosystem approach to fisheries. Issues, terminology, principles, institutional foundations, implementation and outlook. FAO Fisheries Technical Paper. No. 443: 71 p. Garcia S. M. , Kolding J., Rice J., Rochet M. J., Zhou S., Arimoto T., Beyer J. E., et al. 2012 . Reconsidering the consequences of selective fisheries . Science , 335 : 1045 – 1047 . Google Scholar Crossref Search ADS PubMed WorldCat Gilman E. , Brothers N., Kobayashi D. R. 2005 . Principles and approaches to abate seabird by‐catch in longline fisheries . Fish and Fisheries , 6 : 35 – 49 . Google Scholar Crossref Search ADS WorldCat Gilman E. , Passfield K., Nakamura K. 2014 . Performance of regional fisheries management organizations: ecosystem‐based governance of bycatch and discards . Fish and Fisheries , 15 : 327 – 351 . Google Scholar Crossref Search ADS WorldCat Gilman E. , Zollett E., Beverly S., Nakano H., Davis K., Shiode D., Dalzell P., et al. 2006 . Reducing sea turtle by‐catch in pelagic longline fisheries . Fish and Fisheries , 7 : 2 – 23 . Google Scholar Crossref Search ADS WorldCat Gjedrem T. 1983 . Genetic variation in quantitative traits and selective breeding in fish and shellfish . Aquaculture , 33 : 51 – 72 . Google Scholar Crossref Search ADS WorldCat Godfray H. C. J. , Beddington J. R., Crute I. R., Haddad L., Lawrence D., Muir J. F., Pretty J., et al. 2010 . Food security: the challenge of feeding 9 billion people . Science , 327 : 812 – 818 . Google Scholar Crossref Search ADS PubMed WorldCat Goethel D. R. , Kerr L. A., Cadrin S. X. 2012 . Incorporating Spatial Population Structure in Stock Assessment Models of Marine Species . Sea Grant College Program, Massachusetts Institute of Technology , Boston, MA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gordon H. S. 1954 . The economic theory of a common-property resource: the fishery . Journal of Political Economy , 62 : 124 – 142 . Google Scholar Crossref Search ADS WorldCat Hare S. R. , Mantua N. J. 2000 . Empirical evidence for North Pacific regime shifts in 1977 and 1989 . Progress in Oceanography , 47 : 103 – 146 . Google Scholar Crossref Search ADS WorldCat Harley S. , Davies N., Hampton J., McKechnie S. 2014 . Stock assessment of bigeye tuna in the western and central Pacific. WCPFC-SC10-2014/SA-WP-01 Rev1 25 July. Noumea, New Caledonia. 115 p. https://www.wcpfc.int/node/18975 (last accessed 20 August 2016). Harrison A. L. 2012 . A synthesis of marine predator migrations, distribution, species overlap, and use of Pacific Ocean Exclusive Economic Zones. Ph.D. dissertation, University of California at Santa Cruz. Heino M. , Godø O. R. 2002 . Fisheries-induced selection pressures in the context of sustainable fisheries . Bulletin of Marine Science , 70 : 639 – 656 . OpenURL Placeholder Text WorldCat Hernandez-Milian G. , Goetz S., Varela-Dopico C., Rodriguez-Gutierrez J., Romón-Olea J., Fuertes-Gamundi J. R., Ulloa-Alonso E., et al. 2008 . Results of a short study of interactions of cetaceans and longline fisheries in Atlantic waters: environmental correlates of catches and depredation events . Hydrobiologia , 612 : 251 – 268 . Google Scholar Crossref Search ADS WorldCat Hinke J. T. , Kaplan I. C., Aydin K., Watters G. M., Olson R. J., Kitchell J. F. 2004 . Visualizing the food-web effects of fishing for tunas in the Pacific Ocean . Ecology and Society , 9 : 10. Google Scholar Crossref Search ADS WorldCat Holmlund C. M. , Hammer M. 1999 . Ecosystem services generated by fish populations . Ecological Economics , 29 : 253 – 268 . Google Scholar Crossref Search ADS WorldCat Hsieh C. H. , Reiss C. S., Hunter J. R., Beddington J. R., May R. M., Sugihara G. 2006 . Fishing elevates variability in the abundance of exploited species . Nature , 443 : 859 – 862 . Google Scholar Crossref Search ADS PubMed WorldCat HSVAR . 2016 . The Food and Agriculture Organization of the United nations, High Seas Vessels Authorization Record (HSVAR). Version 2015-2016. http://www.fao.org/fishery/collection/hsvar/en (last accessed 9 August 2016). Hunt G. L. , McKinnell S. 2006 . Interplay between top-down, bottom-up, and wasp-waist control in marine ecosystems . Progress in Oceanography , 68 : 115 – 124 . Google Scholar Crossref Search ADS WorldCat Hutchings J. A. , Minto C., Ricard D., Baum J. K., Jensen O. P. 2010 . Trends in the abundance of marine fishes . Canadian Journal of Fisheries and Aquatic Sciences , 67 : 1205 – 1210 . Google Scholar Crossref Search ADS WorldCat ISC Pacific Bluefin Tuna Working Group . 2016 . Executive Summary of the 2016 Pacific Bluefin Tuna Stock Assessment. 16th Meeting of the ISC Plenary, July 2016 (ISC16), Japan. IUCN . 2015 . The IUCN Red List of Threatened Species. Version 2015-4. http://www.iucnredlist.org. (last accessed 6 April 2016) Jackson J. B. , Kirby M. X., Berger W. H., Bjorndal K. A., Botsford L. W., Bourque B. J., Bradbury R. H., et al. 2001 . Historical overfishing and the recent collapse of coastal ecosystems . Science , 293 : 629 – 637 . Google Scholar Crossref Search ADS PubMed WorldCat Jennings S. , Kaiser M. J. 1998 . The effects of fishing on marine ecosystems . Advances in Marine Biology , 34 : 201 – 352 . Google Scholar Crossref Search ADS WorldCat Jorgensen C. , Enberg K., Dunlop E. S., Arlinghaus R., Boukal D. S., Brander K., Ernande B., et al. 2007 . Ecology-managing evolving fish stocks . Science , 318 : 1247 – 1248 . Google Scholar Crossref Search ADS PubMed WorldCat Juan-Jordá M. J. , Mosqueira I., Cooper A. B., Freire J., Dulvy N. K. 2011 . Global population trajectories of tunas and their relatives . Proceedings of the National Academy of Sciences of the United States of America , 108 : 20650 – 20655 . Google Scholar Crossref Search ADS PubMed WorldCat Kelleher K. 2005 . Discards in the world’s marine fisheries: an update. FAO Fisheries Technical Paper No. 470. Rome. 131 http://www.fao.org/docrep/008/y5936e/y5936e00.htm (last accessed 20 August 2016). Kenny A. J. , Skjoldal H. R., Engelhard G. H., Kershaw P. J., Reid J. B. 2009 . An integrated approach for assessing the relative significance of human pressures and environmental forcing on the status of Large Marine Ecosystems . Progress in Oceanography , 81 : 132 – 148 . Google Scholar Crossref Search ADS WorldCat Kitchell J. F. , Essington T. E., Boggs C. H., Schindler D. E., Walters C. J. 2002 . The role of sharks and longline fisheries in a pelagic ecosystem of the central Pacific . Ecosystems , 5 : 202 – 216 . Google Scholar Crossref Search ADS WorldCat Kitchell J. F. , Martell S. J. D., Walters C. J., Jensen O. P., Kaplan I. C., Watters J., Essington T. E., et al. 2006 . Billfishes in an ecosystem context . Bulletin of Marine Science , 79 : 669 – 682 . OpenURL Placeholder Text WorldCat Klaer N. , Polacheck T. 1997 . By-catch of albatrosses and other seabirds by Japanese longline fishing vessels in the Australian Fishing Zone from April 1992 to March 1995 . Emu , 97 : 150 – 167 . Google Scholar Crossref Search ADS WorldCat Koslow J. A. , Boehlert G. W., Gordon J. D. M., Haedrich R. L., Lorance P., Parin N. 2000 . Continental slope and deep‐sea fisheries: implications for a fragile ecosystem . ICES Journal of Marine Science , 57 : 548 – 557 . Google Scholar Crossref Search ADS WorldCat Kuparinen A. , Merilä J. 2007 . Detecting and managing fisheries-induced evolution . Trends in Ecology & Evolution , 22 : 652 – 659 . Google Scholar Crossref Search ADS PubMed WorldCat Lam V. W. , Cheung W. W., Reygondeau G., Sumaila U. R. 2016 . Projected change in global fisheries revenues under climate change . Scientific Reports , 6 : 32607. OpenURL Placeholder Text WorldCat Larkin P. A. 1979 . Predator-prey relations in fishes: an overview of the theory. Predator-prey Systems in Fisheries Management . Sport Fishing Institute , Washington DC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Law R. , Grey D. R. 1989 . Evolution of yields from populations with age-specific cropping . Evolutionary Ecology , 3 : 343 – 359 . Google Scholar Crossref Search ADS WorldCat Law R. 2000 . Fishing, selection, and phenotypic evolution . ICES Journal of Marine Science , 57 : 659 – 668 . Google Scholar Crossref Search ADS WorldCat Lewison R. L. , Crowder L. B., Wallace B. P., Moore J. E., Cox T., Zydelis R., McDonald S., et al. 2014 . Global patterns of marine mammal, seabird, and sea turtle bycatch reveal taxa-specific and cumulative megafauna hotspots . Proceedings of the National Academy of Sciences of the United States of America , 111 : 5271 – 5276 . Google Scholar Crossref Search ADS PubMed WorldCat Lewison R. L. , Freeman S. A., Crowder L. B. 2004 . Quantifying the effects of fisheries on threatened species: the impact of pelagic longlines on loggerhead and leatherback sea turtles . Ecology Letters , 7 : 221 – 231 . Google Scholar Crossref Search ADS WorldCat Limpus C. J. , Limpus D. J. 2003 . The loggerhead turtle, Caretta caretta, in the Equatorial and Southern Pacific Ocean: a species in decline. In Loggerhead Sea Turtles, 199–209 . Ed. by Bolten A. B., Witherington B. E. S.. Smithsonian Institution Press , Washington, DC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Limpus C. J. , Walker T. A., West J. 1995 . Posthatchling specimens and records from the Australian region. In Proceedings for Marine Turtle Conservation Workshop. Ed. by James, R. [Compiler]. Australian National Parks and Wildlife Service, Canberra, 86 p. Lindegren M. , Möllmann C., Hansson L. A. 2010 . Biomanipulation: a tool in marine ecosystem management and restoration? Ecological Applications , 20 : 2237 – 2248 . Google Scholar Crossref Search ADS PubMed WorldCat Link J. S. 2002 . Does food web theory work for marine ecosystems? Marine Ecology Progress Series , 230 : 1 – 9 . Google Scholar Crossref Search ADS WorldCat Maguire J. J. 2006 . The State of World Highly Migratory, Straddling and Other High Seas Fishery Resources and Associated Species (No. 495). Food & Agriculture Organization, Rome. Marshall C. T. , Browman H. I. 2007 . Disentangling the causes of maturation trends in exploited fish populations . Marine Ecology Progress Series , 335 : 249 – 251 . Google Scholar Crossref Search ADS WorldCat Maunder M. N. , Sibert J. R., Fonteneau A., Hampton J., Kleiber P., Harley S. J. 2006 . Interpreting catch per unit effort data to assess the status of individual stocks and communities . ICES Journal of Marine Science , 63 : 1373 – 1385 . Google Scholar Crossref Search ADS WorldCat McCauley D. J. , Pinsky M. L., Palumbi S. R., Estes J. A., Joyce F. H., Warner R. R. 2015 . Marine defaunation: animal loss in the global ocean . Science , 347 : 1255641. Google Scholar Crossref Search ADS PubMed WorldCat Menge B. A. 2000 . Top-down and bottom-up community regulation in marine rocky intertidal habitats . Journal of Experimental Marine Biology and Ecology , 250 : 257 – 289 . Google Scholar Crossref Search ADS PubMed WorldCat Merrie A. , Dunn D. C., Metian M., Boustany A. M., Takei Y., Elferink A. O., Ota Y., et al. 2014 . An ocean of surprises–Trends in human use, unexpected dynamics and governance challenges in areas beyond national jurisdiction . Global Environmental Change 27 : 19 – 31 . Google Scholar Crossref Search ADS WorldCat Mills C. E. 1995 . Medusae, siphonophores and ctenophores as planktivorous predators in changing global ecosystems . ICES Journal of Marine Science 52 : 575 – 581 . Google Scholar Crossref Search ADS WorldCat Molina J. M. , Cooke S. J. 2012 . Trends in shark bycatch research: current status and research needs . Reviews in Fish Biology and Fisheries 22 : 719 – 737 . Google Scholar Crossref Search ADS WorldCat Möllmann C. , Diekmann R., Muller-Karulis B., Kornilovs G., Plikshs M., Axe P. 2009 . Reorganization of a large marine ecosystem due to atmospheric and anthropogenic pressure: a discontinuous regime shift in the Central Baltic Sea . Global Change Biology , 15 : 1377 – 1393 . Google Scholar Crossref Search ADS WorldCat Möllmann C. , Diekmann R. 2012 . Marine ecosystem regime shifts induced by climate and overfishing: a review for the northern hemisphere . Advances in Ecological Research , 47 : 303 . Google Scholar Crossref Search ADS WorldCat Mulder C. , Boit A., Mori S., Arie Vonk J., Dyer S. D., Faggiano L., Geisen S., et al. 2012 . Distributional (in) congruence of biodiversity-ecosystem functioning . Advances in Ecological Research , 46 : 1. Google Scholar Crossref Search ADS WorldCat Murawski S. , Methot R., Tromble G. 2007 . Biodiversity loss in the ocean: how bad is it? Science , 316 : 1281 – 1284 . Google Scholar Crossref Search ADS PubMed WorldCat Myers R. A. , Worm B. 2003 . Rapid worldwide depletion of predatory fish communities . Nature , 423 : 281 – 283 . Google Scholar Crossref Search ADS WorldCat Neilson J. D. , Loefer J., Prince E. D., Royer F., Calmettes B., Gaspar P., Lopez R., et al. 2014 . Seasonal distributions and migrations of northwest Atlantic Swordfish: inferences from integration of pop-up satellite archival tagging studies . PLoS One , 9 : e112736. Google Scholar Crossref Search ADS PubMed WorldCat Nicol S. , Croaxall J., Tratahn P., Gales N., Murphy E. 2007 . Paradigm misplaced? Antarctic marine ecosystems are affected by climate change as well as biological processes and harvesting . Antarctic Science , 19 : 291 – 295 . Google Scholar Crossref Search ADS WorldCat Oguz T. , Gilbert D. 2007 . Abrupt transitions of the top-down controlled Black Sea pelagic ecosystem during 1960–2000: evidence for regime-shifts under strong fishery exploitation and nutrient enrichment modulated by climate-induced variations . Deep Sea Research Part I: Oceanographic Research Papers , 54 : 220 – 242 . Google Scholar Crossref Search ADS WorldCat Oliver S. , Braccini M., Newman S. J., Harvey E. S. 2015 . Global patterns in the bycatch of sharks and rays . Marine Policy , 54 : 86 – 97 . Google Scholar Crossref Search ADS WorldCat Olsen E. M. , Heino M., Lilly G. R., Morgan M. J., Brattey J., Ernande B., Dieckmann U. 2004 . Maturation trends indicative of rapid evolution preceded the collapse of northern cod . Nature , 428 : 932 – 935 . Google Scholar Crossref Search ADS PubMed WorldCat Olson R. J. , Watters G. M. 2003 . A model of the pelagic ecosystem in the eastern tropical Pacific Ocean . Inter-American Tropical Tuna Commission Bulletin , 22 : 135 – 218 . OpenURL Placeholder Text WorldCat Ottersen G. , Hjermann D. Ø., Stenseth N. C. 2006 . Changes in spawning stock structure strengthen the link between climate and recruitment in a heavily fished cod (Gadus morhua) stock . Fisheries Oceanography , 15 : 230 – 243 . Google Scholar Crossref Search ADS WorldCat Pace M. L. , Cole J. G., Carpenter S. R., Kitchell J. F. 1999 . Trophic cascades revealed in diverse ecosystems . Trends in Ecology & Evolution , 14 : 483 – 488 . Google Scholar Crossref Search ADS PubMed WorldCat Paine R. T. 1980 . Food webs: linkage, interaction strength and community infrastructure . Journal of Animal Ecology , 49 : 667 – 685 . Google Scholar Crossref Search ADS WorldCat Pauly D. , Christensen V., Dalsgaard J., Froese R., Torres F. J. Jr. 1998 . Fishing down marine food webs . Science , 279 : 860 – 863 . Google Scholar Crossref Search ADS PubMed WorldCat Pauly D. , Zeller D. (Eds.), 2015 . Sea Around Us Concepts, Design and Data (seaaroundus.org) . University of British Columbia . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Pauly D. , Zeller D. 2016 . Catch reconstructions reveal that global marine fisheries catches are higher than reported and declining . Nature Communications , 7 : OpenURL Placeholder Text WorldCat Pauly D. , Watson R., Alder J. 2005 . Global trends in world fisheries: impacts on marine ecosystems and food security . Philosophical Transactions of the Royal Society B: Biological Sciences , 360 : 5 – 12 . Google Scholar Crossref Search ADS WorldCat Peacor S. D. , Werner E. E. 2008 . Nonconsumptive effects of predators and trait‐mediated indirect effects. eLS. doi: 10.1002/9780470015902.a0021216 Peckarsky B. L. , Abrams P. A., Bolnick D. I., Dill L. M., Grabowski J. H., Luttbeg B., Orrock J. L., et al. 2008 . Revisiting the classics: considering nonconsumptive effects in textbook examples of predator-prey interactions . Ecology , 89 : 2416 – 2425 . Google Scholar Crossref Search ADS PubMed WorldCat Pepperell J. G. , Harvey G. 2010 . Fishes of the Open-Ocean . University of Chicago Press , Chicago, IL . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Pinnegar J. K. , Polunin N. V. C., Francour P., Badalamenti F., Chemello R., Harmelin-Vivien M. L., Hereu B., et al. 2000 . Trophic cascades in benthic marine ecosystems: lessons for fisheries and protected-area management . Environmental Conservation , 27 : 179 – 200 . Google Scholar Crossref Search ADS WorldCat Polovina J. J. , Woodworth-Jefcoats P. A. 2013 . Fishery-induced changes in the subtropical Pacific pelagic ecosystem size structure: observations and theory . PLoS One , 8 : e62341. Google Scholar Crossref Search ADS PubMed WorldCat Postel S. , Bawa K., Kaufman L., Peterson C. H., Carpenter S., Tillman D., Dayton P., et al. 2012 . Nature's Services: Societal Dependence on Natural Ecosystems . Island Press , Washington, DC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Queiroz N. , Humphries N. E., Mucientes G., Hammerschlag N., Lima F. P., Scales K. L., Miller P. I., et al. 2016 . Ocean-wide tracking of pelagic sharks reveals extent of overlap with longline fishing hotspots . Proceedings of the National Academy of Sciences of the United States of America , 113 : 1582 – 1587 . Google Scholar Crossref Search ADS PubMed WorldCat Roberts C. 2002 . Deep impact: the rising toll of fishing in the deep sea . Trends in Ecology and Evolution , 17 : 242 – 245 . Google Scholar Crossref Search ADS WorldCat Robertson G. , Gales R. 1998 . Albatross Biology and Conservation . Surrey Beatty & Sons, Chipping Norton . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Rochet M. J. 1998 . Short-term effects of fishing on life history traits of fishes . ICES Journal of Marine Science , 55 : 371 – 391 . Google Scholar Crossref Search ADS WorldCat Rochet M. J. , Benoît E. 2011 . Fishing destabilizes the biomass flow in the marine size spectrum. Proceedings of the Royal Society of London B: Biological Sciences, 279: 284–292. Ryman N. , Utter F., Laikre L. 1995 . Protection of intraspecific biodiversity of exploited fishes . Reviews in Fish Biology and Fisheries , 5 : 417 – 446 . Google Scholar Crossref Search ADS WorldCat Scheffer M. , Carpenter S. R. 2003 . Catastrophic regime shifts in ecosystems: linking theory to observation . Trends in Ecology & Evolution , 18 : 648 – 656 . Google Scholar Crossref Search ADS WorldCat Scheffer M. , Carpenter S., Foley J. A., Folke C., Walker B. 2001 . Catastrophic shifts in ecosystems . Nature , 413 : 591 – 596 . Google Scholar Crossref Search ADS PubMed WorldCat Sharpe D. M. , Hendry A. P. 2009 . SYNTHESIS: life history change in commercially exploited fish stocks: an analysis of trends across studies . Evolutionary Applications , 2 : 260 – 275 . Google Scholar Crossref Search ADS PubMed WorldCat Shiomoto A. , Tadokoro K., Nagasawa K., Ishida Y. 1997 . Trophic relations in the subarctic North Pacific ecosystem: possible feeding effect from pink salmon . Marine Ecology Progress Series , 150 : 75 – 85 . Google Scholar Crossref Search ADS WorldCat Sibert J. , Hampton J., Kleiber P., Maunder M. 2006 . Biomass, size, and trophic status of top predators in the Pacific Ocean . Science , 314 : 1773 – 1776 . Google Scholar Crossref Search ADS PubMed WorldCat Skomal G. B. , Mandelman J. W. 2012 . The physiological response to anthropogenic stressors in marine elasmobranch fishes: a review with a focus on the secondary response . Comparative Biochemistry and Physiology A , 162 : 146 – 155 . Google Scholar Crossref Search ADS WorldCat Smout S. , Asseburg C., Matthiopoulos J., Fernández C., Redpath S., Thirgood S., Harwood J. 2010 . The functional response of a generalist predator . PLoS One , 5 : e10761. Google Scholar Crossref Search ADS PubMed WorldCat Soule M., E. , Wilcox B. A. 1980 . Conservation biology. An evolutionary-ecological perspective . Sinauer Associates, Inc , Sunderland, MA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Spotila J. R. , Reina R. D., Steyermark A. C., Plotkin P. T., Paladino F. V. 2000 . Pacific leatherback turtles face extinction . Nature , 405 : 529 – 530 . Google Scholar Crossref Search ADS PubMed WorldCat Steele J. H. 1998 . From carbon flux to regime shift . Fisheries Oceanography , 7 : 176 – 181 . Google Scholar Crossref Search ADS WorldCat Steele J. H. 2004 . Regime shifts in the ocean: reconciling observations and theory . Progress in Oceanography , 60 : 135 – 141 . Google Scholar Crossref Search ADS WorldCat Steele M. , Forrester G., Almany G. 1998 . Influences of predators and conspecifics on recruitment of a tropical and a temperate reef fish . Marine Ecology Progress Series , 172 : 115 – 125 . Google Scholar Crossref Search ADS WorldCat Steneck R. S. , Sala E. A. 2005 . Large marine carnivores: trophic cascades and top-down controls in coastal ecosystems past and present. Large Carnivores and the Conservation of Biodiversity , pp. 110 – 137 . Ed. by Ray J., Redford K., Steneck R., Berger J. Island Press , Washington, DC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Stokes K. , Law R. 2000 . Fishing as an evolutionary force . Marine Ecology-Progress Series , 208 : 307 – 309 . OpenURL Placeholder Text WorldCat Strong D. , Frank K. 2010 . Human involvement in food webs . Annual Review of Environment and Resources , 35: 1 – 23 . Google Scholar Crossref Search ADS WorldCat Suding K. , Gross K., Houseman G. 2004 . Alternative states and positive feedbacks in restoration ecology . Trends in Ecology & Evolution , 19 : 46 – 53 . Google Scholar Crossref Search ADS PubMed WorldCat Sumaila U. R. , Lam V. W., Miller D. D., Teh L., Watson R. A., Zeller D., Cheung W. W., et al. 2015 . Winners and losers in a world where the high seas is closed to fishing . Scientific Reports , 5 : 8481. Google Scholar Crossref Search ADS PubMed WorldCat Swartz W. , Sala E., Tracey S., Watson R., Pauly D. 2010 . The spatial expansion and ecological footprint of fisheries (1950 to present) . PLoS One , 5 : e15143. Google Scholar Crossref Search ADS PubMed WorldCat Tasker M. L. , Becker P. H. 1992 . Influences of human activities on seabird populations in the North Sea . Netherlands Journal of Aquatic Ecology , 26 : 59 – 73 . Google Scholar Crossref Search ADS WorldCat Taylor N. G. , McAllister M. K., Lawson G. L., Carruthers T., Block B. A. 2011 . Atlantic bluefin tuna: a novel multistock spatial model for assessing population biomass . PLoS One , 6 : e27693. Google Scholar Crossref Search ADS PubMed WorldCat Tuck G. N. , Polacheck T., Croxall J. P., Weimerskirch H., Tuck G. N., Polacheck T. O. M., Croxall J. P. 2001 . Modelling the impact of fishery by-catches on albatross populations . Journal of Applied Ecology , 38 : 1182 – 1196 . Google Scholar Crossref Search ADS WorldCat UNCLOS. 1982 . United Nations Convention on the Law of the Sea, opened for signature 10 December 1982. 1833 UNTS397 (entered into force 10 November 1994). UNFSA . 1995 . United Nations Conference on Straddling Fish Stocks and Highly Migratory Fish Stocks, July 24–August 4, 1995, Agreement for the Implementation of the Provisions of the United Nations Convention on the Law of the Sea of 10 December Relating to the Conservation and Management of Straddling Fish Stocks and Highly Migratory Fish Stocks, U.N. DOCA/Conf. 164/37. Valls A. , Coll M., Christensen V. 2015 . Keystone species: toward an operational concept for marine biodiversity conservation . Ecological Monographs , 85 : 29 – 47 . Google Scholar Crossref Search ADS WorldCat Verity P. G. , Smetacek V. 1996 . Organism life cycles, predation, and the structure of marine pelagic ecosystems . Marine Ecology Progress Series , 130 : 277 – 293 . Google Scholar Crossref Search ADS WorldCat Walker T. A. , Parmenter C. J. 1990 . Absence of a pelagic phase in the life cycle of the flatback turtle, Natator depressa (Garman) . Journal of Biogeography , 275 – 278 . OpenURL Placeholder Text WorldCat Wallace B. P. , Lewison R. L., McDonald S. L., McDonald R. K., Kot C. Y., Kelez S., Bjorkland R. K., et al. 2010 . Global patterns of marine turtle bycatch . Conservation Letters , 3 : 131 – 142 . Google Scholar Crossref Search ADS WorldCat Walsh M. R. , Munch S. B., Chiba S., Conover D. O. 2006 . Maladaptive changes in multiple traits caused by fishing: impediments to population recovery . Ecology Letters , 9 : 142 – 148 . Google Scholar Crossref Search ADS PubMed WorldCat Walters C. 2003 . Folly and fantasy in the analysis of spatial catch rate data . Canadian Journal of Fisheries and Aquatic Sciences , 60 : 1433 – 1436 . Google Scholar Crossref Search ADS WorldCat Walters C. J. , Kitchell J. F. 2001 . Cultivation/depensation effects on juvenile survival and recruitment: implications for the theory of fishing . Canadian Journal of Fisheries and Aquatic Sciences , 58 : 39 – 50 . Google Scholar Crossref Search ADS WorldCat Ward P. , Myers R. A. 2005 . Shifts in open-ocean fish communities coinciding with the commencement of commercial fishing . Ecology , 86 : 835 – 847 . Google Scholar Crossref Search ADS WorldCat Ware D. M. , Thomson R. E. 2005 . Bottom-up ecosystem trophic dynamics determine fish production in the northeast Pacific . Science , 308 : 1280 – 1284 . Google Scholar Crossref Search ADS PubMed WorldCat Webb T. J. , Berghe E., V., O'Dor R. 2010 . Biodiversity's big wet secret: the global distribution of marine biological records reveals chronic under-exploration of the deep pelagic ocean . PLoS One , 5 : e10223. Google Scholar Crossref Search ADS PubMed WorldCat Woodward G. , Ebenman B., Emmerson M., Montoya J. M., Olesen J. M., Valido A., Warren P. H. 2005 . Body size in ecological networks . Trends in Ecology & Evolution , 20 : 402 – 409 . Google Scholar Crossref Search ADS PubMed WorldCat Wooster W. S. , Zhang C. I. 2004 . Regime shifts in the North Pacific: early indications of the 1976–1977 event . Progress in Oceanography , 60 : 183 – 200 . Google Scholar Crossref Search ADS WorldCat Worm B. , Barbier E. B., Beaumont N., Duffy J. E., Folke C., Halpern B. S., Jackson J., et al. 2006 . Impacts of biodiversity loss on ocean ecosystem services . Science , 314 : 787 – 790 . Google Scholar Crossref Search ADS PubMed WorldCat Worm B. , Myers R. A. 2003 . Meta-analysis of cod-shrimp interactions reveals top-down control in oceanic food webs . Ecology , 84 : 162 – 173 . Google Scholar Crossref Search ADS WorldCat Worm B. , Sandow M., Oschlies A., Lotze H. K., Myers R. A. 2005 . Global patterns of predator diversity in the open-oceans . Science , 309 : 1365 – 1369 . Google Scholar Crossref Search ADS PubMed WorldCat Worm B. , Tittensor D. P. 2011 . Range contraction in large pelagic predators . Proceedings of the National Academy of Science of the United States of America , 108 : 11942 – 11947 . Google Scholar Crossref Search ADS WorldCat Zug G. R. , Balazs G. H., Wetherall J. A. 1995 . Growth in juvenile loggerhead sea turtles (Caretta caretta) in the north Pacific pelagic habitat . Copeia , 1995 : 484 – 487 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Melbourne-Thomas, Jessica; Constable, Andrew J; Fulton, Elizabeth A; Corney, Stuart P; Trebilco, Rowan; Hobday, Alistair J; Blanchard, Julia L; Boschetti, Fabio; Bustamante, Rodrigo H; Cropp, Roger; Everett, Jason D; Fleming, Aysha; Galton-Fenzi, Ben; Goldsworthy, Simon D;

Turner, Joseph A; Babcock, Russell C; Hovey, Renae; Kendrick, Gary A

doi: 10.1093/icesjms/fsx085pmid: N/A

Abstract Mesophotic coral ecosystems (MCEs) occur at depths beyond those typically associated with coral reefs. Significant logistical challenges associated with data collection in deep water have resulted in a limited understanding of the ecological relevance of these deeper coral ecosystems. We review the trends in this research, covering the geographic spread of MCE research, the focus of these studies, the methods used, how MCEs differ in terms of species diversity and begin to assess connectivity of coral populations. Clear locational biases were observed, with studies concentrated in a few discrete areas mainly around the Atlantic region. The focus of MCE studies has diversified in recent years and more detailed aspects of MCE ecology are now being investigated in particular areas of research. Advances in technology are also reflected in the current range of research, with a wider variety of methods now employed. However, large information gaps are present in entire regions and particularly in relation to the threats, impacts and subsequent management of MCEs. Analysis of species diversity shows that initial definitions based on depth alone may not be appropriate globally, while further taxonomic resolution may also be required to deduce the full biodiversity of major groups in certain regions. Genetic studies to date show species-specific results, although distinct deeper populations do appear to exist, which raises questions regarding the potential of MCEs to act as refugia. Introduction Coral reefs are in worldwide decline, due to increased mass disturbance events brought about by climate change and anthropogenic activities (Hughes et al., 2003; Bellwood et al., 2004; Hoegh-Guldberg et al., 2007). However, a majority of the data on which these projections are based are from coral reefs shallower than 20–30 m, while the trends below this depth remain unknown (Bak et al., 2005; Bridge et al., 2013). Deeper mesophotic coral ecosystems (MCEs) are defined as tropical and sub-tropical light-dependent communities occurring from approximately 30 m to the lower limit of the photic zone, extending as deep as 150 m in some locations (Hinderstein et al., 2010). These reefs are perceived as continuations of the shallow reef communities, with a similarly diverse range of taxa (Lesser et al., 2009; Hinderstein et al., 2010). Communities are primarily structured by light (Sheppard, 1982; Lesser et al., 2009) although there are other factors at play, including topography (Bridge et al., 2010), temperature (Kahng et al., 2012), sedimentation, and water movement (Goreau and Goreau, 1973; Sheppard, 1982). Mesophotic communities have been broadly described (Busby, 1966; Goreau and Goreau, 1973; Bouchon, 1981; Sheppard, 1982; Fricke and Meischner, 1985; Colin et al., 1986; Fricke and Knauer, 1986; Thresher and Colin, 1986) but remain relatively unexplored compared to shallow water reefs, particularly in relation to ecological characteristics and functions. This is primarily due to their location, lying beyond recreational SCUBA diving limits, and therefore posing increased logistical challenges (Lesser et al., 2009; Kahng et al., 2010). Mesophotic reefs are starting to gain more attention as modern technological advances make them increasingly accessible (Lesser et al., 2009; Kahng et al., 2010). Advances in habitat mapping and technologies such as Remotely Operated underwater Vehicles (ROVs) and Autonomous Underwater Vehicles (AUVs) can provide a useful platform for monitoring these systems (Singh et al., 2004; Armstrong et al., 2006; Bridge et al., 2011a). Increased interest in MCEs is evident in the exponential increase in publications following recent workshops and special journal theme sections (Loya et al., 2016). During the past decade in particular, knowledge of these systems has moved on significantly. MCEs have only been studied in a few areas of the world resulting in little generalizable knowledge of the drivers of their structure, function, connectivity and refugia role for shallow reefs globally. There is a poor understanding of the role environmental factors have in influencing spatial patterns in community structure, and therefore how MCEs respond to anthropogenic threats and climate change (Puglise et al., 2009; Kahng et al., 2014). MCEs can harbour diverse biological assemblages of corals, fish and other invertebrates consisting of a range of “deep-specialist” and “depth-generalist” species (Bongaerts et al., 2010a; Kahng et al., 2014). Some species are endemic to these systems, highlighting the importance of MCEs in contributing to and maintaining global biodiversity (Heyward et al., 2010; Bridge et al., 2011b; Kane et al., 2014; Muir et al., 2015). As more studies are completed, the limitations to our current knowledge have become evident. For example, studies investigating mesophotic areas of the Great Barrier Reef identified that submerged reef habitat may have been underestimated by as much as 100% (Harris et al., 2012) and new species records for Australia have also been found (Muir et al., 2015). It has been suggested that MCEs function as refugia, where communities are sheltered from perturbations in shallow waters including high temperature, sedimentation, storm damage and fishing and so may re-seed more frequently disturbed shallow reefs (Bongaerts et al., 2010a; Hinderstein et al., 2010). Larval connectivity needs to be understood, in order to assess the extent of re-seeding potential, including whether species are present in deep and shallow water and how they are connected through the movement of currents (Lesser et al., 2009; Slattery et al., 2011; Baker et al., 2016). A zone that harbours both shallow reef and mesophotic species appears to occur in a number of the locations studied, generally between 30 and 60 m (Lesser et al., 2009; Slattery et al., 2011), often termed the upper-mesophotic. However, in deeper areas, high levels of depth-endemism would suggest greater habitat specialization, and possibly limited larval exchange with shallower waters and a reduced ability to replenish shallow habitats (Slattery et al., 2011). If the community structures between deep mesophotic and shallow coral reefs are different then re-seeding will not be possible. Information on the distribution and extent of MCEs, the factors that determine their distributions, and the organisms found in these ecosystems, are all critical to inform biodiversity management (Puglise et al., 2009; Baker et al., 2016). The spatial distributions of rare and ecologically important habitats are required to adequately design networks of Marine Protected Areas and ensure representation of all habitat types (Bridge et al., 2016a). Mesophotic reefs are likely to provide similar ecosystem services to those of shallow water reefs and can contribute to fisheries, tourism, and pharmaceutical uses (Eyal et al., 2015; Baker et al., 2016). Identifying the key ecosystem services provided by these systems is important in order to gain support for their protection (Puglise et al., 2009). Adopting a broad, ecosystem-wide approach that encompasses deep reefs is most likely to have many environmental, social and economic benefits (Bridge et al., 2013). This review investigates the current literature regarding MCEs. The term “deep coral reef” is often used to refer to much deeper water ecosystems of aphotic species associated with colder water, e.g. Lophelia pertusa, defined as living without light (Freiwald et al., 2004). While these ecosystems may occasionally occur in mesophotic depth ranges at high latitudes, such as in Norwegian fjords, or at similar latitudes but at much greater depths (Roberts et al., 2006), they function differently from shallower coral ecosystems; with the term cold-water corals coined to differentiate them from tropical coral reefs (Freiwald and Roberts, 2005). For this review, MCEs will be defined as in Hinderstein et al. (2010) as light-dependant coral-dominated systems in tropical regions that form extensions of shallow coral reefs. The aims are to: (1) Characterize study locations and global hotspots of MCE research, (2) Identify trends in MCE research topics, (3) Identify the methods used, including how they have changed over time, (4) Describe how mesophotic biodiversity may differ between locations, and (5) Describe connectivity trends across shallow reefs to mesophotic depths. Assessing the work done so far will allow us to identify and characterize the key aspects of MCEs as well as identifying the key gaps in our understanding to inform future research direction. Methods A literature review was carried out following the systematic methods outlined in Pickering and Byrne (2014) and Pickering et al. (2015). The databases Google Scholar, Web of Science, and Scopus in May 2016 and February 2017 were searched using the search terms: mesophotic AND reef OR coral OR fish OR sponge OR connectivity OR ecolog* OR community OR recruit* OR impact OR disturbance The specialist database at mesophotic.org (http://www.mesophotic.org/publications/), maintained by field experts, was also utilized, and all papers were screened for content. As “mesophotic” is a relatively new term to be applied to reef ecosystems, defined in Puglise et al. (2009), we further checked references from recent review articles (Lesser et al., 2009; Kahng et al., 2010; Kahng et al., 2014; Baker et al., 2016; Loya et al., 2016) to ensure all relevant papers were acquired. Still, the search was conservative and some papers that did not use the search terms we utilized would not have been identified. Results were limited to those with an English title and abstract. Studies were screened to ensure relevance in a two-step process outlined below, and results are shown in Figure 1. Titles and abstracts were required to mention or contain information on: mesophotic or deep/twilight reef, tropical habitats, and coral reef ecology. Following screening, the full texts of relevant articles were obtained and reviewed. Studies were excluded if the main aims did not concentrate on mesophotic depths and communities (e.g. “deep” areas can relate to less than 10 m in some studies and so would not meet the aims of this review). A study had to encompass a proportion of the 30–150 m depth band defined for mesophotic areas in order to be considered relevant. In order to gather data to meet Aims 1–3 the following information was recorded for all papers: Authors and Title Geographic location (including coordinates), split into region [based on those used by Burke et al. (2011)], country and study area Year Primary research focus (Table 1) Table 1. Primary research focus categories and descriptions. Category . Description . Anthropogenic impact Study focuses on identifying the effects of a specific anthropogenic impact (e.g. dredge disposal, fishing) on the ecosystem Descriptive A characterizing study, identifying the communities present Ecosystem function Study focuses on specific aspects of the ecosystem or the biology of a particular group/taxa Geomorphology Study focuses on physical structural features relating to the underlying geology Life History Study focuses on life history parameters, such as reproduction and growth characteristics Management Study focuses on the ecosystem from a management perspective Methods Study focuses on comparing two or more methods Molecular ecology Investigation of macromolecules, specifically including genetic studies Natural impacts Study focuses on identifying the effects of a specific natural impact (e.g. bleaching, storms) on the ecosystem Review Study is a review paper Structuring variables Study specifically investigates abiotic or biotic variables that structure the community along a gradient Taxonomy Study is specifically focussed on the identification of (new) species Category . Description . Anthropogenic impact Study focuses on identifying the effects of a specific anthropogenic impact (e.g. dredge disposal, fishing) on the ecosystem Descriptive A characterizing study, identifying the communities present Ecosystem function Study focuses on specific aspects of the ecosystem or the biology of a particular group/taxa Geomorphology Study focuses on physical structural features relating to the underlying geology Life History Study focuses on life history parameters, such as reproduction and growth characteristics Management Study focuses on the ecosystem from a management perspective Methods Study focuses on comparing two or more methods Molecular ecology Investigation of macromolecules, specifically including genetic studies Natural impacts Study focuses on identifying the effects of a specific natural impact (e.g. bleaching, storms) on the ecosystem Review Study is a review paper Structuring variables Study specifically investigates abiotic or biotic variables that structure the community along a gradient Taxonomy Study is specifically focussed on the identification of (new) species Open in new tab Table 1. Primary research focus categories and descriptions. Category . Description . Anthropogenic impact Study focuses on identifying the effects of a specific anthropogenic impact (e.g. dredge disposal, fishing) on the ecosystem Descriptive A characterizing study, identifying the communities present Ecosystem function Study focuses on specific aspects of the ecosystem or the biology of a particular group/taxa Geomorphology Study focuses on physical structural features relating to the underlying geology Life History Study focuses on life history parameters, such as reproduction and growth characteristics Management Study focuses on the ecosystem from a management perspective Methods Study focuses on comparing two or more methods Molecular ecology Investigation of macromolecules, specifically including genetic studies Natural impacts Study focuses on identifying the effects of a specific natural impact (e.g. bleaching, storms) on the ecosystem Review Study is a review paper Structuring variables Study specifically investigates abiotic or biotic variables that structure the community along a gradient Taxonomy Study is specifically focussed on the identification of (new) species Category . Description . Anthropogenic impact Study focuses on identifying the effects of a specific anthropogenic impact (e.g. dredge disposal, fishing) on the ecosystem Descriptive A characterizing study, identifying the communities present Ecosystem function Study focuses on specific aspects of the ecosystem or the biology of a particular group/taxa Geomorphology Study focuses on physical structural features relating to the underlying geology Life History Study focuses on life history parameters, such as reproduction and growth characteristics Management Study focuses on the ecosystem from a management perspective Methods Study focuses on comparing two or more methods Molecular ecology Investigation of macromolecules, specifically including genetic studies Natural impacts Study focuses on identifying the effects of a specific natural impact (e.g. bleaching, storms) on the ecosystem Review Study is a review paper Structuring variables Study specifically investigates abiotic or biotic variables that structure the community along a gradient Taxonomy Study is specifically focussed on the identification of (new) species Open in new tab Methods used for data collection Depth range investigated Figure 1. Open in new tabDownload slide Summary of numbers of papers included/excluded in the process. Figure 1. Open in new tabDownload slide Summary of numbers of papers included/excluded in the process. Only a subset of papers provided information to investigate Aims 4 and 5. Mesophotic species diversity, deepest records, or the depth at which significant changes in community structure occur was included in the database when available. Any study that concentrated on genetic differences was investigated for Aim 5. This is currently the most effective method to quantify connectivity between deep and shallow areas: we extracted information on species, whether there was a genetic change with depth and if so what depth the changes occurred. Data manipulation and analysis was conducted in R (R Core Team, 2010) and figures were constructed using the ggplot2 package (Wickham, 2009). Aims 1–3 involved summarizing the information by location (map produced in ArcGIS 10.4), research focus and method. To address Aim 4 summary statistics were calculated for species diversity and transition depths between regions. Due to the relatively few data points, a rigorous statistical analysis was not possible for Aims 4 and 5. Results A total of 349 papers were classified in this study, spanning from 1966 to 2017. A majority of the studies on mesophotic reefs have been completed since 2010 (56%) (Figure 2), with 54 studies (15%) completed in 2016 alone. Research is concentrated in specific regions and countries (Figure 3) with over half (57%) of global mesophotic studies having been carried out in the Atlantic region, particularly in the Caribbean. Figure 2. Open in new tabDownload slide Frequency of publications focussing on MCEs. Figure 2. Open in new tabDownload slide Frequency of publications focussing on MCEs. Figure 3. Open in new tabDownload slide Global MCE research distribution (number of studies per country). Figure 3. Open in new tabDownload slide Global MCE research distribution (number of studies per country). Research into mesophotic reefs is globally very regionally localized. For example, while research spans a number of countries in the Atlantic (Table 2) effort is disproportionally split across them. Additionally, studies can concentrate in specific countries; with Israel (Middle East) and Hawaii, USA (Pacific) contributing to 91 and 71% of the entire studies for that region respectively. A single country, the USA has the greatest number of studies (18%, split over two regions) although they are focussed in geographically small areas with almost all studies occurring in Hawaii (Pacific) and Florida (Atlantic). Australia (13% of global studies) has observed significant modern research interest with 70% of Australian studies occurring since 2010. Again, studies are localized with 50% occurring on the Great Barrier Reef. The Indian Ocean and Southeast Asia are significantly under-represented (1 and 2% of global studies, respectively). Table 2. Studies completed on mesophotic reefs by region and country. Region . Country . Number of studies . Atlantic Bahamas 26 7.2% Barbados 1 0.3% Belize 1 0.3% Bermuda 8 2.2% Bonaire 5 1.4% Brazil 19 5.3% Cayman Islands 7 1.9% Curacao 26 7.2% Guinea 1 0.3% Honduras 3 0.8% Jamaica 11 3.1% Mexico 2 0.6% Panama 2 0.6% Puerto Rico 36 10.0% US Virgin Islands 33 9.2% USA 19 5.3% Australia Australia 50 13.9% Indian Ocean Chagos 2 0.6% Réunion 1 0.3% Middle East Egypt 1 0.3% Israel 30 8.3% Saudi Arabia 1 0.3% Sudan 1 0.3% Pacific Cook Islands 1 0.3% French Polynesia 3 0.8% Marshall Islands 6 1.7% Micronesia 6 1.7% Panama 1 0.3% Samoa 2 0.6% USA 47 13.1% Southeast Asia Brunei 1 0.3% Japan 4 1.1% Papua New Guinea 1 0.3% Philippines 1 0.3% Taiwan 1 0.3% Region . Country . Number of studies . Atlantic Bahamas 26 7.2% Barbados 1 0.3% Belize 1 0.3% Bermuda 8 2.2% Bonaire 5 1.4% Brazil 19 5.3% Cayman Islands 7 1.9% Curacao 26 7.2% Guinea 1 0.3% Honduras 3 0.8% Jamaica 11 3.1% Mexico 2 0.6% Panama 2 0.6% Puerto Rico 36 10.0% US Virgin Islands 33 9.2% USA 19 5.3% Australia Australia 50 13.9% Indian Ocean Chagos 2 0.6% Réunion 1 0.3% Middle East Egypt 1 0.3% Israel 30 8.3% Saudi Arabia 1 0.3% Sudan 1 0.3% Pacific Cook Islands 1 0.3% French Polynesia 3 0.8% Marshall Islands 6 1.7% Micronesia 6 1.7% Panama 1 0.3% Samoa 2 0.6% USA 47 13.1% Southeast Asia Brunei 1 0.3% Japan 4 1.1% Papua New Guinea 1 0.3% Philippines 1 0.3% Taiwan 1 0.3% Open in new tab Table 2. Studies completed on mesophotic reefs by region and country. Region . Country . Number of studies . Atlantic Bahamas 26 7.2% Barbados 1 0.3% Belize 1 0.3% Bermuda 8 2.2% Bonaire 5 1.4% Brazil 19 5.3% Cayman Islands 7 1.9% Curacao 26 7.2% Guinea 1 0.3% Honduras 3 0.8% Jamaica 11 3.1% Mexico 2 0.6% Panama 2 0.6% Puerto Rico 36 10.0% US Virgin Islands 33 9.2% USA 19 5.3% Australia Australia 50 13.9% Indian Ocean Chagos 2 0.6% Réunion 1 0.3% Middle East Egypt 1 0.3% Israel 30 8.3% Saudi Arabia 1 0.3% Sudan 1 0.3% Pacific Cook Islands 1 0.3% French Polynesia 3 0.8% Marshall Islands 6 1.7% Micronesia 6 1.7% Panama 1 0.3% Samoa 2 0.6% USA 47 13.1% Southeast Asia Brunei 1 0.3% Japan 4 1.1% Papua New Guinea 1 0.3% Philippines 1 0.3% Taiwan 1 0.3% Region . Country . Number of studies . Atlantic Bahamas 26 7.2% Barbados 1 0.3% Belize 1 0.3% Bermuda 8 2.2% Bonaire 5 1.4% Brazil 19 5.3% Cayman Islands 7 1.9% Curacao 26 7.2% Guinea 1 0.3% Honduras 3 0.8% Jamaica 11 3.1% Mexico 2 0.6% Panama 2 0.6% Puerto Rico 36 10.0% US Virgin Islands 33 9.2% USA 19 5.3% Australia Australia 50 13.9% Indian Ocean Chagos 2 0.6% Réunion 1 0.3% Middle East Egypt 1 0.3% Israel 30 8.3% Saudi Arabia 1 0.3% Sudan 1 0.3% Pacific Cook Islands 1 0.3% French Polynesia 3 0.8% Marshall Islands 6 1.7% Micronesia 6 1.7% Panama 1 0.3% Samoa 2 0.6% USA 47 13.1% Southeast Asia Brunei 1 0.3% Japan 4 1.1% Papua New Guinea 1 0.3% Philippines 1 0.3% Taiwan 1 0.3% Open in new tab Descriptive studies dominate the MCE literature (Table 3). However, research focus has shifted over time (Figure 4). The proportion of descriptive studies drops from 92% before 1980 to 33% post-2010. Research focus has also diversified, with increases observed in all other research categories between 2001 and 2011 onwards. Studies investigating molecular ecology have seen the largest increase, from zero before 2000 to 7% then 16% in 2001–2010 and post-2010, respectively. These studies are mostly conducted in the Atlantic (47%, exclusively in the Caribbean), Australia (24%) and the Pacific (21%). Life history studies and research focussing on impacts are in their infancy, only occurring since 2000. This work is currently highly concentrated in the Atlantic region with 78% of life history and 59% of impact (including natural and anthropogenic) studies taking place there. Table 3. Primary research focus of studies on mesophotic coral ecosystems. Research focus . Number of studies . Descriptive 137 39.3% Molecular Ecology 38 10.9% Taxonomy 33 9.5% Structuring Variables 30 8.6% Review 23 6.6% Ecosystem Function 20 5.7% Geomorphology 18 5.2% Life History 18 5.2% Methods 11 3.2% Natural Impacts 10 2.9% Anthropogenic Impact 7 2.0% Management 4 1.1% Research focus . Number of studies . Descriptive 137 39.3% Molecular Ecology 38 10.9% Taxonomy 33 9.5% Structuring Variables 30 8.6% Review 23 6.6% Ecosystem Function 20 5.7% Geomorphology 18 5.2% Life History 18 5.2% Methods 11 3.2% Natural Impacts 10 2.9% Anthropogenic Impact 7 2.0% Management 4 1.1% Open in new tab Table 3. Primary research focus of studies on mesophotic coral ecosystems. Research focus . Number of studies . Descriptive 137 39.3% Molecular Ecology 38 10.9% Taxonomy 33 9.5% Structuring Variables 30 8.6% Review 23 6.6% Ecosystem Function 20 5.7% Geomorphology 18 5.2% Life History 18 5.2% Methods 11 3.2% Natural Impacts 10 2.9% Anthropogenic Impact 7 2.0% Management 4 1.1% Research focus . Number of studies . Descriptive 137 39.3% Molecular Ecology 38 10.9% Taxonomy 33 9.5% Structuring Variables 30 8.6% Review 23 6.6% Ecosystem Function 20 5.7% Geomorphology 18 5.2% Life History 18 5.2% Methods 11 3.2% Natural Impacts 10 2.9% Anthropogenic Impact 7 2.0% Management 4 1.1% Open in new tab Figure 4. Open in new tabDownload slide Research focus of mesophotic coral ecosystem studies over time. Figure 4. Open in new tabDownload slide Research focus of mesophotic coral ecosystem studies over time. A variety of methods are used to study MCEs (Figure 5). Although lying beyond recreational diving depths, SCUBA diving is the most common method used for most year categories (Second to Submersibles in the 1980s and 1990s). Even prior to 1980, 69% of the studies were completed using SCUBA-based observations, before advances in technical diving such as closed-circuit rebreathers (CCRs), and when health and safety regulations were less conservative. Methods have diversified widely since 2001 as more techniques have become available. Recent studies appear to be utilizing a number of methods as the research focus diversifies. Experimental and genetic labwork are now used more widely, as finer ecological details of MCEs are explored, with large increases in the use of these methods observed from 2011. Technological advances are observed with the arrival of ROVs, AUVs, and Baited Remote Underwater Video (BRUVs) from 2001 onwards. SCUBA remains the most popular method, accounting for 33% of studies post-2010 although labwork, including identification, experiments, and genetics, totals 31%. Figure 5. Open in new tabDownload slide Methods used in mesophotic studies. SCUBA = Self-Contained Underwater Breathing Apparatus, ROV = Remotely Operated Vehicle, AUV = Autonomous Underwater Vehicle, BRUV = Baited Remote Underwater Video. Figure 5. Open in new tabDownload slide Methods used in mesophotic studies. SCUBA = Self-Contained Underwater Breathing Apparatus, ROV = Remotely Operated Vehicle, AUV = Autonomous Underwater Vehicle, BRUV = Baited Remote Underwater Video. Like their shallow-water counterparts, MCEs vary with respect to their location around the globe (Baker et al., 2016). While accurate species numbers are rarely reported, results that could be extracted are summarized in Table 4. No data were available for the Indian Ocean and South-East Asia regions but MCEs harbour high benthic and fish biodiversity in all other regions. Atlantic MCEs are less species-rich in terms of scleractinian corals, but macroalgal and sponge diversity is high. Challenges in species identification are a likely cause of low numbers of sponge and algal studies, particularly as remote methods become increasingly used. Reported transition depths, defined as the boundary where significant changes in species composition are observed, thus representing the transition between upper and lower mesophotic communities, appear variable (Table 5). Fish transition depths appear shallower, however, this is probably an artefact of most fish studies including surveys at shallower depths (Mean start depth = 22.7 ± 4.4 m) with 61% of studies completed using SCUBA. This could be interpreted better as the transition between shallow and “mesophotic associated” fish communities. Most mesophotic benthic studies start at greater depths (Mean start depth = 40.5 ± 4.6 m) covering the entire mesophotic range (Mean end depth = 218.3 ± 37.6 m). Benthic communities in the Atlantic transition to more deep-specialized communities at 60 m. This does not hold globally, with this change occurring at greater depths in the Pacific and Australia. Table 4. Mean species richness at mesophotic depths (>30 m) for each region ± Standard Error. Region . Macroalgae . Scleractinian coral . Fish . Sponge . Atlantic 90.3 (± 25.9, n = 4) 16.3 (±1.7, n = 18) 77.3 (±9.2, n = 16) 79.6 (±28.8, n = 9) Australia 32.0 (±3.0, n = 3) 240.4 (±37.7, n = 5) Middle East 48.0 (±20.3, n = 3) 139 (n = 1) Pacific 69.8 (± 7.6, n = 4) 27.6 (±6.6, n = 5) 132.7 (±6.6, n = 7) Region . Macroalgae . Scleractinian coral . Fish . Sponge . Atlantic 90.3 (± 25.9, n = 4) 16.3 (±1.7, n = 18) 77.3 (±9.2, n = 16) 79.6 (±28.8, n = 9) Australia 32.0 (±3.0, n = 3) 240.4 (±37.7, n = 5) Middle East 48.0 (±20.3, n = 3) 139 (n = 1) Pacific 69.8 (± 7.6, n = 4) 27.6 (±6.6, n = 5) 132.7 (±6.6, n = 7) Blanks show no data available for that region. Open in new tab Table 4. Mean species richness at mesophotic depths (>30 m) for each region ± Standard Error. Region . Macroalgae . Scleractinian coral . Fish . Sponge . Atlantic 90.3 (± 25.9, n = 4) 16.3 (±1.7, n = 18) 77.3 (±9.2, n = 16) 79.6 (±28.8, n = 9) Australia 32.0 (±3.0, n = 3) 240.4 (±37.7, n = 5) Middle East 48.0 (±20.3, n = 3) 139 (n = 1) Pacific 69.8 (± 7.6, n = 4) 27.6 (±6.6, n = 5) 132.7 (±6.6, n = 7) Region . Macroalgae . Scleractinian coral . Fish . Sponge . Atlantic 90.3 (± 25.9, n = 4) 16.3 (±1.7, n = 18) 77.3 (±9.2, n = 16) 79.6 (±28.8, n = 9) Australia 32.0 (±3.0, n = 3) 240.4 (±37.7, n = 5) Middle East 48.0 (±20.3, n = 3) 139 (n = 1) Pacific 69.8 (± 7.6, n = 4) 27.6 (±6.6, n = 5) 132.7 (±6.6, n = 7) Blanks show no data available for that region. Open in new tab Table 5. Mean transition depth between benthic and fish communities for each region ± Standard Error. Region . Benthic . Fish . Atlantic 60.9 m (±4.68, n = 12) 63.1 m (± 4.9, n = 8) Australia 75.3 m (±3.4, n = 15) 49 m (n = 1) Middle East 50.0 m (±0.0, n = 2) Pacific 74.7 m (±3.6, n = 7) 61.3 m (±16.1, n = 9) Region . Benthic . Fish . Atlantic 60.9 m (±4.68, n = 12) 63.1 m (± 4.9, n = 8) Australia 75.3 m (±3.4, n = 15) 49 m (n = 1) Middle East 50.0 m (±0.0, n = 2) Pacific 74.7 m (±3.6, n = 7) 61.3 m (±16.1, n = 9) Blanks show no data available for that region. Open in new tab Table 5. Mean transition depth between benthic and fish communities for each region ± Standard Error. Region . Benthic . Fish . Atlantic 60.9 m (±4.68, n = 12) 63.1 m (± 4.9, n = 8) Australia 75.3 m (±3.4, n = 15) 49 m (n = 1) Middle East 50.0 m (±0.0, n = 2) Pacific 74.7 m (±3.6, n = 7) 61.3 m (±16.1, n = 9) Region . Benthic . Fish . Atlantic 60.9 m (±4.68, n = 12) 63.1 m (± 4.9, n = 8) Australia 75.3 m (±3.4, n = 15) 49 m (n = 1) Middle East 50.0 m (±0.0, n = 2) Pacific 74.7 m (±3.6, n = 7) 61.3 m (±16.1, n = 9) Blanks show no data available for that region. Open in new tab We have reviewed studies looking at genetic differences between corals and their associated Symbiodinium to describe connectivity patterns between MCEs and shallow reefs, and these studies showed distinct differences with depth (58% of records), between and within genera (Table 6; Figure 6). Most genetic differences appear to occur below 30 m potentially implying shallow and deep populations. Six genera (Acropora, Eusmilia, Helioseris, Meandrina, Montipora, and Mycetophyllia) showed no genetic differences across depth, though most were only sampled in a single study (except n = 2 for Helioseris). Table 6. Numbers of records of whether genetic differences in corals of their Symbiodinium occur (Yes) or do not occur (No) with depth. Genera . Species . No . Yes . Depth range investigated . References . Acropora Acropora elegans 1 10–48 m (Bongaerts et al., 2011c) Agaricia Agaricia agaricites 2 5–50 m (Bongaerts et al., 2013a, 2015a) Agaricia fragilis 1 12–40 m (Bongaerts et al., 2017) Agaricia grahamae 1 1 15–90 m (Bongaerts et al., 2013a, 2015b) Agaricia lamarcki 3 10–70 m (Bongaerts et al., 2013a, 2015a; Lucas et al., 2016) Agaricia undata 1 15–90 m (Bongaerts et al., 2015b) Echinophyllia Echinophyllia aspera 1 10–62 m (Bongaerts et al., 2011c) Eusmilia Eusmilia fastigiata 1 5–40 m (Bongaerts et al., 2015a) Galaxea Galaxea astreata 1 10–55 m (Bongaerts et al., 2011c) Helioseris Helioseris cucullata 2 25–45 m (Bongaerts et al., 2013a, 2015a) Leptoseris Leptoseris hawaiiensis 1 10–70 m (Bongaerts et al., 2011c) Leptoseris spp. 1 3 1–127 m (Chan et al., 2009; Luck et al., 2013; Pochon et al., 2015; Ziegler et al., 2015) Madracis Madracis carmabi 1 5–40 m (Frade et al., 2008b) Madracis decatis 2 5–40 m (Frade et al., 2008b; Bongaerts et al., 2015a) Madracis formosa 2 1 5–60 m (Frade et al., 2008a,b; Bongaerts et al., 2015a) Madracis mirabilis 1 1 5–40 m (Bongaerts et al., 2015a), Madracis pharensis 5 5–90 m (Frade et al., 2008a,b; Bongaerts et al., 2015a,b) Madracis senaria 2 5–40 m (Frade et al., 2008a,b) Meandrina Meandrina meandrites 1 5–40 m (Bongaerts et al., 2015a) Montastrea Montastrea cavernosa 3 3–91 m (Lesser et al., 2010; Brazeau et al., 2013; Bongaerts et al., 2015a) Montipora Montipora spp. 1 10–70 m (Bongaerts et al., 2011c) Mycetophyllia Mycetophyllia ferox 1 25–40 m (Bongaerts et al., 2015a) Orbicella Orbicella faveolata 1 5–25 m (Bongaerts et al., 2015a) Orbicella franksi 1 10–25 m (Bongaerts et al., 2015a) Pachyseris Pachyseris speciosa 1 2 1–62 m (Bongaerts et al., 2011c; Cooper et al., 2011; Ziegler et al., 2015) Pavona Pavona spp. 1 10–59 m (Bongaerts et al., 2011c) Porites Porites astreoides 2 3 2–30 m (Bongaerts et al., 2015a; Serrano et al., 2016; Reich et al., 2017) Porites spp. 1 1 1–70 m (Bongaerts et al., 2011c; Ziegler et al., 2015) Seriatopora Seriatopora hystrix 3 4 2–57 m (Bongaerts et al., 2010b,2011b,c; Cooper et al., 2011; Nir et al., 2011; van Oppen et al., 2011) Siderastrea Siderastrea siderea 1 2–50 m (Bongaerts et al., 2015a) Stephanocoenia Stephanocoenia intersepta 1 1 10–60 m (Bongaerts et al., 2015a, 2017) Genera . Species . No . Yes . Depth range investigated . References . Acropora Acropora elegans 1 10–48 m (Bongaerts et al., 2011c) Agaricia Agaricia agaricites 2 5–50 m (Bongaerts et al., 2013a, 2015a) Agaricia fragilis 1 12–40 m (Bongaerts et al., 2017) Agaricia grahamae 1 1 15–90 m (Bongaerts et al., 2013a, 2015b) Agaricia lamarcki 3 10–70 m (Bongaerts et al., 2013a, 2015a; Lucas et al., 2016) Agaricia undata 1 15–90 m (Bongaerts et al., 2015b) Echinophyllia Echinophyllia aspera 1 10–62 m (Bongaerts et al., 2011c) Eusmilia Eusmilia fastigiata 1 5–40 m (Bongaerts et al., 2015a) Galaxea Galaxea astreata 1 10–55 m (Bongaerts et al., 2011c) Helioseris Helioseris cucullata 2 25–45 m (Bongaerts et al., 2013a, 2015a) Leptoseris Leptoseris hawaiiensis 1 10–70 m (Bongaerts et al., 2011c) Leptoseris spp. 1 3 1–127 m (Chan et al., 2009; Luck et al., 2013; Pochon et al., 2015; Ziegler et al., 2015) Madracis Madracis carmabi 1 5–40 m (Frade et al., 2008b) Madracis decatis 2 5–40 m (Frade et al., 2008b; Bongaerts et al., 2015a) Madracis formosa 2 1 5–60 m (Frade et al., 2008a,b; Bongaerts et al., 2015a) Madracis mirabilis 1 1 5–40 m (Bongaerts et al., 2015a), Madracis pharensis 5 5–90 m (Frade et al., 2008a,b; Bongaerts et al., 2015a,b) Madracis senaria 2 5–40 m (Frade et al., 2008a,b) Meandrina Meandrina meandrites 1 5–40 m (Bongaerts et al., 2015a) Montastrea Montastrea cavernosa 3 3–91 m (Lesser et al., 2010; Brazeau et al., 2013; Bongaerts et al., 2015a) Montipora Montipora spp. 1 10–70 m (Bongaerts et al., 2011c) Mycetophyllia Mycetophyllia ferox 1 25–40 m (Bongaerts et al., 2015a) Orbicella Orbicella faveolata 1 5–25 m (Bongaerts et al., 2015a) Orbicella franksi 1 10–25 m (Bongaerts et al., 2015a) Pachyseris Pachyseris speciosa 1 2 1–62 m (Bongaerts et al., 2011c; Cooper et al., 2011; Ziegler et al., 2015) Pavona Pavona spp. 1 10–59 m (Bongaerts et al., 2011c) Porites Porites astreoides 2 3 2–30 m (Bongaerts et al., 2015a; Serrano et al., 2016; Reich et al., 2017) Porites spp. 1 1 1–70 m (Bongaerts et al., 2011c; Ziegler et al., 2015) Seriatopora Seriatopora hystrix 3 4 2–57 m (Bongaerts et al., 2010b,2011b,c; Cooper et al., 2011; Nir et al., 2011; van Oppen et al., 2011) Siderastrea Siderastrea siderea 1 2–50 m (Bongaerts et al., 2015a) Stephanocoenia Stephanocoenia intersepta 1 1 10–60 m (Bongaerts et al., 2015a, 2017) Open in new tab Table 6. Numbers of records of whether genetic differences in corals of their Symbiodinium occur (Yes) or do not occur (No) with depth. Genera . Species . No . Yes . Depth range investigated . References . Acropora Acropora elegans 1 10–48 m (Bongaerts et al., 2011c) Agaricia Agaricia agaricites 2 5–50 m (Bongaerts et al., 2013a, 2015a) Agaricia fragilis 1 12–40 m (Bongaerts et al., 2017) Agaricia grahamae 1 1 15–90 m (Bongaerts et al., 2013a, 2015b) Agaricia lamarcki 3 10–70 m (Bongaerts et al., 2013a, 2015a; Lucas et al., 2016) Agaricia undata 1 15–90 m (Bongaerts et al., 2015b) Echinophyllia Echinophyllia aspera 1 10–62 m (Bongaerts et al., 2011c) Eusmilia Eusmilia fastigiata 1 5–40 m (Bongaerts et al., 2015a) Galaxea Galaxea astreata 1 10–55 m (Bongaerts et al., 2011c) Helioseris Helioseris cucullata 2 25–45 m (Bongaerts et al., 2013a, 2015a) Leptoseris Leptoseris hawaiiensis 1 10–70 m (Bongaerts et al., 2011c) Leptoseris spp. 1 3 1–127 m (Chan et al., 2009; Luck et al., 2013; Pochon et al., 2015; Ziegler et al., 2015) Madracis Madracis carmabi 1 5–40 m (Frade et al., 2008b) Madracis decatis 2 5–40 m (Frade et al., 2008b; Bongaerts et al., 2015a) Madracis formosa 2 1 5–60 m (Frade et al., 2008a,b; Bongaerts et al., 2015a) Madracis mirabilis 1 1 5–40 m (Bongaerts et al., 2015a), Madracis pharensis 5 5–90 m (Frade et al., 2008a,b; Bongaerts et al., 2015a,b) Madracis senaria 2 5–40 m (Frade et al., 2008a,b) Meandrina Meandrina meandrites 1 5–40 m (Bongaerts et al., 2015a) Montastrea Montastrea cavernosa 3 3–91 m (Lesser et al., 2010; Brazeau et al., 2013; Bongaerts et al., 2015a) Montipora Montipora spp. 1 10–70 m (Bongaerts et al., 2011c) Mycetophyllia Mycetophyllia ferox 1 25–40 m (Bongaerts et al., 2015a) Orbicella Orbicella faveolata 1 5–25 m (Bongaerts et al., 2015a) Orbicella franksi 1 10–25 m (Bongaerts et al., 2015a) Pachyseris Pachyseris speciosa 1 2 1–62 m (Bongaerts et al., 2011c; Cooper et al., 2011; Ziegler et al., 2015) Pavona Pavona spp. 1 10–59 m (Bongaerts et al., 2011c) Porites Porites astreoides 2 3 2–30 m (Bongaerts et al., 2015a; Serrano et al., 2016; Reich et al., 2017) Porites spp. 1 1 1–70 m (Bongaerts et al., 2011c; Ziegler et al., 2015) Seriatopora Seriatopora hystrix 3 4 2–57 m (Bongaerts et al., 2010b,2011b,c; Cooper et al., 2011; Nir et al., 2011; van Oppen et al., 2011) Siderastrea Siderastrea siderea 1 2–50 m (Bongaerts et al., 2015a) Stephanocoenia Stephanocoenia intersepta 1 1 10–60 m (Bongaerts et al., 2015a, 2017) Genera . Species . No . Yes . Depth range investigated . References . Acropora Acropora elegans 1 10–48 m (Bongaerts et al., 2011c) Agaricia Agaricia agaricites 2 5–50 m (Bongaerts et al., 2013a, 2015a) Agaricia fragilis 1 12–40 m (Bongaerts et al., 2017) Agaricia grahamae 1 1 15–90 m (Bongaerts et al., 2013a, 2015b) Agaricia lamarcki 3 10–70 m (Bongaerts et al., 2013a, 2015a; Lucas et al., 2016) Agaricia undata 1 15–90 m (Bongaerts et al., 2015b) Echinophyllia Echinophyllia aspera 1 10–62 m (Bongaerts et al., 2011c) Eusmilia Eusmilia fastigiata 1 5–40 m (Bongaerts et al., 2015a) Galaxea Galaxea astreata 1 10–55 m (Bongaerts et al., 2011c) Helioseris Helioseris cucullata 2 25–45 m (Bongaerts et al., 2013a, 2015a) Leptoseris Leptoseris hawaiiensis 1 10–70 m (Bongaerts et al., 2011c) Leptoseris spp. 1 3 1–127 m (Chan et al., 2009; Luck et al., 2013; Pochon et al., 2015; Ziegler et al., 2015) Madracis Madracis carmabi 1 5–40 m (Frade et al., 2008b) Madracis decatis 2 5–40 m (Frade et al., 2008b; Bongaerts et al., 2015a) Madracis formosa 2 1 5–60 m (Frade et al., 2008a,b; Bongaerts et al., 2015a) Madracis mirabilis 1 1 5–40 m (Bongaerts et al., 2015a), Madracis pharensis 5 5–90 m (Frade et al., 2008a,b; Bongaerts et al., 2015a,b) Madracis senaria 2 5–40 m (Frade et al., 2008a,b) Meandrina Meandrina meandrites 1 5–40 m (Bongaerts et al., 2015a) Montastrea Montastrea cavernosa 3 3–91 m (Lesser et al., 2010; Brazeau et al., 2013; Bongaerts et al., 2015a) Montipora Montipora spp. 1 10–70 m (Bongaerts et al., 2011c) Mycetophyllia Mycetophyllia ferox 1 25–40 m (Bongaerts et al., 2015a) Orbicella Orbicella faveolata 1 5–25 m (Bongaerts et al., 2015a) Orbicella franksi 1 10–25 m (Bongaerts et al., 2015a) Pachyseris Pachyseris speciosa 1 2 1–62 m (Bongaerts et al., 2011c; Cooper et al., 2011; Ziegler et al., 2015) Pavona Pavona spp. 1 10–59 m (Bongaerts et al., 2011c) Porites Porites astreoides 2 3 2–30 m (Bongaerts et al., 2015a; Serrano et al., 2016; Reich et al., 2017) Porites spp. 1 1 1–70 m (Bongaerts et al., 2011c; Ziegler et al., 2015) Seriatopora Seriatopora hystrix 3 4 2–57 m (Bongaerts et al., 2010b,2011b,c; Cooper et al., 2011; Nir et al., 2011; van Oppen et al., 2011) Siderastrea Siderastrea siderea 1 2–50 m (Bongaerts et al., 2015a) Stephanocoenia Stephanocoenia intersepta 1 1 10–60 m (Bongaerts et al., 2015a, 2017) Open in new tab Figure 6. Open in new tabDownload slide Depths at which genetic changes in corals and their Symbiodinium occur. Figure 6. Open in new tabDownload slide Depths at which genetic changes in corals and their Symbiodinium occur. Discussion Studies of Mesophotic coral ecosystems (MCEs) are currently highly location and region specific and not represented in all oceans globally. While this is also the case with shallow reefs (Fisher et al., 2011) the imbalance is not as great. A result of the strong locational bias is that there is not enough evidence to suggest an understanding of the ecological role of MCEs in a global context. Data collection in these ecosystems is still relatively expensive, as most methods require specialized equipment and training. It seems likely that this is the main reason why mesophotic studies are concentrated in areas where the initial investments have been made and equipment is available to enable specific research groups to explore these ecosystems. Huge regional gaps are apparent, showing that almost no studies have been conducted in the Indian Ocean and South-East Asia regions. This is of particular concern given the known high biodiversity of shallow coral ecosystems in these regions and the threats they face (Burke et al., 2011). MCE research has been mainly focussed in the exploratory phase, aiming to characterize the communities in different locations. What we know from these descriptive studies is that there is a common depth/light attenuation pattern in MCE benthic community structure indicating that upper mesophotic depths have a dominance, in terms of percentage cover, of phototrophic taxa, predominantly corals, shifting to primarily heterotrophic communities, made up of sponges and octocorals, of the lower mesophotic (Lesser et al., 2009; Bongaerts et al., 2010a; Kahng et al., 2010; Baker et al., 2016). It is also well understood that light, topography, and temperature stand out as three main factors that influence the structure of MCE communities. Light is the major factor, with the deepest zooxanthellate coral records associated with areas known for clear water (Kahng et al., 2010; Baker et al., 2016). Topography is also important, with local bathymetric features, such as slope, influencing benthic community structure (Bridge et al., 2010; Locker et al., 2010; Sherman et al., 2010; Englebert et al., 2017). Temperature is influenced by local upwelling (Bridge et al., 2010) and internal waves (Kahng and Kelley, 2007; Kahng et al., 2012). This affects depth limits of organisms (Kahng et al., 2012) particularly at higher latitudes (Grigg, 2005) where corals are already residing close to their physiological limits. As the processes associated with MCEs have become better understood, fewer descriptive studies are being carried out and a greater proportion are focused on understanding ecological processes. Moving forward there is more need for studies to be targeted in locations with varying combinations of these influencing factors, as well as proving these trends hold in currently unstudied regions. We still know little about the pressures that MCEs face, from both anthropogenic and natural sources. The “deep reef refugia” hypothesis suggests that mesophotic areas are more remote from these threats and may re-seed impacted shallow areas (Bongaerts et al., 2010a; Hinderstein et al., 2010). Locational differences occur for natural impacts, for example coral bleaching and disease are reported mainly in the Caribbean (Garcia-Sais et al., 2007; Nemeth et al., 2008; Smith et al., 2015) whereas storm impacts are common in western Pacific areas (Harmelin-Vivien and Laboute, 1986; Bongaerts et al., 2013b; White et al., 2013). Human impacts are currently poorly documented and although localized studies are occurring (Appeldoorn et al., 2015), not enough evidence is available to discuss global or regional trends and threats. Additionally, recovery rates appear to be largely unknown. This is inevitable given that current impacts on MCEs are likely unnoticed or unquantified. This kind of longitudinal information is crucial for effective management of these systems. Additional gaps lie around the direct measurements of life history characteristics and post-settlement processes of benthic organisms at mesophotic depths. Further work into life history dynamics of mesophotic organisms will give an insight into resilience and recovery when faced with disturbances. Conflicting results have been found in terms of fecundity and spawning synchrony of mesophotic coral colonies (Holstein et al., 2015; Prasetia et al., 2016) that also vary between species and locations (Eyal-Shaham et al., 2016). These variations highlight that we know little regarding this subject, which is a concern for managers. Technological advances have made a range of techniques available for studying MCEs; however, the cost of many of these techniques impacts on the extent to which they are used for data collection. Technical SCUBA diving, despite the training and equipment required, tends to be a cheaper option hence its popularity. The advantage of diving is that it permits investigation of organisms in situ, allowing easier species identification and more precise sample collection. ROVs are commonly utilized for sample collection in the more inaccessible Cold-Water Coral (CWC) ecosystems where the simultaneous use of video allows increased sampling precision and minimal damage (Fosså et al., 2005). There is a lack of precise benthic species diversity data, in particular for difficult to identify macroalgae and sponges, so it is important that this taxonomic detail is not lost, and rather targeted to assess specific community structures. However, diving only allows small areas to be surveyed, which may not meet management goals, and additional health and safety concerns associated with technical diving may mean remote methods are more appropriate in some areas. Acoustic methods cover large areas and allow geophysical variables to be measured at fine scales, where reef corals show distinct bathymetric signatures (Brown et al., 2011). Acoustic data have proven to be highly successful for identifying the extent of CWC ecosystems (Fosså et al., 2005; Roberts et al., 2009; Buhl-Mortensen et al., 2015). Estimates of total habitat area can be deduced and the information can be used for habitat suitability modelling to identify areas of likely occurrence, which has performed well when applied to mesophotic habitats, given the knowledge of the key structuring variables (Bridge et al., 2012; Costa et al., 2015). Detailed bathymetric information allows for planning of future surveys and can assist with ROV navigation, particularly in areas of high rugosity (Fosså et al., 2005). ROVs are often used to provide qualitative visual information to explore new areas (Kahng and Kelley, 2007; Bongaerts et al., 2011a; Blythe-Skyrme et al., 2013; Englebert et al., 2014), commonly prior to committing divers or to survey depths >150 m. Samples collected from ROVs have allowed the detailed taxonomy of mesophotic corals (Muir et al., 2015) as well as further lab experiments (van Oppen et al., 2011). CWC ecosystems have utilized ROVs to deploy additional equipment and set up in situ experiments (Roberts et al., 2009) which are an approach that should be considered for MCEs. AUVs, while unable to collect samples, can offer a more quantitative approach to obtaining imagery. Hundreds of thousands of accurately georeferenced images may be collected, as well as accompanying environmental information (Williams et al., 2012; Pizarro et al., 2013), while also having the advantage of running independently to the deployment vessel. AUVs have the capability to accurately perform repeat monitoring surveys and relocate colonies (Pizarro et al., 2013; Ferrari et al., 2016) which enable an insight into processes such as growth rates in the future. In terms of costs per area surveyed remote methods may be cheaper, although their ability to fill data gaps surrounding life history traits may be limited; however they may be complemented by diving surveys and experiments. High biodiversity is common across MCEs of all regions (Baker et al., 2016) although there is still further biodiversity to be discovered as mentioned above. High taxonomic resolution is required to assess connectivity, as will be discussed, where species-specific differences are observed. Depths at which communities change appear to be area-specific and evidence appears to suggest that using depth alone as a basis for universal definitions may not be appropriate. The depth of the transition zone, representing a shift in upper and lower mesophotic assemblages also varies between locations. While 60 m is commonly reported (Fricke and Meischner, 1985; Liddell and Ohlhorst, 1988; Bongaerts et al., 2010a; Bridge et al., 2010; Slattery et al., 2011) on average this only applies to the Atlantic region. In the Pacific and the Coral Sea, the transition zone depth extends past 80 m (Kahng and Kelley, 2007; Pyle et al., 2016; Englebert et al., 2017). Equally, mesophotic depths are shallower for locations with lower light regimes, such as Ningaloo, Australia (Rees et al., 2004) or Pohnpei, Micronesia (Muir and Wallace, 2016) and reduced temperature, such as Bermuda (Fricke and Meischner, 1985). This raises questions about the ecological relevance (Laverick et al., 2016) for the global definition of the transition to MCEs of 30–40 m (Puglise et al., 2009; Hinderstein et al., 2010; Baker et al., 2016). Most mesophotic fish surveys use SCUBA-based methods and often make use of lengthy decompression schedules to collect accompanying shallow water data (Lombardi and Godfrey, 2011; Andradi-Brown et al., 2016a) so as to allow comparisons to be made. High proportions of fish species are common to both shallow and lower mesophotic areas across regions (Bejarano et al., 2014; Wagner et al., 2014; Lindfield et al., 2015), while genetic similarities are also described (Tenggardjaja et al., 2014). Ontogenic movements are also reported (Brokovich et al., 2006; Rosa et al., 2015; Andradi-Brown et al., 2016b) suggesting movement is common across depths. Given these findings, reported transition depths seem to represent the change from shallow water to mesophotic associated fish communities, in contrast to benthic communities where transition depths represent the change from upper to lower mesophotic; having already seen a shift from shallow waters. While changes in the benthic composition are likely to affect distributions of fish species (Garcia-Sais et al., 2007; Brokovich et al., 2008; Garcia-Sais, 2010): corals may decrease but sponges and other benthic organisms can provide structural habitat (Bell et al., 2013) at depth. Other factors may be structuring fish communities, such as food availability, given the distinct changes in functional groups observed (Bridge et al., 2016b). Assessing connectivity between shallow and deep reefs is a primary focus in the published literature, and more studies are being undertaken in this research area. Vertical connectivity will ultimately determine whether MCEs can re-seed shallow coral reefs following chronic disturbances. Questions do remain over how genetic changes in Symbiodinium correlate with that of their hosts. However, given that host specificity is common and specific adaptation to environmental conditions are likely to have evolved (LaJeunesse et al., 2004; Frade et al., 2008b; Finney et al., 2010) differences probably indicate genetic separation of shallow and MCEs (Bongaerts et al., 2010b,c). Populations below 30 m are reported as unconnected to shallower conspecifics, with distinct shallow and deep genetic populations found (Brazeau et al., 2013). Deeper coral populations are specialized to lower light conditions, showing changes in morphology (Fricke and Meischner, 1985; Einbinder et al., 2009; Nir et al., 2011), photosynthetic efficiency (Lesser et al., 2010; Mass et al., 2010; Nir et al., 2011; Einbinder et al., 2016), and alternative nutrient sources (Muscatine et al., 1989; Einbinder et al., 2009; Crandall et al., 2016). Isolated reefs appear to have higher vertical genetic connectivity, possibly due to the importance of localized recruitment for sustaining populations (Serrano et al., 2016) or reduced competition following disturbance which may prevent localized extinctions (van Oppen et al., 2011; Sinniger et al., 2012; Muir et al., 2015). However, using Bermuda as an example, not all species show this pattern of strong vertical connectivity (Bongaerts et al., 2017). The reproductive mode may give some insight, with broadcast spawning genera generally showing reduced genetic partitioning with depth (Bongaerts et al., 2011c,, 2017) although this is not exclusive (Bongaerts et al., 2015a). Local environmental conditions also play a role, and light levels will ultimately influence the upper and lower limits of coral species and their Symbiodinium types due to functional adaptations (Frade et al., 2008a,c). This again calls into question the use of only depth to define deep and shallow MCEs. Overall, the findings in this review show that differences in vertical connectivity patterns at species and genera level are common across MCEs globally. These results highlight our limited knowledge, and the need for these studies to be done at both localized scales, for a detailed analysis of local populations, and across biogeographic ranges. Conclusions The importance of mesophotic areas is now recognized in the scientific community. There is a clear locational bias of the existing research to the Atlantic, and specifically the Caribbean, which makes the extrapolation of findings to the rest of the world difficult. Definitions coined from data in this region alone need to be redefined as more studies are completed globally. A clear priority is to collect data for MCEs in South East Asia and the Indian Ocean. Remote methods are clearly advancing research in this field, though it is important to not lose taxonomic detail, given the apparent species and location specificity of connectivity patterns. If shallow and deep populations are separate, then management plans need to accommodate this in order to conserve the different biodiversity of both of these light-mediated ecosystems. The current lack of information about the threats and impacts on MCEs needs to be addressed immediately so that they can be identified at local, regional, and global scales so that effective management can be implemented. Further prioritization of such studies, as well as those investigating connectivity at both local and regional scales, is clearly required, to ensure adequate protection of these ecosystems and their shallow water counterparts, for which relying on MCEs as refugia may not be appropriate. Acknowledgements We thank our funding agency BHP Billiton-CSIRO Ningaloo Outlook Marine Research Partnership for support of this work. The views expressed herein are those of the authors and do not necessarily reflect the views of BHP Billiton or CSIRO. We would also like to thank Tom Bridge and Tim Cooper for their comments on the manuscript prior to submission. References Andradi-Brown D. , East A., Shepherd L., Stockdale E., Rogers A. 2016a . Challenges and opportunities in mesophotic reef research . Reef Encounter: The News Journal of the International Society for Reef Studies , 31 : 26 – 31 . Google Scholar OpenURL Placeholder Text WorldCat Andradi-Brown D. , Gress E., Wright G., Exton D., Rogers A. 2016b . Reef fish community biomass and trophic structure changes across shallow to upper-mesophotic reefs in the Mesoamerican barrier reef, Caribbean . PLoS One , 11 : e0156641 . Google Scholar Crossref Search ADS WorldCat Appeldoorn R. , Ballantine D., Bejarano I., Carlo M., Nemeth M., Otero E., Pagan F., et al. 2015 . Mesophotic coral ecosystems under anthropogenic stress: a case study at Ponce, Puerto Rico . Coral Reefs , 35 : 63 – 75 . Google Scholar Crossref Search ADS WorldCat Armstrong R. A. , Singh H., Torres J., Nemeth R. S., Can A., Roman C., Eustice R., et al. 2006 . Characterising the deep insular shelf coral reef habitat of the Hind Bank Marine Conservation District (US Virgin Islands) using the Seabed Autonomous Underwater Vehicle . Continental Shelf Research , 26 : 194 – 205 . Google Scholar Crossref Search ADS WorldCat Bak R. P. M. , Nieuwland G., Meesters E. H. 2005 . Coral reef crisis in deep and shallow reefs: 30 years of constancy and change in reefs of Curacao and Bonaire . Coral Reefs , 24 : 475 – 479 . Google Scholar Crossref Search ADS WorldCat Baker E. K. , Puglise K. A., Harris P. T. 2016 . Mesophotic Coral Ecosystems—A Lifeboat for Coral Reefs? The United Nations Environment Programme and GRID-Arendal, Nairobi and Arendal. Bejarano I. , Appeldoorn R. S., Nemeth M. 2014 . Fishes associated with mesophotic coral ecosystems in La Parguera, Puerto Rico . Coral Reefs , 33 : 313 – 328 . Google Scholar Crossref Search ADS WorldCat Bell J. J. , Davy S. K., Jones T., Taylor M. W., Webster N. S. 2013 . Could some coral reefs become sponge reefs as our climate changes? Global Change Biology , 19 : 2613 – 2624 . Google Scholar Crossref Search ADS PubMed WorldCat Bellwood D. R. , Hughes T. P., Folke C., Nystrom M. 2004 . Confronting the coral reef crisis . Nature , 429 : 827 – 833 . Google Scholar Crossref Search ADS PubMed WorldCat Blythe-Skyrme V. J. , Rooney J., Parrish F., Boland R. 2013 . Mesophotic Coral Ecosystems—Potential Candidates as Essential Fish Habitat and Habitat Areas of Particular Concern . Pacific Island Fisheries Science Center, National Marine Fisheries Science Center Administrative Reports , H-13-02 : 53 . p. Google Scholar OpenURL Placeholder Text WorldCat Bongaerts P. , Bridge T. C. L., Kline D. I., Muir P. R., Wallace C. C., Beaman R. J., Hoegh-Guldberg O. 2011a . Mesophotic coral ecosystems on the walls of Coral Sea atolls . Coral Reefs , 30 : 335 – 335 . Google Scholar Crossref Search ADS WorldCat Bongaerts P. , Carmichael M., Hay K. B., Tonk L., Frade P. R., Hoegh-Guldberg O. 2015a . Prevalent endosymbiont zonation shapes the depth distributions of scleractinian coral species . Royal Society Open Science , 2 : 140297 . Google Scholar Crossref Search ADS WorldCat Bongaerts P. , Frade P. R., Hay K. B., Englebert N., Latijnhouwers K. R., Bak R. P., Vermeij M. J., et al. 2015b . Deep down on a Caribbean reef: lower mesophotic depths harbor a specialized coral-endosymbiont community . Scientific Reports , 5 : 7652 . Google Scholar Crossref Search ADS WorldCat Bongaerts P. , Frade P. R., Ogier J. J., Hay K. B., van Beijswijk J., Englebert N., Vermeij M. J., et al. 2013a . Sharing the slope: depth partitioning of agariciid corals and associated Symbiodinium across shallo and mesophotic habitats (2-60 m) on a Caribbean reef . BMC Evolutionary Biology , 13 : 205 . Google Scholar Crossref Search ADS WorldCat Bongaerts P. , Muir P., Englebert N., Bridge T. C. L., Hoegh-Guldberg O. 2013b . Cyclone damage at mesophotic depths on Myrmidon Reef (GBR) . Coral Reefs , 32 : 935 – 935 . Google Scholar Crossref Search ADS WorldCat Bongaerts P. , Ridway T., Sampayo E. M., Hoegh-Guldberg O. 2010a . Assessing the ‘deep reef refugia’ hypothesis: focus on Caribbean reefs . Coral Reefs , 29 : 309 – 327 . Google Scholar Crossref Search ADS WorldCat Bongaerts P. , Riginos C., Brunner R., Englebert N., Smith S. R., Hoegh-Guldberg O. 2017 . Deep reefs are not universal refuges: Reseeding potential varies among coral species . Science Advances , 3 : e1602373 . Google Scholar Crossref Search ADS PubMed WorldCat Bongaerts P. , Riginos C., Hay K. B., van Oppen M. J., Hoegh-Guldberg O., Dove S. 2011b . Adaptive divergence in a scleractinian coral: physiological adaptation of Seriatopora hystrix to shallow and deep reef habitats . BMC Evolutionary Biology , 11 : 303 . Google Scholar Crossref Search ADS WorldCat Bongaerts P. , Riginos C., Ridgway T., Sampayo E. M., van Oppen M. J. H., Englebert N., Vermeulen F., et al. 2010b . Genetic divergence across habitats in the widespread coral Seriatopora hystrix and its associated Symbiodinium . PLoS One 5 : e10871 . Google Scholar Crossref Search ADS WorldCat Bongaerts P. , Sampayo E. M., Bridge T. C. L., Ridgway T., Vermeulen F., Englebert N., Webster J. M., et al. 2011c . Symbiodinium diversity in mesophotic coral communities on the Great Barrier Reef: a first assessment . Marine Ecology Progress Series , 439 : 117 – 126 . Google Scholar Crossref Search ADS WorldCat Bouchon C. 1981 . Quantitative study of the Scleractinian coral communities of a fringing reef of Reunion Island (Indian Ocean) . Marine Ecology Progress Series , 4 : 273 – 288 . Google Scholar Crossref Search ADS WorldCat Brazeau D. A. , Lesser M. P., Slattery M. 2013 . Genetic structure in the coral, Montastraea cavernosa: assessing genetic differentiation among and within Mesophotic reefs . PLoS One , 8 : e65845 . Google Scholar Crossref Search ADS PubMed WorldCat Bridge T. , Beaman R., Done T., Webster J. 2012 . Predicting the location and spatial extent of submerged coral reef habitat in the Great Barrier Reef world heritage area, Australia . PLoS One , 7 : e48203 . Google Scholar Crossref Search ADS PubMed WorldCat Bridge T. C. , Grech A. M., Pressey R. L. 2016a . Factors influencing incidental representation of previously unknown conservation features in marine protected areas . Conservation Biology , 30 : 154 – 165 . Google Scholar Crossref Search ADS WorldCat Bridge T. C. , Luiz O. J., Coleman R. R., Kane C. N., Kosaki R. K. 2016b . Ecological and morphological traits predict depth-generalist fishes on coral reefs . Proceedings of the Royal Society of London B: Biological Sciences , 283 : 20152332 . Google Scholar Crossref Search ADS WorldCat Bridge T. C. L. , Done T. J., Beaman R. J., Friedman A., Williams S. B., Pizarro O., Webster J. M. 2010 . Topography, substratum and benthic macrofaunal relationships on a tropical mesophotic shelf margin, central Great Barrier Reef, Australia . Coral Reefs , 30 : 143 – 153 . Google Scholar Crossref Search ADS WorldCat Bridge T. C. L. , Done T. J., Friedman A., Beaman R. J., Williams S. B., Pizarro O., Webster J. M. 2011a . Variability in mesophotic coral reef communities along the Great Barrier Reef, Australia . Marine Ecology Progress Series , 428 : 63 – 75 . Google Scholar Crossref Search ADS WorldCat Bridge T. C. L. , Fabricius K. E., Bongaerts P., Wallace C. C., Muir P. R., Done T. J., Webster J. M. 2011b . Diversity of Scleractinia and Octocorallia in the mesophotic zone of the Great Barrier Reef, Australia . Coral Reefs , 31 : 179 – 189 . Google Scholar Crossref Search ADS WorldCat Bridge T. C. L. , Hughes T. P., Guinotte J. M., Bongaerts P. 2013 . Call to protect all coral reefs . Nature Climate Change , 3 : 528 – 530 . Google Scholar Crossref Search ADS WorldCat Brokovich E. , Einbinder S., Kark S., Shashar N., Kiflawi M. 2006 . A deep nursery for juveniles of the zebra angelfish Genicanthus caudovittatus . Environmental Biology of Fishes , 80 : 1 – 6 . Google Scholar Crossref Search ADS WorldCat Brokovich E. , Einbinder S., Shashar N., Kiflawi M., Kark S. 2008 . Descending to the twilight-zone: changes in coral reef fish assemblages along a depth gradient down to 65 m . Marine Ecology Progress Series , 371 : 253 – 262 . Google Scholar Crossref Search ADS WorldCat Brown C. J. , Smith S. J., Lawton P., Anderson J. T. 2011 . Benthic habitat mapping: a review of progress towards improved understanding of the spatial ecology of the seafloor using acoustic techniques . Estuarine, Coastal and Shelf Science , 92 : 502 – 520 . Google Scholar Crossref Search ADS WorldCat Buhl-Mortensen L. , Buhl-Mortensen P., Dolan M. J. F., Gonzalez-Mirelis G. 2015 . Habitat mapping as a tool for conservation and sustainable use of marine resources: some perspectives from the MAREANO Programme, Norway . Journal of Sea Research , 100 : 46 – 61 . Google Scholar Crossref Search ADS WorldCat Burke L. , Reytar K., Spalding M., Perry A. 2011 . Reefs at Risk Revisited . World Resources Institute , Washington, DC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Busby R. 1966 . Ocean Bottom Reconnaissance Off the East Coast of Andros Island, Bahamas. US Navel Oceanographic Office, Technical Report 20390, Washington DC. Chan Y. L. , Pochon X., Fisher M. A., Wagner D., Concepcion G. T., Kahng S. E., Toonen R. J., et al. 2009 . Generalist dinoflagellate endosymbionts and host genotype diversity detected from mesophotic (67-100 m depths) coral Leptoseris . BMC Ecology , 9 : 21 . Google Scholar Crossref Search ADS PubMed WorldCat Colin P. L. , Devaney M., Hills-Colinvaux L., Suchanek T. H., Harrison J. T. 1986 . Geology and biological zonation of the reef slope, 50-360 m depth, at Enewetak Atoll, Marshall Islands . Bulletin of Marine Science , 38 : 111 – 128 . Google Scholar OpenURL Placeholder Text WorldCat Cooper T. F. , Ulstrup K. E., Dandan S. S., Heyward A. J., Kuhl M., Muirhead A., O'Leary R. A., et al. 2011 . Niche specialization of reef-building corals in the mesophotic zone: metabolic trade-offs between divergent Symbiodinium types . Proceedings of the Royal Society of London B: Biological Sciences , 278 : 1840 – 1850 . Google Scholar Crossref Search ADS WorldCat Costa B. , Kendall M. S., Parrish F. A., Rooney J., Boland R. C., Chow M., Lecky J., et al. 2015 . Identifying suitable locations for mesophotic hard corals offshore of Maui, Hawai'i . PLoS One , 10 : e0130285 . Google Scholar Crossref Search ADS PubMed WorldCat Crandall J. B. , Teece M. A., Estes B. A., Manfrino C., Ciesla J. H. 2016 . Nutrient acquisition strategies in mesophotic hard corals using compound specific stable isotope analysis of sterols . Journal of Experimental Marine Biology and Ecology , 474 : 133 – 141 . Google Scholar Crossref Search ADS WorldCat Einbinder S. , Gruber D. F., Salomon E., Liran O., Keren N., Tchernov D. 2016 . Novel adaptive photosynthetic characteristics of mesophotic symbiotic microalgae within the reef-building coral, Stylophora pistillata . Frontiers in Marine Science , 3 (Article 175): 1–9. Google Scholar OpenURL Placeholder Text WorldCat Einbinder S. , Mass T., Brokovich E., Dubinsky Z., Erez J., Tchernov D. 2009 . Changes in morphology and diet of the coral Stylophora pistillata along a depth gradient . Marine Ecology Progress Series , 381 : 167 – 174 . Google Scholar Crossref Search ADS WorldCat Englebert N. , Bongaerts P., Muir P., Hay K. B., Hoegh-Guldberg O. 2014 . Deepest zooxanthellate corals of the Great Barrier Reef and Coral Sea . Marine Biodiversity , 45 : 1 – 2 . Google Scholar Crossref Search ADS WorldCat Englebert N. , Bongaerts P., Muir P. R., Hay K. B., Pichon M., Hoegh-Guldberg O. 2017 . Lower mesophotic coral communities (60-125 m depth) of the northern Great Barrier Reef and Coral Sea . PLoS One , 12 : e0170336 . Google Scholar Crossref Search ADS PubMed WorldCat Eyal-Shaham L. , Eyal G., Tamir R., Loya Y. 2016 . Reproduction, abundance and survivorship of two Alveopora spp. in the mesophotic reefs of Eilat, Red Sea . Scientific Reports , 6 : 20964 . Google Scholar Crossref Search ADS PubMed WorldCat Eyal G. , Wiedenmann J., Grinblat M., D’Angelo C., Kramarsky-Winter E., Treibitz T., Ben-Zvi O., et al. 2015 . Spectral diversity and regulation of coral fluorescence in a mesophotic reef habitat in the Red Sea . PLoS One , 10 : e0128697 . Google Scholar Crossref Search ADS PubMed WorldCat Ferrari R. , Bryson M., Bridge T., Hustache J., Williams S. B., Byrne M., Figueira W. 2016 . Quantifying the response of structural complexity and community composition to environmental change in marine communities . Global Change Biology , 22 : 1965 – 1975 . Google Scholar Crossref Search ADS PubMed WorldCat Finney J. C. , Pettay D. T., Sampayo E. M., Warner M. E., Oxenford H. A., LaJeunesse T. C. 2010 . The relative significance of host–habitat, depth, and geography on the ecology, endemism, and speciation of coral endosymbionts in the genus Symbiodinium . Microbial Ecology , 60 : 250 – 263 . Google Scholar Crossref Search ADS PubMed WorldCat Fisher R. , Radford B. T., Knowlton N., Brainard R. E., Michaelis F. B., Caley M. J. 2011 . Global mismatch between research effort and conservation needs of tropical coral reefs . Conservation Letters , 4 : 64 – 72 . Google Scholar Crossref Search ADS WorldCat Fosså J. H. , Lindberg B., Christensen O., Lundälv T., Svellingen I., Mortensen P. B., Alvsvåg J. 2005 . Mapping of Lophelia reefs in Norway: experiences and survey methods. In Cold-Water Corals and Ecosystems , pp. 359 – 391 . Ed. by Freiwald A., Roberts J. M.. Springer Berlin Heidelberg , Berlin, Heidelberg . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Frade P. R. , Bongaerts P., Winkelhagen A. J. S., Tonk L., Bak R. P. M. 2008a . In situ photobiology of corals over large depth ranges: a multivariate analysis on the roles of environment, host, and algal symbiont . Limnology and Oceanography , 53 : 2711 – 2723 . Google Scholar Crossref Search ADS WorldCat Frade P. R. , De Jongh F., Vermeulen F., Van Bleijswijk J., Bak R. P. M. 2008b . Variation in symbiont distribution between closely related coral species over large depth ranges . Molecular Ecology , 17 : 691 – 703 . Google Scholar Crossref Search ADS WorldCat Frade P. R. , Englebert N., Faria J., Visser P. M., Bak R. P. M. 2008c . Distribution and photobiology of Symbiodinium types in different light environments for three colour morphs of the coral Madracis pharensis: is there more to it than total irradiance? Coral Reefs , 27 : 913 – 925 . Google Scholar Crossref Search ADS WorldCat Freiwald A. , Fosså J. H., Grehan A., Koslow T., Roberts J. M. 2004 . Cold-Water Coral Reefs . UNEP-WCMC , Cambridge, UK . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Freiwald A. , Roberts J. M. 2005 . Cold-Water Corals and Ecosystems: Preface . Springer , Heidelberg , VII – XII . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Fricke H. W. , Knauer B. 1986 . Diversity and spatial pattern of coral communities in the Red Sea upper twilight zone . Oecologia , 71 : 29 – 37 . Google Scholar Crossref Search ADS PubMed WorldCat Fricke H. W. , Meischner D. 1985 . Depth limits of Bermudan scleractinian corals: a submersible survey . Marine Biology , 88 : 175 – 187 . Google Scholar Crossref Search ADS WorldCat Garcia-Sais J. R. 2010 . Reef habitats and associated sessile-benthic and fish assemblages across a euphotic–mesophotic depth gradient in Isla Desecheo, Puerto Rico . Coral Reefs , 29 : 277 – 288 . Google Scholar Crossref Search ADS WorldCat Garcia-Sais J. R. , Castro R., Sabater-Clavell J., Carlo M., Esteves R. 2007 . Characterization of benthic habitats and associated reef communities at Bajo de Sico Seamount, Mona Passage, Puerto Rico. Report to the Caribbean Fishery Management Council. Goreau T. F. , Goreau N. I. 1973 . The ecology of Jamaican coral reefs. II. Geomorphology, zonation, and sedimentary phases . Bulletin of Marine Science , 23 : 400 – 464 . Google Scholar OpenURL Placeholder Text WorldCat Grigg R. W. 2005 . Depth limit for reef building corals in the Au’au Channel, S.E. Hawaii . Coral Reefs , 25 : 77 – 84 . Google Scholar Crossref Search ADS WorldCat Harmelin-Vivien M. L. , Laboute P. 1986 . Catastrophic impact of hurricanes on atoll outer reef slopes in the Tuamote (French Polynesia) . Coral Reefs , 5 : 55 – 62 . Google Scholar Crossref Search ADS WorldCat Harris P. T. , Bridge T. C. L., Beaman R. J., Webster J. M., Nichol S. L., Brooke B. P. 2012 . Submerged banks in the Great Barrier Reef, Australia, greatly increase available coral reef habitat . ICES Journal of Marine Science , 70 : 284 – 293 . Google Scholar Crossref Search ADS WorldCat Heyward A. , Fromont J., Schönberg C. H. L., Colquhoun J., Radford B., Gomez O. 2010 . The sponge gardens of Ningaloo Reef, Western Australia . The Open Marine Biology Journal , 4 : 3 – 11 . Google Scholar Crossref Search ADS WorldCat Hinderstein L. M. , Marr J. C. A., Martinez F. A., Dowgiallo M. J., Puglise K. A., Pyle R. L., Zawada D. G., et al. 2010 . Theme section on “Mesophotic Coral Ecosystems: Characterization, Ecology, and Management” . Coral Reefs , 29 : 247 – 251 . Google Scholar Crossref Search ADS WorldCat Hoegh-Guldberg O. , Mumby P. J., Hooten A. J., Steneck R. S., Greenfield P., Gomez E., Harvell C. D., et al. 2007 . Coral reefs under rapid climate change and ocean acidification . Science , 318 : 1737 – 1742 . Google Scholar Crossref Search ADS PubMed WorldCat Holstein D. M. , Smith T. B., Gyory J., Paris C. B. 2015 . Fertile fathoms: deep reproductive refugia for threatened shallow corals . Scientific Reports , 5 : 12407 . Google Scholar Crossref Search ADS PubMed WorldCat Hughes T. P. , Baird A. H., Bellwood D. R., Card M., Connolly S. R., Folke C., Grosberg R., et al. 2003 . Climate change, human impacts, and the resilience of coral reefs . Science , 301 : 929 – 933 . Google Scholar Crossref Search ADS PubMed WorldCat Kahng S. E. , Copus J. M., Wagner D. 2014 . Recent advances in the ecology of mesophotic coral ecosystems (MCEs) . Current Opinion in Environmental Sustainability , 7 : 72 – 81 . Google Scholar Crossref Search ADS WorldCat Kahng S. E. , Garcia-Sais J. R., Spalding H. L., Brokovich E., Wagner D., Weil E., Hinderstein L., et al. 2010 . Community ecology of mesophotic coral reef ecosystems . Coral Reefs , 29 : 255 – 275 . Google Scholar Crossref Search ADS WorldCat Kahng S. E. , Kelley C. D. 2007 . Vertical zonation of megabenthic taxa on a deep photosynthetic reef (50–140 m) in the Au’au Channel, Hawaii . Coral Reefs , 26 : 679 – 687 . Google Scholar Crossref Search ADS WorldCat Kahng S. E. , Wagner D., Lantz C., Vetter O., Gove J., Merrifield M. 2012 . Temperature related depth limits of warm-water corals. Proceedings of the 12th International Coral Reef Symposium, Cairns, Australia. Kane C. , Kosaki R. K., Wagner D. 2014 . High levels of mesophotic reef fish endemism in the Northwestern Hawaiian Islands . Bulletin of Marine Science , 90 : 693 – 703 . Google Scholar Crossref Search ADS WorldCat LaJeunesse T. C. , Thornhill D. J., Cox E. F., Stanton F. G., Fitt W. K., Schmidt G. W. 2004 . High diversity and host specificity observed among symbiotic dinoflagellates in reef coral communities from Hawaii . Coral Reefs , 23 : 596 – 603 . Google Scholar OpenURL Placeholder Text WorldCat Laverick, J. H., Andradi-Brown, D. A., Exton, D. A., Bongaerts, P., Bridge, T. C. L., Lesser, M. P., Pyle, R. L., et al. 2016. To what extent do mesophotic coral ecosystems and shallow reefs share species of conservation interest? Environmental Evidence, 5: 16. Lesser M. P. , Slattery M., Leichter J. J. 2009 . Ecology of mesophotic coral reefs . Journal of Experimental Marine Biology and Ecology , 375 : 1 – 8 . Google Scholar Crossref Search ADS WorldCat Lesser M. P. , Slattery M., Stat M., Ojimi M., Gates R. D., Grottoli A. 2010 . Photoacclimatization by the coral Montastraea cavernosa in the mesophotic zone: light, food, and genetics . Ecology , 91 : 990 – 1003 . Google Scholar Crossref Search ADS PubMed WorldCat Liddell D. , Ohlhorst S. L. 1988 . Hard Substrata Community Patterns, 1-120 M, North Jamaica . Palaios , 3 : 413 – 423 . Google Scholar Crossref Search ADS WorldCat Lindfield S. J. , Harvey E. S., Halford A. R., McIlwain J. L. 2015 . Mesophotic depths as refuge areas for fishery-targeted species on coral reefs . Coral Reefs , 35 : 125 – 137 . Google Scholar Crossref Search ADS WorldCat Locker S. D. , Armstrong R. A., Battista T. A., Rooney J. J., Sherman C., Zawada D. G. 2010 . Geomorphology of mesophotic coral ecosystems: current perspectives on morphology, distribution, and mapping strategies . Coral Reefs , 29 : 329 – 345 . Google Scholar Crossref Search ADS WorldCat Lombardi M. , Godfrey J. 2011 . In-Water Strategies for Scientific Diver-Based Examinations of the Vertical Mesophotic Coral Ecosystem (vMCE) from 50 to 150 meters. In Diving for Science 2011. Proceedings of the American Academy of Underwater Sciences 30th Symposium. Ed. by Pollock, N.W. AAUS, Dauphin Island, AL. Loya Y. , Eyal G., Treibitz T., Lesser M. P., Appeldoorn R. 2016 . Theme section on mesophotic coral ecosystems: advances in knowledge and future perspectives . Coral Reefs , 35 : 1 – 9 . Google Scholar Crossref Search ADS WorldCat Lucas M. Q. , Stat M., Smith M. C., Weil E., Schizas N. V. 2016 . Symbiodinium (internal transcribed spacer 2) diversity in the coral host Agaricia lamarcki (Cnidaria: Scleractinia) between shallow and mesophotic reefs in the Northern Caribbean (20–70 m) . Marine Ecology , 37 : 1079 – 1087 . Google Scholar Crossref Search ADS WorldCat Luck D. G. , Forsman Z. H., Toonen R. J., Leicht S. J., Kahng S. E. 2013 . Polyphyly and hidden species among Hawai'i's dominant mesophotic coral genera, Leptoseris and Pavona (Scleractinia: Agariciidae) . Peer Journal , 1 : e132 . Google Scholar Crossref Search ADS WorldCat Mass T. , Kline D. I., Roopin M., Veal C. J., Cohen S., Iluz D., Levy O. 2010 . The spectral quality of light is a key driver of photosynthesis and photoadaptation in Stylophora pistillata colonies from different depths in the Red Sea . The Journal of Experimental Biology , 213 : 4084 – 4091 . Google Scholar Crossref Search ADS PubMed WorldCat Muir P. , Wallace C., Bridge T. C., Bongaerts P. 2015 . Diverse staghorn coral fauna on the mesophotic reefs of north-east Australia . PLoS One , 10 : e0117933 . Google Scholar Crossref Search ADS PubMed WorldCat Muir P. R. , Wallace C. C. 2016 . A rare ‘deep-water’ coral assemblage in a shallow lagoon in Micronesia . Marine Biodiversity , 46 : 543 – 544 . Google Scholar Crossref Search ADS WorldCat Muscatine L. , Porter J. W., Kaplan I. R. 1989 . Resource partitioning by reef corals as determined from stable isotope composition . Marine Biology , 100 : 185 – 193 . Google Scholar Crossref Search ADS WorldCat Nemeth M. , Smith T. B., Blondeau J., Kadison E., Calnan J. M., Gass J. 2008 . Characterization of deep water reef communities within the marine conservation district, St. Thomas, US Virgin Islands. Report to the Caribbean Fisheries Management Council. Nir O. , Gruber D. F., Einbinder S., Kark S., Tchernov D. 2011 . Changes in scleractinian coral Seriatopora hystrix morphology and its endocellular Symbiodinium characteristics along a bathymetric gradient from shallow to mesophotic reef . Coral Reefs , 30 : 1089 – 1100 . Google Scholar Crossref Search ADS WorldCat Pickering C. , Byrne J. 2014 . The benefits of publishing systematic quantitative literature reviews for PhD candidates and other early-career researchers . Higher Education Research & Development , 33 : 534 – 548 . Google Scholar Crossref Search ADS WorldCat Pickering C. , Grignon J., Steven R., Guitart D., Byrne J. 2015 . Publishing not perishing: how research students transition from novice to knowledgeable using systematic quantitative literature reviews . Studies in Higher Education , 40 : 1756 – 1769 . Google Scholar Crossref Search ADS WorldCat Pizarro O. , Williams S. B., Jakuba M. V., Johnson-Roberson M., Mahon I., Bryson M., Steinberg D., et al. 2013 . Benthic monitoring with robotic platforms—the experience of Australia. Underwater Technology Symposium (UT), 2013 IEEE International: 1 – 10 . Pochon X. , Forsman Z. H., Spalding H. L., Padilla-Gamino J. L., Smith C. M., Gates R. D. 2015 . Depth specialization in mesophotic corals (Leptoseris spp.) and associated algal symbionts in Hawai'i . Royal Society Open Science , 2 : 140351 . Google Scholar Crossref Search ADS PubMed WorldCat Prasetia R. , Sinniger F., Harii S. 2016 . Gametogenesis and fecundity of Acropora tenella (Brook 1892) in a mesophotic coral ecosystem in Okinawa, Japan . Coral Reefs , 35 : 53 – 62 . Google Scholar Crossref Search ADS WorldCat Puglise K. A. , Hinderstein L. M., Marr J. C. A., Dowgiallo M. J., Martinez F. A. 2009 . Mesophotic coral ecosystems research strategy: international workshop to prioritize research and management needs for mesophotic coral ecosystems, Jupiter, Florida, 12–15 July 2008. NOAA Technical Memorandum NOS NCCOS 98 and OAR OER 2. Pyle R. L. , Boland R., Bolick H., Bowen B. W., Bradley C. J., Kane C., Kosaki R. K., et al. 2016 . A comprehensive investigation of mesophotic coral ecosystems in the Hawaiian Archipelago . Peer Journal , 4 : e2475 . Google Scholar Crossref Search ADS WorldCat R Core Team . 2010 . R: A Language and Environment for Statistical Computing . R Core Team , Vienna, Austria . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Rees M. , Heyward A., Cappo M., Speare P., Smith L. 2004 . Ningaloo Marine Park—Initial Survey of Seabed Biodiversity in Intermediate and Deeper Waters. Australian Institute of Marine Science , Crawley, Australia . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Reich H. G. , Robertson D. L., Goodbody-Gringley G. 2017 . Do the shuffle: changes in Symbiodinium consortia throughout juvenile coral development . PLoS One , 12 : e0171768 . Google Scholar Crossref Search ADS PubMed WorldCat Roberts J. M. , Wheeler A. J., Freiwald A. 2006 . Reefs of the deep: the biology and geology of cold-water coral ecosystems . Science , 312 : 543 – 547 . Google Scholar Crossref Search ADS PubMed WorldCat Roberts J. M. , Wheeler A. J., Freiwald A., Cairns S. D. 2009 . Cold-water Corals: The Biology and Geology of Deep-sea Coral Habitats . Cambridge University Press , Cambridge , 334 p . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Rosa M. R. , Alves A. C., Medeiros D. V., Coni E. O. C., Ferreira C. M., Ferreira B. P., de Souza Rosa R., et al. 2015 . Mesophotic reef fish assemblages of the remote St. Peter and St. Paul’s Archipelago, Mid-Atlantic Ridge, Brazil . Coral Reefs , 35 : 113 – 123 . Google Scholar Crossref Search ADS WorldCat Serrano X. M. , Baums I. B., Smith T. B., Jones R. J., Shearer T. L., Baker A. C. 2016 . Long distance dispersal and vertical gene flow in the Caribbean brooding coral Porites astreoides . Scientific Reports , 6 : 21619 . Google Scholar Crossref Search ADS PubMed WorldCat Sheppard C. 1982 . Coral populations on reef slopes and their major controls . Marine Ecology Progress Series , 7 : 83 – 115 . Google Scholar Crossref Search ADS WorldCat Sherman C. , Nemeth M., Ruíz H., Bejarano I., Appeldoorn R., Pagán F., Schärer M., et al. 2010 . Geomorphology and benthic cover of mesophotic coral ecosystems of the upper insular slope of southwest Puerto Rico . Coral Reefs , 29 : 347 – 360 . Google Scholar Crossref Search ADS WorldCat Singh H. , Armstrong R. A., Gilbes F., Eustice R., Roman C., Pizarro O., Torres J. 2004 . Imaging coral I: imaging coral habitats with the SeaBED AUV . Subsurface Sensing Technologies and Applications , 5 : 25 – 42 . Google Scholar Crossref Search ADS WorldCat Sinniger F. , Morita M., Harii S. 2012 . "Locally extinct" coral species Seriatopora hystrix found at upper mesophotic depths in Okinawa . Coral Reefs , 32 : 153 – 153 . Google Scholar Crossref Search ADS WorldCat Slattery M. , Lesser M. P., Brazeau D., Stokes M. D., Leichter J. J. 2011 . Connectivity and stability of mesophotic coral reefs . Journal of Experimental Marine Biology and Ecology , 408 : 32 – 41 . Google Scholar Crossref Search ADS WorldCat Smith T. B. , Gyory J., Brandt M. E., Miller W. J., Jossart J., Nemeth R. S. 2015 . Caribbean mesophotic coral ecosystems are unlikely climate change refugia . Global Change Biology , 22 : 2756 – 2765 . Google Scholar Crossref Search ADS WorldCat Tenggardjaja K. A. , Bernardi G., Bowen B. W. 2014 . Vertical and Horizontal Genetic Connectivity in Chromis verater, an Endemic Damselfish Found on Shallow and Mesophotic Reefs in the Hawaiian Archipelago and Adjacent Johnston Atoll . Figshare. PLoS One . 9 : e1154963 . Google Scholar Crossref Search ADS WorldCat Thresher R. E. , Colin P. L. 1986 . Trophic structure, diversity and abundance of fishes of the deep reef (30-300 m) at Enewetak, Marshall Islands . Bulletin of Marine Science , 38 : 253 – 272 . Google Scholar OpenURL Placeholder Text WorldCat van Oppen M. J. , Bongaerts P., Underwood J. N., Peplow L. M., Cooper T. F. 2011 . The role of deep reefs in shallow reef recovery: an assessment of vertical connectivity in a brooding coral from west and east Australia . Molecular Ecology , 20 : 1647 – 1660 . Google Scholar Crossref Search ADS PubMed WorldCat Wagner D. , Kosaki R. K., Spalding H. L., Whitton R. K., Pyle R. L., Sherwood A. R., Tsuda R. T., et al. 2014 . Mesophotic surveys of the flora and fauna at Johnston Atoll, Central Pacific Ocean . Marine Biodiversity Records , 7 : e68 . Google Scholar Crossref Search ADS WorldCat White K. N. , Ohara T., Fujii T., Kawamura I., Mizuyama M., Montenegro J., Shikiba H., et al. 2013 . Typhoon damage on a shallow mesophotic reef in Okinawa, Japan . Peer Journal , 1 : e151 . Google Scholar Crossref Search ADS WorldCat Wickham H. 2009 . ggplot2: Elegant Graphics for Data Analysis . Springer-Verlag , New York . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Williams S. B. , Pizarro O., Jakuba M. V., Johnson C. R., Barrett N., Babcock R., Kendrick G. A., et al. 2012 . Monitoring of benthic reference sites using an autonomous underwater vehicle . Robotics & Automation Magazine , 19 : 73 – 84 . Google Scholar Crossref Search ADS WorldCat Ziegler M. , Roder C. M., Buchel C., Voolstra C. R. 2015 . Mesophotic coral depth acclimatization is a function of host-specific symbiont physiology . Frontiers in Marine Science , 2 : 4 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: [email protected]

Gerlotto,, François

doi: 10.1093/icesjms/fsx086pmid: N/A

Abstract Fish stock assessments based solely on energy flow through the ecosystem are not good predictors of population dynamics. To accurately forecast the response of populations within one or more ecological regimes, consideration must be given to non-trophic mechanisms allowing interactions inside the system, and fish behaviour in response to changes in their habitats. The example of the grey triggerfish (Balistes capriscus) in West Africa shows that fisheries biology is unable to model satisfactorily the life of a fish population. The Ecosystem Approach to Fisheries improves the models but does not overcome this fundamental limitation. Data from direct observations of fish biology and behaviour must be added to the catch and environmental data to help to design energetic-cybernetic models in order to anticipate non-linear and chaotic dynamics. This requires adding data collected by fishers (e.g. underwater acoustics) to scientific data bases, conceiving environmental indicators (e.g. habitat), and using scenarios to anticipate the reactions of populations to regime shifts. It also requires a good understanding of the population structures and strategies. We developed the concept of “pelagic metapopulation” which, through comparative analysis with the jack mackerel (Trachurus murphyi), allowed us to propose a hypothesis explaining the history of the grey triggerfish population. Introduction Naïve question from the audience: if fisheries biology did not exist, would the present status of the world’s fisheries be different? Most likely the answer is, “not substantially”. We must take two points into consideration: (i) as stated by Zwolinski and Demer (2012) “It is widely recognized that many fish stocks worldwide have collapsed because of overexploitation, [and also] perhaps because of cyclical environmental factors, anthropogenic factors, or both” (Pauly et al., 2003; Myers and Worms, 2005a; Hilborn et al., 2005; Coll et al., 2008; Hilborn, 2011); (ii) almost all the large pelagic stocks worldwide have suffered big changes (collapses and/or recoveries) at least once in their history (Hutchings and Reynolds, 2004), which were not foreseen by the assessment experts (Myers and Worms, 2005b; Worm et al., 2009). The story that I present here is the result of a long series of personal experiences which slowly opened my mind to these questions and allowed me to formalize some hypotheses that I tested, with both successes and failures. Analysing such a long process accumulated during one’s career may help to understand how hypotheses, which cannot be considered as spontaneously created by the mind, arise: personal history plays an important role in their conception. This was my [Lesson 1]: if science may be pure logic, research is more likely a craftsman’s trade. It essentially works through empirical feed-backs between what the scientist knows and what he/she experiments on, observes and measures. Back to the beginning As a student I attended the University of Paris-Sud. My first experience in fisheries came from a student job on side trawlers in Brittany—several 2-week fishing trips collecting data. At that point of my career, fishers taught me a series of important lessons. [Lesson 2]: data quality depends on many things, not necessarily linked to the fishery itself (seasickness being one of them!); [Lesson 3]: catch data alone do not adequately represent what happens during a fishing operation; [Lesson 4]: being on the deck every 4 h during 15 days of landing, sorting, discarding, cleaning, and storing fish is a unique experience for learning what a fishery is all about; [Lesson 5]: fishers, who are at sea most of the time, are an extraordinary albeit—until recently—largely underexploited source of knowledge (Hind, 2015) and data (Karp, 2007; Melvin et al., 2016a,b; Stephenson et al., 2016); [Lesson 6]: fish behaviour represents a major issue in fisheries research, but at this time was completely ignored by the assessment studies. When I was recruited in 1973 by ORSTOM (Office de la Recherche Scientifique et Technique Outre-Mer, then IRD—Institut de Recherche pour le Développement—since 2001: www.ird.fr), my background was of a zoologist with some limited skills in fish population dynamics and marine ecology. In November, 1973, ORSTOM took me to Ivory Coast to a work part-time in a project studying the fisheries resources and exploitation of the large lagoons surrounding Abidjan; and part-time under Emile Marchal’s direction to evaluate the marine stocks along the West African coast with a new technique: fisheries acoustics (Forbes and Nakken, 1972; Stéquert and Gerlotto, 1977). The priority in the 1970s in West Africa was “exploration”, for two reasons. First, in the 1960s, the French colonial empire was being dismantled. African countries became independent and began to evaluate their own economically important assets, including their marine resources. Second, the idea of creating “Exclusive Economic Zones” (formalized in 1982 with the UN Convention on the Law of the Sea) arose, and evaluating these still largely underexploited, if not completely unknown, offshore resources allowed these young countries to negotiate treaties to either sell fishing rights or develop their own fisheries. Once in Abidjan, I had the unique opportunity to undertake pioneer activities in the Ebrié lagoon, as it was the first time fisheries research was conducted there, although a few studies on natural history and ecology had been performed some years or even decades earlier (Monod, 1950; Fernandes et al., 1951; Daget and Durand, 1968). Ebrié lagoon is 150 km long and 10–20 km wide, with a yearly catch around 10 000 tonnes (Gerlotto et al., 1976a, b). The only other important research in the lagoon at this time was Serge Garcia’s on Penaeus duorarum (Garcia, 1977). The project allowed me to practice, on a small scale, all the steps usually performed separately by big teams in fisheries research: counting canoes and fishing gear; preparing a catalogue of fishing methods and techniques; evaluating the potential fishing effort; embarking with fishers; defining the catch per unit effort (CPUE) and the fishing seasons; studying the commercialization systems (Gerlotto et al., 1976a,b). I established a network of data collectors, prepared data bases for statistics, and data processing. I also had to obtain the biological information needed for fisheries research: ecological stratification in the lagoon, and growth curves, fecundity, distribution, spawning periods, spawning grounds, migrations and behaviour for the most important species. I eventually worked up the data and published the results (Albaret and Gerlotto, 1976; Gerlotto, 1976, 1979; Durand et al., 1978; Gerlotto and Stéquert, 1978). This rather short (1973–1978) but exhaustive experience allowed me to discover in detail all of the elements collected and calculated for an assessment analysis of a stock. I received there my second series of lessons. [Lesson 7] taught me that even sound decisions could have unexpected ecological consequences. The lagoon is a complex ecosystem within which the fishery is one actor among others, and changing one single compartment of this system was likely to induce a cascade of many other unexpected (and often not desirable) changes. Let me describe a personal (unpublished) observation. Fishers around the lagoon were also farmers. Reducing fishing effort for conservation reasons implied that they had to increase their effort in agriculture to maintain their activity, which they did by burning the forest to get more cultivable lands, with negative impact on the terrestrial ecosystem. Here, a conservative decision for the lagoon led to an ecological concern in a completely different area. One thing is to define the recommendation of a model in your office; another one is to measure its actual effects on the field. [Lesson 8] was that you should not accept the “obvious” if you don’t check it by yourself. I discovered that the artisanal fishery, usually considered a small and traditional activity in tiny villages, was quite different from this “obvious” view in West Africa. On the contrary, it was probably the most effective fishing activity, even more international, modern and profitable than the national fishing industry in Ivory Coast at this time (Gerlotto et al., 1980). Fishing gear and outboard engines were imported from Japan, huge dugout canoes were purchased in Ghana, some of the fishers used to come seasonally from Benin, part of the fishery migrated seasonally to Senegal, etc. (Gerlotto et al., 1979, 1980). Part of the catch (frozen shrimp) was sold in Europe and the smoked fish was exported over long distances in Africa, from Abidjan up to Ouagadougou or Niamey, i.e. thousands of km from the fishing grounds. The fish market was probably as important (in volume and costs), international, reactive, complex, speculative, and sophisticated as any other in the world. That made me extremely reluctant to accept the “obvious” idea that artisanal fishery is always the “good guy” and industrial fishery the “bad guy”. [Lesson 9] was that data are not “the truth”, but are built according to preliminary, often implicit, hypotheses. Just think about this: by convention, a fishing activity is almost always inferred by the catch. This means that anything else is ignored (fishing conditions, market, weather, fishers’ experience, fish behaviour, competition, predation, fish learning, precision of measurements, vessel noise, hydrological conditions, multi-specific structure of the catch, etc.). The “catch-data-only” approach implicitly assumes that fishing activity can be correctly described through the local abundance of fish and the fishing effort, which can be measured using the weight of the catch. Why the weight and not any other information? Most likely because catches in weight were already collected by the fisheries administrations (for their own needs) and the scientist just took the data and continued the series, following the rule “if it works, don’t fix it”. I will detail later what strong hypothesis is hidden behind the catch data. This is often forgotten in a world where, thanks to the internet, databases become independent from their sources. Forgetting this link may produce silly results; data become absolute truth, and this can be disastrous. A good example is given by Pauly (2016) on how such faith in international catch data bases led to incorrect interpretation of the history of global fisheries. My other activity was in fisheries acoustics. ORSTOM received a research vessel in June 1972, the RV “Capricorne”, equipped with SIMRAD analogue instruments: an EKS echo sounder and a QM echo integrator. After a series of preliminary surveys from Congo to Senegal, my first usable results were obtained in July 1974 on the pelagic distribution of micronekton density inside the mid-Atlantic “Deep Scattering Layer” (Gerlotto, 1975; Marchal et al., 1993). It was a revelation. I could provide relative abundance indices and dynamic distribution of the biomass in relation to 3-D hydrodynamic characteristics of the ocean environment. Fisheries acoustics was opening large perspectives, not yet explored. Later, from 1976 to 1980, we were able to obtain biomass estimates in Senegal (Gerlotto et al., 1976a,b), Guinea Bissau (Stéquert et al., 1977), Ivory Coast (Marchal and Picault, 1977), and Guinea (Marchal et al., 1979, 1980). The mysterious history of the grey triggerfish in West Africa Being fortuitously in the right place at the right time I was among the first scientists to explore these ecosystems using modern methods and techniques, and to make a series of (modest) discoveries, among which, one in particular influenced the trajectory of my scientific career: the expansion of the grey triggerfish (Balistes capriscus) population of West Africa during the 1970s. Before 1972, the major stock on the Ghanaian-Ivorian continental shelf was the round sardine Sardinella aurita. It was principally exploited by artisanal fisheries using big dugout canoes with large purse seines (Gerlotto and Stéquert, 1978; Gerlotto et al. 1979a,b). The average catch was around 25 000 tonnes per year (mean 1963–1971). In 1972, the catch increased substantially and reached 72 000 tonnes (close to 95 000 tonnes when including the small industrial fishery operating in the area). Then it dropped dramatically to 4700 tonnes in 1973 and 1400 tonnes in 1974, and for a few years it remained at this order of magnitude. In parallel, the formerly scattered and insignificant population of grey triggerfish increased enormously after 1972, expanding eventually to the entire West African shelf from Nigeria to Senegal. We estimated its biomass peaked at around one million tonnes, making this species the most abundant in West Africa (Caverivière et al., 1981; Caverivière, 1982; Binet et al., 1991). This expansion took approximately 5–6 years and several generations of fish: we captured schools of juvenile triggerfish in 1978 in Guinea (Marchal et al., 1979, 1980), which were too young to be born elsewhere. Moreover, the biology of this fish changed. The grey triggerfish is usually known as a solitary sedentary demersal fish, displaying territorial behaviour around its nest during the spawning season (Simmons and Szedlmayer, 2012). After the explosion, the fish became pelagic, gregarious, migratory (Caverivière et al., 1980, 1981; Caverivière, 1982), living in big schools: the biggest catch we performed on a single school weighted 27 tonnes (Robertson, 1977). Then, the stock dropped off and although not exploited except in Ghana, triggerfish abundance returned to its original level, i.e. sporadic distribution of a few individuals all along the shelf. Since the 1990s, the global situation has been as it was before 1972 (Figure 1). Figure 1. Open in new tabDownload slide Catches of the Ghanaian artisanal fishery (from Aggrey-Fynn, 2007; Binet et al., 1991; Caverivière, 1991, for B. capriscus data before 1972). Figure 1. Open in new tabDownload slide Catches of the Ghanaian artisanal fishery (from Aggrey-Fynn, 2007; Binet et al., 1991; Caverivière, 1991, for B. capriscus data before 1972). How did scientists explain this story? A dedicated international workshop was held in 1976 (Ansa-Emmin and Marcille, 1976). It concluded that the round sardine would likely take decades to recover from such a collapse, because the stock suffered a combination of strong overfishing, occupation of its niche by B. capriscus, and changes in the ecosystem (Binet, 1982; Binet et al., 1991). Actually, the recovery was much faster; from 1978 to 2000, the average catch was around 59 000 tonnes, i.e. twofold higher than before 1972 (Binet et al., 1991). Obviously the conclusions were not accurate. In 1991, the synthesis of a series of workshops in West Africa was published (Cury and Roy, 1991), which included new keywords: instability, variability, and environmental changes. Climatic or environmental variables began to be included in the models (Fréon, 1989; Cury and Roy, 1989). This effort eventually led to the “Ecosystem Approach to Fisheries” (Garcia, 1996). This new approach was particularly welcome in tropical ecosystems which presented a high sensitivity to climatic changes, strong behavioural reactions of fish to external stimuli, a rather low pressure (at that time) from fisheries, displaying clearly the impact of environment on stock dynamics. But even this series of improvements did not explain the variations in triggerfish biomass and behaviour; changes did not have anything to do with human activities and environment could explain the beginning of the story, but not the full story. Why? Such questions oriented assessment scientists’ activities towards more sophisticated models, adding environmental hypotheses and metrics, but always using mostly fisheries data; they oriented mine towards a rather different domain: fish behavioural ecology. Contributions to fisheries acoustics and fish biology My activities took two different but complementary routes. The first was to develop tools and methods able to provide good and abundant pieces of information. The second was to build some conceptual models based on these biological and behavioural questions. But this kind of activity cannot be performed alone, and there was a need to discuss the results, the questions and the hypotheses with the international community. From 1986 to 1991 my colleagues Fréon, Soria and myself developed an ORSTOM project (EICHOANT) devoted to studying the effect of fish behaviour on fish stock assessment, by way of experiments and surveys in Martinique (FWI) and Venezuela. We submitted the important series of results obtained between 1985 and 1990 (Anonymous, 1990a,b) to the international community and became active members of the ICES groups, and more specifically the ICES Working Group on Fisheries Acoustics Science and Technology (FAST), which was (and still is) the most important world forum in this discipline. Belonging to this community was especially critical for us who were somehow outsiders. First of all, at this time, the English language was not considered indispensable for doing good science in France, making the contacts (and the literature) limited. Second, our work area and major objectives were apparently different from those of the “northern” community. Anderson (2015) describes ICES very clearly and how critical its role is for the fisheries research community. After 10 years of membership in the FAST, I received from Peter Stewart, the then chair of the Fisheries Technology Committee (FTC), the parent Committee of FAST, the following suggestion: “François, in life there is a time to be young and a time to be old; a time to be a son and a time to be a father; a time to be a student and a time to be a teacher; it is time for you now to consider being chairman of FAST”. I complied, applied for the job and was elected the fourth chair (1997–2000) following Kjell Olsen, Jim Traynor and John Simmonds in that capacity (Fernandes et al., 2002). Fréon and Misund (1999) summarized our research results in a book which is still a reference. I later was elected chair of the Fisheries Technology Committee (2005–2007), which gave us even greater access to the ICES forum where our hypotheses could be discussed. I included this story because it is linked with [Lesson 10]: full integration in the international community is essential in a scientist’s life, especially for those who are not from English-speaking countries. Most of my career was in Africa (Ivory Coast, Senegal), the Caribbean (Martinique, Cuba, Venezuela) and South America (Chile, Peru). I discovered that one of the strongest limitations in these countries, as far as research was concerned, was the timidity of their young scientists, handicapped by their poor practice of English. Scientists from English-speaking countries can hardly realize how difficult it is to belong to the international community when one is not fluent in English, and how thoughts expressed in one’s mother language, whatever their quality, lose their strength when they have to be translated into (poor) English. Institutions like ICES, with their long practice of international exchanges, play a critical role, and I strongly recommend that these “non-English-speaking” scientists should become involved in such groups. Echo-integration was progressing quickly. A history of fisheries acoustics in ICES was published by Fernandes et al. (2002), who related all of this collective adventure. In the 1990s, technique was no longer a real issue: evidently, many improvements were still needed, but because the electronics, computer facilities, and theoretical concepts already existed and were rapidly progressing, their development was just a matter of time and money. Methodology was another story. It still presented theoretical and fundamental limitations and prevented acoustic results from describing the stock structure and absolute abundance estimates with the degree of accuracy required by assessment models. In particular, two types of scientific studies were needed to which I could contribute: statistics and fish behaviour. Statistical meaning of acoustic data Before the 1980s, statistical methods applied to echo-integration results were very poor. Apart from the works of Bazigos (1975, 1981) and Shotton and Bazigos (1984), no reference work had been published on this topic. Scientists simply applied conventional statistics, violating various conditions for their application. Acoustic samplings present a series of characteristics that theoretically forbid the application of standard variance estimates. Samplings in transects are systematic, auto-correlated, continuous in time and space, and much more detailed in the direction of transects than perpendicular to them (anisotropy). Under these conditions, standard confidence intervals have no meaning (Aglen, 1983; Gerlotto and Stequert, 1983). Some studies intended to find solutions (Johannesson and Mitson, 1983) but the result was often worse, as these solutions actually added more violations to the list, such as post-stratification using the data themselves. Jolly and Hampton (1991) published the only effective method that introduced stochasticity in the data by determining randomly the starting point of the survey and the inter-transect distances. They produced more acceptable variance measurements. But these methods, and especially the random inter-transect distance, did not improve the estimate of the biomass value, as wide inter-transect spaces may become unexplored (Fernandes et al., 2002). Knowing that distribution laws of fish concentrations are highly asymmetric (Fréon et al., 1991; Mullon and Pichon, 1991), the risk of missing during a single survey the small spaces where the bulk of the biomass was concentrated was higher. The cost of statistical acceptability was an increased risk of bias in the abundance estimate. FAST formed a group of scientists to work on this question and publish a Cooperative Research Report (Simmonds et al., 1992) which was probably the first document describing the state-of-the-art and recommendations on acoustic sampling methods. In the early 1980s, Marchal suggested the use of geostatistics in fisheries acoustics. A first paper introducing this method was presented (in French) at the ICES/FAO Symposium on Fisheries Acoustics held in Bergen, Norway (Norway, 1982; Laloë, 1985), but remained completely ignored. Then, we submitted some very preliminary results at the FAST (Gerlotto and Marchal, 1985; Marchal and Gerlotto, 1985), but again with very poor success. The fact that we had a very low level of English, that geostatistics was a French concept developed by the École des Mines de Paris (Matheron, 1970), that the acoustic community was not really interested in statistics, and that we were not geostatisticians, did not help our case. We had to write stronger papers in better English (Gerlotto and Marchal, 1987; Anonymous, 1990a, Petitgas, 1990; Gerlotto, 1993), and to plead almost 10 years before to becoming convincing. Incidentally, I learned [Lesson 11] that could be named the “10 year law”, i.e. the usual time needed for a new concept to be accepted by society. As far as science is concerned, we could define this law as: “Between the moment a new (and potentially fruitful) hypothesis is proposed and the moment it has impact on research, whatever the supporting evidence, a precautionary delay of ten years is implicitly applied by the scientific community”. Although this is highly frustrating when we know that our hypothesis is good because it is supported by serious scientific work, we could consider this delay as a precautionary adaptation of human society; there is need for a 10-year delay to be sure that a new rule is pertinent. How many apparently strong research studies and hypotheses did not survive 10 years, just because they were fundamentally weak or wrong, despite their convincing descriptions (Davenas et al., 1988)? After this long lobbying, ICES organized a workshop in Reykjavik in 1990 which analysed the use of geostatistics and concluded that this method was likely to resolve most of the statistical problems of acoustic surveys (Anonymous, 1990b). Later, a reference book was published by Rivoirard et al. (2008). Nowadays, confidence intervals are meaningful and geostatistics has become a standard method for statistical analysis of acoustic surveys. Exploring school behaviour: the multibeam systems The second source of uncertainty in fisheries acoustics was the existence of biases at many levels, as was detailed early (Simmonds et al., 1992) and confirmed many times (see Fréon and Misund, 1999, for a summary). The most important among them was fish avoidance, first observed by Olsen (1969, 1990). During the 1980s, the capacity to observe schools was very limited. Schools were only recorded as black spots on the echogram. Due to this weak knowledge, contradictory results existed, e.g. between density evaluations by visual methods (counting on photographic or video recordings) and by acoustics (Simmonds et al., 1992). Acoustics provided density values orders of magnitude below visual counting, and contradicted the universal visual observation of fish organization inside a school (individuals separated by around one to three body-length distances). I obtained the answer to this contradiction through an unexpected personal observation of small schools in Martinique. I was working on this French island at the time and used to spend week-ends with my family on a small beach called Grande-Anse. I once had the opportunity to swim during a couple of hours in shallow waters (4 m depth) over a small school of Harengula sp., practically flat and observable in two dimensions. I could see that the school structure was highly heterogeneous, fish being effectively separated by around one body-length to each other, but schools presenting also large empty areas very similar to vacuoles in a cell. Here was the reason for the contradiction. Photographic or video observations usually do not record these vacuoles, while acoustics average them with the dense parts of the school (Fréon et al., 1992). That was [Lesson 12], confirming what many naturalists and ecologist used to say. A discovery is not only obtained through the use of sophisticated experiments; it can come from any observation and at any moment as long as you know what you are looking for and you are prepared to receive it (Fabre, 1924). A discovery is almost always due to the particular capacity of the brain to make analogies and this can happen at any moment and for any reason. Always being open to inputs from the external world is essential in research. Anyway, this observation gave us a hypothesis on school structure that we could test once digital echo-integrators became able to process the acoustic signal received from inside the school. In the late 1980s, we produced the very first acoustic cross section of a school that confirmed its high heterogeneity, showing nuclei and vacuoles (Fréon et al., 1992). Schools appeared to be more complex organizations than expected. At the same time, a pioneering study on school avoidance using long range omnidirectional sonar was presented by Diner and Masse (1987), who described and measured the avoidance behaviour of schools in front of a vessel. Following this first experiment, we developed a series of studies (Gerlotto and Fréon 1988; Fréon and Misund 1999 for a summary). We concluded that (i) avoidance was a major source of bias for a vertical echosounding survey, and (ii) multibeam systems were likely the only tool able to evaluate it (Gerlotto et al., 1999). I suggested adapting one of the new multibeam echosounders (MBE) already developed for bathymetry to fisheries research, and we submitted the project AVITIS to the EU with the objective of building an appropriate instrument (Anonymous, 2000). The project was approved and the Reson 6012 SEABAT Multibeam Echo Sounder became the first one available for fisheries research in the early 2000s (Gerlotto et al., 2000). We first used MBEs to measure school avoidance. The results showed that biases due to avoidance were much higher than expected; in some cases, up to 80% of the schools were avoiding the survey vessel (Soria et al., 1996). But this avoidance reaction also appeared to be extremely variable (Brehmer et al., 2000; Gerlotto et al., 2004), and in some cases no avoidance was observed (Figure 2). Modeling avoidance (Soria et al., 2003) and, thereby, predicting it during a given survey, remains intractable due to the huge number of factors driving avoidance. Figure 2. Open in new tabDownload slide Use of multibeam echo sounder for evaluating the school avoidance. Top, left: description of the method; down, left: results from a survey in the Catalan Sea. All the schools recorded are represented by a dot located at the co-ordinates of their gravity centre. Scales in m. Right: histograms of school numbers related to their distances from the vessel (in m), for Senegal, Ivory Coast, Chile, compared to the Catalan Sea (results from left, down). All regions display a similar avoidance pattern except Chile (no avoidance). Figure 2. Open in new tabDownload slide Use of multibeam echo sounder for evaluating the school avoidance. Top, left: description of the method; down, left: results from a survey in the Catalan Sea. All the schools recorded are represented by a dot located at the co-ordinates of their gravity centre. Scales in m. Right: histograms of school numbers related to their distances from the vessel (in m), for Senegal, Ivory Coast, Chile, compared to the Catalan Sea (results from left, down). All regions display a similar avoidance pattern except Chile (no avoidance). The series of experiments we performed gave me my [Lesson 13]: beyond specific results e.g. avoidance, schooling behaviour was a key to understanding fisheries biology. For instance, contrary to the standard hypothesis in stock assessment, each individual fish may display a different catchability pattern, as they are able to learn and present individual adaptive reactions to exploitation and particular fishing pressure (Pyanov, 1993). Soria (1994) showed in tank experiments in Martinique that trained fish from one school were even able to transmit the learned knowledge to another school. In situ studies confirmed this fact. Heavily exploited populations of S. aurita in West Africa exhibited significantly higher avoidance reactions to the same trawler (RV “Antea”) than populations in non-exploited Venezuelan waters (Brehmer et al. 2000). With the use of MBEs we could better understand what schools were and how they behaved (Gerlotto and Paramo, 2003; Gerlotto et al., 2004, 2006; Soria et al., 2007,; Gerlotto et al., 2010). Collectively, these results allowed us to build a conceptual model, presented in Figure 3 (Bertrand et al., 2008). Figure 3. Open in new tabDownload slide A conceptual model describing the relative importance of factors regulating aggregation of gregarious fish as a function of scale. There are two y-axis in relative units, one based on self-organization, the other on environmental forcing (from Bertrand et al., 2008). Figure 3. Open in new tabDownload slide A conceptual model describing the relative importance of factors regulating aggregation of gregarious fish as a function of scale. There are two y-axis in relative units, one based on self-organization, the other on environmental forcing (from Bertrand et al., 2008). This model shows that schools balance two motivations: maintaining a stable social structure; and exploiting a favourable habitat. We also see that if large organizations can be explained mostly by trophic patterns, things are quite different at small scales. We can now define a school as the smallest collective and coherent structure that allows coordinated reactions to changes in fish physiology and environment. Towards a different approach of fisheries biology In the mid-1990s, the knowledge of pelagic fish behavioural ecology was already sufficient to deliver useful information to the assessment people. But the bad surprise was that, except biomass estimates, our results were never used for assessment purposes. Why? My analysis was the following. Ecology involves two processes: (i) energetic and (ii) cybernetic. Energetics is the dissipation of (mostly solar) energy through successive biochemical steps. Solar energy is stored and transported by metabolizing complex biochemical molecules through photosynthesis (phytoplankton). These high energy level molecules concentrated in phytoplankton organisms are then slowly catabolized through a succession of predators from zooplankton to apex predators, ending with bacteria which release low energy level molecules (CO2, H2O…) that are recycled, etc. Cybernetics (Frontier et al., 2008) refers to mechanisms such as fish behaviour, spatial organization, interactions with other species, which exist in an ecosystem but do not belong to the biochemical metabolism process. Their function is to allow the (intra- and inter-specific) individuals in the ecosystem to interact with each other. Although they do not feed the ecosystem, they sustain it. Conventional fisheries biology mainly considers energetics and not cybernetics. Consequently, the data used are related to energy flows. As I said above, there are hypotheses behind data. The one behind catch data is that these data represent the energy produced by the stock. They synthesize the thermodynamic exchanges and are assumed to describe exhaustively the relationships between the stock and the ecosystem. Most models are based on food chains and productivity patterns and can only work using energy-flow data. On the contrary, direct observations, and especially acoustic data, mostly consider the interactions inside the ecosystem: organizations, adaptations, spatial structures, relationships between compartments, etc. The only exploitable information that conventional assessment can receive from acoustics is energetic: the biomass. However, fish catches or abundance estimates are alone unable to describe the interactions between the exploited population and its habitat, as the history of B. capriscus told us. In this context, we decided to introduce non trophic and behavioural indicators into models, for two reasons. First we hypothesized that behaviour allows the fish to adapt to their environment (Figure 3), affecting population dynamics. And second, we highlighted the idea that forecasting capabilities are not the privilege of humans. Marchetti (1998) claimed that “Every living thing has or is a machinery for learning, remembering, and forecasting. The objective is to provide anticipatory reactions to the interactions with the external world” (Marchetti, 1998). If behavioural patterns are the adaptive answer to variations in environmental parameters, then they are integrative and may display easy-to-measure characteristics synthesizing reactions to a number of non-measured or non-measurable environmental variables. These analyses were shared with a few colleagues inside IRD and IFREMER in France (Massé and Gerlotto, 2003) but did not evoke any positive reaction in our Institutions in the 1990s. The objective of assembling a multidisciplinary study on stock assessment and fish behaviour was considered either useless or impossible, and the projects that I submitted to IRD or to the European Commission were rejected. Besides entering into the endless debate between “reductionist vs. holistic” concepts, we simply faced the technical difficulties of crossing borders of scientific disciplines. This was [Lesson 14]: research institutes are organized vertically by disciplines. Therefore, they don’t have the administrative instruments to deal with horizontal multidisciplinary projects: where and how in the institution structure can one evaluate their proposals, methodologies, budgets, results, publications, even the scientists themselves? Things evolved in the early 2000s when climate change became a real issue and required multidisciplinary projects. In 2001, with a new generation of scientists interested in this analysis, I was able to create a research unit inside IRD (ACTIVE, see www.ird.fr) gathering scientists from France, Belgium, Chile, Peru, Hawaii, and La Réunion Island to study fish behaviour as an indicator of how climate, hydrology and the whole ecosystem, including fisheries, impact fish populations. Once this group was formed, we began to consider the situation of “fisheries ecology”. We discarded the “mostly deterministic” vision of dynamics of populations, knowing that fish populations often respond non-linearly to physical forcing (Hilborn et al., 1994; Hsieh et al 2005). With this criterion in mind, the critical question of the predictive capacity of assessment models became an issue. We had to find a way to overcome this oxymoron: providing some anticipatory recommendations to managers on a system characterized by non-linear or chaotic dynamics (in the sense defined from Poincaré’s works: small changes in the starting conditions produce outcomes of great magnitude, making the system unpredictable, e.g. Boccaro, 2010), which prevent predictions over the long term. Marchetti (1998) showed that resolving this contradiction is a major output of natural selection: “DNA is an active memory that learns through hypothesis (mutation in a broad sense) and experiment (survival value of the mutated offspring). (…) The only link to the external world can be the survival feedback”. Survival being the only objective and reward, successful species are those that have developed by selection (through million years of evolution) strategies allowing adaptation to regime shifts, which means some capacity to predict and anticipate them. Then, studying these population strategies, we could get pieces of answers to this apparent contradiction. One possible solution was to work on scenarios. Indeed, a regime shift is not extremely frequent, and between two such events (i.e. up to decades), the dynamics of the population can be considered as reasonably deterministic, following a scenario. But at a given, unpredictable moment, due to regime shift the scenario is replaced by another one. Then we must be able to (i) describe the different scenarios a population can follow, (ii) define which scenario is currently in use according to the existing regime, and (iii) obtain indicators from the population which could inform that, due to regime shift, the system would likely soon move from one scenario to the other. Under these conditions, we could advise the managers that, as long as the scenario remains unchanged, the population is likely to respond linearly to variations in fishing effort, and its dynamics can be anticipated; but we should also be able to ring the alarm when a regime shift is likely to occur (Gerlotto, 2007). The preceding requires obtaining much more environmental and behavioural data than is currently collected. Costs then become an issue for scientific institutes. We addressed this constraint in two ways. First, we extracted all the existing information from data already collected. For instance we could measure the precise position of the oxycline from the recorded echograms in Peru (Ballón et al., 2011) and classify echoes in several groups by comparing acoustic frequencies which allowed us to evaluate zooplankton as well as fish abundance quantitatively all along transects (Ballón et al., 2011). Second, we integrated and empowered the fishers as observers and actors in the management of the environment that they exploit (Melvin et al., 2016a, b). This is technically possible because the instruments used on board fishing vessels are now exactly the same as those used by researchers (echosounders and sonars, underwater sensors of physical parameters, etc.). During my chairmanship of the ICES Fisheries Technology Committee I encouraged the works of a study group studying this question and a Cooperative Research Report was published in 2007 (Karp, 2007) which concluded that fishers’ data could be useful for scientific purposes. Actually, this question of involving the fishers is not only a matter of data collection: they must be full “actors of science” (Massé et al., 2016), for a number of obvious reasons. They have deep knowledge of fish behaviour (Hind, 2015); no improvement in the ecosystem-based monitoring (EBM) can been obtained without their active participation; their willingness to respect regulation depends on their approval of the research and its conclusions; they are part of the ecosystem and as such are both producing and suffering from the changes. Since the early 2000s, IRD has been co-operating with the Marine Institute of Peru (IMARPE, Instituto del Mar del Peru). There, we discovered the richness of a long series of acoustic surveys operated by fishing vessels, the famous EUREKA operations which began in the 1960s and were conducted on a yearly basis (Villanueva, 1971; Fernandes et al., 2002). This was later applied by other countries and laboratories (Chile, USA, Canada, etc. e.g. Melvin et al., 2016a), but in most cases, as in Peru, fishing vessel sampling strategies were decided by scientists. The question now concerned the acoustic data collected during autonomous fishing trips. Only very few occurrences existed in the late 2000 (Barbeaux et al., 2013; Niklitschek, 2016), but we may consider that fishers’ data are currently exploited, as evidenced in the special issue of the journal Fisheries Research that we edited on this topic in 2016 (Melvin et al., 2016b), ten years after the Karp (2007) proposal, i.e. after the already mentioned delay imposed by the “10 year law”. Fishing vessels as scientific platforms Another preliminary question had to be answered: can fishers be considered as “natural” predators or are they completely outside of the ecosystem mechanisms? The reason behind this question was the need to study fishers’ strategies and their relationships with fish distributions. As said above, the weight of a catch is a poor descriptor of a fishing operation, and any attempt to go further requires knowledge on the fisher’s behaviour. A series of important results were obtained in Lima, during the early 2000s (Bertrand et al., 2005), which completely changed our vision of fishers’ behaviour. Not only were the fishers similar in their hunting strategies to other predators (e.g. seabirds), but their trajectories were, as for every predator, highly sophisticated and could be formalized as a Lévi flight equation (Bertrand et al., 2005, 2007). Later, this specific equation for trajectories was included in a wider system, being modelled by the Hidden Semi-Markov Models (Joo et al., 2013). These results confirmed that the fishers’ trajectories could be formalized, extracted and exploited for scientific research. For instance, Joo et al. (2015) presented proxies of anchovy distribution maps using the fishers’ trajectories: the results were comparable to acoustic estimates. Once confirmed that fishers were a scientifically valid source of information, we could develop projects that made them partners in fisheries research, beginning with the use of their own acoustic data. At this time, a new Regional Fisheries Management Organization, the SPRFMO (South Pacific RFMO) was created to regulate and manage the exploitation of international stocks, especially the South Pacific jack mackerel, Trachurus murphyi (www.sprfmo.int). I worked on this species in co-operation with the National Fisheries Society of Peru (SNP, Sociedad Nacional de Pesquerías) and inside the EU delegation to SPRFMO (Hintzen et al., 2014). Acoustic data on jack mackerel were collected on board fishing vessels, processing methods were developed, and annual workshops were organized (Gerlotto et al., 2016a). Applying that concept to anchovy, the Peruvian scientists adapted the Hilborn et al. (2001) recommendation and developed an “Adaptive Precautionary Management” model (Chávez et al., 2008; Gutierrez, pers. comm.) which looks extremely efficient and is likely one of the first assessment models taking fish behaviour and general acoustic information (including fishers’) into consideration. Habitat as indicator The observation and measurement of fish behaviour helps to better understand (and anticipate) the dynamics of their populations. The next step was to find the best synthetic indicators to follow the dynamic changes of the population. A correct study of fish behaviour would require sampling such a large number of metrics that it is simply impossible, as observations on fish avoidance told us. Integrative indicators are necessary. We proposed the use of habitat as an ecological indicator, based on the following hypothesis. The fundamental motivation of fish behaviour is to guarantee the survival of the species. This means maintaining the population in a favourable environment, i.e. a given suitable habitat. This habitat is recognized by the species through the synthesis of a number of metrics, which we do not necessarily record. Then, habitat definition and evolution could be one synthetic indicator for understanding, analysing, and predicting the dynamics of the population (Zwolinski et al., 2010; Bertrand et al., 2016). Large stable habitats favour the development of the population; shrinking habitats are likely to reduce the population abundance. In order to test these hypotheses, IMARPE and IRD worked on habitat of the Peruvian anchovy (Engraulis ringens), and Bertrand et al. (2010, 2011) demonstrated for this species that (i) direct observations give enough pieces of information to describe, measure and map this habitat in three dimensions, and (ii) that the population dimension of the anchovy is directly linked to the habitat dimension (Bertrand et al., 2014). End of my story: the triggerfish mystery revealed We had now developed alternative hypotheses and instruments to propose a more realistic approach to fish population dynamics. But one element of the puzzle was lacking: we described the need to define the scenarios a species may select according to environmental conditions, but what should be included in a scenario? Information on the environment, the fishery, and fish behaviour was already available. The missing piece of information for writing a complete scenario concerned the population strategy of a species. Depending on their population strategies, species may react differently to environmental changes. For instance, existing models of reproductive strategies, e.g. the r- and K-selections (MacArthur and Wilson, 1967) consider the selection of a reproduction pattern; from them, a series of hypotheses have been drawn, e.g. the BOFFFF (Big Old Fat Fecund Female Fish, e.g. Hixon et al., 2014) which have strong implications on assessment. Nevertheless, they do not take into account the dynamic strategies acquired through natural selection on the population. There was a need to define the structure of the population and to find the mechanisms explaining why a species like B. capriscus is able to so dramatically change its behaviour, abundance, and spatial distribution. The concept of habitat that we described gave us some potential elements of an answer. It seems logical that, if habitat drives population abundance, it also drives its structure. Unfortunately, B. capriscus stocks are not considered profitable enough in fisheries to justify any funding for such research. Instead, we were urged to test this hypothesis on jack mackerel. This fish represented the most important exploited stock in the world in the 1980s and 1990s, when its biomass reached almost 25 million tonnes and catch in the 1990s was close to 5 million tonnes. By the 2010s, the biomass was less than 5 million tonnes, and production was less than half a million tonnes (www.sprfmo.int). The population structure of T. murphyi was studied over several years by teams from the different fishing countries, but no consensus was obtained. SPRFMO organized a dedicated workshop in 2008 (SPRFMO, 2008), and a series of hypotheses were listed, from a wide single stock to many discrete stocks all over its distribution area. I received here my [Lesson 15]: scientific results are very often influenced (or biased) by political interests. The conclusions of scientists from the different countries of the South Pacific were generally more in phase with the wishes of their respective governments than with actual scientific evidences. I fully understand that politicians decide not to take into consideration scientific recommendations. After all, this is their raison d’être: to make choices. But we cannot accept pressures on the scientists. The easiest recommendation is to say “resist”, but this is often extremely difficult. At least, scientists must know that this pressure will permanently exist when economic interests are involved. We worked on the population structure of T. murphyi and proposed a metapopulation hypothesis (Gerlotto et al., 2012), based on the definitions listed by Kritzer and Sales (2004) for marine organisms from the model developed by Levin (1969). For this study, we used the considerable sum of biological and ecological information accumulated on jack mackerel (Gerlotto and Dioses, 2013). This hypothesis was not completely satisfactory, because some of the conditions for metapopulations found in the literature were not fully respected, and we decided to go further in this study. A project submitted in 2011 to the EU was approved and after two years of work we concluded again that metapopulation was the most likely structure (Hintzen et al., 2014). In this project, we focused on jack mackerel habitat (Bertrand et al., 2016). Although more convincing, some conditions were still not fulfilled. We had to question whether the existing definition and conditions for metapopulation were applicable to large pelagic populations (McQuinn, 1997). We established a wider definition for metapopulations, making a distinction between two types according to their habitat characteristics: (i) the species living in “territory-bounded habitat” (TBH) and (ii) those living in “environment-bounded habitat” (EBH), pelagic metapopulations belonging to this last group (Gerlotto et al., 2016b). Animals living in TBH are unable to cross the geographical borders of their territory (e.g. mountains, lakes, islands …) and extend their habitat outside of the territory, even though environmental conditions would allow it (except for a few individuals going from one TBH to the other; Cury, 1994). On the contrary, populations living in EBH, where no such geographical borders exist, can increase their area of distribution by filling a suitable habitat only limited by favourable environmental conditions. Under this definition, jack mackerel inside its EBH was organized into a pelagic metapopulation. There are at least three selective advantages to this adaptive population strategy: (i) the fish is able to recolonize the niches that have been lost during the low abundance periods (paleoecology says on a scale of centuries, e.g. Sifeddine et al., 2008); (ii) separate subpopulations are small enough to adapt rapidly to changing conditions of the environment; and (iii) when all the subpopulations merge, the favourable genetic mutations which occurred in one given subpopulation are transmitted to the whole species, which allows it to adapt more rapidly to a highly variable environment. And now, back to the beginning: this EBH hypothesis, developed for T. murphyi, finally enlightens the “triggerfish mystery” when applied to B. capriscus. Generalizing the observations made on the jack mackerel, we can now suggest the following hypotheses: the triggerfish, although not pelagic during its period of low abundance, is not limited by territorial borders, and can be considered as an EBH metapopulation. The natural changes in triggerfish biomass could be described through an EBH metapopulation strategy for this species: a small climatic event occurred in 1972 in West Africa which favoured a successful recruitment. This high abundance induced a change in B. capriscus behaviour, making it possible to expand all over the new suitable pelagic habitat and to cover the entire Gulf of Guinea. This expansion lasted several generations and allowed the species to recover all of its lost niches and to homogenize its genetic pool. Eventually, the abundance decreased and the fish returned to its “natural” benthic behaviour and went back to its historical distribution and abundance. Incidentally, if this hypothesis is correct, the natural depletion of the triggerfish could indicate that the “collapse” of jack mackerel in the South Pacific Ocean is perhaps more the result of a population strategy than overfishing (although the two effects can be additive). Moreover, it questions (among other points) the meaning of a fixed natural mortality coefficient and of a “virgin biomass” (B0), which are currently used in assessment models. The distinction between natural depletion and overfishing collapse as defined by Petitgas et al. (2010), may also become an issue for assessment analysis. I would like to conclude this paper with Lesson [16]: I hope my story demonstrates that scientific research is a long process depending not only on intelligence, knowledge, ability to design hypotheses etc., but also on the availability of adapted techniques, intrinsic delays, and maturation of ideas. Unfortunately, modern life favours mostly short time scales, even in science where administrations impose short-duration (1–4 years) projects and a frenetic rhythm of publication production, evaluated quantitatively regardless of their actual value. There are some visible effects, such as the increasing creation of profit-driven scientific journals, the enormous inflation of meaningless articles, the almost complete absence of references older than 5 years in the bibliographies, and many others (cf. Lawrence, 2016). I am not sure that this short memory strategy is good news for research. In any case I am convinced that today, a 40-year quest for exploring the triggerfish mystery would not be accepted. Young scientists may have to comply with the bureaucratic requirements of the modern organization of research, but they must know that doing good science is another (long) story. Acknowledgements Without the help of Dr David A. Demer, Dr Richard L. O’Driscoll and an anonymous referee, who corrected both my pitiful English and my scientific overstatements, this document would not be readable. My warmest thanks to Dr Howard I. Browman who encouraged me to write this text and guided me from its first to its last version. This gives me opportunity for an additional [Lesson 17]: a great editor is the one you hate during the process of editing your paper and you are grateful to for the rest of your life once it is published. References Aggrey-Fynn J. 2007 . The fishery of Balistes capriscus (Balistidae) in Ghana and possible reasons for its collapse. Thesis Doctor of Natural Sciences Faculty 2 (Biology/Chemistry), University of Bremen, Bremen. October, 2007: 110 p. Albaret J. J. , Gerlotto F. 1976 . Biologie de l’Ethmalose, Ethmalosa fimbriata (Bowdich), en Côte d’Ivoire. 1-Description de la reproduction et des premiers stades larvaires. Documents Scientifiques, Centre de Recherches Océanographiques . Abidjan , 7 : 113 – 133 . OpenURL Placeholder Text WorldCat Aglen A. 1983 . Random errors of acoustic fish abundance estimates in relation to the survey grid density applied . FAO Fisheries Report ( 300 ): 293 – 298 . OpenURL Placeholder Text WorldCat Anderson E. D. 2015 . Lessons from a career in fisheries science . ICES Journal of Marine Science , 72 : 2169 – 2179 . Google Scholar Crossref Search ADS WorldCat Anonymous . 1990a . Collected reprints of the main contributed papers of EICHOANT program (Evaluation of Behaviour Influence on Fishery Biology and Acoustic Observations in Tropical Open Sea) presented during congresses from 1/1/87 to 4/30/90. Document Scientifique—Pôle de Recherche Océanologique et Halieutique Caraïbe; Fort de France (FWI), Le Robert: ORSTOM, IFREMER, 1990, (26): 29 p. Anonymous . 1990b . Report of the workshop on the applicability of spatial statistical techniques to acoustic survey data. ICES/CIEM, CM, 1990/D: 34 . Anonymous 2000 . Project AVITIS: “Three dimension Analysis and visualization of the spatial structure of fish schools using multibeam sonar image processing”. DG XIV FAIR CT96-1717 Final report (1 January 1997–31 December. 1999) Ansa-Emmin M. , Marcille J. 1976 . Rapport du groupe de travail sur la sardinelle (S. aurita) des côtes ivoiro-ghanéennes. Abidjan, 28 juin–3 juillet 1976. ORSTOM, Paris, 89 p. Ballón M. , Bertrand A., Lebourges-Dhaussy A., Gutiérrez M., Ayón P., Grados D., Gerlotto F. 2011 . Is there enough zooplankton to feed forage fish populations off Peru? An acoustic (positive) answer . Progress in Oceanography , 91 : 2011. 360 – 381 . Google Scholar Crossref Search ADS WorldCat Barbeaux S. J. , Horne J. K., Dorn M. W. 2013 . Characterizing walleye pollock (Theragra chalcogramma) winter distribution from opportunistic acoustic data . ICES Journal of Marine Science , 70 : 1162 – 1173 . Google Scholar Crossref Search ADS WorldCat Bazigos G. P. 1975 . The statistical efficiency of echo surveys with special reference to Lake Tanganyika. FAO Fisheries Technical Paper No. 139. Bazigos G. P. 1981 . A manual on acoustic surveys. UNDP/FAO. Series: CECAF/ECAF SERIES 80/17. Bertrand A. , Ballón M., Chaigneau A. 2010 . Acoustic observation of living organisms reveals the upper limit of the oxygen minimum zone . PLoS One , 5 : e10330. Google Scholar Crossref Search ADS PubMed WorldCat Bertrand A. , Chaigneau A., Peraltilla S., Ledesma J., Graco M., et al. 2011 . Oxygen: a fundamental property regulating pelagic ecosystem structure in the coastal southeastern tropical Pacific . PLoS One , 6 : e29558 . Google Scholar Crossref Search ADS PubMed WorldCat Bertrand A. , Habasque J., Hattab T., Hintzen N. T., Oliveros-Ramos R., Gutiérrez M., Demarcq H., Gerlotto F. 2016 . 3-D habitat suitability of jack mackerel Trachurus murphyi in the Southeastern Pacific, a comprehensive study . Progress in Oceanography , 146 : 199 – 211 . 2016. Google Scholar Crossref Search ADS WorldCat Bertrand A. , Gerlotto F., Bertrand S., Gutiérrez M., Alza L., Chipollini A., Díaz E., Espinoza P., Ledesma J., Quesquén R., et al. 2008 . Schooling behaviour and environmental forcing in relation to anchoveta distribution: an analysis across multiple spatial scales . Progress in Oceanography , 79 : 2008. 264 – 277 . Google Scholar Crossref Search ADS WorldCat Bertrand A. , Grados D., Colas F., Bertrand S., Capet X., Chaigneau A., Vargas G., Mousseigne A., Fablet R. 2014 . Broad impacts of fine-scale dynamics on seascape structure from zooplankton to seabirds . Nature Communications , 5 : 5239 . Google Scholar Crossref Search ADS PubMed WorldCat Bertrand S. , Burgos J. M., Gerlotto F., Atiquipa J. 2005 . Lévy trajectories of Peruvian purse-seiners as an indicator of the spatial distribution of anchovy (Engraulis ringens) . ICES Journal of Marine Science , 62 : 477 – 482 . Google Scholar Crossref Search ADS WorldCat Bertrand S. , Bertrand A., Guevara Carrasco R., Gerlotto F. 2007 . Scale-invariant movements of fishermen: the same foraging strategy as natural predators . Ecological Applications , 17 : 331 – 337 . Google Scholar Crossref Search ADS PubMed WorldCat Binet D. 1982 . Influence des variations climatiques sur la pêcherie des Sardinella aurita ivoiro-ghanéennes: relation sécheresse-surpêche . Oceanologica Acta , 5 : 443 – 452 . OpenURL Placeholder Text WorldCat Binet D. , Marchal E., Pezennec O. 1991 . Ivory Coast and Ghana Sardinella aurita fisheries variations and climatic changes: biological responses . Journal of Marine Biological Association of United Kingdom , 70 : 669 . OpenURL Placeholder Text WorldCat Boccaro N. 2010 . Modeling complex systems. Graduate texts in physics . Springer-Verlay , New York , 180 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Brehmer P. , Gerlotto F., Samb B. 2000 . Measuring fish school avoidance during acoustic surveys. In: Incorporation of external factors in marine resource surveys. ICES CM 2000: Annual Science Conference, 2000/09/24–27, Bruges, 13 p. Caverivière A. 1982 . Le baliste des côtes africaines, (Balistes carolinensis): biologie, prolifération et possibilités d’exploitation . Oceanologica Acta , 5 : 453 – 459 . OpenURL Placeholder Text WorldCat Caveriviére, A. 1991 . L'explosion démographique du baliste (Balistes carolinensis) en Afrique de l'Ouest et son évolution en relation avec les tendances climatiques. In Pêcheries ouest africaines : variabilité, instabilité et changement. Ed. by P. Cury and C. Roy. ORSTOM, Paris, pp. 354–367. Caverivière A. , Kulbicki M., Konan J., Gerlotto F. 1981 . Bilan des connaissances actuelles sur Balistes carolinensis dans le golfe de Guinée . Oceanologica Acta , 5 : 453 – 459 . OpenURL Placeholder Text WorldCat Caverivière A. , Gerlotto F., Stéquert B. 1980 . Balistes carolinensis, nouveau stock africain. La Pêche Maritime, août, 1980: 3 – 8 . Chávez F. P. , Bertrand A., Guevara-Carrasco R., Soler P., Csirke J. 2008 . The northern Humboldt Current System: Brief history, present status and a view towards the future . Progress in Oceanography , 79 : 95 – 105 . Google Scholar Crossref Search ADS WorldCat Coll M. , Libralato S., Tudela S., Palomera I., Pranovi F. 2008 . Ecosystem overfishing in the ocean . PLoS One , 3 : e3881 . Google Scholar Crossref Search ADS PubMed WorldCat Cury P. 1994 . Obstinate nature: an ecology of individuals. Thoughts on reproductive behavior and biodiversity . Canadian Journal of Fisheries and Aquatic Sciences , 51 : 1664 – 1673 . Google Scholar Crossref Search ADS WorldCat Cury P. , Roy C. 1989 . Optimal environmental window and pelagic fish recruitment success in upwelling areas . Canadian Journal of Fisheries and Aquatic Sciences , 46 : 670 – 680 . Google Scholar Crossref Search ADS WorldCat Cury P. , Roy C. 1991 . Pêcheries ouest-africaines: variabilité, instabilité et changement. ORSTOM, Paris, 527 p. Davenas E. , Beauvais F., Amara J., Oberbaum M., Robinzon B., Miadonnai A., Tedeschi A., Pomeranz B., Fortner P., Belon P., et al. 1988 . Human basophil degranulation triggered by very dilute antiserum against IgE . Nature , 333 : 816 – 818 . Google Scholar Crossref Search ADS PubMed WorldCat Diner N. , Masse J. 1987 . Fish school behaviour during echo survey observed by acoustic devices . ICES CM , 30: 28. OpenURL Placeholder Text WorldCat Daget J. , Durand J.-R. 1968 . Etude du peuplement de poissons d’un milieu saumâtre tropical poïkilohalin: la baie de Cocody en Côte d’Ivoire . Cahiers ORSTOM. Série Hydrobiologie , 2 : 91 – 111 . OpenURL Placeholder Text WorldCat Durand J.-R. , Amon-Kothias J.-B., Ecoutin J.-M., Gerlotto F., Hie-Dare J.-P., Laë R. 1978 . Statistiques de pêche en lagune Ebrié (Côte d’Ivoire) (1976 et 1977) . Document Scientifique Centre de Recherches Océanographiques Abidjan , 9 : 67 – 114 . OpenURL Placeholder Text WorldCat Fabre J.-H. 1924 . Souvenirs Entomologiques. Edition définitive illustrée, 10 volumes. Librairie Delagrave, Paris. Fernandes V. , Monod T., Mauny R. 1951 . Description de la côte occidentale d’Afrique: Sénégal au Cap de Monte, Archipels (Vol. 11). Centro de Estudos da Guiné Portuguesa. Fernandes P. G. , Gerlotto F., Holliday D. V., Nakken O., Simmonds E. J. 2002 . Acoustic applications in fisheries science: the ICES contribution . ICES Marine Science Symposia , 215 : 483 – 492 . OpenURL Placeholder Text WorldCat Forbes S. T. , Nakken 0. 1972 . Manual of methods for fisheries resource survey and appraisal. Part 2. The use of acoustic instruments for fish detection and abundance estimation . FAO Manual Fisheries Science , 5 : e3881. OpenURL Placeholder Text WorldCat Fréon P. 1989 . Climprod: a fully interactive expert system software for choosing and adjusting a global production model which account for changes in environmental factors. International Symposium on the Long-Term Variability of the Pelagic Fish Populations and their Environment, 1989/11/14-18, Sendai: 10 p. Fréon P. , Gerlotto F., Soria M. 1992 . Changes in school structure according to external stimuli: description and influence on acoustic assessment . Fisheries Research 15 : 45 – 66 . Google Scholar Crossref Search ADS WorldCat Fréon P. , Gerlotto F., Mullon C. 1991 . Les changements d’échelles en halieutique: l’exemple des petits pélagiques côtiers. In Le transfert d’échelle. Séminfor 4: Quatrième Séminaire Informatique de l’ORSTOM, 4, Brest. Colloques et Séminaires , pp. 343 – 364 . Ed. by Mullon C. ORSTOM , Paris . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Fréon P. , Misund O. A. 1999 . Dynamics of pelagic fish distribution and behaviour: effects on fisheries and stock assessment. Fishing News Books, Oxford. Frontier S. , Pichod-Viale D., Leprêtre A., Davoult D., Luczak C. 2008 . Écosystèmes (4e éd): Structure, Fonctionnement, Évolution . Dunod , Paris , 549 p. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Garcia S. 1977 . Biologie et dynamique des populations de crevettes roses (Penaeus duorarum notialis Perez-Farfante, 1967) en Côte d’Ivoire . Thèse Doctorat ès sciences , Univ. Marseille , 265 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Garcia S. M. 1996 . The precautionary approach to fisheries and its implications for fishery research, technology and management. An updated review. FAO Fisheries Technical Paper, 350/2: 75 p. Gerlotto F. 1975 . Note sur les biomasses pélagiques évaluées par écho-intégration dans la zone équatoriale du golfe de Guinée: premiers résultats. Document Scientifique Centre de Recherches Océanographiques, Abidjan, 2: 119 – 138 . Gerlotto F. 1976 . Biologie de Ethmalosa fimbriata (Bowdich) en Côte d’Ivoire. 2- Étude de la croissance en lagune par la méthode de Petersen. Documents Scientifiques Centre de Recherches Océanographiques, Abidjan, 7 : 1 – 27 . Gerlotto F. 1979 . Biologie de Ethmalosa Fimbriata (Bowdich) en Côte d’Ivoire. 3- Étude des Migrations en Lagune Ebrié . Documents Scientifiques Centre de Recherches Océanographiques, Abidjan , 10 : 3 – 41 . OpenURL Placeholder Text WorldCat Gerlotto, F. 1993 . Identification and spatial stratification of tropical fish concentrations using acoustic populations . Aquatic Living Resources , 6 : 243 – 254 . Crossref Search ADS WorldCat Gerlotto F. 2007 . L’observation directe du comportement: une information indispensable dans une approche écosystémique de la gestion des stocks halieutiques. Diplôme d’Habilitation à Diriger des Recherches, Université Montpellier 2, École Doctorale Systèmes Intégrés en Biologie, Agronomie, Géosciences, Hydrosciences, Environnement (SIBAGHE), Montpellier, 14 Décembre 2007: 225 p. Gerlotto F. , Dioses T. 2013 . Bibliographical synopsis on the main traits of life of Trachurus murphyi in the South Pacific Ocean. SPRFMO, SC-01-INF-17, 1st Meeting of the Scientific Committee, La Jolla, 21–27 October 2013, 217 p. Gerlotto F. , Fréon P. 1988 . Influence of the structure and behaviour of fish school in acoustic assessment . ICES C.M ., 8 : 53 . OpenURL Placeholder Text WorldCat Gerlotto F. , Marchal E. 1985 . The concept of acoustic populations as an aid for biomass identification. Comm. ICES/FAST working group, Tromsoe, 22–24 May, 7 p. Gerlotto F. , Marchal E. 1987 The concept of acoustic populations: its use for analyzing the results of acoustic cruises. Communication au 3ème Congrès d’Acoustique Halieutique du Conseil International pour l’Exploitation de la Mer, 22–26 Juin, Seattle. Gerlotto F. , Paramo J. 2003 . The three-dimensional morphology and internal structure of clupeid schools as observed using vertical scanning multibeam sonar . Aquatic Living Resources , 16 : 113 – 122 . Google Scholar Crossref Search ADS WorldCat Gerlotto F. , Stéquert B. 1978 . La pêche maritime artisanale en Afrique de l’Ouest. Caractéristiques générales . La Pêche Maritime , 1202 : 278 – 285 . OpenURL Placeholder Text WorldCat Gerlotto F. , Stequert B. 1983 . Une méthode de simulation pour étudier les densités de poissons. Application à deux cas réels [Simulation method to determine fish density distribution: two case studies]. FAO Fisheries Report (300): 278 – 292 . Gerlotto F. , Hem S., Briet R. 1976a . Statistiques de pêche en lagune Ebrié: année 1975. Documents Scientifiques Centre de Recherches Océanographiques, Abidjan (1976) Série Statistiques CRO; 2: 39 . Gerlotto F. , Gutiérrez M., Bertrand A. 2012 . Insight on population structure of Chilean jack mackerel (Trachurus murphyi) . Aquatic Living Resources , 2012. 25 : 341 – 355 . Google Scholar Crossref Search ADS WorldCat Gerlotto F. , Mensah M. A., Stéquert B. 1979a . La pêche maritime artisanale en Afrique de l’Ouest: la pêche au Ghana . La Pêche Maritime , 1210 : 7. OpenURL Placeholder Text WorldCat Gerlotto F. , Soria M., Fréon P. 1999 . From two dimensions to three: the use of multibeam sonar for a new approach in fisheries acoustics . Canadian Journal of Fisheries and Aquatic Sciences , 56 : 6 – 12 . Google Scholar Crossref Search ADS WorldCat Gerlotto F. , Stéquert B., Brugge W. J. 1979b . La pêche maritime artisanale en Afrique de l’Ouest: la pêche au Sénégal. La Pêche Maritime, 1211, février, 12 p. Gerlotto F. , Stéquert B.L., Philippe V. 1976b . Répartition et abondance des poissons pélagiques côtiers du plateau continental sénégambien évaluées par écho-intégration en Avril-Mai 1976 (Campagne Cap 7605). Documents Scientifiques du CRO D.-T., ORSTOM, 62, October, 64 p. Gerlotto F. , Verdeaux F., Stéquert B. 1980 . La pêche maritime artisanale en Afrique de l’Ouest: évolution et impact socio-économique à travers l’exemple de la pêche en lagune en Côte d’Ivoire. La Pêche Maritime, Janvier, 8 p. Gerlotto F. , Simmonds E. J., Georgakarakos S. 2000 . Current state of the art of sonar techniques. In Report on Echo Trace Classification . Ed. by Reid D. G.. ICES Cooperative Research Report , 238 , Mars , 79 – 96 Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gerlotto F. , Bertrand S., Bez N., Gutierrez M. 2006 . Waves of agitation inside anchovy schools observed with multibeam sonar: a way to transmit information in response to predation . ICES Journal of Marine Science , 63 : 1405 – 1417 . Google Scholar Crossref Search ADS WorldCat Gerlotto F. , Jones E., Bez N., Reid D. G. 2010 . When good neighbours become good friends: observing small scale structures in fish aggregations using multibeam sonar . Aquatic Living Resources , 23 : 143 – 151 . Google Scholar Crossref Search ADS WorldCat Gerlotto F. , Castillo J., Saavedra A., Barbieri M. A., Espejo M., Cotel P. 2004 . Three-dimensional structure and avoidance behaviour of anchovy and common sardine schools in central southern Chile . ICES Journal of Marine Science , 61 : 1120 – 1126 . Google Scholar Crossref Search ADS WorldCat Gerlotto F. , Gutierrez M., Vasquez C., Peraltilla S., Aliaga A., Bernales R. 2016a . Fisheries and fishers’ acoustic data. 1. Collecting acoustic data from fishing vessels: methods, metrics and indicators. ICES Working Group on Fisheries, Acoustics, Science and Technology (WGFAST), Vigo, 19–22 April 2016. Gerlotto F. , Hintzen N. T., Habasque J., Corten A., Gutierrez M., Bertrand A. 2016b . The concept of “Pelagic metapopulation” as exemplified by the case of Jack mackerel Trachurus murphyi in the South Pacific Ocean. SPRFMO, 4th meeting of the Scientific Committee, The Hague, 10–15 October 2016, SC-04-JM-02 (www.sprfmo.int). Hilborn R. 2011 . Overfishing: What Everyone Needs to Know ? Oxford University Press , Oxford . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hilborn R. , Orensanz J. M., Parma A. M. 2005 . Institutions, incentives and the future of fisheries . Philosophical Transactions: Biological Sciences, 360: 47 – 57 . OpenURL Placeholder Text WorldCat Hilborn R. C. , Coppersmith S., Mallinckrodt A. J., McKay S. 1994 . Chaos and nonlinear dynamics: an introduction for scientists and engineers . Computers in Physics , 8 : 689. Google Scholar Crossref Search ADS WorldCat Hilborn R. , Maguire J.-J., Parma A. M., Rosenberg A. A. 2001 . The precautionary approach and risk management: can they increase the probability of success in fishery management? Canadian Journal of Fisheries and Aquatic Sciences , 58N : 99 – 107 . Google Scholar Crossref Search ADS WorldCat Hind E. J. 2015 . A review of the past, the present, and the future of fishers’ knowledge research: a challenge to established fisheries science . ICES Journal of Marine Science , 72 : 341 – 358 . Google Scholar Crossref Search ADS WorldCat Hintzen N. T. , Corten A., Gerlotto F., Habasque J., Bertrand A., Lehodey P., Brunel T., Dragon A.-C., Senina I. 2014 . Hydrography and Jack mackerel stock in the South Pacific—Final report. Studies for carrying out the Common Fisheries Policy, Open call for tenders No MARE/2011/16 Lot 1. Report number C176.14, 65 p. Hixon M. A. , Johnson D. W., Sogard S. M. 2014 . BOFFFFs: on the importance of conserving old-growth age structure in fishery populations . ICES Journal of Marine Science , 71 : 2171 – 2185 . Google Scholar Crossref Search ADS WorldCat Hsieh C. H. , Glaser S. M., Lucas A. J., Sugihara G. 2005 . Distinguishing random environmental fluctuations from ecological catastrophes for the North Pacific Ocean . Nature , 435 : 336 – 340 . Google Scholar Crossref Search ADS PubMed WorldCat Hutchings J. A. , Reynolds J. D. 2004 . Marine fish population collapses: consequences for recovery and extinction risk . BioScience , 54 : 297 – 309 . Google Scholar Crossref Search ADS WorldCat Johannesson K. A. , Mitson R. B. 1983 . Fisheries acoustics: a practical manual for aquatic biomass estimation. FAO Fisheries Technical Paper, 240, 230 p. Jolly G. M. , Hampton I. 1991 . Some problems in the statistical design and analysis of acoustic surveys to assess fish biomass . Rapports et procès-verbaux des réunions / Conseil permanent international pour l'exploration de la mer , 189 : 415 – 421 . OpenURL Placeholder Text WorldCat Joo R. , Bertrand S., Tam J., Fablet R. 2013 . Hidden Markov models: the best models for forager movements? PLoS One , 8 : e71246 . Google Scholar Crossref Search ADS PubMed WorldCat Joo R. , Salcedo O., Gutierrez M., Fablet R., Bertrand S. 2015 . Defining fishing spatial strategies from VMS data: insights from the world’s largest monospecific fishery . Fisheries Research , 164 : 223 – 230 . Google Scholar Crossref Search ADS WorldCat Karp, W. A. (Ed.) 2007 Collection of acoustic data from fishing vessels. ICES Cooperative Research Report, 287, August, 84 p. Kritzer J. P. , Sales P. F. 2004 . Metapopulation ecology in the sea: from Levin’s model to marine ecology and fisheries science . Fish and Fisheries , 5 : 131 – 140 . Google Scholar Crossref Search ADS WorldCat Laloë F. 1985 . Contribution à l’étude de la variance d’estimateurs de biomasse de poissons obtenus par écho-intégration . Océanographie Tropicale , 20 : 161 – 169 . OpenURL Placeholder Text WorldCat Lawrence P. A. 2016 . Chapter Thirty-Six—The Last 50 Years: Mismeasurement and Mismanagement Are Impeding Scientific Research. Essays on Developmental Biology, Part A . Current Topics in Developmental Biology , 116 : 617 – 631 . Google Scholar Crossref Search ADS PubMed WorldCat Levin R. 1969 . Some demographic and genetic consequences of environmental heterogeneity for biological control . Bulletin of Entomological Society of America , 15 : 237 – 240 . Google Scholar Crossref Search ADS WorldCat MacArthur R. , Wilson E. O. 1967 . The Theory of Island Biogeography. Princeton University Press, Princeton, NJ. [2001 reprint]. Marchal E. , Gerlotto F. 1985 . The use of a digital echo-integrator to describe the echograms. Comm. ICES/FAST Working Group, Tromsoe, May 22–24 1985, 3 p. Marchal E. , Picault J. 1977 . Répartition et abondance évaluées par échointégration des poissons du plateau ivoiro-ghanéen en relation avec les upwellings locaux . Journal De Recherche Océanographique , 2 : 39 – 57 . OpenURL Placeholder Text WorldCat Marchal E. , Burczynski J., Gerlotto F. 1979 . République de Guinée: évaluation acoustique des ressources pélagiques le long des côtes de Guinée, Sierra Léone et Guinée-Bissau (N/O Capricorne: novembre-décembre 1978). FAO, FI: GUI/74/024/2, mars 1979, 100 p. Marchal E. , Gerlotto F., Stéquert B. 1993 . On the relationship between scattering layer and tuna abundance in the Eastern Atlantic equatorial current system . Oceanologica Acta , 16 : 261 – 272 . OpenURL Placeholder Text WorldCat Marchal E. , Burczynski J., Gerlotto F., Stéquert B., Varlet F. 1980 . Rapport sur une évaluation acoustique des ressources en poissons pélagiques dans la sous-région guinéenne: deuxième campagne du N/O Capricorne, mars 1979. Fao/gcp/gui/003(nor), 1980: 80 p. Marchetti C. 1998 . Notes on the limits to knowledge explored with Darwinian logic . Complexity , 3 : 22 – 35 . Google Scholar Crossref Search ADS WorldCat Massé J. , Gerlotto F. 2003 . Introducing nature in fisheries research: the use of underwater acoustics for an ecosystem approach of fish population . Aquatic Living Resources , 16 : 107 – 112 . Google Scholar Crossref Search ADS WorldCat Massé J. , Sanchez F., Delauney D., Robert J.-M., Petitgas P. 2016 . A partnership between science and industry for a monitoring of anchovy and sardine in the Bay of Biscay: when fishermen are actors of science . Fisheries Research , 178 : 26 – 38 . Google Scholar Crossref Search ADS WorldCat Matheron G. 1970 . La théorie des variables régionalisées et ses applications . Cahiers de Morphologie mathématique, Fontainebleau, École des Mines de Paris, fascicule , 5 : 212 p. OpenURL Placeholder Text WorldCat McQuinn I. 1997 . Metapopulations and the Atlantic Herring . Revue of Fish Biology and Fisheries , 7 : 297 – 329 . Google Scholar Crossref Search ADS WorldCat Melvin G. D. , Kloser R., Honkalehto T. 2016a . The adaptation of acoustic data from commercial fishing vessels in resource assessment and survey monitoring . Fisheries Research , 178 : 13 – 25 . Google Scholar Crossref Search ADS WorldCat Melvin G. D. , Gerlotto F., Lang C., Trillo P. 2016b . Fishing vessels as scientific platforms: an introduction . Fisheries Research , 178 : 1 – 2 . Google Scholar Crossref Search ADS WorldCat Monod T. 1950 . Notes d’ichtyologie ouest-africaine . Bulletin De L’Institut Français D’Afrique Noir , 12 : 1 – 71 . OpenURL Placeholder Text WorldCat Mullon C. , Pichon G. 1991 . Problèmes statistiques de la très grande variabilité. In Le transfert d’échelle , pp. 183 – 194 . Ed. by Mullon C. Paris : ORSTOM , . (Colloques et Séminaires). Séminfor 4, Quatrième Séminaire Informatique de l’ORSTOM, 4, Brest. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Myers R. A. , Worms B. 2005a . Rapid worldwide depletion of predatory fish communities . Nature , 423 : 280 – 283 . Google Scholar Crossref Search ADS WorldCat Myers R. A. , Worms B. 2005b . Extinction, survival or recovery of large predatory fishes . Philosophical Transactions of the Royal Society of Britain , 360 : 13 – 20 . Google Scholar Crossref Search ADS WorldCat Niklitschek E. J. 2016 . Distribution, density and relative abundance of Antarctic krill estimated by maximum likelihood geostatistics on acoustic data collected during commercial fishing operations . Fisheries Research , 178 : 114 – 121 . Google Scholar Crossref Search ADS WorldCat Olsen K. 1969 . Directional responses in herring to sound and noise stimuli. ICES CM 1969/B:20, 8 pp. Olsen K. 1990 . Fish behaviour and acoustic sampling . Rapports Et Procès-Verbaux Des Réunions Du Conseil International Pour L’Exploration De La Mer , 189 : 147 – 158 . OpenURL Placeholder Text WorldCat Pauly D. 2016 . Having to science the hell out of it . ICES Journal of Marine Science , 73 : 2156 – 2166 . Google Scholar Crossref Search ADS WorldCat Pauly D. , Alder J., Bennett E., Christensen V., Tyedmers P., Watson R. 2003 . The future for fisheries . Science , 302 : 1359 – 1361 . Google Scholar Crossref Search ADS PubMed WorldCat Petitgas P. 1990 . Geostatistics for fish acoustic surveys: precision of the abundance estimate and survey efficiency. Rapports et Procès-Verbaux des Réunions du Conseil International pour l’Exploration de la Mer, Copenhague, 4–9 Octobre 1990, 27 p. Petitgas P. , Secor D. H., McQuinn I., Huse G., Lo N. 2010 . Stock collapses and their recovery: mechanisms that establish and maintain lifecycle closure in space and time . ICES Journal of Marine Science , 67 : 1841 – 1848 . Google Scholar Crossref Search ADS WorldCat Pyanov A. I. 1993 . Fish learning in response to trawl fishing . ICES Marine Science Symposia , 196 : 12 – 16 . OpenURL Placeholder Text WorldCat Rivoirard J. , Simmonds J., Foote K. G., Fernandes P., Bez N. 2008 . Geostatistics for Estimating Fish Abundance . John Wiley and Sons, New York . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Robertson J. B. 1977 . Summary report: FIOLENT 1976 Eastern Central Atlantic coastal fishery resource survey, southern sector. CECAF/TECH/77/2 (En), FAO, Rome, 115 p Shotton R. , Bazigos G. P. 1984 . Techniques and considerations in the design of acoustic surveys . Rapports Et Procès-Verbaux Des Réunions Du Conseil International Pour L’Exploration De La Mer , 184 : 34 – 57 . OpenURL Placeholder Text WorldCat Sifeddine A. , Gutiérrez D., Ortlieb L., Boucher H., Velazco F., Field D., Vargas G., Boussafir M., Salvatteci R., Ferreira V., et al. 2008 . Laminated sediments from the central Peruvian continental slope: a 500 year record of upwelling system productivity, terrestrial runoff and redox conditions . Progress in Oceanography , 79 : 190 – 197 . Google Scholar Crossref Search ADS WorldCat Simmonds E. J. , Williamson N., Gerlotto F., Aglen A. 1992 . Acoustic survey design and analysis procedure: a comprehensive review of current practice. ICES Cooperative Research Report no. 187. ICES, 136 p. Simmons C. M. , Szedlmayer S. T. 2012 . Territoriality, reproductive behavior, and parental care in grey triggerfish, Balistes capriscus, from the northern Gulf of Mexico . Bulletin of Marine Science , 197 : 197 – 209 . Google Scholar Crossref Search ADS WorldCat Soria M. 1994 . Structure et stabilité des bancs et agrégations de poissons pélagiques côtiers tropicaux: application halieutique. Thèse Doctorat Univ. Paris, 284 p. (ORSTOM, Travaux et Documents Microédités; 125). Soria M. , Bahri T., Gerlotto F. 2003 . Effect of external factors (environment and survey vessel) on fish school characteristics observed by echosounder and multibeam sonar in the Mediterranean Sea . Aquatic Living Resources , 16 : 145 – 157 . Google Scholar Crossref Search ADS WorldCat Soria M. , Fréon P., Gerlotto F. 1996 . Analysis of vessel influence on spatial behaviour of fish schools using a multi-beam sonar and consequences for biomass estimates by echo-sounder . ICES Journal of Marine Science , 53 : 453 – 458 . Google Scholar Crossref Search ADS WorldCat Soria M. , Fréon P., Chabanet P. 2007 . Schooling properties of an obligate and a facultative fish species . Journal of Fish Biology , 71 : 1257 – 1269 . Google Scholar Crossref Search ADS WorldCat SPRFMO , 2008 . Report of the South Pacific Regional Fisheries Management Organization Chilean Jack Mackerel Workshop, 30 June–4 July 2008, Santiago, Chile: 00.-SPRFMO-JM-2008-WORKSHOP-REPORT-FINAL-6: 70 p. Stephenson R. L. , Paul S., Pastoors M. A., Kraan M., Holm P., Wiber M., Mackinson S., Dankel D. J., Brooks K., Benson A. 2016 . Integrating fishers’ knowledge research in science and management . ICES Journal of Marine Research , 73 : 1459 – 1465 . Google Scholar Crossref Search ADS WorldCat Stéquert B. , Gerlotto F. 1977 . Une méthode acoustique rapide d’évaluation des stocks de poissons pélagiques côtiers: l’écho-intégration . La Pêche Maritime, Mars , 1977 : 3 – 8 . OpenURL Placeholder Text WorldCat Stéquert B. , Gerlotto F., Le Philippe V. 1977 . Campagne d’écho-intégration Echoproc: résultats d’observations . Archives Du CRODT [Dakar] , 51 : 60 . OpenURL Placeholder Text WorldCat Villanueva R. 1971 . The Peruvian Eureka program of rapid acoustic survey. In Modern Fishing Gear of the World, vol. 3, pp. 20–24 . Ed. by Kristjonsson H.. Fishing News (Books) Ltd ., London , 537 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Worm B. , Hilborn R., Baum J. K., Branch T. A., Collie J. S., Costello C., Fogarty M. J., Fulton E. A., Hutchings J. A., Jennings S., et al. 2009 . Rebuilding Global Fisheries . Science , 325 : 578 – 585 . Google Scholar Crossref Search ADS PubMed WorldCat Zwolinski, J. P., and Demer, D. A. 2012 . A cold oceanographic regime with high exploitation rates in the Northeast Pacific forecasts a collapse of the sardine stock . Proceedings of the National Academy of Sciences , 109 : 4175 – 4180 . Crossref Search ADS WorldCat Zwolinski J. P. , Emmett R. L., Demer D. A. 2010 . Predicting habitat to optimize sampling of Pacific sardine (Sardinops sagax) . ICES Journal of Marine Science , 68 : 867 – 879 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Dickey-Collas,, Mark;McQuatters-Gollop,, Abigail;Bresnan,, Eileen;Kraberg, Alexandra, C;Manderson, John, P;Nash, Richard D, M;Otto, Saskia, A;Sell, Anne, F;Tweddle, Jacqueline, F;Trenkel, Verena, M

doi: 10.1093/icesjms/fsx158

Geoffroy,, Maxime;Cottier, Finlo, R;Berge,, Jørgen;Inall, Mark, E

doi: 10.1093/icesjms/fsw158pmid: N/A

Abstract An autonomous underwater vehicle (AUV) carrying 614 kHz RDI acoustic doppler current profilers (ADCPs) was deployed at four locations over the West Spitsbergen outer shelf in July 2010. The backscatter signal recorded by the ADCPs was extracted and analysed to investigate the vertical distribution and patchiness of pelagic organisms during midnight sun. At the northernmost locations (Norskebanken and Woodfjorden), fresher and colder water prevailed in the surface layer (0–20 m) and scatterers (interpreted as zooplankton and micronekton) were mainly distributed below the pycnocline. In contrast, more saline and warmer Atlantic Water dominated the surface layer at Kongsfjordbanken and Isfjordbanken and scatterers were concentrated in the top 20 m, above the pycnocline. Pelagic scatterers formed patchy aggregations at all locations, but patchiness generally increased with the density of organisms and decreased at depths >80 m. This study contributes to our understanding of the vertical distribution of pelagic organisms in the Arctic, and the spatial coverage of the AUV has extended early acoustic studies limited to Arctic fjords from 1D observations to a broader offshore coverage. Neither synchronized nor unsynchronized vertical migrations were detected, but autonomous vehicles with limited autonomy (<1 day) may not be as effective as long-term mooring deployments or long-range AUVs to study vertical migrations. Short-term AUV-based acoustic surveys of the pelagic communities are nonetheless highly complementary to Eulerian studies, in particular by providing spatial measurements of patchiness. Compared with ship-based or moored acoustic instruments, the 3D trajectory of AUVs also allows using acoustic instruments with higher frequencies and better size resolution, as well as the detection of organisms closer to the surface. Introduction Fundamental aspects of the abundance, lifecycle, vertical distribution, and migratory behaviour of zooplankton and nekton in the Arctic have been studied using traditional net techniques (e.g. Falk-Petersen et al., 2007; Eisner et al., 2013; Darnis and Fortier, 2014) and through the use of acoustics (e.g. La et al., 2015; Geoffroy et al., 2016). For instance, Acoustic Doppler Current Profilers (ADCPs) have been used to document the variations in behaviour of pelagic scatterers with temporal resolution ranging from minutes to seasons (Wallace et al., 2010; Last et al., 2016). The community composition of assemblages detected by acoustics has been estimated from net samples or sediment trap content (e.g. Cottier et al., 2006; Wallace et al., 2010; Berge et al., 2014). ADCPs are primarily deployed to measure current velocity, but their backscatter data can reveal detailed information about the pelagic ecosystem when multi-frequency scientific echosounders are not available (Brierley et al., 2006; Valle-Levinson et al., 2014). However, most ADCP studies on the vertical distribution of pelagic scatterers in the Arctic have been based on Eulerian sampling and lack spatial resolution (e.g. Cottier et al., 2006; Berge et al., 2014; Last et al., 2016). Spatial patchiness remains particularly difficult to measure using data from nets or moored instruments. Autonomous underwater vehicles (AUVs) represent an alternative to Eulerian platforms and allow spatial surveys of the water column (Fernandes et al., 2003; Schofield et al., 2010; Berge et al., 2012). AUVs have a longer operational range and are less vulnerable to bad weather than remotely operated vehicles. They access areas too shallow for scientific vessels (An et al., 2001), and can survey under an ice cover (Brierley et al., 2002). Acoustic devices mounted on AUVs can survey closer to the surface (Boyd et al., 2010) or seabed compared with moored or ship-mounted instruments, thus reducing the surface blind zone and bottom dead zone (i.e. blind areas respectively created by the near-field and the conical shape of the acoustic beam; Scalabrin et al., 2009). In addition, a 3D trajectory allows AUVs to approach targets close enough to use higher frequency acoustic instruments with better size resolution (Fernandes et al., 2003). In July 2010, an AUV fitted with turbulence sensors, ADCPs, and a conductivity-temperature-depth sensor (CTD) was deployed at four locations to study the physical oceanographic environment over the West Spitsbergen outer shelf (Steele et al., 2012). Here, we analyse data from the downward- and upward-looking ADCPs to investigate vertical distributions and patchiness of pelagic scatterers over a larger geographical area than previous studies limited to an Arctic fjord (Cottier et al., 2006; Berge et al., 2014). Specifically, we aim to test the hypotheses that (i) vertical migrations are limited to unsynchronized behaviour during midnight sun (Cottier et al., 2006); and (ii) hydrography determines the depth of pelagic organisms when they are not migrating (Berge et al., 2014). Advantages and limitations of using AUV-mounted ADCPs for biological studies are further discussed. Material and methods Study design and area A Kongsberg Hydroid REMUS AUV, depth rated to 600 m, was deployed in the NW sector of Spitsbergen at four locations on five different occasions (Figure 1) between 6 and 20 July 2010 (Table 1). The oceanographic conditions in this region are dominated by the presence of relatively warm and saline Atlantic Water (AW: T > 3.0°C, S > 34.65), carried northward along the slope by the West Spitsbergen Current (Saloranta and Haugan, 2001; Cottier et al., 2005). On the shelf, and forming a front with the AW, is a seasonally varying presence of cooler and fresher Arctic Water (ArW: −1.5°C < T < 1.0°C, 34.30 < S < 34.80) (Svendsen et al., 2002; Cottier and Venables, 2007). Table 1 Details of the AUV deployments Location . Date . Time (local) . Bottom depth (m) . NB 18 July 2010 06:12–12:35 ∼500 WF 16 July 2010 13:41–19:11 134 IF 20 July 2010 20:04–02:04 225 KF 6 July 2010 13:44–17:27 ∼550 KF 12 July 2010 09:51–16:45 ∼550 Location . Date . Time (local) . Bottom depth (m) . NB 18 July 2010 06:12–12:35 ∼500 WF 16 July 2010 13:41–19:11 134 IF 20 July 2010 20:04–02:04 225 KF 6 July 2010 13:44–17:27 ∼550 KF 12 July 2010 09:51–16:45 ∼550 Table 1 Details of the AUV deployments Location . Date . Time (local) . Bottom depth (m) . NB 18 July 2010 06:12–12:35 ∼500 WF 16 July 2010 13:41–19:11 134 IF 20 July 2010 20:04–02:04 225 KF 6 July 2010 13:44–17:27 ∼550 KF 12 July 2010 09:51–16:45 ∼550 Location . Date . Time (local) . Bottom depth (m) . NB 18 July 2010 06:12–12:35 ∼500 WF 16 July 2010 13:41–19:11 134 IF 20 July 2010 20:04–02:04 225 KF 6 July 2010 13:44–17:27 ∼550 KF 12 July 2010 09:51–16:45 ∼550 Figure 1 Open in new tabDownload slide Map of the study area indicating bathymetry and the limits of the AUV deployments (black boxes) at NB, WF, KF, and IF. Figure 1 Open in new tabDownload slide Map of the study area indicating bathymetry and the limits of the AUV deployments (black boxes) at NB, WF, KF, and IF. For each deployment, AUV-based sampling consisted of four to seven horizontal transects, each of 5–10 km and conducted at depths ranging from 10 to 170 m (Figure 2a–e). The AUV surfaced at the completion of each transect to acquire a GPS position and to communicate with the AUV operators by WiFi or Iridium. In total, the survey covered an area of ∼24 km2 over the outer shelf (Figure 2a–e; right column). The sun remained above the horizon throughout the study giving continuous (though not constant) illumination. Deployments at Norskebanken (NB), Woodfjorden (WF), and Kongsfjordbanken (KF) were conducted in the middle of the day, when the sun elevation was between 22 and 35°. The deployment at Isfjordbanken (IF) was conducted around midnight, when the sun elevation was between 13 and 15° (http://www.sunearthtools.com; accessed on 17 April 2016). Figure 2 Open in new tabDownload slide Left column: Continuous volume backscattering strength (Sv in dB re 1 m−1) at (a) NB on 18 July, (b) WF on 16 July, (c) KF on 6 July, (d) KF on 12 July, and (e) IF on 20 July. Local time of deployments and retrievals are indicated on the x-axis. The solid black line represents the trajectory of the AUV and the dashed black lines demarcate the SL, the IL, and the DL. Right column: Position (lat/long) and depth along the trajectory of each deployment. Figure 2 Open in new tabDownload slide Left column: Continuous volume backscattering strength (Sv in dB re 1 m−1) at (a) NB on 18 July, (b) WF on 16 July, (c) KF on 6 July, (d) KF on 12 July, and (e) IF on 20 July. Local time of deployments and retrievals are indicated on the x-axis. The solid black line represents the trajectory of the AUV and the dashed black lines demarcate the SL, the IL, and the DL. Right column: Position (lat/long) and depth along the trajectory of each deployment. Acoustic and environmental data collection The AUV recorded acoustic data, temperature, and salinity along transects (see Steele et al., 2012 for further details). Two RDI 614 kHz ADCPs mounted on the AUV, one looking upward and another downward, recorded the raw acoustic backscatter to about 42 m both above and below the vehicle. The AUV cruised at three to four knots and the ping rate of the ADCPs varied from 1 ping each 6 to 7.7 s, resulting in a horizontal resolution between 9 and 16 m. A CTD mounted on the AUV recorded temperature-salinity profiles to calculate (i) speed of sound; (ii) the coefficient of absorption; and (iii) density gradient profiles used to determine the depth and water density at the pycnocline. In the analysis of backscatter data, we followed Cottier et al. (2006) and partitioned the water column into three layers: (i) the Surface Layer (SL; 0–20 m), an Intermediate Layer (IL; 20–80 m), and a Deeper Layer (DL; >80 m). Backscatter data The acoustic volume backscattering strength (Sv in dB re 1 m−1) is an indication of the density of scatterers in a given volume. Because the 614 kHz ADCP signal can detect single targets as small as ∼2.4 mm (i.e. wavelength at c = 1500 m·s−1), most of the backscatter measured here can likely be attributed to meso- and macrozooplankton (Lorke et al., 2004). Although fish are better detected at higher frequencies, micronekton also likely contributed to a portion of the backscatter (e.g. Benoit-Bird, 2009). Sv was calculated from raw data using the SONAR equation adapted for ADCPs (Deines, 1999). The coefficient of absorption (α) used to calculate the Time-Varied-Gain (TVG = 40 logR + 2αR, where R is the range from the transducer) was estimated from mean temperature and salinity values recorded with the AUV-mounted CTD. The inclusion of a maximum Sv threshold of −45 dB discarded potential stronger echoes from large targets and noise. A time-varied-threshold (TVT = 20 logR + 2αR − 142), selected with an iteration process on echoes typical of noise, was added to offset noise amplification at depth by the TVG (e.g. Benoit et al., 2008; Geoffroy et al., 2016). Data from the upward looking ADCP in KF on 6 July were polluted by noise and removed from the analysis. For each ping, Sv values were calculated over 4 m vertical bins to be consistent with previous ADCP-based studies (Cottier et al., 2006; Wallace et al., 2010; Berge et al., 2014). For each deployment, linear sv values from all bins of the same depth were averaged and associated with mean temperature and salinity at each depth. Vertical velocity anomalies To verify the occurrence of unsynchronized vertical migration, vertical velocity anomalies (w’) were calculated for each bin by subtracting the average vertical speed for the entire deployment from the vertical speed within that bin (Cottier et al., 2006). A positive mean w’ for a given bin corresponds to an overall upward migration, while negative values indicate downward migration. To limit biases from the vertical movement of the AUV, only vertical speed measurements collected at fixed depths were used for these calculations and aberrant values (>15 mm·s−1 or >4-fold mean speed) were discarded. Estimation of density and calculation of the patchiness index To calculate patchiness indices, we derived an estimate of the density of scatterers (ρv in ind·m−3) within each bin (1 ping horizontally × 4 m vertically): ρv=svσbs(1) sv is the linear volume scattering strength (m2·m−3) and σbs the cross-sectional area of the average scatterer (Parker-Stetter et al., 2009). No net samples were collected in the vicinity of the AUV deployments, but as the 614 kHz signal is likely dominated by zooplankton we estimated an average target strength (TS) of −89.94 dB re 1 m2 based on the average zooplankton scatterer captured by Cottier et al. (2006) and using the randomly oriented fluid bent-cylinder model (Stanton et al., 1994). The corresponding σbs was 1 × 10−9 m2 (Equation 2): σbs =10TS10(2) For each deployment, the Lloyd’s patchiness index P (Lloyd, 1967) within the SL, IL, and DL was then calculated using Equation (3): P=ρv¯+s2ρv¯-1ρv¯(3) where ρv¯ represents the mean density of individuals within a given layer and s2 is the sample variance. P depends on the spatial distribution of scatterers and describes how many other individuals are in the sample relative to a random distribution. P < 1 indicates a uniform distribution, P = 1 corresponds to a random (i.e. Poisson) distribution, and P > 1 indicates an aggregating behaviour. The index increases with increased patchiness. For instance, P = 2 if individuals are twice as crowded compared with a random distribution (Lloyd, 1967; Houde and Lovdal, 1985; De Robertis, 2002). The spatial scale of patchiness measurements corresponds to the sampling scale, in our case 9–16 m horizontally (i.e. one ping) by 4 m vertically. Results Water masses and vertical distribution of pelagic scatterers At the northern sites (NB and WF), salinity and temperature in the SL were lower (S < 34.58, T < 5.07°C; Figure 3a and b) than at the southernmost locations (KF and IF), indicating less influence of AW. Backscatter values higher than the mean sv for the entire deployment were concentrated within the first 4 m and below the 1027.84 and 1027.63 kg·m−3 isopycnal lines, respectively (Figure 4a and b; left column). These water densities coincide with a stabilization in the density gradient profiles, and thus roughly correspond to the base of the pycnocline (BOP; Figure 4a and b; right column). In contrast, surface water (0–20 m) at KF and IF was more saline and warmer (S < 35.20, T < 6.60°C; Figure 3a and b) than at the northernmost locations, and backscatter values higher than average were concentrated above the 1027.7 and 1027.8 kg·m−3 isopycnal lines (Figure 4c–e; left column), which also roughly correspond to the BOP (Figure 4c–e; right column). Figure 3 Open in new tabDownload slide Indicative profiles of salinity (a) and temperature (b) reconstructed from the five AUV deployments. The vertical resolution of the profiles is 10 m. Figure 3 Open in new tabDownload slide Indicative profiles of salinity (a) and temperature (b) reconstructed from the five AUV deployments. The vertical resolution of the profiles is 10 m. Figure 4 Open in new tabDownload slide Left column: temperature-salinity diagrams for each deployment where the data points are the mean T-S value within a 4 m depth range corresponding to the ADCP bins. An isopycnal line (in kg·m−3) demarcating the 4 m bins with backscatter values higher (orange asterisks) and lower (black dots) than average is drawn. Right column: Vertical profiles of density gradient with a 4 m vertical resolution. The grey line is the depth of the isopycnal line in the left panel. Hatched orange lines indicate sections of the profiles with backscatter values higher than average. Note that the scale of the x-axis is one order of magnitude lower in (d) and (e). Figure 4 Open in new tabDownload slide Left column: temperature-salinity diagrams for each deployment where the data points are the mean T-S value within a 4 m depth range corresponding to the ADCP bins. An isopycnal line (in kg·m−3) demarcating the 4 m bins with backscatter values higher (orange asterisks) and lower (black dots) than average is drawn. Right column: Vertical profiles of density gradient with a 4 m vertical resolution. The grey line is the depth of the isopycnal line in the left panel. Hatched orange lines indicate sections of the profiles with backscatter values higher than average. Note that the scale of the x-axis is one order of magnitude lower in (d) and (e). No isolated dense echoes typical of fish schools were detected, supporting the idea that the pelagic scattering layers were mainly composed of zooplankton. The backscatter at NB remained low (<85 dB) from the surface to the maximum sampling depth of 150 m (Figure 5a), indicating low densities of scatterers. At WF, the backscatter reached maximal values at the surface, decreased down to 40 m, increased until 70 m, and decreased at greater depths (Figure 5b). At the southernmost locations (KF and IF), Sv values were significantly higher in the SL than in the IL and DL (Tuckey HSD; p < 0.001) (Figure 5c–e). Maximal backscatter occurred near the surface and decreased linearly with depth until 80 m (Sv = −0.2 × Depth-78.3; r2 = 0.73; p < 1 × 10–15; n = 60) (Figure 5f). Figure 5 Open in new tabDownload slide Profiles of volume backscattering strength (Sv in dB re 1 m–1) averaged over 4 m vertical bins. The dashed grey lines demarcate the SL, the IL, and the DL. Data from KF and IF are pooled in panel f, where a regression line was added for the SL and IL (dashed black line). Figure 5 Open in new tabDownload slide Profiles of volume backscattering strength (Sv in dB re 1 m–1) averaged over 4 m vertical bins. The dashed grey lines demarcate the SL, the IL, and the DL. Data from KF and IF are pooled in panel f, where a regression line was added for the SL and IL (dashed black line). The vertical distribution of the backscatter was similar at both southernmost locations, despite the fact that data were collected during midday at KF and around midnight at IF. Mean linear backscatter did not differ significantly within the SL (Kruskal-Wallis; p = 0.54) or the DL (Kruskal-Wallis; p = 0.63), although the median was slightly higher in the SL at midnight (Figure 6a and c). In the IL, mean backscatter was similar between the first deployment at KF (6 July) and the deployment at IF, but was significantly higher at KF on 12 July (Kruskal-Wallis; p = 0.007; Figure 6b). However, the backscatter variance was high for all deployments (Figure 6a–c). Figure 6 Open in new tabDownload slide Box plots comparing the average backscatter in linear form (m2·m−3) for deployments around midday (KF) and midnight (IF) in the (a) SL, (b) IL, and (c) DL. The black line is the median, bottom and top of the rectangle are lower and upper quartiles, bottom and top whiskers are minimum and maximum values (excluding the outliers). Empty dots are outliers (more than 1.5 times the upper quartile). Figure 6 Open in new tabDownload slide Box plots comparing the average backscatter in linear form (m2·m−3) for deployments around midday (KF) and midnight (IF) in the (a) SL, (b) IL, and (c) DL. The black line is the median, bottom and top of the rectangle are lower and upper quartiles, bottom and top whiskers are minimum and maximum values (excluding the outliers). Empty dots are outliers (more than 1.5 times the upper quartile). Positive and negative vertical velocity anomaly values (w’) were measured at all depths and all locations (Figure7a–e; left column). Upward movement (positive w’ values) of scatterers was mainly measured above 80 m at NB, 40 m at WF, and 90 m at IF, while downward migration (negative w’ values) was measured deeper (Figure 7a, b, e; right column). The direction was inverted at KF, with downward migration above 80 m (6 July) or 40 m (12 July) and upward movement at greater depths (Figure 7c, d; right column). Although time-averaged w’ measured within each 4 m changed between the surface layers and at depth, suggesting different migration directions, variance was high (typically ± 2 mm s−1) and average w’ values were low (typically much < ±1 mm·s−1) at all locations (Figure 7; right column). Figure 7 Open in new tabDownload slide Left column: vertical velocity anomalies (w’ in mm s−1) along the trajectory of the AUV (solid black line). Right column: Corresponding profiles of w’ with a resolution of 4 m (thick black lines) ± one standard deviation (grey polygons). The vertical dashed lines indicate 0 mm·s−1 and the horizontal dashed black lines demarcate the SL, the IL, and the DL. Figure 7 Open in new tabDownload slide Left column: vertical velocity anomalies (w’ in mm s−1) along the trajectory of the AUV (solid black line). Right column: Corresponding profiles of w’ with a resolution of 4 m (thick black lines) ± one standard deviation (grey polygons). The vertical dashed lines indicate 0 mm·s−1 and the horizontal dashed black lines demarcate the SL, the IL, and the DL. Density and patchiness The estimated mean density of scatterers at the northernmost locations was more uniform with depth compared with the southernmost sites (Figure 8; left column). The estimated density remained between 0.9 and 1.0 ind·m−3 at NB (Figure 8a; left column), and between 2.4 and 4.0 ind·m−3 at WF (Figure 8b; left column). At KF and IF, the estimated density varied from 9.1 to 13.6 ind·m−3 in the SL, from 2.2 to 6.3 ind·m−3 in the IL, and from 0.6 to 0.8 ind·m−3 at greater depths (Figure 8c–e; left column). Lloyd’s index of patchiness (P) was > 1 in the SL at all locations, indicating patchy distributions near the surface (Figure 8a–e; right column). Distributions were generally less patchy in the IL, and at NB the distribution was uniform in the IL (P < 1: Figure 8a; right column). In contrast, at KF the patchiness increased in the IL compared with the SL (Figure 8c–d; right column). When compared with the SL, patchiness in the DL decreased at all locations with uniform distributions at NB and IF (Figure 8a and e; right column). The patchiness index was over one order of magnitude higher in the SL at NB and WF than anywhere else, indicating ten times patchier distributions (Figure 8a and b; right column). Apart from these two observations, patchiness was significantly correlated with the density of scatterers (Spearman rank correlation; ρ = 0.56; p = 0.016) (Figure 9). Figure 8 Open in new tabDownload slide Left column: bar plots of the mean density of pelagic scatterers (ind·m−3) estimated for each layer. Right column: Corresponding bar plots of the Lloyd’s patchiness index (P) for each layer. The dashed grey lines indicate the limit between a uniform (P < 1) and a patchy distribution (P > 1). Note the cut in the x-axis for NB and WF. Figure 8 Open in new tabDownload slide Left column: bar plots of the mean density of pelagic scatterers (ind·m−3) estimated for each layer. Right column: Corresponding bar plots of the Lloyd’s patchiness index (P) for each layer. The dashed grey lines indicate the limit between a uniform (P < 1) and a patchy distribution (P > 1). Note the cut in the x-axis for NB and WF. Figure 9 Open in new tabDownload slide The Lloyd’s patchiness index (P) against the mean estimated density of pelagic scatterers (ind·m−3) for each layer of each deployment. Figure 9 Open in new tabDownload slide The Lloyd’s patchiness index (P) against the mean estimated density of pelagic scatterers (ind·m−3) for each layer of each deployment. Discussion The 3D trajectory of the AUV allowed documenting the 614 kHz backscatter from <1.5 m below the surface to vertical ranges up to 200 m (Figure 2). In comparison, the surface blind zone of ship-based surveys reaches ∼15 m (Scalabrin et al., 2009), and if a similar ADCP had been installed on a mooring at depth the vertical range would not have been greater than 40 m. The extended vertical and spatial ranges conferred by the 3D trajectory of the AUV allowed obtaining valuable insights into synchronized and unsynchronized vertical migrations during midnight sun, documenting the vertical distribution of pelagic scatterers in relation to hydrography, and demonstrating that their patchiness increased with the density of organisms. Synchronized and unsynchronized vertical migrations during midnight sun The vertical distributions of backscatter during midday and around midnight at the two southernmost locations were statistically similar (Figure 6) and interpreted as an absence of synchronized Diel Vertical Migration (DVM), as generally reported during periods of continuous illumination in the Arctic (Fischer and Visbeck, 1993; Blachowiak-Samolyk et al., 2006; Cottier et al., 2006). Although synchronized DVM does not generally occur during continuous illumination at high latitudes, an alternate behaviour of unsynchronized vertical migration, with animals migrating independently of each other in response to their individual needs, has been reported from May to July in Arctic fjord environments (Cottier et al., 2006; Wallace et al., 2010). This migration occurs continuously during a 24-h period and do not modify the total abundance of scatterers within each layer. However, unsynchronized migration can be identified in ADCP records when the mean direction of migration in the SL is downward (indicated by negative w’ values) and the mean direction of migration in the IL and DL is upward (indicated by positive w’ values; details in Cottier et al., 2006). In this study, mean values of w’ were positive (upward) in the SL and negative (downward) in the DL at most locations, except for KF where the opposite occurred. Even at KF, variance was high and w’ measurements were low compared with previous studies that have documented unsynchronized migration (e.g. −8 to 8 mm·s−1; Cottier et al., 2006). In contrast to previous observations in Arctic fjords, our data thus suggest that pelagic scatterers do not perform clear unsynchronized migration over the outer shelf during midnight sun. Accordingly, their contribution to the biological pump is likely reduced at that time of the year (Tarling and Johnson, 2006; Wallace et al., 2013). It is important to note that the period of averaging w’ during this study (<7 h) was <5% that of Cottier et al. (2006) and Wallace et al. (2013) (7 days). Given the high variance in w’, the detection of unsynchronized migratory behaviours of planktonic organisms may require longer duration surveys. Furthermore, as most AUVs cannot cover 24-h cycles, the detection of DVM in the Arctic using this technique is limited to comparisons between midday and midnight surveys. Hence, even though our results suggest an absence of unsynchronized and synchronized vertical migrations in the outer shelf environment during midnight sun, such migrations could possibly occur. Long-range AUVs (e.g. Hobson et al., 2012) were recently developed and they could overcome this issue by combining the benefits of AUVs to that of multi-day deployments on Eulerian platforms. Vertical distribution of pelagic scatterers in relation to hydrography Although the vertical distributions of pelagic organisms, in particular zooplankton, are mainly related to changes in light intensity, Berge et al. (2014) suggested that hydrographic structures can determine resting depth of zooplankton between migration events. As no vertical migrations were detected during this study, it is likely that other factors, including hydrography, influenced the vertical distribution of scatterers. With the exception of a few patchy aggregations in the top 4 m, scatterers at the northernmost locations were distributed below the pycnocline, as previously documented for Arctic fjords (Berge et al., 2014) and during laboratory experiments (Lougee et al., 2002). These small pelagic organisms likely avoided colder and fresher surface waters to remain in denser and deeper water masses, where higher viscosities require less energy to hold position (Harder, 1968) and temperatures are closer to thermal preferences (Berge et al., 2014). In contrast, density and temperature were higher at the southernmost locations so scatterers remained within and above the pycnocline. We surmise that discrepancies in vertical distributions of the pelagic scattering layers between the northernmost and southernmost locations derived in part from different hydrographic regimes, in addition to other factors such as variations in the zooplankton assemblages and in primary production (Blachowiak-Samolyk et al., 2008). Furthermore, this study supports the idea that the pycnocline acts as a physical barrier limiting vertical migrations of small pelagic organisms and contributing to their retention in either the SL or at depth (Lougee et al., 2002). Therefore, in addition to continuous solar irradiance, the strong density gradient prevailing during Arctic summer may contribute to the absence of vertical migrations between different water masses. Increased patchiness with density Due to increased spatial range, AUV-mounted ADCPs provide better spatial resolution of patchiness than moored ADCPs (e.g. Brierley et al., 2006) or multi-net samplers (e.g. Vogedes et al., 2014). Our results are nonetheless consistent with previous observations of an aggregating behaviour for Calanus spp. in Isfjorden in July (Vogedes et al., 2014). However, our mean density estimates remained below 14 ind·m−3, while previous plankton net-based studies conducted in fjords reported zooplankton densities from 76 to >200 ind·m−3 in the first 100 m of the water column (Kwasniewski et al., 2003; Cottier et al., 2006; Berge et al., 2014). These results suggest considerably lower abundances of pelagic scatterers over the outer shelf than within fjords, supporting previous work by Daase and Eiane (2007) in northern Spitsbergen. If patchiness increases with density (Figure 9), then patchy aggregations are expected to be more abundant in fjords compared with outer shelf locations. Lloyd (1967) developed the patchiness index P (Equation 3) to study the “mean crowding” of animals or plants. In the marine environment, the index proved useful to document the patchiness of fish eggs and ichthyoplankton (e.g. McQuinn et al., 1983; Houde and Lovdal, 1985; Maynou et al., 2006) and zooplankton (e.g. George, 1981; De Robertis, 2002; Greer et al., 2013). Bez (2000) indicated that the Lloyd’s patchiness index is biased when calculated from densities rather than counts, as in this study. Nonetheless, by comparing the index calculated from zooplankton backscatter data (density) with P computed from the total number of targets in a simulated acoustic image (counts), De Robertis (2002) demonstrated that, despite sampling biases resulting in conservative values, P can efficiently be used as a measure of aggregation at low target densities, such as those observed here. Biases could also originate from the average cross section of scatterers used for calculations, which was based on the average copepod cross section at Kongsfjorden (Cottier et al., 2006). The mean cross section (⁠ σbs ⁠) of scatterers could have been different offshore, which would have biased density and patchiness calculations. The patchiness index calculated here nonetheless provides a relative measure between vertical layers (SL, IL, and DL) and acts as a baseline indicator for the patchiness of pelagic organisms in the Arctic. The scatterers exhibited a strong aggregating behaviour, most likely to dilute predation risk by visual predators, maximize food capture, and optimize energy expenditure (Folt and Burns, 1999; Ritz, 2000). The very high patchiness indices in the SL at NB and WF resulted from a generally low density with few dense and small aggregations just below the surface (Figure 2a and b; left column), although patchiness generally increased with scatterer density. Patchiness may also partly explain the significantly higher backscatter in the IL at KF on 12 July compared with 6 July (Figure 6b), as a non-uniform distribution is likely to result in variations among deployments. Another possible explanation for variations in density and patchiness in the IL between deployments at KF might be the paucity of samples at certain depths on 06 July. Some sections of the water column were then only surveyed during ascent or descent of the AUV and patches of zooplankton or micronekton could have been missed (Figure 2c and d). At small scales (metres), physical turbulence can also determine the spatial distribution of pelagic organisms and facilitates the formation of aggregations (Mackas et al., 1997; De Robertis, 2002). During the survey, turbulence was higher in the SL and decreased with depth (Steele et al., 2012). Because patchiness followed a similar trend, it is possible that it was correlated with turbulence, in addition to the density of scatterers. Conclusions The use of an AUV allowed investigating key aspects of the distribution and behaviour of Arctic pelagic organisms over larger spatial scales than previously reported. The AUV also enabled measurements of additional spatial variables, such as patchiness indices. This study supports the hypothesis that, in the absence of vertical migrations, hydrographic structures influence vertical distributions of pelagic organisms on a regional scale. In particular, the pycnocline could represent a physical barrier that retains organisms in either the surface layer or below the strongest density gradient. Scatterers consistently formed patchy aggregations in the top 20 m, which stresses both the ecological importance of this layer for predators and the need for prudent interpretations when calculating abundances from stationary net deployments. AUV-based acoustic surveys of the pelagic communities are complementary to Eulerian studies, for instance by providing spatial measurements of patchiness. The 3D trajectory of AUVs allows approaching targets sufficiently close to use high frequency acoustic instruments with high size resolution and, by reducing the surface blind zone to <1.5 m, enables detection of aggregations close to the surface. However, future surveys of vertical migrations by planktonic organisms would benefit from the deployment of long-range AUVs to cover several daily cycles. Acknowledgements This work results from an internship financially supported by Québec-Océan at Université Laval, Canada, and hosted by the Scottish Association for Marine Science. We thank the Master and crew of the RSS James Clark Ross and the scientists and technical support staff of Cruise JR219, in particular Tim Boyd, Colin Griffiths and Estelle Dumont for AUV deployments. Field work was funded by the UK Natural Environment Research Council (under the Oceans 2025 programme and National Capability support for the Scottish Marine Robotics Facility). We are also grateful to Matt Toberman and Laura Hobbs for advice while developing the data processing algorithm, and to Gérald Darnis for reviewing the article. This study is a contribution to the Norges Forskningsråd project Arctic ABC number 244319. References An E. , Dhanak M. R., Shay L. K., Smith S., Van Leer J. 2001 . Coastal oceanography using a small AUV . Journal of Atmospheric and Oceanic Technology , 18 : 215 – 234 . Google Scholar Crossref Search ADS WorldCat Benoit D. , Simard Y., Fortier L. 2008 . Hydroacoustic detection of large winter aggregations of Arctic cod (Boreogadus saida) at depth in ice-covered Franklin Bay (Beaufort Sea) . Journal of Geophysical Research: Oceans , 113 . OpenURL Placeholder Text WorldCat Benoit-Bird K. J. 2009 . The effects of scattering-layer composition, animal size, and numerical density on the frequency response of volume backscatter . ICES Journal of Marine Science , 66 : 582 – 593 . Google Scholar Crossref Search ADS WorldCat Berge J. , Batnes A. S., Johnsen G., Blackwell S. M., Moline M. A. 2012 . Bioluminescence in the high Arctic during the polar night . Marine Biology , 159 : 231 – 237 . Google Scholar Crossref Search ADS PubMed WorldCat Berge J. , Cottier F., Varpe Ø., Renaud P. E., Falk-Petersen S., Kwasniewski S., Griffiths C., et al. 2014 . Arctic complexity: a case study on diel vertical migration of zooplankton . Journal of Plankton Research , 36 : 1279 – 1297 . Google Scholar Crossref Search ADS PubMed WorldCat Bez N. 2000 . On the use of Lloyd's index of patchiness . Fisheries Oceanography , 9 : 372 – 376 . Google Scholar Crossref Search ADS WorldCat Blachowiak-Samolyk K. , Søreide J. E., Kwasniewski S., Sundfjord A., Hop H., Falk-Petersen S., Hegseth E. N. 2008 . Hydrodynamic control of mesozooplankton abundance and biomass in northern Svalbard waters (79-81 degrees N) . Deep-Sea Research Part II , 55 : 2210 – 2224 . Google Scholar Crossref Search ADS WorldCat Blachowiak-Samolyk K. , Kwasniewski S., Richardson K., Dmoch K., Hansen E., Hop H., Falk-Petersen S., Mouritsen L. T. 2006 . Arctic zooplankton do not perform diel vertical migration (DVM) during periods of midnight sun . Marine Ecology Progress Series , 308 : 101 – 116 . Google Scholar Crossref Search ADS WorldCat Boyd T. , Inall M., Dumont E., Griffiths C. 2010 . AUV observations of mixing in the tidal outflow from a Scottish sea loch. In Autonomous Underwater Vehicles (AUV) , pp. 1 – 9 . IEEE . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Brierley A. S. , Fernandes P. G., Brandon M. A., Armstrong F., Millard N. W., McPhail S. D., Stevenson P., et al. 2002 . Antarctic krill under sea ice: Elevated abundance in a narrow band just south of ice edge . Science , 295 : 1890 – 1892 . Google Scholar Crossref Search ADS PubMed WorldCat Brierley A. S. , Saunders R. A., Bone D. G., Murphy E. J., Enderlein P., Conti S. G., Demer D. A. 2006 . Use of moored acoustic instruments to measure short-term variability in abundance of Antarctic krill . Limnology and Oceanography Methods , 4 : 18 – 29 . Google Scholar Crossref Search ADS WorldCat Cottier F. R. , Tarling G. A., Wold A., Falk-Petersen S. 2006 . Unsynchronized and synchronised vertical migration of zooplankton in a high Arctic fjord . Limnology and Oceanography , 51 : 2586 – 2599 . Google Scholar Crossref Search ADS WorldCat Cottier F. R. , Tverberg V., Inall M., Svendsen H., Nilsen F., Griffiths C. 2005 . Water mass modification in an Arctic fjord through cross-shelf exchange: The seasonal hydrography of Kongsfjorden, Svalbard . Journal of Geophysical Research: Oceans , 110 : C12005. Google Scholar Crossref Search ADS WorldCat Cottier F. R. , Venables E. J. 2007 . On the double-diffusive and cabbeling environment of the Arctic Front, West Spitsbergen . Polar Research , 26 : 152 – 159 . Google Scholar Crossref Search ADS WorldCat Daase M. , Eiane K. 2007 . Mesozooplankton distribution in northern Svalbard waters in relation to hydrography . Polar Biology , 30 : 969 – 981 . Google Scholar Crossref Search ADS WorldCat Darnis G. , Fortier L. 2014 . Temperature, food and the seasonal vertical migration of key Arctic copepods in the thermally stratified Amundsen Gulf (Beaufort Sea, Arctic Ocean) . Journal of Plankton Research , 36 : 1092 – 1108 . Google Scholar Crossref Search ADS WorldCat Deines K. L. 1999 . Backscatter estimation using broadband acoustic Doppler current profilers. In Current measurement , pp. 249 – 253 . IEEE . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC De Robertis A. 2002 . Small-scale spatial distribution of the euphausiid Euphausia pacifica and overlap with planktivorous fishes . Journal of Plankton Research , 24 : 1207 – 1220 . Google Scholar Crossref Search ADS WorldCat Eisner L. , Hillgruber N., Martinson E., Maselko J. 2013 . Pelagic fish and zooplankton species assemblages in relation to water mass characteristics in the northern Bering and southeast Chukchi seas . Polar Biology 36 : 87 – 113 . Google Scholar Crossref Search ADS WorldCat Falk-Petersen S. , Pavlov V., Timofeev S., Sargent J. R. 2007 . Climate variability and possible effects on Arctic food chains: the role of Calanus. In Arctic alpine ecosystems and people in a changing environment , pp. 147 – 166 . Ed. by Ørbæk J. B., Kallenborn R., Tombre I., Hegseth E. N., Falk-Petersen S., Hoel A. H.. Springer , New York . 433 pp. Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Fernandes P. G. , Stevenson P., Brierley A. S., Armstrong F., Simmonds E. J. 2003 . Autonomous underwater vehicles: future platforms for fisheries acoustics . ICES Journal of Marine Science , 60 : 684 – 691 . Google Scholar Crossref Search ADS WorldCat Fischer J. , Visbeck M. 1993 . Seasonal variation of the daily zooplankton migration in the Greenland Sea . Deep Sea Research Part I , 40 : 1547 – 1557 . Google Scholar Crossref Search ADS WorldCat Folt C. L. , Burns C. W. 1999 . Biological drivers of zooplankton patchiness . Trends in Ecology and Evolution , 14 : 300 – 305 . Google Scholar Crossref Search ADS PubMed WorldCat Geoffroy M. , Majewski A., LeBlanc M., Gauthier S., Walkusz W., Reist J. D., Fortier L. 2016 . Vertical segregation of age-0 and age-1+ polar cod (Boreogadus saida) over the annual cycle in the Canadian Beaufort Sea . Polar Biology , 39 : 1023 – 1037 . Google Scholar Crossref Search ADS WorldCat George D. 1981 . Zooplankton patchiness . Report from the Freshwater Biology Association , 49 : 32 – 44 . OpenURL Placeholder Text WorldCat Greer A. T. , Cowen R. K., Guigand C. M., Mcmanus M. A., Sevadjian J. C., Timmerman A. H. V. 2013 . Relationships between phytoplankton thin layers and the fine-scale vertical distributions of two trophic levels of zooplankton . Journal of Plankton Research , 35 : 939 – 956 . Google Scholar Crossref Search ADS WorldCat Harder W. 1968 . Reaction of plankton organisms to water stratification . Limnology and Oceanography , 13 : 156 – 168 . Google Scholar Crossref Search ADS WorldCat Hobson B. W. , Bellingham J. G., Kieft B., McEwen R., Godin M., Zhang Y. 2012 . Tethys-class long-range AUVs-extending the endurance of propeller-driven cruising AUVs from days to weeks. In Autonomous Underwater Vehicles (AUV) , pp. 1 – 8 . IEEE . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Houde E. D. , Lovdal J. D. A. 1985 . Patterns of variability in ichthyoplankton occurence and abundance in Biscayne Bay, Florida . Estuarine, Coastal and Shelf Science , 20 : 79 – 103 . Google Scholar Crossref Search ADS WorldCat Kwasniewski S. , Hop H., Falk-Petersen S., Pedersen G. 2003 . Distribution of Calanus species in Kongsfjorden, a glacial fjord in Svalbard . Journal of Plankton Research , 25 : 1 – 20 . Google Scholar Crossref Search ADS WorldCat La H. S. , Kang M., Dahms H. U., Ha H. K., Yang E. J., Lee H., Kim Y. N., et al. 2015 . Characteristics of mesozooplankton sound-scattering layer in the Pacific Summer Water, Arctic Ocean . Deep Sea Research Part II , 120 : 114 – 123 . Google Scholar Crossref Search ADS WorldCat Last K. S. , Hobbs L., Berge J., Brierley A. S., Cottier F. 2016 . Moonlight drives ocean-scale mass vertical migration of zooplankton during the Arctic winter . Current Biology , doi: 10.1016/j.cub.2015.11.038 OpenURL Placeholder Text WorldCat Lloyd M. 1967 . Mean crowding . Journal of Animal Ecology , 36 : 1 – 30 . Google Scholar Crossref Search ADS WorldCat Lorke A. , McGinnis D. F., Spaak P., Wueest A. 2004 . Acoustic observations of zooplankton in lakes using a Doppler current profiler . Freshwater Biology , 49 : 1280 – 1292 . Google Scholar Crossref Search ADS WorldCat Lougee L. A. , Bollens S. M., Avent S. R. 2002 . The effects of haloclines on the vertical distribution and migration of zooplankton . Journal of Experimental Marine Biology and Ecology , 278 : 111 – 134 . Google Scholar Crossref Search ADS WorldCat Mackas D. L. , Kieser R., Saunders M., Yelland D. R., Brown R. M., Moore D. F. 1997 . Aggregation of euphausiids and Pacific hake (Merluccius productus) along the outer continental shelf off Vancouver Island . Canadian Journal of Fisheries and Aquatic Sciences , 54 : 2080 – 2096 . Google Scholar Crossref Search ADS WorldCat Maynou F. , Olivar M. P., Emelianov M. 2006 . Patchiness of eggs, larvae and juveniles of European hake Merluccius merluccius from the NW Mediterranean . Fisheries Oceanography , 15 : 390 – 401 . Google Scholar Crossref Search ADS WorldCat McQuinn I. H. , Fitzgerald G. J., Powles H. 1983 . Environmental effects on embryos and larvae of the Isle Verte stock of Atlantic herring (Clupea harengus harengus) . Le Naturaliste Canadien , 110 : 343 – 353 . OpenURL Placeholder Text WorldCat Parker-Stetter S. L. , Rudstam L. G., Sullivan P. J., Warner D. M. 2009 . Standard operating procedures for fisheries acoustic surveys in the Great Lakes , 1st edn. Great Lakes Fishery Commission , Ann Arbor , 170 . pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Ritz D. A. 2000 . Is social aggregation in aquatic crustaceans a strategy to conserve energy? . Canadian Journal of Fisheries and Aquatic Sciences , 57 : 59 – 67 . Google Scholar Crossref Search ADS WorldCat Saloranta T. A. , Haugan P. M. 2001 . Interannual variability in the hydrography of Atlantic water northwest of Svalbard . Journal of Geophysical Research , 106 : 931 – 943 . Google Scholar Crossref Search ADS WorldCat Scalabrin C. , Marfia C., Boucher J. 2009 . How much fish is hidden in the surface and bottom acoustic blind zones? . ICES Journal of Marine Science , 66 : 1355 – 1363 . Google Scholar Crossref Search ADS WorldCat Schofield O. , Glenn S., Orcutt J., Arrott M., Meisinger M., Gangopadhyay A., Brown W., et al. 2010 . Automated sensor network to advance ocean science . Eos, Transactions American Geophysical Union , 91 : 345 – 346 . Google Scholar Crossref Search ADS WorldCat Stanton T. K. , Wiebe P. H., Chu D., Benfield M. C., Scanlon L., Martin L., Eastwood R. L. 1994 . On acoustic estimates of zooplankton biomass . ICES Journal of Marine Science , 51 : 505 – 512 . Google Scholar Crossref Search ADS WorldCat Steele E. , Boyd T., Inall M., Dumont E., Griffiths C. 2012 . Cooling of the West Spitsbergen Current: AUV-based turbulence measurements west of Svalbard. In Autonomous Underwater Vehicles (AUV) , pp. 1 – 7 . IEEE . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Svendsen H. , Beszczynska, Møller A., Hagen J. O., Lefauconnier B., Tverberg V., Gerland S., Ørbøk J. B. 2002 . The physical environment of Kongsfjorden–Krossfjorden, an Arctic fjord system in Svalbard . Polar Research , 21 : 133 – 166 . OpenURL Placeholder Text WorldCat Tarling G. A. , Johnson M. L. 2006 . Satiation gives krill that sinking feeling . Current Biology , 16 : 83 – 84 . Google Scholar Crossref Search ADS WorldCat Valle-Levinson A. , Castro L., Cáceres M., Pizarro O. 2014 . Twilight vertical migrations of zooplankton in a Chilean fjord . Progress in Oceanography , 129 : 114 – 124 . Google Scholar Crossref Search ADS WorldCat Vogedes D. , Eiane K., Batnes A. S., Berge J. 2014 . Variability in Calanus spp. abundance on fine- to mesoscales in an Arctic fjord: implications for little auk feeding . Marine Biology Research , 10 : 437 – 448 . Google Scholar Crossref Search ADS WorldCat Wallace M. I. , Cottier F. R., Berge J., Tarling G. A., Griffiths C., Brierley A. S. 2010 . Comparison of zooplankton vertical migration in an ice-free and a seasonally ice-covered Arctic fjord: an insight into the influence of sea ice cover on zooplankton behaviour . Limnology and Oceanography , 55 : 831 – 845 . Google Scholar Crossref Search ADS WorldCat Wallace M. I. , Cottier F. R., Brierley A. S., Tarling G. A. 2013 . Modelling the influence of copepod behaviour on faecal pellet export at high latitudes . Polar Biology , 36 : 579 – 592 . Google Scholar Crossref Search ADS WorldCat Author notes " † Please note that this paper should have published in the 6th Zooplankton Production Symposium, in volume 74, issue 7. The publisher apologizes for this error. © International Council for the Exploration of the Sea 2016. All rights reserved. For Permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Vollset, Knut, Wiik;Halttunen,, Elina;Finstad,, Bengt;Karlsen,, Ørjan;Bjørn, Pål, Arne;Dohoo,, Ian

doi: 10.1093/icesjms/fsx090pmid: N/A

Abstract Impacts of sea lice (Lepeophtheirus salmonis or Caligus spp.) on wild salmonids is currently one of the most important issues facing management of fish farms in salmon producing countries in the northern hemisphere. Surveillance of sea lice on wild Atlantic salmon (Salmo salar) is often hampered by the ability to catch enough migrating post-smolts. Therefore, sea lice abundance on anadromous trout (Salmo trutta) is often used to infer sea lice abundance on migrating salmon post-smolt. However, the assumption that there is a relationship between the abundance of lice on salmon and trout has never been tested. Here we use a dataset of sea lice on salmon post-smolt and sea trout that have been caught simultaneously in trawl hauls, to evaluate the correlation in abundance of sea lice between the two species using various statistical models. We demonstrate that trout generally has higher abundances of sea lice than salmon. Average lice per gram fish on sea trout (log transformed) predicted the abundance of lice on salmon best. Negative binomial models of lice counts were preferable to using trout lice counts as direct estimates of salmon lice abundance, and they had better predictive ability than logit models of high (vs. low) lice counts. Including the size of the salmon increased the predictive ability of the model, but these data are not generally available. The effect of salmon weight may have been a direct effect of body size, or an indirect effect of time spent in marine waters. Finally, we predict lower salmon lice counts on migrating salmon with our selected binomial model than with the current method of using trout lice counts as a direct estimator on salmon lice counts, and demonstrate that management advice would change considerably depending on the chosen method. Introduction The impact of sea lice (here referring to Lepeophtheirus salmonis and Caligus spp.) on wild salmon stocks are currently one of the most controversial topics in the debate surrounding the marine survival and conservation of salmon species. Sea louse (or salmon louse when only referring to the most prevalent lice on salmonids, L. salmonis) is an ecto-parasite that attaches to the surface of the skin of salmonids and creates lesions which causes osmoregulatory stress for the host (Grimnes and Jakobsen, 1996; Finstad et al., 2000; Wagner et al., 2008). This can lead to secondary infections (Wells et al., 2007), higher risk of predation mortality (Godwin et al., 2015; Peacock et al., 2015), or in conditions of high infection levels, death (Birkeland, 1996; Grimnes et al., 1996; Bjoern and Finstad, 1997, 1998; Finstad and Grimnes, 1997; Finstad et al., 2000). The reason for the controversy related to the interaction between lice and salmon is that salmon farms have been identified as an important source of infestations on wild fish (Bjorn et al., 2011; Serra-Llinares et al., 2014, 2016). This can lead to unnaturally high levels of lice, and affect the seasonal timing of infestations (Krkosek et al., 2006; Vollset and Barlaup, 2014) and disrupt the natural migratory allopatry of adult and young fish which helps protect young fish from infestations (Krkosek et al., 2006). Moreover, several studies have demonstrated that lice can lead to reduction of returning adult spawners (Krkosek et al., 2013; Skilbrei et al., 2013; Vollset et al., 2015), although the magnitude of this impact on population levels has been debated (Jackson et al., 2014; Krkošek et al., 2014). In addition, sub-lethal effects of salmon lice such as reduced growth and increased age at return have also been demonstrated (Skilbrei et al., 2013; Vollset et al., 2014). For this reason Norwegian salmon farms are obliged to adhere to strict national regulations, not exceeding a certain number of female or mobile lice and in some regions conduct coordinated delousing to keep infestation pressure on migrating wild post-smolts to a minimum (Torrissen et al., 2013). To monitor infestation levels of lice, wild fish are caught using trawling with specialized equipment (FISH-LIFT, Holst and McDonald, 2000), trap nets (Barlaup et al., 2013), seine nets or gill-nets (Bjorn et al., 2011). These levels can then be compared with threshold levels of lice which are believed to be physiological harmful or lethal based on laboratory studies, modelling studies or expert opinions (or a combination of the three) (Taranger et al., 2015). In the case of Atlantic salmon (Salmo salar), which in the Atlantic is the species with the highest cultural, recreational and commercial interest, direct sampling of migratory young salmon is difficult, costly and time consuming. Most efficient is the use of pelagic trawls using specialized equipment (FISH-LIFT, Holst and McDonald, 2000). However, even though salmon post-smolts are caught most years, most of the trawls are often done over only a few days, usually whenever the boat happens to encounter an aggregation of migrating post-smolts (Finstad et al., 2000). Consequently, these samples are usually lumped together in time and space and thus are not representative of the entire smolt-run. In addition, due to cost and time, trawling data are only available from a select few fjords, and in some years catches are very small or absent. Sea trout, on the other hand, are easily catchable most likely because they have different marine behavior compared with Atlantic salmon. In contrast to salmon, who swim relatively fast off-shore, sea trout stay close to the coast throughout their marine sojourn where they are easily catchable using low cost methods such as gill nets or trap nets (Thorstad et al., 2016). The availability of sea trout and the lack of good data from migratory salmon post-smolts, has forced management to utilize salmon lice levels on sea trout as a proxy indicator for infestation levels of salmon. For example, Taranger et al. (2015) suggested a method of estimating percent likelihood of survival for salmon based on lice levels on sea trout during the time of migration of salmon post-smolt. However, to date no data has been presented that can test if this relationship is valid. To test the assumption that lice levels of sea trout reflect the lice levels on salmon we collected trawl data where salmon and sea trout has been caught simultaneously and attempted to correlate different parasite measurement across the two species. Thus, this is the first attempt to validate the assumption that sea trout can be used as a proxy indicator for parasite load on Atlantic salmon post-smolts. Material and methods Sampling The Institute of Marine Research (IMR) and the Norwegian Institute for Nature Research are, to the best of our knowledge, the only two institutions that have conducted trawling for post-smolts in Norway. All available data were obtained from these two institutions and it was evident that both trawling methods and data recording had changed over time. However, trawling was performed every year when wild salmon post-smolts migrate from the rivers to the sea during weeks 18–28 in years 1998–2015. A specially designed FISH-LIFT trawl (Holst and McDonald, 2000) was used for sampling in order to avoid loss of both scales and lice on the caught fish. FISH–LIFT sorts the fish into a closed aquarium, connected to the trawl cod-end, such that it allows large numbers of fish to continue to swim unharmed in the aquarium once caught. The ca. 15 m long trawl was deployed once or twice per day with speeds of ca. 3 knots ∼4 h at a time, for distances of 6–20 nautical miles. The trawled stretch varied according to the weather, currents and the amount of by-catch. In order to sample fish which represented the accumulated infestation over the whole fjord migration route, the trawling was concentrated in the outer parts of the sampled fjords, as close as possible to the shore line. The fish were transferred from the floating aquarium to a basin on board the research vessel, retrieved rapidly with a small hand held net, put into individual plastic bags, and euthanized with a blow to the head. The lice on the fish were counted either immediately, or frozen and counted later in the lab. The fish were handled with utmost care to avoid loss of scales and lice. However, apparent scale loss or other injuries were noted along with other individual characteristics (species, length and weight) during lice counting. Stages of salmon lice were distinguished to the stages copepodites, chalimus, preadult, adult male and adult female. The results are reported yearly by IMR (Svåsand et al., 2015; Nilsen et al., 2016). Data selection criteria for analysis The original data file contained records from the years 1998 to 2015 in nine regions (Hardangerfjorden, Trondheimsfjorden, Namsenfjorden, Osterfjorden, Sognefjorden, Frohavet, Nordfjorden, Altafjorden, Malangen). In some of these regions trawling was not done consistently in the same geographic location and the definition of a region is therefore somewhat ambiguous. In the data from the later periods, the exact geographic track of the trawls was available, in others, only the start and stop location and time was available. In addition, in some cases the exact location of the trawling was not possible to identify in the old data. For simplicity and due to relatively few data points we have decided to broadly categorize the trawl hauls into “fjords” where the trawling had taken place. Furthermore, while the catches of trawls consisted of several trawl hauls, we decided to pool data within a week in a year in order to get sufficient fish number at each data points. As a result, the data were pooled into “groups” which consisted of all fish caught in 1 week within a given fjord and our research objective was to compare lice counts from trout and salmon within the same group. Reviewing the original data it was also clear that trout catch data were in many cases missing. Personnel involved in trawling indicated that there had been some inconsistency in whether or not trout had been kept for lice counting. This was especially evident in the older data, where no data on trout existed. After excluding groups without trout or salmon counts, the remaining dataset contained data from weeks 18–28 in years 2004–2015 in six fjords (Hardangerfjorden, Trondheimsfjorden, Namsenfjorden, Nordfjorden, Osterfjorden, Sognefjorden). In all these data trawling had been conducted in the outer part of the fjord. Salmon louse, which is an external parasite, can for various reasons be scraped off during trawling or handling. To avoid using individuals with high loss of salmon lice we excluded individuals that had a documented scale loss of >50% (n = 122). Furthermore, in some instances only total lice were recorded (not divided into attached and mobile), or weight and length of the fish was not recorded. These cases were all excluded (n = 254). Another 15 trout that were very large >750 g were also excluded. The final dataset contained 99 groups from 6 fjords over a period of 12 years (2004–2015) with a total of 2762 individuals (2474 salmon and 288 trout). For statistical reasons, our final exclusion criterion was that each group should contain at least three specimens from each species. This final criterion reduced the dataset to 316 salmon and 228 trout, and the number of fjords was reduced to 4 (Hardangerfjorden, Trondheimsfjorden, Sognefjorden and Namsenfjorden) over a period of 10 years. Data analysis Descriptive statistics were computed and quantile plots used to compare the distributions of trout and salmon lice counts. A scatterplot of group mean counts (salmon vs. trout) was used to explore the unconditional relationship in the final dataset. In general, our goal was to attempt to predict the number of lice on individual salmon post-smolt based on the average number of lice on trout in that group. To correct for the fact that different groups had very different number of trout records (meaning the predictor was measured with highly variable precision), each data point was weighted according to the number of trout records in the group. Three types of statistical models were fitted to the data; (i) negative-binomial, (ii) logistic, and (iii) linear. In all models, a random effect for group (i.e. each week/fjord/year combination) was included to account for the lack of independence among lice counts on salmon. In the negative-binomial models the number of lice per salmon was the response variable. Predictors evaluated are described below. To evaluate the predictive ability of the negative binomial model, correlations between observed and predicted salmon lice counts were computed and scatterplots created. Diagnostics plots were used to evaluate the normality and homoscedasticity of the random effects and residuals were examined for extreme values. The logistic model was used to estimate the odds that a salmon had a lice count above 0.1 lice/g. This threshold was chosen because it is believed to be the threshold were salmon post-smolts first experience physiological impacts from salmon lice, once the lice develops into mobile stages (Wagner et al., 2008). In practical terms, this would mean 2 lice on 20 g salmon post-smolt, or 5 lice on a 50 g salmon post-smolt. To present the predictive ability of the logistic model, sensitivity (Se) and specificity (Sp)were calculated for the raw data and for the best fitting model along a range of cut-points. Model diagnostics were similar to those for the negative binomial model. In the linear model we attempted to predict the lice per gram for salmon post-smolt. However, these models were discarded for a couple of reasons. It was not possible to appropriately weight the trout data in the multilevel linear model, and the assumption of normality of residuals at all levels was not well met, so no further results from these models are presented. In total seven competing models with different predictors were used to predict the lice levels of salmon. These included: ∼average total lice per trout ∼log (average total lice per trout) ∼log (average (total lice per trout/trout weight)) ∼log (average (total lice per trout/trout weight)) + salmon weight ∼log (average attached lice per trout) ∼log (average (attached lice per trout/trout weight)) ∼log (average (attached lice per trout/trout weight)) + salmon weight Given that the predictors were all continuous, the linearity of the relationships between each predictor and the relevant outcome was evaluated using lowess smoothed curves and by adding quadratic terms to the models. For the negative binomial models, results from models in which the trout lice counts were included as linear and quadratic functions are presented. Finally, we also modelled attached lice on salmon post-smolt, replicating all models that were built for the total lice counts on salmon post-smolt. However, these results were almost identical to the results using total lice. The reason for this was that total and attached lice counts were very highly correlated (rho = 0.976). Consequently, the results from the modelling exercise using attached lice are not presented here. Estimates of model fit (Akaike information criterion; AIC and r2) may not reliably reflect the predictive ability of the model because the estimates were based on the same data used to build the model. In order to validate the model a cross-validation procedure of the best fitting model was carried out as follows. The selected model (M3) was fit using data from 20 of the 21 groups and this model was then used to predict values in the one group omitted. The procedure was repeated until predicted lice counts were obtained for all groups in the dataset. The correlation between these values and the observed mean salmon lice counts was computed. Implications for management advice In Norway, sea trout captured in trap-nets are used to predict population level effects in both salmon and trout populations through a simple multinomial relationship between lice per gram fish and likelihood of mortality (in percent) due to the lice infestation (Svåsand et al., 2015; Taranger et al., 2015). In epidemiology, this percent mortality is referred to as the attributable fraction (i.e. the reduction in the probability of survival given that the individual does not die from another cause). Samples of trout that are assumed to represent salmon post-smolt (mostly taken in trap-nets or gill nets) are taken during the period of time when salmon are thought to be migrating. For salmon, only lice counts from trout under 150 g are used. These counts are divided into four categories based on number of lice/gram fish weight (0–0.1, 0.1–0.2, 0.2–0.3, >0.3) that are assumed to be related to probabilities of mortality of 0, 20, 50, and 100%, respectively. A weighted average mortality for the entire sample is then calculated. These overall estimates (deemed “population level effects”) are further categorized according to assumed sustainability (0–10% green, 10–30% yellow, >30% red) (Taranger et al., 2012). In contrast, our approach attempts to use the average number of lice on trout as a predictor of number of lice on salmon post-smolt, instead of using the trout counts directly as a predictor of lice on salmon. To illustrate what effect our approach would have on management advice, we used an independent dataset of trout collected with a trap net in 2009–2015 in the Herdlefjorden outside Bergen (https://doi.pangaea.de/10.1594/PANGAEA.873663) to calculate the “population level effects” as calculated in the management system (see above). The sampling area was the outer fjords of the Osterfjord, where several salmon populations migrate past, and it is believed that this is the region where migrating salmon mainly encounters salmon lice (Vollset et al., 2016). The estimated “population level effects” derived directly from trout lice counts was compared with those based on predicted salmon lice counts derived from our best-fitting negative binomial model. Results In the original dataset (numbers of salmon and trout, from nine regions over a 17-year period) most fish had zero lice and the two species had comparable maximum lice counts (salmon max = 177, trout max = 189). However, the trout had clearly fewer zeroes and overall higher lice counts (Figure 1). This finding corresponds to earlier observations that sea trout generally have higher levels of lice than salmon (Nilsen et al. 2016). The max lice counts in the subset of data used in the analysis (316 salmon and 228 trout from four fjords over a 10-year period) were somewhat lower (salmon max = 104, trout max = 106). Figure 2 presents the quantile plot comparison of all the remaining study groups. Figure 3 presents a scatter plot of the group average lice counts of salmon vs. trout. Figure 1. Open in new tabDownload slide Quantile plot of all salmon and trout lice counts in original dataset. Data from 2474 salmon and 288 trout in 98 groups. Figure 1. Open in new tabDownload slide Quantile plot of all salmon and trout lice counts in original dataset. Data from 2474 salmon and 288 trout in 98 groups. Figure 2. Open in new tabDownload slide Quantile plots of all salmon and trout lice counts from dataset used in analysis. Data from 316 salmon and 228 trout in 21 groups which contained a minimum of 3 salmon and 3 trout. Figure 2. Open in new tabDownload slide Quantile plots of all salmon and trout lice counts from dataset used in analysis. Data from 316 salmon and 228 trout in 21 groups which contained a minimum of 3 salmon and 3 trout. Figure 3. Open in new tabDownload slide Scatterplot of salmon total lice counts vs. trout counts from dataset used in analysis. Point size represents weight assigned to point. Data from 316 salmon and 228 trout in 21 groups which contained a minimum of 3 salmon and 3 trout. Figure 3. Open in new tabDownload slide Scatterplot of salmon total lice counts vs. trout counts from dataset used in analysis. Point size represents weight assigned to point. Data from 316 salmon and 228 trout in 21 groups which contained a minimum of 3 salmon and 3 trout. Negative binomial models The predictors, AIC and r2 for the seven models are presented in Table 1. The table also includes the results from models with quadratic terms for the various measures of lice count, as these rendered a slightly lower AIC and (in most cases a slightly higher r2). The main results can be summarized as follows; (i) log transforming the counts on trout clearly increased the model fit, (ii) using lice per gram trout was a better predictor than using only counts of lice as a predictor, (iii) correcting for the size of the salmon increased the predictive ability of the model, and (iv) using attached lice per trout instead of total lice (M5–M7) gave a better prediction before correcting for size of the trout but a poorer prediction when correcting for size of the trout. The correlation coefficient (r2) between observed and fitted values for the best model (M4) was 0.72 for the linear model and 0.74 for the model including the quadratic term. Table 1. Comparison of negative binomial models for total lice counts on salmon with various forms and combinations of predictors included in the model. Model . Outcome . Predictor(s) . Quadratic . Linear . AIC . Corr.a . AIC . Corr.a . M1 total lice total lice per fish (tlpf)b 10466 0,49 10481 0,51 M2 " log(total lice per fish (tlpf))c 10438 0,55 10442 0,57 M3* " log(tlpf) adjusted for weight 10200 0,71 10205 0,67 M4 " M3 + salmon weight 10088 0,74 10092 0,72 M5 " log(trout attached lice per fish (tapf))d 10385 0,59 10389 0,60 M6 " log(tapf) adjusted for weight 10269 0,62 10275 0,68 M7 " M6 + salmon weight 10145 0,68 10152 0,71 Model . Outcome . Predictor(s) . Quadratic . Linear . AIC . Corr.a . AIC . Corr.a . M1 total lice total lice per fish (tlpf)b 10466 0,49 10481 0,51 M2 " log(total lice per fish (tlpf))c 10438 0,55 10442 0,57 M3* " log(tlpf) adjusted for weight 10200 0,71 10205 0,67 M4 " M3 + salmon weight 10088 0,74 10092 0,72 M5 " log(trout attached lice per fish (tapf))d 10385 0,59 10389 0,60 M6 " log(tapf) adjusted for weight 10269 0,62 10275 0,68 M7 " M6 + salmon weight 10145 0,68 10152 0,71 a correlation between observed and predicted log(counts). b correlations based on average predicted value for the group. c mean total lice count on trout in the group. d mean attached lice count on trout in the group. Table 1. Comparison of negative binomial models for total lice counts on salmon with various forms and combinations of predictors included in the model. Model . Outcome . Predictor(s) . Quadratic . Linear . AIC . Corr.a . AIC . Corr.a . M1 total lice total lice per fish (tlpf)b 10466 0,49 10481 0,51 M2 " log(total lice per fish (tlpf))c 10438 0,55 10442 0,57 M3* " log(tlpf) adjusted for weight 10200 0,71 10205 0,67 M4 " M3 + salmon weight 10088 0,74 10092 0,72 M5 " log(trout attached lice per fish (tapf))d 10385 0,59 10389 0,60 M6 " log(tapf) adjusted for weight 10269 0,62 10275 0,68 M7 " M6 + salmon weight 10145 0,68 10152 0,71 Model . Outcome . Predictor(s) . Quadratic . Linear . AIC . Corr.a . AIC . Corr.a . M1 total lice total lice per fish (tlpf)b 10466 0,49 10481 0,51 M2 " log(total lice per fish (tlpf))c 10438 0,55 10442 0,57 M3* " log(tlpf) adjusted for weight 10200 0,71 10205 0,67 M4 " M3 + salmon weight 10088 0,74 10092 0,72 M5 " log(trout attached lice per fish (tapf))d 10385 0,59 10389 0,60 M6 " log(tapf) adjusted for weight 10269 0,62 10275 0,68 M7 " M6 + salmon weight 10145 0,68 10152 0,71 a correlation between observed and predicted log(counts). b correlations based on average predicted value for the group. c mean total lice count on trout in the group. d mean attached lice count on trout in the group. In the preceding models, fjord was included as a random effect. To explore the role of fjord on the predictive outcome, we added fjord as fixed effects. This had little impact on the parameters. Likewise, we explored the effect of between-year variation by adding year as a fixed effect. This strongly reduced the between group variance as there were 10 years and only 21 groups (fjord/week/year combinations). Fortunately, it had very little effect on the parameter estimates, and we are therefore relatively confident that our estimates are robust. Including salmon weight (M4) improved the model fit; it resulted in a slight decrease in the coefficient for the trout lice count variable from 1.59 to 1.46, and the linear and quadratic coefficients for salmon weight were +0.096 and −0.0007, respectively. However, in the following text we have chosen to focus on the model not including salmon weight. The rationale behind this is that our aim is to use trout in a sampling area where we lack samples of salmon to predict number of lice on salmon. Consequently, in these areas we will not have information on the size of the salmon, thus it will not be possible to base predictions on a model, which requires knowledge of salmon weights. The correlation coefficients of the observed and fitted values for the model, not including size of salmon (M3), was 0.67 for the linear model and 0.71 for the quadratic model. As seen in Figure 4 adding the quadratic term had very little impact on the actual predictions the model made with the exception that the quadratic model produced some much smaller predicted values for three groups. For simplicity we have, in the following text, focused on the linear model (M3). The parameter estimates for model M3 are shown in Table 2 The cross validation exercise (model M3—linear) showed relatively little reduction in the correlation between predicted and observed salmon lice counts (r dropped from 0.67 to 0.59), suggesting that the estimate of the predictive ability was only biased in a positive manner to limited extent. This positive bias would have been present for all models so would not have affected the ranking of models in terms of predictive ability. Table 2. Model parameters from logistic model M3 from Table 1. Fixed effect . . . . . . Coef. Std. Err. z P > |z| [95% CI] Constant 1.59 0.08 20.44 <0.01 [1.43, 1.74] LN (TLPGM) 3.69 0.16 23.16 <0.01 Over-dispersion parameter 1.20 0.04 27.41 <0.01 [1.18, 1.30] Random effect Coef. Std. Err. [95% CI] Group 0.54 0.09 [0.39, 0.75] Fixed effect . . . . . . Coef. Std. Err. z P > |z| [95% CI] Constant 1.59 0.08 20.44 <0.01 [1.43, 1.74] LN (TLPGM) 3.69 0.16 23.16 <0.01 Over-dispersion parameter 1.20 0.04 27.41 <0.01 [1.18, 1.30] Random effect Coef. Std. Err. [95% CI] Group 0.54 0.09 [0.39, 0.75] The model was based on 316 individuals in 21 groups. LN (TLPGM) is the log normal lice counts on trout divided by the weight of the trout. Table 2. Model parameters from logistic model M3 from Table 1. Fixed effect . . . . . . Coef. Std. Err. z P > |z| [95% CI] Constant 1.59 0.08 20.44 <0.01 [1.43, 1.74] LN (TLPGM) 3.69 0.16 23.16 <0.01 Over-dispersion parameter 1.20 0.04 27.41 <0.01 [1.18, 1.30] Random effect Coef. Std. Err. [95% CI] Group 0.54 0.09 [0.39, 0.75] Fixed effect . . . . . . Coef. Std. Err. z P > |z| [95% CI] Constant 1.59 0.08 20.44 <0.01 [1.43, 1.74] LN (TLPGM) 3.69 0.16 23.16 <0.01 Over-dispersion parameter 1.20 0.04 27.41 <0.01 [1.18, 1.30] Random effect Coef. Std. Err. [95% CI] Group 0.54 0.09 [0.39, 0.75] The model was based on 316 individuals in 21 groups. LN (TLPGM) is the log normal lice counts on trout divided by the weight of the trout. Figure 4. Open in new tabDownload slide Plots of observed mean lice counts on salmon vs. predicted values. (Based on M3 linear and M3 quadratic models). Lines of linear fit and observation identifiers have been included. Figure 4. Open in new tabDownload slide Plots of observed mean lice counts on salmon vs. predicted values. (Based on M3 linear and M3 quadratic models). Lines of linear fit and observation identifiers have been included. Logistic models The AIC and Se and Sp for the logistic models are shown in Table 3. The results mirror those of the negative-binomial models; (i) log transforming the number of lice on trout increased predictive ability (M11 vs. M12), (ii) adjusting for weight of the trout increased the predictive ability (M12 vs. M13), (iii) adding weight of the salmon increased the predictive ability, and (iv) using attached lice as a predictor rather than total lice did not give a better prediction after correcting the size of the trout. Table 3. Comparison of logistic models for total lice counts on salmon with various forms and combinations of predictors included in the model. Model . Outcome . Predictor(s) . AIC . Sea . Spb . M11 high licec total lice per fish (tlpfd) 2832 57 52 M12 " log(total lice per fish (tlpf)) 2831 66 47 M13* " log(tlpf) adjusted for weight 2673 99 55 M14 " M13 + salmon weight 2485 75 63 M15 " log(trout attached lice per fish (tapfe)) 2795 66 61 M16 " log(tapf) adjusted for weight 2693 99 55 M17 " M16 + salmon weight 2493 82 64 Model . Outcome . Predictor(s) . AIC . Sea . Spb . M11 high licec total lice per fish (tlpfd) 2832 57 52 M12 " log(total lice per fish (tlpf)) 2831 66 47 M13* " log(tlpf) adjusted for weight 2673 99 55 M14 " M13 + salmon weight 2485 75 63 M15 " log(trout attached lice per fish (tapfe)) 2795 66 61 M16 " log(tapf) adjusted for weight 2693 99 55 M17 " M16 + salmon weight 2493 82 64 a Probability of correctly classifying a positive salmon. b Probability of correctly classifying a negative salmon. c High lice burden was >0.1 lice/g of salmon. d Mean total lice count on trout in the group. e Mean attached lice count on trout in the group. Table 3. Comparison of logistic models for total lice counts on salmon with various forms and combinations of predictors included in the model. Model . Outcome . Predictor(s) . AIC . Sea . Spb . M11 high licec total lice per fish (tlpfd) 2832 57 52 M12 " log(total lice per fish (tlpf)) 2831 66 47 M13* " log(tlpf) adjusted for weight 2673 99 55 M14 " M13 + salmon weight 2485 75 63 M15 " log(trout attached lice per fish (tapfe)) 2795 66 61 M16 " log(tapf) adjusted for weight 2693 99 55 M17 " M16 + salmon weight 2493 82 64 Model . Outcome . Predictor(s) . AIC . Sea . Spb . M11 high licec total lice per fish (tlpfd) 2832 57 52 M12 " log(total lice per fish (tlpf)) 2831 66 47 M13* " log(tlpf) adjusted for weight 2673 99 55 M14 " M13 + salmon weight 2485 75 63 M15 " log(trout attached lice per fish (tapfe)) 2795 66 61 M16 " log(tapf) adjusted for weight 2693 99 55 M17 " M16 + salmon weight 2493 82 64 a Probability of correctly classifying a positive salmon. b Probability of correctly classifying a negative salmon. c High lice burden was >0.1 lice/g of salmon. d Mean total lice count on trout in the group. e Mean attached lice count on trout in the group. Similar to the negative binomial model, we focus on the model that does not correct for the weight of the salmon. The model parameters for model M13 are shown in Table 4. The predicted probability of a salmon having high total lice count (>0.1 lice/g) across range of mean trout lice counts shown in Figure 5. At ∼0.1 lice/g trout, the expected prevalence of high lice counts in salmon was about 14%, while at ∼0.3 lice/g trout the expected prevalence of high lice counts in salmon was 56%. Table 4. Model parameters from logistic model M13 from Table 3. Fixed effect . . . . . . . Coef. Std. Err. z P>|z| [95% CI] Constant 2.44 0.25 9.59 <0.01 [1.94, 2.94] LN (TLPGM) 1.81 0.14 13.23 <0.01 [1.55, 2.08] Random effect Coef. Std. Err. [95% CI] Group 1.21885 0.212472 [0.86, 1.72] Fixed effect . . . . . . . Coef. Std. Err. z P>|z| [95% CI] Constant 2.44 0.25 9.59 <0.01 [1.94, 2.94] LN (TLPGM) 1.81 0.14 13.23 <0.01 [1.55, 2.08] Random effect Coef. Std. Err. [95% CI] Group 1.21885 0.212472 [0.86, 1.72] The model was based on 316 individuals in 21 groups. LN (TLPGM) is the log normal lice counts on trout divided by the weight of the trout. Table 4. Model parameters from logistic model M13 from Table 3. Fixed effect . . . . . . . Coef. Std. Err. z P>|z| [95% CI] Constant 2.44 0.25 9.59 <0.01 [1.94, 2.94] LN (TLPGM) 1.81 0.14 13.23 <0.01 [1.55, 2.08] Random effect Coef. Std. Err. [95% CI] Group 1.21885 0.212472 [0.86, 1.72] Fixed effect . . . . . . . Coef. Std. Err. z P>|z| [95% CI] Constant 2.44 0.25 9.59 <0.01 [1.94, 2.94] LN (TLPGM) 1.81 0.14 13.23 <0.01 [1.55, 2.08] Random effect Coef. Std. Err. [95% CI] Group 1.21885 0.212472 [0.86, 1.72] The model was based on 316 individuals in 21 groups. LN (TLPGM) is the log normal lice counts on trout divided by the weight of the trout. Figure 5. Open in new tabDownload slide Predicted probability of a salmon having a high lice count (>0.1 lice/g) across the range of mean trout lice values (based on model 13 with log transformed, weight adjusted lice counts as sole predictor). Figure 5. Open in new tabDownload slide Predicted probability of a salmon having a high lice count (>0.1 lice/g) across the range of mean trout lice values (based on model 13 with log transformed, weight adjusted lice counts as sole predictor). Figure 6. Open in new tabDownload slide Plot of Se and Sp estimates derived at cutpoint probabilities (%) ranging from 0 to 100 (lower x-axis)—based on model M13 with log transformed, weight adjusted lice counts as the sole predictor. Upper axis shows equivalent total lice counts (per g) at cutpoints (1, 20, 40, 60, and 80%). Figure 6. Open in new tabDownload slide Plot of Se and Sp estimates derived at cutpoint probabilities (%) ranging from 0 to 100 (lower x-axis)—based on model M13 with log transformed, weight adjusted lice counts as the sole predictor. Upper axis shows equivalent total lice counts (per g) at cutpoints (1, 20, 40, 60, and 80%). The Se and Sp of the best fitting model (M14) were 75 and 63, respectively, while the Se and Sp of model M13 were 99 and 55, respectively (both based on predicted probability cutpoints of 0.01). In comparison, when trout counts in each group were used directly by choosing cut offs of 0.1 or 0.2 lice/g trout, the Sp and Se were 20.9/97.8 and 49.4/59.3, respectively. To explore how the Se and Sp (i.e. using the model as a diagnostic test), varied with the predicted probability cutpoint chosen, we plotted how Se and Sp changed with different cutoff points (and their corresponding trout lice count cut offs). Essentially, this asks the following question: how sensitive and specific would a “diagnostic” test be at predicting salmon with high or low lice counts, given a predictive cutoff point of lice per gram trout? In Figure 6, it is easy to see that a balance between Se and Sp is achieved at a cutpoint of approximately 0.2 (equivalent to a trout lice count of 12.1). However, at this cutpoint, both the Se and Sp are quite low (∼65%) meaning that salmon with high and low counts are both misclassified ∼35% of the time. Figure 7. Open in new tabDownload slide Scatter plot showing index values based on observed counts in trout under 150 g and predicted counts in salmon. Dashed lines demarcate edge of “green zone” (10% mortality), solid lines demarcate edge of “red zone” (30% mortality) and dash-dot line shows linear fit of the points. Figure 7. Open in new tabDownload slide Scatter plot showing index values based on observed counts in trout under 150 g and predicted counts in salmon. Dashed lines demarcate edge of “green zone” (10% mortality), solid lines demarcate edge of “red zone” (30% mortality) and dash-dot line shows linear fit of the points. Implications for management advice In the second dataset (from Herdlefjorden) the estimated likelihood of mortality of migrating salmon post-smolt (in %) based on the weighted average mortality calculated directly from lice counts on sampled trout below 150 grams (method by Taranger et al. 2015 described earlier), for 1 May–31 July in years 2009–2015 divided into five 2-week periods, is presented in Table 5a. In Table 5b, we present the same estimated likelihood of mortality based on predicted lice counts [predictions based on model 3 using all sampled trout data, adjusted for the average weight of the salmon post-smolts in the trawl data (23 g)]. The pairwise comparison of the two sets of mortality estimates are also plotted in Figure 7. These demonstrate how the mortality estimates based on the model predictions using all trout are clearly lower than the weighted average mortality using direct lice counts on trout under 150 g. Out of the 35 samples, 4 groups went from category “yellow” to category “green”, while 6 went from category “red” to category “yellow”. In addition, one sample went from category “green” to category “yellow”. Table 5. Estimated “population level effect” of salmon lice based on the method described in Taranger et al. (2015). (a) . . . . . . Year . 1–15 May . 16–31 May . 1–15 June . 16–30 June . 1–31 July . 2009 0.0 (3) 2.4 (17) 18.8 (17) 13.3 (9) 50.0 (2) 2010 0.0 (4) 26.0 (55) 55.7 (23) 83.0 (30) (0) 2011 6.7 (3) (0) 10.0 (12) 0.0 (2) (0) 2012 14.2 (12) 31.8 (17) 58.5 (20) 77.8 (9) 5.0 (2) 2013 0.0 (1) 0.0 (5) 0.0 (8) 36.9 (16) (0) 2014 0.0 (6) 13.3 (9) 73.3 (3) 97.4 (57) (0) 2015 0.0 (4) 100.0 (3) 48.0 (5) 0.0 (1) 30.0 (4) (b) Year 1–15 May 16–31 May 1–15 June 16–30 June 1– 31 July 2009 0.0 (7) 0.9 (23) 7.0 (23) 3.3 (12) 25.0 (4) 2010 0.0 (7) 15.2 (69) 38.8 (32) 65.2 (33) (0) 2011 0.0 (17) 6.3 (8) 5.8 (24) 0.0 (4) (0) 2012 4.7 (34) 24.8 (31) 40.0 (27) 58.8 (16) 17.8 (9) 2013 0.0 (20) 0.0 (23) 0.0 (21) 18.0 (25) (0) 2014 1.3 (15) 6.1 (23) 50.0 (8) 90.3 (63) (0) 2015 18.5 (20) 21.5 (20) 12.1 (19) 0.0 (8) 11.7 (12) (a) . . . . . . Year . 1–15 May . 16–31 May . 1–15 June . 16–30 June . 1–31 July . 2009 0.0 (3) 2.4 (17) 18.8 (17) 13.3 (9) 50.0 (2) 2010 0.0 (4) 26.0 (55) 55.7 (23) 83.0 (30) (0) 2011 6.7 (3) (0) 10.0 (12) 0.0 (2) (0) 2012 14.2 (12) 31.8 (17) 58.5 (20) 77.8 (9) 5.0 (2) 2013 0.0 (1) 0.0 (5) 0.0 (8) 36.9 (16) (0) 2014 0.0 (6) 13.3 (9) 73.3 (3) 97.4 (57) (0) 2015 0.0 (4) 100.0 (3) 48.0 (5) 0.0 (1) 30.0 (4) (b) Year 1–15 May 16–31 May 1–15 June 16–30 June 1– 31 July 2009 0.0 (7) 0.9 (23) 7.0 (23) 3.3 (12) 25.0 (4) 2010 0.0 (7) 15.2 (69) 38.8 (32) 65.2 (33) (0) 2011 0.0 (17) 6.3 (8) 5.8 (24) 0.0 (4) (0) 2012 4.7 (34) 24.8 (31) 40.0 (27) 58.8 (16) 17.8 (9) 2013 0.0 (20) 0.0 (23) 0.0 (21) 18.0 (25) (0) 2014 1.3 (15) 6.1 (23) 50.0 (8) 90.3 (63) (0) 2015 18.5 (20) 21.5 (20) 12.1 (19) 0.0 (8) 11.7 (12) The upper table (a) is based on estimates using trout <150 g, while the lower (b) is based on trout predictions from model 3. Number inside brackets indicates number of fish used to calculate the “population level effect”. Note that N is lower for (a) compared with (b) because they only utilize trout smaller than 150 g. Table 5. Estimated “population level effect” of salmon lice based on the method described in Taranger et al. (2015). (a) . . . . . . Year . 1–15 May . 16–31 May . 1–15 June . 16–30 June . 1–31 July . 2009 0.0 (3) 2.4 (17) 18.8 (17) 13.3 (9) 50.0 (2) 2010 0.0 (4) 26.0 (55) 55.7 (23) 83.0 (30) (0) 2011 6.7 (3) (0) 10.0 (12) 0.0 (2) (0) 2012 14.2 (12) 31.8 (17) 58.5 (20) 77.8 (9) 5.0 (2) 2013 0.0 (1) 0.0 (5) 0.0 (8) 36.9 (16) (0) 2014 0.0 (6) 13.3 (9) 73.3 (3) 97.4 (57) (0) 2015 0.0 (4) 100.0 (3) 48.0 (5) 0.0 (1) 30.0 (4) (b) Year 1–15 May 16–31 May 1–15 June 16–30 June 1– 31 July 2009 0.0 (7) 0.9 (23) 7.0 (23) 3.3 (12) 25.0 (4) 2010 0.0 (7) 15.2 (69) 38.8 (32) 65.2 (33) (0) 2011 0.0 (17) 6.3 (8) 5.8 (24) 0.0 (4) (0) 2012 4.7 (34) 24.8 (31) 40.0 (27) 58.8 (16) 17.8 (9) 2013 0.0 (20) 0.0 (23) 0.0 (21) 18.0 (25) (0) 2014 1.3 (15) 6.1 (23) 50.0 (8) 90.3 (63) (0) 2015 18.5 (20) 21.5 (20) 12.1 (19) 0.0 (8) 11.7 (12) (a) . . . . . . Year . 1–15 May . 16–31 May . 1–15 June . 16–30 June . 1–31 July . 2009 0.0 (3) 2.4 (17) 18.8 (17) 13.3 (9) 50.0 (2) 2010 0.0 (4) 26.0 (55) 55.7 (23) 83.0 (30) (0) 2011 6.7 (3) (0) 10.0 (12) 0.0 (2) (0) 2012 14.2 (12) 31.8 (17) 58.5 (20) 77.8 (9) 5.0 (2) 2013 0.0 (1) 0.0 (5) 0.0 (8) 36.9 (16) (0) 2014 0.0 (6) 13.3 (9) 73.3 (3) 97.4 (57) (0) 2015 0.0 (4) 100.0 (3) 48.0 (5) 0.0 (1) 30.0 (4) (b) Year 1–15 May 16–31 May 1–15 June 16–30 June 1– 31 July 2009 0.0 (7) 0.9 (23) 7.0 (23) 3.3 (12) 25.0 (4) 2010 0.0 (7) 15.2 (69) 38.8 (32) 65.2 (33) (0) 2011 0.0 (17) 6.3 (8) 5.8 (24) 0.0 (4) (0) 2012 4.7 (34) 24.8 (31) 40.0 (27) 58.8 (16) 17.8 (9) 2013 0.0 (20) 0.0 (23) 0.0 (21) 18.0 (25) (0) 2014 1.3 (15) 6.1 (23) 50.0 (8) 90.3 (63) (0) 2015 18.5 (20) 21.5 (20) 12.1 (19) 0.0 (8) 11.7 (12) The upper table (a) is based on estimates using trout <150 g, while the lower (b) is based on trout predictions from model 3. Number inside brackets indicates number of fish used to calculate the “population level effect”. Note that N is lower for (a) compared with (b) because they only utilize trout smaller than 150 g. Discussion This study is the first study that demonstrates that there is a correlation between lice levels on sea trout and migrating salmon post-smolts. This result should be highly relevant for management, which for a number of years has estimated the impact of salmon lice on Atlantic salmon using lice counts from sea trout (Svåsand et al., 2015; Taranger et al., 2015). However, it must be stressed that there is large variance around the relationship between the lice levels of the two species, and that the predictive ability of using lice levels on trout to predict lice levels on salmon post-smolts is relatively low so predictions will be rather imprecise. One of our main findings was that the sea trout generally had higher levels of lice than salmon. A plausible explanation for this can be that the two species have very different marine life-history strategies and near-shore habitat use. While Most trout (and especially smaller trout) spend all of their time near shore (Thorstad et al., 2016), salmon post-smolt migrate relatively fast from their river of origin and outwards towards the open sea. The average progression rate of salmon during their early marine migration is somewhere around 0.4–3.0 body lengths per second (Thorstad et al., 2012; Vollset et al., 2016). This means that salmon post-smolt can spend everything from a few days up to several weeks (sometimes over a month) migrating in near shore environment (depending on topography) where they are most likely to encounter salmon lice. Thus, most salmon post-smolt caught by a trawl will have had lice that have not yet developed to mobile stages (Finstad et al. 2000). Trout on the other hand, may have been residing in the marine near shore environment for several months depending on the life-history strategy of the individual, and therefore amassed a wider range of lice life-stages. However, we suspect that even though we do catch a lot of trout with high levels of lice in the trawls, we might not catch the most affected trout due to mortality and premature return migration to the fresh water (Birkeland, 1996; Birkeland and Jakobsen, 1997). Therefore, even the highest levels of lice on trout might be underestimations. Different species and populations of salmonids are known to have different susceptibility and immune response to lice. Dawson et al. (1997) found that significantly more lice settled on hatchery-reared sea trout compared with hatchery-reared salmon (∼400–500 g). Although, salmon lice numbers declined more rapidly on sea trout than on salmon, the number at the end of the experiment was higher on trout than on salmon. If salmon lice prefer trout as a host compared with salmon, this may explain the lower abundance of lice on salmon compared with trout in our study. From a life-history perspective trout may be a high risk—high gain host from the perspective of the parasite, as they remain closer to the coast where the likelihood of encountering another host is high, but the likelihood of survival in areas affected by freshwater runoff is lower (Thorstad et al., 2015). In contrast Glover et al. (2003) found higher lice abundance on a farmed salmon group compared with a different population of trout, contrasting the finding by Dawson et al. (1997). However, farmed fish may differ from wild salmon, and in a follow up study Glover et al. (2004) found that wild fish from the river Dale had lower susceptibility than farmed salmon and wild salmon from the nearby river Vosso. Another mechanism that may amplify the lice burden on individual fish is that fish infested with salmon lice are also more susceptible to new infestations (Ugelvik et al., 2016). Interestingly, lice levels on salmon post-smolt were strongly affected by the size of the fish. Increased size may be linked to swimming speed and consequently the encounter rate with lice. For example, Samsing et al. (2015) demonstrated that lice encounter is dome-shaped with the highest encounter rate at intermediate swimming speeds. A larger individual will also create a larger pulse of water around the head, which may trigger the salmon lice copepodite to swim towards the fish. Heuch et al. (2007) demonstrated that salmon lice reacts to the pulse of water from a model salmon head pushed forward into a tank, jumping towards the head. Higher lice numbers on larger fish have also been observed in experimental infestation studies (Glover et al., 2003,, 2004,, 2005). Another plausible explanation is that size reflects growth during the marine phase. Consequently, size will be correlated with time spent in marine waters and therefore larger post-smolt will have a longer exposure history than smaller fish. If this is the case, the trawl samples a distribution of fish with various exposure histories. Thus, the abundance on lice on the sampled fish must be viewed as an underestimation for populations with long exposure histories and an overestimation for populations with a low exposure history. Ideally, trawling should be conducted after the post-smolt have left the exposure area for lice from fish farms. However, the few attempts to capture fish in open waters after they have left fjords have generally been unsuccessful with very small catches (Bengt Finstad, pers. comm.). Even though our binary approach that modelled the likelihood that a salmon had either high (>0.1 lice/g fish weight) or low (<0.1 lice/g fish weight) did support the hypothesis that lice on trout was a significant predictor of high or low lice counts on salmon, the results from this method were not promising. The advantage of using such a binary model is that it can be used as diagnostic tool to define a fish in a category of either “sick” or “not-sick”, and with this one can calculate the Sp and Se of the model. This provides information on how good the diagnostic tests are. However, clearly the optimal balance between Se and Sp indicated that the model was a poor diagnostic tool. For example, increasing the Sp to more than 60% would strongly compromise the Se and vice versa. Futhermore, a binary system does not allow for a calculation of estimated mortality on salmon post-smolt as suggested in Taranger et al. (2015). Therefore, our recommendation is to utilize the numerical values of counts (from the negative binomial model described earlier) to predict lice numbers on salmon instead of predicting it categorically. Currently a new management system is being implemented in Norway where the allowable production of salmon in different regions in Norway will be based on (among other things) advice on the impact of salmon lice on wild fish (Karlsen et al., 2016). More specifically, the current threshold values that are to be implemented state that the production biomass is not allowed to increase if the estimated mortality of local salmonid populations based on lice counts from wild caught fish exceeds 10%, and the allowable biomass will be reduced if the estimated mortality exceeds 30%. Using an independent dataset on trap-net caught trout, we demonstrated how our model would lead to a clear reduction in the estimated mortality for salmon post-smolts compared with the currently used method, and that in many cases the advice given based on the above mentioned threshold would change. In this study, we have used the relationship between salmon lice on sea trout and Atlantic salmon as a global relationship and applied it to another system and sampling method when calculating the infestation pressure on salmon. We are aware that this has limitations. The relationship between salmon lice on sea trout and Atlantic salmon are most likely strongly dependent on temporal and spatial scales. Fjord came out as an important random effect in our analysis indicating that lice counts varied substantially among fjords and it is quite plausible that the trout lice count—salmon lice count relationship is different at different overall lice levels. In theory, the relationship between any specific trawling station and trap-net location may be inherently different because of a range of reasons, e.g. the geography/bathymetry and position of the sample locations, the rivers in the vicinity, and salmonid population attributes in these rivers. However, to date there are not enough available data to do an analysis that can unravel this complexity. Although there seems to be a consensus that in fish farm intensive areas there are higher infestation levels of lice on wild fish (Bjorn et al., 2011; Serra-Llinares et al., 2014, 2016), documenting negative population level effects on salmonids in general, and on Atlantic salmon particularly, remains controversial. There are several reasons for this controversy, but the difficulty of getting accurate estimates of sea lice loads on migrating wild post-smolt salmon is perhaps the largest, as this leads to uncertainties in estimating the lice induced mortality of salmon. Our advice is to increase the trawling effort on salmon post-smolts in the Norwegian fjords. More data can thereafter be used to validate to what degree the relationship found in this study is globally valid and to validate potential spatially explicit salmon lice models such as hydrodynamic particle tracking models. Also, based on this analysis, we suggest that using sea trout lice counts directly to predict expected mortality on salmon is not appropriate. However, trout lice counts may be used to predict salmon lice counts, with these estimates then being combined with information from other sources for decision making in the management system. Acknowledgements Karin Kroon Boxaspen, Rune Nilsen, Nils Arne Hvidsten, Julius Dahle, Arild Refsnes, and Arve Kristiansen are acknowledged for providing the trawling data used in this study. Arve Kristiansen is also acknowledged for providing information on trawling practices. The trawl data was obtained by the National Salmon Lice Monitoring Program on wild Salmonids in Norway funded by the Institute of Marine Research, Norway, the Norwegian Food Safety Authority, the Norwegian Environment Agency and the Norwegian Institute for Nature Research. Funding Knut Wiik Vollset and Ian Dohoo were supported by the Norwegian Research Council (Project. Nr. 243912/E50) during the writing and analysis of this study. All authors from IMR and NINA funded by internal grants from the Institute of Marine Research and the Norwegian Institute for Nature Research, respectively. References Barlaup B. T. , Gabrielsen S. E., Loyland J., Schlappy M. L., Wiers T., Vollset K. W., Pulg U. 2013 . Trap design for catching fish unharmed and the implications for estimates of sea lice (Lepeophtheirus salmonis) on anadromous brown trout (Salmo trutta) . Fisheries Research , 139 : 43 – 46 . Google Scholar Crossref Search ADS WorldCat Birkeland K. 1996 . Consequences of premature return by sea trout (Salmo trutta) infested with the salmon louse (Lepeophtheirus salmonis Kroyer): Migration, growth, and mortality . Canadian Journal of Fisheries and Aquatic Sciences/Journal Canadien des Sciences Halieutiques et Aquatiques. Ottawa 53 : 2808 – 2813 . Google Scholar Crossref Search ADS WorldCat Birkeland K. , Jakobsen P. J. 1997 . Salmon lice, Lepeophtheirus salmonis, infestation as a causal agent of premature return to rivers and estuaries by sea trout, Salmo trutta, juveniles . Environmental Biology of Fishes , 49 : 129 – 137 . Google Scholar Crossref Search ADS WorldCat Bjoern P. A. , Finstad B. 1997 . The physiological effects of salmon lice infection on sea trout post smolts . Nordic Journal of Freshwater Research , 73 : 60 – 72 . OpenURL Placeholder Text WorldCat Bjoern P. A. , Finstad B. 1998 . The development of salmon lice (Lepeophtheirus salmonis) on artificially infected post smolts of sea trout (Salmo trutta) . Canadian Journal of Zoology/Revue Canadien de Zoologie , 76 : 970 – 977 . Google Scholar Crossref Search ADS WorldCat Bjorn P. A. , Sivertsgard R., Finstad B., Nilsen R., Serra-Llinares R. M., Kristoffersen R. 2011 . Area protection may reduce salmon louse infection risk to wild salmonids . Aquaculture Environment Interactions , 1 : 233 – 244 . Google Scholar Crossref Search ADS WorldCat Dawson L. H. J. , Pike A. W., Houlihan D. F., McVicar A. H. 1997 . Comparison of the susceptibility of sea trout (Salmo trutta L.) and Atlantic salmon (Salmo salar L.) to sea lice (Lepeophtheirus salmonis (Krøyer, 1837)) infections . Ices Journal of Marine Science 54 : 1129 – 1139 . OpenURL Placeholder Text WorldCat Finstad B. , Bjoern P., Grimnes A., Hvidsten N. 2000 . Laboratory and field investigations of salmon lice [Lepeophtheirus salmonis (Kroyer)] infestation on Atlantic salmon (Salmo salar L.) post-smolts . Aquaculture Research , 31 : 795 – 803 . Google Scholar Crossref Search ADS WorldCat Finstad B. , Grimnes A. 1997 . Registreringer av lakselus på laks, sjøørret og sjøøye i 1996 (Registrations of salmon lice on Atlantic salmon, sea trout and Arctic charr in 1996) . NINA Oppdragsmelding , 485 : 1 – 27 . Glover K. A. , Aasmundstad T., Nilsen F., Storset A., Skaala O. 2005 . Variation of Atlantic salmon families (Salmo salar L.) in susceptibility to the sea lice Lepeophtheirus salmonis and Caligus elongatus . Aquaculture , 245 : 19 – 30 . Google Scholar Crossref Search ADS WorldCat Glover K. A. , Hamre L. A., Skaala Ø., Nilsen F. 2004 . A comparison of sea louse (Lepeophtheirus salmonis) infection levels in farmed and wild Atlantic salmon (Salmo salar L.) stocks . Aquaculture , 232 : 41 – 52 . Google Scholar Crossref Search ADS WorldCat Glover K. A. , Skaala Ø., Nilsen F., Olsen R., Teale A. J., Taggart J. B. 2003 . Differing susceptibility of anadromous brown trout (Salmo trutta L.) populations to salmon louse (Lepeophtheirus salmonis (Krøyer, 1837)) infection . Ices Journal of Marine Science , 60 : 1139 – 1148 . Google Scholar Crossref Search ADS WorldCat Godwin S. C. , Dill L. M., Reynolds J. D., Krkosek M. 2015 . Sea lice, sockeye salmon, and foraging competition: lousy fish are lousy competitors . Canadian Journal of Fisheries and Aquatic Sciences , 72 : 8. Google Scholar Crossref Search ADS WorldCat Grimnes A. , Finstad B., Bjorn P. A. 1996 . Økologiske og fysiologiske konsekvenser av lus på laksefisk i fjordsystem (Ecological and physiological consequences of lice on salmonids in the fjord ecosystems) . NINA Oppdragsmelding , 381 : 1 – 37 . Grimnes A. , Jakobsen P. J. 1996 . The physiological effects of salmon lice infection on post-smolt of Atlantic salmon . Journal of Fish Biology , 48 : 1179 – 1194 . Google Scholar Crossref Search ADS WorldCat Heuch P. A. , Doall M. H., Yen J. 2007 . Water flow around a fish mimic attracts a parasitic and deters a planktonic copepod . Journal of Plankton Research , 29 : I3 – I16 . Google Scholar Crossref Search ADS WorldCat Holst J. C. , McDonald A. 2000 . FISH-LIFT: a device for sampling live fish with trawls . Fisheries Research , 48 : 87 – 91 . Google Scholar Crossref Search ADS WorldCat Jackson D. , Cotter D., Newell J., ODonohoe P., Kane F., McDermott T., Kelly S. et al. 2014 . Response to M. Krkošek, C. W. Revie, B. Finstad, and C. D. Todd’s comment on Jackson et al. Impact of Lepeophtheirus salmonis infestations on migrating Atlantic salmon, Salmo salar L., smolts at eight locations in Ireland with an analysis of lice-induced marine mortality . Journal of Fish Diseases , 37 : 419 – 421 . Google Scholar Crossref Search ADS PubMed WorldCat Karlsen Ø. , Finstad B., Ugedal O., Svåsand T. 2016 . Kunnskapsstatus som grunnlag for kapasitetsjustering innen produksjons-områder basert på lakselus som indikator . Rapport fra Havforskningen nr. 14-2016 . pp. 137 . Krkosek M. , Lewis M. A., Morton A., Frazer L., Volpe J. P. 2006 . Epizootics of wild fish induced by farm fish . Proceedings of the National Academy of Sciences of the United States of America , 103 : 15506 – 15510 . Google Scholar Crossref Search ADS PubMed WorldCat Krkošek M. , Revie C. W., Finstad B., Todd C. D., Comment on Jackson . et al. 2014 . Impact of Lepeophtheirus salmonis infestations on migrating Atlantic salmon, Salmo salar L., smolts at eight locations in Ireland with an analysis of lice-induced marine mortality’ . Journal of Fish Diseases , 37 : 415 – 417 . Google Scholar Crossref Search ADS PubMed WorldCat Krkosek M. , Revie C. W., Gargan P. G., Skilbrei O. T., Finstad B., Todd C. D. 2013 . Impact of parasites on salmon recruitment in the Northeast Atlantic Ocean . Proceedings of the Royal Society B-Biological Sciences , 280 : 20122359 . Google Scholar Crossref Search ADS WorldCat Nilsen R. , Bjørn P. A., Serra-Llinares R. M., Asplin L., Sandvik A. D., Johnsen I. A., Karlsen Ø. et al. 2016 . Lakselusinfeksjonen på vill laksefisk langs norskekysten i 2015. Enn fullskalatest av modellbasert varsling og tilstandsbekreftelse . Rapport fra Havforskningen nr. 2-2016 . pp. 60 . Peacock S. J. , Krkošek M., Bateman A. W., Lewis M. A. 2015 . Parasitism and food web dynamics of juvenile Pacific salmon . Ecosphere , 6 : 1 – 16 . Google Scholar Crossref Search ADS WorldCat Samsing F. , Solstorm D., Oppedal F., Solstorm F., Dempster T. 2015 . Gone with the flow: current velocities mediate parasitic infestation of an aquatic host . International Journal for Parasitology , 45 : 559 – 565 . Google Scholar Crossref Search ADS PubMed WorldCat Serra-Llinares R. M. , Bjorn P. A., Finstad B., Nilsen R., Harbitz A., Berg M., Asplin L. 2014 . Salmon lice infection on wild salmonids in marine protected areas: an evaluation of the Norwegian ‘National Salmon Fjords’ . Aquaculture Environment Interactions , 5 : 1 – 16 . Google Scholar Crossref Search ADS WorldCat Serra-Llinares R. M. , Bjørn P. A., Finstad B., Nilsen R., Asplin L. 2016 . Nearby farms are a source of lice for wild salmonids: a reply to Jansen et al. (2016) . Aquaculture Environment Interactions , 8 : 351 – 356 . Google Scholar Crossref Search ADS WorldCat Skilbrei O. T. , Finstad B., Urdal K., Bakke G., Kroglund F., Strand R. 2013 . Impact of early salmon louse, Lepeophtheirus salmonis, infestation and differences in survival and marine growth of sea-ranched Atlantic salmon, Salmo salar L., smolts 1997-2009 . Journal of Fish Diseases , 36 : 249 – 260 . Google Scholar Crossref Search ADS PubMed WorldCat Svåsand T. , Boxaspen K. K., Karlsen Ø., Kvamme B. O., Stien L. H., Taranger G. L. 2015 . Risikovurdering norsk fiskeoppdrett-2013. Fisken og havet, særnummer: 2-2015. Taranger G. L. , Karlsen Ø., Bannister R. J., Glover K. A., Husa V., Karlsbakk E., Kvamme B. O. et al. 2015 . Risk assessment of the environmental impact of Norwegian Atlantic salmon farming . ICES Journal of Marine Science: Journal du Conseil , 72 : 997 – 1021 . Google Scholar Crossref Search ADS WorldCat Taranger G. L. , Svåsand T., Bjørn P. A., Jansen P. A., Heuch P. A., Grøntvedt R. N., Asplin L. et al. 2012 . Forslag til førstegenerasjons målemetode for miljøeffekt (effektindikatorer) med hensyn til genetisk påvirkning fra oppdrettslaks til villaks, og påvirkning av lakselus fra oppdrett på viltlevende laksefiskbestander. Thorstad E. B. , Todd C. D., Uglem I., Bjorn P. A., Gargan P. G., Vollset K. W., Halttunen E. et al. 2015 . Effects of salmon lice Lepeophtheirus salmonis on wild sea trout Salmo trutta-a literature review . Aquaculture Environment Interactions , 7 : 91 – 113 . Google Scholar Crossref Search ADS WorldCat Thorstad E. B. , Todd C. D., Uglem I., Bjørn P. A., Gargan P. G., Vollset K. W., Halttunen E. et al. 2016 . Marine life of the sea trout . Marine Biology , 163 : 1 – 19 . Google Scholar Crossref Search ADS WorldCat Thorstad E. B. , Whoriskey F., Uglem I., Moore A., Rikardsen A. H., Finstad B. 2012 . A critical life stage of the Atlantic salmon Salmo salar: behaviour and survival during the smolt and initial post-smolt migration . Journal of Fish Biology , 81 : 500 – 542 . Google Scholar Crossref Search ADS PubMed WorldCat Torrissen O. , Jones S., Asche F., Guttormsen A., Skilbrei O. T., Nilsen F., Horsberg T. E. et al. 2013 . Salmon lice - impact on wild salmonids and salmon aquaculture . Journal of Fish Diseases , 36 : 171 – 194 . Google Scholar Crossref Search ADS PubMed WorldCat Ugelvik M. S. , Mo T., Mennerat A., Skorping A. 2016 . Atlantic salmon infected with salmon lice are more susceptible to new lice infections . Journal of Fish Diseases , 40 : 311 – 317 . Google Scholar Crossref Search ADS PubMed WorldCat Vollset K. W. , Barlaup B. T. 2014 . First report of winter epizootic of salmon lice on sea trout in Norway . Aquaculture Environment Interactions , 5 : 5. Google Scholar Crossref Search ADS WorldCat Vollset K. W. , Barlaup B. T., Mahlum S., Bjørn P. A., Skilbrei O. T. 2016 . Estimating the temporal overlap between post-smolt migration of Atlantic salmon and salmon lice infestation pressure from fish farms . Aquaculture Environment Interactions , 8 : 511 – 525 . Google Scholar Crossref Search ADS WorldCat Vollset K. W. , Barlaup B. T., Skoglund H., Normann E. S., Skilbrei O. T. 2014 . Salmon lice increase the age of returning Atlantic salmon . Biological Letter , 10 : 20130896. Google Scholar Crossref Search ADS WorldCat Vollset K. W. , Krontveit R. I., Jansen P. A., Finstad B., Barlaup B. T., Skilbrei O. T., Krkošek M. et al. 2015 . Impacts of parasites on marine survival of Atlantic salmon: a meta-analysis. Fish and Fisheries , 17 : 714 – 730 . Wagner G. N. , Fast M. D., Johnson S. C. 2008 . Physiology and immunology of Lepeophtheirus salmonis infections of salmonids . Trends in Parasitology , 24 : 176 – 183 . Google Scholar Crossref Search ADS PubMed WorldCat Wells A. , Grierson C. E., Marshall L., MacKenzie M., Russon I. J., Reinardy H., Sivertsgard R. et al. 2007 . Physiological consequences of “premature freshwater return” for wild sea-run brown trout (Salmo trutta) postsmolts infested with sea lice (Lepeophtheirus salmonis) . Canadian Journal of Fisheries and Aquatic Sciences/Journal Canadien des Sciences Halieutiques et Aquatiques , 64 : 1360 – 1369 . Google Scholar Crossref Search ADS WorldCat Author notes " Vollset, K. W., Halttunen, E., Finstad, B., Karlsen, Ø., Bjørn, P. A., and Dohoo, I. 2017. Salmon lice infestations on sea trout predicts infestations on migrating salmon post-smolts. – ICES Journal of Marine Science, 00:000–000. © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Richerson, Kate; Holland, Daniel S

doi: 10.1093/icesjms/fsx093pmid: N/A

Abstract As anthropogenic changes interact with natural climate cycles, the variability of marine ecosystems is likely to increase. Changes in productivity of particular fisheries might be expected to lead not only to direct impacts within a fishery but to economic and ecological effects on other fisheries if there is substantial cross-participation by fishers. We use data from the US West Coast salmon troll fishery before, during, and after a large-scale closure to illustrate how altered resource availability influences the behaviour of fishing vessels in heterogeneous ways. We find that vessels were less likely to participate in fishing of any type during the closure, with >40% of vessels ceasing fishing temporarily and 17% exiting permanently. Vessels that were more dependent on salmon were more likely to cease fishing while more diversified vessels were more likely to continue. In spite of a high level of cross-participation, we find limited evidence that vessels increased their participation in other fisheries to offset lost salmon revenue. Ports that obtained more of their revenue from salmon troll vessels saw larger decreases in their revenue during the closure. Ocean conditions from 2013 to 2015 suggest the possibility of another highly restricted salmon fishing season in 2017. Our models predict that such restrictions would cause another economic disaster and lead to a large fraction of vessels exiting fishing but suggest that effects on fisheries linked by cross-participation are likely to be low. Introduction Fishers are an integral part of nearly all marine ecosystems. Their behaviour both drives and is driven by changes in the ecosystem, resulting in complex interactions between climate, fishing, harvested populations, and other associated components of the system. The interdependence of fishing, ecology, and climate can be particularly important for upwelling-driven systems where climate fluctuations are strongly linked to population-level processes across many marine species (Doney et al., 2012). The high variability in productivity in these fisheries can motivate fishers to maintain complex fishing portfolios that enable them to adjust to ups and downs in individual fisheries (Kasperski and Holland, 2013), and this has the potential to transmit climate related shocks across species that are not linked ecologically. The California Current Large Marine Ecosystem (CCLME), reaching from southern British Colombia to Baja Mexico, is a highly productive, variable upwelling system that supports many interlinked fisheries. Basin-scale atmospheric and oceanic processes, particularly the El Niño Southern Oscillation, Pacific Decadal Oscillation (PDO), North Pacific Gyre Oscillation, and North Pacific High pressure system, influence temperature, wind patterns, circulation, upwelling, and other ocean conditions. These physical conditions in turn influence primary productivity, zooplankton, and higher trophic levels, including fish, birds, and marine mammals through both direct (physiological) and indirect (trophic) effects (Mantua et al., 1997; Di Lorenzo et al., 2008; Schwing et al., 2010; King et al., 2011; Schroeder et al., 2013; Sydeman et al., 2014). The success of many US West Coast fisheries is also tied to biophysical conditions in the CCLME at various spatial and temporal scales. For example, the famous collapse of the lucrative California sardine fishery in the late 1940s has been (at least partially) attributed to the shift from a warm to cool PDO regime across the Pacific (Zwolinski and Demer, 2012). More recently, in 2008, the collapse of the Sacramento River fall-run Chinook (SRFC; Oncorhynchus tshawytscha), in combination with low coho salmon (Oncorhynchus kisutch) returns, lead to unprecedented restrictions on salmon fishing along much of the US West Coast, with a complete closure of the ocean Chinook fishery south of Cape Falcon, Oregon and a limited season elsewhere. These restrictions extended into the 2009 season, and resulted in the declaration of a West Coast-wide federal disaster and the release of US $170 million in aid to salmon fishers and other salmon-dependent businesses. The collapse of the SRFC was attributed to poor ocean conditions in 2005 and 2006, with weak upwelling and warm temperatures that resulted in limited prey availability and low survival for the 2004 and 2005 brood years (Lindley et al., 2009). Though the extent of the closure was unprecedented, the collapse was not the first coast-wide salmon disaster tied to poor ocean conditions in recent history, with a federally declared disaster in 1994–1995 attributed to El Niño, drought, flooding, reduced upwelling, and poor ocean conditions. In addition, in 2006 a disaster was declared for the Klamath River basin salmon fishery, where low returns were blamed on poor stream and ocean conditions. Changes in ocean conditions can have a large effect on salmon populations that are already struggling after many decades of reduced and degraded stream habitat. This in turn has serious implications for the salmon fishery. As a mixed stock fishery, the ocean season can be sharply curtailed to protect weaker stocks (many of which are protected under the Endangered Species Act), even when other stocks remain relatively strong. Anomalous conditions off the West Coast from 2013 to 2015 have raised the possibility of another poor salmon fishing season in 2017 and beyond. The large mass of unusually warm water known as “the Blob” resulted in exceptionally high sea surface temperatures off much of the coast beginning in late 2013 and lasting through 2015 and was associated with greatly reduced upwelling and low productivity in many areas during 2014–2015 (Bond et al., 2015; Peterson et al., 2015a,b). As the Blob faded, the El Niño event of 2015–2016 again brought increased temperatures and decreased upwelling (Jacox et al., 2016). This El Niño is among the strongest in recent history, with the Oceanic Niño Index tying with the previous record set in 1997–1998. In addition to changes brought by El Niño and the Blob, the PDO changed from negative to positive sign in 2014, indicating a shift to a warmer, lower-productivity phase that is associated with lower salmon returns along the West Coast (Mantua et al., 1997). Together, these ocean conditions have already brought about serious economic consequences for the Dungeness crab, Pacific whiting, market squid, salmon, and pink shrimp fisheries (Peterson et al., 2016). For salmon, these conditions are likely to result in reduced survival of cohorts returning in the next several years (Leising et al., 2015). Knowledge of the potential impacts of a limited salmon season may aid decision making for managers, fishers, and other stakeholders confronted with the potential of greatly reduced salmon availability, or potential closures, over the next few years. For example, a model of the coho salmon fishery indicated that accurate El Niño forecasts 1–1.5 years in advance of the event could result in up to $1 million in increased revenue (Costello et al., 1998). However, this may greatly underestimate the value of forecasts because it only accounts for the change in direct profits in the fishery resulting from optimizing escapement over time. It does not account for impacts on other fisheries that might be managed better or the potential for fishers to capitalize on advance knowledge of fishery productivity to better utilize their fishing assets and labour. Predicting the effects of a fishery collapse or closures is complex. The variable strategies and characteristics of individual fishers can result in differing behavioural responses to policy changes and/or altered resource availability (e.g. Zhang and Smith, 2011). This fleet-level heterogeneity may be especially important in fisheries like the West Coast salmon fishery, which includes a wide range of vessel sizes with diverse fishing strategies and involvement in a range of other fisheries. These vessels are unlikely to respond to anomalous events in a uniform manner; thus, it is useful to quantify how the effects of the most recent closure of the fishery varied across vessels, with the goal of identifying the characteristics of vessels that are more vulnerable to the effects of reduced salmon availability. In addition, changes in salmon availability may cause fishers to divert some or all of their effort into alternate fisheries, and this is likely to vary across vessels and depend on which other fisheries they participate in. Such shifts may then affect the profitability and sustainability of other fisheries. Changes in fishing patterns may also affect coastal economies as the revenue brought in by vessels shifts in magnitude, timing, and/or space. We conduct an analysis of the direct and indirect effects of the 2008–2009 West Coast Salmon fishery closure with the goal of identifying which fishers were affected, how they responded to the closure, and whether there were substantial indirect effects on other fisheries. We develop and apply methods that can be used to predict impact of potential future closures, which can provide fishery managers, participants, and other stakeholders, with an opportunity to prepare and perhaps mitigate impacts. We focus on the behaviour of troll fishing vessels before, during, and after the 2008–2009 closure, with the goal of identifying the characteristics of fishers that are linked to increased vulnerability or resilience to the closure. In addition, we explore whether the salmon troll fleet altered their effort in other fisheries during the closure, and quantify the effects of the closure on port-level revenues. Specifically, we model the vessel-level decision to fish each year as a function of fisher characteristics including revenue level, diversification, dependence on salmon, and spatial descriptors of area and range. We further model relative revenue each year as a function of fisher characteristics, and address whether the closure differentially impacted vessel fishing behaviour and revenue. Because some vessels appear to have exited fishing entirely, we model the vessel-level decision to stay or leave fishing following the closure as a function of vessel characteristics and examine differences between the “stayers” and “leavers”. In addition, we examine whether vessels altered their participation in non-salmon fisheries during the closure. Finally, we examine whether relative port-level dependence on salmon was correlated with reduced revenue during the closure. Methods Background on the study system Major fisheries on the US West Coast include salmon, Dungeness crab, whiting, non-whiting groundfish (including flatfish, rockfish, and sablefish), albacore tuna, squid, coastal pelagics, shrimp, and other shellfish. There is a high degree of interdependency among fisheries (including salmon), as they are linked by common environmental and economic factors, as well as by the high degree of overlap in participation. Fishers are thought to move between fisheries across the season according to changes in profitability, species distributions, and regulations. Most fisheries are limited-entry, meaning that fishers must buy an existing license before entering, and many already hold licenses in multiple fisheries. These fisheries are governed by a complex mixture of state and federal management, with a few straddling stocks also covered under international treaties. In Washington, the salmon and steelhead fisheries are co-managed by American Indian tribes who have rights to half of the catch in their usual and accustomed fishing grounds. Though current salmon runs are only a fraction of their historical sizes, the salmon fishery is one of the more important fisheries on the West Coast, with landings valued at ∼US$35 million during the most recent peak in 2013 (PMFC, 2016). There are five species of Pacific salmon on the US West Coast, each of which is comprised of a number of runs associated with certain freshwater spawning habitats and (for Chinook) spawning migration seasons. Chinook salmon dominate ocean commercial catches, though some coho and small numbers of pink, sockeye, and chum salmon are also taken. Ocean catches are highest in California in most years, followed by Oregon and Washington. Hatchery supplementation is used to enhance production in the face of declining natural stocks, particularly for Chinook and coho. Due to their anadromous life history, both freshwater and marine conditions affect salmon survival, and environmental variability can cause large fluctuations in salmon returns across space and time. The ocean salmon season must be implemented such that it meets escapement and rebuilding goals across runs, and as mentioned earlier the fishery may be highly restricted to protect stocks with low projected returns. The SRFC stock is particularly important for the fishery, historically providing 80–95% of the California ocean harvest (CDFW, 2013) as well as a portion of catches in Oregon. The commercial ocean salmon fishery relies on trolling, while river, sound, and estuarine fisheries use gillnets and purse seines. Well over 1000 fishing vessels participate in West Coast salmon troll fisheries in most years, typically in combination with other fisheries (particularly Dungeness crab, non-whiting groundfish, and albacore tuna). These fisheries are seasonal, meaning that some are complimentary (e.g. the winter Dungeness crab fishery and the summer salmon fishery), and some may be at least partially substitutable (e.g. the summer albacore and salmon fisheries). Definition of salmon ocean troll fleet When modelling fishing behaviour and impacts of the closure on the salmon fleet our unit of observation is the fishing vessel. In most cases vessels belong to an owner-operator, though in some cases firms own multiple vessels or vessels are owned by more than one person. Although it might be preferable to model firm or individual behaviour, data availability limits us to modelling vessel behaviour. To define our group of focal vessels (henceforth referred to as the salmon troll fleet), we first identified any commercial vessels that participated in the salmon troll fishery in any capacity from 2001 to 2007 (the year before the closure). Of these vessels, we selected vessels that met the following criteria: Total annual revenue from salmon troll fishing averaged at least $1000 per year over 2001–2007 (excluding any years when the vessel did not fish). Revenue from salmon troll averaged at least 5% of total vessel revenue per year over 2001–2007. The vessel fished at least 3 of 7 years in 2001–2007. Vessels that met these criteria formed a cohort of focal vessels that we followed through time in order to quantify their responses to the closure. All data were taken from the Pacific States Marine Fisheries Commission’s PacFin database (PacFin, 2016). Revenues were adjusted for inflation relative to 2005 using the personal consumption expenditure series (http://www.bea.gov/national/consumer_spending.htm). Any revenue from Dungeness crab landed in November and December was grouped with revenue in the next calendar year. This is because the Dungeness season typically runs from mid-November or December through the following spring and summer, and this method of grouping ensures that an entire fishing season is represented in each year. Fisher/vessel characteristics To identify vessel characteristics that may modulate the effects of the closure, for each vessel in each year we calculated 5-year moving averages (excluding the current year) of (i) total annual vessel revenue; (ii) percent of revenue from salmon troll; (iii) latitudinal centre of gravity (LCG); (iv) latitudinal inertia (LI); and (iv) Herfindahl-Hirschman index (HHI; Hirschman, 1964). We chose 5-year lagged averages in order to smooth out interannual variation while still identifying vessel characteristics that are likely to affect fishing behaviour in the current year. We did not include years when the vessel did not fish in calculating mean annual revenue. LCG for a given vessel in a given year is calculated as LCG=Σirevi⋅latiΣirevi(1) where revi is total revenue landed in port i in that year and lati is the latitude of that port. This can be thought of as a measure of a vessel’s typical landing location along the West Coast. Similarly, we calculated annual latitudinal revenue inertia for each vessel as LI=Σirevi(lati−LCG)2Σirevi(2) This describes the dispersion around the centre of gravity and can be thought of as a measure of how far a vessel tends to range from its mean landing location. The HHI, also called the Simpson diversity index, is a measure of income diversification that ranges from near zero to 10 000, with higher values indicating less diversified income sources. It is defined as HHI=Σjpj2(3) where pj is the percent of annual revenue from species group j landed by a given vessel in a given year. For details of the species groups used see Kasperski and Holland (2013). In order to facilitate interpretation, in all further analyses we use the inverse HHI (i.e. 1/HHI), such that a higher value indicates higher diversification. Models of fishing behaviour and revenue Previous modelling of fishery participation has focused primarily on the decision of when or where to fish conditional on participating in a particular fishery (Eales and Wilen, 1986; Curtis and Hicks, 2000; Hicks and Schnier, 2008; Abbott and Wilen, 2010, 2011; Haynie and Layton, 2010; Zhang and Smith, 2011). A smaller literature has allowed for non-participation within a season (Berman et al., 1997; Smith and Wilen, 2003; Kahui and Alexander, 2008) or longer-term decisions of fishery participation or entry-exit behaviour (Bockstael and Opaluch, 1983; Ward and Sutinen, 1994), but largely in a single-fishery context. These models generally assume participation or location choices are based on the expected revenue in the object fishery and/or location. They do not address the more complex fleet dynamics of West Coast fisheries where fishers may participate in multiple fisheries over the year, and the decisions of which fisheries to participate in—or whether to sit out fishing entirely in a given year—are likely highly interdependent. We are unaware of prior models that address how a closure of one fishery affects decisions of fishers to continue participation in other non-contemporaneous fisheries. We posit that the decision to continue fishing in other fisheries during the closure year, the fishing revenue generated by vessels that do continue fishing in the closure year, and the decision to permanently exit the fishery, will depend on how much revenue the vessels have been deriving from fishing (including salmon and non-salmon revenues) as well as the share of that revenue coming from salmon (i.e. dependence on salmon). While we do not have information on non-fishing income, we posit that vessels with low overall fishing income are likely to have non-fishery sources of income they may rely on in the event of a closure making their exit from the fishery more likely. We posit that vessels with a range of other fishing opportunities (as measured by their fishery revenue diversification) are more likely to continue fishing in other fisheries during the salmon closure. We also posit that the location of the vessel’s previous fishing area will impact fishing behaviour in the closure year since some vessels were able to continue fishing salmon north of the closed area. Finally, we hypothesize that the vessel’s spatial range may affect its behaviour and revenue, as more mobile vessels may be more flexible and able to utilize a wider range of fishing grounds. We took a two-step approach to modelling vessel behaviour and revenue. First, we fit binomial generalized linear mixed models (GLMMs) with a logit link to model the decision to fish each year. GLMMs are hierarchical models that allow for non-normal response variables as well as correlated data (e.g. repeated measures over time of the same individuals). Mixed models where the intercept is allowed to vary randomly across individuals or groups are similar to random-effects panel models used in econometrics, where unobserved individual or group effects are uncorrelated with the regressors. In our case, we hypothesized that the decision to fish in a given year would be influenced by a vessel’s characteristics (revenue level, revenue diversification, spatial descriptors, dependence on salmon, and number of years fished in the prior 5 years) as well as constants for the year and whether or not the ocean fishery closure was in effect. In addition, we hypothesized that vessel characteristics may modulate the response of vessels to the closure. Thus, we modelled the probability pyi that vessel i fishes in year y as  logit(pyi)=α+β1y+β2closure+β3mean.revenuei+β4mean.HHIi+β5mean.percent.trolli+β6mean.LCGi+β7mean.LIi+β8years.fishedi+β9closure⋅mean.revenuei+β10closure⋅mean.HHIi +β11closure⋅mean.percent.trolli+β12closure⋅mean.LCGi+β13closure⋅mean.LIi+β14closure⋅years.fishedi+ai(4) where closure is a dummy variable that takes on a value of 1 during the closure (2008 and 2009) and zero otherwise (In Equation (4) as well as (6–9), note that the variables on the right hand side are constructed with data from the prior 5 years to avoid endogeneity.). We define year as number of years since the first year in the time period we are modelling (2001–2015). The random intercept term ai is normally distributed. All numerical explanatory variables were centred and scaled by their standard deviation to facilitate model fitting and interpretation (Schielzeth, 2010). We calculated variance inflation factors to assess collinearity in the model and found that all values were <2.5, indicating no concerning collinearity. In addition, we calculated the condition number κ of the design matrix and found κ = 4.6, again indicating low collinearity. GLMMs were fitted using the package lme4 (Bates et al., 2015) in R 3.2.3 (R Development Core Team, 2015). Marginal and conditional pseudo-R2 values were calculated according to Nakagawa and Schielzeth (2013). To assess the predictive power of our model, we used the area under curve (AUC) of the receiver operating characteristic (ROC). This method, which originates in the signal processing literature and is commonly used to assess the fit or predictive power of dichotomous models, has increasingly been used in ecology, medicine, and other fields (Fielding and Bell, 1997). To construct the ROC we used k-fold cross-validation, where we randomly divided the vessels into groups of 100. We trained the model on all but one group, then tested it on the selected group. This was repeated on each group. To evaluate model performance we use the common rule of thumb where AUC ≤ 0.6 is failed, 0.6 < AUC ≤ 0.7 is poor, 0.7 < AUC ≤ 0.8 is fair, 0.8 < AUC ≤ 0.9 is good, and 0.9 < AUC ≤ 1.0 is excellent. For vessels that fished in a given year, we calculated their revenue anomaly zyi as the total revenue of fishing vessel i minus its long-term (2001–2015) mean revenue, scaled by the long-term standard deviation: zyi=revenueyi−mean.revenueisd.revenuei(5) This serves as a measure of vessel revenue relative to its average revenue over the entire time period. As the second step in our approach, we then model the revenue anomaly as zyi=α+β1y+β2closure+β3mean.revenuei+β4mean.HHIi+β5mean.percent.trolli+β6mean.LCGi+β7mean.LIi+β8years.fishedi+β9closure⋅mean.revenuei+β10closure⋅mean.HHIi+β11closure⋅mean.percent.trolli+β12closure⋅mean.LCGi+β13closure⋅mean.LIi+β14closure⋅years.fishedi+ε(6) We use a simple linear model here, as standardizing revenue using each vessel’s long-term mean and standard deviation accounts for individual vessel-specific effects. As our primary focus is on the change in revenue during the closure, we excluded vessels that did not fish in 2008 or 2009 from this analysis. For predictive purposes, we also modelled untransformed annual vessel revenue using the linear model revenueyi=α+β1y+β2closure+β3mean.revenuei+β4mean.HHIi+β5mean.percent.trolli+β6mean.LCGi+β7mean.LIi+β8years.fishedi+β9closure⋅mean.revenuei+β10closure⋅mean.HHIi+β11closure⋅mean.percent.trolli+β12closure⋅mean.LCGi+β13closure⋅mean.LIi+β14closure⋅years.fishedi+ε(7) Because diagnostic plots indicated heavy-tailed residuals, we took a robust regression approach using the package robustbase (Todorov and Filzmoser, 2009; Rousseeuw et al., 2016). Stayers and leavers We hypothesized that the salmon ocean fishery closure may have caused some vessels to exit fishing entirely. To identify vessels that permanently left fishing during 2008–2009, any vessels that did not have any commercial landings in 2008–2015 were classified as “leavers”, while vessels that did have commercial landings during this time were classified as “stayers” (Hackett et al., 2015). We then used a binomial generalized linear model (GLM) with a logit link (i.e. a logistic model) to model the decision to leave the fishery and not return following the closure. Thus, we modelled the probability p of staying in the fishery as  logit(p)=α+β1mean.revenue+β2mean.HHI+β3mean.percent.troll+β4mean.LCG+β5mean.LI+β6years.fished(8) where the predictor variables are as defined above. We also compared the vessel characteristics of stayers and leavers using Wilcoxon-Mann-Whitney tests to determine whether the characteristics of these vessels differed significantly. Specifically, we examined differences in mean revenue, mean HHI, mean centre of gravity, mean inertia, and number of years fished across vessels that remained in the fishery and vessels that left fishing completely during the closure. Predictions Unfavourable ocean conditions from 2013 to 2015 suggest that low salmon returns are likely in the next few years. We demonstrate the utility of the models above by predicting the impacts of a potential closure in 2017. To do so, we first identify the current salmon troll fleet using the criteria outlined above. This includes vessels that participated in the salmon troll vessels over the 2009–2015, accounting for vessels that dropped out during the closure and vessels that entered during or since the closure. We then predict the proportion of vessels that will not fish, and the revenue of those that remain. For the binomial models, we chose cutoff values that maximized the sum of the model sensitivity and specificity. For simplicity, we assume that the extent and location of the closure is the same as that in 2008. We also compare mean vessel characteristics of the current fleet to characteristics of the pre-closure fleet in order to identify any changes in overall fleet characteristics that may influence responses to a closure. Port-level analyses Reductions in fishing revenue are likely to resonate across fishing communities, but the responses may differ across locations. To identify the dependence of individual ports on revenue from salmon troll vessels, we calculated the mean proportion of annual revenue from salmon troll vessels prior to the closure (2001–2007) for each port. Then, in each year we calculated the port-level revenue anomaly analogously to the vessel-level anomaly described above, using mean and standard deviation of total annual revenue landed in each port over 2001–2015. We then regressed port dependence on salmon troll vessels against port revenue anomaly in each year to determine if ports that were more dependent on salmon tended to have lower revenue during the closure. Participation in other fisheries We hypothesized that in years when salmon availability is limited, vessels that participate in the salmon fishery may shift some of their effort into other fisheries. This is based on the observation that most of our focal vessels participate in multiple fisheries, meaning they could potentially attempt to make up for lost salmon revenue by increasing effort in other fisheries. To explore whether vessels in the salmon troll fleet altered their participation in other fisheries during the salmon troll closure, we first identified vessels that participated in the groundfish, crab, and/or highly migratory species (HMS, mainly albacore tuna) fisheries in the 7 years preceding the closure. This was done to identify boats that likely had the ability (in terms of gear, expertise, and permits) to participate in those fisheries during the closure. We chose these three fisheries because they represent the main non-salmon fisheries that our focal fleet participates in. We then constructed a set of binomial GLMMs to examine the probability a vessel (within the subset of the fleet identified above) participates in each of these fisheries each year, given that it chooses to fish at all. The probability p that vessel i participates in a given fishery in year y is  logit(pyi)=α+β1y+β2closure+β3mean.revenuei+β4mean.HHIi+β5mean.percent.trolli+β6areai+β7mean.LIi+β8years.fishedi+β9closure⋅mean.revenuei+β10closure⋅mean.HHIi+β11closure⋅mean.percent.trolli+β12closure⋅areai+β13closure⋅mean.LIi+β14closure⋅years.fishedi+ai(9) Here the predictor variables are the same as the variables in the salmon fishery participation model, with the exception of area. This is a categorical value representing the location of the vessel’s mean centre of gravity (southern California, northern California, Oregon, or Washington) and was chosen to characterize typical vessel location while avoiding model convergence issues. Northern and southern California were defined as the areas within California north and south of Pt. Arena (39°N), respectively. In addition to the models above, we also examined total effort in other fisheries across all vessels in the fleet. This is important because the impact on other fisheries is likely mostly driven by the total change in effort across the fleet, regardless of changes in individual behaviour. To do so, we constructed time series of total number of trips undertaken by our focal fleet in the crab, groundfish, and HMS fisheries. We then used the method of Chen and Liu (1993) to detect outliers in these time series that might correspond to shifts in effort that occurred during the closure. This method identifies four different kinds of outliers (additive outliers, innovation outliers, level shifts, and temporary changes) in autoregressive moving average (ARMA) time series models. This was implemented using the package tsoutliers version 0.6 (López-de-Lacalle, 2015). In order to disentangle the potential effects of reduced participation during the closure (i.e. the fact that fewer boats overall participated in fishing), we also performed this analysis on the subset of vessels that fished at least 1 year in 2008 and 2009. We also hypothesized that the salmon closure may have altered the fleet-level seasonal distribution of effort across fisheries. That is, if vessels cannot fish for salmon during their usual season, they may fish for other species during that time, altering the intra-annual pattern of participation in that fishery. To test this hypothesis, we created monthly time series of the number of trips undertaken in each fishery (crab, groundfish, HMS, and salmon) and calculated the monthly share of total annual trips for each month. We then constructed seasonal dummy regressions where share of trips s taken in a fishery is smonth=α+β⋅month+ε(10) where month is a dummy variable. We then tested for structural change in these regressions using both the supF test (which tests for a single unknown breakpoint using F-statistics; Andrews, 1993) and the Bai–Perron test (Bai and Perron, 1998, 2003), which tests for the presence of multiple unknown breakpoints using the Bayesian information criterion and the residual sum of squares. These analyses were done using the strucchange package in R (Zeileis et al., 2002). Results Characteristics of the focal vessels We identified 1214 vessels that met the criteria for belonging to the salmon troll fishery prior to the closure, forming our focal group of vessels. In a given year, the number of these vessels participating in fishing of any kind ranged from a high of 1141 in 2004 to a low of 691 in 2009, not including any new entrants during or after the closure (Figure 1). These vessels were quite diverse with wide ranges of annual revenue and varying portfolios of target fisheries. Mean total annual per-vessel revenue from 2001–2015 ranged from $610 to $549 881, with a median of $24 273 (excluding years in which vessels did not fish; Figure 2). On average, vessels obtained 55% of their annual revenue from salmon, ranging from 2 to 100% (Figure 2). Other fisheries that these vessels participated in include groundfish (particularly sablefish), crab (almost exclusively Dungeness crab), and HMS (mainly albacore). A smaller number of vessels targeted shrimp (mainly spotted prawns and pink shrimp), coastal pelagics (mainly market squid and Pacific herring), or other species (e.g. halibut, hagfish, white seabass, and California spiny lobster). Mean vessel inverse HHI values indicated that most vessels were somewhat diversified, but 77 vessels fished salmon exclusively over this time period, giving them the minimum mean inverse HHI of 0.0001 (Figure 2). The distribution of vessel centres of gravity indicates that these vessels tend to be concentrated in central-north California, northern Oregon, and to a lesser extent northern Washington. Mean inertia values were generally low (median = 0.08), indicating that most vessels did not range very far from their centre of gravity. Vessels fished a median of 12 years during 2001–2015 (Figure 2). Figure 1. Open in new tabDownload slide Number of focal salmon troll vessels that participated in fishing each year. Figure 2. Open in new tabDownload slide Histograms of vessel characteristics averaged across 2001–2015. Total annual revenue from these vessels combined ranged from US$31 583 026 in 2009 to US$73 289 081 in 2012. Total revenue in 2008 was US$35 432 584 while the average annual revenue in the 5 years before the closure was US$56 200 864, indicating that the closure was associated with a reduction of US$45 386,118 in total revenue during the 2-year period relative to the preceding time period. Mean annual revenue from salmon troll among these vessels was US$19 041 815 in 2003–2007, compared with US$1 473 554 in 2008 and US$1 777 552 in 2008, indicating a loss of US$34 832 524 in salmon troll revenue during 2008–2009. Crab and salmon brought in the most revenue to these vessels, followed by groundfish and HMS, with relatively small overall contributions from coastal pelagics, shellfish, shrimp, and other groups (though these were important sources of revenue for a small number of individual vessels). Crab revenue was relatively low during the closure, likely due to natural fluctuations in crab abundance (Figure 3). Figure 3. Open in new tabDownload slide Total annual revenue from each management group harvested by vessels in the salmon troll fleet. Binomial fishing behaviour model The binomial GLMM indicated that all predictors were significant at the p = 0.05 level, with the exception of inertia and the main effect of percent of revenue from salmon troll. We did not drop inertia from the model as doing so resulted in slightly higher Akaike information criterion value (11521.1 vs. 11519.2). The marginal pseudo-R2 was 0.45 and the conditional pseudo-R2 was 0.62, suggesting that the model fixed effects explained ∼45% of the variance while the fixed and random effects together explained ∼62%. The AUC was 0.84, demonstrating good model fit. Year, closure, and centre of gravity all had significantly negative main effects, indicating that they decreased the probability a vessel fished in a given year. Revenue, diversification (inverse HHI), and the number of years fished in the past 5 years had significant positive effects (Supplementary Table S1; Figure 4a). The interaction between closure and revenue was negative, indicating that the positive effect of revenue was lessened during the closure. The interaction between closure and percent of revenue from salmon troll was also negative, suggesting that vessels more dependent on salmon troll revenue were less likely to fish during the closure years, though this vessel characteristic has no significant impact outside of the closure. The interaction between closure and diversification was positive, demonstrating that more diversified vessels were relatively more likely to fish during the closure years. The effect of number of years fished was also more positive during the closure years. Though the main effect of centre of gravity was negative, the interaction of closure and centre of gravity was positive, indicating that the closure affected vessels differentially across space, with more northern vessels more likely to fish during the closure years. Overall, vessel revenue, the presence of a closure, and the interaction between closure and vessel dependence on salmon troll had the largest effects, suggesting they have the greatest impact on whether or not a vessel fishes in a given year. Figure 4. Open in new tabDownload slide Coefficient estimates from models of (a) annual fishing participation; (b) revenue anomaly; (c) untransformed revenue; (d) decision to stay in fishing following the closure. Horizontal bars represent 95% CIs. Asterisks represent p-values (***p < 0.001, **p < 0.01, *p < 0.05). Revenue models Relative vessel revenue varied greatly across years, with the lowest median anomalies occurring during the closure (Figure 5). During these years, 82–83% of vessels had negative revenue anomalies, indicating that the large majority of vessels that fished in the closure years made less money than average. However, though our model identified significant predictors of vessel revenue anomaly (Supplementary Table S2; Figure 4b), overall fit was poor, with R2 = 0.09. The presence of the closure had the largest effect on revenue anomaly, and significantly negative interactions with closure indicate that vessels with higher revenue, greater dependence on salmon, and more years fished saw greater declines in relative revenue during the closure. Figure 5. Open in new tabDownload slide Annual vessel revenue anomaly. Points represent individual vessel values, while boxplots show the quartiles. The model of untransformed revenue had better fit, with R2 = 0.75. Unsurprisingly, mean revenue over the past 5 years was the strongest predictor of annual vessel revenue (Supplementary Table S3; Figure 4c). The closure was associated with a significantly negative effect on revenue, as was the interaction between closure and mean revenue. Stayers and leavers We found that 209 vessels exited fishing completely during the closure, representing ∼17% of the fleet. The GLM of the probability of remaining in the fishery suggested that vessels with higher revenue and a greater number of years fished were more likely to stay, while vessels that had a higher proportion of revenue from salmon troll were more likely to exit (Supplementary Table S4; Figure 4d). AUC was 0.79, indicating fair model fit. Though the model did not identify diversification or inertia as significant predictors, we found that on average, leavers were significantly less diverse and had lower inertia than stayers (Table 1). Leavers also tended to have lower mean revenue, a higher proportion of revenue from salmon troll, and fewer years fished. Diversification is positively correlated with average revenues, which could increase the standard error and reduce significance of this variable. Centre of gravity was not significantly different between stayers and leavers. Table 1. Average characteristics of stayers and leavers and results from Wilcoxon-Mann-Whitney tests. Vessel characteristic . Mean stayer value . Mean leaver value . p value . Number of years fished 4.54 3.74 <0.001 Mean percent salmon troll 0.59 0.82 <0.001 Mean revenue 53 541.65 16 630.56 <0.001 Mean centre of gravity 41.60 41.36 0.36 Mean inertia 0.74 0.29 <0.001 Mean inverse HHI 0.000148 0.000124 <0.001 Vessel characteristic . Mean stayer value . Mean leaver value . p value . Number of years fished 4.54 3.74 <0.001 Mean percent salmon troll 0.59 0.82 <0.001 Mean revenue 53 541.65 16 630.56 <0.001 Mean centre of gravity 41.60 41.36 0.36 Mean inertia 0.74 0.29 <0.001 Mean inverse HHI 0.000148 0.000124 <0.001 Open in new tab Table 1. Average characteristics of stayers and leavers and results from Wilcoxon-Mann-Whitney tests. Vessel characteristic . Mean stayer value . Mean leaver value . p value . Number of years fished 4.54 3.74 <0.001 Mean percent salmon troll 0.59 0.82 <0.001 Mean revenue 53 541.65 16 630.56 <0.001 Mean centre of gravity 41.60 41.36 0.36 Mean inertia 0.74 0.29 <0.001 Mean inverse HHI 0.000148 0.000124 <0.001 Vessel characteristic . Mean stayer value . Mean leaver value . p value . Number of years fished 4.54 3.74 <0.001 Mean percent salmon troll 0.59 0.82 <0.001 Mean revenue 53 541.65 16 630.56 <0.001 Mean centre of gravity 41.60 41.36 0.36 Mean inertia 0.74 0.29 <0.001 Mean inverse HHI 0.000148 0.000124 <0.001 Open in new tab Current fleet and predictions We identified 1089 vessels belonging to the current salmon troll fleet, indicating that the current fleet is ∼10% smaller than the pre-closure fleet. This accounts for the 209 vessels that exited fishing entirely, as well as 297 vessels that continued fishing but no longer met our criteria for being salmon troll vessels. It also includes 391 vessels that entered the salmon troll fleet from 2008 onwards. On average, the newer fleet is less dependent on salmon but also slightly less diverse (Table 2). Though mean centre of gravity does not differ, the newer fleet has greater inertia, suggesting more movement across space. Vessels in the new fleet fished fewer years, but this may be driven by new entrants that have a short fishing history. We show the average characteristics of the new entrants in Table 3. Table 2. Mean vessel characteristics of the pre- and post-closure fleets and results from Wilcoxon-Mann-Whitney tests. Vessel characteristic . Mean 2001–2007 fleet . Mean 2009–2015 fleet . p value . Mean revenue 45 268 43 253 0.107 Mean inverse HHI 0.0001412 0.0001403 0.020 Mean percent salmon troll 0.532 0.486 <0.001 Mean centre of gravity 41.55 41.59 0.32 Mean inertia 0.6302 0.6319 <0.001 Number of years 3.831 3.410 <0.001 Vessel characteristic . Mean 2001–2007 fleet . Mean 2009–2015 fleet . p value . Mean revenue 45 268 43 253 0.107 Mean inverse HHI 0.0001412 0.0001403 0.020 Mean percent salmon troll 0.532 0.486 <0.001 Mean centre of gravity 41.55 41.59 0.32 Mean inertia 0.6302 0.6319 <0.001 Number of years 3.831 3.410 <0.001 Open in new tab Table 2. Mean vessel characteristics of the pre- and post-closure fleets and results from Wilcoxon-Mann-Whitney tests. Vessel characteristic . Mean 2001–2007 fleet . Mean 2009–2015 fleet . p value . Mean revenue 45 268 43 253 0.107 Mean inverse HHI 0.0001412 0.0001403 0.020 Mean percent salmon troll 0.532 0.486 <0.001 Mean centre of gravity 41.55 41.59 0.32 Mean inertia 0.6302 0.6319 <0.001 Number of years 3.831 3.410 <0.001 Vessel characteristic . Mean 2001–2007 fleet . Mean 2009–2015 fleet . p value . Mean revenue 45 268 43 253 0.107 Mean inverse HHI 0.0001412 0.0001403 0.020 Mean percent salmon troll 0.532 0.486 <0.001 Mean centre of gravity 41.55 41.59 0.32 Mean inertia 0.6302 0.6319 <0.001 Number of years 3.831 3.410 <0.001 Open in new tab Table 3. Mean vessel characteristics of new entrants. Vessel characteristic . Mean . Mean revenue 40 671 Mean inverse HHI 0.0001347 Mean percent salmon troll 0.447 Mean centre of gravity 42.11 Mean inertia 0.4905 Number of years 3.085 Vessel characteristic . Mean . Mean revenue 40 671 Mean inverse HHI 0.0001347 Mean percent salmon troll 0.447 Mean centre of gravity 42.11 Mean inertia 0.4905 Number of years 3.085 Open in new tab Table 3. Mean vessel characteristics of new entrants. Vessel characteristic . Mean . Mean revenue 40 671 Mean inverse HHI 0.0001347 Mean percent salmon troll 0.447 Mean centre of gravity 42.11 Mean inertia 0.4905 Number of years 3.085 Vessel characteristic . Mean . Mean revenue 40 671 Mean inverse HHI 0.0001347 Mean percent salmon troll 0.447 Mean centre of gravity 42.11 Mean inertia 0.4905 Number of years 3.085 Open in new tab Our model predicts that in the event of another closure, only ∼47% of vessels in the current fleet would fish at all during that year (Figure 6). Of the vessels that we predict would fish, our models predict a mean annual revenue of $65 750 (five vessels were predicted to have negative revenue, so we did not include these vessels in further calculations). In contrast, if there is no closure, we predict that 82% of vessels would fish, and that the average vessel would make $71 154. Overall, the fleet is predicted to have total annual revenue of $63 825 400 without a closure and $33 466 993 with a closure, suggesting a loss of over $30 million in a single year (Figure 6). Assuming the magnitude of the closure would be similar to the last closure, our model predicts that 303 vessels would exit fishing completely, representing nearly 30% of the fleet. Figure 6. Open in new tabDownload slide Time series of actual and predicted number of vessels (top) and total revenue (bottom). Solid lines represent actual values, dashed lines are predicted values in the absence of a closure, and dotted lines are predicted values in the presence of a closure. Grey bands are bootstrapped 95% prediction intervals. Port level analyses Regressions of importance of the salmon troll fleet to ports (i.e. mean proportion of total annual fishery revenue from salmon troll vessels) against port revenue anomaly were non-significant in all years except 2008–2010 (negative relationship) and 2015 (positive relationship; Figure 7). This indicates that ports that gained more revenue from salmon troll vessels had significantly lower revenue during the closure, and that this effect persisted into the year after the closure. These ports tended to be in central California, northern California, and Oregon. However, these relationships are noisy; indicating that dependence on salmon only drives part of the fluctuations in port-level revenue. Total revenue across time and latitude is shown in Figure 8. Figure 7. Open in new tabDownload slide Regressions of mean percent of port revenue from salmon troll vessels versus revenue anomaly. Only years with significant relationships are shown. Figure 8. Open in new tabDownload slide Total revenue from salmon troll vessels across time and latitude. Participation in other fisheries Within the salmon fleet, we identified 496 vessels that participated in the crab fishery, 378 vessels that participated in the groundfish fishery, and 428 vessels that participated in the HMS fishery prior to the closure. The proportion of each of these sub-fleets participating in each fishery annually is show in Figure 9 (proportion of the entire fleet participating in the salmon fishery is shown for comparison). Our models indicated that vessels were significantly more likely to participate in the crab fishery during the closure, though this effect was modulated for Oregon and Washington vessels (Supplementary Table S5; Figure 10a). Our model also suggested that vessels with a lower inertia were more likely to harvest crab during the closure. There was little evidence that the closure increased participation in the groundfish (Supplementary Table S6; Figure 10b) or HMS fisheries (Supplementary Table S7; Figure 10c), though there was some indication that more diversified vessels were less likely to harvest groundfish during the closure. In addition, there was some suggestion that northern California and Oregon vessels, as well as high inertia vessels were more likely to harvest HMS during the closure, but these effects were not statistically significant. Figure 9. Open in new tabDownload slide Proportion of subfleets that participate in major alternate fisheries. Figure 10. Open in new tabDownload slide Coefficients from model of participation in the (a) crab fishery; (b) groundfish fishery; and (c) HMS fishery. Horizontal bars represent 95% CIs. Asterisks represent p-values (***p < 0.001; **p < 0.01; *p < 0.05). Analysis of total fleet-wide trips in the crab, groundfish, and HMS fisheries did not identify any outliers in each of these time series (Figure 11). This was true even if we extended our time series back to 1995, and if we limited the vessels to those that fished during the closure. This indicates that there was little overall change in effort (in terms of number of trips) in other fisheries over 2001–2015, in spite of a significant proportion of vessels exiting temporarily during the closure and/or permanently following the closure. In contrast, analysis of the fleet-wide number of salmon trips identified a level shift in 2006 and a temporary change in 2008. Extending the time series to 1995 resulted in the additional identification of temporary change in 2004 and an additive outlier in 2007 (Supplementary Figure S1). Figure 11. Open in new tabDownload slide Time series of number of trips in the main fisheries targeted by the salmon troll fleet. Left panel shows trips from all vessels and right panel shows trips by vessels that fished during the closure. In the seasonal time series of fleet-wide proportion of trips, F-statistics suggest a structural change in the seasonality of groundfish trips between 2003 and 2006 (with the sup-F test identifying the breakpoint at June 2003), and of salmon trips around 2007 and 2009–2010 (with the sup-F test identifying the breakpoint at May 2010; Figure 12, Supplementary Figure S2). The Bai and Perron method did not identify any breakpoints in the groundfish seasonal time series, but again found support for a structural change in salmon trip seasonality in May 2010 (Figure 12; Supplementary Figure S3). Neither method found evidence of structural changes in crab or HMS trip seasonality. Thus, there is little evidence that the salmon fishery closure altered the intra-seasonal allocation of effort in other fisheries, though it may have impacted the seasonality of the salmon fishery itself. Figure 12. Open in new tabDownload slide Proportion of total annual trips taken in each month in each management group. Dashed line shows the breakpoint in the groundfish time series identified by the sup-F test and dotted line shows the breakpoint in the salmon time series identified by both the sup-F test and Bai–Perron test. Discussion As the earth’s climate continues to change, oceanographic variability is likely to have increasing impacts on marine ecosystems. Changes in ocean conditions that alter productivity of particular fisheries can alter behaviour of fishers affecting their harvest, not only of the species directly impacted by oceanic changes, but other species they target. Fishery closures or collapses may have impacts that persist long after the fishery recovers. Anticipating and effectively responding to oceanic variability thus requires understanding whether and how fishers will react to changes by altering their harvesting activities in and beyond directly impacted fisheries, and in the short and long term. We analysed the 2008–2009 salmon ocean troll fishery closure to quantify how a large fluctuation in resource availability may affect salmon fishers, other fisheries, and fishing communities. Our results indicate that most salmon fishers in the California Current typically have limited alternatives in the face of sharply reduced salmon availability, and this is particularly true for vessels with low revenue diversification. Fishing vessels that are highly dependent on salmon are likely to cease fishing (some permanently) rather than attempting to move into other fisheries. This may be because of difficulty obtaining permits, the cost of retrofitting boats for other fisheries, and/or because fishers gain personal satisfaction from salmon fishing that they would not find elsewhere. The vessels that exited fishing permanently tended to be more dependent on salmon, less diversified, and lower revenue, indicating that these types of vessels may be less resilient to a closure or decline in salmon availability. Vessels that participate in multiple fisheries are more likely to remain active during a salmon fishery closure, albeit often with lower revenue relative to average levels. Though diversification was positively associated with being active in fishing during the closure, we found only limited evidence that these vessels increased their participation in other fisheries. There is evidence that those who previously fished for Dungeness crab were more likely to do so, though the net change in overall effort in the crab fishery resulting from this was probably not large. Dungeness crab recruitment and landings (which are tightly linked to the size of fishable crab population) can fluctuate by an order of magnitude across years due environmental conditions (Shanks, 2013), and crab catches were relatively low in 2008–2009. Low catch rates in the crab fishery may have reduced the degree to which fishers affected by the salmon closure participated in the crab fishery. The limited displacement of effort into other fisheries during the salmon fishery closure may also partly stem from the complimentary nature of the fishing portfolios often targeted by fishers. For example, while the ocean salmon fishery peaks in the summer, the crab fishery usually opens in December and the majority of catches are usually landed in the first 4–6 weeks of the season, meaning there is little temporal overlap in these fisheries. The groundfish fishery peaks in the summer; however, bimonthly catch limits (for the limited entry fishery) and daily/weekly trip limits (for the open-access fishery) may have limited the ability of vessels to increase their groundfish landings to compensate for the lack of salmon. The albacore fishery peaks in the late summer/early fall, but vessels (particularly smaller vessels) may be limited by weather, hold size, and range, as well as the varying migration patterns of albacore (Dotson, 1980). Access to most West Coast fisheries is also restricted to a limited number of license holders, so fishers cannot freely enter them unless they already have a license or can purchase an existing license from another vessel. Consequently, though many fishers have relatively diversified fishing portfolios, they may already be maximizing their investment in other available fisheries, and have few feasible fishing alternatives during the salmon season. This is supported by our results finding no abrupt changes in the seasonal pattern of trips in the groundfish, crab, and HMS fisheries during 2008–2009. This is also supported at least anecdotally by fishers; for example, one who was interviewed by Ackerman et al. (2016) said in reference to the salmon closures, The impact for that was tremendous, in that you have communities, for example Garibaldi, where the fisherman, who depend on salmon fishing as part of their fisheries income could not go fishing. If you take away one of those income streams then it’s not like you can create more by increasing your catch with albacore or crab. Thus, though most salmon fishers also participate in other fisheries, those fisheries appear to provide a limited buffer against the effects of poor or no salmon fishing. For vessels that remain in the fishery, the primary strategy in the face of a poor season may be to wait it out, rather than attempt to greatly alter their investment in other fisheries. The adaptive strategies available to fishers are not limited to their allocation of effort across fisheries. Many of the vessels included in this analysis are relatively low-revenue vessels belonging to small independent owner–operators, and many of these fishers likely have alternate sources of income outside of fishing. This may allow them to cease fishing when conditions are poor, and/or to supplement their fishing income during bad years. However, we currently lack data to quantify these alternate forms of employment and their potential effects on fishing behaviour. Future planned work includes surveys of fishers to gather information about employment outside of fishing, non-monetary benefits of fishing, and other factors that may influence fishing behaviour. This will provide a more complete picture of the factors motivating behaviour and allow us to create more robust models of movement between multiple fisheries, as well as other sectors. Although this study indicated fairly limited indirect impacts on other fisheries resulting from effort displacement following the salmon fishery closure, the same may not be true for other fisheries subject to impacts of climate variation. For example, in 2015–2016 there was a delay of the California Dungeness crab fishery due to high levels of domoic acid resulting from a harmful algal bloom (HAB) that has been attributed to the Blob (McKibben et al., 2017). Total California crab landings in the 2015–2016 season were down about 50% from the prior year, and there were reports that many crab fishers shifted their effort into the daily trip limit sablefish fishery, resulting in an in-season action by the Pacific Fishery Management Council to lower trip limits to avoid an early closure of the fishery. Researchers and policymakers should explore the behaviour of participants in this and other fisheries that are affected by environmental changes. Responses and effort displacement into other fisheries may differ from what we observed with the salmon closure. Though we focus our analyses on the closure of 2008–2009, crises in the West Coast salmon fishery have been occurring for the past several decades, including a partial closure in Oregon in 2006. As a consequence of that event, as well as a relatively poor season in 2007, much of the salmon fishery was already struggling when the unprecedented closure of 2008–2009 came about. The coupled marine and terrestrial environments associated with the California Current have already become increasingly variable over the past decades (Black et al., 2014), and future events are likely to include more poor salmon periods of varying spatial and temporal extents. While our analyses focus on the particular events of 2008–2009, we believe that the general patterns we identified are likely to apply to future scenarios, including perhaps the next few years. The methods used in this study should also be generalizable to other fisheries subject to fishery closures associated low productivity or periodic events such HABs. The results of this study are especially timely as the changes brought by drought, El Niño, the Blob, and the switch to a positive PDO indicate poor conditions for salmon returning in 2016–2018. If there is another closure of similar magnitude (as it appears there will be in 2017), our results suggest that the current salmon troll fleet is likely to face similar consequences. This would mean a large portion of the fleet would cease fishing temporarily, and some would likely leave fishing permanently, with large implications for the fishery and associated industries and communities. However, there is little to suggest that salmon fishers would greatly increase their effort in other fisheries, meaning that direct impacts on the crab, groundfish, and HMS fisheries are likely to be relatively low. It is worth noting that the anomalous environmental conditions over the past few years could lead to negative impacts on fisheries that persist longer than the 2-year closure we focus on in this study. More productive ocean conditions may not return until late 2017 or 2018 (Peterson et al., 2016) meaning salmon returns are not likely to improve for several years following that. Though many fishers in this study were able to survive a 2-year closure, a longer period of highly restricted fishing may cause more attrition and/or movement to other fisheries. Fishers may also need to develop new adaptive strategies, such as diversifying their fishing and non-fishing revenue streams. In the longer term, changes in climate, stream, and oceanographic conditions may mean that highly restricted salmon seasons may become more common, and other species may face similar changes. As noted earlier, the Blob was associated unprecedented delays in the crab season, as well as negative impacts on other ocean fisheries. The conditions brought by the Blob have been suggested to foreshadow future conditions in the Pacific, with serious consequences for the CCLME and the livelihoods that depend on it. Managers and policymakers may want to consider how fishers are likely to respond to these events, and how policy changes could potentially mitigate some of the unwanted impacts. For example, policymakers might weigh the benefits of facilitating movement among substitute fisheries with the potential impacts on those fisheries and the wider ecosystem. Though a closure may only last 1–2 years, ecosystem and economic effects may persist over much longer time periods, necessitating careful consideration of trade-offs among management options. Supplementary data Supplementary material is available at the ICESJMS online version of the manuscript. Funding This work was funded by the National Oceanic and Atmospheric Administration (NOAA), the NOAA Integrated Ecosystem Assessment Program, and the National Science Foundation (grant number 1616821). References Andrews D. W. K. 1993 . Tests for parameter instability and structural change with unknown change point . Econometrica , 61 : 821 – 856 . Google Scholar Crossref Search ADS WorldCat Abbott J. K. , Wilen J. E. 2010 . Voluntary cooperation in the commons? Evaluating the Sea State program with reduced form and structural models . Land Economics , 86 : 131 – 154 . Google Scholar Crossref Search ADS WorldCat Abbott J. K. , Wilen J. E. 2011 . Dissecting the tragedy: a spatial model of behavior in the commons . Journal of Environmental Economics and Management , 62 : 386 – 401 . Google Scholar Crossref Search ADS WorldCat Ackerman R. , Neuenfeldt R., Eggermont T., Burbidge M., Lehrman J., Wells N., Chen X. 2016 . Resilience of Oregon Coastal Communities in Response to External Stressors . Master’s Thesis, University of Michigan, Ann Arbor, Michigan . Google Scholar Bai J. , Perron P. 1998 . Estimating and testing linear models with multiple structural changes . Econometrica , 66 : 47 – 78 . Google Scholar Crossref Search ADS WorldCat Bai J. , Perron P. 2003 . Computation and analysis of multiple structural change models . Journal of Applied Econometrics , 18 : 1 – 22 . Google Scholar Crossref Search ADS WorldCat Bates D. , Mächler M., Bolker B., Walker S. 2015 . Fitting linear mixed-effects models using lme4 . Journal of Statistical Software , 67 : 48. Google Scholar Crossref Search ADS WorldCat Berman M. , Haley S., Kim H. 1997 . Estimating net benefits of reallocation: discrete choice models of sport and commercial fishing . Marine Resource Economics , 12 : 307 – 327 . Google Scholar Crossref Search ADS WorldCat Black, B. A., Sydeman, W. J., Frank, D. C., Griffin, D., Stahle, D. W., García-Reyes, M., Rykaczewski, R. R., et al. 2014. Six centuries of variability and extremes in a coupled marine-terrestrial ecosystem. Science, 345: 1498–1502. Bockstael N. E. , Opaluch J. J. 1983 . Discrete modelling of supply response under uncertainty: the case of the fishery . Journal of Environmental Economics and Management , 10 : 125 – 137 . Google Scholar Crossref Search ADS WorldCat Bond N. A. , Cronin M. F., Freeland H., Mantua N. 2015 . Causes and impacts of the 2014 warm anomaly in the NE Pacific . Geophysical Research Letters , 42 : 3414 – 3420 . Google Scholar Crossref Search ADS WorldCat CDFW (California Department of Fish and Wildlife) . 2013 . Status of the fisheries report: an update through 2011. Report to the California Fish and Game Commission as directed by the Marine Life Management Act of 1998. Chen C. , Liu L.-M. 1993 . Joint estimation of model parameters and outlier effects in time series . Journal of the American Statistical Association , 88 : 284 – 297 . Google Scholar OpenURL Placeholder Text WorldCat Costello C. J. , Adams R. M., Polasky S. 1998 . The value of El Niño forecasts in the management of salmon: a stochastic dynamic assessment . American Journal of Agricultural Economics , 80 : 765 – 777 . Google Scholar Crossref Search ADS WorldCat Curtis R. , Hicks R. L. 2000 . The cost of sea turtle preservation: the case of Hawaii’s pelagic longliners . American Journal of Agricultural Economics , 82 : 1191 – 1197 . Google Scholar Crossref Search ADS WorldCat Di Lorenzo E. , Schneider N., Cobb K., Franks P., Chhak K., Miller A., McWilliams J. et al. 2008 . North Pacific Gyre Oscillation links ocean climate and ecosystem change . Geophysical Research Letters , 35 : L08607. Google Scholar Crossref Search ADS WorldCat Doney S. C. , Ruckelshaus M., Duffy J. E., Barry J. P., Chan F., English C. A., Galindo H. M. et al. 2012 . Climate change impacts on marine ecosystems . Marine Science , 4 : 11 – 37 . Google Scholar Crossref Search ADS WorldCat Dotson R. C. 1980 . Fishing methods and equipment of the US West Coast albacore fleet . NOAA-NMFS Technical Memorandum NOAA-TM-NMFS-SWFC-8 . Google Scholar Eales J. , Wilen J. E. 1986 . An examination of fishing location choice in the pink shrimp fishery . Marine Resource Economics , 2 : 331 – 351 . Google Scholar Crossref Search ADS WorldCat Fielding A. H. , Bell J. F. 1997 . A review of methods for the assessment of prediction errors in conservation presence/absence models . Environmental Conservation , 24 : 38 – 49 . Google Scholar Crossref Search ADS WorldCat Hackett S. , Pitchon A., Hansen D. 2015 . Economic attributes of stayers and leavers in four California fisheries . CalCOFI Reports , 56 : 1 – 10 . Google Scholar OpenURL Placeholder Text WorldCat Haynie A. C. , Layton D. F. 2010 . An expected profit model for monetizing fishing location choices . Journal of Environmental Economics and Management , 59 : 165 – 176 . Google Scholar Crossref Search ADS WorldCat Hicks R. L. , Schnier K. E. 2008 . Eco-labeling and dolphin avoidance: a dynamic model of tuna fishing in the Eastern Tropical Pacific . Journal of Environmental Economics and Management , 56 : 103 – 116 . Google Scholar Crossref Search ADS WorldCat Hirschman A. O. 1964 . The paternity of an index . The American Economic Review , 54 : 761 – 762 . Google Scholar OpenURL Placeholder Text WorldCat Jacox M. G. , Hazen E. L., Zaba K. D., Rudnick D. L., Edwards C. A., Moore A. M., Bograd S. J. 2016 . Impacts of the 2015–2016 El Niño on the California Current System: Early assessment and comparison to past events . Geophysical Research Letters , 7072 – 7080 . Google Scholar OpenURL Placeholder Text WorldCat Kahui V. , Alexander W. R. J. 2008 . A bioeconomic analysis of marine reserves for Paua (Abalone) Management at Stewart Island, New Zealand . Environmental and Resource Economics , 40 : 339 – 367 . Google Scholar Crossref Search ADS WorldCat Kasperski S. , Holland D. S. 2013 . Income diversification and risk for fishermen . Proceedings of the National Academy of Sciences of the United States of America , 110 : 2076 – 2081 . Google Scholar Crossref Search ADS PubMed WorldCat King J. R. , Agostini V. N., Harvey C. J., McFarlane G. A., Foreman M. G., Overland J. E., Di Lorenzo E., et al. 2011 . Climate forcing and the California Current ecosystem . ICES Journal of Marine Science: Journal du Conseil , 68 : 1199 – 1216 . Google Scholar Crossref Search ADS WorldCat Leising A. W. , Schroeder I. D., Bograd S. J., Abell J., Durazo R., Gaxiola-Castro G., Bjorkstedt E. P., et al. 2015 . State of the California Current 2014-15: Impacts of the Warm-Water “Blob”. CalCOFI Reports, 56: 31–68. Lindley S. T. , Grimes C. B., Mohr M. S., Peterson W., Stein J., Anderson J. T., Botsford L. W., et al. 2009 . What caused the Sacramento River fall Chinook stock collapse? NOAA Tech Memo NMFS-SWFSC, NOAA-TM-NMFS-SWFSC-447. López-de-Lacalle, J. 2015. tsoutliers: Detection of Outliers in Time Series. R package version 0.6. Available at https://cran.r-project.org/web/packages/tsoutliers/tsoutliers.pdf (last accessed 27 April 2017). Mantua N. J. , Hare S. R., Zhang Y., Wallace J. M., Francis R. C. 1997 . A Pacific interdecadal climate oscillation with impacts on salmon production . Bulletin of the American Meteorological Society , 78 : 1069 – 1079 . Google Scholar Crossref Search ADS WorldCat McKibben S. M. , Peterson W., Wood A. M., Trainer V. L., Hunter M., White A. E. 2017 . Climatic regulation of the neurotoxin domoic acid . Proceedings of the National Academy of Sciences of the United States of America , 114 : 239 – 244 . Google Scholar Crossref Search ADS PubMed WorldCat Nakagawa S. , Schielzeth H. 2013 . A general and simple method for obtaining R2 from generalized linear mixed-effects models . Methods in Ecology and Evolution , 4 : 133 – 142 . Google Scholar Crossref Search ADS WorldCat PacFin (Pacific Fisheries Information Network). 2016 . retrieval dated 3/16/2016, Pacific States Marine Fisheries Commission (PSFMC) , Portland, Oregon (www.psmfc.org). Peterson W. , Bond N., Robert M. 2016 . The Blob is gone but has morphed into a strongly positive PDO/SST pattern. PICES Press, 24: 46. Peterson W. , Robert M., Bond N. 2015a . The warm blob continues to dominate the ecosystem of the northern California current. PICES Press, 23: 44. Peterson W. , Robert M., Bond N. 2015b . The warm blob-Conditions in the northeastern Pacific Ocean. PICES Press, 23: 36. PMFC (Pacific Fishery Management Council) . 2016 . Review of 2015 Ocean Salmon Fisheries: Stock Assessment and Fishery Evaluation Document for the Pacific Coast Salmon Fishery Management Plan. (Document prepared for the Council and its advisory entities), Pacific Fishery Management Council, 7700 NE Ambassador Place, Suite 101, Portland, Oregon 97220-1384. R Development Core Team . 2015 . R: A Language and Environment for Statistical Computing . R Foundation for Statstical Computing , Vienna, Austria . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Rousseeuw P. , Croux C., Todorov V., Ruckstuhl A., Salibian-Barrera M., Verbeke T., Koller M. et al. 2016 . robustbase: Basic Robust Statistics. R package version 0.92-6. Schielzeth H. 2010 . Simple means to improve the interpretability of regression coefficients . Methods in Ecology and Evolution , 1 : 103 – 113 . Google Scholar Crossref Search ADS WorldCat Schroeder I. D. , Black B. A., Sydeman W. J., Bograd S. J., Hazen E. L., Santora J. A., Wells B. K. 2013 . The North Pacific High and wintertime pre-conditioning of California current productivity . Geophysical Research Letters , 40 : 541 – 546 . Google Scholar Crossref Search ADS WorldCat Schwing F. B. , Mendelssohn R., Bograd S. J., Overland J. E., Wang M., Ito S-i. 2010 . Climate change, teleconnection patterns, and regional processes forcing marine populations in the Pacific . Journal of Marine Systems , 79 : 245 – 257 . Google Scholar Crossref Search ADS WorldCat Shanks A. L. 2013 . Atmospheric forcing drives recruitment variation in the Dungeness crab (Cancer magister), revisited . Fisheries Oceanography , 22 : 263 – 272 . Google Scholar Crossref Search ADS WorldCat Smith M. D. , Wilen J. E. 2003 . Economic impacts of marine reserves: the importance of spatial behavior . Journal of Environmental Economics and Management , 46 : 183 – 206 . Google Scholar Crossref Search ADS WorldCat Sydeman W. J. , Thompson S. A., García-Reyes M., Kahru M., Peterson W. T., Largier J. L. 2014 . Multivariate ocean-climate indicators (MOCI) for the central California Current: Environmental change, 1990–2010 . Progress in Oceanography , 120 : 352 – 369 . Google Scholar Crossref Search ADS WorldCat Todorov V. , Filzmoser P. 2009 . An Object-Oriented Framework for Robust Multivariate Analysis . Journal of Statistical Software , 32 : 1 – 47 . Google Scholar Crossref Search ADS WorldCat Ward J. M. , Sutinen J. G. 1994 . Vessel entry-exit behavior in the Gulf of Mexico shrimp fishery . American Journal of Agricultural Economics , 76 : 916 – 923 . Google Scholar Crossref Search ADS WorldCat Zeileis A. , Leisch F., Hornik K., Kleiber C. 2002 . strucchange: an R package for testing for structural change in linear regression models . Journal of Statistical Software , 7 : 1 – 38 . Google Scholar Crossref Search ADS WorldCat Zhang J. , Smith M. D. 2011 . Heterogeneous response to marine reserve formation: a sorting model approach . Environmental and Resource Economics 49 : 311 – 325 . Google Scholar Crossref Search ADS WorldCat Zwolinski J. P. , Demer D. A. 2012 . A cold oceanographic regime with high exploitation rates in the Northeast Pacific forecasts a collapse of the sardine stock . Proceedings of the National Academy of Sciences of the United States of America , 109 : 4175 – 4180 . Google Scholar Crossref Search ADS PubMed WorldCat Published by International Council for the Exploration of the Sea 2017. This work is written by US Government employees and is in the public domain in the US. Published by International Council for the Exploration of the Sea 2017. This work is written by US Government employees and is in the public domain in the US.

Guan,, Lisha;Chen,, Yong;Staples, Kevin, W;Cao,, Jie;Li,, Bai

doi: 10.1093/icesjms/fsx064pmid: N/A

Abstract Atlantic cod (Gadus morhua) in the Gulf of Maine (GOM) is an iconic marine fishery stock that has experienced a substantial distributional shift since the mid-1990s. A geostatistical delta-generalized linear mixed model was utilized to hindcast yearly season-specific distributions of GOM cod. These distributions were calculated using the spring and fall bottom trawl survey data for the stock, along with cell-based bathymetry and bottom temperature data for the study area for the years 1982–2013. The centre of stock distribution (the centre of gravity), spatial extent in latitude and longitude, area occupied and median habitat temperature were estimated annually to quantify changes in the spatial dynamics of GOM cod. Time series of these distributional metrics were then used to evaluate the influences of climate change and density-dependent habitat selection on GOM cod’s distribution. Results showed that the rapid southwestward shift in the stock distribution after the late 1990s could not simply be attributed to decreasing stock abundance or warming bottom temperatures. The observed shift in cod distribution requires further investigation on whether it is possibly a result of other factors, like fluctuating productivity among subpopulations. Introduction Understanding of fish stock dynamics is essential for promoting sustainable and efficient use of fisheries resources. Conventional fisheries stock assessment focuses on the estimation of temporal dynamics in stock abundance, spawning stock biomass (SSB), recruitment and sustainable yield (Hilborn and Walters, 1992). Spatial dynamics (i.e. time-varying spatial patterns in the distribution and abundance) are often ignored, assuming individual fish within a stock are either homogeneously distributed or well-mixed within the management unit. Most marine fish populations, however, have complex spatial structures and are not evenly distributed over their distinct geographic ranges (Kritzer and Sale, 2004). Limited knowledge of population structure and spatial variability in key demographics has hindered the development of stock assessment models that account for such complex population structures. In this context, the spatial dynamics of fishery stocks are rarely well evaluated. Ignoring the spatial structure and dynamics of a stock can reduce the precision of abundance estimates and lead to a risk of overexploitation or local extirpation (Ying et al., 2011; Kerr et al., 2014; Thorson et al., 2015). Evaluating the spatial dynamics of marine fish populations can provide insights into their distinct population structures and responses to climate change. For a given population, if the geographic range has a strong positive association with the population size, changes in the spatial distribution are generally considered as a result of density dependent habitat selection (MacCall, 1990; Spencer, 2008). In this context, individuals within the population move freely to the most suitable habitats first, with less suitable habitats becoming occupied as the population size increases. As a contrast, when a fish stock consists of several subpopulations that are made up of smaller spawning components, shifts in the stock distribution for a specific season may be caused by fluctuations in the productivity of different subpopulations over years or even longer time spans (Hilborn et al., 2003; Ames and Lichter, 2013). For example, spawning groups in the southern range of the northern cod stock have rebounded after an absence of 15 years, perhaps triggered by immigration (Rose and Rowe, 2015). The area occupied by such a stock is not necessarily proportional to its total abundance, as it takes time to recolonize local areas with depleted spawning components. In addition, distributional shifts in marine fishes may also reflect the influence of large scale climatic or oceanographic processes like changing temperature (Brander et al., 2003; Landa et al., 2014). For example, nearly two-thirds of both exploited and nonexploited fish species in the North Sea have shifted in mean latitude or depth, possibly towards cooler temperatures in response to increases in sea temperature from 1977 to 2001 (Perry et al., 2005). However, if the temperatures experienced by a population changes uniformly with the overall temperature within its distribution range, shifts in the distribution may not be a result of changing temperature (Swain and Benoît, 2006). Atlantic cod (Gadus morhua) in the Gulf of Maine (GOM) is an iconic marine fishery stock that has experienced a substantial southwestward distributional shift since the mid-1990s (Palmer, 2014; Richardson et al., 2014). The species was widely distributed across the GOM before the 1920s, whereas high abundance and spawning activities are now more frequently observed in the western region (Rich, 1929; Ames, 2004; Zemeckis et al., 2014b). Reductions in landings and fishing effort based on classical stock assessments have not prevented the stock decline (NEFSC, 2013, 2015; Palmer, 2014). Similarly, the implementation of inshore GOM fishery closures since the late 1990s served to reduce fishing mortality on seasonal aggregations of cod, but have not successfully promoted rebuilding or prevented shifts in the distribution of the stock and fishing activity (Armstrong et al., 2013; Richardson et al., 2014). These facts call for a careful examination of the spatial dynamics of GOM cod, which may reflect the stock’s variations in population structure and responses to overfishing, climate change, or prey availability. In ecology, cod in the GOM are generally found in waters below 10 °C, but can be present in warmer temperatures up to 13.3 °C during the fall season (Helser and Brodziak, 1996; Zemeckis et al., 2017). Juvenile cod mostly appear in depths of 25–75 m, whereas adults prefer depths between 40 and 130 m, with deeper waters as overwintering habitats (Fahay et al., 1999; Zemeckis et al., 2017). A tagging study in the northern Gulf of St Lawrence has indicated that Atlantic cod move around in schools, rather than individually (Tamdrari et al., 2012). At larger scales, they form local aggregations during the spawning seasons (i.e. April to July in spring and November to February in winter) and exhibit strong spawning site fidelity (Howell et al., 2008; Siceloff and Howell, 2013; Zemeckis et al., 2014a, b). Common metrics for evaluating changes in species’ distribution include the centre of distribution [i.e. centre of gravity, (COG)], spatial extent in latitude and longitude, area occupied and habitat associations (i.e. cumulative density distributions in various environmental variables) (e.g. Swain and Wade, 1993; Perry et al., 2005; Swain and Benoît, 2006). Conventional approaches directly use survey data to estimate these metrics, ignoring potential changes in the spatial distribution of sampling effort (Swain and Benoît, 2006; Pinsky et al., 2013). Recently, Thorson et al. (2016) proposed an alternative species distribution function (SDF) approach, which calculates COG, stock boundaries and area occupied from the predicted species distribution or density function. The SDF approach generally provides unbiased estimates of the distribution metrics and standard errors of these estimates, regardless of spatial-varying sampling efforts. In this study, the SDF approach was utilized to calculate the time-varying COG, stock boundaries, area occupied, and median habitat temperature of GOM cod, in order to quantify the distributional shift from 1982 to 2013. Furthermore, the time series for these distributional metrics were used to evaluate two mechanisms potentially driving the distributional shift of the stock: (i) density-dependent habitat selection and (ii) direct effects of temperature change. We hypothesize that these two mechanisms did not play a major role on the distributional shift of the GOM cod stock. Material and methods Data compilation The area of this study covers the offshore strata (>30 m in depth) in the bottom trawl surveys for the GOM cod stock, which have been conducted biannually by the Northeast Fisheries Science Centre (NEFSC), National Oceanic and Atmospheric Administration (NOAA) since the 1960s (Figure 1). The inshore strata were excluded, because they are often occupied by the lobster trap fishery and have not been continuously sampled. Ocean depths in the study area generally range from 30 to 400 m. The NEFSC seasonal surveys follow a stratified random sampling design. For each trawl, the abundance in number and biomass in weight of each species are recorded, along with maximum depth during the gear deployment, bottom temperature, and salinity. Figure 1 Open in new tabDownload slide Cell-based study area: offshore strata in the NEFSC spring and fall bottom trawl survey for Atlantic cod in the GOM. Figure 1 Open in new tabDownload slide Cell-based study area: offshore strata in the NEFSC spring and fall bottom trawl survey for Atlantic cod in the GOM. For this study, a total of 3941 successful tows falling within the offshore GOM strata producing abundance indices by number of individuals and associated geographic coordinates (beginning latitudes and longitudes) were extracted from the NEFSC spring and fall bottom trawl surveys from 1982 to 2013 (Figure 2). Cod were absent from 2235 of the 3941 stations, with 1018 in spring tows and 1127 in fall tows. For the other 1706 stations with cod catches, 974 and 822 of which were in spring and fall surveys, respectively. The abundance indices of cod with calibrations (Miller et al., 2010; NEFSC, 2013) were obtained for these stations which caught cod (M. Palmer, pers. comm.). Figure 2 Open in new tabDownload slide Decadal distributions of calibrated cod abundance indices (number per two) from the NEFSC surveys from 1982 to 2013 in spring and fall. Figure 2 Open in new tabDownload slide Decadal distributions of calibrated cod abundance indices (number per two) from the NEFSC surveys from 1982 to 2013 in spring and fall. We split the study area into a total of 3027 square cells (each cell was 0.05° × 0.05° for an area of 22–23 km2) and compiled information on the bathymetry and bottom thermal field at the cell scale. The primary data source for obtaining bathymetric information was a raster map file of the Northeast Atlantic offshore bathymetry provided by the National Centers for Environmental Information, NOAA (1999). We used ArcMap 10.2.2 (ESRI, 2014), a geospatial processing program, to extract the mean depths for most square cells. For the cells outside the raster map file coverage, the mean depth for each cell was estimated by averaging depth records for trawls that started within the cell from the NEFSC bottom trawl surveys (1982–2013). The bottom thermal field was estimated by season and year based on monthly hindcast data from the Northeast Coastal Ocean Forecast System (NECOFS) (Beardsley et al., 2013; NeCOFS, 2013). The hindcast provides hourly and monthly mean values of oceanographic variables with varying horizontal resolutions ranging from 0.02 to 10 km for the northeast US coastal region (Chen et al., 2006). The accuracy and precision of the NECOFS hindcast have been validated by comparing to measurements recorded during trawl surveys (Guan et al., 2017; Liu et al., 2017). NECOFS stations within each study area cell were searched for and identified as neighbour stations. If no NECOFS station appeared, the nearest station to the centre of the cell would be identified as the neighbour station. Then, bottom temperatures for the neighbour station(s) for four months (i.e. April, May, October, and November) were extracted from the NECOFS monthly hindcast (1982–2013), because the NEFSC bottom trawl surveys were mainly conducted in these months. Finally, bottom temperatures of the neighbour station(s) were averaged for each cell by season and year to estimate a dynamic bottom thermal field for the study area. Geostatisitical modelling to quantify spatial dynamics of cod Geostatistical delta-generalized linear mixed models (GLMMs) (Shelton et al., 2014; Thorson et al., 2015) were used to fit the survey abundance index data and environmental variables by season. The geostatistical model is a two-stage GLMM, which models the probability of nonzero catches and positive abundance indices separately. Unlike conventional two-stage GLMMs, geostatistical models treat the habitat of target species as multiple surfaces and split the habitat into hundreds of small patches according to the spatial distribution of survey abundance index. Each measured habitat variable is considered as a separate surface and included in the model as a fixed effect. All unmeasured habitat variables are combined as another surface. Geostatistical models consider the effects of unmeasured variables as random effects via spatial and spatio-temporal correlations. The basic idea is that population densities at nearby sites are more similar than densities at geographically remote sites. In this study, encounter probabilities p and positive abundance indices λ were modelled as a function of a combination of linear predictors. The linear predictors included average density dt in year t, the associations (βD and βBT) of depth and bottom temperature with p or λ, spatially correlated variability ωs in p or λ at a patch s that is persistent over years, spatially correlated variability εs,t at a patch s in year t. The following model was used to approximate encounter probability pi for tow i at patch s(i), with a binomial error distribution and a logit link function: pi=logit-1dtip+βDp×Dsi+βBTp×BTsi,t(i)+ωsip+ɛsi,tip,(1) where t(i) is the year of sampling tow i, s(i) represents the patch where tow i is located in the study area, Ds(i) is the average depth at s(i) and BTs(i),t(i) is the average bottom temperature at s(i) in t(i). Similarly, we approximated positive abundance index λi for tow i with a Gamma error distribution and a log link: λi=wi×expdtiλ+βDλ×Dsi+βBTλ×BTsi+ωsiλ+ɛsi,tiλ,(2) where wi represents the area swept for tow i, which has been standardized as 0.038 km2. The models were fitted and validated using the geostatistical model (version geo_index_v3l) developed by Dr James Thorson in the R programme (https://github.com/nwfsc-assess/geostatistical_delta-GLMM). Finally, two geostatistical models were built for predicting seasonal distributions of GOM cod by year from 1982 to 2013, one for spring and the other for fall. Subsequently, we used the models to estimate seasonal encounter probabilities and positive abundance indices for each identified square cell in the study area for each year. Average cod density dj,t at square cell j at year t was estimated as follows: d^j,t=p^s(j),t×λ^s(j),t,(3) Where s(j) denotes the patch that square cell j is located in. The total relative abundance index across the study area at year t was then calculated as: b^t=∑j=1njaj×d^j,t,(4) where nj is the number of square cells identified in the study area, and aj is the area (in km2) of square cell j. Based on Thorson’s SDF approach (Thorson et al., 2016), the predicted distribution of GOM cod was used to calculate season-specific COG, spatial extent in latitude and longitude and area occupied. We estimated latitudinal and longitudinal COG at year t for spring or fall as: Latitudet¯=∑j=1njdj,t×Latitudej∑j=1njdj,t,(5) Longitudet¯=∑j=1njdj,t×Longitudej∑j=1njdj,t,(6) where Latitudej and Longitudej are latitude and longitude for square cell j. To estimate the spatial extent in latitude and longitude, we calculated the cumulative density distributions in latitude and longitude as: qtδ=∑j=1njdj,tILatitudej<δ∑j=1njdj,t or ∑j=1njdj,tILongitudej<δ∑j=1njdj,t,(7) where I(Latitudej < δ) and I(Longitudej< δ) are indicator functions that equal one if Latitudej or Longitudej is less than δ and zero otherwise. In the northeast of the GOM, there is a small area 5Yb defined by the North Atlantic Fisheries Organization. About 75% of Atlantic cod tagged in area 5Yb were recaptured in the Canadian 4X stock in the northeast (Hunt and Neilson, 1993; Clark and Emberley, 2010). Similarly, spawning components in the Great South Channel, Nantucket Shoals, southern New England, and Middle Atlantic that are currently classified into the Georges Bank management unit exhibit more connectivity with spawning components in the GOM (Tallack, 2011; Zemeckis et al., 2014a). In order to decrease the impact of this connectivity on the quantification of GOM cod distribution, we did a sensitivity analysis using different percentiles (5th to 25th percentiles stepped by 5 and 95th to 75th percentiles stepped by 5) in latitude or longitude as the stock extents. The 20th and 80th percentiles were chosen to estimate the stock extents, as temporal trend in each of the four stock extents does not change for different scenarios, but their fluctuations continuously declined as the percentiles used got closer to the 20th or 80th percentiles before stablizing. For area occupied, we estimated a population kernel which approximated the distribution of GOM cod and then calculated the volume of the kernel as area occupied (Woillez et al., 2009; Thorson et al., 2016). The distribution of GOM cod within the thermal field determined the species habitat temperature. The cumulative density distributions in temperature were calculated for estimating the species habitat temperature, qtδ=∑j=1njdj,tIBTj<δ∑j=1njdj,t,(8) where I(BTj < δ) is also an indicator function that equal one if BTj is less than δ and zero otherwise. We identified the 50th percentile of this distribution as the median habitat temperature for GOM cod. Analysis of the metrics for distributional shift The time series of total abundance indices and area occupied were used to test the density-dependent effect on the distributional shift of GOM cod. A linear regression analysis was conducted, with area occupied as the response variable and total abundance indices as the explanatory variable. The slope coefficient and r2 value given by the linear regression were used to evaluate how changes in stock abundance might affect the area occupied. If the slope coefficient was not significantly different from 0, changes in the total abundance indices would have negligible influences on area occupied. R-squared values indicated the percentages of variance in area occupied explained by the variance in total abundance indices. The seasonal thermal fields for the study area from 1982 to 2013 were quantitatively analysed and compared with the time series of season-specific median habitat temperatures occupied by GOM cod to investigate the relationship between warming temperature and distributional shifts of GOM cod. The 5th, 50th, and 95th percentiles of bottom temperatures for all identified square cells were calculated separately for the spring and fall for each year. Temporal trends in the median temperature of the overall study area were estimated using the slope coefficient given by a linear regression model with year as the explanatory variable (X) and the 50th percentiles of the environmental temperatures by season as the response variable (Y). Similarly, we performed linear regression analysis to estimate temporal trends in the median habitat temperatures for GOM cod. If the median habitat temperature for GOM cod follows the median temperature in the study area, this stock possibly has a wide tolerance of temperatures and thereby the hypothesis would be supported that shifts in the stock distribution were not directly attributed to changing temperatures. On the contrary, if the median habitat temperature for GOM cod does not change with the environment, a critical comparison needs between the seasonal distribution of cod and the fine scale thermal field in the study area for rejecting the hypothesis. Thus, for each season, mean bottom temperatures were calculated for each cell in the study area and temporal trends at the fine scale thermal field were estimated with a linear regression model. Results Predicted relative density distributions for Atlantic cod in the GOM were displayed by season and year in Supplementary Figures S1 and 2. The season-specific total abundance indices estimated from these predicted relative densities showed similar trends with the stratified mean estimates used for the recent 2014 GOM cod stock assessment (Figure 3). The two kinds of estimates fluctuated concurrently, but stratified estimates were more variable than that in the geostatistical estimates (Figure 3). Moreover, the stratified estimates of total abundance index had generally larger standard errors than the geostatistical estimates. Extremely large values and uncertainties appeared in the stratified estimates for the springs of 2007 and 2008, and the falls of 1982, 1988, and 2002, which was not the case for the geostatistical estimates. Figure 3 Open in new tabDownload slide Season-specific estimates of total relative abundance from the geostatistical model developed by this study and the stratified model currently used in the stock assessment of GOM cod. Figure 3 Open in new tabDownload slide Season-specific estimates of total relative abundance from the geostatistical model developed by this study and the stratified model currently used in the stock assessment of GOM cod. Figure 4 shows the centres of gravity for each of the three periods (i.e. 1982–1991, 1992–2001, and 2002–2013) by season. The COG for cod distribution in latitude and longitude indicated clear southwestward shifts in spring and fall. Standard deviation in the COG for each time period increased in spring, but often decreased in fall. This result tended to be robust with respect to choices of data grouping of time periods. Figure 4 Open in new tabDownload slide Southwestward shifts in the latitudinal and longitudinal centers of gravity for the spatial distribution of Atlantic cod in the GOM by season, year and time period. The squares show standard deviations in the COG three time periods: 1982–1991 (red and solid square), 1992–2001 (green and dash square) and 2001–2013 (blue and dot square). Figure 4 Open in new tabDownload slide Southwestward shifts in the latitudinal and longitudinal centers of gravity for the spatial distribution of Atlantic cod in the GOM by season, year and time period. The squares show standard deviations in the COG three time periods: 1982–1991 (red and solid square), 1992–2001 (green and dash square) and 2001–2013 (blue and dot square). Temporal trends in the northern and eastern extents display by season separately (Figure 5a). In general, the northern extent of GOM cod showed a southern shift to low latitudes. The eastern extent moved westward by >1° in longitude for three times after the 21st century. The southern and western spatial extents almost remained in the same latitudes/longitudes during the period of 1982–2013. Meanwhile, the size of the area occupied by GOM cod showed a decreasing trend since 1982, with increasing fluctuations after 2000 and a slightly increasing trend since 2009 (Figure 5b). Figure 5 Open in new tabDownload slide Temporal trends in in the (a) northern and eastern extents, and (b) area occupied by GOM Atlantic cod in spring and fall. Figure 5 Open in new tabDownload slide Temporal trends in in the (a) northern and eastern extents, and (b) area occupied by GOM Atlantic cod in spring and fall. The slope coefficient and r2-value given by the linear regression between total abundance indices and areas occupied seem to refute density-dependent effects on the distribution of GOM cod. The slope coefficient was estimated as −0.019 in spring and −0.062 in fall, which did not significantly differ from 0. Very small r2 values (0.001 in spring and 0.015 in fall) indicated that variability in the area occupied could hardly be explained by variability in the total abundance indices. Furthermore, temporal trends in the median habitat temperatures of GOM cod and the median bottom temperature of the study area are displayed by season in Supplementary Figure S3. In spring, the 50th percentile of the environmental temperatures increased at a rate of 0.029 °C per year, whereas the species’ median habitat temperature seemed to follow the lower limit (5th percentile) of the environmental temperatures, displaying a nonsignificant decreasing trend. This indicates that cod tend to inhabit lower temperatures (<6 °C) than available to them in the GOM during the spring. As a contrast, the median habitat temperature of GOM cod significantly increased at an estimated rate of 0.032 °C per year in fall, with p-value = 0.033. Similar is the rate of increase in the 50th percentile of the environmental temperatures in fall, with an estimate of 0.027 and p-value = 0.054. As for fine-scale spatio-temporal variations in the thermal field of the study area, the eastern GOM generally had warmer bottom temperatures than western areas in both the spring and fall (Figure 6a and b). The GOM has been warming since 1982 and the increases in bottom temperature exhibited substantial spatial heterogeneity (Figure 6c and d). In the spring of 1982–2013, nearshore areas along the GOM did not show a significant increase in bottom temperature, whereas eastern GOM between 66.5°W and 68°W was getting warmer at a rate of 0.032–0.044 °C per year (Figure 6c). In contrast, the eastern GOM did not get significantly warmer in fall. Fall bottom temperatures in the western nearshore areas warmed at a rate of 0.033–0.053 °C per year (Figure 6d). Figure 6 Open in new tabDownload slide Grid-based distribution of (a) mean bottom temperatures (BT) in spring, (b) mean BT in fall, (c) increasing rates of BT in spring and (d) increasing rates of BT in fall from 1982 to 2013 in offshore GOM areas. Figure 6 Open in new tabDownload slide Grid-based distribution of (a) mean bottom temperatures (BT) in spring, (b) mean BT in fall, (c) increasing rates of BT in spring and (d) increasing rates of BT in fall from 1982 to 2013 in offshore GOM areas. Discussion This study provided model-based estimates of total abundance indices for GOM cod stock, which showed similar trends to the stratified mean estimates used in the stock assessment (Palmer, 2014). These model-based estimates seem more accurate and precise, as the season-specific geostatistical models explain more of the stock’s spatial dynamics and the stratified mean estimates rely on an assumption of homogeneous distribution of the fish within each survey stratum (NEFC, 1988; Thorson et al., 2015). In addition, the stratified mean estimates show sudden and sharp increases with large uncertainties for some years, especially 2007 and 2008, because some survey stations have especially large catches. The sharp increases almost double the abundance index estimates of previous years, a result that is biologically unrealistic, leading us to believe that the stratified mean estimates tend to be less reliable in those years. This is a major reason that leads to the huge shift from nonoverfished to overfished in perception of stock status between the 2008 and 2011 benchmark stock assessments. In this context, considering the spatial dynamics of GOM cod should be able to improve the stock assessment. Noticeable decreases in the area occupied of the GOM cod stock have occurred with considerable shifts to the south or west in the northern and eastern extents and the COG since the late 1990s. These changes in the spatial dynamics of GOM cod indicate an overall contraction of the stock to the southwest over the last two decades, which is supported by the annual Gini indices (i.e. concentration indices) and decadal distributions for GOM cod from the NEFSC surveys (NEFSC, 2013; Palmer, 2014). The stock as a complex metapopulation consists of several subpopulations that are made up of multiple spawning components (Zemeckis et al., 2014b). Almost half of the historical spawning components for the 1920s were critically depleted by the late 1940s (Ames, 2004). Recently, a GOM-wide tagging study for Atlantic cod found very few tag returns in the northern coastal area, suggesting a continued localized depletion (Tallack, 2007). This study revealed a rapid contraction of GOM cod to the southwest in the 2000s, corroborating the tagging study results. There may be various contributors to the recent local depletion after the late 1990s. First, the lipid-rich prey for GOM cod sharply declined in the same period, as the closure of fishways on the St. Croix River in 1995 triggered the collapse of Maine’s largest remaining alewife population, while an industrial-scale fishery for Atlantic herring was rapidly developed in coastal New England (Willis et al., 2006; Ames and Lichter, 2013). Secondly, global warming may have led to reduced recruitment and increased mortality in the stock, whereas the management has failed to recognize these impacts, contributing to overfishing especially for less productive subpopulations (Pershing et al., 2015). What’s more, the GOM cod SSB has been below half of the maximum sustainable SSB proxy since 1982 (Palmer, 2014). Under this condition, other factors could be easily contributing to the continued decline in abundance, like overfishing resultant from retrospective patterns in the stock assessment (NEFSC, 2015) and Allee effect at low biomass (Keith and Hutchings, 2012). Yet, it is possible that fishing has not directly contributed to the local depletion, because the catch-weighted centre of fishing activity contracted to the southwest and not much fishing remained in the eastern GOM after the mid-1990s (Palmer, 2014). An interesting result shown in this study is that variance in the COG of GOM cod distribution increased in the spring, but displayed a sharp decline from the 1982–1991 to the 1992–2001 time block and a modest increase from the 1992–2001 to the 2002–2013 time block in fall. This may imply shifts in the productivity of two major subpopulations in western GOM: northern spring coastal spawners and southern spawners that spawn in winter and early spring (Zemeckis et al., 2014b). The northern spawners form large spawning aggregations specifically in Bigelow Bight, Ipswich Bay, and Massachusetts Bay during spring (April to July), whereas the southern spawners start their prespawning concentrations in Ipswich Bay, Massachusetts Bay, and on offshore Jeffreys Ledge, Stellwagen Bank at the end of Fall (November) (Howell et al., 2008; Zemeckis et al., 2014b). In this context, widely spreading COG in spring and relatively concentrating COG in fall after 2002 might suggest a continuous decline in the northern spring spawners and a relatively increased productivity in the southern spawning complex. To test this suggestion, future studies like genetic mixed-stock analysis (Wang et al., 2015) are needed to evaluate temporal variations in the contribution of different subpopulations to the GOM cod stock. No significant relationship was found between the total relative abundance index and the area occupied of GOM cod, which is consistent with the study by Nye et al. (2009). They examined the relationship between area occupied and survey abundance for each of 36 fish stocks on the northeast US continental shelf. Results showed no strong relationship for some stocks including GOM cod, though stock size was more often correlated with the area occupied by each species. Therefore, density-dependent effects are likely minimal on the distributional shift of GOM cod. In response to climate change, assemblages of fish species on the northeast US continental shelf have exhibited various patterns in individual distribution shifts since the mid-20th century (Nye et al., 2009; Kleisner et al., 2016). Fish stocks do not have a coherent response to changing temperature because of differences in life history traits and geographical affinities (Brander et al., 2003; Tu et al., 2015). The GOM is a semi-closed basin on the Northwest Atlantic continental shelf and fed by two major sources of water masses. Scotian Shelf Water from the Nova Scotian shelf enters the gulf at the surface, and relatively warmer and more saline Slope Water enters at depth and along the bottom through the Northeast Channel (Townsend et al., 2015). The proportions of these water masses determine the oceanography of the GOM region, which can be affected by multiple climatic phenomena (Xu et al., 2015; Saba et al., 2016). Many mid-trophic level species in the GOM have shifted west-southwest and to shallower waters in response to climate changes, possibly for cooler temperatures (Kleisner et al., 2016). Alewives and herrings, as major lipid-rich prey for GOM cod (Ames and Lichter, 2013), are among these species. Availability of such lipid-rich prey could largely influence growth, condition and reproductive capacity of Atlantic cod (Rose and O’Driscoll, 2002). It seems unlikely that GOM cod have shifted their distribution to follow alewives and herrings, as their migration patterns have remained quite constant across decades (Ames, 2004; Tallack, 2007). On the contrary, shifts in the stock distribution might result from variations in the productivity of different subpopulations that caused by spatio-temporal changes in the availability of lipid-rich prey. Additionally, the median habitat temperature for cod appeared to increase at a similar speed with the median temperature in the study area during the fall, which suggests that changing temperature may not directly contribute to recent distribution shifts in GOM cod. This point is supported by the species’ capability of tolerating a wide range of temperatures (Righton et al., 2010), but seems to contradict the result in spring that GOM cod appeared to inhabit much lower temperatures of their preference than available to them. This inconsistency has triggered a deep investigation into the GOM cod distribution and fine scale thermal field in the study area in spring. During the last three decades, the eastern GOM had generally warmer bottom temperatures than southwestern areas. Moreover, spring bottom temperatures in the northeast were warming at high rates (Mills et al., 2013). This might have prevented stocks in the GOM from shifting to the north, but hardly lead to the southwestward shifts after the late 1990s. Atlantic cod mostly migrated to nearshore colder areas along the GOM coastline for feeding or spawning in spring (Howell et al., 2008; Ames and Lichter, 2013), with a consequence of low median habitat temperatures close to lower limit of the environmental temperatures. This is to say, warming bottom temperatures in northeastern GOM would not impede cod migrations to nearshore areas in the north or force the stock to shift west-southwest. Thus, it is believed that warming temperature has not directly driven the rapid changes in the spatial dynamics of the GOM cod stock. In conclusion, this study quantitatively analysed the distributional shift of GOM cod and testified that the rapid changes in the spatial dynamics after the late 1990s was not a simply result of decreasing stock abundance or climate change. The observed shift in cod distribution possibly resulted from other factors, like fluctuating productivity among subpopulations, which needs further investigation. Climate changes are not heterogeneous in space and may have quite varying influences on the productivity of various subpopulations via affecting the distribution of their important prey species. More studies on quantitatively analysing distributional shifts of commercially or ecologically important fish stocks should be expected in the future. For most stocks, we do not understand their spatial structure as well as GOM cod, and cannot improve our understanding without massive efforts from tagging, genetics, morphology, or related stock identification studies. However, there are ample fisheries-independent survey abundance index data for many species and related environmental information, which can be used for hindcasting their density distribution through statistical modelling. Quantitative analysis of the predicted density distribution function may help determine whether a given stock has complex structure, detect potential local depletion, and evaluate the influence of climate change on the stock distribution. Supplementary data Supplementary material is available at the ICESJMS online version of the article. Acknowledgements We thank James Thorson for sharing his Geostatistical models’ codes and Robert Johnston, Paul Kostovick, Nancy McHugh, and Michael Palmer for providing NEFSC bottom trawl surveys data. Also, we would like to thank the Marine Ecosystem Dynamics Modeling Laboratory, School for Marine Science and Technology, University of Massachusetts-Dartmouth (MEDML/SMAST/UMASSD) for providing the NECOFS hindcast fields dataset. Finally, sincere thanks to China Scholarship Council (CSC) and University of Maine (UMaine) Graduate School for supporting Lisha Guan’s graduate study at UMaine. References Ames E. P. 2004 . Atlantic Cod Stock Structure in the Gulf of Maine . Fisheries , 29 : 10 – 28 . Google Scholar Crossref Search ADS WorldCat Ames E. P. , Lichter J. 2013 . Gadids and Alewives: Structure within complexity in the Gulf of Maine . Fisheries Research , 141 : 70 – 78 . Google Scholar Crossref Search ADS WorldCat Armstrong M. P. , Dean M. J., Hoffman W. S., Zemeckis D. R., Nies T. A., Pierce D. E., Diodati P. J. et al. 2013 . The application of small scale fishery closures to protect Atlantic cod spawning aggregations in the inshore Gulf of Maine . Fisheries Research , 141 : 62 – 69 . Google Scholar Crossref Search ADS WorldCat Beardsley R. C. , Chen C., Xu Q. 2013 . Coastal flooding in Scituate (MA): a FVCOM study of the 27 December 2010 nor’easter . Journal of Geophysical Research: Oceans , 118 : 6030 – 6045 . Google Scholar Crossref Search ADS WorldCat Brander K. M. , Blom M. F., Borges M. 2003 . Changes in fish distribution in the eastern North Atlantic: Are we seeing a coherent response to changing temperature? ICES Marine Science Symposia , 219 : 261 – 270 . OpenURL Placeholder Text WorldCat Chen C. , Beardsley R. C., Cowles G. 2006 . An unstructured-grid finite-volume coastal ocean model (FVCOM) System . Oceanography , 19 : 78 – 89 . Google Scholar Crossref Search ADS WorldCat Clark D. S. , Emberley J. 2010 . Assessment of Cod in Division 4X in 2008. DFO Can. Sci. Advis. Sec. Res. Doc. 2009/018. vi + 101 pp. Environmental Systems Research Institute (ESRI) . 2014 . ArcGIS Release 10.2.2. Redlands, CA. Fahay M. P. , Berrien P. L., Johnson D. L., Morse W. W. 1999 . Essential fish habitat source document: Atlantic cod, Gadus morhua, life history and habitat characteristics. US NOAA Technical Memorandum NMFS-NE-124: 41 pp. Guan L. , Chen Y., Wilson J. A. 2017 . Evaluating spatio-temporal variability in the habitat quality of Atlantic cod (Gadus morhua) in the Gulf of Maine . Fisheries Oceanography , 26 : 83 – 96 . Google Scholar Crossref Search ADS WorldCat Helser T. E. , Brodziak J. K. 1996 . Influence of temperature and depth on distribution and catches of yellowtail flounder, Atlantic cod, and haddock in NEFSC bottom trawl surveys. NEFSC Ref. Doc. 96-05e, 20 pp. Hilborn R. , Quinn T. P., Schindler D. E., Rogers D. E. 2003 . Biocomplexity and fisheries sustainability . Proceedings of the National Academy of Sciences of the United States of America , 100 : 6564 – 6568 . Google Scholar Crossref Search ADS PubMed WorldCat Hilborn R. , Walters C. J. 1992 . Quantitative Fisheries Stock Assessment: Choice, Dynamics and Uncertainty . Chapman and Hall , New York . 570 pp. Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Howell W. H. , Morin M., Rennels N., Goethel D. 2008 . Residency of adult Atlantic cod (Gadus morhua) in the western Gulf of Maine . Fisheries Research , 91 : 123 – 132 . Google Scholar Crossref Search ADS WorldCat Hunt J. J. , Neilson J. D. 1993 . Is there a separate stock of Atlantic cod in the western side of the Bay of Fundy? . North American Journal of Fisheries Management , 13 : 421 – 436 . Google Scholar Crossref Search ADS WorldCat Keith D. M. , Hutchings J. A. 2012 . Population dynamics of marine fishes at low abundance . Canadian Journal of Fisheries and Aquatic Sciences , 69 : 1150 – 1163 . Google Scholar Crossref Search ADS WorldCat Kerr L. A. , Cadrin S. X., Kovach A. I. 2014 . Consequences of a mismatch between biological and management units on our perception of Atlantic cod off New England . ICES Journal of Marine Science , 71 : 1366 – 1381 . Google Scholar Crossref Search ADS WorldCat Kleisner K. M. , Fogarty M. J., McGee S., Barnett A., Fratantoni P., Greene J., Hare J. A., et al. 2016 . The effects of sub-regional climate velocity on the distribution and spatial extent of marine species assemblages . PLoS One , 11 : 1 – 21 . Google Scholar Crossref Search ADS WorldCat Kritzer J. P. , Sale P. F. 2004 . Metapopulation ecology in the sea:from Levins’ model to marine ecology and fisheries science . Fish and Fisheries , 5 : 131 – 140 . Google Scholar Crossref Search ADS WorldCat Landa C. S. , Ottersen G., Sundby S., Dingsør G. E., Stiansen J. E. 2014 . Recruitment, distribution boundary and habitat temperature of an arcto-boreal gadoid in a climatically changing environment: a case study on Northeast Arctic haddock (Melanogrammus aeglefinus) . Fisheries Oceanography , 23 : 506 – 520 . Google Scholar Crossref Search ADS WorldCat Liu C. , Cowles G. W., Zemeckis D. R., Cadrin S. X., Dean M. J. 2017 . Validation of a hidden Markov model for the geolocation of Atlantic cod . Canadian Journal of Fisheries and Aquatic Sciences , doi: 10.1139/cjfas-2016-0376. OpenURL Placeholder Text WorldCat MacCall A. D. 1990 . Dynamic Geography of Marine Fish Populations . Washington Sea Grant Program, Seattle , Washington . 153 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Miller T. J. , Das C., Politis P. J., Miller A. S., Lucey S. M., Legault C. M., Brown R. W., et al. 2010 . Estimation of Albatross IV to Henry B. Bigelow calibration factors. Northeast Fisheries Science Center Reference Document, 10–5: 233 pp. Mills K. E. , Pershing A. J., Brown C. J., Chen Y., Chiang F.-S., Holland D. S., Lehuta S., et al. 2013 . Fisheries management in a changing climate lessons from the 2012 ocean heat wave in the Northwest Atlantic . Oceanography , 26 : 191 – 195 . Google Scholar Crossref Search ADS WorldCat NeCOFS . 2013 . Northeast Coastal Ocean Forecasting System (NeCOFS) Main Portal. http://fvcom.smast.umassd.edu/research_projects/NECOFS. NEFC . 1988 . An evaluation of the bottom trawl survey program of the Northeast Fisheries Center. US NOAA Technical Memorandum NMFS-F/NEC-52: 94 pp. NEFSC . 2013 . 55th Northeast Regional Stock Assessment Workshop (55th SAW) Assessment Report. US Department Commerce, Northeast Fisheries Science Center Reference Document 13-11. Woods Hole, Massachusetts. 434 pp. NEFSC . 2015 . Operational Assessment of 20 Northeast Groundfish Stocks, Updated Through 2014. US Dept Commer, Northeast Fish Sci Cent Ref Doc. 15-24; 251 pp. Available from: National Marine Fisheries Service, 166 Water Street, Woods Hole, MA 02543-1026, or online at http://www.nefsc.noaa.gov/publications/. NOAA National Centers for Environmental Information. 1999 . U.S. Coastal Relief Model. Available at https://www.ngdc.noaa.gov/mgg/coastal/crm.html (last accessed 1 April 2014). Nye J. A. , Link J. S., Hare J. A., Overholtz W. J. 2009 . Changing spatial distribution of fish stocks in relation to climate and population size on the Northeast United States continental shelf . Marine Ecology Progress Series , 393 : 111 – 129 . Google Scholar Crossref Search ADS WorldCat Palmer M. C. 2014 . 2014 Assessment Update Report of the Gulf of Maine Atlantic Cod Stock. US Dept Commer, Northeast Fish Sci Cent Ref Doc. 14-14; 119 pp. doi: 10.7289/V5V9862C. Perry A. L. , Low P. J., Ellis J. R., Reynolds J. D. 2005 . Climate Change and Distribution Shifts in Marine Fishes . Science , 308 : 1912 – 1915 . Google Scholar Crossref Search ADS PubMed WorldCat Pershing A. J. , Alexander M. A., Christina M., Kerr L. A., Bris A. L., Mills K. E., Nye J. A., et al. 2015 . Slow adaptation in the face of rapid warming leads to collapse of the Gulf of Maine cod fishery . Science , 350 : 809 – 812 . Google Scholar Crossref Search ADS PubMed WorldCat Pinsky M. L. , Worm B., Fogarty M. J., Sarmiento J. L., Levin S. A. 2013 . Marine taxa track local climate velocities . Science , 341 : 1239 – 1242 . Google Scholar Crossref Search ADS PubMed WorldCat Rich W. H. 1929 . Fishing Grounds of the Gulf of Maine . U.S. Commissioner of Fisheries , Washington, D.C . 152 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Richardson D. , Palmer M., Smith B. 2014 . The influence of forage fish abundance on the aggregation of Gulf of Maine Atlantic cod (Gadus morhua) and their catchability in the fishery . Canadian Journal of Fisheries and Aquatic Sciences , 71 : 1349 – 1362 . Google Scholar Crossref Search ADS WorldCat Righton D. A. , Andersen K. H., Neat F., Thorsteinsson V., Steingrund P., Svedäng H., Michalsen K., Hinrichsen H. H., et al. 2010 . Thermal niche of Atlantic cod Gadus morhua: limits, tolerance and optima . Marine Ecology Progress Series , 420 : 1 – 13 . Google Scholar Crossref Search ADS WorldCat Rose G. A. , Rowe S. 2015 . Northern cod comeback . Canadian Journal of Fisheries and Aquatic Sciences , 72 : 1789 – 1798 . Google Scholar Crossref Search ADS WorldCat Rose G. , O’Driscoll R. L. 2002 . Capelin are good for cod: can the northern stock rebuild without them? . ICES Journal of Marine Science , 59 : 1018 – 1026 . Google Scholar Crossref Search ADS WorldCat Saba V. S. , Griffies S. M., Anderson W. G., Winton M., Alexander M. A., Delworth T. L., Hare J. A., et al. 2016 . Enhanced warming of the Northwest Atlantic Ocean under climate change . Journal of Geophysical Research: Oceans , 121 : 118 – 132 . Google Scholar Crossref Search ADS WorldCat Shelton A. O. , Thorson J. T., Ward E. J., Feist B. E. 2014 . Spatial semiparametric models improve estimates of species abundance and distribution . Canadian Journal of Fisheries and Aquatic Sciences , 71 : 1655 – 1666 . Google Scholar Crossref Search ADS WorldCat Siceloff L. , Howell W. H. 2013 . Fine-scale temporal and spatial distributions of Atlantic cod (Gadus morhua) on a western Gulf of maine spawning ground . Fisheries Research , 141 : 31 – 43 . Google Scholar Crossref Search ADS WorldCat Spencer P. D. 2008 . Density-independent and density-dependent factors affecting temporal changes in spatial distributions of eastern Bering Sea flatfish . Fisheries Oceanography , 17 : 396 – 410 . Google Scholar Crossref Search ADS WorldCat Swain D. P. , Benoît H. P. 2006 . Change in habitat associations and geographic distribution of thorny skate (Amblyraja radiata) in the southern Gulf of St Lawrence: Density-dependent habitat selection or response to environmental change? . Fisheries Oceanography , 15 : 166 – 182 . Google Scholar Crossref Search ADS WorldCat Swain D. , Wade E. 1993 . Density-dependent geographic distribution of Atlantic cod (Gadus morhua) in the southern Gulf of St. Lawrence . Canadian Journal of Fisheries and Aquatic Sciences , 50 : 725 – 733 . Google Scholar Crossref Search ADS WorldCat Tallack S. 2007 . A description of tagging data from the Northeast Regional Cod Tagging Program (WP3A) and preliminary applications of weighting and mixing analysis (WP3C). Draft report submitted to the National Marine Fisheries Service, Northeast Fisheries Science Center. Woods Hole, Massachusetts. 60 pp. Tallack S. M. L. 2011 . Stock identification applications of conventional tagging data for Atlantic cod in the Gulf of Maine. In Proceedings from the 2nd International Symposium on Advances in Fish Tagging and Marking Techniques, pp. 1–15. Ed. by J. McKenzie, B. Parsons, A. C. Seitz, R. K. Kopf, M. Mesa, and Q. Phelps. American Fisheries Society, Auckland, NZ. Tamdrari H. , Brêthes J. C., Castonguay M., Duplisea D. E. 2012 . Homing and group cohesion in Atlantic cod Gadus morhua revealed by tagging experiments . Journal of Fish Biology , 81 : 714 – 727 . Google Scholar Crossref Search ADS PubMed WorldCat Thorson J. T. , Pinsky M. L., Ward E. J. 2016 . Model-based inference for estimating shifts in species distribution, area occupied and centre of gravity . Methods in Ecology and Evolution , 7 : 990 – 1002 . Google Scholar Crossref Search ADS WorldCat Thorson J. T. , Shelton A. O., Ward E. J., Skaug H. J. 2015 . Geostatistical delta-generalized linear mixed models improve precision for estimated abundance indices for West Coast groundfishes . ICES Journal of Marine Science , 72 : 1297 – 1310 . Google Scholar Crossref Search ADS WorldCat Townsend D. W. , Pettigrew N. R., Thomas M. A., Neary M. G., Mcgillicuddy D. J., Donnell J. O. 2015 . Water masses and nutrient fluxes to the Gulf of Maine . Journal of Marine Research , 73 : 93 – 122 . Google Scholar Crossref Search ADS PubMed WorldCat Tu C.-Y. , Tian Y., Hsieh C.-H. 2015 . Effects of climate on temporal variation in the abundance and distribution of the demersal fish assemblage in the Tsushima Warm Current region of the Japan Sea . Fisheries Oceanography , 24 : 177 – 189 . Google Scholar Crossref Search ADS WorldCat Wang L. , Liu S., Zhuang Z., Lin H., Meng Z. 2015 . Mixed-stock analysis of small yellow croaker Larimichthys polyactis providing implications for stock conservation and management . Fisheries Research , 161 : 86 – 92 . Google Scholar Crossref Search ADS WorldCat Willis T. V. , Bentzen P., Paterson I. G. 2006 . Two Reports on Alewives in the St. Croix River. Hallowell, Maine. 68 pp. https://www.fws.gov/GOMCP/pdfs/MaineRiversStCroixReportFinal.pdf. Woillez M. , Rivoirard J., Petitgas P. 2009 . Notes on survey-based spatial indicators for monitoring fish populations . Aquatic Living Resources , 22 : 155 – 164 . Google Scholar Crossref Search ADS WorldCat Xu H. , Kim H.-M., Nye J. A., Hameed S. 2015 . Impacts of the North Atlantic Oscillation on sea surface temperature on the Northeast US Continental Shelf . Continental Shelf Research , 105 : 60 – 66 . Google Scholar Crossref Search ADS WorldCat Ying Y. , Chen Y., Lin L., Gao T., Quinn T. 2011 . Risks of ignoring fish population spatial structure in fisheries management . Canadian Journal of Fisheries and Aquatic Sciences , 68 : 2101 – 2120 . Google Scholar Crossref Search ADS WorldCat Zemeckis D. R. , Hoffman W. S., Dean M. J., Armstrong M. P., Cadrin S. X. 2014a . Spawning site fidelity by Atlantic cod (Gadus morhua) in the Gulf of Maine: implications for population structure and rebuilding . ICES Journal of Marine Science , 71 : 1356 – 1365 . Google Scholar Crossref Search ADS WorldCat Zemeckis D. R. , Liu C., Cowles G. W., Dean M. J., Hoffman W. S., Martins D., Cadrin S. X. 2017 . Seasonal movements and connectivity of an Atlantic cod (Gadus morhua) spawning component in the western Gulf of Maine . ICES Journal of Marine Science , 74 : 1780 – 1796 . Google Scholar Crossref Search ADS WorldCat Zemeckis D. R. , Martins D., Kerr L. A., Cadrin S. X. 2014b . Stock identification of Atlantic cod (Gadus morhua) in US waters: an interdisciplinary approach . ICES Journal of Marine Science , 71 : 1490 – 1506 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Irgens,, Christian;Kjesbu, Olav, S;Folkvord,, Arild

doi: 10.1093/icesjms/fsx088pmid: N/A

Abstract This study documents how settlement of juvenile Atlantic cod (Gadus morhua) in the Barents Sea affects otolith growth and morphology. A simple method to objectively discriminate between age 0 and age 1 cod sampled in late summer was demonstrated by using only two otolith morphometric descriptors: area and perimeter. In the pre-settled 0-group cod, otolith lobe formation clearly increased with fish size, resulting in high otolith crenulation. This trend was disrupted during settlement, resulting in noticeably less crenulated otoliths of the settled 1-group cod sampled in winter. Combined observations of otolith shape, fish size, and body condition suggest that environmental factors associated with settlement during autumn, particularly reduced food intake, directly affect lobe formation leading to less crenulated otoliths. Comparably reduced body condition and otolith crenulation of 0-group cod in bottom trawls (vs. pelagic trawls) may indicate early settlement or vertical exploratory behaviour in the Barents Sea Ecosystem Survey (in August–September) and, thus, an underrepresentation of 0-group cod from pelagic trawling. Introduction The Barents Sea (BS) Atlantic cod (Gadus morhua) is one of the most important commercial fish stocks in the North Atlantic and currently represents the largest stock of Atlantic cod (Ottersen et al., 2014). Large fluctuations in stock size and recruitment (Kjesbu et al., 2014) highlight the importance of reliable abundance estimates at an early stage, i.e. pelagic post-larval cod (hereafter called 0-group cod) (Sundby et al., 1989). BS cod 0-group indices provide an early indication of the level of recruitment (Dragesund and Olsen, 1965; Dragesund et al., 2008) and are, therefore, used in recruitment variability studies (Ottersen and Loeng, 2000) and considered in recruitment models (ICES, 2016). Knowledge of ontogeny, behaviour, and ecology of 0-group cod, combined with environmental factors influencing growth and survival, constitute, therefore, highly important input for proper stock management as well as for assessing energy transfer through the Barents Sea ecosystem (Eriksen et al., 2011), where cod is an important apex predator (Steneck, 2012). A critical event for 0-group cod, like for many other marine demersal fish species, is when juveniles settle to the bottom where they must compete for food and suitable habitats (Hüssy et al., 2003a). Unlike the Norwegian coastal (NC) (Salvanes et al., 1994) or Baltic (Hüssy et al., 2003a) cod stocks that settle in relatively shallow nearshore benthic habitats, the BS cod juveniles descend from relatively warm surface waters to the dark and cold deepwater bottom (Ottersen et al., 2002) with vastly different diet composition and prey accessibility (Ponomarenko, 1965; Bergstad et al., 1987; Dalpadado and Bogstad, 2004). The 0-group BS cod have been reported to settle in September–October (Dragesund and Olsen, 1965; Bergstad et al., 1987) or even closer to winter (Sundby et al., 1989; Hylen et al., 2000). However, since growth and ontogeny influence the timing of settlement, spatiotemporal settling dynamics are likely to vary between years (Salvanes et al., 1994; Bastrikin et al., 2014). These variations may influence growth and survival, but also challenge the survey design for monitoring the pelagic BS 0-group cod (Dingsør, 2005). Several studies of cod populations collectively show that neither timing nor size at settlement can be generalized (NC cod: Salvanes et al., 1994; Northwest Atlantic cod: Tupper and Boutilier, 1995; Baltic cod: Rehberg-Haas et al., 2012; North Sea cod: Bastrikin et al., 2014). Variations may be attributed to the wide geographical distribution of North Atlantic cod stocks resulting in a diversity of physical and biological environments and different life histories (Brander, 2005), but also stock genetics may play an important role, i.e. of settlement depth preferences (Fevolden et al., 2012). Otoliths are not only essential in ageing fish (Campana and Thorrold, 2001), but also in providing information on individual growth rates throughout life (Colloca et al., 2003) and timing of critical life history events such as settlement (Rehberg-Haas et al., 2012). Despite a general strong correlation between somatic growth and otolith growth, changes in temperature and food consumption may lead to a decoupling of this relationship (Neat et al., 2008). For the same reasons, a decoupling in otolith growth may be triggered by settlement, shown in Baltic cod (Hüssy et al., 2003b), challenging the reliability of backcalculation techniques (Campana and Neilson, 1985) and emphasizing the importance of understanding settling dynamics. Change in otolith shape during settlement has also been attributed to shifts in temperature (Villegas-Hernández et al., 2008), depth (Gauldie and Crampton, 2002), and food consumption (Hüssy, 2008) in a wide range of species. This introduces the possibility of using otolith shape to infer settlement in fish. Because otolith shape is stock-specific in many species, including Atlantic cod (Cardinale et al., 2004; Jónsdóttir et al., 2006; Stransky et al., 2008), the impact from settlement on otolith shape ontogeny may appear different between cod stocks. Altogether, this highlights the need for studies of settling dynamics explicitly on the high-latitude BS cod stock. In this study, we will examine how settlement of 0-group BS cod affects otolith shape, somatic body growth, and condition, and specifically we aim to elucidate how basic otolith morphometric parameters may be used to discriminate between age 0 and age 1 cod. Moreover, by comparing 0-group cod from pelagic and bottom trawl catches, we will use otolith shape, fish size, and body condition to infer the extent of potential settling behaviour already in August–September. Material and methods Sampling effort and methods In total, 3138 juvenile cod were sampled in the Barents Sea as part of two survey series: the Barents Sea Ecosystem Survey (BESS) conducted in August and September, and the winter survey in February and early March, by the research vessels “G.O. Sars” and “Johan Hjort” (Institute of Marine Research, Norway) (Table 1). All samples were taken from trawl stations distributed from 70–78°N and 13–44°E in the Barents Sea (Figure 1). In BESS surveys, the analysed 0-group and 1-group cod were sampled each year during 2005–2008. Two standard survey trawls were used: (i) pelagic Harstad trawl with a 10-mm (stretched) inner mesh in the codend, and (ii) bottom Campelen 1800 trawl with a 22-mm (stretched) mesh codend (Anonymous, 2008). Pelagic trawling was normally conducted at three depth ranges: 0–20, 20–40, and 40–60 m for 10 min at each depth, equivalent to a distance of 0.5 nautical miles at a speed of 3 knots (Mjanger et al., 2013). When echo registrations showed layers of 0-group fish deeper than 60 m, similar trawling at 60–80 m was also included. Although bottom trawling in the BESS does not deliberately target 0-group fish that typically remain in upper pelagic layers during the survey, 0-group cod were found in bottom trawl catches, although less frequent than in the pelagic trawl. These bottom trawl caught 0-group cod were sorted from larger juvenile and adult cod, but 0-group overlapped in size with a proportion of smaller 1-group cod and, therefore, were included in the samples. In the winter survey, newly settled cod juveniles (now 1-group) were sampled in 2006 and 2009, where only bottom trawling was conducted using Campelen 1800, as in the BESS. In both survey series, up to 100 cod were randomly sampled from the total catch at each station and frozen (–18 °C) for later measurements and analysis. Handling of trawl catches was conducted according to standard procedures (Mjanger et al., 2013). All geographical and technical survey data used in this study were retrieved from the IMR Central Database. All cod measured on board up to 22 cm (rounded down to the nearest cm) are presented (Figure 2). Table 1. Overview of the sampled juvenile cod from the Barents Sea Ecosystem Survey (BESS) in August–September and the winter survey (WS) in February–March. Year Survey No. stations No. fish samples No. otoliths image analysed Age group Cohort TL (mm) WW (g) RCF CRI Calendar day Mean SD Mean SD Mean SD Mean SD Mean SD 2005 BESS 6 231 169 24 1+s* 2004 140.0 18.3 19.3 7.7 0.88 0.04 0.89 0.06 245.3 2.4 145 0s 2005 92.2 12.1 6.7 3.2 1.10 0.09 1.22 0.13 235.6 4.9 2006 WS 4 67 63 63 1w 2005 120.9 12.3 12.0 3.6 0.89 0.06 0.90 0.05 56.7 7.0 2006 BESS 19 684 579 27 1+s* 2005 182.9 18.6 49.2 15.5 0.98 0.07 1.16 0.09 236.0 3.6 552 0s 2006 97.9 13.8 8.0 3.4 1.09 0.10 1.28 0.14 239.3 8.7 2007 BESS 9 402 182 32 1+s* 2006 169.3 19.6 38.4 13.4 0.97 0.07 1.05 0.07 260.2 10.1 150 0s 2007 84.6 20.3 5.4 4.1 1.03 0.08 1.11 0.14 245.9 12.4 2008 BESS 36 1680 632 632 0s 2008 87.6 9.8 4.9 1.8 0.97 0.08 1.09 0.11 254.8 2.2 2009 WS 3 74 60 60 1w 2008 117.7 12.1 10.8 3.7 0.85 0.05 0.90 0.05 45.3 5.5 Total 77 3138 1685 1479 0s All 91.6 13.9 6.3 3.2 1.03 0.11 1.17 0.15 246.3 10.5 123 1w All 119.3 12.3 3.7 3.7 0.87 0.06 0.90 0.05 51.2 8.5 89 1+s* All 165.3 25.4 36.4 17.4 0.95 0.07 0.94 0.09 248.0 12.3 Year Survey No. stations No. fish samples No. otoliths image analysed Age group Cohort TL (mm) WW (g) RCF CRI Calendar day Mean SD Mean SD Mean SD Mean SD Mean SD 2005 BESS 6 231 169 24 1+s* 2004 140.0 18.3 19.3 7.7 0.88 0.04 0.89 0.06 245.3 2.4 145 0s 2005 92.2 12.1 6.7 3.2 1.10 0.09 1.22 0.13 235.6 4.9 2006 WS 4 67 63 63 1w 2005 120.9 12.3 12.0 3.6 0.89 0.06 0.90 0.05 56.7 7.0 2006 BESS 19 684 579 27 1+s* 2005 182.9 18.6 49.2 15.5 0.98 0.07 1.16 0.09 236.0 3.6 552 0s 2006 97.9 13.8 8.0 3.4 1.09 0.10 1.28 0.14 239.3 8.7 2007 BESS 9 402 182 32 1+s* 2006 169.3 19.6 38.4 13.4 0.97 0.07 1.05 0.07 260.2 10.1 150 0s 2007 84.6 20.3 5.4 4.1 1.03 0.08 1.11 0.14 245.9 12.4 2008 BESS 36 1680 632 632 0s 2008 87.6 9.8 4.9 1.8 0.97 0.08 1.09 0.11 254.8 2.2 2009 WS 3 74 60 60 1w 2008 117.7 12.1 10.8 3.7 0.85 0.05 0.90 0.05 45.3 5.5 Total 77 3138 1685 1479 0s All 91.6 13.9 6.3 3.2 1.03 0.11 1.17 0.15 246.3 10.5 123 1w All 119.3 12.3 3.7 3.7 0.87 0.06 0.90 0.05 51.2 8.5 89 1+s* All 165.3 25.4 36.4 17.4 0.95 0.07 0.94 0.09 248.0 12.3 Means and SD of fish length (TL), fish weight (WW), relative condition factor (RCF), otolith crenulation index (CRI), and time of sampling (Calendar day) are summarized for the age groups of “0s”, “1w”, and “1+s” (“s” for summer and “w” for winter) for the 2004–2008 cod cohorts. * The mean values of the 1+s-group do not represent the true population means since 1+s-group cod was not targeted during sampling. Table 1. Overview of the sampled juvenile cod from the Barents Sea Ecosystem Survey (BESS) in August–September and the winter survey (WS) in February–March. Year Survey No. stations No. fish samples No. otoliths image analysed Age group Cohort TL (mm) WW (g) RCF CRI Calendar day Mean SD Mean SD Mean SD Mean SD Mean SD 2005 BESS 6 231 169 24 1+s* 2004 140.0 18.3 19.3 7.7 0.88 0.04 0.89 0.06 245.3 2.4 145 0s 2005 92.2 12.1 6.7 3.2 1.10 0.09 1.22 0.13 235.6 4.9 2006 WS 4 67 63 63 1w 2005 120.9 12.3 12.0 3.6 0.89 0.06 0.90 0.05 56.7 7.0 2006 BESS 19 684 579 27 1+s* 2005 182.9 18.6 49.2 15.5 0.98 0.07 1.16 0.09 236.0 3.6 552 0s 2006 97.9 13.8 8.0 3.4 1.09 0.10 1.28 0.14 239.3 8.7 2007 BESS 9 402 182 32 1+s* 2006 169.3 19.6 38.4 13.4 0.97 0.07 1.05 0.07 260.2 10.1 150 0s 2007 84.6 20.3 5.4 4.1 1.03 0.08 1.11 0.14 245.9 12.4 2008 BESS 36 1680 632 632 0s 2008 87.6 9.8 4.9 1.8 0.97 0.08 1.09 0.11 254.8 2.2 2009 WS 3 74 60 60 1w 2008 117.7 12.1 10.8 3.7 0.85 0.05 0.90 0.05 45.3 5.5 Total 77 3138 1685 1479 0s All 91.6 13.9 6.3 3.2 1.03 0.11 1.17 0.15 246.3 10.5 123 1w All 119.3 12.3 3.7 3.7 0.87 0.06 0.90 0.05 51.2 8.5 89 1+s* All 165.3 25.4 36.4 17.4 0.95 0.07 0.94 0.09 248.0 12.3 Year Survey No. stations No. fish samples No. otoliths image analysed Age group Cohort TL (mm) WW (g) RCF CRI Calendar day Mean SD Mean SD Mean SD Mean SD Mean SD 2005 BESS 6 231 169 24 1+s* 2004 140.0 18.3 19.3 7.7 0.88 0.04 0.89 0.06 245.3 2.4 145 0s 2005 92.2 12.1 6.7 3.2 1.10 0.09 1.22 0.13 235.6 4.9 2006 WS 4 67 63 63 1w 2005 120.9 12.3 12.0 3.6 0.89 0.06 0.90 0.05 56.7 7.0 2006 BESS 19 684 579 27 1+s* 2005 182.9 18.6 49.2 15.5 0.98 0.07 1.16 0.09 236.0 3.6 552 0s 2006 97.9 13.8 8.0 3.4 1.09 0.10 1.28 0.14 239.3 8.7 2007 BESS 9 402 182 32 1+s* 2006 169.3 19.6 38.4 13.4 0.97 0.07 1.05 0.07 260.2 10.1 150 0s 2007 84.6 20.3 5.4 4.1 1.03 0.08 1.11 0.14 245.9 12.4 2008 BESS 36 1680 632 632 0s 2008 87.6 9.8 4.9 1.8 0.97 0.08 1.09 0.11 254.8 2.2 2009 WS 3 74 60 60 1w 2008 117.7 12.1 10.8 3.7 0.85 0.05 0.90 0.05 45.3 5.5 Total 77 3138 1685 1479 0s All 91.6 13.9 6.3 3.2 1.03 0.11 1.17 0.15 246.3 10.5 123 1w All 119.3 12.3 3.7 3.7 0.87 0.06 0.90 0.05 51.2 8.5 89 1+s* All 165.3 25.4 36.4 17.4 0.95 0.07 0.94 0.09 248.0 12.3 Means and SD of fish length (TL), fish weight (WW), relative condition factor (RCF), otolith crenulation index (CRI), and time of sampling (Calendar day) are summarized for the age groups of “0s”, “1w”, and “1+s” (“s” for summer and “w” for winter) for the 2004–2008 cod cohorts. * The mean values of the 1+s-group do not represent the true population means since 1+s-group cod was not targeted during sampling. Figure 1. Open in new tabDownload slide Geographic distribution of the trawl station samples in the Barents Sea from 2005 to 2009; 70 stations during the Barents Sea Ecosystem survey (BESS; white marks) and 7 during the winter survey (WS; black marks). Figure 1. Open in new tabDownload slide Geographic distribution of the trawl station samples in the Barents Sea from 2005 to 2009; 70 stations during the Barents Sea Ecosystem survey (BESS; white marks) and 7 during the winter survey (WS; black marks). Figure 2. Open in new tabDownload slide Total length distributions of cod smaller than 22 cm measured on board during the Barents Sea Ecosystem survey (BESS) in 2005–2008 (n = 41 085) and the winter survey (WS) in 2006–2009 (n = 19 263). Data derived from IMR Central Database. Figure 2. Open in new tabDownload slide Total length distributions of cod smaller than 22 cm measured on board during the Barents Sea Ecosystem survey (BESS) in 2005–2008 (n = 41 085) and the winter survey (WS) in 2006–2009 (n = 19 263). Data derived from IMR Central Database. Fish measurements In the laboratory, frozen samples were thawed for ca 1 h prior to taking individual fish measurements of wet weight (WW, in gram), standard length (SL, in mm), and total length (TL, in mm). In cases of damaged caudal fins (∼3%), TL was estimated from the TL/SL relationship (r2 = 0.999, p < 0.001), where TL was, on an average, 9.8% larger than SL. The relative condition factor (RCF) of each juvenile was calculated by the formula: RCF=WWa TLb= WWobsWWest (1) where the coefficients a and b were determined empirically from the consulted data (n = 3138) by linear regression on the log10-transformed variables of TL and WW (Le Cren, 1951). The linear regression equation was: log WWest = –5.38 + 3.12 × log TL (r2 = 0.97, p < 0.001). Otolith imaging and shape analysis Sagittal otoliths (henceforward only referred to as otoliths) were extracted from up to 20 random individuals from each trawl sample. In the 2005 and 2006 surveys, up to 48 otoliths per trawl sample were extracted. Each otolith pair was photographed on the proximal side with a Nikon DS-Fi1 digital camera with controller DS-U2 mounted on a Leica MZ9.5 stereo microscope, using Nikon NIS Elements F (version 2.30) imaging software. The magnification was adjusted to the size of the otoliths, using either 1.6× or 1.25×. The images (JP2000, 2560 × 1920 pixels) were slightly overexposed on a black background to achieve a sharp white otolith edge under reflected light from two pipe lights (one on each side of the otoliths). The camera was set with a gain of 1.0 and a shutter time of 0.02–0.06 s, depending on the proximity of the light arms. Left otoliths were used since there is no systematic difference in size or shape between left and right otoliths in Atlantic cod (Suthers, 1996; Neat et al., 2008). Calibration images were captured at the beginning of each work session and in cases of changes magnification. In total, otoliths from 1685 fish were photographed and measured for otolith area (OA), otolith perimeter (OP), otolith length (OL), otolith width (OW), and circularity (CI) using ImageJ software v.1.51g (Rasband, 2016). OL and OW were based on maximum and minimum Feret (calliper) diameter, respectively, and CI was based on the OA–OP relationship as follows: CI= 4π ×OAOP2 (2) However, both otoliths with an elliptic shape or a crenulated perimeter will show low circularity values. Therefore, a modified CI descriptor (CImod) was developed by multiplying CI with the OL/OW component, which differentiate ellipticity from crenulation. The resulting index will then be relatively lower for crenulated shapes than elongated shape, compared to the circularity index: CImod=CI ×OLOW (3) The CImod values were then reflected over the mean value, so numbers increased with otolith crenulation (or lobe formation), hereafter referred to as CRI (crenulation index), where CRI=2n∑i=1nCImodi-CImod (4) If the left otolith was damaged, deformed, or crystalline (vaterite), the right otolith was used. Otolith transverse sections and age verification In the search for macrostructural settlement checks in 52 randomly selected otoliths, transverse sections of otoliths were examined for growth from core to edge, annotating potential checks or onset of winter annulus to the edge (0-group n = 19, and winter-sampled 1-group cod, n = 33) for both the repeatedly sampled cohorts of 2005 and 2008. The otoliths were embedded in epofix resin with hardener (Struers A/S, Ballerup, Denmark) and cut in 400–500 μm thick transverse sections through the core using an Isomet 1000 low-speed saw (Buehler Inc., Lake Bluff, IL). Sections were ground and polished to a thickness of 100–300 μm and photographed under a dissection microscope at 6.0× magnification, with the same camera setup as used to photograph the whole otolith. To ascertain that shape discrimination could also be applied (see Results section), an additional 20 otoliths from the summer-sampled juveniles were transversally cut and aged based on presence or absence of annuli; 10 of 0-group and 10 of 1-group cod. Statistics Generalized linear models (GLMs) were used to test for differences in mean body sizes (WW and TL), body condition (RCF), and otolith shape (CRI) between age groups (d.f. = 2), cohorts (d.f. = 3) or trawl type member (d.f. =1). A Hartleys’ Fmax was consulted to test for similarities in variance. GLM ANOVA was used to test for differences in regression coefficient and intercepts between age groups of the log OA-log OP and TL/CRI relationship and cohorts in the regression of TL/CRI (only 0-group). Due to unbalanced samples of 0-group cod between cohorts and between trawl types, and because the mean OL differed between these groups, the OL effect on the CRI was eliminated by using the residuals of the OL/CRI relationship of 0-group otoliths >2.4 mm OL (RESOL/CRI). Therefore, using GLMs, RESOL/CRI was used (instead of CRI) when testing differences in means between cohorts, and GLM ANCOVA when testing difference in regressions between trawls in the calendar day–RESOL/CRI relationship, only in 0-group cod. Kolmogorov–Smirnov (K–S) two-sample tests were applied on cumulative distributions of TL, RCF, and CRI, testing differences between 0-group and 1-group cod from the 2005 and 2008 cohorts. Statistical analyses were carried out in Statistica 13 (Dell Inc., 2015). Results Length distribution from surveys The use of fish length was not adequate to discriminate between age 0 and age 1 cod on the BESS survey because of size overlap (between 11 and 14 cm TL) despite a significant difference (GLM, p < 0.001). A similar body size overlap was confirmed in the overall survey data from the BESS and the winter surveys in the period 2005–2009, which is based on more extensive on-board fish measurements (n(TL < 22 cm) = 60 348) than in the present study (Figure 2). Age discrimination The present discrimination criteria resulted in three groups of fish defined by age and season (of the year): (i) 0-group and (ii) 1+-group cod that were sampled during the BESS survey, hereafter referred to as 0s-group and 1+s-group cod (“s” for summer), respectively, and (iii) 1-group cod sampled during the winter survey, hereafter referred to as 1w-group cod (“w” for winter). These age groups could be identified based on the OP–OA relationship, where two noticeable different otolith shapes in all otoliths (n = 1685) were observed in the log OP–log OA scatterplot (Figure 3). Both the winter sampled 1w-group otoliths, together with a smaller proportion of the BESS-sampled otoliths, had a significant lower slope in the OP–OA regression than the larger group of the BESS otoliths (GLM ANCOVA, p < 0.001). The group of BESS otoliths with the lower regression slope was regarded as 1+s-group (n = 83) because of their similarity to the 1w-group otoliths (n = 123) (GLM ANCOVA, p = 0.16). These 1+s individuals were also significantly largest (GLM, p < 0.001). The dominant group of BESS otoliths with the higher OP–OA regression slope were regarded as 0s-group (n = 1479) because 0-group cod were the main target for the sampling at the BESS survey (thus dominating in the material; Table 1) and their smaller sized otoliths. Figure 3. Open in new tabDownload slide Relationship between log otolith area (OA) and log otolith perimeter (OP) of otoliths of juvenile age 0 and 1 Barents Sea cod from the Barents Sea Ecosystem Survey (BESS; circles) and the winter survey (WS; triangles). Two groups of otolith shapes were non-overlapping and could be separated by a fitted linear cut-off line (log OP = 0.549 + 0.658 × log OA). Figure 3. Open in new tabDownload slide Relationship between log otolith area (OA) and log otolith perimeter (OP) of otoliths of juvenile age 0 and 1 Barents Sea cod from the Barents Sea Ecosystem Survey (BESS; circles) and the winter survey (WS; triangles). Two groups of otolith shapes were non-overlapping and could be separated by a fitted linear cut-off line (log OP = 0.549 + 0.658 × log OA). Age verification from seasonal otolith growth patterns The shape discrimination method was supported by age reading of transverse otolith sections from the 10 largest 0s-group cod and the 10 smallest 1+s-group cod, which would have the highest probability of being discriminated to an incorrect age group. Age was found to correspond unambiguously with the above otolith-shape-designated age groups. No annuli were found in the 0s-group otoliths (nor any settling checks that could be confused with an annulus), while a clear annulus were found in each of the 1+s-group otoliths. Additionally, age was confirmed in the 52 randomly selected otoliths for macrostructural examination. All 0s-group otoliths were lacking settling checks, while translucent marginal increments were found in 22 of 33 1w-group otoliths that probably represented the onset of the first annulus, but could also potentially have coincided with time of settlement. Macrostructural settlement checks prior to onset of first annulus were, however, not detected in 1w-group otoliths. Otolith lobe formation and settlement Overall, the distinct lobate 0s-group otoliths had significantly higher CRI than the 1w-group and 1+s-group otoliths (GLM, p < 0.001), which had a less distinct lobe and a smoother surface (Figure 4). This difference was also significant within each of the two cohorts (2005 and 2008) that were sampled in both summer and winter represented in both age groups (GLM, both p < 0.001). The otolith shape development throughout the juvenile period (0s-group at 4–6 months of age, 1w-group 10–11 months, and 1+s-group 16–18 months) showed, however, that CRI increased with fish size in all the age groups, especially in the 0s-group (Figure 5). However, this trend was interrupted by a noticeable drop in CRI in fish between 10 and 13 cm during autumn. This explains why the 1w- and 1+s-group were less lobed than the 0-group otolith. The rapid increase in CRI with TL of 0s-group cod was evident in all years (GLM, all r2 > 0.40, all p < 0.001), although between-year variations did occur (GLM ANOVA, p < 0.001). Throughout age 1, from first winter (1w-group) to next summer/early autumn (1+s-group), the CRI was relative stable, but still somewhat increasing with size and age (GLM, r2 = 0.26, p < 001). However, among the 0s-group and 1+s-group fish that were overlapping, CRI was still noticeably lower in the 1+s-group otoliths and not overlapping with the 0s-group, of which group that were sampled simultaneously at the BESS. For the sake of completeness, overviews of all mean values of TL, WW, RCF, and CRI of each cohort-specific age group and sampling days were presented (Table 1). Figure 4. Open in new tabDownload slide Left otolith from three specially selected juvenile Barents Sea cod of similar size (TL = 127 mm) from the 0s-, 1w-, and 1+s-group (“s” for summer and “w” for winter). Otoliths are photographed with sulcus (proximal side) facing camera and rostrum pointing up. Scale bar = 1 mm. Figure 4. Open in new tabDownload slide Left otolith from three specially selected juvenile Barents Sea cod of similar size (TL = 127 mm) from the 0s-, 1w-, and 1+s-group (“s” for summer and “w” for winter). Otoliths are photographed with sulcus (proximal side) facing camera and rostrum pointing up. Scale bar = 1 mm. Figure 5. Open in new tabDownload slide The relationship between total length (TL) and otolith crenulation index (CRI) in juvenile Barents Sea cod from the age groups 0s-group (circles, “s” for summer); 1w-group (triangles, “w” for winter); and 1+s-group (squares). The dashed trend line indicates the direction of change in CRI with growth. Figure 5. Open in new tabDownload slide The relationship between total length (TL) and otolith crenulation index (CRI) in juvenile Barents Sea cod from the age groups 0s-group (circles, “s” for summer); 1w-group (triangles, “w” for winter); and 1+s-group (squares). The dashed trend line indicates the direction of change in CRI with growth. Age and cohort-specific differences Although total length, body condition, and CRI were different between the 2005 and 2008 cohorts, the difference from summer to winter (0s-group vs. 1w-group) were larger in all the variables (Figure 6a–c). Fish size with increased from summer to winter (29 mm in 2005 and 30 mm in 2008) equally in all size proportions (K–S2005, Dmax = 0.79, p < 0.001; K–S2008, Dmax = 0.89, p < 0.001). RCF was reduced from summer to winter in both cohorts, however, somewhat more in the cohort of 2005 (K–S2005, Dmax = 0.85, p < 0.001; K–S2008, Dmax = 0.69, p < 0.001). Similarly, the CRI also decreased in both cohorts (K–S2005, Dmax = 0.94, p < 0.001; K–S2008, Dmax = 0.82, p < 0.001), but from a significantly higher CRI of the 2005 0s-group otoliths vs. the 2008 cohort (K–S, Dmax = 0.45, p < 0.001) to almost identical CRIs of the 1w-group otolith (K–S, Dmax = 0.15, p > 0.05). Figure 6. Open in new tabDownload slide Cumulative distributions of total length (TL) (a), body condition (RFC) (b), and otolith crenulation (CRI) (c) from 0s-group (thick lines, “s” for summer) and 1w-group (thin lines, “w” for winter) Barents Sea cod from the 2005 (dashed) and 2008 (solid) cohorts. Figure 6. Open in new tabDownload slide Cumulative distributions of total length (TL) (a), body condition (RFC) (b), and otolith crenulation (CRI) (c) from 0s-group (thick lines, “s” for summer) and 1w-group (thin lines, “w” for winter) Barents Sea cod from the 2005 (dashed) and 2008 (solid) cohorts. Pelagic vs. bottom trawling The 0s-group juveniles were generally bigger (TL) in bottom-trawl catches in all years (GLM, p < 0.01), except for 2005 (GLM, p = 0.38) (Figure 7). The RCF was, however, significantly lower in the bottom trawl (despite the bigger fish size) in all years (GLM, p < 0.01) except for 2007 (GLM, p = 0.32). Also, the RESOL/CRI of 0s-group cod otoliths was lower in the bottom trawl than in the pelagic trawl (GLM, year 2005–2007, p < 0.001 and year 2008, p < 0.05). In contrast to the strong and positive CRI/TL relationship in the 0s-group fish, the RESOL/CRI decreased steadily with time during the BESS, indicating an early reduction in otolith lobes, before the clear drop between later in autumn. This reduction was evident in both trawl types (Figure 8), but decreased somewhat faster in fish from the pelagic trawl (r2 = 0.53, p < 0.001, RESOL/CRI = 1.87−0.0075 × calendar day) than in the bottom trawl (r2 = 0.30, p < 0.001, RESOL/CRI = 1.41−0.0058 × calendar day) (GLM ANCOVA, p < 0.001). Figure 7. Open in new tabDownload slide Means of total length (TL) (a), body condition (RCF) (b), and relative otolith crenulation at otolith size (RESOL/CRI) (c) from 0-group cod from catches of the pelagic trawls (open) and the bottom trawls (filled) during the Barents Sea Ecosystem Survey (BESS). Boxes indicate ± 2 × SE and whiskers indicate ± SD. Levels of significance are indicated above the plots; ns (p > 0.05), * (p < 0.05), ** (p < 0.01), *** (p < 0.001). Figure 7. Open in new tabDownload slide Means of total length (TL) (a), body condition (RCF) (b), and relative otolith crenulation at otolith size (RESOL/CRI) (c) from 0-group cod from catches of the pelagic trawls (open) and the bottom trawls (filled) during the Barents Sea Ecosystem Survey (BESS). Boxes indicate ± 2 × SE and whiskers indicate ± SD. Levels of significance are indicated above the plots; ns (p > 0.05), * (p < 0.05), ** (p < 0.01), *** (p < 0.001). Figure 8. Open in new tabDownload slide Relative otolith crenulation at otolith size (RESOL/CRI) of 0-group cod otoliths (otolith lengths > 2.4 mm) from catches of pelagic trawls (open circles) and bottom trawls (filled triangles) during the Barents Sea Ecosystem Survey (BESS) in August–September. Figure 8. Open in new tabDownload slide Relative otolith crenulation at otolith size (RESOL/CRI) of 0-group cod otoliths (otolith lengths > 2.4 mm) from catches of pelagic trawls (open circles) and bottom trawls (filled triangles) during the Barents Sea Ecosystem Survey (BESS) in August–September. Discussion In this study, marked temporal changes in otolith shape of juvenile BS cod were documented and used to objectively discriminate between age 0 and age 1 cod from the BESS survey. Discrimination between fish species (Yu et al., 2014), fish stocks (Campana and Casselman, 1993; Stransky et al., 2008), spawning grounds (Galley et al., 2006), and age (Beyer and Szedlmayer, 2010) based on otolith shape has been demonstrated in various data sets, also with respect to ageing of Atlantic cod (Doering-Arjes et al., 2008). Common for these studies is that the discrimination methods are based on complex Fourier descriptors and/or multiple morphometric descriptors, typically using multivariate analysis or principal component analysis. However, in this study, only two otolith morphometric descriptors (otolith area and perimeter) were found necessary to discriminate between age 0 and age 1 cod, which later was verified by traditional age reading. Consequently, this allowed the 1685 cod juveniles to be objectively classified by their respective groups of age 0 and age 1+ without the need for traditional age reading based on otolith increment counts. The current use of digital otolith-shape analysis for age discrimination has not only the potential of being automated for time- and cost-effectiveness, but it may also remove uncertainties that may be introduced through subjective age reading, i.e. that potential otolith checks or false annuli may bias the age estimation (Wright et al., 2002; Neat et al., 2008). The increased CRI with total length prior to settlement was evidently a result of progressing otolith lobes, as they originate from secondary growth centres (accessory primordia) that are typically associated with the larval–juvenile stage transition (Gartner, 1991). Similar to the otoliths from the pre-settled BS 0s-group cod (5–12 cm TL) in the present study, lobe formation was documented to increase in early juvenile Baltic cod of 3–7 cm SL (Hüssy, 2008) (Note: in our analysis, TL was found to be 9.8% larger than SL, see above). Moreover, the progression of lobes was suggested to continue beyond the early life stages and well into the late juvenile period of Baltic cod, up to 20–30 cm SL. This differs from BS cod, where transient but distinct reduction in CRI evidently occurred during autumn/early winter, after the BESS survey in August/September, but before the winter survey in February/March, in juveniles of 9–14 cm TL. When the 1w-group cod were sampled, otolith lobes were noticeably less distinct. The fact that the disruption in lobe development coincides with the period of juvenile settlement in autumn suggests that environmental factors associated with the shift in habitat also affect the otolith shape in cod. Settlement has been suggested to influence otolith shape in the coral reef fish Stegastes partitus by means of changed temperature and food accessibility (Villegas-Hernández et al., 2008), which typically influence otolith growth. In BS 0-group cod, settling is typically associated with changes in prey type and availability (Ponomarenko, 1965, 1979; Renkawitz et al., 2011), and predation (Tupper and Boutilier, 1995), which are among the most important factors for growth and post-settlement survival. However, settlement appears to be a gradual process in the Barents Sea (Ottersen et al., 2014) and in other shelf areas (Hüssy et al., 2003b; Bastrikin et al., 2014), characterized by frequent exploratory migrations towards the seabed before becoming near-bottom associated. Consequently, a change in diet may, therefore, happen gradually. In pelagic 0-group BS cod, just prior to settlement, the diet consists of planktivorous prey items dominated by krill and amphipods (and copepods earlier in summer) (Dalpadado and Bogstad, 2004; Dalpadado et al., 2009), while the diet of settled 0-group cod seems to include more shrimps, polychaetes, and small fish, but also with krill and amphipods as important contributors (Ponomarenko, 1965; Dalpadado and Bogstad, 2004). Thus, it seems unlikely that diet change alone can explain the major change in otolith shape. Furthermore, fish body condition was significantly reduced during autumn, suggesting reduced food intake while settling. Settlement forces juveniles to adapt to both greater depths with less available light and reduced daylight hours or near total darkness in winter (Sundby et al., 2016). Since cod is mainly a visual predator, this is expected to greatly affect its feeding behaviour and its ability to detect and catch prey items (Meager et al., 2010). Feeding conditions in the post-settled period contrasts with the pre-settled pelagic period in summer, the latter having much longer daylight at high latitudes (70–78°N), and thus longer feeding time and overall higher growth (Suthers and Sundby, 1996; Helle, 2000). Because food intake and metabolism affect general otolith growth (Gauldie and Nelson, 1990), reduced food intake and growth during settlement can possibly contribute to reduced otolith lobe formation, as found in this study. Recent feeding history has been shown to affect otolith shape in juvenile Baltic cod (Hüssy, 2008) and juvenile coral reef fish (Gagliano and McCormick, 2004). Higher food consumption has also been suggested to increase otolith growth in the lobe areas (Hüssy, 2008), where otolith lobes became more opaque due to higher protein levels than in the between-lobe areas. Thus, progressing lobe formation was assumed to be a response to increased protein synthesis with higher food intake. Under low feeding conditions in settling BS cod, the accretion rate in lobe areas may, therefore, be expected to be reduced, permitting the intralobe areas to be filled through regular otolith growth. This would reduce the overall surface crenulations, as observed in the 1w-group cod otoliths. Early settlement of BS 0-group cod has previously been indicated by an underrepresentation of 0-group cod in the pelagic trawl towards the end of the BESS (Dingsør, 2006). Compared to 0-group fish in pelagic trawls, 0-group fish in bottom trawls had lower body condition, despite being bigger, and they had less crenulated otoliths. This suggested that a proportion of the 0-group cod had settled or were undertaking vertical exploratory migrations during this survey. However, due to interannual variations in the factors that may influence timing of fish settlement, spatiotemporal settlement dynamics are also likely to vary between years (Bastrikin et al., 2014), and thus, may influence 0-group indices differently between years. Numerous studies on cod show stock-specific otolith shape where both genetics and environmental factors are suggested to explain the differences (Cardinale et al., 2004; Jónsdóttir et al., 2006). Otolith lobe formation seen in BS cod should not uncritically be generalized between different stocks. To which extent stock characteristics of otolith shape are attributed to genetic contributions and environmental factors, or a combination of these, is difficult to study in wild fish and is not well understood. Genetic influence on otolith growth has been demonstrated experimentally in co-reared juvenile BS and NC cod (Otterlei et al., 2002), and otolith growth rate has been shown to highly influence otolith shape (Campana and Casselman, 1993). Yet, the different environments inhabited by the two stocks is thought to account for most of the otolith-shape characteristics that distinguish BS and NC cod otoliths (Stransky et al., 2008). Different preferences of settling area and depth have been reported for pelagic juveniles of BS and NC cod off the coast of northern Norway (Fevolden et al., 2012). It is, therefore, likely that otolith growth and shape may be influenced differently between fish from the two stocks, independently of genetics. However, because both BS and NC cod juveniles were observed in bottom-trawl catches, it could indicate a colocation of post-settled juveniles from both stocks (Fevolden et al., 2012). This would eliminate environmental influence as a confounding factor, allowing for future investigations of genetic effects on otolith development over settlement. Conclusions In this study, we have documented changes in otolith shape of juvenile BS cod and demonstrated a simple method to objectively discriminate between age 0 and age 1 cod sampled in August–September (in the BESS) by using only two otolith morphometric descriptors: otolith area and perimeter. A major reduction in CRI (less pronounced otolith lobes) and fish body condition during autumn and early winter was identified, which contrasts with the positive relationship between fish size and CRI seen in pre-settled 0-group cod. This suggests that circumstances associated with settling influence otolith shape and cause otolith lobes of 0-group BS cod to nearly disappear. Decreasing trends in lobe formation combined with lower body condition (despite bigger 0-group fish size) in the bottom trawl compared to the pelagic trawl in the BESS survey indicated early settling behaviour in August–September, which could lead to an underrepresentation of 0-group cod in pelagic trawl catches. How temporal and spatial variations in settlement occur and to what extent they influence the pelagic trawl catches of 0-group cod in the BESS remain to be clarified. Longer term studies are needed, however, to unravel key factors underlying spatiotemporal settlement dynamics and post-settlement growth and success in BS cod. Comparative studies between colocated BS and NC cod could also help disentangle the genetic role in ontogenetic otolith development and lobe formation. Acknowledgements We are grateful to F. Midtøy and J. Skadal for their effort in sampling and processing fish and otoliths during the trawl surveys and in the laboratory, and the crews on the research vessels “Johan Hjort” and “G. O. Sars” for making the field sampling possible. We also thank R. D. M. Nash and A. J. Geffen for valuable comments and considerations to this study. References Anonymous . 2008 . Håndbok for vitenskapelig tråling, versjon 3.0 . Institute of Marine Research , Bergen . 94 pp. WorldCat COPAC Bastrikin D. K. , Gallego A. , Millar C. P. , Priede I. G. , Jones E. G. 2014 . Settlement length and temporal settlement patterns of juvenile cod (Gadus morhua), haddock (Melanogrammus aeglefinus), and whiting (Merlangius merlangus) in a northern North Sea coastal nursery area . ICES Journal of Marine Science , 71 : 2101 – 2113 . Google Scholar Crossref Search ADS WorldCat Bergstad O. A. , Jørgensen T. , Dragesund O. 1987 . Life-history and ecology of the gadoid resources of the Barents Sea . Fisheries Research , 5 : 119 – 161 . Google Scholar Crossref Search ADS WorldCat Beyer S. G. , Szedlmayer S. T. 2010 . The use of otolith shape analysis for ageing juvenile red snapper, Lutjanus campechanus . Environmental Biology of Fishes , 89 : 333 – 340 . Google Scholar Crossref Search ADS WorldCat Brander K. M. E (Ed). 2005 . Spawning and life history information for North Atlantic cod stocks. ICES Cooperative Research Report No. 274, 152 pp. Campana S. E. , Casselman J. M. 1993 . Stock discrimination using otolith shape analysis . Canadian Journal of Fisheries and Aquatic Sciences , 50 : 1062 – 1083 . Google Scholar Crossref Search ADS WorldCat Campana S. E. , Neilson J. D. 1985 . Microstucture of fish otoliths . Canadian Journal of Fisheries and Aquatic Sciences , 42 : 1014 – 1032 . Google Scholar Crossref Search ADS WorldCat Campana S. E. , Thorrold S. R. 2001 . Otoliths, increments, and elements: key to a comprehensive understanding of fish populations . Canadian Journal of Fisheries and Aquatic Sciences , 58 : 30 – 38 . Google Scholar Crossref Search ADS WorldCat Cardinale M. , Doering-Arjes P. , Kastowsky M. , Mosegaard H. 2004 . Effects of sex, stock, and environment on the shape of known-age Atlantic cod (Gadus morhua) otoliths . Canadian Journal of Fisheries and Aquatic Sciences , 61 : 158 – 167 . Google Scholar Crossref Search ADS WorldCat Colloca F. , Cardinale M. , Marcello A. , Ardizzone G. D. 2003 . Tracing the life history of red gurnard (Aspitrigla cuculus) using validated otolith annual rings . Journal of Applied Ichthyology , 19 : 1 – 9 . Google Scholar Crossref Search ADS WorldCat Dalpadado P. , Bogstad B. 2004 . Diet of juvenile cod (age 0-2) in the Barents Sea in relation to food availability and cod growth . Polar Biology , 27 : 140 – 154 . Google Scholar Crossref Search ADS WorldCat Dalpadado P. , Bogstad B. , Eriksen E. , Rey L. 2009 . Distribution and diet of 0-group cod (Gadus morhua) and haddock (Melanogrammus aeglefinus) in the Barents Sea in relation to food availability and temperature . Polar Biology , 32 : 1583 – 1596 . Google Scholar Crossref Search ADS WorldCat Dell Inc . 2015 . Statistica (Data Analysis Software System) , 13th edn. Dell Inc ., Texas, USA. www.software.dell.com/products/statistica/. WorldCat COPAC Dingsør G. E. 2005 . Estimating abundance indices from the international 0-group fish survey in the Barents Sea . Fisheries Research , 72 : 205 – 218 . Google Scholar Crossref Search ADS WorldCat Dingsør G. E. 2006 . Influence of spawning stock size and environment on abundance and survival of juveniles in commercially important fish stocks in the Barents Sea. Ph.D. Thesis. Department of Biology, University of Bergen, Norway. 41 pp. Doering-Arjes P. , Cardinale M. , Mosegaard H. 2008 . Estimating population age structure using otolith morphometrics: a test with known-age Atlantic cod (Gadus morhua) individuals . Canadian Journal of Fisheries and Aquatic Sciences , 65 : 2342 – 2350 . Google Scholar Crossref Search ADS WorldCat Dragesund O. , Olsen S. 1965 . On the possibility of estimating year-class strength by measuring echo-abundance of 0-group fish . Fiskeridirektoratets Skrifter, Serie Havundersøkelser , 13 : 48 – 75 . WorldCat Dragesund O. , Hylen A. , Olsen S. , Nakken O. 2008 . The Barents Sea 0-group surveys; a new concept of pre-recruitment studies. In Norwegian Spring-Spawning Herring and Northeast Arctic Cod. 100 Years of Research and Management, 1st edn, pp. 119–136 . Ed. by Nakken O. . Tapir Academic Press , Trondheim . 177 pp. Google Preview WorldCat COPAC Eriksen E. , Bogstad B. , Nakken O. 2011 . Ecological significance of 0-group fish in the Barents Sea ecosystem . Polar Biology , 34 : 647 – 657 . Google Scholar Crossref Search ADS WorldCat Fevolden S. E. , Westgaard J. I. , Pedersen T. , Præbel K. 2012 . Settling-depth vs. genotype and size vs. genotype correlations at the Pan I locus in 0-group Atlantic cod Gadus morhua . Marine Ecology Progress Series , 468 : 267 – 278 . Google Scholar Crossref Search ADS WorldCat Gagliano M. , McCormick M. I. 2004 . Feeding history influences otolith shape in tropical fish . Marine Ecology Progress Series , 278 : 291 – 296 . Google Scholar Crossref Search ADS WorldCat Galley E. A. , Wright P. , Gibb F. 2006 . Combined methods of otolith shape analysis improve identification of spawning areas of Atlantic cod . ICES Journal of Marine Science , 63 : 1710 – 1717 . Google Scholar Crossref Search ADS WorldCat Gartner J. V. 1991 . Life histories of 3 species of lanternfishes (Pisces: Myctophidae) from the eastern Gulf of Mexico. 1. Morphological and microstructural analysis of sagittal otoliths . Marine Biology , 111 : 11 – 20 . Google Scholar Crossref Search ADS WorldCat Gauldie R. W. , Crampton J. S. 2002 . An eco-morphological explanation of individual variability in the shape of the fish otolith: comparison of the otolith of Hoplostethus atlanticus with other species by depth . Journal of Fish Biology , 60 : 1204 – 1221 . Google Scholar Crossref Search ADS WorldCat Gauldie R. W. , Nelson D. G. A. 1990 . Otolith growth in fishes . Comparative Biochemistry and Physiology A - Comparative Physiology , 97 : 119 – 135 . Google Scholar Crossref Search ADS WorldCat Helle K. 2000 . Does the midnight sun increase the feeding rate and hence the growth rate of early juvenile Arcto-Norwegian cod Gadus morhua in the Barents Sea? Marine Ecology Progress Series , 197 : 293 – 297 . Google Scholar Crossref Search ADS WorldCat Hüssy K. 2008 . Otolith shape in juvenile cod (Gadus morhua): ontogenetic and environmental effects . Journal of Experimental Marine Biology and Ecology , 364 : 35 – 41 . Google Scholar Crossref Search ADS WorldCat Hüssy K. , Mosegaard H. , Hinrichsen H.-H. , Böttcher U. 2003a . Factors determining variations in otolith microincrement width of demersal juvenile Baltic cod Gadus morhua . Marine Ecology Progress Series , 258 : 243 – 251 . Google Scholar Crossref Search ADS WorldCat Hüssy K. , Mosegaard H. , Hinrichsen H.-H. , Böttcher U. 2003b . Using otolith microstructure to analyse growth of juvenile Baltic cod Gadus morhua . Marine Ecology Progress Series , 258 : 233 – 241 . Google Scholar Crossref Search ADS WorldCat Hylen A. , Nakken O. , Nedreaas K. 2000 . Northeast Arctic cod: fisheries life history, stock fluctuations and mangement. In Norwegian Spring-Spawning Herring and Northeast Arctic Cod. 100 Years of Research and Management, pp. 83–118 . Ed. by Nakken O. . Tapir Academic Press , Trondheim . 177 pp. Google Preview WorldCat COPAC ICES . 2016 . Report of the Arctic Fisheries Working Group (AFWG), Dates 19–25 April 2016, ICES HQ, Copenhagen, Denmark. ICES Document CM 2016/ACOM: 06. 621 pp. Jónsdóttir I. G. , Campana S. E. , Marteinsdottir G. 2006 . Otolith shape and temporal stability of spawning groups of Icelandic cod (Gadus morhua L.) . ICES Journal of Marine Science , 63 : 1501 – 1512 . Google Scholar Crossref Search ADS WorldCat Kjesbu O. S. , Bogstad B. , Devine J. A. , Gjøsæter H. , Howell D. , Ingvaldsen R. B. , Nash R. D. M. et al. 2014 . Synergies between climate and management for Atlantic cod fisheries at high latitudes . Proceedings of the National Academy of Sciences of the United States of America , 111 : 3478 – 3483 . Google Scholar Crossref Search ADS PubMed WorldCat Le Cren E. D. 1951 . The length-weight relationship and seasonal cycle in gonad weight and condition in the perch (Perca fluviatilis) . Journal of Animal Ecology , 20 : 201 – 219 . Google Scholar Crossref Search ADS WorldCat Meager J. J. , Moberg O. , Strand E. , Utne-Palm A. C. 2010 . Effects of light intensity on visual prey detection by juvenile Atlantic cod (Gadus morhua L.) . Marine and Freshwater Behaviour and Physiology , 43 : 99 – 108 . Google Scholar Crossref Search ADS WorldCat Mjanger H. , Hestenes K. , Svendsen B. V. , de Lange W. T. 2013 . Håndbok for prøvetaking av fisk og krepsdyr, 3.16 edn . Institute of Marine Research , Bergen . 169 pp. Google Preview WorldCat COPAC Neat F. C. , Wright P. J. , Fryer R. J. 2008 . Temperature effects on otolith pattern formation in Atlantic cod Gadus morhua . Journal of Fish Biology , 73 : 2527 – 2541 . Google Scholar Crossref Search ADS WorldCat Otterlei E. , Folkvord A. , Nyhammer G. 2002 . Temperature dependent otolith growth of larval and early juvenile Atlantic cod (Gadus morhua) . ICES Journal of Marine Science , 59 : 401 – 410 . Google Scholar Crossref Search ADS WorldCat Ottersen G. , Loeng H. 2000 . Covariability in early growth and year-class strength of Barents Sea cod, haddock, and herring: the environmental link . ICES Journal of Marine Science , 57 : 339 – 348 . Google Scholar Crossref Search ADS WorldCat Ottersen G. , Bogstad B. , Yaragina N. A. , Stige L. C. , Vikebø F. B. , Dalpadado P. 2014 . A review of early life history dynamics of Barents Sea cod (Gadus morhua) . ICES Journal of Marine Science , 71 : 2064 – 2087 . Google Scholar Crossref Search ADS WorldCat Ottersen G. , Helle K. , Bogstad B. 2002 . Do abiotic mechanisms determine interannual variability in length-at-age of juvenile Arcto-Norwegian cod? Canadian Journal of Fisheries and Aquatic Sciences , 59 : 57 – 65 . Google Scholar Crossref Search ADS WorldCat Ponomarenko I. Y. 1965 . Comparative characteristic of some biological indices of the bottom stages of 0-group cod belonging to the 1956, 1958, 1959, 1960 and 1961 year-classes . ICNAF Special Publication , 6 : 349 – 354 . WorldCat Ponomarenko I. Y. 1979 . Distribution, feeding, growth and survival of cod fingerlings of the abundant 1970 year-class . Trudy PINRO (in Russian) , 43 : 77 – 114 . WorldCat Rasband W. 2016 . ImageJ, 1.51g edn. National Institutes of Health, USA. Rehberg-Haas S. , Hammer C. , Hillgruber N. , Hüssy K. , Temming A. 2012 . Otolith microstructure analysis to resolve seasonal patterns of hatching and settlement in western Baltic cod . ICES Journal of Marine Science , 69 : 1347 – 1356 . Google Scholar Crossref Search ADS WorldCat Renkawitz M. D. , Gregory R. S. , Schneider D. C. 2011 . Habitat dependant growth of three species of bottom settling fish in a coastal fjord . Journal of Experimental Marine Biology and Ecology , 409 : 79 – 88 . Google Scholar Crossref Search ADS WorldCat Salvanes A. G. V. , Giske J. , Nordeide J. T. 1994 . Life history approach to habitat shifts for coastal cod . Aquaculture and Fisheries Management , 25 : 215 – 228 . WorldCat Steneck R. S. 2012 . Apex predators and trophic cascades in large marine ecosystems: learning from serendipity . Proceedings of the National Academy of Sciences of the United States of America , 109 : 7953 – 7954 . Google Scholar Crossref Search ADS PubMed WorldCat Stransky C. , Baumann H. , Fevolden S. E. , Harbitz A. , Høie H. , Nedreaas K. H. , Salberg A. B. et al. 2008 . Separation of Norwegian coastal cod and Northeast Arctic cod by outer otolith shape analysis . Fisheries Research , 90 : 26 – 35 . Google Scholar Crossref Search ADS WorldCat Sundby S. , Bjørke H. , Soldal A. V. , Olsen S. 1989 . Mortality rates during the early life stages and year-class strength of northeast Arctic cod (Gadus morhua L.) . Rapports Et Procès-Verbaux Des Réunions Du Conseil International Pour L’Exploration De La Mer , 191 : 351 – 358 . WorldCat Sundby S. , Drinkwater K. F. , Kjesbu O. S. 2016 . The North Atlantic spring-bloom system – where the changing climate meets the winter dark . Frontiers in Marine Science , 3 : 28. Google Scholar Crossref Search ADS WorldCat Suthers I. 1996 . A guide to the extraction and interpretation of otholiths from larval and pelagic juvenile Arcto-Norwegian Cod (Gadus morhua) . Fisken Og Havet , 13 : 1 – 13 . WorldCat Suthers I. M. , Sundby S. 1996 . Role of the midnight sun: comparative growth of pelagic juvenile cod (Gadus morhua) from the Arcto-Norwegian and a Nova Scotian stock . ICES Journal of Marine Science , 53 : 827 – 836 . Google Scholar Crossref Search ADS WorldCat Tupper M. , Boutilier R. G. 1995 . Size and priority at settlement determine growth and competitive success of newly settled Atlantic cod . Marine Ecology Progress Series , 118 : 295 – 300 . Google Scholar Crossref Search ADS WorldCat Villegas-Hernández H. , González-Salas C. , Aguilar-Perera A. , López-Gómez M. J. 2008 . Settlement dynamics of the coral reef fish Stegastes partitus, inferred from otolith shape and microstructure analysis . Aquatic Biology , 1 : 249 – 258 . Google Scholar Crossref Search ADS WorldCat Wright P. J. , Panfili J. , Morales-Nin B. , Geffen A. J. 2002 . Otoliths. In Manual of Fish Sclerochronology , pp. 31–57. Ed. by Panfili J. , de Pontual H. , Troadec H. , Wright P. J. . Ifremer-lRD coedition , Brest . 464 pp. Google Preview WorldCat COPAC Yu X. , Cao L. , Liu J. , Zhao B. , Shan X. , Dou S. 2014 . Application of otolith shape analysis for stock discrimination and species identification of five goby species (Perciformes: Gobiidae) in the northern Chinese coastal waters . Chinese Journal of Oceanology and Limnology , 32 : 1060 – 1073 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2017. All rights reserved. For Permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)