ICES Journal of Marine Science: Journal du Conseil

Stock, Brian C; Heppell, Scott A; Waterhouse, Lynn; Dove, India C; Pattengill-Semmens, Christy V; McCoy, Croy M; Bush, Phillippe G; Ebanks-Petrie, Gina; Semmens, Brice X

doi: 10.1093/icesjms/fsaa221pmid: N/A

Abstract Fish spawning aggregations (FSAs) are vulnerable to overexploitation, yet quantitative assessments of FSA populations are rare. We document an approach for how to conduct such an assessment, evaluating the response of Critically Endangered Nassau Grouper (Epinephelus striatus) to protections in the Cayman Islands. We assessed pre-protection status on all islands using length data from fishery catch. We then used 17 years of noninvasive length-frequency data, collected via diver-operated laser calipers, to estimate recruitment and spawning biomass of Nassau Grouper on Little Cayman following protection. Bimodal length distributions in 2017–2019 indicated a large recruitment pulse (4–8× average) derived from spawning in 2011. Biomass recovered to 90–106% of the pre-exploitation level after 16 years, largely driven by the strong 2011 year class. Length distributions were also bimodal in 2017–2019 on nearby Cayman Brac, implying a synchronous recruitment pulse occurred on both islands. Our results demonstrate that: (i) in situ length data can be used to monitor protected FSAs; (ii) spatiotemporal FSA closures can be effective, but success takes time if population recovery depends upon sporadic recruitment; and (iii) FSA fishery management targets may need to be higher than commonly recommended (i.e. spawning potential ratio >0.6 instead of 0.4). Introduction Fisheries management based on traditional stock assessments and effort controls has been effective at reducing overfishing for many fish populations (Beddington et al., 2007; Worm et al., 2009; Hilborn et al., 2020). However, one longstanding challenge occurs when fish exhibit aggregating behaviour that sustains high total catch and catch rates [i.e. catch-per-unit-effort (CPUE)] while abundance steeply declines. The formation of transient fish spawning aggregations (FSAs) is one such behaviour that leads to extremely high biomass density, readily predictable in space and time, and therefore often targeted by fisheries (Johannes, 1998; Claro and Lindeman, 2003; Sadovy and Domeier, 2005; Sadovy de Mitcheson and Erisman, 2012). Failing to recognize the “hyperstability” of FSA fisheries has contributed to notable collapses, such as for northern cod (Rose and Kulka, 1999) and orange roughy (Clark, 2001). Although less well publicized and smaller scale, many fisheries that target FSAs of large-bodied tropical reef species such as grouper (Epinephelidae) and snapper (Lutjanidae) have also collapsed (Johannes, 1998; Sadovy de Mitcheson et al., 2008; Claro et al., 2009; Robinson et al., 2015). In response to these declines, spatiotemporal closures are increasingly advocated as a practical and enforceable way to reduce fishing mortality (F) at FSAs, which can be very intense over small temporal and spatial scales (e.g. days and 100 s of metres; Russell et al., 2012; Sadovy de Mitcheson, 2016). Many spatiotemporal closures have been implemented to conserve FSAs, and cost-effective data collection and assessment methodologies are urgently needed to evaluate these protections (Claro and Lindeman, 2003; Grüss et al., 2014; Sherman et al., 2016). In most cases, populations continue to decline after protection or their status is unknown (Table 11.1 in Russell et al., 2012). Well-documented cases of FSA population increase following spatiotemporal protection do exist (Russ and Alcala, 2004; Nemeth, 2005; Luckhurst and Trott, 2009; Hamilton et al., 2011; Sadovy de Mitcheson and Colin, 2012; Waterhouse et al., 2020), but these are uncommon and rarely based on population dynamics models that can help explain the mechanisms underlying recovery through estimation of F and recruitment. Fisheries-dependent indices of abundance (e.g. CPUE) data can be cost effective to collect for non-protected FSAs but must be interpreted with caution due to concerns about hyperstability (Rose and Kulka, 1999; Robinson et al., 2015). Fisheries-independent surveys based on underwater visual census, mark-recapture, or acoustics techniques are informative but require more resources. Length-frequency data are relatively cost-effective to collect, and length-based analyses of FSAs soon after protections have shown increases in the mean size of fish, as expected due to reduced F (Beets and Friedlander, 1999; Nemeth, 2005; Luckhurst and Trott, 2009). Longer-term population recovery, however, is indicated by the recruitment of smaller fish and a trend toward an unfished length structure with a broad range of sizes (Heppell et al., 2012). Ideally, length-frequency data could be collected over multiple years and used in an assessment model that can distinguish between changes in size due to changes in F vs. changes in recruitment (e.g. Rudd and Thorson, 2018). For protected FSAs, scientists can use noninvasive methods such as underwater visual census, stereo cameras, or laser calipers to obtain length measurements (Colin, 2012a; Heppell et al., 2012). At FSAs where harvest is allowed, scientists can also collect length and weight data from the catch. Study species: Nassau Grouper Overfishing of FSAs has driven striking declines of Nassau Grouper (Epinephelus striatus) throughout the Caribbean Sea, providing a classic example of the challenge that aggregation behaviour poses for fisheries management (Sala et al., 2001; Sadovy de Mitcheson et al., 2020). Like many large-bodied (mature adults range from 45 to 90 cm total length), long-lived (at least 29 years), high trophic level reef fish, Nassau Grouper, are both highly sought after and vulnerable to fisheries (Sadovy and Eklund, 1999; Patrick et al., 2010; Hobday et al., 2011). Nassau Grouper are territorial and solitary outside spawning season but form extremely dense FSAs at highly predictable sites and times to spawn (e.g. 30 000 individuals in a 100 m × 500 m area; Smith, 1972). Nassau Grouper FSAs historically supported one of the most important finfish fisheries in the Caribbean, but intense and uncontrolled FSA fishing has led to the disappearance of two-thirds of known Nassau Grouper FSAs and a Critically Endangered listing by the IUCN (Sadovy de Mitcheson et al., 2008; Sadovy et al., 2018). The United States has prohibited take and possession of Nassau Grouper since 1990 and recently listed the species as Threatened under the U.S. Endangered Species Act (NMFS, 2016). Several governments, including Mexico, Belize, the Cayman Islands, and The Bahamas, have instituted spatial protections at known Nassau Grouper FSA sites and/or temporal protections covering the spawning season. These efforts have generally been successful at reducing F, but recovery has been variable and quantitative estimates of population responses—either abundance or size frequency—are rare (Ehrhardt and Deleveaux, 2007; Heppell et al., 2012; Sadovy de Mitcheson and Colin, 2012; Cheung et al., 2013; Waterhouse et al., 2020). In the Cayman Islands, a UK Overseas Territory in the Caribbean Sea, Nassau Grouper FSAs historically formed at five known locations (Figure 1). An additional FSA exists at Pickle Bank, an offshore seamount whose political jurisdiction is unclear due to the overlapping Exclusive Economic Zones of the Cayman Islands and Cuba. Fishermen have targeted Cayman FSAs with small boats and handlines around the full moons in January and February since the early 1900s (Bush et al., 2006). Responding to fishermen’s concerns over declining numbers and size of Nassau Grouper, in 1985 the Cayman Islands government restricted fishing FSAs to only residents using hook-and-line gear. In 1987, the Cayman Islands Department of the Environment (CI-DoE) began monitoring CPUE and collecting biological data (length, weight, sex, and age; Bush and Ebanks-Petrie, 1994). This study produced the most complete growth curve and oldest recorded individual for the species (29 years) and showed that total catch, CPUE, and mean size declined at all the main Cayman FSAs from 1987 to 2001 (Bush et al., 2006). Figure 1. Open in new tabDownload slide Map showing the location of historic and current Nassau Grouper spawning aggregations in the Cayman Islands. (a) Location of the Cayman Islands within the western Caribbean Sea. Aggregations (FSAs, black points in b) are found either at shelf edges near reef promontories (Grand Cayman, Little Cayman, Cayman Brac) or offshore seamounts that rise to within 30 m of the surface (Twelve-Mile Bank, Pickle Bank). The FSA off the west end of Little Cayman (black triangle in b, hatched area in c) is currently the largest and the focus of this study. In (b), bathymetry is shown with grey contour lines at 50, 100, 200, 500, and 1000 m. Acoustically tagged adult Nassau Grouper have not crossed deep water between islands (>200 m). In (c), the FSA site (hatched area) is defined by three moorings (points) and the shelf edge at 30–40 m. Figure 1. Open in new tabDownload slide Map showing the location of historic and current Nassau Grouper spawning aggregations in the Cayman Islands. (a) Location of the Cayman Islands within the western Caribbean Sea. Aggregations (FSAs, black points in b) are found either at shelf edges near reef promontories (Grand Cayman, Little Cayman, Cayman Brac) or offshore seamounts that rise to within 30 m of the surface (Twelve-Mile Bank, Pickle Bank). The FSA off the west end of Little Cayman (black triangle in b, hatched area in c) is currently the largest and the focus of this study. In (b), bathymetry is shown with grey contour lines at 50, 100, 200, 500, and 1000 m. Acoustically tagged adult Nassau Grouper have not crossed deep water between islands (>200 m). In (c), the FSA site (hatched area) is defined by three moorings (points) and the shelf edge at 30–40 m. In 2001, local fishermen discovered a new FSA off the west end of Little Cayman and divers estimated that the aggregation had roughly 7000 fish at the time of discovery (Bush et al., 2006; Whaylen et al., 2007). Intense fishing by local fishermen using handlines removed around 4000 fish in two consecutive 1-week spawning seasons (ca. 2000 in 2001 and 1934 in 2002; Whaylen et al., 2004). In 2003, the Cayman Islands Marine Conservation Board banned fishing on the aggregation site. Since 2003, the Grouper Moon Project, a collaboration between the CI-DoE, Reef Environmental Education Foundation (REEF), and academic scientists, has published several findings relevant to FSA assessment: (i) acoustically tagged adult fish on Little Cayman and Cayman Brac do not cross deep water to other islands and the vast majority of reproductive fish attend the one FSA on their island to spawn (Semmens et al., 2007, 2009); (ii) a decrease in mean length coupled with an increase in size range from 2004 to 2010 suggests that recruitment occurred on Little Cayman (Heppell et al., 2012); and (iii) since protection, fish numbers have increased substantially on Little Cayman, tentatively on Cayman Brac, and not at all on Grand Cayman (Waterhouse et al., 2020). Based on this evidence, the Cayman Islands government renewed the initial FSA fishing bans and subsequently enacted comprehensive Nassau Grouper management via legislation (no take during spawning months, bag and slot limits away from FSAs in the rest of the year; Cayman Islands Cabinet, 2016; Waterhouse et al., 2020). Management is currently settled by this legislation, and the primary objective is to ensure viability of the FSAs (i.e. FSAs form and spawning is observed) while allowing small amounts of recreational and artisanal catch outside of spawning season. We present a case study highlighting the value of using length-frequency data to evaluate the response of Cayman Islands Nassau Grouper populations to 16 years of spatiotemporal FSA protection. We first analyse pre-protection fishery catch data to estimate growth, mortality, and sspawning potential ratio (SPR) at FSAs throughout the Cayman Islands. We then combine 17 years of in situ length-frequency data with an estimate of abundance into a length-based stock assessment model for the Little Cayman FSA. We specifically address the following: (I) What was the population status at FSAs throughout the Cayman Islands before and after protections? (II) How have population size structure and biomass changed on Little Cayman and Cayman Brac since protection? (III) How variable was recruitment during the recovery monitoring period? (IV) Did growth rates differ by island before or after protections? Methods Historical Cayman FSA sites Of the five FSAs which supported fisheries before the closures in 2003, the CI-DoE collected fishery-dependent data (described below) primarily at the three with the vast majority of the catch, located on the east ends of Grand Cayman, Cayman Brac, and Little Cayman (Figure 1b; Bush et al., 2006). Few data were collected from the other two FSAs near Grand Cayman: the southwest FSA was no longer fished after it disappeared in 1990 and Twelve-Mile Bank was sporadically exploited and yielded lower numbers of fish. We excluded these data from our analyses. Pickle Bank is not regularly exploited, but 159 fish were opportunistically caught and measured in 2000. West end Little Cayman FSA site Since 2003, the Grouper Moon Project has logged thousands of diver-hours observing spawning behaviour and collecting fishery-independent data (described below) at the new FSA off the west end of Little Cayman (Figure 1c; Whaylen et al., 2004, 2007). For 7–10 days following the full moons in January and February, Nassau Grouper aggregate in a roughly 300 m × 50 m area well defined by three project-placed moorings, the insular shelf edge, and dive navigation lines connecting the three moorings to the shelf edge (Figure 1c). Stock structure assumption: FSAs represent populations Acoustic tagging of mature fish on- and off-FSA sites on Little Cayman and Cayman Brac has directly shown that only one FSA forms on each island, the vast majority (98%) of fish attend the FSA on their island, and fish do not cross deep water to other islands (Semmens et al., 2007, 2009). Thus, we assumed that distinct populations exist on each island (possibly linked via larval dispersal), and that data collected from the FSAs represent the entirety of the adult population on each island. Given that Pickle Bank is smaller than either Little Cayman or Cayman Brac, surrounded by deep water, and far from either the Cayman Islands or Cuba (Figure 1), the same is likely true for Pickle Bank. Although Grand Cayman is larger and formerly supported two FSAs, the southwestern FSA disappeared by 1990 and we assumed that the 1988–1997 data from the northeastern FSA represented the entire reproductive population. Fisheries-dependent biological data We used biological data collected from fisheries catch at the three main FSAs between 1978 and 2002 before protections were implemented (Colin et al., 1987; Bush and Ebanks-Petrie, 1994; Bush et al., 2006). These data included total length, weight, and sex of commercial catch and sagittal otoliths for ageing. Colin et al. (1987) measured standard lengths at the Little Cayman FSA in 1978, and we converted these to total lengths using the published relationship with largest sample size and nearest proximity to the Cayman Islands (Claro 1990 in Sadovy and Eklund, 1999). The ageing method was validated by Bush et al. (1996) using oxytetracycline injections in captive fish. Following the methods of Bush and Ebanks-Petrie (1994) and Bush et al. (2006), we added 1 year to otolith ring counts because all fish were caught on FSAs and had “plus” growth. Finally, total catch estimates were available from the newly discovered west end Little Cayman FSA for the 2 years it was heavily fished (∼2000 fish in 2001 and 1934 in 2002; Whaylen et al., 2004). In situ length-frequency data For each year from 2003 to 2019, we collected noninvasive length data from the west end Little Cayman FSA using diver-operated laser calipers. In addition to the aluminium bracket system described in Heppell et al. (2012), we created a new system with two red laser diodes placed 20 cm apart inside a waterproof acrylic housing, with a GoPro Hero™ video camera attached in front (Figure 2). We aimed to collect 500–1000 length measurements per year because simulation studies of length-based assessment methods show a performance plateau above this sample size (Hordyk et al., 2015b; Rudd and Thorson, 2018). The number of dives and days necessary to achieve this sample size varied by year, primarily depending on dive conditions and currents. We also collected in situ length data from the Cayman Brac FSA in 2017–2019, although we were not able to collect large sample sizes (n = 107, 125, and 115) because there were fewer and more dispersed fish, and the site typically has challenging boat operation and dive conditions (high seas, strong currents, deeper site). See Supplementary material for details of laser caliper and stereo camera calibration, measurement error estimation, video collection, and data processing (Supplementary Figures S1–S3). Figure 2. Open in new tabDownload slide Laser caliper system used to measure fish lengths in situ. Two parallel laser diodes are placed 20 cm apart inside a custom-machined acrylic housing, with a GoPro Hero™ camera in separate housing mounted in front. (a) A diver using the system. (b) Example still-frame image with the two laser dots on a fish perpendicular to the camera. The known distance between the laser dots is used as a scale to measure total length. Figure 2. Open in new tabDownload slide Laser caliper system used to measure fish lengths in situ. Two parallel laser diodes are placed 20 cm apart inside a custom-machined acrylic housing, with a GoPro Hero™ camera in separate housing mounted in front. (a) A diver using the system. (b) Example still-frame image with the two laser dots on a fish perpendicular to the camera. The known distance between the laser dots is used as a scale to measure total length. Fishery-independent abundance estimate, Little Cayman post-protection In the context of assessing small-scale FSA fisheries, it is rare to have fishery-independent data on abundance. We were fortunate to have access to yearly estimates of the number of mature fish at the west end Little Cayman FSA from 2005 to 2018 (Waterhouse et al., 2020), which we used as an index of abundance in a length-based stock assessment model for the Little Cayman population (described below). Waterhouse et al. (2020) fit a state-space model of the number of spawners, modelling the population as a random walk with drift in log-space, i.e. log⁡St+1=log⁡St+μ+εt ⁠, where St is the number of spawners in year t, μ is the mean population growth rate, and εt is the annual deviation in growth rate in year t. The model was then fit to diver-collected mark-resight and video census data using Markov Chain Monte Carlo sampling. Since the assessment model (described below) assumed that the yearly abundance estimates were independent, we re-analysed the mark-resight data without the assumption that population growth is a function of population size, i.e. we removed Eq. 1 from Waterhouse et al. (2020) and simply estimated St using the number of fish tagged, Kt ⁠, and the proportion of tagged sides of fish in mark-resight surveys, pt ⁠: St=Kt/2pt ⁠. We also allowed for the possibility that the credible interval widths were too small to use as the index observation error, σI ⁠, in the assessment. We approximated σI as the mean of the approximate Z-scores from the Waterhouse et al. (2020) median posterior estimates of the number of spawners, S50% ⁠: S97.5%-S2.5%2×1.96×S50%, where S97.5%-S2.5% is the 95% CI width. We then considered this value, σI = 0.103, as a lower bound in our assessment model and conducted a sensitivity analysis on σI (Supplementary Figure S4). Estimating island-specific growth To estimate growth, we used the length-age data from 1988 to 1992 collected at the three main historic FSAs (n = 99, 132, and 246 from Little Cayman, Cayman Brac, and Grand Cayman, respectively; Figure 3; Bush et al., 2006). We also included 25 lengths of un-aged fish presumed to be 1-year old in February 2012 on Little Cayman from a large recruitment event from February 2011 spawning (Camp et al., 2013; Semmens et al., 2013). We modelled the length-at-age of fish i, L[ai] ⁠, using the von Bertalanffy function: Lai=L∞1-e-Kai-a0+ εi, εi ∼ N0,σεi2, σεi=CVL×L∞1-e-Kai-a0, where L∞ is the asymptotic length, K is the growth coefficient, a0 is the theoretical age when length is zero, and the variation of length-at-age increases with mean length and is normally distributed with variance, σε2 ⁠. Since exploratory analysis showed potential differences in length-at-age between the three islands, we fit a series of hierarchical growth models that allowed for island-specific deviations from the overall mean parameters (Kimura, 1980; Helser and Lai, 2004). These ranged in complexity from the simplest model, m1, with no island-specific deviations, to the full model, m8, with all parameters varying by island (Table 1; Ogle, 2016). As in Helser and Lai (2004), we modelled the growth parameter vectors for each island j, θj=(L∞j, lnKj, a0j) ⁠, as random effects assumed to follow a multivariate normal distribution with mean μ=(L∞, lnK, a0) and covariance matrix Σ ⁠, i.e.: θj= L∞jlnKjt0j ∼ MVNμ, Σ, Σ= σL∞2σL∞lnKσL∞a0σL∞lnKσlnK2σlnKt0σL∞a0σlnKt0σt02. Figure 3. Open in new tabDownload slide Island-specific Nassau Grouper growth curves from the Cayman Islands. Fish age-4 and older were sampled from 1988 to 1992 FSA catch on Little Cayman, Cayman Brac, and Grand Cayman (n = 99, 132, and 246). Little Cayman age-1 fish (n = 25) were sampled away from the FSA in February 2012. The black dashed line and shading depict the predicted length at age and 95% CI from the overall growth curve. Growth curve parameter estimates are given in Table 2. Figure 3. Open in new tabDownload slide Island-specific Nassau Grouper growth curves from the Cayman Islands. Fish age-4 and older were sampled from 1988 to 1992 FSA catch on Little Cayman, Cayman Brac, and Grand Cayman (n = 99, 132, and 246). Little Cayman age-1 fish (n = 25) were sampled away from the FSA in February 2012. The black dashed line and shading depict the predicted length at age and 95% CI from the overall growth curve. Growth curve parameter estimates are given in Table 2. Table 1. Hierarchical von Bertalanffy growth models for Cayman Islands Nassau Grouper, where L^a=L∞j1-e-Kja-a0j for island j. Model . Island-specific parameters . No. fixed effect parameters . Converged and pos. def. Hessian . AIC . ΔAIC . m1 – 4 Yes 3 112.8 34.4 m2 K 5 Yes 3 078.4 0 m3 L∞ 5 Yes 3 082.2 3.8 m4 a0 5 Yes 3 087.9 9.5 m5 K, L∞ 7 No 3 090.7 12.3 m6 L∞, a0 7 No 3 086.4 8.0 m7 K, a0 7 No 3 088.7 10.3 m8 K, L∞, a0 10 No – – Model . Island-specific parameters . No. fixed effect parameters . Converged and pos. def. Hessian . AIC . ΔAIC . m1 – 4 Yes 3 112.8 34.4 m2 K 5 Yes 3 078.4 0 m3 L∞ 5 Yes 3 082.2 3.8 m4 a0 5 Yes 3 087.9 9.5 m5 K, L∞ 7 No 3 090.7 12.3 m6 L∞, a0 7 No 3 086.4 8.0 m7 K, a0 7 No 3 088.7 10.3 m8 K, L∞, a0 10 No – – Open in new tab Table 1. Hierarchical von Bertalanffy growth models for Cayman Islands Nassau Grouper, where L^a=L∞j1-e-Kja-a0j for island j. Model . Island-specific parameters . No. fixed effect parameters . Converged and pos. def. Hessian . AIC . ΔAIC . m1 – 4 Yes 3 112.8 34.4 m2 K 5 Yes 3 078.4 0 m3 L∞ 5 Yes 3 082.2 3.8 m4 a0 5 Yes 3 087.9 9.5 m5 K, L∞ 7 No 3 090.7 12.3 m6 L∞, a0 7 No 3 086.4 8.0 m7 K, a0 7 No 3 088.7 10.3 m8 K, L∞, a0 10 No – – Model . Island-specific parameters . No. fixed effect parameters . Converged and pos. def. Hessian . AIC . ΔAIC . m1 – 4 Yes 3 112.8 34.4 m2 K 5 Yes 3 078.4 0 m3 L∞ 5 Yes 3 082.2 3.8 m4 a0 5 Yes 3 087.9 9.5 m5 K, L∞ 7 No 3 090.7 12.3 m6 L∞, a0 7 No 3 086.4 8.0 m7 K, a0 7 No 3 088.7 10.3 m8 K, L∞, a0 10 No – – Open in new tab This model was attractive because it accounts for parameter correlations and borrows strength across islands to estimate island-specific growth parameters, despite some islands having few samples of older or younger fish. We implemented the hierarchical growth model in Template Model Builder, which uses Laplace approximation to provide maximum likelihood estimates (MLEs) of the fixed effects and empirical Bayes estimates of the random effects (Kristensen et al., 2016). We assessed model convergence by confirming that the Hessian was positive definite and that the absolute values of all final gradients were <0.0001. To select the most parsimonious of the eight nested models, we used marginal Akaike’s Information Criterion, AIC= -2 log ℓ+2p ⁠, where ℓ is the marginal likelihood evaluated at the MLE and p is the number of estimated parameters (Table 1; Burnham and Anderson, 2002). Finally, we fit the length-weight relationship as a log-linear model, log(W) = α + β log(L), with the R function “lm”. Estimating natural mortality We estimated natural mortality, M, using catch-curve analysis of the length data from Pickle Bank (n = 159). First, we converted length data to ages using the mean parameters from the best-fit hierarchical growth model (Table 2 and Figure 3), and then followed the guidance of Smith et al. (2012) to use the Chapman-Robson estimator (Chapman and Robson, 1960), implemented in the “FSA” R package (Ogle et al., 2018). This estimate of M depends on the assumption that Pickle Bank is unexploited, and M will be biased upwards if this is not true. We consider the assumption that fishing pressure on Pickle bank is low to be reasonable given how small and isolated Pickle Bank is from the Cayman Islands and Cuba (Figure 1). In addition, adult Nassau Grouper are extremely unlikely to leave Pickle Bank and experience fishing pressure elsewhere, given that they do not appear to move between Little Cayman and Cayman Brac (acoustic and floy tagging data not shown) despite the islands being separated only by 8 km and 250 m deep water. Furthermore, the substantial proportion of large fish caught on Pickle Bank indicates a high probability of a natural age structure (Figure 4). Figure 4. Open in new tabDownload slide Nassau Grouper catch length distributions from FSA fisheries in the Cayman Islands before protections. Data are pooled across years for Cayman Brac (1990–1995, 1998, 2000), Grand Cayman (1988–1989, 1993, 1995, 1997), and Pickle Bank (2000). Data from Little Cayman were divided into two separate periods, 1987–1995 and 2002, because (i) no FSA fishing occurred for 6 years between 1995 and 2001 and (ii) data from 1987 to 1995 are from the historic east end site while data from 2002 are from the rediscovered west end site. Dashed lines indicate the mean total length for each FSA. Figure 4. Open in new tabDownload slide Nassau Grouper catch length distributions from FSA fisheries in the Cayman Islands before protections. Data are pooled across years for Cayman Brac (1990–1995, 1998, 2000), Grand Cayman (1988–1989, 1993, 1995, 1997), and Pickle Bank (2000). Data from Little Cayman were divided into two separate periods, 1987–1995 and 2002, because (i) no FSA fishing occurred for 6 years between 1995 and 2001 and (ii) data from 1987 to 1995 are from the historic east end site while data from 2002 are from the rediscovered west end site. Dashed lines indicate the mean total length for each FSA. Table 2. Estimated von Bertalanffy growth function parameters from model m2, which allowed Kj to vary by island, i.e. L^a=L∞1-e-Kja-a0 for island j. Parameter . Overall/mean . Little Cayman . Cayman Brac . Grand Cayman . L∞ (cm) 80.2 (76.8, 83.7) – – – Kj (1/year) 0.155 (0.134, 0.175) 0.140 (0.125, 0.156) 0.160 (0.143, 0.178) 0.164 (0.146, 0.182) a0 −0.832 (−0.984, −0.680) – – – CVL 0.092 (0.087, 0.098) – – – No. fish 502 124 132 246 Parameter . Overall/mean . Little Cayman . Cayman Brac . Grand Cayman . L∞ (cm) 80.2 (76.8, 83.7) – – – Kj (1/year) 0.155 (0.134, 0.175) 0.140 (0.125, 0.156) 0.160 (0.143, 0.178) 0.164 (0.146, 0.182) a0 −0.832 (−0.984, −0.680) – – – CVL 0.092 (0.087, 0.098) – – – No. fish 502 124 132 246 95% confidence interval limits are given in parentheses. Open in new tab Table 2. Estimated von Bertalanffy growth function parameters from model m2, which allowed Kj to vary by island, i.e. L^a=L∞1-e-Kja-a0 for island j. Parameter . Overall/mean . Little Cayman . Cayman Brac . Grand Cayman . L∞ (cm) 80.2 (76.8, 83.7) – – – Kj (1/year) 0.155 (0.134, 0.175) 0.140 (0.125, 0.156) 0.160 (0.143, 0.178) 0.164 (0.146, 0.182) a0 −0.832 (−0.984, −0.680) – – – CVL 0.092 (0.087, 0.098) – – – No. fish 502 124 132 246 Parameter . Overall/mean . Little Cayman . Cayman Brac . Grand Cayman . L∞ (cm) 80.2 (76.8, 83.7) – – – Kj (1/year) 0.155 (0.134, 0.175) 0.140 (0.125, 0.156) 0.160 (0.143, 0.178) 0.164 (0.146, 0.182) a0 −0.832 (−0.984, −0.680) – – – CVL 0.092 (0.087, 0.098) – – – No. fish 502 124 132 246 95% confidence interval limits are given in parentheses. Open in new tab Length-based assessment models To assess the status of all Cayman Islands Nassau Grouper FSAs before protections, when only fishery length data were available, we used the Length-Based Spawning Potential Ratio (LBSPR) model developed by Hordyk et al. (2015a,b, 2016). LBSPR is a promising method for populations with limited monitoring data, since SPR can be calculated from life history parameters and length-frequency data under the assumptions of logistic selectivity and maturity. In a comparison of several length-based assessment methods, Chong et al. (2020) showed that LBSPR outperformed others using only one length distribution. We fit LBSPR to pre-protection catch length-frequency data from four FSAs: Little Cayman, Cayman Brac, Grand Cayman, and Pickle Bank (Hordyk et al., 2016). On the two islands for which we had length data following protections, Little Cayman and Cayman Brac, we also fit the LBSPR model to in situ length-frequency data to compare pre- and post-protection status. We used the island-specific parameters from the best-fit hierarchical growth model and the “LBSPR” R package (Table 3; Hordyk, 2017). Table 3. Parameters used to fit the LBSPR model. Parameter . LC . CB . GC . PB . Source . L∞ Asymptotic length (cm) 80.2 80.2 80.2 80.2 This study (Table 2 and Figure 3) K Growth coefficient (1/year) 0.140 0.160 0.164 0.155 This study (Table 2 and Figure 3) M Natural mortality (1/year) 0.276 This study α Length-weight intercept 3.725 × 10−6 This study β Length-weight slope 3.384 This study L50 Length at 50% maturity (cm) 47.4 Sadovy and Eklund (1999) L95 Length at 95% maturity (cm) 55.7 Sadovy and Eklund (1999) CVL Coefficient of variation of L 0.096 This study (Supplementary Figure S8) Bin width (cm) 1 Maximum length (cm) 100 Minimum length (cm) 1 Parameter . LC . CB . GC . PB . Source . L∞ Asymptotic length (cm) 80.2 80.2 80.2 80.2 This study (Table 2 and Figure 3) K Growth coefficient (1/year) 0.140 0.160 0.164 0.155 This study (Table 2 and Figure 3) M Natural mortality (1/year) 0.276 This study α Length-weight intercept 3.725 × 10−6 This study β Length-weight slope 3.384 This study L50 Length at 50% maturity (cm) 47.4 Sadovy and Eklund (1999) L95 Length at 95% maturity (cm) 55.7 Sadovy and Eklund (1999) CVL Coefficient of variation of L 0.096 This study (Supplementary Figure S8) Bin width (cm) 1 Maximum length (cm) 100 Minimum length (cm) 1 Island abbreviations: CB, Cayman Brac; GC, Grand Cayman; LC, Little Cayman; PB, Pickle Bank. Length-weight parameters were fit to the log-linear model, log(Wi) = α + β log(Li), with weight in kg and length in cm. Open in new tab Table 3. Parameters used to fit the LBSPR model. Parameter . LC . CB . GC . PB . Source . L∞ Asymptotic length (cm) 80.2 80.2 80.2 80.2 This study (Table 2 and Figure 3) K Growth coefficient (1/year) 0.140 0.160 0.164 0.155 This study (Table 2 and Figure 3) M Natural mortality (1/year) 0.276 This study α Length-weight intercept 3.725 × 10−6 This study β Length-weight slope 3.384 This study L50 Length at 50% maturity (cm) 47.4 Sadovy and Eklund (1999) L95 Length at 95% maturity (cm) 55.7 Sadovy and Eklund (1999) CVL Coefficient of variation of L 0.096 This study (Supplementary Figure S8) Bin width (cm) 1 Maximum length (cm) 100 Minimum length (cm) 1 Parameter . LC . CB . GC . PB . Source . L∞ Asymptotic length (cm) 80.2 80.2 80.2 80.2 This study (Table 2 and Figure 3) K Growth coefficient (1/year) 0.140 0.160 0.164 0.155 This study (Table 2 and Figure 3) M Natural mortality (1/year) 0.276 This study α Length-weight intercept 3.725 × 10−6 This study β Length-weight slope 3.384 This study L50 Length at 50% maturity (cm) 47.4 Sadovy and Eklund (1999) L95 Length at 95% maturity (cm) 55.7 Sadovy and Eklund (1999) CVL Coefficient of variation of L 0.096 This study (Supplementary Figure S8) Bin width (cm) 1 Maximum length (cm) 100 Minimum length (cm) 1 Island abbreviations: CB, Cayman Brac; GC, Grand Cayman; LC, Little Cayman; PB, Pickle Bank. Length-weight parameters were fit to the log-linear model, log(Wi) = α + β log(Li), with weight in kg and length in cm. Open in new tab LBSPR assumes an equilibrium population state and only considers one length distribution at a time (either 1 year of data or multiple years pooled). Rudd and Thorson (2018) relaxed this equilibrium assumption in their Length-based Integrated Mixed Effects (LIME) model. LIME estimates time-varying recruitment and fishing mortality in a state-space framework and can be run using only length data (as with the LBSPR) or include fishery catch and an index of abundance if they exist. Otherwise, LIME makes the same assumptions as LBSPR. We chose to use LBSPR to assess pre-protection status because only length data were available, and LBSPR has been shown to outperform LIME when fit to only 1 year of length data (Chong et al., 2020). However, LIME was appropriate to assess the Little Cayman FSA after protections because it capitalizes on the available time series of length, abundance estimates, and catch to relax the assumption that the population is at equilibrium. For both LBSPR and LIME, we assumed that the gears used before (hook and line, catch) and after (laser calipers, non-extractive) protections had logistic selectivity and that the selectivity was the same for both gears. These assumptions seemed reasonable because large fish, >70 cm, were well-represented in the Little Cayman catch length data from 2002 (Figure 4), as well as in the sample caught by hook and line for acoustic tagging (12/144 greater than L∞ ⁠, data not shown). In addition, we observed similar proportions of smaller fish, 40–50 cm, in the laser caliper and fishery catch data aggregated across years (Figure 5). Based on the behaviour of fish at the FSA and our data collection protocol, we believe that the probability of measuring a fish with the laser calipers was independent of size, given the fish was mature and at the FSA (Supplementary material). Figure 5. Open in new tabDownload slide Length distributions from Little Cayman Nassau Grouper spawning aggregations (FSAs). Pre-protection data are from fisheries catch (dark shading, 1978–2002), and post-protection data are from in situ laser calipers (light shading, 2003–2019). The size structure recovery following 5 years of no FSA fishing (1996–2000) and subsequent protection is shown by wider distributions from 2002 to 2019. The 2017–2019 distributions are bimodal with wide range, showing a pulse of recruits. The sample sizes (n) for each year are displayed at right. Figure 5. Open in new tabDownload slide Length distributions from Little Cayman Nassau Grouper spawning aggregations (FSAs). Pre-protection data are from fisheries catch (dark shading, 1978–2002), and post-protection data are from in situ laser calipers (light shading, 2003–2019). The size structure recovery following 5 years of no FSA fishing (1996–2000) and subsequent protection is shown by wider distributions from 2002 to 2019. The 2017–2019 distributions are bimodal with wide range, showing a pulse of recruits. The sample sizes (n) for each year are displayed at right. We fit LIME to 17 years of in situ length-frequency data (this study), a 14-year estimate of absolute abundance (numbers of mature fish; Waterhouse et al., 2020), and 2 years of catch data (ca. 2000 fish in 2001 and 1934 fish in 2002; Whaylen et al., 2004). The main purpose for using LIME was to estimate recruitment and depletion (SSB/SSB0) of the Little Cayman population through time following protections, which do not depend on the biomass scale. Still, we included the 2 years of catch data to inform the model about the very high F in 2001–2002. We used the value of M estimated from the catch-curve analysis and conducted sensitivity runs using M ± 0.05/year. We chose to estimate annual recruitment deviations directly without incorporating a stock–recruit relationship, i.e. we set steepness (h) at 1, because the LIME model was not intended to calculate MSY-based reference points or generate catch advice. Nevertheless, we also conducted a sensitivity run using h = 0.7. Finally, we explored the sensitivity of LIME to data weighting parameters—the observation errors for the index, σI ⁠, and catch, σC ⁠, as well as the length composition likelihood. LIME uses the Dirichlet-multinominal distribution by default, which estimates an effective sample size for the length-frequency data that can be lower than the input sample size. We also fit LIME using the multinomial distribution with effective sample sizes calculated using Francis weighting (TA1.8 in Francis, 2011). We used the parameters in Table 4 and the “LIME” R package (Rudd, 2018), starting the model in 1999 to include 4 years of roughly known, extreme variation in F before protection: 2 years in which F was near zero (1999–2000), followed by 2 years of high F (2001–2002). Table 4. Parameters used to fit the LIME models to assess the Little Cayman FSA. Parameter . Model . Source . LIME-fixed-K . LIME integrated . L∞ von Bertalanffy asymptotic length (cm) 80.2 a81.2 (77.9, 84.3) This study (Table 2 and Figure 3) K von Bertalanffy growth coefficient (1/year) 0.140 a0.141 (0.126, 0.156) This study (Table 2 and Figure 3) a0 Age at zero length −0.832 a−0.802 (−0.951, −0.654) This study (Table 2 and Figure 3) M Natural mortality (1/year) 0.276 This study L50 Length at 50% maturity (cm) 47.4 Sadovy and Eklund (1999) L95 Length at 95% maturity (cm) 55.7 Sadovy and Eklund (1999) S50 Length at 50% selectivity (cm) a61.9 (59.4, 64.4) a61.8 (59.2, 64.4) S95 Length at 95% selectivity (cm) a66.1 (59.9, 72.4) a66.0 (59.4, 72.6) α Length-weight intercept 3.725 × 10−6 This study β Length-weight slope 3.384 This study σF Fishing mortality process error 0.3 Default σC Catch observation error 0.2 Sensitivity analysis (Supplementary Figure S4) σI Abundance index observation error 0.175 Sensitivity analysis (Supplementary Figure S4) σR Recruitment process error a0.87 (0.60, 1.25) a0.83 (0.57, 1.21) CVL Growth curve coefficient of variation 0.096 Likelihood profile (Supplementary Figure S8) q Abundance index catchability 1 h Steepness of Beverton–Holt 1 Default Bin width (cm) 1 Maximum length (cm) 100 Minimum length (cm) 1 Parameter . Model . Source . LIME-fixed-K . LIME integrated . L∞ von Bertalanffy asymptotic length (cm) 80.2 a81.2 (77.9, 84.3) This study (Table 2 and Figure 3) K von Bertalanffy growth coefficient (1/year) 0.140 a0.141 (0.126, 0.156) This study (Table 2 and Figure 3) a0 Age at zero length −0.832 a−0.802 (−0.951, −0.654) This study (Table 2 and Figure 3) M Natural mortality (1/year) 0.276 This study L50 Length at 50% maturity (cm) 47.4 Sadovy and Eklund (1999) L95 Length at 95% maturity (cm) 55.7 Sadovy and Eklund (1999) S50 Length at 50% selectivity (cm) a61.9 (59.4, 64.4) a61.8 (59.2, 64.4) S95 Length at 95% selectivity (cm) a66.1 (59.9, 72.4) a66.0 (59.4, 72.6) α Length-weight intercept 3.725 × 10−6 This study β Length-weight slope 3.384 This study σF Fishing mortality process error 0.3 Default σC Catch observation error 0.2 Sensitivity analysis (Supplementary Figure S4) σI Abundance index observation error 0.175 Sensitivity analysis (Supplementary Figure S4) σR Recruitment process error a0.87 (0.60, 1.25) a0.83 (0.57, 1.21) CVL Growth curve coefficient of variation 0.096 Likelihood profile (Supplementary Figure S8) q Abundance index catchability 1 h Steepness of Beverton–Holt 1 Default Bin width (cm) 1 Maximum length (cm) 100 Minimum length (cm) 1 a Estimated in model (MLE with 95% CI in parentheses). Otherwise fixed at specified value. Open in new tab Table 4. Parameters used to fit the LIME models to assess the Little Cayman FSA. Parameter . Model . Source . LIME-fixed-K . LIME integrated . L∞ von Bertalanffy asymptotic length (cm) 80.2 a81.2 (77.9, 84.3) This study (Table 2 and Figure 3) K von Bertalanffy growth coefficient (1/year) 0.140 a0.141 (0.126, 0.156) This study (Table 2 and Figure 3) a0 Age at zero length −0.832 a−0.802 (−0.951, −0.654) This study (Table 2 and Figure 3) M Natural mortality (1/year) 0.276 This study L50 Length at 50% maturity (cm) 47.4 Sadovy and Eklund (1999) L95 Length at 95% maturity (cm) 55.7 Sadovy and Eklund (1999) S50 Length at 50% selectivity (cm) a61.9 (59.4, 64.4) a61.8 (59.2, 64.4) S95 Length at 95% selectivity (cm) a66.1 (59.9, 72.4) a66.0 (59.4, 72.6) α Length-weight intercept 3.725 × 10−6 This study β Length-weight slope 3.384 This study σF Fishing mortality process error 0.3 Default σC Catch observation error 0.2 Sensitivity analysis (Supplementary Figure S4) σI Abundance index observation error 0.175 Sensitivity analysis (Supplementary Figure S4) σR Recruitment process error a0.87 (0.60, 1.25) a0.83 (0.57, 1.21) CVL Growth curve coefficient of variation 0.096 Likelihood profile (Supplementary Figure S8) q Abundance index catchability 1 h Steepness of Beverton–Holt 1 Default Bin width (cm) 1 Maximum length (cm) 100 Minimum length (cm) 1 Parameter . Model . Source . LIME-fixed-K . LIME integrated . L∞ von Bertalanffy asymptotic length (cm) 80.2 a81.2 (77.9, 84.3) This study (Table 2 and Figure 3) K von Bertalanffy growth coefficient (1/year) 0.140 a0.141 (0.126, 0.156) This study (Table 2 and Figure 3) a0 Age at zero length −0.832 a−0.802 (−0.951, −0.654) This study (Table 2 and Figure 3) M Natural mortality (1/year) 0.276 This study L50 Length at 50% maturity (cm) 47.4 Sadovy and Eklund (1999) L95 Length at 95% maturity (cm) 55.7 Sadovy and Eklund (1999) S50 Length at 50% selectivity (cm) a61.9 (59.4, 64.4) a61.8 (59.2, 64.4) S95 Length at 95% selectivity (cm) a66.1 (59.9, 72.4) a66.0 (59.4, 72.6) α Length-weight intercept 3.725 × 10−6 This study β Length-weight slope 3.384 This study σF Fishing mortality process error 0.3 Default σC Catch observation error 0.2 Sensitivity analysis (Supplementary Figure S4) σI Abundance index observation error 0.175 Sensitivity analysis (Supplementary Figure S4) σR Recruitment process error a0.87 (0.60, 1.25) a0.83 (0.57, 1.21) CVL Growth curve coefficient of variation 0.096 Likelihood profile (Supplementary Figure S8) q Abundance index catchability 1 h Steepness of Beverton–Holt 1 Default Bin width (cm) 1 Maximum length (cm) 100 Minimum length (cm) 1 a Estimated in model (MLE with 95% CI in parentheses). Otherwise fixed at specified value. Open in new tab We modified LIME in three ways. First, we integrated the best-fit hierarchical growth model, m2, with LIME to estimate L∞ ⁠, a0 ⁠, island-specific Kj ⁠, and σK2 internally. This “LIME-integrated” model thus explicitly accounts for uncertainty in growth parameters, addressing the commonly cited concern that data-limited assessments assume life history parameters are known without error (Pons et al., 2019, 2020). We compared these results with LIME run with growth parameters fixed at the values estimated externally, as usual, which we refer to as the “LIME-fixed-K” model. Second, LIME includes a penalty on annual F deviations as a random walk, Ft+1 ∼ NFt,σF2 to facilitate convergence. As this was not appropriate for the extreme F fluctuations from 1999 to 2003, we modified LIME to penalize F deviations only beginning in 2004. Last, we specified that the index was in units of number of spawners, as opposed to total (or spawning) biomass, by replacing the predicted spawning biomass in year t, B^t ⁠, with the predicted number of spawners in year t, S^t ⁠, in the equation for the predicted index in year t, i.e. I^t=qS^t instead of I^t=qB^t ⁠, where S^t=∑aN^t,aMata ⁠, N^t,a is the number of age a fish at time t, and Mata is the maturity-at-age a. We then fixed catchability, q, at 1 because the Waterhouse et al. (2020) model directly estimates St in absolute, not relative, numbers. We admitted the possibility that the abundance index could be biased 10% low or high by conducting sensitivity runs using q = 0.9 and q = 1.1. See Table 5 for a summary of the data used to fit each model. The data and code underlying our analysis are available at https://github.com/brianstock/cayman-grouper-assess. Table 5. Summary of the types, years, and locations of data used to fit each model. Model . Estimates . Data type . Result . Quantity . Island . Pre/post- protection . Length . Index . Catch . Age . Growth L∞, a0, and K LC, CB, GC Pre 1988–1992, 2012b 1988–1992 Tables 1 and 2 Figures 3 and 6 Catch curve M PB Pre 2000 Tables 3 and 4 LBSPR SPR LC Pre, post 1978–1995, 2002–2019 Figure 7 SPR CB Pre, post 1990–2000 Figure 7 SPR GC Pre 1988–1997 Figure 7 SPR PB Pre 2000 Figure 7 LIME-fixed-K SSB/SSB0, F, recruitment LC Post 2002–2019 2005–2018 2001–2002 Figure 9 LIME integrated SSB/SSB0, F, recruitment LC Post 2002–2019, 1988–1992a 2005–2018 2001–2002 1988–1992a Figure 8 Model . Estimates . Data type . Result . Quantity . Island . Pre/post- protection . Length . Index . Catch . Age . Growth L∞, a0, and K LC, CB, GC Pre 1988–1992, 2012b 1988–1992 Tables 1 and 2 Figures 3 and 6 Catch curve M PB Pre 2000 Tables 3 and 4 LBSPR SPR LC Pre, post 1978–1995, 2002–2019 Figure 7 SPR CB Pre, post 1990–2000 Figure 7 SPR GC Pre 1988–1997 Figure 7 SPR PB Pre 2000 Figure 7 LIME-fixed-K SSB/SSB0, F, recruitment LC Post 2002–2019 2005–2018 2001–2002 Figure 9 LIME integrated SSB/SSB0, F, recruitment LC Post 2002–2019, 1988–1992a 2005–2018 2001–2002 1988–1992a Figure 8 Island/FSA abbreviations: CB, Cayman Brac; GC, Grand Cayman; LC, Little Cayman; PB, Pickle Bank. “Pre/post” refers to FSA protection status (FSAs were protected in 2003). Prior to protection, length data were collected from fishery catch, and after protection, length data were collected in situ via diver-operated laser calipers. a Age-length data (1988–1992) were used in the LIME-integrated model to fit the growth parameters, but these data did not contribute to the likelihood of the population length composition (model years: 1999–2019). b Lengths of 25 un-aged fish presumed to be 1-year old were recorded in 2012 on Little Cayman from a large recruitment event from 2011 spawning (Camp et al., 2013; Semmens et al., 2013). Open in new tab Table 5. Summary of the types, years, and locations of data used to fit each model. Model . Estimates . Data type . Result . Quantity . Island . Pre/post- protection . Length . Index . Catch . Age . Growth L∞, a0, and K LC, CB, GC Pre 1988–1992, 2012b 1988–1992 Tables 1 and 2 Figures 3 and 6 Catch curve M PB Pre 2000 Tables 3 and 4 LBSPR SPR LC Pre, post 1978–1995, 2002–2019 Figure 7 SPR CB Pre, post 1990–2000 Figure 7 SPR GC Pre 1988–1997 Figure 7 SPR PB Pre 2000 Figure 7 LIME-fixed-K SSB/SSB0, F, recruitment LC Post 2002–2019 2005–2018 2001–2002 Figure 9 LIME integrated SSB/SSB0, F, recruitment LC Post 2002–2019, 1988–1992a 2005–2018 2001–2002 1988–1992a Figure 8 Model . Estimates . Data type . Result . Quantity . Island . Pre/post- protection . Length . Index . Catch . Age . Growth L∞, a0, and K LC, CB, GC Pre 1988–1992, 2012b 1988–1992 Tables 1 and 2 Figures 3 and 6 Catch curve M PB Pre 2000 Tables 3 and 4 LBSPR SPR LC Pre, post 1978–1995, 2002–2019 Figure 7 SPR CB Pre, post 1990–2000 Figure 7 SPR GC Pre 1988–1997 Figure 7 SPR PB Pre 2000 Figure 7 LIME-fixed-K SSB/SSB0, F, recruitment LC Post 2002–2019 2005–2018 2001–2002 Figure 9 LIME integrated SSB/SSB0, F, recruitment LC Post 2002–2019, 1988–1992a 2005–2018 2001–2002 1988–1992a Figure 8 Island/FSA abbreviations: CB, Cayman Brac; GC, Grand Cayman; LC, Little Cayman; PB, Pickle Bank. “Pre/post” refers to FSA protection status (FSAs were protected in 2003). Prior to protection, length data were collected from fishery catch, and after protection, length data were collected in situ via diver-operated laser calipers. a Age-length data (1988–1992) were used in the LIME-integrated model to fit the growth parameters, but these data did not contribute to the likelihood of the population length composition (model years: 1999–2019). b Lengths of 25 un-aged fish presumed to be 1-year old were recorded in 2012 on Little Cayman from a large recruitment event from 2011 spawning (Camp et al., 2013; Semmens et al., 2013). Open in new tab Results Growth and natural mortality In the pre-protection period with fishery data, 1988–1992, Nassau Grouper were smaller at given age on Little Cayman than the other islands (Figure 3). Of the hierarchical growth models, only those that allowed one or fewer parameters to vary by island converged (Table 1). Model m2, with island-specific random effects on the growth coefficient, Kj ⁠, and shared L∞ and a0 ⁠, had the lowest AIC and estimated a lower growth coefficient on Little Cayman (0.140/year, 95% CI: 0.125–0.156) than Cayman Brac (0.160/year, 95% CI: 0.143–0.178) or Grand Cayman (0.164/year, 95% CI: 0.146 − 0.182; Tables 1–2 and Figure 3). We estimated natural mortality as M = 0.276/year (95% CI: 0.17–0.38). This estimate falls within the range reported by previous catch-curve analysis for Nassau Grouper (M = 0.17–0.30/year; Thompson and Munro, 1978) and is very close to estimates from methods recommended in a recent meta-analysis (⁠ MHoenignls = 0.224/year using tmax = 29 years; MPaulynls-T = 0.245/year; Then et al., 2015). Length-frequency analysis The pre-protection catch length distributions from the three historic FSAs on Little Cayman, Cayman Brac, and Grand Cayman were similar, except that lengths from Grand Cayman had a smaller range and were about 3 cm larger on average (Figure 4). Pickle Bank had much larger fish—the average was 10 cm larger than the three main historic Cayman Islands FSAs. Individuals >70 cm were rare at the three historic FSAs, whereas they comprised roughly half of the catch on Pickle Bank (Figure 4). All recent years (2002–2019) of length distributions from the west end Little Cayman FSA had wider range and larger individuals than catches at the historic east end Little Cayman FSA (1978–1995, Figure 5). This was true both for fisheries catch immediately before protection (2002) and for in situ laser caliper data after protection (2003–2019) and reflects the fact that there was no FSA fishing from 1995 to 2001. The 2017–2019 length distributions were bimodal on both Little Cayman and Cayman Brac, with a pulse of small fish 45–55 cm not seen in the other 22 years (Figure 6). On Little Cayman, the modes (local maxima, dashed lines in Figure 6) clearly stepped right from 2017 to 2019 on both islands, as expected if the modes represent growth of a single strong cohort. The Little Cayman modes aligned well with the predicted lengths from the Little Cayman growth curve assuming the cohort was spawned in 2011 (i.e. age 6 in 2017, age 7 in 2018, age 8 in 2019; Figure 6). The 2018 and 2019 modes were slightly lower than the growth curve predictions for age-7 and age-8 fish, by 0.91 and 2.05 cm, respectively. In all years, the modes were larger on Cayman Brac than Little Cayman, which is consistent with the higher estimated growth coefficient on Cayman Brac (Figure 6 and Table 2). The 2017–2018 Cayman Brac modes aligned well with the growth curve predictions under the assumption the cohort was spawned in 2011, although this was not the case for 2019. Figure 6. Open in new tabDownload slide Bimodal length distributions from Little Cayman and Cayman Brac for the years 2017–2019. The modes (local maxima, dashed lines) shift right from 2017 to 2019 on both islands, and the modes are larger on Cayman Brac (CB) than Little Cayman (LC) in all years. Predicted lengths from the Cayman Brac growth curve are: 48.7 cm at age 5 (CB circle), 53.4 cm at age 6 (CB triangle), 57.3 cm at age 7 (CB square), and 60.7 cm at age 8 (CB cross). Predicted lengths from the Little Cayman growth curve are: 44.8 cm at age 5 (LC circle), 49.5 cm at age 6 (LC triangle), 53.5 cm at age 7 (LC square), and 57.0 cm at age 8 (LC cross). Figure 6. Open in new tabDownload slide Bimodal length distributions from Little Cayman and Cayman Brac for the years 2017–2019. The modes (local maxima, dashed lines) shift right from 2017 to 2019 on both islands, and the modes are larger on Cayman Brac (CB) than Little Cayman (LC) in all years. Predicted lengths from the Cayman Brac growth curve are: 48.7 cm at age 5 (CB circle), 53.4 cm at age 6 (CB triangle), 57.3 cm at age 7 (CB square), and 60.7 cm at age 8 (CB cross). Predicted lengths from the Little Cayman growth curve are: 44.8 cm at age 5 (LC circle), 49.5 cm at age 6 (LC triangle), 53.5 cm at age 7 (LC square), and 57.0 cm at age 8 (LC cross). Length-based assessment models Before protections, LBSPR-calculated SPR at the three historic FSAs ranged from 0.45 to 0.53 (Figure 7). The lightly exploited Pickle Bank FSA had higher SPR at 0.73 (95% CI: 0.59–0.87). SPR for the Little Cayman population in 2002, following 6 years of no FSA fishing, was estimated to increase from 0.48 to 1.00. Following 2 years of intense aggregation fishing in 2001 and 2002, the Little Cayman SPR decreased to 0.60 (95% CI: 0.56–0.63) before increasing to 0.94 (95% CI: 0.86–1.00) in recent years. SPR for Cayman Brac also dramatically increased following protection, from 0.50 to 1.00. Figure 7. Open in new tabDownload slide Estimated SPR for Cayman Islands Nassau Grouper spawning aggregations before and after protections implemented in 2003. Pre-protection length data (circles, white background) were collected by sampling FSA catch on Little Cayman, Cayman Brac, Grand Cayman, and Pickle Bank. Although not formally protected until 2003, no FSA fishing occurred on Little Cayman for 6 years between 1995 and 2001. Post-protection length data (squares, grey background) were collected from diver-operated laser calipers on Little Cayman (2003–2019, pooled into 5-year bins) and Cayman Brac (2017–2019). Figure 7. Open in new tabDownload slide Estimated SPR for Cayman Islands Nassau Grouper spawning aggregations before and after protections implemented in 2003. Pre-protection length data (circles, white background) were collected by sampling FSA catch on Little Cayman, Cayman Brac, Grand Cayman, and Pickle Bank. Although not formally protected until 2003, no FSA fishing occurred on Little Cayman for 6 years between 1995 and 2001. Post-protection length data (squares, grey background) were collected from diver-operated laser calipers on Little Cayman (2003–2019, pooled into 5-year bins) and Cayman Brac (2017–2019). In the LIME models for the Little Cayman population from 1999 to 2019, including different amounts of uncertainty in the growth parameters and length composition data led to slightly different but consistent trends in spawning biomass, numbers, mean size, and recruitment (Table 6). Francis weighting gave less weight to the length composition data than the default Dirichlet-multinomial (w = 0.08 compared to w = 0.54, where w is the multiplier for effective sample size). This substantially improved the model fit to both the mean length and the index (Supplementary Figure S5). We found that σI=0.175 and σC=0.20 resulted in the lowest negative log-likelihood and therefore considered the LIME-integrated model with these values and Francis weighting the final model (Supplementary Figure S4). LIME generally fit the abundance index and length-frequency data well, including the bimodal distributions in 2017–2019 (Supplementary Figures S6–S7). LIME models that attempted to estimate the growth curve coefficient of variation, CVL, did not converge. However, a grid search with CVL ranging from 0.08 to 0.11 found that the negative log-likelihood was minimized at CVL = 0.096 and this value was used in the final LIME-integrated model (Supplementary Figure S8). Table 6. LIME model estimates of min/max depletion (SSB/SSB0), 2011 recruitment pulse, and growth parameters for the Little Cayman FSA, assuming different amounts of uncertainty in the growth parameters and length composition data. Estimated quantity . LIME integrated . LIME-fixed-K . Francis weighting . Dirichlet-multinomial . Francis weighting . Dirichlet-multinomial . 2019 SSB/SSB0 0.90 (0.65, 1.25) 0.96 (0.73, 1.25) 1.06 (0.80, 1.40) 1.04 (0.81, 1.36) 2009 SSB/SSB0 0.23 (0.17, 0.32) 0.29 (0.23, 0.38) 0.27 (0.20, 0.36) 0.33 (0.26, 0.42) 2002 SSB/SSB0 0.46 (0.39, 0.54) 0.55 (0.50, 0.61) 0.53 (0.46, 0.61) 0.59 (0.55, 0.63) 2011 Rec/mean Rec 5.3 (0.6, 9.9) 7.8 (3.1, 12.5) 4.4 (1.6, 7.1) 5.6 (3.6, 7.6) L∞ 81.2 (78.1, 84.3) 79.3 (77.3, 81.4) 80.2a 80.2a k 0.141 (0.126, 0.156) 0.146 (0.135, 0.158) 0.140a 0.140a a0 −0.80 (−0.95, −0.65) −0.78 (−0.90, −0.66) −0.83a −0.83a Estimated quantity . LIME integrated . LIME-fixed-K . Francis weighting . Dirichlet-multinomial . Francis weighting . Dirichlet-multinomial . 2019 SSB/SSB0 0.90 (0.65, 1.25) 0.96 (0.73, 1.25) 1.06 (0.80, 1.40) 1.04 (0.81, 1.36) 2009 SSB/SSB0 0.23 (0.17, 0.32) 0.29 (0.23, 0.38) 0.27 (0.20, 0.36) 0.33 (0.26, 0.42) 2002 SSB/SSB0 0.46 (0.39, 0.54) 0.55 (0.50, 0.61) 0.53 (0.46, 0.61) 0.59 (0.55, 0.63) 2011 Rec/mean Rec 5.3 (0.6, 9.9) 7.8 (3.1, 12.5) 4.4 (1.6, 7.1) 5.6 (3.6, 7.6) L∞ 81.2 (78.1, 84.3) 79.3 (77.3, 81.4) 80.2a 80.2a k 0.141 (0.126, 0.156) 0.146 (0.135, 0.158) 0.140a 0.140a a0 −0.80 (−0.95, −0.65) −0.78 (−0.90, −0.66) −0.83a −0.83a LIME-integrated included uncertainty in growth parameters while LIME-fixed-K did not. Francis data weighting included more uncertainty in the length composition data than using the Dirichlet-multinomial likelihood. Values are MLE with 95% CI in parentheses. a Fixed in model, not estimated. Open in new tab Table 6. LIME model estimates of min/max depletion (SSB/SSB0), 2011 recruitment pulse, and growth parameters for the Little Cayman FSA, assuming different amounts of uncertainty in the growth parameters and length composition data. Estimated quantity . LIME integrated . LIME-fixed-K . Francis weighting . Dirichlet-multinomial . Francis weighting . Dirichlet-multinomial . 2019 SSB/SSB0 0.90 (0.65, 1.25) 0.96 (0.73, 1.25) 1.06 (0.80, 1.40) 1.04 (0.81, 1.36) 2009 SSB/SSB0 0.23 (0.17, 0.32) 0.29 (0.23, 0.38) 0.27 (0.20, 0.36) 0.33 (0.26, 0.42) 2002 SSB/SSB0 0.46 (0.39, 0.54) 0.55 (0.50, 0.61) 0.53 (0.46, 0.61) 0.59 (0.55, 0.63) 2011 Rec/mean Rec 5.3 (0.6, 9.9) 7.8 (3.1, 12.5) 4.4 (1.6, 7.1) 5.6 (3.6, 7.6) L∞ 81.2 (78.1, 84.3) 79.3 (77.3, 81.4) 80.2a 80.2a k 0.141 (0.126, 0.156) 0.146 (0.135, 0.158) 0.140a 0.140a a0 −0.80 (−0.95, −0.65) −0.78 (−0.90, −0.66) −0.83a −0.83a Estimated quantity . LIME integrated . LIME-fixed-K . Francis weighting . Dirichlet-multinomial . Francis weighting . Dirichlet-multinomial . 2019 SSB/SSB0 0.90 (0.65, 1.25) 0.96 (0.73, 1.25) 1.06 (0.80, 1.40) 1.04 (0.81, 1.36) 2009 SSB/SSB0 0.23 (0.17, 0.32) 0.29 (0.23, 0.38) 0.27 (0.20, 0.36) 0.33 (0.26, 0.42) 2002 SSB/SSB0 0.46 (0.39, 0.54) 0.55 (0.50, 0.61) 0.53 (0.46, 0.61) 0.59 (0.55, 0.63) 2011 Rec/mean Rec 5.3 (0.6, 9.9) 7.8 (3.1, 12.5) 4.4 (1.6, 7.1) 5.6 (3.6, 7.6) L∞ 81.2 (78.1, 84.3) 79.3 (77.3, 81.4) 80.2a 80.2a k 0.141 (0.126, 0.156) 0.146 (0.135, 0.158) 0.140a 0.140a a0 −0.80 (−0.95, −0.65) −0.78 (−0.90, −0.66) −0.83a −0.83a LIME-integrated included uncertainty in growth parameters while LIME-fixed-K did not. Francis data weighting included more uncertainty in the length composition data than using the Dirichlet-multinomial likelihood. Values are MLE with 95% CI in parentheses. a Fixed in model, not estimated. Open in new tab On Little Cayman, the LIME-integrated model estimated a two-step decline in spawning biomass, driven by both decreasing size and numbers of fish during two periods of high fishing mortality (2001–2002 and 2006–2009, Figure 8). The 2 years of heavy FSA fishing immediately prior to protection, 2001–2002, reduced SSB by 54% (95% CI: 46–61%). Depletion (SSB/SSB0) reached a low of 0.23 (95% CI: 0.17–0.32) in 2009 and then dramatically increased to 0.90 (95% CI: 0.65–1.25) in 2019 (Figure 8d and Table 6). The number of spawners similarly reached a minimum in 2008, increased slightly from 2008 to 2012 due in part to slightly higher recruitment in the 2003–2006 period, and then increased dramatically from 2014 to 2018 (Figure 8b and c). Mean size went through five alternative periods of decrease (2000–2002, 2005–2010, 2016–2018) and increase (2003–2005, 2010–2016) corresponding to pulses in F and recruitment (Figure 8a). Figure 8. Open in new tabDownload slide LIME-integrated model output for the Little Cayman Nassau Grouper FSA: (a) mean length (cm), (b) recruitment (age-0), (c) abundance (number of spawners), (d) depletion (SSB/SSB0), (e) fishing mortality, and (f) selectivity. Black triangles show the input data with 95% confidence intervals. Green points, lines, and shading depict the MLEs and 95% confidence intervals. In (a) and (c), model estimates for years without data are distinguished by green lines without points. In (b) and (e), recruitment deviations and fishing mortality were fixed at 0 for years without points. Figure 8. Open in new tabDownload slide LIME-integrated model output for the Little Cayman Nassau Grouper FSA: (a) mean length (cm), (b) recruitment (age-0), (c) abundance (number of spawners), (d) depletion (SSB/SSB0), (e) fishing mortality, and (f) selectivity. Black triangles show the input data with 95% confidence intervals. Green points, lines, and shading depict the MLEs and 95% confidence intervals. In (a) and (c), model estimates for years without data are distinguished by green lines without points. In (b) and (e), recruitment deviations and fishing mortality were fixed at 0 for years without points. We estimated a large recruitment pulse from 2011 spawning on Little Cayman that was 5.3 times average (95% CI: 0.6–9.9, Figure 8b). This was robust to assuming different amounts of uncertainty in the growth parameters and length composition data (Figure 9), as well as values of M (Supplementary Figure S9 and Supplementary Table S1) and steepness (Supplementary Table S2). Under these various parameterizations, the magnitude of the 2011 recruitment pulse varied from 4.4–7.8 times average recruitment (Table 6 and Supplementary Tables S1 and S2). Including uncertainty in the growth parameters and downweighting the length composition data increased the uncertainty in the timing of the recruitment pulse as well as the magnitude (wider confidence intervals for 2010 and 2012 recruitment in Figure 9). Figure 9. Open in new tabDownload slide Estimated recruitment for the Little Cayman Nassau Grouper FSA from models with a range of uncertainty and data weighting. The LIME-integrated model (a, c) incorporates uncertainty in growth parameters by estimating them internally, whereas the LIME-fixed-K model (b, d) fixes growth parameters at values estimated externally. Francis weighting (a, b) gives less weight to the length data (w = 0.08) than the Dirichlet-multinomial (c, d; w = 0.54). Green points, lines, and shading depict the MLEs and 95% confidence intervals. Recruitment deviations are fixed at 0 in years without points (2016–2019) because the data are uninformative (age at first capture >4). Figure 9. Open in new tabDownload slide Estimated recruitment for the Little Cayman Nassau Grouper FSA from models with a range of uncertainty and data weighting. The LIME-integrated model (a, c) incorporates uncertainty in growth parameters by estimating them internally, whereas the LIME-fixed-K model (b, d) fixes growth parameters at values estimated externally. Francis weighting (a, b) gives less weight to the length data (w = 0.08) than the Dirichlet-multinomial (c, d; w = 0.54). Green points, lines, and shading depict the MLEs and 95% confidence intervals. Recruitment deviations are fixed at 0 in years without points (2016–2019) because the data are uninformative (age at first capture >4). Discussion On Little Cayman and Cayman Brac, Nassau Grouper population size structure and SPR clearly recovered following 16 years of FSA protections. Using a time series of in situ length data is an effective method for monitoring protected FSAs, where all mature fish aggregate at high density and can be efficiently measured. Both the laser caliper and stereo camera systems were accurate enough to detect bimodal length distributions on Little Cayman and Cayman Brac in 2017–2019, which implied recruitment of a very strong year class spawned in 2011. On Little Cayman, spawning biomass was reduced by 54% in 2 years of intense FSA fishing and then took 16 years to recover to near pre-exploitation levels (Figure 8d). This recovery was largely driven by the one strong year of recruitment in 2011. These results attest to the value of monitoring FSA size structure in addition to numbers of fish—combining length and abundance data in an assessment allow for a more complete picture of population status and can attribute increases or decreases to changes in fishing mortality vs. recruitment. The differences in growth coefficients (Table 2 and Figure 3) and 2017–2019 length distributions (Figure 6) between Little Cayman and Cayman Brac strongly suggest synchrony in the 2011 large recruitment event on both islands. There is, nevertheless, an alternate hypothesis for the difference in length modes between the two islands: fish grew at the same rate on both islands but are 1 year older on Cayman Brac, i.e. a large recruitment event took place on Cayman Brac in 2011 followed by the same on Little Cayman in 2012. This possibility is not supported by the above model outputs, and the presence of numerous 1–1.5-year-old (12–23 cm) juveniles on Little Cayman in February–July 2012 (Camp et al., 2013; Semmens et al., 2013) provides further evidence against two major recruitment events. Nearly zero juveniles were sighted in all years 2004–2017 except for 2012 (Semmens et al., 2013). Furthermore, it is not unreasonable that Little Cayman and Cayman Brac would show strong recruitment in the same year, because they are only separated by 8 km and late-stage larvae are easily capable of swimming this distance against currents (Leis et al., 2009). The most likely explanation is that recruitment on both islands was paired. If true, direct and indirect evidence indicates that growth coefficients were consistently lower on Little Cayman during two separate time periods, 1987–1992 and 2011–2019. It is not clear why Nassau Grouper growth would be slower on Little Cayman than Cayman Brac or Grand Cayman. The biomass of Nassau Grouper on Little Cayman increased nearly fourfold from 2008 to 2019 and was much higher than on Cayman Brac despite similar habitat area (Supplementary Figure S10, McCoy, 2019). Thus, growth may plausibly be slower on Little Cayman in recent years if there are density-dependent growth effects. This could also help explain why the Little Cayman 2018–2019 length modes were 0.9–2.0 cm smaller than the growth curve predictions for age-7 and age-8 fish (Figure 6) since the age-length data used to fit the model are from a period with lower population density. However, while slower growth at higher density is consistent with density-dependent growth, it is not direct evidence, and future work would be necessary to evaluate the hypothesis. Furthermore, density dependence is only one possible mechanism underlying the slower growth on Little Cayman vs. Cayman Brac in recent years, and it does not explain why growth was also slower on Little Cayman compared to the other islands in 1987–1992. Cayman Brac has relatively more “spur and groove” and patch reef habitat, and differences in benthos may be related to prey density and growth rate (McCoy, 2019). Instead of intraspecific competition, Nassau Grouper on Little Cayman may have less access to food because there are more large snappers and groupers in general, and thus, interspecific competition may be greater. Prey may be harder to capture on Little Cayman, requiring Nassau Grouper to allocate more energy to active metabolism and less to growth. Behaviour may differ—when DOE scientists attempted in-water capture to acoustically tag Nassau Grouper in situ, they noted that Cayman Brac fish were markedly warier and more skittish than their counterparts on Little Cayman (B. Johnson, pers. comm.). Nassau Grouper on Little Cayman may have matured earlier than their counterparts on Cayman Brac, which would result in an energy reallocation from somatic growth to reproductive growth. Of these alternative explanations, we can only rule out different ages at maturity because the 2011 cohort was absent from the 2016 length distribution on Little Cayman. Still, there are many conceivable mechanisms behind the difference in estimated growth coefficients, and future work could test for these possibilities. Another obvious question is: What was special about conditions in 2011 that led to a major recruitment pulse on both islands? Like many reef fish, Nassau Grouper are benthic but are pelagic broadcast spawners, and successful recruitment may largely depend on favourable currents bringing larvae close to suitable reef habitat. The prevailing current around the Cayman Islands flows east-northeast to west-southwest, but the mean current is weak and looping eddies that retain water for months are common (Richardson, 2005). Thus, self-recruitment within the Caymans is a very likely possibility (Colin et al., 1987; Heppell et al., 2009, 2011; Colin, 2012b; Sadovy de Mitcheson and Colin, 2012). Future work could model larval dispersal from Little Cayman using archived remote sensing data and compare 2011 against low recruitment years. Alternatively, strong recruitment in 2011 could have been related to abundant prey or fewer predators at critical space and time scales for larval survival (e.g. Cushing’s match–mismatch hypothesis, with a strong “match” in 2011; Cushing, 1990). It is also possible for physical forcing to positively affect larval dispersal and survival simultaneously (Checkley et al., 1988). While intriguing, these possibilities are difficult to test in hindsight. Nassau Grouper at different locations throughout the Caribbean appear to spawn during months when the average temperature is around 26°C (Table 2 in Tucker et al., 1993). Sea surface temperatures near Little Cayman were indeed relatively cool in 2011, ∼26°C (Supplementary Figure S11). However, temperatures were also lower in 2005, 2006, and 2009, and no large recruitment events were observed in these years. Water temperature around 26°C may be a necessary but insufficient condition for recruitment success. Future work could investigate a possible temperature effect on recruitment, which may be acting as a proxy for effects related to currents, prey, or predators. Aggregation status and management Out of 17 years of monitoring on Little Cayman, we saw only one year of strong recruitment. This is not surprising since Nassau Grouper are periodic strategists (Winemiller and Rose, 1992): long-lived and highly fecund, capable of withstanding years of recruitment failure sporadically punctuated by large successful spawning events. Given the dramatic decline of Nassau Grouper throughout the Caribbean, it is possible that external recruitment (i.e. from other FSAs) is more sporadic now than in the past. When FSAs were far greater in size and number, the probability of any population receiving larvae from a different FSA was likely higher, and therefore, recruitment less variable. Now, with fewer and smaller FSAs, the remaining FSAs may be more dependent on self-recruitment. Whether or not this is true, we observed high recruitment variability for the Little Cayman FSA that had no correlation with spawning stock biomass. This is important information for management as it implies that long recovery timelines for this species should be expected. On Little Cayman, where biomass was reduced by 53% in 2 years of intense FSA fishing, the recovery to pre-exploited levels took 16 years and was largely driven by the one strong year of recruitment in 2011. Waterhouse et al. (2020) reported that numbers of Nassau Grouper on Cayman Brac have likely increased since protection but tempered their conclusions due to sparse observations. Two of our results strengthen confidence that the Cayman Brac population has, in fact, increased. First, pre- and post-protection length data used to estimate SPR show a substantial improvement in population status between 1990–2000 and 2017–2019 (Figure 7). Second, the bimodal 2017–2019 length distributions imply that a large recruitment pulse occurred on Cayman Brac as well as Little Cayman (Figure 6). Thus, while we cannot map the Cayman Brac population trajectory in fine detail as we did for Little Cayman, our length data do support the increase in abundance described by Waterhouse et al. (2020). The LBSPR model estimated SPR ranging from 0.45 to 0.53 at the three historic FSAs in the 1980–1990s (Figure 7). These SPR estimates were above 0.40, which is often recommended as a risk-averse reference point in cases where the stock–recruit relationship is not estimable (Clark, 1993, 2002; Mace, 1994; Hordyk et al., 2015b; Rudd and Thorson, 2018). Yet, Nassau Grouper populations in the Cayman Islands subsequently declined; total catch, CPUE, and mean size all decreased to very low levels by 2001 (Bush et al., 2006), and fishermen decided to stop fishing at the Little Cayman east end FSA by 1995. Particularly concerning is that the Grand Cayman FSA, with an estimated SPR of 0.53 (95% CI: 0.48–0.57) in the 1988–1997 period, has shown no sign of recovery despite 16 years of protection. Recovery on Grand Cayman was a reasonable expectation because (i) SPR on all three islands was similar before protections, (ii) substantial increases in biomass and SPR have occurred on both Little Cayman and Cayman Brac over the same time period with the same management measures (i.e. the FSA closures sufficiently reduced F on the other islands), and (iii) Grand Cayman is much larger than either Little Cayman or Cayman Brac (roughly 2.5 times available reef habitat; McCoy, 2019), so should have a higher carrying capacity and potential for rebuilding. However, the few post-protection observations we have from Grand Cayman suggest that the population remains depleted (Waterhouse et al., 2020). All the above strongly suggest that the Grand Cayman population was not being fished at sustainable levels before 2001 (with SPR estimated at 0.53), was very depleted by 2001, and remains depleted. Together with the species’ history of exploitation (range-wide dramatic declines in catch, disappearance of FSAs, and failure of lost FSAs to re-form once protected), this indicates that SPR <0.6 may be an unwise reference point for managing Nassau Grouper spawning aggregation fisheries. It is plausible that a sustainable SPR for Nassau Grouper could be higher than 0.40. Clark (2002) and Brooks et al. (2010) demonstrated that the appropriate SPR depends on the slope of the stock–recruit curve at low stock size, and that for less resilient species (i.e. lower stock–recruit steepness) SPR in the range 0.60–0.86 could be warranted. Zhou et al. (2020) modelled SPR at MSY (SPRMSY) as a function of life history parameters for 185 stocks and found that nearly two-thirds require SPRMSY > 40%. In addition, the species’ life history may not follow the assumptions underlying the typical SPR reference point guidelines. First, the guidelines are derived from Beverton–Holt or Ricker stock–recruit relationships that do not admit the possibility of an Allee effect (i.e. depensation, lower recruits per spawner at low stock size; Brooks et al., 2010). The spawning aggregation behaviour of Nassau Grouper may well be a “strong” Allee effect mechanism, whereby FSAs no longer form at population sizes (or densities) below a threshold and few, if any, recruits are produced (Courchamp et al., 2008; Sadovy de Mitcheson, 2016). If such a threshold exists, Nassau Grouper stock sizes need to be kept above it. Second, Nassau Grouper recruitment may be more driven by environmental stochasticity, including variable larval dispersal, and only weakly related to stock size. This is the case for many, if not most, managed fish stocks in the world (Szuwalski et al., 2015). Of course, both these mechanisms may act in concert, such that recruitment is very low or zero at low stock sizes when FSAs cease to form, and then unrelated to stock size above a threshold. This highlights the need for fisheries assessment and management tools to be adapted for aggregating species’ life history (Sadovy de Mitcheson, 2016). Conclusion The Cayman Islands government should be commended for acting quickly to protect the Nassau Grouper FSAs. While roughly half of the Little Cayman spawning biomass was harvested in the 2 years before protection, the remaining individuals continued to form a spawning aggregation. Had the Caymanians not acted quickly, then recovery, had it occurred at all, would likely have been even more protracted than it was; Nassau Grouper recovery is almost nonexistent at sites throughout the Caribbean where aggregating behaviour has ceased. The recovery of these historic sites may depend on getting a pulse of larvae from a healthy FSA—it is possible this occurred on Cayman Brac in 2011. The Nassau Grouper FSA on Little Cayman is currently the largest spawning aggregation known for the species, and the status of the Cayman Brac FSA is markedly improved. FSA protections are increasingly common in the Caribbean, and region-wide recovery of Nassau Grouper depends on population responses to these protections. In the Cayman Islands, scientific monitoring following temporary FSA protections bolstered the necessary political will to extend these protections and make them permanent through legislation (no take during spawning months, bag and slot limits away from FSAs in the rest of the year; Cayman Islands Cabinet, 2016; Waterhouse et al., 2020). We found that time series of in situ length data is an especially effective method for assessing protected FSAs and was even able to detect the recruitment of strong year classes and differences in growth between islands. The methods demonstrated here are useful for assessing FSAs and lend themselves to efforts aimed at managing sustainable reef fisheries. Supplementary data Supplementary material is available at the ICESJMS online version of the manuscript. Data availability The data and code underlying our analysis are available at https://github.com/brianstock/cayman-grouper-assess. Acknowledgements We thank the hard-working CI-DoE staff who have provided years of field support: Bradley Johnson, Ivan Montieth, Keith Neale, Robert Walton, James Gibb, Cody Panton, Michael Guderian, Kevin Jackson, and Chris Dixon. Many REEF staff and volunteer divers helped collect laser caliper video, including Leslie Whaylen, Steve Gittings, Hal Peterson, Todd Bohannon, Cody Panton, Tammi Warrender, and Josh Stewart. Peter Hillenbrand (Southern Cross Club), Jason Belport (Little Cayman Beach Resort), and Cayman Airways generously provided accommodations and logistical support. We thank Patrick L. Colin and Douglas Y. Shapiro for access to the original data of 1978 from Little Cayman. Their work was done under permit from the Cayman Islands government. The manuscript was vastly improved by constructive comments from three anonymous reviewers. Funding BCS received support from the Cooperative Institute for Marine Ecosystems and Climate (CIMEC) and the National Science Foundation Graduate Research Fellowship under Grant No. DGE-1144086. ICD was funded by the Scripps Institution of Oceanography Director’s Office in support of the 2017 Scripps Undergraduate Research Fellowship (SURF) REU program (NSF-OCE 1659793). References Beddington J. R. , Agnew D. J. , Clark C. W. 2007 . Current problems in the management of marine fisheries . Science , 316 : 1713 – 1716 . Google Scholar Crossref Search ADS PubMed WorldCat Beets J. , Friedlander A. 1999 . Evaluation of a conservation strategy: a spawning aggregation closure for red hind, Epinephelus guttatus, in the U.S. Virgin Islands . Environmental Biology of Fishes , 55 : 91 – 98 . Google Scholar Crossref Search ADS WorldCat Brooks E. N. , Powers J. E. , Cortes E. 2010 . Analytical reference points for age-structured models: application to data-poor fisheries . ICES Journal of Marine Science , 67 : 165 – 175 . Google Scholar Crossref Search ADS WorldCat Burnham K. P. , Anderson D. R. 2002 . Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach . Springer , New York . 488 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Bush P. , Ebanks-Petrie G. . 1994 . The Cayman Islands Nassau Grouper study—a progress report. In Proceedings of the 43rd Annual Gulf and Caribbean Fisheries Institute, pp. 387 – 388 . Charleston, SC, USA. Bush P. G. , Lane E. D. , Ebanks G. C. 1996 . Validation of Ageing Technique for Nassau Grouper (Epinephelus striatus) in the Cayman Islands. In Biology, Fisheries and Culture of Tropical Snappers and Groupers. Proceedings EPOMEX/ICLARM International Workshop on Tropical Snappers and Groupers , pp. 150 – 157 . Ed. by Arrequin-Sanchez F. A. , Munro J. L. , Balgos M. C. , Pauly D. . October 1993. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Bush P. G. , Lane E. D. , Ebanks-Petrie G. C. , Luke K. , Johnson B. , McCoy C. , Bothwell J. , Parsons E. 2006 . The Nassau Grouper spawning aggregation fishery of the Cayman Islands—an historical and management perspective. In Proceedings of the 57th Annual Gulf and Caribbean Fisheries Institute, pp. 515 – 524 . St. Petersburg, FL, USA. Camp E. F. , Lohr K. E. , Barry S. C. , Bush P. G. , Jacoby C. A. , Manfrino C. 2013 . Microhabitat associations of late juvenile Nassau Grouper (Epinephelus striatus) off Little Cayman, BWI . Bulletin of Marine Science , 89 : 571 – 581 . Google Scholar Crossref Search ADS WorldCat Cayman Islands Cabinet . 2016 . The National Conservation (General) Regulations. http://www.gov.ky/portal/pls/portal/docs/1/12326595.PDF (last accessed 21 December 2020). Chapman D. , Robson D. 1960 . The analysis of a catch curve . Biometrics , 16 : 354 – 368 . Google Scholar Crossref Search ADS WorldCat Checkley D. M. Jr , Raman S. , Maillet G. L. , Mason K. M. 1988 . Winter storm effects on the spawning and larval drift of a pelagic fish . Nature , 335 : 346 – 348 . Google Scholar Crossref Search ADS WorldCat Cheung W. , Sadovy Y. , Braynen M. , Gittens L. 2013 . Are the last remaining Nassau grouper Epinephelus striatus fisheries sustainable? Status quo in the Bahamas . Endangered Species Research , 20 : 27 – 39 . Google Scholar Crossref Search ADS WorldCat Chong L. , Mildenberger T. K. , Rudd M. B. , Taylor M. H. , Cope J. M. , Branch T. A. , Wolff M. , et al. 2020 . Performance evaluation of data-limited, length-based stock assessment methods . ICES Journal of Marine Science , 77 : 97 – 108 . 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 W. G. 1993 . The effect of recruitment variability on the choice of a target level of spawning biomass per recruit. In Proceedings of the International Symposium on Management Strategies for Exploited Fish Populations, pp. 233 – 246 . University of Alaska Fairbanks. Clark W. G. 2002 . F35% revisited ten years later . North American Journal of Fisheries Management , 22 : 251 – 257 . Google Scholar Crossref Search ADS WorldCat Claro R. , de Mitcheson Y. S. , Lindeman K. C. , García-Cagide A. R. 2009 . Historical analysis of Cuban commercial fishing effort and the effects of management interventions on important reef fishes from 1960–2005 . Fisheries Research , 99 : 7 – 16 . Google Scholar Crossref Search ADS WorldCat Claro R. , Lindeman K. C. 2003 . Spawning aggregation sites of snapper and grouper species (Lutjanidae and Serranidae) on the Insular Shelf of Cuba . Gulf and Caribbean Research , 14 : 91 – 106 . Google Scholar Crossref Search ADS WorldCat Colin P. L. 2012a . Studying and monitoring aggregating species. In Reef Fish Spawning Aggregations: Biology, Research and Management , pp. 285 – 329 . Ed. by Sadovy de Mitcheson Y. , Colin P. L. . Springer Netherlands , Dordrecht . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Colin P. L. 2012b . Timing and location of aggregation and spawning in reef fishes. In Reef Fish Spawning Aggregations: Biology, Research and Management , pp. 117 – 158 . Ed. by Sadovy de Mitcheson Y. , Colin P. L. . Springer Netherlands , Dordrecht . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Colin P. L. , Shapiro D. Y. , Weiler D. 1987 . Aspects of the reproduction of two groupers, Epinephelus guttatus and E. striatus in the West Indies . Bulletin of Marine Science , 40 : 220 – 230 . Google Scholar OpenURL Placeholder Text WorldCat Courchamp F. , Berec L. , Gascoigne J. 2008 . Allee Effects in Ecology and Conservation . Oxford University Press . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Cushing D. 1990 . Plankton production and year-class strength in fish populations: an update of the match/mismatch hypothesis . Advances in Marine Biology , 26 : 249 – 293 . Google Scholar Crossref Search ADS WorldCat Ehrhardt N. M. , Deleveaux V. K. W. 2007 . The Bahamas’ Nassau grouper (Epinephelus striatus) fishery—two assessment methods applied to a data—deficient coastal population . Fisheries Research , 87 : 17 – 27 . Google Scholar Crossref Search ADS WorldCat Francis R. I. C. C. 2011 . Data weighting in statistical fisheries stock assessment models . Canadian Journal of Fisheries and Aquatic Sciences , 68 : 1124 – 1138 . Google Scholar Crossref Search ADS WorldCat Grüss A. , Robinson J. , Heppell S. S. , Heppell S. A. , Semmens B. X. 2014 . Conservation and fisheries effects of spawning aggregation marine protected areas: what we know, where we should go, and what we need to get there . ICES Journal of Marine Science , 71 : 1515 – 1534 . Google Scholar Crossref Search ADS WorldCat Hamilton R. J. , Potuku T. , Montambault J. R. 2011 . Community-based conservation results in the recovery of reef fish spawning aggregations in the Coral Triangle . Biological Conservation , 144 : 1850 – 1858 . Google Scholar Crossref Search ADS WorldCat Helser T. E. , Lai H.-L. 2004 . A Bayesian hierarchical meta-analysis of fish growth: with an example for North American largemouth bass, Micropterus salmoides . Ecological Modelling , 178 : 399 – 416 . Google Scholar Crossref Search ADS WorldCat Heppell S. A. , Semmens B. X. , Pattengill-Semmens C. V. , Bush P. G. , Johnson B. C. , McCoy C. M. , Paris C. et al. 2009 . Tracking potential larval dispersal patterns from Nassau grouper aggregation sites: Evidence for local retention and the “importance of place”. In Proceedings of the 61st Gulf and Caribbean Fisheries Institute , pp. 325 – 327 . Gosier, Guadeloupe , French West Indies . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Heppell S. , Semmens B. X. , Pattengill-Semmens C. V. , Bush P. G. , Johnson B. C. , McCoy C. M. , Gibb J. et al. 2011 . Oceanographic patterns associated with Nassau grouper aggregation spawn timing: shifts in surface currents on the nights of peak spawning. In Proceedings of the 63rd Gulf and Caribbean Fisheries Institute, pp. 152 – 154 . San Juan, Puerto Rico. Heppell S. A. , Semmens B. X. , Archer S. K. , Pattengill-Semmens C. V. , Bush P. G. , McCoy C. M. , Heppell S. S. , et al. 2012 . Documenting recovery of a spawning aggregation through size frequency analysis from underwater laser calipers measurements . Biological Conservation , 155 : 119 – 127 . Google Scholar Crossref Search ADS WorldCat Hilborn R. , Amoroso R. O. , Anderson C. M. , Baum J. K. , Branch T. A. , Costello C. , de Moor C. L. et al. 2020 . Effective fisheries management instrumental in improving fish stock status . Proceedings of the National Academy of Sciences of the United States of America , 117 : 2218 – 2224 . Google Scholar Crossref Search ADS PubMed WorldCat Hobday A. J. , Smith A. D. M. , Stobutzki I. C. , Bulman C. , Daley R. , Dambacher J. M. , Deng R. A. , et al. 2011 . Ecological risk assessment for the effects of fishing . Fisheries Research , 108 : 372 – 384 . Google Scholar Crossref Search ADS WorldCat Hordyk A. 2017 . LBSPR: Length-Based Spawning Potential Ratio. v0.1.2. https://CRAN.R-project.org/package=LBSPR (last accessed 21 December 2020). Hordyk A. , Ono K. , Sainsbury K. , Loneragan N. , Prince J. 2015a . Some explorations of the life history ratios to describe length composition, spawning-per-recruit, and the spawning potential ratio . ICES Journal of Marine Science , 72 : 204 – 216 . Google Scholar Crossref Search ADS WorldCat Hordyk A. , Ono K. , Valencia S. , Loneragan N. , Prince J. 2015b . A novel length-based empirical estimation method of spawning potential ratio (SPR), and tests of its performance, for small-scale, data-poor fisheries . ICES Journal of Marine Science , 72 : 217 – 231 . Google Scholar Crossref Search ADS WorldCat Hordyk A. R. , Ono K. , Prince J. D. , Walters C. J. 2016 . A simple length-structured model based on life history ratios and incorporating size-dependent selectivity: application to spawning potential ratios for data-poor stocks . Canadian Journal of Fisheries and Aquatic Sciences , 73 : 1787 – 1799 . Google Scholar Crossref Search ADS WorldCat Johannes R. E. 1998 . The case for data-less marine resource management: examples from tropical nearshore finfisheries . Trends in Ecology & Evolution , 13 : 243 – 246 . Google Scholar Crossref Search ADS PubMed WorldCat Kimura D. K. 1980 . Likelihood methods for the von Bertalanffy growth curve . Fishery Bulletin , 77 : 765 – 776 . Google Scholar OpenURL Placeholder Text WorldCat Kristensen K. , Nielsen A. , Berg C. , Skaug H. , Bell B. M. 2016 . TMB: automatic differentiation and Laplace approximation . Journal of Statistical Software , 70 : 1 – 21 . Google Scholar Crossref Search ADS WorldCat Leis J. , Hay A. , Howarth G. 2009 . Ontogeny of in situ behaviours relevant to dispersal and population connectivity in larvae of coral-reef fishes . Marine Ecology Progress Series , 379 : 163 – 179 . Google Scholar Crossref Search ADS WorldCat Luckhurst B. E. , Trott T. M. 2009 . Seasonally-Closed Spawning Aggregation Sites for Red Hind (Epinephelus guttatus): Bermuda’s Experience over 30 Years (1974–2003). In Proceedings of the 61st Gulf and Caribbean Fisheries Institute, pp. 331–336. Gosier, Guadeloupe, French West Indies. Mace P. M. 1994 . Relationships between common biological reference points used as thresholds and targets of fisheries management strategies . Canadian Journal of Fisheries and Aquatic Sciences , 51 : 110 – 122 . Google Scholar Crossref Search ADS WorldCat McCoy C. M. 2019 . Cayman Islands Marine Protected Areas, Their Status, Effects and Future . Bangor University , Bangor, Wales, UK . 236 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Nemeth R. 2005 . Population characteristics of a recovering US Virgin Islands red hind spawning aggregation following protection . Marine Ecology Progress Series , 286 : 81 – 97 . Google Scholar Crossref Search ADS PubMed WorldCat NMFS . 2016 . Endangered and threatened wildlife and plants: final listing determination on the proposal to list the Nassau Grouper as threatened under the Endangered Species Act. Technical Report 50 CFR Part 223, U.S. Dept. Commerce, National Oceanic and Atmospheric Administration, National Marine Fisheries Service. Ogle D. H. 2016 . Individual growth. In Introductory Fisheries Analyses with R , pp. 221 – 249 . Chapman and Hall/CRC , New York . 10.1201/9781315371986 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Crossref Ogle D. H. , Wheeler P. , Dinno A. 2018 . FSA: Fisheries Stock Analysis. R package version 0.8.22. https://github.com/droglenc/FSA (last accessed 21 December 2020). Patrick W. S. , Spencer P. , Link J. , Cope J. , Field J. , Kobayashi D. , Lawson P. , et al. 2010 . Using productivity and susceptibility indices to assess the vulnerability of United States fish stocks to overfishing . Fishery Bulletin , 108 : 305 – 322 . Google Scholar OpenURL Placeholder Text WorldCat Pons M. , Kell L. , Rudd M. B. , Cope J. M. , Frédou F. L. 2019 . Performance of length-based data-limited methods in a multifleet context: application to small tunas, mackerels, and bonitos in the Atlantic Ocean . ICES Journal of Marine Science , 76 : 960 – 973 . Google Scholar Crossref Search ADS WorldCat Pons M. , Cope J. M. , Kell L. T. 2020 . Comparing performance of catch-based and length-based stock assessment methods in data-limited fisheries . Canadian Journal of Fisheries and Aquatic Sciences , 77 : 1026 – 1012 . Google Scholar Crossref Search ADS WorldCat Robinson J. , Graham N. A. J. , Cinner J. E. , Almany G. R. , Waldie P. 2015 . Fish and fisher behaviour influence the vulnerability of groupers (Epinephelidae) to fishing at a multispecies spawning aggregation site . Coral Reefs , 34 : 371 – 382 . Google Scholar Crossref Search ADS WorldCat Rose G. A. , Kulka D. W. 1999 . Hyperaggregation of fish and fisheries: how catch-per-unit-effort increased as the Northern cod (Gadus morhua) declined . Canadian Journal of Fisheries and Aquatic Science 56 : 10 . Google Scholar Crossref Search ADS WorldCat Rudd M. B. 2018 . Length-Based Integrated Mixed Effects Model. v2.0.7. https://github.com/merrillrudd/LIME (last accessed 21 December 2020). Rudd M. B. , Thorson J. T. 2018 . Accounting for variable recruitment and fishing mortality in length-based stock assessments for data-limited fisheries . Canadian Journal of Fisheries and Aquatic Sciences , 75 : 1019 – 1035 . Google Scholar Crossref Search ADS WorldCat Russ G. R. , Alcala A. C. 2004 . Marine reserves: long-term protection is required for full recovery of predatory fish populations . Oecologia , 138 : 622 – 627 . Google Scholar Crossref Search ADS PubMed WorldCat Russell M. W. , Luckhurst B. E. , Lindeman K. C. 2012 . Management of spawning aggregations. In Reef Fish Spawning Aggregations: Biology, Research and Management , pp. 371 – 404 . Ed.by Sadovy de Mitcheson Y. and Colin P. L. . Springer Netherlands, Dordrecht . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Sadovy Y. , Eklund A.-M. 1999 . Synopsis of Biological Data on the Nassau Grouper, Epinephelus striatus (Bloch, 1792), and the Jewfish, E. itajara (Lichtenstein, 1822). Technical Report NOAA Technical Report NMFS 146. A Technical Report of the Fishery Bulletin. FAO Fisheries Synopsis 157 . U.S. Department of Commerce. Sadovy Y. , Domeier M. 2005 . Are aggregation-fisheries sustainable? Reef fish fisheries as a case study . Coral Reefs , 24 : 254 – 262 . Google Scholar Crossref Search ADS WorldCat Sadovy Y. , Aguilar-Perera A. , Sosa-Cordero E. 2018 . Epinephelus striatus . The IUCN Red List of Threatened Species , 2018 : e.T7862A46909843. 10.2305/IUCN.UK .2018-2.RLTS.T7862A46909843.en (last accessed 21 December 2020). Google Scholar OpenURL Placeholder Text WorldCat Crossref Sadovy de Mitcheson Y. 2016 . Mainstreaming fish spawning aggregations into fishery management calls for a precautionary approach . BioScience , 66 : 295 – 306 . Google Scholar Crossref Search ADS WorldCat Sadovy de Mitcheson Y. , Colin P. L. 2012 . Species case studies. In Reef Fish Spawning Aggregations: Biology, Research and Management , pp. 405 – 565 . Ed. by Sadovy de Mitcheson Y. and Colin P. L. . Springer Netherlands, Dordrecht . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Sadovy de Mitcheson Y. , Cornish A. , Domeier M. , Colin P. L. , Russell M. , Lindeman K. C. 2008 . A global baseline for spawning aggregations of reef fishes . Conservation Biology , 22 : 1233 – 1244 . Google Scholar Crossref Search ADS PubMed WorldCat Sadovy de Mitcheson Y. , Erisman B. 2012 . Fishery and biological implications of fishing spawning aggregations, and the social and economic importance of aggregating fishes. In Reef Fish Spawning Aggregations: Biology, Research and Management , pp. 225 – 284 . Ed. by Sadovy de Mitcheson Y. and Colin P. L. . Springer Netherlands, Dordrecht . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Sadovy de Mitcheson Y. , Prada Triana M. C. , Azueta J. O. , Lindeman K. C. 2020 . Regional Fish Spawning Aggregation Fishery Management Plan: Focus on Nassau Grouper and Mutton Snapper . Western Central Atlantic Fishery Commission , San Juan, Puerto Rico . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Sala E. , Ballesteros E. , Starr R. M. 2001 . Rapid decline of Nassau Grouper spawning aggregations in Belize: fishery management and conservation needs . Fisheries , 26 : 23 – 30 . Google Scholar Crossref Search ADS WorldCat Semmens B. X. , Luke K. E. , Bush P. G. , Pattengill-Semmens C. , Johnson B. , McCoy C. , Heppell S. 2007 . Investigating the reproductive migration and spatial ecology of Nassau Grouper (Epinephelus striatus) on Little Cayman Island using acoustic tags—an overview. In Proceedings of the 58th Annual Gulf and Caribbean Fisheries Institute, pp. 199 – 206 . San Andres, Columbia. Semmens B. X. , Bush P. , Heppell S. , Johnson B. , McCoy C. , Pattengill-Semmens C. , Heppell S. S. 2009 . The spatial ecology of a remnant Nassau Grouper stock on Cayman Brac, Cayman Islands. In Proceedings of the 61st Gulf and Caribbean Fisheries Institute . Gosier , Guadeloupe , French West Indies. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Semmens B. , Heppell S. , Bush P. , Pattengill-Semmens C. , McCoy C. , Johnson B. 2013 . Evidence of episodic mass recruitment of Nassau Grouper in the Cayman Islands. In Proceedings of the 65th Gulf and Caribbean Fisheries Institute, p. 199 . Santa Marta, Columbia. Sherman K. , Dahlgren C. , Stevens J. , Tyler C. 2016 . Integrating population biology into conservation management for endangered Nassau grouper Epinephelus striatus . Marine Ecology Progress Series , 554 : 263 – 280 . Google Scholar Crossref Search ADS WorldCat Smith C. L. 1972 . A spawning aggregation of Nassau Grouper, Epinephelus striatus (Bloch) . Transactions of the American Fisheries Society , 101 : 257 – 261 . Google Scholar Crossref Search ADS WorldCat Smith M. W. , Then A. Y. , Wor C. , Ralph G. , Pollock K. H. , Hoenig J. M. 2012 . Recommendations for catch-curve analysis . North American Journal of Fisheries Management , 32 : 956 – 967 . Google Scholar Crossref Search ADS WorldCat Szuwalski C. S. , Vert-Pre K. A. , Punt A. E. , Branch T. A. , Hilborn R. 2015 . Examining common assumptions about recruitment: a meta-analysis of recruitment dynamics for worldwide marine fisheries . Fish and Fisheries , 16 : 633 – 648 . Google Scholar Crossref Search ADS WorldCat Then A. Y. , Hoenig J. M. , Hall N. G. , Hewitt D. A. 2015 . Evaluating the predictive performance of empirical estimators of natural mortality rate using information on over 200 fish species . ICES Journal of Marine Science , 72 : 82 – 92 . Google Scholar Crossref Search ADS WorldCat Thompson R. , Munro J. L. 1978 . Aspects of the biology and ecology of Caribbean reef fishes: Serranidae (hinds and groupers) . Journal of Fish Biology , 12 : 115 – 146 . Google Scholar Crossref Search ADS WorldCat Tucker J. , Bush P. G. , Slaybaugh S. T. 1993 . Reproductive patterns of Cayman Islands Nassau Grouper (Epinephelus striatus) populations . Bulletin of Marine Science , 52 : 961 – 969 . Google Scholar OpenURL Placeholder Text WorldCat Waterhouse L. , Heppell S. A. , Pattengill-Semmens C. V. , McCoy C. , Bush P. , Johnson B. C. , Semmens B. X. 2020 . Recovery of critically endangered Nassau grouper (Epinephelus striatus) in the Cayman Islands following targeted conservation actions . Proceedings of the National Academy of Sciences of the United States of America , 117 : 1587 – 1589 . Google Scholar Crossref Search ADS PubMed WorldCat Whaylen L. , Bush P. , Johnson B. , Luke K. E. , McCoy C. , Semmens B. , Boardman M. 2007 . Aggregation dynamics and lessons learned from five years of monitoring at a Nassau grouper (Epinephelus striatus) spawning aggregation in Little Cayman, Cayman Islands, BWI. In Proceedings of the 59th Annual Gulf and Caribbean Fisheries Institute, pp. 479 – 488 . Belize City, Belize. Whaylen L. , Pattengill-Semmens C. V. , Semmens B. X. , Bush P. G. , Boardman M. R. 2004 . Observations of a Nassau grouper, Epinephelus striatus, spawning aggregation site in Little Cayman, Cayman Islands, including multi-species spawning information . Environmental Biology of Fishes , 70 : 305 – 313 . Google Scholar Crossref Search ADS WorldCat Winemiller K. O. , Rose K. A. 1992 . Patterns of life-history diversification in North American fishes: implications for population regulation . Canadian Journal of Fisheries and Aquatic Sciences , 49 : 2196 – 2218 . Google Scholar Crossref Search ADS WorldCat Worm B. , Hilborn R. , Baum J. K. , Branch T. A. , Collie J. S. , Costello C. , Fogarty M. J. , et al. 2009 . Rebuilding global fisheries . Science , 325 : 578 – 585 . Google Scholar Crossref Search ADS PubMed WorldCat Zhou S. , Punt A. E. , Lei Y. , Deng R. A. , Hoyle S. D. 2020 . Identifying spawner biomass per-recruit reference points from life-history parameters . Fish and Fisheries , 21 : 760 – 714 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2021. All rights reserved. For permissions, please email: [email protected] This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Bekkevold, Dorte; Piper, Adam; Campbell, Ronald; Rippon, Philip; Wright, Ros M; Crundwell, Charles; Wysujack, Klaus; Stevens, Jamie R; King, R Andrew; Aarestrup, Kim; Maltby, Alistair

doi: 10.1093/icesjms/fsaa240

Jang, Ho Geun; Yamazaki, Satoshi; Kiyama, Shoichi; Higashida, Keisaku; Tinch, Dugald

doi: 10.1093/icesjms/fsab001pmid: N/A

Abstract We studied how local environmental and socio-economic factors impact fish supply and the price outcome of small-scale fisheries. We exploited day-to-day variations in sea surface temperature (SST) and cross-sectional differences in market distance and age of fishers to measure consumers’ responsiveness to price changes in a small-scale octopus fishery in Japan. Using the estimated demand parameters, we quantified the economic consequences of local socio-environmental factors in terms of changes in fishing revenue and consumer surplus. We found that increasing SST and an aging workforce increase the octopus supply and that consumers are responsive to price changes due to such supply shocks. Our results suggest that increasing SST and an aging workforce have positive net effects on fishing revenue and consumer surplus within the fishery. The octopus fishery provides a temporary source of income in the off-season of other species, smoothing the seasonal income variation of elderly fishers in the community. Introduction How much do local weather conditions influence the economic outcome of small-scale fisheries? Do improving the accessibility of fish markets and moderating population aging make the local small-scale fishing community better off? This article quantifies how fish supply and the price outcome of a small-scale fishery are impacted by seasonal variation in sea surface temperature (SST) and cross-sectional differences in market distance and fishers’ age. Small-scale fisheries are closely tied to the development of coastal communities around the world by providing food, labour, and income to the local population (Béné et al., 2007, 2016; FAO, 2015). Over 80% of global fishing activity is conducted by non-motorized or small-motorized vessels, and 90% of employment in capture fisheries is attributed to small-scale fisheries (Watson, 2017; FAO, 2018; Rousseau et al., 2019). However, the sustainable management of these fisheries is complex, as it requires considering both anthropogenic and environmental factors at multiple scales in time and space. Among environmental factors, long-term interannual environmental changes, such as global warming and El Nino Southern Oscillation, influence the growth, distribution, and migration of fish (Lehodey et al., 2006). Likewise, local environmental factors, such as seasonal weather and oceanographic variations, directly impact fishery productivity (Angrist et al., 2000; Wang et al., 2010). Although the root causes of these environmental changes are beyond the control of individuals, they have significant social and economic implications for small-scale fishing communities (Madin et al., 2012; Gattuso et al., 2015; Free et al., 2019). In addition to environmental factors, the productivity of small-scale fisheries is sensitive to the socio-economic structure of local fishing communities. For example, broad literature suggests that demographic changes, such as population aging and decreasing fertility, negatively impact economic productivity and growth (Maestas et al., 2016; Katagiri, 2018; Aksoy et al., 2019). The fishery literature, however, suggests that the net impact of such demographic changes is ambiguous. This is because an aging workforce in a fishery does not necessarily decrease labour productivity, as fishing skills and experience often increase with the age of fishers (Kirkley et al., 1998; Pascoe and Coglan, 2002; Tingley et al., 2005). Similarly, previous research suggests that accessibility to local fish markets plays a crucial role in determining the behaviour of small-scale fishers, who typically operate in remote areas with limited market access (Béné, 2009; Schmitt and Kramer, 2009). On one hand, improved market access may expand income-generating opportunities within the local community and incentivise fishers to increase productive efficiency. At the same time, however, increasing market access has been observed in conjunction with increased fishing pressure and declining fish stocks (Cinner and McClanahan, 2006; Cinner et al., 2013; Stevens et al., 2014). The net economic impact of improved market access on the local fishing community is, therefore, ambiguous and relies on empirical evidence. To assess the economic impact of changes in market conditions and other socio-environmental factors for the fishery, information regarding consumers’ responsiveness to price changes (i.e. price elasticity of demand) is necessary. Either an upward or downward shift in the productivity of fisheries will result in a change in the price of fish. This means that whether fishing communities would be better or worse off as a result of supply shocks depends not only on productivity change in the fishery but also on how consumers respond to the corresponding price change. There are numerous studies that estimate the price elasticity of demand for fish, and the estimates vary significantly across different countries, regions, fish products, types of consumers, and estimation methods. See, for example Asche et al. (2007), Gallet (2009), and Andreyeva et al. (2010), who provided reviews and meta-analyses of studies that estimate the demand for fish. Until now, however, there has been limited research on the market demand for small-scale fisheries and the sensitivity of fish prices to external factors that impact fish supply. This article aims to contribute to this literature by addressing the following two research questions: (i) What is the impact of local weather conditions and socio-economic factors on the fish supply and the price outcome of a small-scale fishery? (ii) What are the economic consequences of these local socio-environmental factors in terms of changes in fishing revenue and consumer surplus? To achieve this, we used an instrumental variable approach, with daily records of catch and price data for a small-scale octopus fishery in Maizuru Bay, Japan. Following discussions with local fishing operators and management officers in the Kyoto Fishery Cooperatives, day-to-day variation in SST and cross-sectional variation in market distance and age of fishers were selected as three factors (i.e. instruments) that may impact the amount of octopus caught by individual fishers each day (and, hence, the price). The insights gained from this case study are highly relevant to small-scale fisheries elsewhere. The management of small-scale fisheries is complicated by the fact that production is not solely driven by internal factors within a fishery but also depends on factors outside the domain of the fishery. These external factors include inherent variation in natural environments, as well as changes in market conditions and the socio-cultural characteristics of fishing communities (Cinner et al., 2005; Aswani and Sabetian, 2010; Hall, 2011; Yamazaki et al., 2018). Using an instrumental variable approach, we first estimated the expected change in the price of octopus due to the external supply shocks of local socio-environmental changes. We then identified the consumers’ responsiveness to these price changes. We used the estimated demand parameters (i.e. own-price elasticity of demand) to evaluate the extent to which fishing revenue and consumer surplus are affected when local socio-environmental factors influence the octopus supply. While the instrumental variable estimation of demand parameters is common in economics (Angrist and Krueger, 2001; Greene, 2018), it is used less frequently for the estimation of fish demand compared to other methods, such as the almost ideal demand system model (Wellman, 1992). The few exceptions include Angrist et al. (2000), Graddy (2006), and Tokunaga (2018), who used weather conditions at sea and fluctuations in landings and auctioned volume as exogeneous supply shocks to estimate the demand in two of the world’s largest wholesale seafood markets: the Fulton Fish Market in New York (United States) and the Tsukiji Market in Tokyo (Japan). Background and data Maizuru Bay octopus fishery Maizuru Bay, with a bay area of 23 km2 and a maximum depth of 30 m, is in the northern Kyoto prefecture of Japan (Figure 1). Octopus fishing is predominantly conducted on an individual basis by fishers who are granted fishing rights (gyogyo ken) in Maizuru. Like many other fishing communities in Japan (Yamashita, 2012; Kamoey, 2015), the octopus fishing population of Maizuru confronts the issue of an aging workforce. In 2017, the median age of octopus fishers was 61 years. This means that about 50% were over the age at which the national pension can be granted (OECD, 2019). While the main fishing methods used in the fishery are traps and diving with a spear, fishers’ choice of fishing methods is also associated with their age. Young fishers tend to use a combination of traps and diving, whereas elderly fishers typically use traps only. Figure 1. Open in new tabDownload slide Location of the Maizuru Bay octopus fishery. The green point is the fish market, and the red points are the residential postcodes of octopus fishers in Maizuru. The mean distance to the market is 4.9 km, and the minimum and maximum distances are 0.2 and 10.5 km, respectively. The dark shaded area is where fishers have access with their fishing rights, while fishers typically catch octopus within the bay. Figure 1. Open in new tabDownload slide Location of the Maizuru Bay octopus fishery. The green point is the fish market, and the red points are the residential postcodes of octopus fishers in Maizuru. The mean distance to the market is 4.9 km, and the minimum and maximum distances are 0.2 and 10.5 km, respectively. The dark shaded area is where fishers have access with their fishing rights, while fishers typically catch octopus within the bay. The fishery is managed by a community-based resource management system that has no restrictions on fishing gear or total allowable catch; however, entry to the fishery is limited to local commercial fishers who are granted fishing rights. While fishers in Maizuru Bay generally target multiple species, there is no technical difference in the availability of alternative species among fishers, as they all have access to the same bay and operate under the same regulatory system. Individual differences in harvesting behaviours may reflect the difference in economic needs and, hence, the opportunity cost of targeting different species. Octopus caught in Maizuru Bay are sold to registered middlemen through auctions held at the local fish market, which is managed by the Kyoto Fishery Cooperatives. While they are considered a single product in the market, there are two octopus species caught in Maizuru Bay: common octopus (Octopus vulgaris) and East Asian common octopus (Octopus sinensis). The quality of some pelagic and demersal fish sometimes depends on fishing method (McConnell and Strand, 2000; Asche and Guillen, 2012; Sogn-Grundvåg et al., 2013; Lee, 2014). However, there is no empirical or anecdotal evidence that shows the quality or consumers’ preferences of octopus are influenced by the fishing methods in the current study area. Interviews with local fishers and management officers in the Kyoto Fishery Cooperatives also indicate that there are price differences based on size, and the possibility exists that more experienced fishers are able to catch high-value octopus. This size-dependent pricing and targeting behaviour has also been observed in fisheries elsewhere (Zimmermann and Heino, 2013; Asche et al., 2015). Data sources We retrieved data from three different sources. First, daily data on catch (in kg) and price (in JPY/kg) for all fishers registered in Maizuru in 2017 were taken from a database administrated by the Kyoto Fishery Cooperatives. We used daily data on catch and price in 2017 at the level of individual fishers, meaning that our analysis relies on both time and cross-sectional variations (i.e. repeated cross-sectional data). We used data in 2017 because it was the latest data available for a full year at the time of the study (2018). However, while we note that one full year of daily observations is sufficient to capture seasonal variations within a year, this does not consider long-term interannual changes. We classified fishers based on the species caught, and only those who caught octopus (i.e. octopus fishers) in 2017 were included in the final dataset. Additionally, daily SST data for Maizuru Bay were retrieved from the Maizuru Fisheries Research Station of Kyoto University. We also collected information about the octopus fishers’ age, main fishing method and residential postcode from the Kyoto Fishery Cooperatives. The postcode was used to construct a proxy for market distance between each fisher’s residence and the Maizuru fish market. Specifically, market distance was measured by calculating the great-circle distance between a fisher’s residence and the Maizuru fish market. The physical distance between a fisher’s residence (postcode) and the Maizuru fish market is a reasonable proxy for market distance because (i) there are multiple ports around the bay, and fishers use the port nearest to their residence, (ii) ground transportation is typically used to transport fish, and (iii) most (if not all) catches are sold to the single market. However, we also note that the great-circle distance measures the shortest distance between two points along the surface of a sphere and is not the exact travel path along the actual road. Table 1 shows the descriptive statistics of all the variables used in this paper. Table 1. Descriptive statistics. Variable . Unit . Mean . Median . Min . Max . SD . Obs. . Catch Kg/day 10.3 7.0 0.3 80.0 10.1 619 Price JPY/kg 1,023 983 50 3,300 544 619 Market distance km 4.9 5.5 0.2 10.5 2.5 619 Sea surface temperature °C 19.8 19.6 5.5 30.7 6.6 610 Elderly fisher Dummy 0.74 1 0 1 0.44 619 Variable . Unit . Mean . Median . Min . Max . SD . Obs. . Catch Kg/day 10.3 7.0 0.3 80.0 10.1 619 Price JPY/kg 1,023 983 50 3,300 544 619 Market distance km 4.9 5.5 0.2 10.5 2.5 619 Sea surface temperature °C 19.8 19.6 5.5 30.7 6.6 610 Elderly fisher Dummy 0.74 1 0 1 0.44 619 Note: Number of positive octopus catches (fishing days) = 237. Total number of fishers in 2017 = 40. Mean number of fishers per day = 2.84. Elderly fisher = 1 if ≥62 years old, 0 otherwise. Open in new tab Table 1. Descriptive statistics. Variable . Unit . Mean . Median . Min . Max . SD . Obs. . Catch Kg/day 10.3 7.0 0.3 80.0 10.1 619 Price JPY/kg 1,023 983 50 3,300 544 619 Market distance km 4.9 5.5 0.2 10.5 2.5 619 Sea surface temperature °C 19.8 19.6 5.5 30.7 6.6 610 Elderly fisher Dummy 0.74 1 0 1 0.44 619 Variable . Unit . Mean . Median . Min . Max . SD . Obs. . Catch Kg/day 10.3 7.0 0.3 80.0 10.1 619 Price JPY/kg 1,023 983 50 3,300 544 619 Market distance km 4.9 5.5 0.2 10.5 2.5 619 Sea surface temperature °C 19.8 19.6 5.5 30.7 6.6 610 Elderly fisher Dummy 0.74 1 0 1 0.44 619 Note: Number of positive octopus catches (fishing days) = 237. Total number of fishers in 2017 = 40. Mean number of fishers per day = 2.84. Elderly fisher = 1 if ≥62 years old, 0 otherwise. Open in new tab Data description In 2017, there were 40 octopus fishers in total, with 237 days of total fishing operations. While octopus fishing was conducted throughout the year, the number of octopus fishers each day was small (Figure 2). The mean and median numbers of octopus fishers per day were 2.84 and 2, respectively, and fishers on average caught octopus for 11 days in the year. Of the 40 octopus fishers in the year, only seven caught octopus for more than 30 days. SST varied between 5.5 and 30.7°C throughout the year. There is a moderate positive correlation between the number of octopus fishers and SST, with a correlation coefficient of 0.11, while there seems to be an inverted U-shaped relationship between the number of fishers and SST. Specifically, there was only one fisher who caught octopus when SST reached either the maximum temperature of 30.7°C or the minimum temperature of 5.5°C, whereas the highest number of fishers (i.e. 10 fishers) was observed on the day when SST was intermediate at 15°C in May. Figure 2. Open in new tabDownload slide The number of octopus fishers per day (grey bar) and the SST (red solid line) in Maizuru Bay, Japan. The mean number of fishers per day is 2.84 (dashed line). Figure 2. Open in new tabDownload slide The number of octopus fishers per day (grey bar) and the SST (red solid line) in Maizuru Bay, Japan. The mean number of fishers per day is 2.84 (dashed line). The mean daily catch of octopus per fisher was relatively stable at 10.3 kg throughout the year (Figure 3). However, the size of interpersonal variation in the daily catch varied seasonally. For example, there was marked interpersonal variation in the daily catch from mid-May to mid-September when the SST was relatively high. By contrast, the daily catch variation among fishers was relatively small in other seasons when the SST was relatively low. The average price of octopus was more volatile than the catch over the year. In summer, the price and interpersonal variation in price were relatively low, whereas the highest price range was observed in spring when the catch was low compared to other seasons within the year. Figure 3. Open in new tabDownload slide Time series of daily catch (above), price (below), and SST (red line). The box plot shows raw data on the daily distribution of catches and prices. A black solid line shows a cubic fit of each variable, and the shaded area around the black solid line represents the 95% confidence interval. Figure 3. Open in new tabDownload slide Time series of daily catch (above), price (below), and SST (red line). The box plot shows raw data on the daily distribution of catches and prices. A black solid line shows a cubic fit of each variable, and the shaded area around the black solid line represents the 95% confidence interval. Estimation of demand function and welfare effects Demand function Generally, there is a negative correlation between the price and quantity of octopus traded in the market, with a correlation coefficient of −0.18. However, it is important to note that each data point reflects the intersection of demand and supply curves; thus, it is impossible to estimate the demand function based on the observed relationship between quantities and prices per se. In other words, OLS estimation of the demand function using quantity and price data is biased due to the endogeneity of price. To this end, we used the instrumental variables approach of Angrist et al. (2000), who provided a useful means to estimate the demand function for octopus in the Maizuru fish market. Using valid instruments as supply shocks, this approach addresses the endogeneity problems embedded in observed price and quantity data that reflect equilibrium outcomes of the market clearing process. Let qit and pit denote the log-transformed quantity and price for fisher i in day t. We consider the supply and demand functions given as: qitS=α1+β1pitS+zit′γ+pt−′θ1+uitS,(1) qitD=α2+β2pitD+pt−′θ2+uitD,(2) where the superscripts S and D denote supply and demand, αi, βi, θi (i = 1, 2), and γ are parameters and uitS and uitD are the error terms for the supply and demand function, respectively. In the supply function, zit is a vector of the supply shock variables, which are used as instruments in the estimation of the demand parameters. We also included the log-transformed average prices of octopus and oysters on the last market day (⁠ pt− ⁠) as control variables, which may influence both the octopus demand and supply. These variables control for the possibility that fishers and consumers may substitute between species and that their decision in day t depends on what happened on the last market day. At the equilibrium, we have qitS=qitD=qit* and pitS=pitD=pit* ⁠, which are the quantities and prices observed in the data, respectively. Solving (1) and (2) for the equilibrium yields the price equation, only in terms of the exogeneous supply shock variables: pit*=δ1+zit′δ2+pt−′δ3+uit,(3) where δ1=(α1−α2)/(β2−β1) ⁠, δ2=γ/(β2−β1) ⁠, δ3=(θ1−θ2)/(β2−β1) ⁠, and uit=(uitS−uitD)/(β2−β1) ⁠. Equation (3) is the reduced form equation, which is derived from the intersection (equilibrium condition) of the demand and supply functions (1) and (2). Equation (3) was estimated first to obtain the predicted values of the price, p^it* (i.e. first-stage regression). We then estimated the parameter in (2) by OLS while replacing pit with p^it* (i.e. demand function estimation). For the consistent estimation of the demand parameters, the instruments, zit, need to satisfy two assumptions: (i) validity, in which the instruments are uncorrelated with the error term, i.e. E(zituit) = 0; and (ii) relevance, in which the instruments are correlated with the price, pit, i.e. δ2 ≠ 0 in (3). One possible reason for the violation of the first assumption is that the selected instruments affect not only the fish supply but also the demand. In this case, the instruments are endogenous, and the estimators are inconsistent. The second assumption requires that the selected instruments have significant explanatory power for price variation. Following discussions with Maizuru octopus fishers and staff at the Kyoto Fishery Cooperatives, we first selected three candidate instruments that could possibly shift the supply curve (i.e. supply shocks) without affecting the octopus demand, namely market distance, SST, and fishers’ age. Whether these instruments satisfy the assumptions is also tested statistically in the “Supply shocks and price elasticity of demand” section. Supply shock instruments Market distance was selected because it reflects the opportunity cost of selling octopus due to the time and fuel required for transportation, and it is expected to affect fishers’ decisions regarding whether they catch octopus on the day (Cinner et al., 2013). This variable unlikely reflects the price advantage of those who can transport fish faster to the market, as all the catches are normally transported prior to auction commencement. Thus, those who get to the market first do not necessarily receive a higher price than others. Likewise, SST was selected because changes in weather conditions have previously been shown as a relevant shock for fish supply (Angrist et al., 2000). While we initially collected various weather data in the Maizuru area, including SST, wind speed, precipitation and humidity, we considered SST as a candidate instrument here because other weather variables affect not only the supply of octopus but could also influence consumers’ purchasing behaviours. Furthermore, the growth and distribution of octopus are found to be affected by SST (Ramos et al., 2018), making SST a valid candidate as a day-to-day supply shifter. Fishers’ age may also affect the daily amount of octopus caught by each fisher or their ability to target the size of octopus that is highly valued in the market, as it reflects the years of experience, skills in handling the equipment and locating concentrations of octopus. Among the octopus fishers in Maizuru Bay, the age of fishers also reflects the targeting behaviour of each fisher (i.e. composition of species in the total fishery production), as the opportunity cost of fishing may be different between different age groups. This point is further discussed in the “Revenue portfolio for young and elderly fishers” section. We used the age of 62 years (i.e. the earnings-related pension age) to divide the sample into young and elderly fishers (OECD, 2019). Given that the number of instruments is more than the endogenous variable (i.e. price), we used the two-stage least squares (2SLS) estimator. Consumer surplus Using the estimated demand parameters in (2), we examined the economic impact of supply shocks in terms of a change in consumer surplus due to supply shocks. Evaluating a change in consumer surplus helps understand the welfare implications of changes in the octopus supply. Typically, a change in consumer surplus depends on the price elasticity of demand and supply. However, we selected instrumental variables in such a way that they only affect the fish supply but not the demand. We validated this assumption (i.e. overidentifying restrictions) using Sargan’s (1958) and Hansen’s (1982)J-test (see “Supply shocks and price elasticity of demand” section). This assumption is critically important because the current paper focuses on studying how local socio-environmental factors impact fish supply and the price outcome of small-scale fisheries. When this assumption is satisfied, we are not required to consider a demand shift (i.e. price elasticity of supply) to calculate the change in consumer surplus. Let Qd denote the quantity demanded in kg, Pd denote the corresponding price in JPY, and Pt− denote a vector consisting of the average prices of octopus and oysters on the last market day in JPY. Given the estimated parameters α^2 ⁠, β^2 ⁠, and θ^2 in (2), the demand curve for octopus is given as: Qd=exp (α^2)Pdβ^2Pt−θ^2.(4) For a price increase from Pbase to Pnew, the resulting change in consumer surplus is calculated as: ΔCS=∫PnewPbaseQd(Pd)dPd.(5) Given that data on prices and quantities are observed for each fisher on each day, the consumer surplus in (5) is measured in JPY per fisher per day. We estimated a yearly change in the total consumer surplus by multiplying ΔCS by the mean number of fishers per day (i.e. three fishers) and the total number of fishing days in the sample period (i.e. 237 days). Results Supply shocks and price elasticity of demand In the first-stage regression of (3), the estimated coefficient for market distance is positive, suggesting that the greater the distance from a fisher’s residence to the market, the higher the octopus price for the fisher (Table 2). This is consistent with our expectation that market distance is associated with the opportunity cost of fishing. The coefficients for SST and elderly fisher are negative, suggesting that an increase in these variables shifts the supply curve downward. More specifically, a 1°C increase in SST is associated with a 1.6–2.0% decrease in the price of octopus ceteris paribus because of a downward shift of the supply curve. For all the models considered, the weak instrument test (Stock and Yogo, 2005) shows that the instruments are jointly relevant (δ2 ≠ 0), indicating that they have significant explanatory power for price variation. The Durbin–Wu–Hansen test also rejected the null of exogeneity [i.e. H0: E(pituit) = 0], suggesting that OLS estimation of the demand function is inconsistent. Table 2. First-stage regression results. Variable . Model 1 (all) . Model 2 (selected) . Model 3 (market) . Model 4 (SST) . Model 5 (elderly fisher) . Market distance (km) 0.119*** (0.010) 0.128*** (0.010) SST (°C) −0.018***(0.004) −0.020***(0.005) −0.016***(0.005) Elderly fisher (dummy) −0.234***(0.037) −0.406***(0.041) −0.381***(0.041) ln(last market day price, octopus) 0.170***(0.057) 0.166**(0.070) 0.232***(0.054) 0.152**(0.074) 0.253***(0.064) ln(last market day price, oyster) 0.040 (0.066) 0.120 (0.077) −0.108**(0.049) 0.115 (0.081) −0.057 (0.057) Constant 5.287***(0.528) 5.572***(0.622) 5.202***(0.532) 5.314***(0.653) 5.654***(0.618) Durbin–Wu–Hausman test (F-statistic) 29.681*** 54.876*** 6.485** 15.623*** 28.632*** Weak instrument test (F-statistic) 75.4*** 59.1*** 155.4*** 8.59*** 84.816*** Observations 607 607 616 607 616 Variable . Model 1 (all) . Model 2 (selected) . Model 3 (market) . Model 4 (SST) . Model 5 (elderly fisher) . Market distance (km) 0.119*** (0.010) 0.128*** (0.010) SST (°C) −0.018***(0.004) −0.020***(0.005) −0.016***(0.005) Elderly fisher (dummy) −0.234***(0.037) −0.406***(0.041) −0.381***(0.041) ln(last market day price, octopus) 0.170***(0.057) 0.166**(0.070) 0.232***(0.054) 0.152**(0.074) 0.253***(0.064) ln(last market day price, oyster) 0.040 (0.066) 0.120 (0.077) −0.108**(0.049) 0.115 (0.081) −0.057 (0.057) Constant 5.287***(0.528) 5.572***(0.622) 5.202***(0.532) 5.314***(0.653) 5.654***(0.618) Durbin–Wu–Hausman test (F-statistic) 29.681*** 54.876*** 6.485** 15.623*** 28.632*** Weak instrument test (F-statistic) 75.4*** 59.1*** 155.4*** 8.59*** 84.816*** Observations 607 607 616 607 616 Note: This table reports the first-stage regression results of (3) and the test statistics for the Durbin–Wu–Hausman test of endogeneity and the weak instrument test. The dependent variable is ln(Price). Heteroskedasticity-robust standard errors are in parenthesis. ** p < 0.05, *** p < 0.01. Open in new tab Table 2. First-stage regression results. Variable . Model 1 (all) . Model 2 (selected) . Model 3 (market) . Model 4 (SST) . Model 5 (elderly fisher) . Market distance (km) 0.119*** (0.010) 0.128*** (0.010) SST (°C) −0.018***(0.004) −0.020***(0.005) −0.016***(0.005) Elderly fisher (dummy) −0.234***(0.037) −0.406***(0.041) −0.381***(0.041) ln(last market day price, octopus) 0.170***(0.057) 0.166**(0.070) 0.232***(0.054) 0.152**(0.074) 0.253***(0.064) ln(last market day price, oyster) 0.040 (0.066) 0.120 (0.077) −0.108**(0.049) 0.115 (0.081) −0.057 (0.057) Constant 5.287***(0.528) 5.572***(0.622) 5.202***(0.532) 5.314***(0.653) 5.654***(0.618) Durbin–Wu–Hausman test (F-statistic) 29.681*** 54.876*** 6.485** 15.623*** 28.632*** Weak instrument test (F-statistic) 75.4*** 59.1*** 155.4*** 8.59*** 84.816*** Observations 607 607 616 607 616 Variable . Model 1 (all) . Model 2 (selected) . Model 3 (market) . Model 4 (SST) . Model 5 (elderly fisher) . Market distance (km) 0.119*** (0.010) 0.128*** (0.010) SST (°C) −0.018***(0.004) −0.020***(0.005) −0.016***(0.005) Elderly fisher (dummy) −0.234***(0.037) −0.406***(0.041) −0.381***(0.041) ln(last market day price, octopus) 0.170***(0.057) 0.166**(0.070) 0.232***(0.054) 0.152**(0.074) 0.253***(0.064) ln(last market day price, oyster) 0.040 (0.066) 0.120 (0.077) −0.108**(0.049) 0.115 (0.081) −0.057 (0.057) Constant 5.287***(0.528) 5.572***(0.622) 5.202***(0.532) 5.314***(0.653) 5.654***(0.618) Durbin–Wu–Hausman test (F-statistic) 29.681*** 54.876*** 6.485** 15.623*** 28.632*** Weak instrument test (F-statistic) 75.4*** 59.1*** 155.4*** 8.59*** 84.816*** Observations 607 607 616 607 616 Note: This table reports the first-stage regression results of (3) and the test statistics for the Durbin–Wu–Hausman test of endogeneity and the weak instrument test. The dependent variable is ln(Price). Heteroskedasticity-robust standard errors are in parenthesis. ** p < 0.05, *** p < 0.01. Open in new tab Before estimating the demand parameters, we checked the validity of the instruments based on the Sargan–Hansen J-test, with the null hypothesis that a set of candidate instruments satisfy the overidentifying restrictions, i.e. E(zituit) = 0. Test results show that only the combination of SST and elderly fisher failed to reject the null hypothesis of overidentifying restrictions (Table 3). Given these test results, we used SST and elderly fisher jointly as a valid set of instruments to estimate the demand function. Table 3. Sargan–Hansen J-test results for different combinations of the candidate instrumental variables. Candidate instrument . (1) . (2) . (3) . (4) . Market distance O O – O SST O O O – Elderly fisher O – O O J-test statistic 28.281*** 9.994*** 1.113 17.341*** Candidate instrument . (1) . (2) . (3) . (4) . Market distance O O – O SST O O O – Elderly fisher O – O O J-test statistic 28.281*** 9.994*** 1.113 17.341*** Note: This table reports the results of the Sargan–Hansen J-test for different combinations of the candidate instrumental variables. The sign “O” indicates the variables included. Under the null hypothesis, instrumental variables satisfy overidentifying restrictions. *** p < 0.01. Open in new tab Table 3. Sargan–Hansen J-test results for different combinations of the candidate instrumental variables. Candidate instrument . (1) . (2) . (3) . (4) . Market distance O O – O SST O O O – Elderly fisher O – O O J-test statistic 28.281*** 9.994*** 1.113 17.341*** Candidate instrument . (1) . (2) . (3) . (4) . Market distance O O – O SST O O O – Elderly fisher O – O O J-test statistic 28.281*** 9.994*** 1.113 17.341*** Note: This table reports the results of the Sargan–Hansen J-test for different combinations of the candidate instrumental variables. The sign “O” indicates the variables included. Under the null hypothesis, instrumental variables satisfy overidentifying restrictions. *** p < 0.01. Open in new tab The results from the OLS and 2SLS with all the instruments (Model 1) show that the price elasticity of demand for octopus is inelastic, with estimates of −0.101 and −0.557, respectively (Table 4). However, these estimates are likely to be biased because of the endogeneity of prices or the inclusion of the invalid instrument (i.e. market distance) in the estimation. Using the selected set of instruments (SST and elderly fisher) either jointly or individually (Models 2, 4, and 5), the 2SLS estimate of elasticity ranges from −1.251 to −2.096, implying an elastic demand for octopus. Table 4. OLS and 2SLS estimates of demand function. Variable . OLS . 2SLS . Model 1 (all) . Model 2 (selected) . Model 3 (market) . Model 4 (SST) . Model 5 (elderly fisher) . ln(Price) −0.101 −0.557*** −1.417*** −0.338*** −2.096** −1.251*** (0.069) (0.123) (0.222) (0.126) (0.86) (0.253) ln(last market day price, octopus) −0.030 0.086 0.278** 0.023 0.430 0.227* (0.086) (0.099) (0.142) (0.094) (0.268) (0.134) ln(last market day price, oyster) 0.273*** 0.251*** 0.233** 0.266*** 0.219 0.240** (0.081) (0.086) (0.114) (0.082) (0.147) (0.107) Constant 1.136 3.539*** 8.143*** 2.414** 11.773** 7.344*** (0.914) (1.131) (1.698) (1.112) (4.781) (1.806) Observations 616 607 607 616 607 616 Variable . OLS . 2SLS . Model 1 (all) . Model 2 (selected) . Model 3 (market) . Model 4 (SST) . Model 5 (elderly fisher) . ln(Price) −0.101 −0.557*** −1.417*** −0.338*** −2.096** −1.251*** (0.069) (0.123) (0.222) (0.126) (0.86) (0.253) ln(last market day price, octopus) −0.030 0.086 0.278** 0.023 0.430 0.227* (0.086) (0.099) (0.142) (0.094) (0.268) (0.134) ln(last market day price, oyster) 0.273*** 0.251*** 0.233** 0.266*** 0.219 0.240** (0.081) (0.086) (0.114) (0.082) (0.147) (0.107) Constant 1.136 3.539*** 8.143*** 2.414** 11.773** 7.344*** (0.914) (1.131) (1.698) (1.112) (4.781) (1.806) Observations 616 607 607 616 607 616 Note: This table reports the OLS and 2SLS estimates of the demand function. The dependent variable is ln(Catch). Heteroskedasticity-robust standard errors are in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. Open in new tab Table 4. OLS and 2SLS estimates of demand function. Variable . OLS . 2SLS . Model 1 (all) . Model 2 (selected) . Model 3 (market) . Model 4 (SST) . Model 5 (elderly fisher) . ln(Price) −0.101 −0.557*** −1.417*** −0.338*** −2.096** −1.251*** (0.069) (0.123) (0.222) (0.126) (0.86) (0.253) ln(last market day price, octopus) −0.030 0.086 0.278** 0.023 0.430 0.227* (0.086) (0.099) (0.142) (0.094) (0.268) (0.134) ln(last market day price, oyster) 0.273*** 0.251*** 0.233** 0.266*** 0.219 0.240** (0.081) (0.086) (0.114) (0.082) (0.147) (0.107) Constant 1.136 3.539*** 8.143*** 2.414** 11.773** 7.344*** (0.914) (1.131) (1.698) (1.112) (4.781) (1.806) Observations 616 607 607 616 607 616 Variable . OLS . 2SLS . Model 1 (all) . Model 2 (selected) . Model 3 (market) . Model 4 (SST) . Model 5 (elderly fisher) . ln(Price) −0.101 −0.557*** −1.417*** −0.338*** −2.096** −1.251*** (0.069) (0.123) (0.222) (0.126) (0.86) (0.253) ln(last market day price, octopus) −0.030 0.086 0.278** 0.023 0.430 0.227* (0.086) (0.099) (0.142) (0.094) (0.268) (0.134) ln(last market day price, oyster) 0.273*** 0.251*** 0.233** 0.266*** 0.219 0.240** (0.081) (0.086) (0.114) (0.082) (0.147) (0.107) Constant 1.136 3.539*** 8.143*** 2.414** 11.773** 7.344*** (0.914) (1.131) (1.698) (1.112) (4.781) (1.806) Observations 616 607 607 616 607 616 Note: This table reports the OLS and 2SLS estimates of the demand function. The dependent variable is ln(Catch). Heteroskedasticity-robust standard errors are in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. Open in new tab The price range in the raw data is captured only partially by the selected instruments of SST and elderly fisher. This is because we first estimated the first-stage regression of (3) to capture the price variation that is explained by the instruments (i.e. exogeneous supply shocks). The selected instruments only shift the supply curve within the price ranges captured by these variables, and the price range captured is largely the same regardless of whether the valid instruments are used jointly or independently (Table 5). The price in the raw data ranges from 50 to 3330 JPY, while the variation in SST and elderly fisher jointly captures the price range from 556 to 1599 JPY. Likewise, the price range captured by each instrument is from 633 to 1121 JPY for SST and 463 to 1443 JPY for elderly fisher. These price ranges include the mean price of 1023 JPY. Table 5. The price range explained by each supply shock instrument. Supply shock instrument . Price range (JPY/kg) . Price elasticity of demand . Raw data 50–3 300 – Model 1 (all) 358–2 778 0.557 Model 2 (selected) 556–1 599 1.417 Model 3 (market) 320–2 222 0.338 Model 4 (SST) 633–1 121 2.096 Model 5 (elderly fisher) 463–1 443 1.251 Supply shock instrument . Price range (JPY/kg) . Price elasticity of demand . Raw data 50–3 300 – Model 1 (all) 358–2 778 0.557 Model 2 (selected) 556–1 599 1.417 Model 3 (market) 320–2 222 0.338 Model 4 (SST) 633–1 121 2.096 Model 5 (elderly fisher) 463–1 443 1.251 Note: This table reports the range of prices for octopus in JPY/kg explained by the first-stage regression of (3) when the instruments of market distance, SST and elderly fisher are used either jointly or individually. Open in new tab Table 5. The price range explained by each supply shock instrument. Supply shock instrument . Price range (JPY/kg) . Price elasticity of demand . Raw data 50–3 300 – Model 1 (all) 358–2 778 0.557 Model 2 (selected) 556–1 599 1.417 Model 3 (market) 320–2 222 0.338 Model 4 (SST) 633–1 121 2.096 Model 5 (elderly fisher) 463–1 443 1.251 Supply shock instrument . Price range (JPY/kg) . Price elasticity of demand . Raw data 50–3 300 – Model 1 (all) 358–2 778 0.557 Model 2 (selected) 556–1 599 1.417 Model 3 (market) 320–2 222 0.338 Model 4 (SST) 633–1 121 2.096 Model 5 (elderly fisher) 463–1 443 1.251 Note: This table reports the range of prices for octopus in JPY/kg explained by the first-stage regression of (3) when the instruments of market distance, SST and elderly fisher are used either jointly or individually. Open in new tab Economic effects of an increase in SST Figure 4 shows the estimated demand curve for octopus over the range of prices and quantities predicted by the selected model. The baseline price (Pbase) is 888 JPY/kg. An increase in SST shifts the supply curve downward, resulting in a decrease in the price of octopus. Given the elastic demand curve, when an increase in SST shifts the supply curve, the proportional increase in the quantity of octopus traded in the market will be greater than the decrease in price. This means that an increase in SST results in an increase in fishing revenue, as well as an increase in consumer surplus. In Figure 4, this consumer surplus gain associated with an increase in SST is illustrated by the shaded area below the demand curve between the baseline and the new prices. Figure 4. Open in new tabDownload slide Estimated demand curve and changes in the consumer surplus. Pbase is the baseline price of 888 JPY/kg, and P1.0 is the price of 778 JPY/kg when SST increases by one standard deviation. Figure 4. Open in new tabDownload slide Estimated demand curve and changes in the consumer surplus. Pbase is the baseline price of 888 JPY/kg, and P1.0 is the price of 778 JPY/kg when SST increases by one standard deviation. Using the estimated demand curve, we simulated the predicted change in price, quantity, and consumer surplus for each of the four scenarios where SST increases by 0.5, 1.0, 1.5, and 2.0 standard deviations, respectively (Table 6). One standard deviation is 6.6°C (Table 1). The results show that when SST increases by one standard deviation, the octopus supply on average increases by 1.68 kg per fisher per day, resulting in about a 0.7 million JPY gain in consumer surplus. The magnitude of the gain in consumer surplus increases as SST further increases. For example, when SST increases by two standard deviations, the predicted price becomes close to the lower bound of the price range explained by the selected model. In this case, the quantity traded increases by 3.71 kg per fisher per day from the baseline case, resulting in a consumer surplus gain of 1.4 million JPY. Table 6. Changes in consumer surplus due to an increase in SST. Increase in SST . Price (JPY/kg) . Quantity (kg/fisher/ day) . Change in consumer surplus (JPY/year) . Baseline 861.64 8.20 – 0.5σ 831.46 9.00 346 318 1.0σ 778.36 9.88 702 300 1.5σ 728.64 10.85 1 068 215 2.0σ 682.10 11.91 1 444 340 Increase in SST . Price (JPY/kg) . Quantity (kg/fisher/ day) . Change in consumer surplus (JPY/year) . Baseline 861.64 8.20 – 0.5σ 831.46 9.00 346 318 1.0σ 778.36 9.88 702 300 1.5σ 728.64 10.85 1 068 215 2.0σ 682.10 11.91 1 444 340 Note: This table reports predicted changes in the consumer surplus when SST increases by 0.5, 1.0, 1.5, and 2.0 standard deviations (σ). One standard deviation = 6.6°C. The price and quantity are calculated based on Model 2 in Tables 2 and 4. For the baseline, the mean values of SST, elderly fisher, and other control variables are used. The change in consumer surplus is calculated by multiplying the change in consumer surplus per fisher per day in (5) by the mean number of fishers per day and the total number of fishing days in 2017; 3 fishers × 237 days. Open in new tab Table 6. Changes in consumer surplus due to an increase in SST. Increase in SST . Price (JPY/kg) . Quantity (kg/fisher/ day) . Change in consumer surplus (JPY/year) . Baseline 861.64 8.20 – 0.5σ 831.46 9.00 346 318 1.0σ 778.36 9.88 702 300 1.5σ 728.64 10.85 1 068 215 2.0σ 682.10 11.91 1 444 340 Increase in SST . Price (JPY/kg) . Quantity (kg/fisher/ day) . Change in consumer surplus (JPY/year) . Baseline 861.64 8.20 – 0.5σ 831.46 9.00 346 318 1.0σ 778.36 9.88 702 300 1.5σ 728.64 10.85 1 068 215 2.0σ 682.10 11.91 1 444 340 Note: This table reports predicted changes in the consumer surplus when SST increases by 0.5, 1.0, 1.5, and 2.0 standard deviations (σ). One standard deviation = 6.6°C. The price and quantity are calculated based on Model 2 in Tables 2 and 4. For the baseline, the mean values of SST, elderly fisher, and other control variables are used. The change in consumer surplus is calculated by multiplying the change in consumer surplus per fisher per day in (5) by the mean number of fishers per day and the total number of fishing days in 2017; 3 fishers × 237 days. Open in new tab Revenue portfolio for young and elderly fishers Our first-stage regression (Table 2) shows that the dummy of elderly fisher is another significant variable that shifts the supply curve downward. More octopus being caught by elderly fishers than young fishers may reflect the fact that the opportunity cost of targeting different species is different between the two age groups. To identify this, we calculated the revenue portfolio by species for young and elderly octopus fishers, as shown in Figure 5. Panel (a) of the figure shows the mean revenue share of different species for young and elderly fishers, respectively, in accordance with the top five species in the revenue share that elderly fishers caught in 2017. Likewise, panel (b) shows the mean revenue share of different species for young and elderly fishers but in accordance with the top five species for young fishers. Figure 5. Open in new tabDownload slide Revenue portfolio by species for elderly fishers (≥62 years old) and young fishers (<62 years old). The numbers in red indicate the proportion in each revenue portfolio. Of the 40 fishers in the sample, 20 are elderly fishers, and 20 are young fishers. Panel (a) is the top five revenue proportions among all marine species caught by elderly octopus fishers and the corresponding species caught by younger octopus fishers. In contrast, panel (b) is the top five revenue proportions among all marine species caught by young octopus fishers and the corresponding species caught by elderly octopus fishers. Figure 5. Open in new tabDownload slide Revenue portfolio by species for elderly fishers (≥62 years old) and young fishers (<62 years old). The numbers in red indicate the proportion in each revenue portfolio. Of the 40 fishers in the sample, 20 are elderly fishers, and 20 are young fishers. Panel (a) is the top five revenue proportions among all marine species caught by elderly octopus fishers and the corresponding species caught by younger octopus fishers. In contrast, panel (b) is the top five revenue proportions among all marine species caught by young octopus fishers and the corresponding species caught by elderly octopus fishers. While fishers in Maizuru Bay generally targeted multiple species, there was a substantial difference in species that contributed to the fishing revenue of young and elderly fishers. For elderly fishers, the top five species were snow crab (Chionoecetes opilio), flatfish (Glyptocephalus cynoglossus), octopus, shrimp (Marsupenaeus japonicus), and oyster (Crassostrea gigas), accounting for 95% of the total revenue. However, the revenue share of these species for young fishers was only 39%, most of which was attributable to oyster. The top five species for young fishers were oyster, sea cucumber (Stichopus japonicus), Spanish mackerel (Scomberomorus niphonius), horse mackerel (Trachurus japonicus), and beltfish (Trichiurus lepturus). These species accounted for about 90% of the total revenue for young fishers, yet they were worth <5% of the total revenue for elderly fishers. In general, octopus was not the most important species for either elderly or young fishers. Octopus contributed ∼10% of the total revenue for elderly fishers, whereas <1% of the total revenue of young fishers came from octopus. Discussion An understanding of how local environmental and socio-economic factors impact fish supply and the price outcome of small-scale fisheries is limited but crucial for the formulation of effective fishery management policies. In this article, we considered day-to-day variation in SST and cross-sectional variation in market distance and fishers’ age as the potential factors that impact the daily amount of fish caught by each fisher in a small-scale fishery in Japan. Using the instrumental variable approach, the own-price elasticity of demand was estimated. There are numerous studies estimating the price elasticity of demand for different fish products in Japan and other parts of the world (Wessells and Wilen, 1993; Eales et al., 1997; Asche et al., 2005; Tokunaga, 2018). However, most studies have focused on industrial fisheries or national-level consumption figures, and consumers’ responsiveness to price changes in small-scale fisheries remains unclear. We found that octopus prices in the local market are endogenous and that all factors (i.e. SST, market distance, and age of fishers) significantly influence the price. These results indicate that each fisher’s daily supply of octopus is directly impacted by these environmental and socio-economic factors and that consumers are responsive to the price change as a result of the supply change. However, our results also show that only the set of SST and fishers’ age is identified as a valid exogeneous shock to the octopus supply. A potential reason why the inclusion of market distance makes the set of instruments invalid is that fishers’ residence, and hence their market distance, is strategically determined in accordance with the fishers’ productivity or other characteristics of their fishing businesses. This contradicts the expectation that Japan’s small-scale fishing households are not greatly mobile across different communities, as fishing rights for nearshore coastal waters are not transferable, and local fisheries are tied to personal and family identities within the community. Our results also show that both increasing SST and an aging workforce in the fishery increase octopus supply. It is important to note that these positive relationships are shown using intra-annual variation; therefore, they do not necessarily contradict previous studies that show that global ocean warming has had a net negative impact on global fisheries’ productivity (Free et al., 2019) and that population aging decreases the growth rate of national economies (Maestas et al., 2016; Katagiri, 2018; Aksoy et al., 2019). Intra-annual variation in SST may influence the fish supply not only because the migration and feeding behaviour of fish vary seasonally but also because fishers may change their target species at different times of the year. For example, in our study area, the main income-generating species (snow crab, oyster, and sea cucumber) are caught in winter, whereas octopus is caught throughout the year with minimal capital investment required. Therefore, although octopus is not the main target species, it provides a temporary source of income in the off-season of other species, allowing small-scale fishing operators to smooth seasonal income variation. Our results also show that the revenue share of octopus is relatively higher for elderly fishers, while the contribution of octopus to the revenue share of young fishers is <1%. This difference in fishing portfolios may reflect different needs and values and, hence, the opportunity cost of octopus fishing between the two age groups. The role of minor species in small-scale fisheries has received little attention in existing studies. An important implication of the current study is that the availability of such a species is particularly important for elderly fishers, who are well-experienced but reluctant to invest in new capital (such as boats and gear). This is consistent with fishers’ perception that it is important not to have a day without income. Our regression results show that consumers are highly responsive to price changes; thus, the own-price elasticity of octopus demand is elastic, ranging from −1.25 to −2.10. A price elastic demand for octopus is expected, as octopus is not a staple food in households’ daily diets and is highly substitutable with other seafood products. Our estimates are consistent with the estimate of Eales et al. (1997), who used national-level household consumption data in Japan to find an elastic demand for the category of cuttlefish, squid, and octopus. A price elastic demand means that the decreasing price due to an increase in SST or the proportion of elderly fishers in the community is compensated by a more proportional increase in quantity demand relative to the decreasing price. In other words, there are positive net effects on both fishing revenue and consumer surplus from increasing SST and an aging workforce in the fishery. This result, however, must be interpreted with caution, as the effects of these factors on other species, and hence the community as a whole, are not considered in our analysis. For example, the expansion of the home range of octopus has been found with ocean warming (Hamasaki and Morioka, 2002; Higgins et al., 2012; Ramos et al., 2018), whereas the range of other species that are less tolerant to warming has contracted (Neuheimer et al., 2011; Worm and Tittensor, 2011). An important area of further small-scale fisheries research is to investigate the community-wide impacts of changes in environmental and socio-economic factors. This requires interactions between different fish products and market interactions among fishers; thus, a different approach, such as the almost ideal demand system or computable general equilibrium model, may be better suited than the instrumental variable approach. Data availability The data underlying this article were provided by the Kyoto Fishery Cooperatives and the Maizuru Fisheries Research Station of Kyoto University. The data cannot be shared publicly due to the privacy of individuals that participated in the study. Data will be shared on request to the corresponding author with permission of the Kyoto Fishery Cooperatives and the Maizuru Fisheries Research Station of Kyoto University. Funding This research is supported by the University of Tasmania Graduate Research Scholarship. References Aksoy Y. , Basso H. S. , Smith R. P. , Grasl T. 2019 . Demographic structure and macroeconomic trends . American Economic Journal: Macroeconomics , 11 : 193 – 222 . Google Scholar Crossref Search ADS WorldCat Andreyeva T. , Long M. W. , Brownell K. D. 2010 . The impact of food prices on consumption: a systematic review of research on the price elasticity of demand for food . American Journal of Public Health , 100 : 216 – 222 . Google Scholar Crossref Search ADS PubMed WorldCat Angrist J. D. , Graddy K. , Imbens G. W. 2000 . The interpretation of instrumental variables estimators in simultaneous equations models with an application to the demand for fish . Review of Economic Studies , 67 : 499 – 527 . Google Scholar Crossref Search ADS WorldCat Angrist J. D. , Krueger A. B. 2001 . Instrumental variables and the search for identification: from supply and demand to natural experiments . Journal of Economic Perspectives , 15 : 69 – 85 . Google Scholar Crossref Search ADS WorldCat Asche F. , Bjørndal T. , Gordon D. V. 2005 . Demand Structure for Fish. SNF Working Paper NO.3705. Institute for Research in Economics and Business Administration, Bergen. Asche F. , Bjørndal T. , Gordon D. V. , 2007 . Studies in the demand structure for fish and seafood products. In Handbook of Operations Research in Natural Resources , pp. 295 – 314 . Ed. by Weintraub A. , Romero C. , Bjørndal T. , Epstein R. , Miranda J. , Springer , Boston . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Asche F. , Chen Y. , Smith M. D. 2015 . Economic incentives to target species and fish size: prices and fine-scale product attributes in Norwegian fisheries . ICES Journal of Marine Science , 72 : 733 – 740 . Google Scholar Crossref Search ADS WorldCat Asche F. , Guillen J. 2012 . The importance of fishing method, gear and origin: the Spanish hake market . Marine Policy , 36 : 365 – 369 . Google Scholar Crossref Search ADS WorldCat Aswani S. , Sabetian A. 2010 . Implications of urbanization for artisanal parrotfish fisheries in the Western Solomon Islands . Conservation Biology , 24 : 520 – 530 . Google Scholar Crossref Search ADS PubMed WorldCat Béné C. 2009 . Are fishers poor or vulnerable? Assessing economic vulnerability in small-scale fishing communities . Journal of Development Studies , 45 : 911 – 933 . Google Scholar Crossref Search ADS WorldCat Béné C. , Arthur R. , Norbury H. , Allison E. H. , Beveridge M. , Bush S. , Campling L. , et al. 2016 . Contribution of fisheries and aquaculture to food security and poverty reduction: assessing the current evidence . World Development , 79 : 177 – 196 . Google Scholar Crossref Search ADS WorldCat Béné C. , Macfayden G. , Allison E. H. 2007 . Increasing the Contribution of Small-Scale Fisheries to Poverty Alleviation and Food Security. Food and Agriculture Oraganization of the United Nations, Rome. 125 pp. Cinner J. E. , Marnane M. J. , McClanahan T. R. , Clark T. H. , Ben J. 2005 . Trade, tenure, and tradition: influence of sociocultural factors on resource use in Melanesia . Conservation Biology , 19 : 1469 – 1477 . Google Scholar Crossref Search ADS WorldCat Cinner J. E. , McClanahan T. R. 2006 . Socioeconomic factors that lead to overfishing in small-scale coral reef fisheries of Papua New Guinea . Environmental Conservation , 33 : 73 – 80 . Google Scholar Crossref Search ADS WorldCat Cinner J. E. , Graham N. A. J. , Huchery C. , Macneil M. A. 2013 . Global effects of local human population density and distance to markets on the condition of coral reef fisheries . Conservation Biology , 27 : 453 – 458 . Google Scholar Crossref Search ADS PubMed WorldCat Eales J. , Durham C. , Wessells C. R. 1997 . Generalized models of Japanese demand for fish . American Journal of Agricultural Economics , 79 : 1153 – 1163 . Google Scholar Crossref Search ADS WorldCat FAO . 2015 . Voluntary Guidelines for Securing Sustainable Small-Scale Fisheries . Food and Agricultural Organization of the United Nations, Rome . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC FAO . 2018 . The State of World Fisheries and Aquaculture 2018: Meeting the Sustainable Development Goals. Food and Agricultural Organization of the United Nations, Rome. Free C. M. , Thorson J. T. , Pinsky M. L. , Oken K. L. , Wiedenmann J. , Jensen O. P. 2019 . Impacts of historical warming on marine fisheries production . Science , 363 : 979 – 983 . Google Scholar Crossref Search ADS PubMed WorldCat Gallet C. A. 2009 . The demand for fish: a meta-analysis of the own-price elasticity . Aquaculture Economics and Management , 13 : 235 – 245 . Google Scholar Crossref Search ADS WorldCat Gattuso J. P. , Magnan A. , Billé R. , Cheung W. W. L. , Howes E. L. , Joos F. , Allemand D. , et al. 2015 . Contrasting futures for ocean and society from different anthropogenic CO2 emissions scenarios . Science , 349 : aac4722 . Google Scholar Crossref Search ADS PubMed WorldCat Graddy K. 2006 . Markets: the Fulton fish market . The Journal of Economic Perspectives , 20 : 207 – 220 . Google Scholar Crossref Search ADS WorldCat Greene W. H. 2018 Econometric Analysis ,8th edn. Pearson , New York . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hall S. J. 2011 . Climate change and other external drivers in small-scale fisheries: practical steps for responding. In Small-Scale Fisheries Management: Frameworks and Approaches for the Developing World, pp. 132–159. Ed. by R. Pomeroy and N. Andrew. Centre for Agriculture and Bioscience International, Wallingford. Hamasaki K. , Morioka T. 2002 . Effects of temperature on egg incubation period, and paralarval survival and growth of common cctopus, Octopus vulgaris . Aquaculture Science , 50 : 407 – 413 . Google Scholar OpenURL Placeholder Text WorldCat Hansen L. P. 1982 . Large sample properties of generalized method of moments estimators . Econometrica , 50 : 1029 – 1054 . Google Scholar Crossref Search ADS WorldCat Higgins F. A. , Bates A. E. , Lamare M. D. 2012 . Heat tolerance, behavioural temperature selection and temperature-dependent respiration in larval Octopus huttoni . Journal of Thermal Biology , 50 : 407 – 413 . Google Scholar OpenURL Placeholder Text WorldCat Kamoey A. 2015 . The Japanese Market for Seafood . Globefish Research Program , FAO, Rome . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Katagiri M. 2018 . Economic consequences of population aging in japan: effects through changes in demand structure . Singapore Economic Review , 1 – 23 . Google Scholar OpenURL Placeholder Text WorldCat Kirkley J. , Squires D. , Strand I. E. 1998 . Characterizing managerial skill and technical efficiency in a fishery . Journal of Productivity Analysis , 9 : 145 – 160 . Google Scholar Crossref Search ADS WorldCat Lee M. Y. 2014 . Hedonic pricing of Atlantic cod: effects of size, freshness, and gear . Marine Resource Economics , 29 : 259 – 277 . Google Scholar Crossref Search ADS WorldCat Lehodey P. , Alheit J. , Barange M. , Baumgartner T. , Beaugrand G. , Drinkwater K. , Fromentin J. M. , et al. 2006 . Climate variability, fish, and fisheries . Journal of Climate , 19 : 5009 – 5030 . Google Scholar Crossref Search ADS WorldCat Madin E. M. P. , Ban N. C. , Doubleday Z. A. , Holmes T. H. , Pecl G. T. , Smith F. 2012 . Socio-economic and management implications of range-shifting species in marine systems . Global Environmental Change , 22 : 137 – 146 . Google Scholar Crossref Search ADS WorldCat Maestas N. , Mullen K. , Powell D. 2016 . The Effect of Population Aging on Economic Growth, the Labor Force and Productivity . NBER Working Paper 22452, National Bureau of Economic Research, Cambridge . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC McConnell K. E. , Strand I. E. 2000 . Hedonic prices for fish: tuna prices in Hawaii . American Journal of Agricultural Economics , 82 : 133 – 144 . Google Scholar Crossref Search ADS WorldCat Neuheimer A. B. , Thresher R. E. , Lyle J. M. , Semmens J. M. 2011 . Tolerance limit for fish growth exceeded by warming waters . Nature Climate Change , 1 : 110 – 113 . Google Scholar Crossref Search ADS WorldCat OECD . 2019 . Pensions at a Glance 2019: OECD and G20 Indicators. OECD Publishing, Paris. 10.1787/b6d3dcfc-en . Crossref Pascoe S. , Coglan L. 2002 . The contribution of unmeasurable inputs to fisheries production: an analysis of technical efficiency of fishing vessels in the English Channel . American Journal of Agricultural Economics , 84 : 585 – 597 . Google Scholar Crossref Search ADS WorldCat Ramos J. E. , Pecl G. T. , Moltschaniwskyj N. A. , Semmens J. M. , Souza C. A. , Strugnell J. M. 2018 . Population genetic signatures of a climate change driven marine range extension . Scientific Reports , 8 : 9558 . Google Scholar Crossref Search ADS PubMed WorldCat Rousseau Y. , Watson R. A. , Blanchard J. L. , Fulton E. A. 2019 . Defining global artisanal fisheries . Marine Policy , 108 : 103634 . Google Scholar Crossref Search ADS WorldCat Sargan J. D. 1958 . The estimation of economic relationships using instrumental variables . Econometrica , 26 : 393 – 415 . Google Scholar Crossref Search ADS WorldCat Schmitt K. M. , Kramer D. B. 2009 . Road development and market access on Nicaragua’s Atlantic coast: implications for household fishing and farming practices . Environmental Conservation , 36 : 289 – 300 . Google Scholar Crossref Search ADS WorldCat Sogn-Grundvåg G. , Larsen T. A. , Young J. A. 2013 . The value of line-caught and other attributes: an exploration of price premiums for chilled fish in UK supermarkets . Marine Policy , 38 : 41 – 44 . Google Scholar Crossref Search ADS WorldCat Stevens K. , Irwin B. , Kramer D. , Urquhart G. 2014 . Impact of increasing market access on a tropical small-scale fishery . Marine Policy , 50 : 46 – 52 . Google Scholar Crossref Search ADS WorldCat Stock J. H. , Yogo M. 2005 . Testing for weak instruments in Linear Iv regression. In Identification and Inference for Econometric Models: Essays in Honor of Thomas Rothenberg , pp. 80 – 108 . Ed. by Andrews D. W. K. and Stock J. H. . Cambridge University Press , Cambridge . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Tingley D. , Pascoe S. , Coglan L. 2005 . Factors affecting technical efficiency in fisheries: stochastic production frontier versus data envelopment analysis approaches . Fisheries Research , 73 : 363 – 376 . Google Scholar Crossref Search ADS WorldCat Tokunaga K. 2018 . Estimating elasticity of demand for Pacific bluefin tuna in Tsukiji Fish Market . Marine Resource Economics , 33 : 27 – 60 . Google Scholar Crossref Search ADS WorldCat Wang W. , Zhou C. , Shao Q. , Mulla D. J. 2010 . Remote sensing of sea surface temperature and chlorophyll-a: implications for squid fisheries in the north-west pacific ocean . International Journal of Remote Sensing , 31 : 4515 – 4530 . Google Scholar Crossref Search ADS WorldCat Watson R. A. 2017 . A database of global marine commercial, small-scale, illegal and unreported fisheries catch 1950-2014 . Scientific Data , 4 : 170039 . Google Scholar Crossref Search ADS PubMed WorldCat Wellman K. F. 1992 . The US retail demand for fish products: an application of the almost ideal demand system . Applied Economics , 24 : 445 – 457 . Google Scholar Crossref Search ADS WorldCat Wessells C. R. , Wilen J. E. 1993 . Economic analysis of Japanese household demand for salmon . Journal of the World Aquaculture Society , 24 : 361 – 378 . Google Scholar Crossref Search ADS WorldCat Worm B. , Tittensor D. P. 2011 . Range contraction in large pelagic predators . Proceedings of the National Academy of Sciences of the United States of America , 108 : 11942 – 11947 . Google Scholar Crossref Search ADS PubMed WorldCat Yamashita H. 2012 . Social costs of keeping pension fishers in the community: the case of Japan. In Proceedings of the Sixteenth Biennial Conference of the International Institute of Fisheries Economics and Trade. IIFET, Dar es Salaam. Yamazaki S. , Resosudarmo B. P. , Girsang W. , Hoshino E. 2018 . Productivity, social capital and perceived environmental threats in small-island fisheries: insights from Indonesia . Ecological Economics , 152 : 62 – 75 . Google Scholar Crossref Search ADS WorldCat Zimmermann F. , Heino M. 2013 . Is size-dependent pricing prevalent in fisheries? The case of Norwegian demersal and pelagic fisheries . ICES Journal of Marine Science , 70 : 1389 – 1395 . Google Scholar Crossref Search ADS WorldCat Jang, H. G., Yamazaki, S., Kiyama, S., Higashida, K., and Tinch, D. 2021. Economic effects of sea surface temperature, aging population, and market distance on a small-scale fishery. – ICES Journal of Marine Science, 00:000–000. © International Council for the Exploration of the Sea 2021. 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)

Vaughan, Louise; Brophy, Deirdre; O’Toole, Ciar; Graham, Conor; Ó Maoiléidigh, Niall ; Poole, Russell

doi: 10.1093/icesjms/fsaa253pmid: N/A

Abstract Environmental and habitat change can have profound and complex impacts on fish. We examined an unexploited population of European eel (Anguilla anguilla) from a West of Ireland catchment. The population is long-lived and slow-growing compared to many other European eel populations. Von Bertalanffy growth curves showed decadal changes, with a trend towards larger K, and t0 values in both males and females and a smaller L∞ in females. A growth biochronology spanning seven decades (1950s–2010s) was constructed using otolith annual increment measurements. We found evidence of high variability in growth over the course of the time series. A decrease in growth occurred after the early 2000s, potentially driven by habitat and climatic changes. Growth was negatively correlated with early spring and winter temperatures, providing strong evidence that the length of the growing season impacts eel growth. Growth was also positively correlated with summer temperatures and the number of days that exceeded 16˚C (GSL16˚C). The response to temperature was age-dependent; at age one the positive relationship with GSL16˚C was most pronounced and the negative relationship with winter temperatures was not evident. This study demonstrates the impact of climate change and highlights the complexities of eel growth strategies in a changing environment. Introduction Growth is a key parameter in population dynamics and is a biological process that integrates the effects of both intrinsic and extrinsic components (Shelton et al., 2013; Morrongiello and Thresher, 2015). In fish, seasonal growth marks (annuli) in otoliths and other hard parts can be used to estimate age and to reconstruct an organism’s growth trajectory and environmental history. Historical collections of fish hard parts (otoliths/scales/operculi) provide a wealth of information on past ecological variation in fish communities, which has been revealed by building detailed multidecadal biochronologies of growth (Friedland and Haas, 1996; Morrongiello et al., 2012; Yokouchi and Daverat, 2013). Such reconstructed growth time series are particularly valuable in species where long-term data monitoring does not exist (Morrongiello et al., 2012). Relationships between historical biochronologies and available climatic and habitat data can inform predictions of biological responses to future change. In species such as estuary perch (Percalates colonorum) (Morrongiello et al., 2014; Stoessel et al., 2018), Alaskan rockfish (Sebastes polyspinis) (Matta et al., 2018), Atlantic horse mackerel (Trachurus trachurus) (Tanner et al., 2019), European plaice (Pleuronectes platessa) (van der Sleen et al., 2018), and rock flathead (Platycephalus laevigatus) (Barrow et al., 2018) otolith biochronologies have been successfully used to determine the effect of environmental drivers on growth and to examine potential responses to climatic impacts. The European eel follows a catadromous life cycle, spawning in the Sargasso Sea and moving back across the Atlantic Ocean as larvae, with the assistance of ocean currents. Since the 1980s stocks of the European eel have been in serious decline with recruitment decreasing by ∼90–95% since the 1970s (Dekker, 2003). The species was assessed by the International Union for Conservation of Nature (IUCN) in 2008 and formally classified as critically endangered in 2010 (Freyhof and Kottelat, 2010); this classification was reaffirmed in 2014 (Jacoby and Gollock, 2014). In 2007, the European Union adopted a European eel recovery plan under the Regulation (EC) No. 1100/2007 (European Council, 2007; ICES, 2016b). Various reasons have been proposed to explain the decline in European eel populations including overfishing, habitat degradation, pollution, parasites, and oceanic climatic factors that affect migration routes (Dekker, 2003; ICES, 2009). Diadromous fish species such as the European eel are thought to be particularly vulnerable to climate change as they utilise freshwater, brackish, and marine waters, all of which are subject to different and complex effects of climate change (Graham and Harrod, 2009). A pan-European abrupt decline in glass eel recruitment since the early 1980s suggests a broad scale climatic shift may be at least partly responsible, affecting larval survival and spawning success in the ocean and adult growth in freshwater (ICES, 2009, 2016a; Fealy et al., 2014). Climate change and global warming may have different effects on eel populations depending on latitude, local habitat conditions, and life history stage. The European average annual air temperature for the last decade (2009–2018) was between 1.6°C and 1.7°C above the pre-industrial level, making it the warmest decade on record (European Environment Agency, 2019). The global average temperature is predicted to continue to increase throughout this century (IPCC, 2014). This rise in temperatures is thought to be of benefit to European eels as growth rates, particularly in northern latitudes, have been predicted to rise (Daverat et al., 2012; Yokouchi and Daverat, 2013). However, European eels show great plasticity in growth strategies depending on local habitat conditions (Capoccioni et al., 2014). Growth during the juvenile (yellow eel) phase can vary geographically with eels from southern latitudes displaying much faster growth rates than individuals from more northern habitats (Capoccioni et al., 2014). However, increased growth rates can only be realized if sufficient prey is available as increased temperatures will increase metabolic rate (Graham and Harrod, 2009). Migrating silver eels do not feed, relying solely on fat reserves stored during the yellow eel phase (up to 30% of body weight) for energy during the 5000 km journey to the Sargasso Sea (Tesch, 2003). Improved knowledge of population-specific responses to climate change and changes to local habitat conditions is needed to reliably manage and conserve stocks of European eels. The objective of this study was to examine long-term trends in size and individual growth histories of silver eel, captured on their outward migration from the Burrishoole catchment in the West of Ireland. Growth patterns were analysed using individual length and age data as well as a seven-decade biochronology constructed from individual otolith growth measurements. Mixed effects models were used to examine the hierarchical nature of the growth time series. Interactions with seasonal temperatures and length of growing season were also investigated using a local surface water temperature time series. Changes in growth were described and relationships and the possible influence of local habitat conditions considered. Methods Study area The Burrishoole catchment, located in the West of Ireland (53° 56ʹ N, 9° 35ʹ W) (Figure 1), has a total productive wetted area of 474 ha (449 ha lacustrine, 25 ha fluvial). Loughs Feeagh and Bunaveela, the two largest freshwater lakes, have mean depths >12 m, are oligotrophic (TP < 10 µg L−1), coloured (∼ 80 mg L−1 PtCo), and have low alkalinity (<20 mg L−1 CaCO3) and pH (c. 6.7). The climate is strongly influenced by the Atlantic with mild winters and cool summers (Jennings et al., 2000). The catchment drains into Clew Bay in the Northeast Atlantic through Lough Furnace, a saline lagoon, by means of two channels; the Mill Race and the Salmon Leap. Partial upstream and downstream fish trapping facilities have been in place since 1958 and full trapping facilities have been in operation on both channels since 1970. The system has never been commercially fished for eels and no stocking of eels has taken place (Poole et al., 2018). Downstream traps on both channels are of similar design (wolf traps employing horizontal grids with 10 mm gaps on a 1:10 inclination) (Wolf, 1951; McGrath, 1969). Figure 1. Open in new tabDownload slide Location of the Burrishoole catchment on the west coast of Ireland showing the location of the two fish traps where eels migrating from the catchment were sampled. Figure 1. Open in new tabDownload slide Location of the Burrishoole catchment on the west coast of Ireland showing the location of the two fish traps where eels migrating from the catchment were sampled. The European-wide decrease in the number of eels migrating numbers is also evident in the Burrishoole catchment (Poole et al., 2018). Breakpoint analysis has previously indicated a shift in the silver eel production after 1982 with a substantial reduction in the numbers migrating from 4719 to 2821 (Poole et al., 2018). In more recent years, the ratio of silver female to male numbers has increased (from 24 to 45% between 2012 and 2017), following a longer-term decline from >90% males in the 1960s (Poole et al. 1990). The mean size of female eels has declined (from 53 cm in 1997–2005 to 50 cm in 2008–2012) while the mean length of male eels has remained stable at ∼36 cm (2000–2018) (Marine Institute, 2017). Changes in local environmental conditions have also been recorded in the catchment over a similar time period; the annual minimum surface water temperature has increased by 0.34°C decade−1 (Woolway et al., 2019) while water clarity indices have declined (Dalton et al. 2018). The Burrishoole system provided a unique environment for the study of a European eel population. Fish trapping facilities allowed for a comprehensive assessment of silver eel production from the freshwater population and the construction of a seven-decade biochronology in a virtually pristine natural fish habitat with little direct anthropogenic influences. Eel data collection Downstream migrating silver eels were caught in the fish traps at the Mill Race or Salmon Leap (Figure 1). Fish traps were checked daily and all silver eels were counted and recorded. A subsample of descending silver eels was analysed for lengths (to the nearest 0.1 cm) and weights (to the nearest 5 g). Results for the eel length analysis are presented in Poole et al. (2018) (see Supplementary Figures S1 and S2). An additional subsample of eels was killed via an overdose of anaesthetic, dissected, and sagittal otoliths removed. Otoliths were available for growth analysis from 1987, 1988, 1999, 2007, and 2012–2017. Additional age readings were available from a previous study (2001–2003, 2005) without the corresponding increment readings. These ages were used only in the analysis of the population age structure (Figure 2). Younger eels from the 1950s and 1960s are underrepresented in the time series as sampling of silver eels for otoliths only began in 1987. The minimum age at silvering of eels from cohorts in the 1950s and 1960s were 29 and 19, respectively. Due to the long-lived nature of eels in the Burrishoole catchment, many eels achieve ages of >35 (Poole and Reynolds, 1996a, b), allowing growth trajectories spanning over seven decades to be constructed. Figure 2. Open in new tabDownload slide Age structure (years) graphs of silver eels sampled for growth (a) females and (b) males in the Burrishoole catchment. Figure 2. Open in new tabDownload slide Age structure (years) graphs of silver eels sampled for growth (a) females and (b) males in the Burrishoole catchment. Otolith growth measurement Otoliths were extracted by dissecting the eel heads, placed on a glass slide convex side down, secured used clear tape, and then cut in half along the transverse axis through the nucleus using a sharp scalpel. After cutting otoliths were placed into the upper part of a Bunsen flame using a scalpel blade, until they darkened to a grey colour. After burning the otolith halves were mounted on microscope slide using silicon sealant with the reading surface pressed against the slide surface. A more detailed description of the methods is presented in the ICES eel age reading manual (ICES, 2011) after Hu and Todd (1981) and Moriarty (1983). Prepared samples were examined using a stereo microscope (Olympus SZX 12) under reflected light. Images of the prepared otoliths were taken using Image Pro-plus V6 software and examined for annual growth increments (Figure 3). All otoliths were aged by two readers to measure bias between readers and any otoliths that were too difficult to age were excluded from the analysis. Otoliths were deemed unreadable if the annuli were impossible to differentiate into distinct bands due to lack of contrast between light and dark areas or if mounted incorrectly in the silicon medium. This excluded 75 otoliths from the analysis. A summary of excluded otoliths is provided in Table 1. Annual growth measurements were taken from the end of one dark winter band to the end of the next along the axis of widest growth, producing a curved trajectory (Hu and Todd, 1981; Moriarty, 1983; ICES, 2011). All growth prior to the zero band, the first clearly marked band outside the nucleus (Figure 3) is taken to be oceanic growth and was not included in the analysis (ICES, 2011). The first annual growth measurement was taken from the end of the first winter band (zero band) (Poole et al., 1992; Poole et al., 2018 ) to the end of the second winter band. Calculation of growth rates from otoliths gave similar results to growth estimation from mark/recapture studies carried out in the 1980’s and ongoing PIT tag studies over the past decade (Poole and Reynolds, 1996a; Marine Institute, unpublished data, 2010—present). Figure 3. Open in new tabDownload slide Transverse section of a 21-year-old eel burned and cracked otolith from the Burrishoole catchment caught in 2014. Annuli are marked with ticks along transect P1. Figure 3. Open in new tabDownload slide Transverse section of a 21-year-old eel burned and cracked otolith from the Burrishoole catchment caught in 2014. Annuli are marked with ticks along transect P1. Table 1. Summary of sample size, ages, and lengths of all eels subsampled for age and growth. Decade . Years Sampled . Females . Males . Excluded Females . Excluded Males . n . Mean length cm (min–max) . Mean age (min–max) . n . Mean length cm (min–max) . Mean age (min–max) . % . Mean length cm (min–max) . % . Mean length cm (min–max) . 1980s 1987–1988 72 51.12 31.83 40 39.35 23.33 30.9 58.08 14.9 35.94 (42.3–88.9) (19–58) (32.0–43.5) (9–37) (43.7–88.2) (31.3–40.7) 1990s 1999 34 54.2 29.50 29 36.37 28.24 0 – 3.3 36.1 (44.9–65.8) (16–39) (32.6–65.8) (10–45) (36.1–36.1) 2000s 2001–2003, 2005a 168 54.96 30.20 118 36.17 20.50 – – – – (43.5–89.1) (12–46) (29.1–43.9) (7–58) 2007 81 52.11 28.90 61 36.05 19.85 8.0 59.50 9.0 36.57 (42.6–78.5) (16–42) (31.8–45.4) (8–39) (45–91.2) (34.2–41.8) 2010s 2012–2017 156 52.4 30.49 42 36.93 17.64 10.3 50.15 10.6 33.98 (40.5–101.5) (15–48) (32.6–45.3) (8–34) (41.8–60.1) (30.1–37.4) Total 511 53.13 30.27 290 36.71 21.11 – – – – (40.5–101.5) (12–48) (29.1–45.4) (7–58) Total (excluding 2001–2003, 2005) 343 52.24 30.30 172 37.09 21.54 14 55.71 9.9 35.63 (40.5–101.5) (15–48) (31.8–45.4) (8–45) (41.8–91.2) (30.1–40.7) Decade . Years Sampled . Females . Males . Excluded Females . Excluded Males . n . Mean length cm (min–max) . Mean age (min–max) . n . Mean length cm (min–max) . Mean age (min–max) . % . Mean length cm (min–max) . % . Mean length cm (min–max) . 1980s 1987–1988 72 51.12 31.83 40 39.35 23.33 30.9 58.08 14.9 35.94 (42.3–88.9) (19–58) (32.0–43.5) (9–37) (43.7–88.2) (31.3–40.7) 1990s 1999 34 54.2 29.50 29 36.37 28.24 0 – 3.3 36.1 (44.9–65.8) (16–39) (32.6–65.8) (10–45) (36.1–36.1) 2000s 2001–2003, 2005a 168 54.96 30.20 118 36.17 20.50 – – – – (43.5–89.1) (12–46) (29.1–43.9) (7–58) 2007 81 52.11 28.90 61 36.05 19.85 8.0 59.50 9.0 36.57 (42.6–78.5) (16–42) (31.8–45.4) (8–39) (45–91.2) (34.2–41.8) 2010s 2012–2017 156 52.4 30.49 42 36.93 17.64 10.3 50.15 10.6 33.98 (40.5–101.5) (15–48) (32.6–45.3) (8–34) (41.8–60.1) (30.1–37.4) Total 511 53.13 30.27 290 36.71 21.11 – – – – (40.5–101.5) (12–48) (29.1–45.4) (7–58) Total (excluding 2001–2003, 2005) 343 52.24 30.30 172 37.09 21.54 14 55.71 9.9 35.63 (40.5–101.5) (15–48) (31.8–45.4) (8–45) (41.8–91.2) (30.1–40.7) For the years 2001–2003, 2005 only age and length data was available, these years were not included in the growth analysis a No incremental data available, not included in growth analysis. Open in new tab Table 1. Summary of sample size, ages, and lengths of all eels subsampled for age and growth. Decade . Years Sampled . Females . Males . Excluded Females . Excluded Males . n . Mean length cm (min–max) . Mean age (min–max) . n . Mean length cm (min–max) . Mean age (min–max) . % . Mean length cm (min–max) . % . Mean length cm (min–max) . 1980s 1987–1988 72 51.12 31.83 40 39.35 23.33 30.9 58.08 14.9 35.94 (42.3–88.9) (19–58) (32.0–43.5) (9–37) (43.7–88.2) (31.3–40.7) 1990s 1999 34 54.2 29.50 29 36.37 28.24 0 – 3.3 36.1 (44.9–65.8) (16–39) (32.6–65.8) (10–45) (36.1–36.1) 2000s 2001–2003, 2005a 168 54.96 30.20 118 36.17 20.50 – – – – (43.5–89.1) (12–46) (29.1–43.9) (7–58) 2007 81 52.11 28.90 61 36.05 19.85 8.0 59.50 9.0 36.57 (42.6–78.5) (16–42) (31.8–45.4) (8–39) (45–91.2) (34.2–41.8) 2010s 2012–2017 156 52.4 30.49 42 36.93 17.64 10.3 50.15 10.6 33.98 (40.5–101.5) (15–48) (32.6–45.3) (8–34) (41.8–60.1) (30.1–37.4) Total 511 53.13 30.27 290 36.71 21.11 – – – – (40.5–101.5) (12–48) (29.1–45.4) (7–58) Total (excluding 2001–2003, 2005) 343 52.24 30.30 172 37.09 21.54 14 55.71 9.9 35.63 (40.5–101.5) (15–48) (31.8–45.4) (8–45) (41.8–91.2) (30.1–40.7) Decade . Years Sampled . Females . Males . Excluded Females . Excluded Males . n . Mean length cm (min–max) . Mean age (min–max) . n . Mean length cm (min–max) . Mean age (min–max) . % . Mean length cm (min–max) . % . Mean length cm (min–max) . 1980s 1987–1988 72 51.12 31.83 40 39.35 23.33 30.9 58.08 14.9 35.94 (42.3–88.9) (19–58) (32.0–43.5) (9–37) (43.7–88.2) (31.3–40.7) 1990s 1999 34 54.2 29.50 29 36.37 28.24 0 – 3.3 36.1 (44.9–65.8) (16–39) (32.6–65.8) (10–45) (36.1–36.1) 2000s 2001–2003, 2005a 168 54.96 30.20 118 36.17 20.50 – – – – (43.5–89.1) (12–46) (29.1–43.9) (7–58) 2007 81 52.11 28.90 61 36.05 19.85 8.0 59.50 9.0 36.57 (42.6–78.5) (16–42) (31.8–45.4) (8–39) (45–91.2) (34.2–41.8) 2010s 2012–2017 156 52.4 30.49 42 36.93 17.64 10.3 50.15 10.6 33.98 (40.5–101.5) (15–48) (32.6–45.3) (8–34) (41.8–60.1) (30.1–37.4) Total 511 53.13 30.27 290 36.71 21.11 – – – – (40.5–101.5) (12–48) (29.1–45.4) (7–58) Total (excluding 2001–2003, 2005) 343 52.24 30.30 172 37.09 21.54 14 55.71 9.9 35.63 (40.5–101.5) (15–48) (31.8–45.4) (8–45) (41.8–91.2) (30.1–40.7) For the years 2001–2003, 2005 only age and length data was available, these years were not included in the growth analysis a No incremental data available, not included in growth analysis. Open in new tab Temperature data Water temperature was recorded on the Burrishoole outflow near the Mill Race fish trap at midnight each day using a paper chart recorder with a thermocouple (1961–2004), a StowAway TidbiT temperature data logger from Onset (TB132-05 + 37) (2004–2009) or a temperature sensor on an Orpheus Mini Water Level Recorder from OTT Hydrometry (Woolway et al., 2019) (2009–2017). Missing temperature values were estimated from air temperature at the Lough Feeagh weather station using the air2water modelling approach of Piccolroaz (2016) (7.9% of values were estimated using this approach). Annual growing degree days (GDD) were calculated using the below formula: Growing Degree Days (GDD)=∑(Tdaily−Tbase),if Tdaily is greater than TbaseGDD=0, if Tdaily is less than Tbase where: Tdaily = Daily midnight water temperatures from the Mill Race Tbase = Base Temperature The number of days in a year when the temperature was above a designated base temperature provided estimates of growing season length (GSL). Base temperatures of 10˚C, 12˚C, and 16˚C were used to generate three variables: GSL10°C, GSL12°C, and GSL16°C. Analysis of temporal trends in eel population age structure All statistical analyses were conducted in R, version 3.5.1 (R Core Team, 2018). Population age structures were compared between decades using Tukey’s post-hoc and Anderson–Darling tests (Scholz and Zhu, 2019). Growth curves were statistically compared using likelihood ratio tests (Kimura, 1980). Temporal trends in length data were analysed using Mann–Kendall tests. Collinearity of the temperature variables was examined using variance inflation factor (VIF) analysis. No strong collinearity was observed among the included variables (VIF < 3, Pearson correlation coefficient <0.35). Analysis of temporal trends in eel back-calculated lengths-at-age A linear relationship between otolith and eel length for Burrishoole eels was established in previous studies by Moriarty (1983) and Poole (1994) and confirmed in this study (Supplementary Figure S3). Following the methods of Vigliola and Meekan (2009), three models of back-calculation were applied to the dataset (Dahl-Lea, Biological Intercept, and Modified Fry). There was little difference between the three models; however, the Modified Fry model showed a marginally better fit to the Burrishoole data (see Supplementary Figure S3 for further information back-calculation methods). Using back-calculated lengths determined by the three models, growth was described using the Von Bertalanffy (1957) growth equation (Von Bertalanffy, 1957): lt= L∞(1-e-Kt-t0), where: lt = length at time t L∞= the asymptotic length at which growth is zero K = a measure of the rate at which length approaches Lt (growth rate) t0 = the time at which the fish would have been zero size if it had always grown according to the von Bertalanffy equation. The three back-calculation models provided similar von Bertalanffy growth curves. The curves based on the back-calculated lengths from the Modified Fry were used to interpret of temporal changes in the growth of the population (Supplementary Figure S4). Von Bertalanffy growth curves were calculated based on the decade that an individual fish was recruited into the freshwater environment. R packages FSA (Ogle et al., 2020) and nlstools (Baty et al., 2015) were used to compute the von Bertalanffy growth curves. Analysis of temporal trends in eel otolith growth The approaches of Morrongiello and Thresher (2015) and Weisberg et al. (2010) were used to construct hierarchical mixed effect models of the relationships between annual otolith growth and sets of random and fixed explanatory variables. The analysis was run using the lme4 (Bates et al., 2015), AICcmodavg (Mazerolle, 2019), and effects (Fox and Weisberg, 2018) R packages. A series of variables representing intrinsic and extrinsic drivers of growth were included in the hierarchical models (Table 2). The random-effect FishID was included in the models to account for correlations between multiple annual growth measurements within individual fish. Age was also included in the random effects structure to account for age-dependant growth rates and Reader was included as a fixed effect to account for bias between readers. The response variable (Annual growth), and two predictor variables [Age and Age at Capture (AAC)] were natural log-transformed to satisfy the model assumptions. All predictor variables were mean centred to aid in model convergence. Table 2. Description of the parameters used in the analysis of European eel growth. Parameter . Description . Intrinsic fixed effects  Age Age (year) when otolith increment was formed  Sex Male or female  AAC Final age (year) at time of silvering  Year Year in which increment was formed Extrinsic fixed effects  Temperature Average temperature taken from daily water temperatures recorded at the Mill Race (split into Annual, Spring, Summer, Autumn and Winter, and monthly temperatures)  GDD Annual GDD (>10°C/>12°C/>16°C)  GSL Annual GSL (10°C/12°C/16°C)  Reader Increment reader Random effects  FishID (F) Unique fish identifier code  Year (Y) Year in which increment was formed  Cohort Group of individuals recruited into freshwater environment as glass eels in the same year  Age Random age slope on each of FishID, Year, and Cohort random intercepts Parameter . Description . Intrinsic fixed effects  Age Age (year) when otolith increment was formed  Sex Male or female  AAC Final age (year) at time of silvering  Year Year in which increment was formed Extrinsic fixed effects  Temperature Average temperature taken from daily water temperatures recorded at the Mill Race (split into Annual, Spring, Summer, Autumn and Winter, and monthly temperatures)  GDD Annual GDD (>10°C/>12°C/>16°C)  GSL Annual GSL (10°C/12°C/16°C)  Reader Increment reader Random effects  FishID (F) Unique fish identifier code  Year (Y) Year in which increment was formed  Cohort Group of individuals recruited into freshwater environment as glass eels in the same year  Age Random age slope on each of FishID, Year, and Cohort random intercepts Open in new tab Table 2. Description of the parameters used in the analysis of European eel growth. Parameter . Description . Intrinsic fixed effects  Age Age (year) when otolith increment was formed  Sex Male or female  AAC Final age (year) at time of silvering  Year Year in which increment was formed Extrinsic fixed effects  Temperature Average temperature taken from daily water temperatures recorded at the Mill Race (split into Annual, Spring, Summer, Autumn and Winter, and monthly temperatures)  GDD Annual GDD (>10°C/>12°C/>16°C)  GSL Annual GSL (10°C/12°C/16°C)  Reader Increment reader Random effects  FishID (F) Unique fish identifier code  Year (Y) Year in which increment was formed  Cohort Group of individuals recruited into freshwater environment as glass eels in the same year  Age Random age slope on each of FishID, Year, and Cohort random intercepts Parameter . Description . Intrinsic fixed effects  Age Age (year) when otolith increment was formed  Sex Male or female  AAC Final age (year) at time of silvering  Year Year in which increment was formed Extrinsic fixed effects  Temperature Average temperature taken from daily water temperatures recorded at the Mill Race (split into Annual, Spring, Summer, Autumn and Winter, and monthly temperatures)  GDD Annual GDD (>10°C/>12°C/>16°C)  GSL Annual GSL (10°C/12°C/16°C)  Reader Increment reader Random effects  FishID (F) Unique fish identifier code  Year (Y) Year in which increment was formed  Cohort Group of individuals recruited into freshwater environment as glass eels in the same year  Age Random age slope on each of FishID, Year, and Cohort random intercepts Open in new tab Model selection Model selection was on the basis of Akaike’s information criterion corrected for sample size (AICc). First, the optimal random effects structure was identified by fitting the maximal fixed effects model with an increasingly complex random effects structure. Next, the fixed effects structure was refit using different intrinsic fixed effects and the best-fitting model was chosen using AICc (Table 3). Finally, this optimal model was used as the base model for the addition of increasingly complex extrinsic fixed effects. Using the approach of Zuur et al. (2009), the best-fitting models were reanalysed using restricted maximum likelihood estimates of error to produce unbiased parameter estimates. Table 3. Model selection. Model . df . AICc . ΔAIC . LL . R2LMM(m) . R2LMM(c) . Random effects  1|FishID 9 32 911.2 2269.46 −16 447 0.335 0.409  Age|FishID 11 31 355.3 713.56 −15 667 0.358 0.445  Age|FishID + 1|Year 12 30 655.7 13.90 −15 316 0.319 0.453  1|Cohort 9 34 948.8 4306.98 −17 465 0.314 0.326  Age|FishID + Age|Year 14 30 641.8 0.00 −15 307 0.319 0.449 Fixed effects  Age 10 30 921.7 279.89 −15 451 0.292 0.441  Age + Sex 11 30 922.8 281.05 −15 450 0.293 0.442  Age + AAC 11 30 740 98.24 −15 359 0.307 0.447  Age + Sex + AAC 12 30 682.1 40.35 −15 329 0.316 0.451  (Age * Sex) + AAC 13 30 659.7 17.91 −15 317 0.324 0.457  (Age * Sex) + (AAC * Sex) 14 30 641.8 0.00 −15 307 0.319 0.449 Model . df . AICc . ΔAIC . LL . R2LMM(m) . R2LMM(c) . Random effects  1|FishID 9 32 911.2 2269.46 −16 447 0.335 0.409  Age|FishID 11 31 355.3 713.56 −15 667 0.358 0.445  Age|FishID + 1|Year 12 30 655.7 13.90 −15 316 0.319 0.453  1|Cohort 9 34 948.8 4306.98 −17 465 0.314 0.326  Age|FishID + Age|Year 14 30 641.8 0.00 −15 307 0.319 0.449 Fixed effects  Age 10 30 921.7 279.89 −15 451 0.292 0.441  Age + Sex 11 30 922.8 281.05 −15 450 0.293 0.442  Age + AAC 11 30 740 98.24 −15 359 0.307 0.447  Age + Sex + AAC 12 30 682.1 40.35 −15 329 0.316 0.451  (Age * Sex) + AAC 13 30 659.7 17.91 −15 317 0.324 0.457  (Age * Sex) + (AAC * Sex) 14 30 641.8 0.00 −15 307 0.319 0.449 Random effects structures were fitted using the maximal fixed effects structures [(Age * Sex) + (AAC * Sex) + Reader]. Intrinsic fixed effects structures were fitted using the optimal random effects structure ((Age|FishID) + (Age|Year)). All models included a Reader fixed effect to account for reader error. Optimal model shown in bold. LL, log-likelihood; R2LMM(m), marginal R2 (variance explained by fixed effects only); R2LMM(c), conditional R2 (variance explained by fixed and random effects). Open in new tab Table 3. Model selection. Model . df . AICc . ΔAIC . LL . R2LMM(m) . R2LMM(c) . Random effects  1|FishID 9 32 911.2 2269.46 −16 447 0.335 0.409  Age|FishID 11 31 355.3 713.56 −15 667 0.358 0.445  Age|FishID + 1|Year 12 30 655.7 13.90 −15 316 0.319 0.453  1|Cohort 9 34 948.8 4306.98 −17 465 0.314 0.326  Age|FishID + Age|Year 14 30 641.8 0.00 −15 307 0.319 0.449 Fixed effects  Age 10 30 921.7 279.89 −15 451 0.292 0.441  Age + Sex 11 30 922.8 281.05 −15 450 0.293 0.442  Age + AAC 11 30 740 98.24 −15 359 0.307 0.447  Age + Sex + AAC 12 30 682.1 40.35 −15 329 0.316 0.451  (Age * Sex) + AAC 13 30 659.7 17.91 −15 317 0.324 0.457  (Age * Sex) + (AAC * Sex) 14 30 641.8 0.00 −15 307 0.319 0.449 Model . df . AICc . ΔAIC . LL . R2LMM(m) . R2LMM(c) . Random effects  1|FishID 9 32 911.2 2269.46 −16 447 0.335 0.409  Age|FishID 11 31 355.3 713.56 −15 667 0.358 0.445  Age|FishID + 1|Year 12 30 655.7 13.90 −15 316 0.319 0.453  1|Cohort 9 34 948.8 4306.98 −17 465 0.314 0.326  Age|FishID + Age|Year 14 30 641.8 0.00 −15 307 0.319 0.449 Fixed effects  Age 10 30 921.7 279.89 −15 451 0.292 0.441  Age + Sex 11 30 922.8 281.05 −15 450 0.293 0.442  Age + AAC 11 30 740 98.24 −15 359 0.307 0.447  Age + Sex + AAC 12 30 682.1 40.35 −15 329 0.316 0.451  (Age * Sex) + AAC 13 30 659.7 17.91 −15 317 0.324 0.457  (Age * Sex) + (AAC * Sex) 14 30 641.8 0.00 −15 307 0.319 0.449 Random effects structures were fitted using the maximal fixed effects structures [(Age * Sex) + (AAC * Sex) + Reader]. Intrinsic fixed effects structures were fitted using the optimal random effects structure ((Age|FishID) + (Age|Year)). All models included a Reader fixed effect to account for reader error. Optimal model shown in bold. LL, log-likelihood; R2LMM(m), marginal R2 (variance explained by fixed effects only); R2LMM(c), conditional R2 (variance explained by fixed and random effects). Open in new tab The best-fitting random effects structure (using AICc) included a random slope for Age and random intercepts for both FishID and Year. The addition of cohort (year of recruitment into freshwater) resulted in the model failing to converge due to lack of data points. In the selection of the fixed effects, interactions between the temperature variables and age were included to allow for age-dependent temperature responses. Supplementary Table S2 shows the model selection table for temperature effects. Results Temporal trends in temperature in the Burrishoole catchment Temperature data from the Burrishoole system show evidence of decadal change. Supplementary Figure S5 illustrates seasonal changes in water temperature over the time series. Average seasonal surface water temperatures in Lough Feeagh have increased in recent decades with winter (DJF), spring (MAM), summer (JJA), and autumn (SON) temperatures rising from an average of 4.79°C, 7.36°C, 14.84°C, and 11.67°C, respectively, in the 1960s to 6.23°C, 8.93°C, 15.89°C, and 12.43°C in the 2010s. GDD (>10°C) and GDD (>12°C) (Figure 4a and b) increased from the 1960s to the 2010s. The mean number of GDD (>12°C) per year in the 1960s was 341 compared with 430 in the 1990s and 480 in the 2010s (Figure 4b). The first Julian day of the year when water temperatures above 12°C are recorded occurred earlier in the year in more recent decades (1960s: 131 d, 1990s: 120 d, 2010s: 95 d). GGD values for >16°C show a different pattern than >10°C and >12°C (Figure 4c) with the highest levels of GDD seen in the 1980s and the 2010s. GDD (>16°C) is more indicative of summer water temperatures in the Burrishoole system. The high levels of GDD (>16°C) seen in the 1980s seem to be driven by summer heatwave events that occurred in the years 1982–1984 and 1989. Figure 4. Open in new tabDownload slide Average cumulative GDD (a) >10°C, (b) >12˚C, and (c) >16˚C for the Burrishoole catchment based on midnight water temperatures from the Mill Race divided by decade. Figure 4. Open in new tabDownload slide Average cumulative GDD (a) >10°C, (b) >12˚C, and (c) >16˚C for the Burrishoole catchment based on midnight water temperatures from the Mill Race divided by decade. Temporal trends in eel population age structure Age was calculated from the growth dataset for four decades of eel samples (Table 1). A significant difference was seen between decadal groups in the mean age of males [F (3, 286) = 13.23, p < 0.001]. Tukey’s post-hoc tests showed significant differences between all decadal groups with the exception of the 2010s and 2000s and the 2000s and 1980s (Table 4). Male eels showed a general decrease in mean age from 23 in the 1980s to 18 in the 2010s. In the single sample from the 1990s, collected in 1999, the mean age of males was 28, which was higher than in any other decade (Table 1). There were more male eels in the 0–19 age range in samples from the 2000s and 2010s compared to the 1990s and 1980s (Figure 2). Table 4. Post-hoc tests multiple comparisons, Tukey’s HSD, age comparisons of European eel decadal groups (male and female) from the Burrishoole catchment (*p < 0.05; **p < 0.01; ***p < 0.001). Sex . Decade (I) . Decade (J) . Mean difference (I-J) . p-value . 95% confidence intervals . Lower . Upper . Female 1990s 1980s −3.513 0.043* −6.955 −0.072 2000s 1980s −3.234 0.001*** −5.420 −1.048 2010s 1980s −2.526 0.028* −4.859 −0.193 2000s 1990s 0.279 0.995 −2.771 3.329 2010s 1990s 0.987 0.852 −2.170 4.144 2010s 2000s 0.708 0.707 −0.995 2.411 Male 1990s 1980s 4.916 0.042* 0.120 9.713 2000s 1980s −3.046 0.103 −6.485 0.394 2010s 1980s −5.682 0.005** −10.027 −1.337 2000s 1990s −7.962 0.000*** −11.899 −4.025 2010s 1990s −10.599 0.000*** −15.347 −5.850 2010s 2000s −2.636 0.183 −6.009 0.736 Sex . Decade (I) . Decade (J) . Mean difference (I-J) . p-value . 95% confidence intervals . Lower . Upper . Female 1990s 1980s −3.513 0.043* −6.955 −0.072 2000s 1980s −3.234 0.001*** −5.420 −1.048 2010s 1980s −2.526 0.028* −4.859 −0.193 2000s 1990s 0.279 0.995 −2.771 3.329 2010s 1990s 0.987 0.852 −2.170 4.144 2010s 2000s 0.708 0.707 −0.995 2.411 Male 1990s 1980s 4.916 0.042* 0.120 9.713 2000s 1980s −3.046 0.103 −6.485 0.394 2010s 1980s −5.682 0.005** −10.027 −1.337 2000s 1990s −7.962 0.000*** −11.899 −4.025 2010s 1990s −10.599 0.000*** −15.347 −5.850 2010s 2000s −2.636 0.183 −6.009 0.736 Open in new tab Table 4. Post-hoc tests multiple comparisons, Tukey’s HSD, age comparisons of European eel decadal groups (male and female) from the Burrishoole catchment (*p < 0.05; **p < 0.01; ***p < 0.001). Sex . Decade (I) . Decade (J) . Mean difference (I-J) . p-value . 95% confidence intervals . Lower . Upper . Female 1990s 1980s −3.513 0.043* −6.955 −0.072 2000s 1980s −3.234 0.001*** −5.420 −1.048 2010s 1980s −2.526 0.028* −4.859 −0.193 2000s 1990s 0.279 0.995 −2.771 3.329 2010s 1990s 0.987 0.852 −2.170 4.144 2010s 2000s 0.708 0.707 −0.995 2.411 Male 1990s 1980s 4.916 0.042* 0.120 9.713 2000s 1980s −3.046 0.103 −6.485 0.394 2010s 1980s −5.682 0.005** −10.027 −1.337 2000s 1990s −7.962 0.000*** −11.899 −4.025 2010s 1990s −10.599 0.000*** −15.347 −5.850 2010s 2000s −2.636 0.183 −6.009 0.736 Sex . Decade (I) . Decade (J) . Mean difference (I-J) . p-value . 95% confidence intervals . Lower . Upper . Female 1990s 1980s −3.513 0.043* −6.955 −0.072 2000s 1980s −3.234 0.001*** −5.420 −1.048 2010s 1980s −2.526 0.028* −4.859 −0.193 2000s 1990s 0.279 0.995 −2.771 3.329 2010s 1990s 0.987 0.852 −2.170 4.144 2010s 2000s 0.708 0.707 −0.995 2.411 Male 1990s 1980s 4.916 0.042* 0.120 9.713 2000s 1980s −3.046 0.103 −6.485 0.394 2010s 1980s −5.682 0.005** −10.027 −1.337 2000s 1990s −7.962 0.000*** −11.899 −4.025 2010s 1990s −10.599 0.000*** −15.347 −5.850 2010s 2000s −2.636 0.183 −6.009 0.736 Open in new tab There was a significant difference in the mean ages of females between the decadal groups [F (3, 511) = 5.11, p < 0.001]. Tukey’s post-hoc tests showed significant differences between the ages of females in the 1980s and the 1990s and 2010s (Table 4). There was no significant difference between the 2000s and the 2010s. In females, the decrease in mean age was not as large as it was in males, decreasing from 33 in the 1980s to 31 in the 2010s. In the samples from the 1980s, there were females who reached ages of greater than 50; such long-lived fish were absent from samples collected in other decades. An Anderson–Darling K sample distribution test showed a significant difference in age distribution between decades (Male: AD = 8.792, p < 0.01, Female: AD = 6.211, p < 0.01) (Figure 2). Temporal trends in back-calculated lengths-at-age The analysis of von Bertalanffy growth curves showed decadal changes in growth parameters (Figure 5). L∞ decreased in the female eel population over the four decades from 1960s–1990s (Table 5) while K and t0 values increased over the same time period. Male eels showed a similar pattern of increasing t0 and K values. Values of L∞ were variable throughout the decades but did not show the same dramatic drop as was seen in the females. Likelihood ratio tests showed significant differences between growth curves in both males and females (Supplementary Table S1). An examination of growth curves using different models of back-calculated eel lengths is shown in Supplementary Figure S4. All models demonstrated similar decadal changes in growth. Figure 5. Open in new tabDownload slide Von Bertalanffy curves (lines) and observed values (points) for (a) female and (b) male silver eels in the Burrishoole catchment subdivided by decade eels were recruited into the catchment. Figure 5. Open in new tabDownload slide Von Bertalanffy curves (lines) and observed values (points) for (a) female and (b) male silver eels in the Burrishoole catchment subdivided by decade eels were recruited into the catchment. Table 5. Summary table of von Bertalanffy growth curves. Recruitment decade . n . L∞ . K . t0 . Female  1950s 41 85.60 0.023 −4.30  1960s 48 114.99 0.018 −4.14  1970s 90 93.93 0.024 −3.99  1980s 118 70.19 0.042 −2.83  1990s 39 63.07 0.061 −2.34 Male  1950s 11 51.97 0.033 −4.70  1960s 38 55.35 0.037 −3.71  1970s 19 61.79 0.039 −3.65  1980s 34 49.11 0.055 −3.17  1990s 57 55.73 0.058 −2.33  2000s 13 55.23 0.079 −1.64 Recruitment decade . n . L∞ . K . t0 . Female  1950s 41 85.60 0.023 −4.30  1960s 48 114.99 0.018 −4.14  1970s 90 93.93 0.024 −3.99  1980s 118 70.19 0.042 −2.83  1990s 39 63.07 0.061 −2.34 Male  1950s 11 51.97 0.033 −4.70  1960s 38 55.35 0.037 −3.71  1970s 19 61.79 0.039 −3.65  1980s 34 49.11 0.055 −3.17  1990s 57 55.73 0.058 −2.33  2000s 13 55.23 0.079 −1.64 Open in new tab Table 5. Summary table of von Bertalanffy growth curves. Recruitment decade . n . L∞ . K . t0 . Female  1950s 41 85.60 0.023 −4.30  1960s 48 114.99 0.018 −4.14  1970s 90 93.93 0.024 −3.99  1980s 118 70.19 0.042 −2.83  1990s 39 63.07 0.061 −2.34 Male  1950s 11 51.97 0.033 −4.70  1960s 38 55.35 0.037 −3.71  1970s 19 61.79 0.039 −3.65  1980s 34 49.11 0.055 −3.17  1990s 57 55.73 0.058 −2.33  2000s 13 55.23 0.079 −1.64 Recruitment decade . n . L∞ . K . t0 . Female  1950s 41 85.60 0.023 −4.30  1960s 48 114.99 0.018 −4.14  1970s 90 93.93 0.024 −3.99  1980s 118 70.19 0.042 −2.83  1990s 39 63.07 0.061 −2.34 Male  1950s 11 51.97 0.033 −4.70  1960s 38 55.35 0.037 −3.71  1970s 19 61.79 0.039 −3.65  1980s 34 49.11 0.055 −3.17  1990s 57 55.73 0.058 −2.33  2000s 13 55.23 0.079 −1.64 Open in new tab Temporal trends in otolith growth; contribution of intrinsic effects Mixed effects models were developed to examine factors influencing otolith growth (Table 3). The optimal intrinsic effects model of eel growth included Age, AAC Sex, and Reader as fixed effects with a random slope for Age and random intercepts for both FishID and Year as well as the Age*Sex and AAC*Sex interactions: yijk~α0+ αiF+ αkY+ β1 (Ageij)+ β2 (AACi)+β3 (Ageij*Sexi)+ β4 (AACi* Sexi)+ β5 (Readeri)+ b1iFAgeij+ b1kYAgejk+ εijk[αiFb1kF] ~ N(0, Σi), [αiYb1kY] ~ N(0, Σk), εijk ~ N(0, σ2) where: yijk = annual growth increment (µm) y for fish i at age j from year k, α0 = overall mean annual growth intercept, αiF = random intrinsic effect for fish i, αkY = random extrinsic environmental effect for year k, β1 = Age coefficient, β2 = AAC coefficient, β3 = Age * Sex coefficient, β4 = AAC * Sex coefficient, β5 = Reader coefficient, b1iF = random Age slope for fish i, correlated with αiF ⁠, Σ = covariance matrix between random intercept and random slope, b1kY = random Age slope for year k, correlated with αkY ⁠, εijk = normally distributed (⁠ N ⁠) error. The fixed effect portion of the model explained 34.3% of the variation in the dataset. Including a random slope for Age and random intercept for Year and FishID increased the percentage of variation predicted in the model to 45%. Age explained the greatest amount of variation (20%) in growth, as expected, growth rates declined with increasing age (Figure 6a). Females grow faster than males particularly in the earlier years of life (Figure 6a). AAC (age of silvering) accounted for 2.5% of the variation in growth in the intrinsic model. Figure 6. Open in new tabDownload slide Predicted annual otolith growth variation in Burrishoole European eels back-transformed to the original scale. (a) Age trend (mean with 95% CI) for males and females (microns); (b) Age-at-capture trend (mean with 95% CI) for males and females (microns). Figure 6. Open in new tabDownload slide Predicted annual otolith growth variation in Burrishoole European eels back-transformed to the original scale. (a) Age trend (mean with 95% CI) for males and females (microns); (b) Age-at-capture trend (mean with 95% CI) for males and females (microns). Growth at previous ages decreased with increasing Age at Capture (Figure 6b). The eels were captured on their outward migration from the river, so this corresponds to age at silvering. Therefore slow-growing eels silver at an older age than fast-growing eels. This decline in growth with Age at Capture was more pronounced in males than in females. In the optimal model, the random Age slope was positively correlated to the Year random intercept (correlation = 0.12) indicating that in good growth years the slope of the growth–age relationship was shallower (i.e. older fish grew proportionally better) whereas in poor growth years it steepened. In any given year, correlation between individuals was low (0.044) indicating high variability in growth between individual fish. Figure 7a shows the predicted average growth variation over time (Year) after accounting for intrinsic effects. There is considerable annual average growth variation over the time series. From the early 1970s to early 2000s, growth is above the average for the time series. Two peaks in growth can be seen in the mid-1990s and mid-1970s. The last 20 years of the time series (mid-1990s—late-2010s) was characterized by a period of declining growth with the lowest level of growth seen in 2012 (Figure 7a). Figure 7. Open in new tabDownload slide (a) Predicted time-dependent average otolith growth variation (after accounting for intrinsic effects) for Burrishoole European eels back-transformed to the original scale. Annual otolith growth variation represented by Year (1950–2017) random-effect conditional modes [best linear unbiased predictors (BLUPs) ± SE], (b) Cohort (1950–2005) random-effect conditional modes (BLUPs ± SE). Dashed lines represent long-term average otolith growth. Figure 7. Open in new tabDownload slide (a) Predicted time-dependent average otolith growth variation (after accounting for intrinsic effects) for Burrishoole European eels back-transformed to the original scale. Annual otolith growth variation represented by Year (1950–2017) random-effect conditional modes [best linear unbiased predictors (BLUPs) ± SE], (b) Cohort (1950–2005) random-effect conditional modes (BLUPs ± SE). Dashed lines represent long-term average otolith growth. A cohort only model with a random intercept was fitted to investigate variation in growth between eel cohorts. Similar conditions experienced by juveniles recruited in the same year can affect future growth and survival. Intraclass correlations for the random intercept Cohort model showed low levels of among individual correlations (0.02). Cohorts from the 1950s/60s grew more slowly than later cohorts (Figure 7b). The mean age at silvering of females eels that recruited to freshwater from the 1940s to 1960s was 32. Faster growers from these cohorts may have already matured and left the system when sampling commenced in 1987 and were therefore not sampled in the study. Individuals recruited in 1967, 1977, and 1995 showed the highest growth in the time series. Cohorts recruited into the catchment from the late-1970s to mid-1980s showed depressed growth. Temporal trends in otolith growth; contribution of extrinsic effects A series of different temperature models were fitted to investigate whether the temporal variation seen in the Year effect was associated with a change in the temperature regime in the catchment. A period of warming has been evident in the Burrishoole system since 1997/1998 (Poole et al., 2018). In all cases, the inclusion of a temperature variable to the base model improved the fit (Supplementary Table S2) with GSL variables (GSL16˚C/GSL12˚C/GSL10˚C) displaying the best fit of all the temperature variables. GSL16˚C is largely reflective of summer temperatures; 90% of days when temperatures exceeded 16˚C occurred in the summer months (June to August). The GSL10°C and GSL12°C seasons run on average from May to October and June to October, respectively, with 94 and 95% of days greater than 10°C/12°C occurring in these months. Growth was positively related to the length of the growing season at base temperature 16˚C (GSL16˚C) and average summer (June to August) temperatures but negatively related to all other temperature parameters (Table 6). Table 6. Parameter estimates (with SE) and test statistic t for models. Model and parameter . Intrinsic . Cohort . GSL16˚C and monthly temps . Estimate . SE . T . Sig . Estimate . SE . T . Sig . Estimate . SE . t . Sig . Intercept 3.635 0.017 212.417 *** 3.652 0.01 367.144 *** 3.66 0.015 241.161 *** Age −0.309 0.012 −26.529 *** −0.358 0.004 −85.464 *** −0.337 0.013 −24.976 *** AAC −0.402 0.041 −9.774 *** −0.183 0.02 −9.052 *** −0.359 0.042 −8.607 *** Sex (male) −0.118 0.02 −5.808 *** −0.121 0.008 −15.066 *** −0.127 0.02 −6.222 *** Reader (Reader 2) −0.022 0.005 −4.252 *** −0.023 0.006 −3.939 *** −0.02 0.005 −3.602 *** Age * Sex −0.089 0.017 5.285 *** −0.063 0.008 7.961 *** 0.102 0.018 5.819 *** AAC * Sex −0.239 0.051 −4.716 *** −0.342 0.026 −13.07 *** −0.261 0.052 −5.065 *** November Temp. – – – – – – −0.031 0.013 −2.448 ** November Temp. * Age – – – – – – −0.015 0.009 −1.721 . March Temp. – – – – – – −0.024 0.011 −2.087 * March Temp. * Age – – – – – – −0.011 0.009 −1.298 GSL16˚C – – – – – – 0.001 0.0006 1.775 * GSL16˚C * Age – – – – – – −0.0005 0.0004 −1.228 Model and parameter . Intrinsic . Cohort . GSL16˚C and monthly temps . Estimate . SE . T . Sig . Estimate . SE . T . Sig . Estimate . SE . t . Sig . Intercept 3.635 0.017 212.417 *** 3.652 0.01 367.144 *** 3.66 0.015 241.161 *** Age −0.309 0.012 −26.529 *** −0.358 0.004 −85.464 *** −0.337 0.013 −24.976 *** AAC −0.402 0.041 −9.774 *** −0.183 0.02 −9.052 *** −0.359 0.042 −8.607 *** Sex (male) −0.118 0.02 −5.808 *** −0.121 0.008 −15.066 *** −0.127 0.02 −6.222 *** Reader (Reader 2) −0.022 0.005 −4.252 *** −0.023 0.006 −3.939 *** −0.02 0.005 −3.602 *** Age * Sex −0.089 0.017 5.285 *** −0.063 0.008 7.961 *** 0.102 0.018 5.819 *** AAC * Sex −0.239 0.051 −4.716 *** −0.342 0.026 −13.07 *** −0.261 0.052 −5.065 *** November Temp. – – – – – – −0.031 0.013 −2.448 ** November Temp. * Age – – – – – – −0.015 0.009 −1.721 . March Temp. – – – – – – −0.024 0.011 −2.087 * March Temp. * Age – – – – – – −0.011 0.009 −1.298 GSL16˚C – – – – – – 0.001 0.0006 1.775 * GSL16˚C * Age – – – – – – −0.0005 0.0004 −1.228 Significance codes: “***” <0.0001, “**” <0.001, “*” <0.01, “.” < 0.05. Open in new tab Table 6. Parameter estimates (with SE) and test statistic t for models. Model and parameter . Intrinsic . Cohort . GSL16˚C and monthly temps . Estimate . SE . T . Sig . Estimate . SE . T . Sig . Estimate . SE . t . Sig . Intercept 3.635 0.017 212.417 *** 3.652 0.01 367.144 *** 3.66 0.015 241.161 *** Age −0.309 0.012 −26.529 *** −0.358 0.004 −85.464 *** −0.337 0.013 −24.976 *** AAC −0.402 0.041 −9.774 *** −0.183 0.02 −9.052 *** −0.359 0.042 −8.607 *** Sex (male) −0.118 0.02 −5.808 *** −0.121 0.008 −15.066 *** −0.127 0.02 −6.222 *** Reader (Reader 2) −0.022 0.005 −4.252 *** −0.023 0.006 −3.939 *** −0.02 0.005 −3.602 *** Age * Sex −0.089 0.017 5.285 *** −0.063 0.008 7.961 *** 0.102 0.018 5.819 *** AAC * Sex −0.239 0.051 −4.716 *** −0.342 0.026 −13.07 *** −0.261 0.052 −5.065 *** November Temp. – – – – – – −0.031 0.013 −2.448 ** November Temp. * Age – – – – – – −0.015 0.009 −1.721 . March Temp. – – – – – – −0.024 0.011 −2.087 * March Temp. * Age – – – – – – −0.011 0.009 −1.298 GSL16˚C – – – – – – 0.001 0.0006 1.775 * GSL16˚C * Age – – – – – – −0.0005 0.0004 −1.228 Model and parameter . Intrinsic . Cohort . GSL16˚C and monthly temps . Estimate . SE . T . Sig . Estimate . SE . T . Sig . Estimate . SE . t . Sig . Intercept 3.635 0.017 212.417 *** 3.652 0.01 367.144 *** 3.66 0.015 241.161 *** Age −0.309 0.012 −26.529 *** −0.358 0.004 −85.464 *** −0.337 0.013 −24.976 *** AAC −0.402 0.041 −9.774 *** −0.183 0.02 −9.052 *** −0.359 0.042 −8.607 *** Sex (male) −0.118 0.02 −5.808 *** −0.121 0.008 −15.066 *** −0.127 0.02 −6.222 *** Reader (Reader 2) −0.022 0.005 −4.252 *** −0.023 0.006 −3.939 *** −0.02 0.005 −3.602 *** Age * Sex −0.089 0.017 5.285 *** −0.063 0.008 7.961 *** 0.102 0.018 5.819 *** AAC * Sex −0.239 0.051 −4.716 *** −0.342 0.026 −13.07 *** −0.261 0.052 −5.065 *** November Temp. – – – – – – −0.031 0.013 −2.448 ** November Temp. * Age – – – – – – −0.015 0.009 −1.721 . March Temp. – – – – – – −0.024 0.011 −2.087 * March Temp. * Age – – – – – – −0.011 0.009 −1.298 GSL16˚C – – – – – – 0.001 0.0006 1.775 * GSL16˚C * Age – – – – – – −0.0005 0.0004 −1.228 Significance codes: “***” <0.0001, “**” <0.001, “*” <0.01, “.” < 0.05. Open in new tab Years of high growth coincided with high GSL16°C. Average GSL16°C for the time series (1961–2017) was 36 d. The highest rate of eel otolith growth was observed in 1995 (Figure 7a) when temperatures exceeded 16°C for 62 d. Other years of high growth coinciding with high numbers of days above 16°C included 1975 (N days = 74), 1976 (N days = 70), 1996 (N days = 53), and 1997 (N days = 59). During the period 2003–2017, GSL16˚C values were above 50 for 7 out of the 15 years. However, during those years, the peaks of growth seen in other high GSL16˚C years were not observed. A Wilcoxon signed-rank test showed no difference between years with above average GSL16˚C and growth (T = 446, z = −1.14, p = 0.256). Growth was greater in years with below average minimum water temperatures (T = 214, z = −3.04, p < 0.001). When the dataset was subsetted to only include the years with below the average minimum temperatures, a significant difference was seen between high GSL16˚C years and low GSL16˚C years with higher years showing higher growth (T = 140, z = −2.22, p < 0.05). It appeared that in the more recent part of the time series (2003–2017), as minimum water temperatures increased the effect of a high GSL16˚C in summer was less pronounced. To further investigate this interaction between the different temperature variables, additional temperature effects were added to the best-fitting model (GSL16°C). The addition of average March and November water temperatures improved the model fit (ΔAICc = 11.15, Supplementary Table S2). Figure 8 demonstrates the effects of GSL16°C (Figure 8a) and March and November temperatures (Figure 8b and c) from this model. Increasing surface water temperatures in March and November led to a decline in overall silver eel predicted growth. From 2002 onwards, eel otolith growth rates remained below the average for the time series despite the fact that summer temperatures were above average in several years during that period. During this time period, March and November water temperatures were above average (mean temperatures: 6.49°C and 9.94°C, respectively). The average interval that temperatures exceeded 16°C for this time period time ran from 22 June to 29 August, an increase of 8 d over the whole time series average of 17 June to 1 September. Interactions between Age and the temperature variables show that younger eels respond differently to temperature effects than older eels. In 1-year-old eels, the positive relationship between otolith growth GSL16°C is most pronounced (Figure 8d) whereas the negative relationship with spring (Figure 8e) and winter (Figure 8f) temperatures is not. This indicates that the longer growing season is of greatest benefit to young eels while negative responses to warming primarily affect the older age classes. Overall, the results indicate that eel growth has substantially decreased in the last decade and while warm summers have a positive influence on growth, warming occurring in early Spring/Winter can lead to a reduction in growth. Figure 8. Open in new tabDownload slide Predicted growth variation Burrishoole European eels back-transformed from a log scale to the original, temperature trends (a) Number of days >16˚C in a year, (b) Average March surface water temperatures (˚C), (c) Average November surface water temperatures (°C), (d) Age and number of days >16˚C in a year interaction, (e) Age and average March water temperatures (˚C), and (f) Age and average November water temperatures (˚C). Figure 8. Open in new tabDownload slide Predicted growth variation Burrishoole European eels back-transformed from a log scale to the original, temperature trends (a) Number of days >16˚C in a year, (b) Average March surface water temperatures (˚C), (c) Average November surface water temperatures (°C), (d) Age and number of days >16˚C in a year interaction, (e) Age and average March water temperatures (˚C), and (f) Age and average November water temperatures (˚C). Discussion The Burrishoole catchment with its comprehensive fish trapping system provided an excellent environment to study a European eel community in a virtually pristine natural fish habitat. The eel population within the system is unexploited and has never been stocked. There is no industrial activity in the catchment and the predominant land use is forestry and sheep grazing (Doyle et al., 2019). Blanket peatlands make up 52% of the catchment (Doyle et al., 2019) and are largely used as commonage for sheep grazing (Weir, 1996). This study has shown a change in the population age structure, mean lengths, and growth rates over the study period (1950s–2010s) in the absence of any exploitation or stocking. These changes coincide with the decline in silver eel numbers migrating annually from the system (Poole et al., 2018)and also a European-wide decline in eel recruitment numbers (Moriarty, 1990; ICES, 2019). Analysis of age and length distributions and of von Bertalanffy growth curves by decade (1950s–2000s) showed sex-specific changes in growth across the time series. Female age structure has remained relatively stable from the 1980s to the 2010’s, although numbers of old fish (>50 years) have declined and mean length of females has been declining since the mid-2000s (Poole et al., 2018). In contrast, for the males, the length distribution has remained relatively stable while the mean age has declined. In females, L∞ of the recruitment cohorts declined from the 1950s to the 1990s (note: no females from cohorts in the 2000s were captured) while K increased indicating a speeding up of the growth process and a greater slowing down towards maturation than observed in earlier years. Fast growers that recruited to the river population in the 1950s and 1960s may have matured and left the system before sampling commenced in 1987. The oldest fish in the samples may therefore represent the slow growers from that time period. An increase in growth rate was still evident in cohorts from the 1970s to present indicating that the change in growth was not only due to the absence of fast growers in the latter years of the dataset. Models developed to examine growth interactions with water temperature indicated a complex relationship. Relationships between the temperature variables and the annual eel otolith growth (year random effect) showed that in general eel growth increased as summer temperatures increased but decreased with increasingly mild conditions in spring and autumn/early winter. The observed decline in the size of female eels as temperatures increased is consistent with the temperature size rule, which predicts that due to metabolic demands smaller bodied individuals will do better in warm environments than larger-bodied individuals (Atkinson, 1994). Mounting evidence from multiple taxa indicates that a decrease in body size is a universal response to warming temperatures (Daufresne et al., 2009; Baudron et al., 2014) and the phenomenon is reported for multiple fish species in temperate environments (Brunel and Dickey-Collas, 2010; van Walraven et al., 2010; Baudron et al., 2014). The different temporal growth patterns displayed by male and female eels may reflect sex-specific adaptations. To optimize reproductive success males are thought to minimize the time needed to reach maturity while females maximize fecundity (Helfman et al., 1987; Davey and Jellyman, 2005; MacNamara and McCarthy, 2012). Both sexes must reach a sufficiently large body size to survive the spawning migration (Boetius and Boetius, 1980). In males, the success of the spawning migration is maximized by maturing at the smallest size possible (Vollestad and Jonsson, 1988; Davey and Jellyman, 2005). In less productive environments, females may silver at a minimum length that allows for a successful reproductive cycle (Yokouchi et al., 2018) while in more favourable environments migration may be delayed until a larger body size is reached (Davey and Jellyman, 2005). The trade-off for a prolonged growth period in freshwater is an increased risk of pre-spawning mortality (Helfman et al., 1987). In France, Ireland, Hungary, and the Netherlands, poor growing conditions led to earlier onset of maturity in female eels (Yokouchi et al., 2018). As temperatures have increased in the Burrishoole catchment males appear to reach the critical size for successful migration (average length: 36.2 cm ± 2.81) at a younger age (Poole et al., 2018)and are spending less time in freshwater. In females, size at migration has decreased and while mean age at migration has remained relatively stable, eels >50 years old have been absent from samples of the population since the 1980s. This may be indicative of a deterioration in growing conditions and a shift in the life history strategy towards time minimization rather than maximization of fecundity. There has been evidence of change in habitat conditions in the Burrishoole catchment in recent years. Palaeolimnological records show nutrient enrichment and increased sedimentation from the 1950s (Dalton et al., 2014), which have been linked to commercial forestry and overgrazing by livestock (Allott et al., 2005; May et al., 2005). The proportion of organic matter in Lough Feeagh increased from baseline levels of ∼27% before 1960 to 46% in the mid-1990s (de Eyto et al., 2016). While the recent shifts in eel life history strategies cannot be definitively linked to any one ecological feature; we suggest that a widespread study on eel habitat conditions within the catchment would be of benefit to further understanding of these changes. Fluctuations in length, growth, and age at migration may be under the influence of density-dependent processes. Although direct estimates of eel abundance in the entire Burrishoole catchment are not available, recent fyke net surveys in Lough’s Feeagh and Bunaveela combined with a consistent decline in the number of silver eels migrating (Poole et al., 2018)and a European-wide decline in glass eel recruitment (ICES, 2016a, c) are indicative of declining density. Average catch per unit effort CPUE of fyke net surveys carried out from 2009 to 2019 were 1.24 and 0.30 eels per net per night, respectively, in Lough’s Feeagh and Bunaveela. Similar surveys carried out in 1987/1988 (Poole, 1994) showed an average CPUE of 1.49 and 1.05 in Lough’s Feeagh and Bunaveela (Marine Institute, 2009–2019, unpublished data). In other areas, the proportion of females in the population has been shown to increase as densities decrease (Parsons et al., 1977; Helfman et al., 1987; Oliveira, 1999; Oliveira and McCleave, 2002; Tesch, 2003). In the Burrishoole, sex ratios of silver eels have varied from 94% males in the 1960s (Moriarty, 1974; Piggins, 1985) to ∼30% in the 1980s (Poole et al., 1990) and 44% in the 2010s (Poole et al., 2018), providing further evidence of declining abundance. Under the assumptions of density-dependent growth, reductions in abundance would be expected to lead to increases in growth and younger age at migration. Differences in otolith growth between males and female eels are widely reported (Poole and Reynolds, 1996b;, Poole and Reynolds, 1998 ; Oliveira, 1999). This is consistent with the results of this study, which shows that differences in otolith growth between males and females are evident from as early as the first year. The complex temperature–growth relationship indicated by the growth models may be linked to trends in seasonal temperature patterns and to changes in eel phenology. Water temperatures in Lough Feeagh have increased over the last two decades (Fealy et al., 2014). Between 1973 and 2014, the minimum surface temperature in Lough Feeagh increased (increase 0.08°C decade−1) four times more rapidly than the mean July to September surface temperature (Woolway et al., 2019). Average minimum surface water temperatures in this study for the 1960s were 2.4°C compared to 4.7°C in the 2010s. From 1971 to 2015, the silver eel migration run started between mid-July to late-October, on average 8 d earlier each year as the time series progressed (Sandlund et al., 2017). Cannaby and Hüsrevoğlu (2009) investigated sea surface temperatures in the Irish Sea and noted a period of warming from the late-1990s onwards; this warming was also evident in Lough Feeagh (Poole et al., 2018). In contrast with other studies examining eel growth in a warming environment (Daverat et al., 2012; Yokouchi and Daverat, 2013), the analysis of otolith growth measurements indicates that in the Burrishoole population growth has declined in the last two decades. While the analysis of von Bertalanffy curves suggests increasing growth, this is somewhat confounded by the change in the age structure of the population; as the analysis of the otolith annual growth measurements showed, growth at previous ages decreases with increasing age at capture. In the earlier part of the time series, the presence of more older slower growing fish contributes to the lower average growth trajectory for the population. The results of the hierarchical mixed effects modelling reveal that when all other intrinsic drivers of growth variability (Age, AAC, Sex) are equal, the predicted growth of an eel declined markedly after 2000. This demonstrates the power of the hierarchical modelling approach for disentangling complex responses to a changing climate. In a pan-European analysis of eel populations from 146 locations, Daverat et al. (2012) showed that growth rates increased as the number of days when temperatures exceeded 13˚C increased, demonstrating that on a broad scale, increases in growth may occur as water temperatures increase. In the same study, it was postulated that eel growth ceases below 12°C. While we see a similar positive trend with GSL16°C in this study, it is most pronounced at younger ages. A positive relationship between otolith growth and November water temperature was only evident in the youngest age class suggesting that older fish are more vulnerable to the negative effects of a warming environment. The extremely long-lived nature of Burrishoole eels may explain their distinct response to temperature conditions compared to other European populations. Eel’s exhibit dormant periods or torpor over winter months (Nyman, 1972; Walsh, 1983) where they cease feeding and burrow in the mud. Westerberg and Sjöberg (2015) showed that overwintering dormancy of European eels in Lake Mälaren, Sweden began when temperature fell below a range of 4–12°C and that activity resumed at a range of 3°C–7°C. Observations from fyke net surveys in the Burrishoole showed that CPUE was relatively high from April to September and low for the months of October to March (Poole, 1994), indicating a lack of activity over these winter months. Increases in water temperature during early spring/winter may be causing the timing, or level, of winter torpor state to change, leading to increases in metabolic rate and foraging activities at times where prey items may not be as readily available. Better knowledge of eel behaviour and feeding patterns and their relationships with temperature is needed to test this hypothesis. In conclusion, the Burrishoole eel population has shown considerable variation in growth over time with a pronounced decline in growth seen in recent years. Our investigation into the effects of water temperature demonstrates that local effects on eel populations may be more multifaceted than previously thought. Increases in water temperatures in early spring/winter appear to be having a negative effect on growth while warming summer temperatures are having a positive effect. The best-fitting model developed in this study accounted for 46.1% of the variation in growth observed in the Burrishoole silver eel population but could not fully explain the dramatic decline observed of late. Future studies examining habitat partitioning, diet composition, and additional environmental drivers may serve to explain more of this variance. The current study highlights the complexities of eel growth strategies in a changing environment. Globally, declines in eel recruitment are of major concern and the understanding of growing conditions and local habitat effects on eel populations may be of vital importance to future conservation efforts for this species. Knowledge of how local populations are likely to respond to future increases in temperature are useful for predicting the effectiveness of conservation measures such as habitat protection, regulation of fishing activity, removal of barriers to migration or restocking. Future conservation plans for European eels should account for localized responses to environmental change. Data availability The data that support the findings of this study are available from the corresponding author, upon reasonable request. Supplementary data Supplementary material is available at the ICESJMS online version of the manuscript. Funding This study was funded through a Grant-Aid Agreement No. PBA/FS/16/03 (unlocking the archive: using scale and otolith chronologies to resolve climate impacts) awarded to Deirdre Brophy under the Marine Research Programme by the Irish Government. Acknowledgements The authors gratefully acknowledge the work of the field staff of the Marine Institute, Newport (Burrishoole catchment) for their enormous efforts in data collection over the years that make studies like this possible. In particular, we would like to thank Mary Dillane and Elvira DeEyto for their incredible work maintaining the long-term monitoring programme in the Burrishoole catchment. We thank the editor and three anonymous reviewers whose comments/suggestions helped improve and clarify this manuscript. Special thanks to Elizabeth Tray, Conor Dolan, Joe Cooney, and Seán Kelly for help with dissections, imaging, and otolith ageing. References Atkinson D. 1994 . Temperature and organism size—a biological law for ectotherms? In Advances in Ecological Research, pp. 1–58 . Ed. by Begon M. , Fitter A. H. Academic Press, London . 301 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Barrow J. , Ford J. , Day R. , Morrongiello J. 2018 . Environmental drivers of growth and predicted effects of climate change on a commercially important fish, Platycephalus laevigatus . Marine Ecology Progress Series , 598 : 201 – 212 . 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 : 1 – 48 . Google Scholar Crossref Search ADS WorldCat Baty F. , Ritz C. , Charles S. , Brutsche M. , Flandrois J.-P. , Delignette-Muller M.-L. 2015 . A toolbox for nonlinear regression in R: the package nlstools . Journal of Statistical Software , 66 : 1 – 21 . Google Scholar Crossref Search ADS WorldCat Baudron A. R. , Needle C. L. , Rijnsdorp A. D. , Tara Marshall C. 2014 . Warming temperatures and smaller body sizes: synchronous changes in growth of North Sea fishes . Global Change Biology , 20 : 1023 – 1031 . Google Scholar Crossref Search ADS PubMed WorldCat Boetius I. , Boetius J. 1980 . Experimental maturation of female silver eels, Anguilla anguilla . Estimates of fecundity and energy reserves for migration and spawning . Dana , 1 : 1 – 28 . Google Scholar OpenURL Placeholder Text WorldCat Brunel T. , Dickey-Collas M. 2010 . Effects of temperature and population density on von Bertalanffy growth parameters in Atlantic herring: a macro-ecological analysis . Marine Ecology Progress Series , 405 : 15 – 28 . Google Scholar Crossref Search ADS WorldCat Cannaby H. , Hüsrevoğlu Y. S. 2009 . The influence of low-frequency variability and long-term trends in North Atlantic sea surface temperature on Irish waters . ICES Journal of Marine Science , 66 : 1480 – 1489 . Google Scholar Crossref Search ADS WorldCat Capoccioni F. , Lin D.-Y. , Iizuka Y. , Tzeng W.-N. , Ciccotti E. 2014 . Phenotypic plasticity in habitat use and growth of the European eel (Anguilla anguilla) in transitional waters in the Mediterranean area . Ecology of Freshwater Fish , 23 : 65 – 76 . Google Scholar Crossref Search ADS WorldCat Dalton C. , O’Dwyer B. , Taylor D. , de Eyto E. , Jennings E. , Chen G. , Poole R. , et al. 2014 . Anthropocene environmental change in an internationally important oligotrophic catchment on the Atlantic seaboard of western Europe . Anthropocene , 5 : 9 – 21 . Google Scholar Crossref Search ADS WorldCat Dalton C. , Sparber K. , de Eyto E. 2018 . Assessing sedimentation in a temperate dystrophic lake in the NE Atlantic seaboard region . Journal of Paleolimnology , 60 : 117 – 131 . Google Scholar Crossref Search ADS WorldCat Daufresne M. , Lengfellner K. , Sommer U. 2009 . Global warming benefits the small in aquatic ecosystems. Proceedings of the National Academy of Sciences of the United States of America, 106 : 12788 – 12793 . Daverat F. , Beaulaton L. , Poole R. , Lambert P. , Wickström H. , Andersson J. , Aprahamian M. et al. 2012 . One century of eel growth: changes and implications . Ecology of Freshwater Fish , 21 : 325 – 336 . Google Scholar Crossref Search ADS WorldCat Davey A. , Jellyman D. 2005 . Sex determination in freshwater eels and management options for manipulation of sex . Reviews in Fish Biology and Fisheries , 15 : 37 – 52 . Google Scholar Crossref Search ADS WorldCat de Eyto E. , Dalton C. , Dillane M. , Jennings E. , McGinnity P. , O’Dwyer B. , Poole R. et al. 2016 . The response of North Atlantic diadromous fish to multiple stressors, including land use change: a multidecadal study . Canadian Journal of Fisheries and Aquatic Sciences , 73 : 1759 – 1769 . Google Scholar Crossref Search ADS WorldCat Dekker W. 2003 . Did lack of spawners cause the collapse of the European eel. Anguilla anguilla? Fisheries Management and Ecology , 10 : 365 – 376 . Google Scholar Crossref Search ADS WorldCat Doyle B. C. , de Eyto E. , Dillane M. , Poole R. , McCarthy V. , Ryder E. , Jennings E. 2019 . Synchrony in catchment stream colour levels is driven by both local and regional climate . Biogeosciences , 16 : 1053 – 1071 . Google Scholar Crossref Search ADS WorldCat European Council . 2007 . Council Regulation (EC) No 1100/2007 of 18 September 2007 establishing measures for the recovery of the stock of European eel . Official Journal of the European Union , L248 : 1 – 7 . OpenURL Placeholder Text WorldCat European Environment Agency . 2019 . Global and European Temperature. European Environment Agency. https://www.eea.europa.eu/data-and-maps/indicators/global-and-european-temperature-9/assessment (last accessed 24 April 2020). Fealy R. , Allott N. , Broderick C. , de Eyto E. , Dillane M. , Erdil R. M. , Jennings E. , et al. 2014 . RESCALE: Review and simulate climate and catchment responses at Burrishoole. Technical Report. 409 pp. https://oar.marine.ie/handle/10793/944 (last accessed 24 April 2020). Fox J. , Weisberg S. 2018 . Visualizing fit and lack of fit in complex regression models with predictor effect plots and partial residuals . Journal of Statistical Software , 87 : 1 – 27 . Google Scholar Crossref Search ADS WorldCat Freyhof J. , Kottelat M. 2010 . Anguilla anguilla. The IUCN Red List of Threatened Species 2010: e.T60344A12353683. 24 April 2020. Friedland K. D. , Haas R. E. 1996 . Marine post-smolt growth and age at maturity of Atlantic salmon . Journal of Fish Biology , 48 : 1 – 15 . Google Scholar Crossref Search ADS WorldCat Graham C. T. , Harrod C. 2009 . Implications of climate change for the fishes of the British Isles . Journal of Fish Biology , 74 : 1143 – 1205 . Google Scholar Crossref Search ADS PubMed WorldCat Helfman G. S. , Facey D. E. , Stanton Hales L. , Bozeman E. L. 1987 . Reproductive ecology of the American eel. American Fisheries Society Symposium, 1 : 42 – 56 . Hu L. C. , Todd P. R. 1981 . An improved technique for preparing eel otoliths for ageing . New Zealand Journal of Marine and Freshwater Research , 5 : 445 – 446 . Google Scholar Crossref Search ADS WorldCat ICES . 2009 . Report of the Study Group on Anguillid Eels in Saline Waters (SGAESAW), 16–18 March 2009, Sackville, Canada; 3–5 September 2009, Gothenburg, Sweden. ICES Document CM/DFC: 06. 183 pp. ICES . 2011 . Report of the Workshop on Age Reading of European and American Eel (WKAREA2), 22–24 March 2011, Bordeaux, France. ICES Document CM 2011/ACOM: 43. 35 pp. Separate document – Annex 5: Manual for the aging of Atlantic Eel: Otolith preparation methodologies, age interpretation and image storage. IPCC . 2014 : Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team . Ed. by Pachauri R.K. and Meyer L.A. . IPCC , Geneva, Switzerland . 151 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC ICES . 2016a . Report of the Joint EIFAAC/ICES/GFCM Working Group on Eel (WGEEL), 24 November–2 December 2015 , Antalya, Turkey . ICES Document CM 2015/ACOM : 18 . 130 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC ICES . 2016b . Report of the Working Group on Eels (WGEEL), 15–22 September 2016 , Cordoba, Spain . ICES Document CM 2016/ACOM : 19 . 107 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC ICES . 2016c . Report of the Workshop on Growth-increment Chronologies in Marine Fish: climate-ecosystem interactions in the North Atlantic 2 (WKGIC2), 18–22 April 2016 , Esporles, Mallorca, Spain . ICES Document CM 2016/SSGIEOM : 18 . 18 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC ICES . 2019 . European eel (Anguilla anguilla) throughout its natural range. In Report of the ICES Advisory Committee, 2019. ICES Advice 2019, ele.2737.nea. doi: 10.17895/ices.advice.4825. Jacoby D. , Gollock M. 2014 . Anguilla anguilla . The IUCN Red List of Threatened Species , 2014 : e.T60344A45833138. doi: 10.2305/IUCN.UK.2014-1.RLTS.T60344A45833138.en. 24 April 2020. Google Scholar OpenURL Placeholder Text WorldCat Jennings E. , Allott N. , McGinnity P. , Poole R. , Quirke W. , Twomey H. , George G. 2000 . The North Atlantic Oscillation: effects on freshwater systems in Ireland . Biology and Environment: Proceedings of the Royal Irish Academy , 100B : 149 – 157 . Google Scholar OpenURL Placeholder Text WorldCat Kimura D. 1980 . Likelihood methods for the von Bertalanffy growth curve . Fishery Bulletin , 77 : 765 – 776 . Google Scholar OpenURL Placeholder Text WorldCat MacNamara R. , McCarthy T. K. 2012 . Size-related variation in fecundity of European eel (Anguilla anguilla) . ICES Journal of Marine Science , 69 : 1333 – 1337 . Google Scholar Crossref Search ADS WorldCat Marine Institute . 2017 . Newport Research Facility. Annual Report No. 62. Marine Institute, Newport. 73 pp. Matta M. E. , Helser T. E. , Black B. A. 2018 . Intrinsic and environmental drivers of growth in an Alaskan rockfish: an otolith biochronology approach . Environmental Biology of Fishes , 101 : 1571 – 1587 . Google Scholar Crossref Search ADS WorldCat May L. , Place C. , O’Hea B. , Lee M. , Dillane M. , McGinnity P. 2005 . Modelling soil erosion and transport in the Burrishoole catchment, Newport, Co. Mayo, Ireland . Freshwater Forum , 23 : 139 – 154 . Google Scholar OpenURL Placeholder Text WorldCat Mazerolle M. J. 2019 AICcmodavg: model selection and multimodel inference based on (Q)AIC(c). R package version 2.2-2. https://cran.r-project.org/package=AICcmodavg. McGrath C. J. 1969 . The fish trapping installation at the Salmon Leap, Furnace, Co. Mayo. Annual Report of the Salmon Research Trust, Appendix V, 107 pp. Moriarty C. 1974 . Eel research in 1973. An Roinn Talmhaíochta agus Iascaigh . Fishery Leaflet , 60 : 9 pp. Google Scholar OpenURL Placeholder Text WorldCat Moriarty C. 1983 . Age determination and growth rate of eels, Anguilla anguilla (L) . Journal of Fish Biology , 23 : 257 – 264 . Google Scholar Crossref Search ADS WorldCat Moriarty C. 1990 . European catches of elver of 1928–1988 . Internationale Revue Des Gesamtes Hydrobiology , 75 : 701 – 706 . Google Scholar Crossref Search ADS WorldCat Morrongiello J. R. , Thresher R. E. 2015 . A statistical framework to explore ontogenetic growth variation among individuals and populations: a marine fish example . Ecological Monographs , 85 : 93 – 115 . Google Scholar Crossref Search ADS WorldCat Morrongiello J. R. , Thresher R. E. , Smith D. C. 2012 . Aquatic biochronologies and climate change . Nature Climate Change , 2 : 849 – 857 . Google Scholar Crossref Search ADS WorldCat Morrongiello J. R. , Walsh C. T. , Gray C. A. , Stocks J. R. , Crook D. A. 2014 . Environmental change drives long-term recruitment and growth variation in an estuarine fish . Global Change Biology , 20 : 1844 – 1860 . Google Scholar Crossref Search ADS PubMed WorldCat Ogle D. H. , Wheeler P. , Dinno A. , 2020 . FSA: Fisheries Stock Analysis. R package version 0.8.30. https://github.com/droglenc/FSA. Oliveira K. 1999 . Life history characteristics and strategies of the American eel, Anguilla rostrata . Canadian Journal of Fisheries and Aquatic Sciences , 56 : 795 – 802 . Google Scholar Crossref Search ADS WorldCat Oliveira K. , McCleave J. D. 2002 . Sexually different growth histories of the American eel in four rivers in Maine . Transactions of the American Fisheries Society , 131 : 203 – 211 . Google Scholar Crossref Search ADS WorldCat Parsons J. , Vickers K. U. , Warden Y. 1977 . Relationship between elver recruitment and changes in the sex ratio of silver eels Anguilla anguilla L. migrating from Lough Neagh, Northern Ireland . Journal of Fish Biology , 10 : 211 – 229 . Google Scholar Crossref Search ADS WorldCat Piccolroaz S. 2016 . Prediction of lake surface temperature using the air2water model: guidelines, challenges, and future perspectives . Advances in Oceanography and Limnology , 7 : 36 – 50 . Google Scholar Crossref Search ADS WorldCat Piggins D. J. 1985 . The silver eel runs of the Burrishoole river system: 1959-1984; numbers, weights, timing and sex ratios. ICES Document CM 1985/M.5: International Council for the Exploration of the Sea, Copenhagen: 12 pp. Poole W. R. , Diserud O. H. , Thorstad E. B. , Durif C. M. , Dolan C. , Sandlund O. T. , Bergesen K. , et al. 2018 . Long-term variation in numbers and biomass of silver eels being produced in two European river systems . ICES Journal of Marine Science , 75 : 1627 – 1637 . Google Scholar Crossref Search ADS WorldCat Poole W. R. 1994 . A population study of the European eel (Anguilla anguilla (L.)) in the Burrishoole system, Ireland, with special reference to growth and movement, Ph.D. thesis, Dublin University, Dublin, Ireland. Poole W. R. , Reynolds J. D. 1996a . Age and growth of yellow eel, Anguilla anguilla (L.), determined by two different methods . Ecology of Freshwater Fish , 5 : 86 – 95 . Google Scholar Crossref Search ADS WorldCat Poole W. R. , Reynolds J. D. 1996b . Growth rate and age at migration of Anguilla anguilla . Journal of Fish Biology , 48 : 633 – 642 . Google Scholar OpenURL Placeholder Text WorldCat Poole W. R. , Reynolds J. D. 1998 . Variability in Growth Rate in European Eel Anguilla anguilla (L.) in a Western Irish Catchment . Biology and Environment: Proceedings of the Royal Irish Academy , 98B : 141 – 145 . Google Scholar OpenURL Placeholder Text WorldCat Poole W. R. , Reynolds J. D. , Moriarty C. 1990 . Observations on the Silver Eel Migrations of the Burrishoole River System, Ireland, 1959 to 1988 . Internationale Revue der Gesamten Hydrobiologie und Hydrographie , 75 : 807 – 815 . Google Scholar Crossref Search ADS WorldCat Poole W. R. , Reynolds J. D. , Moriarty C. 1992 . Age and growth of eel Anguilla anguilla (L.) in oligotrophic streams . Irish Fisheries Investigations: Series A , 36 : 72 – 77 . Google Scholar OpenURL Placeholder Text WorldCat R Core Team . 2018 . R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. Sandlund O. T. , Diserud O. H. , Poole R. , Bergesen K. , Dillane M. , Rogan G. , Durif C. et al. 2017 . Timing and pattern of annual silver eel migration in two European watersheds are determined by similar cues . Ecology and Evolution , 7 : 5956 – 5966 . Google Scholar Crossref Search ADS PubMed WorldCat Scholz, F., and Zhu, A. 2019 . kSamples: K-Sample Rank Tests and their Combinations . R Package Version 1.2-9. Google Scholar OpenURL Placeholder Text WorldCat Shelton A. O. , Satterthwaite W. H. , Beakes M. P. , Munch S. B. , Sogard S. M. , Mangel M. 2013 . Separating intrinsic and environmental contributions to growth and their population consequences . The American Naturalist , 181 : 799 – 814 . Google Scholar Crossref Search ADS PubMed WorldCat Stoessel D. J. , Morrongiello J. R. , Raadik T. A. , Lyon J. P. , Nicol M. D. 2018 . Determinants of year class strength and growth of estuary perch Macquaria colonorum in a highly regulated system . Marine and Freshwater Research , 69 : 1663 – 1673 . Google Scholar Crossref Search ADS WorldCat Tanner S. E. , Vieira A. R. , Vasconcelos R. P. , Dores S. , Azevedo M. , Cabral H. N. , Morrongiello J. R. 2019 . Regional climate, primary productivity and fish biomass drive growth variation and population resilience in a small pelagic fish . Ecological Indicators , 103 : 530 – 541 . Google Scholar Crossref Search ADS WorldCat Tesch F. W. 2003 . The Eel: Biology and Management of Anguillid Eels. Blackwell Science Ltd , Oxford, England . 408 pp . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC van der Sleen P. , Stransky C. , Morrongiello J. R. , Haslob H. , Peharda M. , Black B. A. 2018 . Otolith increments in European plaice (Pleuronectes platessa) reveal temperature and density-dependent effects on growth . ICES Journal of Marine Science , 75 : 1655 – 1663 . Google Scholar Crossref Search ADS WorldCat van Walraven L. , Mollet F. M. , van Damme C. J. G. , Rijnsdorp A. D. 2010 . Fisheries-induced evolution in growth, maturation and reproductive investment of the sexually dimorphic North Sea plaice (Pleuronectes platessa L.) . Journal of Sea Research , 64 : 85 – 93 . Google Scholar Crossref Search ADS WorldCat Vigliola L. , Meekan M. 2009 . The back-calculation of fish growth from otoliths. In Tropical Fish Otoliths: Information for Assessment, Management and Ecology , pp. 174 –- 211 . Ed. by Green B. S. , Mapstone B. D. , Carlos G. , Begg G. A. . Springer , Dordrecht . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Vollestad L. A. , Jonsson B. 1988 . A 13-year study of the population dynamics and growth of the European eel Anguilla anguilla in a Norwegian river: evidence for density-dependent mortality, and development of a model for predicting yield . Journal of Animal Ecology , 57 : 983 – 997 . Google Scholar Crossref Search ADS WorldCat Von Bertalanffy L. 1957 . Quantitative laws in metabolism and growth . The Quarterly Review of Biology , 32 : 217 – 231 . Google Scholar Crossref Search ADS PubMed WorldCat Walsh P. J. D. , Foster G. D. , Moon T. 1983 . The effects of temperature on metabolism of the American eel Anguilla rostrata (LeSueur): compensation in the summer and torpor in the winter . Physiological Zoology , 56 : 532 – 540 . Google Scholar Crossref Search ADS WorldCat Weir G. 1996 . Sheep overgrazing in the Nephin Begs. MSc thesis, Trinity College Dublin, Dublin, Ireland. Weisberg S. , Spangler G. , Richmond L. S. 2010 . Mixed effects models for fish growth . Canadian Journal of Fisheries and Aquatic Sciences , 67 : 269 – 277 . Google Scholar Crossref Search ADS WorldCat Westerberg H. , Sjöberg N. 2015 . Overwintering dormancy behaviour of the European eel (Anguilla anguilla L.) in a large lake . Ecology of Freshwater Fish , 24 : 532 – 543 . Google Scholar Crossref Search ADS WorldCat Wolf P. 1951 . A trap for the capture of fish and other organisms moving downstream . Transactions of the American Fisheries Society , 80 : 41 – 45 . Google Scholar Crossref Search ADS WorldCat Woolway R. I. , Weyhenmeyer G. A. , Schmid M. , Dokulil M. T. , de Eyto E. , Maberly S. C. , May L. , et al. 2019 . Substantial increase in minimum lake surface temperatures under climate change . Climatic Change , 155 : 81 – 94 . Google Scholar Crossref Search ADS WorldCat Yokouchi K. , Daverat F. 2013 . Modeling individual growth trajectories of the female European eel in relation to temperature and habitat-use history in the Gironde River, France . Aquatic Biology , 19 : 185 – 193 . Google Scholar Crossref Search ADS WorldCat Yokouchi K. , Daverat F. , Miller M. , Fukuda N. , Sudo R. , Tsukamoto K. , Elie P. et al. 2018 . Growth potential can affect timing of maturity in a long-lived semelparous fish . Biology Letters , 14 : 20180269 . Google Scholar Crossref Search ADS PubMed WorldCat Zuur A. , Ieno E. I. , Walker N. , Saveliev A. , Smith G. M. 2009 . Mixed Effect Models and Extensions in Ecology with R . Springer , New York . 574 pp. Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Vaughan, L., Brophy, D., O’Toole, C., Graham, C., Maoiléidigh, N. Ó, and Poole, R. 2020. Growth rates in a European eel Anguilla anguilla (L., 1758) population show a complex relationship with temperature over a seven-decade otolith biochronology. – ICES Journal of Marine Science, 00: 000–000. © International Council for the Exploration of the Sea 2021. 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)

Sobradillo, B; Boyra, G; Pérez-Arjona, I; Martinez, U; Espinosa, V

doi: 10.1093/icesjms/fsaa242pmid: N/A

Abstract This study measures the dorsal aspect target strength (TS; dB re 1 m2) and TS–length (standard length, SL; cm) relationships for European anchovy, attained both ex situ and in situ in two different seasons across 7 years in the Bay of Biscay. The measurements were made at three frequently used acoustic frequencies (38, 120, and 200 kHz). A backscattering model for physostome fish was utilized to help interpret the results. The obtained experimental mean TS for anchovies with an SL of 3.5–19.5 cm was −44.6 (±2.3), −46.9 (±3), and −48.4 (±2.7) dB at 38, 120, and 200 kHz, respectively, yielding b20 values of −66.4, −68.7, and −70.4 dB, respectively. The results were consistent across seasons and between in situ and ex situ conditions, presenting TS–length relationships with statistically significant slopes (p-values <0.05) for all frequencies. This research represents part of a series of efforts planned to obtain a comprehensive TS vs. length and depth relationship to update the acoustic assessment methodology of European anchovy in the Bay of Biscay. Introduction European anchovy (Engraulis encrasicolus; Linnaeus, 1758) is one of the main commercial species in the Bay of Biscay, supporting profitable fisheries for both the Spanish and French fleets. Stock assessment of this resource is based on the so-called catch-Bayesian biomass-based model (CBBM) (Ibaibarriaga et al., 2008; ICES, 2015), which depends on the internationally coordinated scientific advice. This advice is based in an abundance index of the adult stock abundance, derived from a combination of commercial and fisheries independent information. The scientific surveys that contribute to the CBBM are BIOMAN (Massé et al., 2018); based on the daily egg production method in spring; and two acoustic-trawl surveys, PELGAS (Massé et al., 2018) in the spring and JUVENA (Boyra et al., 2013) in autumn. These surveys’ methodologies are discussed and evaluated annually at the International Council for the Exploration of the Sea Working Group of Acoustics and Eggs (ICES WGACEGG) and the results are synthetized at the ICES Working Group on Southern Horse Mackerel, Anchovy and Sardine (ICES, 2017) to produce the CBBM index. Acoustic surveys are considered effective methods for quantifying the distribution and abundance of many pelagic marine fauna (Simmonds and Maclennan, 2005). In most cases the echo-integration technique is used to estimate fish density (MacLennan et al., 1990), necessitating information regarding the dorsal aspect target strength (TS; dB re 1 m2) (Maclennan et al., 2002) of the fish that contributes to the received signal (Jech and Horne, 2001). The TS is a measure of the proportion of the incident intensity that is backscattered by the target (Maclennan et al., 2002). To translate acoustic density measurements into biologically more meaningful measures, such as biomass or abundance, the log-linear relationship between the standard length (SL; cm) of the fish and the backscattered acoustic energy is commonly used. This TS to length (TS–L) equation can be expressed as: TS=a log 10(SL)+b,(1) where the slope, a, and the intercept, b, are generally assumed to be species-specific constants. In the case of physostomous fish such as anchovy, a is normally close to 20 (Love, 1977; Foote, 1980) and (1) is often replaced by: TS=20 log 10(SL)+b20.(2) The acoustic surveys that are currently used to estimate anchovy biomass in the Bay of Biscay utilise different b20 values at 38 kHz: the French survey PELGAS uses −71.2 dB, originally obtained from 19 to 35 cm herring Clupea harengus (ICES, 1982), while the Spanish surveys JUVENA and PELACUS use −72.6 dB from 8 to 32 cm herrings (Degnbol et al., 1985). The lack of unique and ad hoc TS–length relationship for Bay of Biscay anchovy presents an obstacle for the development of an absolute index of abundance for this species. This has been acknowledged by ICES WGACEGG, with the attainment of a common TS–length relationship for the region deemed one of the key objectives of the working group (ICES, 2013). Generally, in situ TS measurements are assumed to deliver accurate results when collected with concurrent reliable biological samples and tilt angle information (Torgersen and Kaartvedt, 2001; Madirolas et al., 2016; Zare et al., 2017). However, measuring the TS of fish in their natural environment may be accompanied by difficulties that result in biased TS values. In particular, during the daytime small pelagic fish such as European anchovy aggregate in schools that are too densely packed to resolve individual targets (Sawada et al., 2009). Nevertheless, various strategies exist to overcome this problem. One mitigation tactic is to lower transducers closer to the fish targets, thus reducing the sampling volume (Ona, 2003; Kang et al., 2009; Sawada et al., 2009; Murase et al., 2011; Fernandes et al., 2016). In the Bay of Biscay, a variant of this technique has been applied to estimate the TS of anchovy (Doray et al., 2016), but the methodology stimulated a change in fish behaviour that seemed to positively bias the mean TS values. Other strategies involve working at night, when most species disperse and migrate near the surface (Glass, 2000). These are most useful when a study area is dominated by the target species (Foote et al., 1987; Barange et al., 1996; Peltonen and Balk, 2005; Zhao et al., 2008) but might prove problematic where there are numerous fish species or in the presence of high abundances of plankton. An alternative strategy is to conduct ex situ experiments (Kang and Hwang, 2003; Kang et al., 2009), providing greater control over density and ensuring the isolation of the target species. However, ex situ experiments pose some concerns such as potentially altering the behaviour of the targets (and hence potentially biasing the mean TS values) or the close proximity of the targets to the transducers, which may cause short-range problems [i.e. TS dependence with distance related to measurements performed inside the near field of the transducer (Simmonds and Maclennan, 2005; Foote, 2014; Chu and Eastland, 2015; Pérez-Arjona et al., 2018), or the extended size of fish that produces an uncertainty in the position inside the beam measured by split-beam echosounders (Kieser et al., 2000)]. To gain insights regarding the influence of behaviour on TS, modelling techniques (Fujino et al., 2009) have often been used in combination with ex situ TS measurements (Henderson and Horne, 2007; Sawada et al., 2011). The objective of this work is to measure the TS values and model TS–length relationships for European anchovy at 38, 120, and 200 kHz frequencies, using acoustic data collected both in situ and ex situ. A reduction in potential multiple target bias was attempted by working during the night and applying a high-density filter (Gauthier and Rose, 2001) to the in situ-measured data as well as controlling the influence of the different sampling volumes related to the pulse durations used at the different ex situ experiments. In addition, an ad hoc calibration experiment was conducted to correct the possible bias derived from the short range of some of the ex situ measurements that were done in the transition range between near and far field. Finally, a backscattering model based on the method of fundamental solutions (MFS) (Fairweather et al., 2003; Pérez-Arjona et al., 2018, 2020a) for physostomous fish, simulating the swimbladder as two-chambered prolate spheroids (Andreeva, 1974; Weston, 1966; Love, 1978; Furusawa, 1988; Ye, 1997) plus the backbone, was utilised to help interpret the empirical results. Most of the numerical methods considered for the simulation of TS values are solely valid when estimating TS in the far field of both the emitting transducer and the scatterer (fish) (Jech et al., 2015). Only the finite element method (FEM) (Lilja et al., 2004) and the boundary-element method (BEM) (Foote and Francis, 2002) provide alternatives at arbitrary close distances, but they (especially FEM) have a high and perhaps even unaffordable computational cost. The MFS is a meshless method that has proved to be useful in estimating the measurable TS of fish and the contributions of the different inner structures of fish to TS. It has similar or even greater accuracy than FEM or BEM (Godinho et al., 2012), yet with reduced computational costs, which is a consideration that is especially important when examining fish models with additional fish structures to a swim bladder (e.g. a fish backbone) (Pérez-Arjona et al., 2018). Material and methods In situ data collection Acoustic-trawl data were collected from 2010 to 2017 via two scientific surveys in the Bay of Biscay at two different times of year (Figure 1). JUVENA (Boyra et al., 2013) was conducted in September and mainly focused on the juvenile component of the anchovy population, while BIOMAN (Santos et al., 2016) was undertaken in May (during the peak of the spawning season) and examined the adult component. Two scientific research vessels were used in each survey: RV “Ramón Margalef” (RM, hereafter) and RV “Emma Bardán” (EB, hereafter). Both collected continuous acoustic data using a Simrad EK60 scientific echosounder, with split-beam transducers of 38 kHz (ES38B-7), 120 kHz (ES120-7C), and 200 kHz (200-7C). The echosounders were calibrated at least once a year, typically at the beginning of the survey following standard procedures (Demer et al., 2015), with intercalibration exercises carried out each year between the two vessels following a standard methodology (Simmonds and Maclennan, 2005). The most relevant calibration parameters of the in situ measurements are described in Supplementary Table S1. Figure 1. Open in new tabDownload slide Area of study in the Bay of Biscay, with the locations where the TS measurements were performed, distinguishing the fishing trawls conducted for the in situ measurements in both surveys and the location of the cage for the ex situ ones. Figure 1. Open in new tabDownload slide Area of study in the Bay of Biscay, with the locations where the TS measurements were performed, distinguishing the fishing trawls conducted for the in situ measurements in both surveys and the location of the cage for the ex situ ones. Ground truth trawl hauls were performed based on the interpretation of the echograms, aiming to identify fish species and to determine their size distribution across the whole area of study. Both vessels performed trawl hauls during JUVENA, but only EB performed trawl hauls during BIOMAN due to the regular activities that take place on the RM in association with the daily egg production method. Trawl samples were obtained using a Gloria HOD 352 pelagic trawl of 15-m vertical opening, with a 10-mm mesh size (bar length) at the cod end. Fishing trawls were performed during the day and the night, between 5- and 300-m depth and at a mean speed of 4 knots. Acoustic data recorded during trawl hauls with predominance (>85%) of anchovy in the catch were selected for TS analysis (Supplementary Table S2). The typical recording range was 200 m, yielding 2–3 pings per second. Lengths were obtained from a random sample of >50 individuals of each haul and measured to 0.5-cm SL classes onboard the research vessel. Ex situ data collection Ex situ TS measurements were obtained from two sets of anchovy individuals, captured in December 2011 (set 1) and July 2013 (set 2) in the Bay of Biscay. Both were captured by the purse seiner Itsas Lagunak and transported in live bait fishing tanks onboard the vessel. The first set comprised 120 anchovies that were kept in water tanks (1 m depth × 3 m diameter) in the Aquaculture School of Mutriku for eight months before being moved to the sea culture cage at the mouth of Mutriku harbour (Gipuzkoa, Spain; 43°18′N, 02°22′W) (Figure 1). The cage was cylindrical with ∼8 m depth and 16 m diameter and a mesh size of 0.4 cm. The second set consisted of ∼5000 anchovies and was transported directly from the purse seiner tanks to the harbour cage. After being moved to the cage, anchovies were left at least two days to settle before the experiments commenced. Unfortunately, after dropping the second set of anchovies to the cage, we noticed that it was contaminated by number with ∼2% of horse mackerel. A diver visually inspected the cage periodically to maintain, feed and monitor the fish. Two groups of measurements (N1, N2) were carried out using the first set of anchovies and another (N3) using the second one. At the end of each set of measurements, 50 specimens were weighed and measured for SL. TS measurements were made using a three-frequency (38, 120, and 200 kHz) Simrad EK60 split-beam scientific echosounder system, installed on a floating 0.6 m × 0.6 m platform ∼20 cm below the sea surface. The floating platform was placed about halfway (∼4 m) between the centre and the border of the cage (Figure 2a). The platform was connected to a logistics boat that housed the ancillary electronic equipment and the 12-V batteries used as the power source. Day- and night-time data were registered during the study, but following a preliminary inspection of the data, only night experiments were used in the analysis. Daytime data yielded significantly higher TS values, probably owing to the greater packing densities reported by the diver and hence likely subjected to a higher probability of detecting unresolved multiple echoes. Figure 2. Open in new tabDownload slide (a) Scheme of the experimental set up in the cage. Transducers were installed in a floating platform near the surface pointing downwards towards the scattered anchovies. Different regions are highlighted with reference to the different acoustic fields (near, near-to-far and far field; shown in B) of the used 38 kHz transducer. The dashed lines mark the critical ranges (Rc and 4Rc) that divide the three regions. The TS analysis was done using decreasing pulse durations to achieve increasing sampling resolution (note that neither anchovies nor sampling volumes are at proportional scale). Targets from the grey areas (i.e. depth <2 and >6 m) were not used for analysis. (b) Theoretical range-dependence of the pressure wave of a circular piston simulating the geometry of a Simrad ES38B transducer (see the text for details), showing both the exact solution (black line) and the far field aproximation (i.e. spherical decay, grey line). The different ranges at which the sphere was calibrated during the near-to-far field calibration experiment are also marked (black circles). The calibration values at each range obtained in this experiment were used to correct the anchovy TS measurements of the corresponding range in the cage. Figure 2. Open in new tabDownload slide (a) Scheme of the experimental set up in the cage. Transducers were installed in a floating platform near the surface pointing downwards towards the scattered anchovies. Different regions are highlighted with reference to the different acoustic fields (near, near-to-far and far field; shown in B) of the used 38 kHz transducer. The dashed lines mark the critical ranges (Rc and 4Rc) that divide the three regions. The TS analysis was done using decreasing pulse durations to achieve increasing sampling resolution (note that neither anchovies nor sampling volumes are at proportional scale). Targets from the grey areas (i.e. depth <2 and >6 m) were not used for analysis. (b) Theoretical range-dependence of the pressure wave of a circular piston simulating the geometry of a Simrad ES38B transducer (see the text for details), showing both the exact solution (black line) and the far field aproximation (i.e. spherical decay, grey line). The different ranges at which the sphere was calibrated during the near-to-far field calibration experiment are also marked (black circles). The calibration values at each range obtained in this experiment were used to correct the anchovy TS measurements of the corresponding range in the cage. Measurements at the cage were performed using different pulse durations (64, 128, 256, and 512 µs) to check whether the values obtained varied at increasing sampling volumes (i.e. decreasing vertical resolutions) due to the greater failure probability of the single target discrimination algorithm for larger volumes (see below). Calibrations were made following standard procedures (Demer et al., 2015) and were repeated for all pulse durations and power settings (Table 1). Table 1. Calibration settings of the ex situ data, performed at 4 m from the transducer. Year . Experiment . Frequency . Transducer radius . Pulse duration . Power . Gain . Sa correction . code . (kHz) . (m) . (μs) . (W) . (dB) . (dB) . 2012 N1 38 0.21 256 800 23.62 −0.66 120 0.08 256 200 23.55 −0.59 200 0.05 256 180 25.74 −0.44 2012 N2 38 0.21 256 800 23.51 −0.65 120 0.08 64 200 25.23 −0.58 200 0.05 64 180 24.74 −0.67 2013 N3 38 0.21 512 800 25.4 −0.75 120 0.08 256 200 26.63 −0.61 200 0.05 128 180 26.07 −0.76 Year . Experiment . Frequency . Transducer radius . Pulse duration . Power . Gain . Sa correction . code . (kHz) . (m) . (μs) . (W) . (dB) . (dB) . 2012 N1 38 0.21 256 800 23.62 −0.66 120 0.08 256 200 23.55 −0.59 200 0.05 256 180 25.74 −0.44 2012 N2 38 0.21 256 800 23.51 −0.65 120 0.08 64 200 25.23 −0.58 200 0.05 64 180 24.74 −0.67 2013 N3 38 0.21 512 800 25.4 −0.75 120 0.08 256 200 26.63 −0.61 200 0.05 128 180 26.07 −0.76 Note that the 200-kHz gain values differ from the observed increasing trend with time because there were two different 200-kHz transducers used. Open in new tab Table 1. Calibration settings of the ex situ data, performed at 4 m from the transducer. Year . Experiment . Frequency . Transducer radius . Pulse duration . Power . Gain . Sa correction . code . (kHz) . (m) . (μs) . (W) . (dB) . (dB) . 2012 N1 38 0.21 256 800 23.62 −0.66 120 0.08 256 200 23.55 −0.59 200 0.05 256 180 25.74 −0.44 2012 N2 38 0.21 256 800 23.51 −0.65 120 0.08 64 200 25.23 −0.58 200 0.05 64 180 24.74 −0.67 2013 N3 38 0.21 512 800 25.4 −0.75 120 0.08 256 200 26.63 −0.61 200 0.05 128 180 26.07 −0.76 Year . Experiment . Frequency . Transducer radius . Pulse duration . Power . Gain . Sa correction . code . (kHz) . (m) . (μs) . (W) . (dB) . (dB) . 2012 N1 38 0.21 256 800 23.62 −0.66 120 0.08 256 200 23.55 −0.59 200 0.05 256 180 25.74 −0.44 2012 N2 38 0.21 256 800 23.51 −0.65 120 0.08 64 200 25.23 −0.58 200 0.05 64 180 24.74 −0.67 2013 N3 38 0.21 512 800 25.4 −0.75 120 0.08 256 200 26.63 −0.61 200 0.05 128 180 26.07 −0.76 Note that the 200-kHz gain values differ from the observed increasing trend with time because there were two different 200-kHz transducers used. Open in new tab In addition to the regular calibration at 4 m range, an additional calibration experiment was conducted to determine and correct the bias effect of working at distances where the far field condition of the corresponding transducer was not achieved (see below). Near-to-far field calibration experiment Due to the limited cage dimensions (Figure 2), some of the acoustical measurements were done closer than the far field distance of the 38-kHz transducer, which posed some initial concerns about the validity of these measurements. Depending on the distance from the transducer face, three main regions can be distinguished: the near field, the far field and a near-to-far transition field (Figure 2). The near field is the region of distances closer than RC, the critical range or Fresnel distance, which for the corresponding flat piston transducer is defined as RC= a2/λ, being a the transducer radius and λ its operative wavelength (Medwin and Clay, 1998; Foote, 2014). Inside the near field (i.e. at distances less than 1.14, 0.46, and 0.33 m, respectively, for the 38, 120, and 200 kHz transducers used in this work; radius of each transducer shown in Table 1) the amplitude oscillates with distance (Figure 2) and quantitative acoustic measures are problematic. Immediately beyond RC lies the near-to-far transition field, where the amplitude does not oscillate with distance and the spreading beam directivity pattern is stable, with the on-axis amplitude decreasing monotonously but not yet following spherical spreading. Finally, beyond the transition zone is the far field, where the amplitude decreases spherically with range. The far field starts somewhere between two to four times RC (depending on the maximum deviation allowed from spherical spreading (Foote, 2014)). In this work, taking a rather conservative assumption (4RC), the far field condition was accomplished at different distances for each frequency: 4.55 m at 38 kHz, 1.85 m at 120 kHz, and 1.33 m at 200 kHz. Therefore, as only targets from 2 to 6 m depth were analysed (Figure 2), only those performed at 38 kHz were susceptible to be biased by this effect. Moreover, since the considered measuring distances were greater than RC (1.14 m at 38 kHz) but smaller than 4Rc, measurements followed a stable pattern. It is for this reason that the corrections can be applied not only at the calibration points but also interpolated to intermediate distances. To test the validity of these data and, when necessary, to obtain an extra depth-dependent gain correction, an ad hoc calibration experiment was carried out, in which the TS of a reference target (a tungsten carbide sphere of 38.1 mm diameter) was measured on axis at different distances from a 38 kHz transducer (2, 3, 4, 6, and 12 m), i.e. covering distances from the near-to-far to the far field (Figure 2b). The TS of the sphere was first measured at 12 m (i.e. well beyond the far field range and, thus, where the target measurements were expected to be free from bias caused by range) and then at all the other ranges. We simulated the pressure waves emitted by a 38 kHz transducer using the analytical solution of the pressure radiation of a circular flat piston (Kinsler et al., 1999, pp. 181–185) at different ranges, to visualize (Figure 2b) and help interpret the results of the near-to-far field experiment. Data processing Target selection In situ data To separate swimbladdered fish from fluid-like organisms or macro-zooplankton, a bi-frequency algorithm (Ballón et al., 2011; Lezama-Ochoa et al., 2011) was used on the in situ experiments, based on the differences in mean volume backscattering strength (Sv, dB re 1 m2/m3) at 38 and 120 kHz. It was possible to use this algorithm because of the high percentage of beam overlap within the in situ measurements ranging from 60 to 86% (see here an interactive application illustrating the beam overlap calculations). A binary matrix was created from the data selected by the algorithm and was applied as a mask to the three frequencies of the study using Echoview Software Pty Ltd, 2013, version 5.2 (Echoview Software, 2013). All the echograms were visually inspected to check the correct performance of the algorithm (Figure 3). The TS values were then derived from the selected echoes using a single target detection algorithm (Soule, 1997; Ona and Barange, 1999), the values of which are presented in Table 2. Figure 3. Open in new tabDownload slide Top panel: example of 38 kHz TS echogram showing 10 min of one of the ex situ experiments (N1) targeting ∼10 cm anchovies at 256 µs pulse duration and −70 dB threshold. Anchovies and plankton are easily distinguished by intensity and length of the echotraces. Bottom panels: example of 38 kHz TS echogram showing 2.5 min of the 165 037 in situ haul targeting ∼12 cm anchovies at 1024 µs pulse duration and −70 dB threshold. The left side shows the raw TS echogram where the anchovies (yellow-red echoes) are surrounded by a plankton layer (grey-blue echoes); and the right side shows the same 2.5 min of acoustic data at the same threshold but after the plankton filtering process, where only fish echoes remain (see the text for details). Figure 3. Open in new tabDownload slide Top panel: example of 38 kHz TS echogram showing 10 min of one of the ex situ experiments (N1) targeting ∼10 cm anchovies at 256 µs pulse duration and −70 dB threshold. Anchovies and plankton are easily distinguished by intensity and length of the echotraces. Bottom panels: example of 38 kHz TS echogram showing 2.5 min of the 165 037 in situ haul targeting ∼12 cm anchovies at 1024 µs pulse duration and −70 dB threshold. The left side shows the raw TS echogram where the anchovies (yellow-red echoes) are surrounded by a plankton layer (grey-blue echoes); and the right side shows the same 2.5 min of acoustic data at the same threshold but after the plankton filtering process, where only fish echoes remain (see the text for details). Table 2. Parameters used in the single target detection algorithm. . Units . In situ . Ex situ . TS threshold dB −70 −70 Filter angles (major/minor) Degrees 3.5 3.5 Min. pulse length ms 0.7 0.7 Max. pulse length ms 1.5 1.5 Max. beam compensation dB 6 6 Max. angle SD (minor/major) Degrees 0.6 0.2 . Units . In situ . Ex situ . TS threshold dB −70 −70 Filter angles (major/minor) Degrees 3.5 3.5 Min. pulse length ms 0.7 0.7 Max. pulse length ms 1.5 1.5 Max. beam compensation dB 6 6 Max. angle SD (minor/major) Degrees 0.6 0.2 Open in new tab Table 2. Parameters used in the single target detection algorithm. . Units . In situ . Ex situ . TS threshold dB −70 −70 Filter angles (major/minor) Degrees 3.5 3.5 Min. pulse length ms 0.7 0.7 Max. pulse length ms 1.5 1.5 Max. beam compensation dB 6 6 Max. angle SD (minor/major) Degrees 0.6 0.2 . Units . In situ . Ex situ . TS threshold dB −70 −70 Filter angles (major/minor) Degrees 3.5 3.5 Min. pulse length ms 0.7 0.7 Max. pulse length ms 1.5 1.5 Max. beam compensation dB 6 6 Max. angle SD (minor/major) Degrees 0.6 0.2 Open in new tab Ex situ data Due to the short depth range available in the experimental cage, the overlapping volume between the 38- and 120-kHz frequencies (42%) was not enough to apply the bi-frequency algorithm (link to the interactive beam overlap illustration). Therefore, it could not be used with the ex situ data. Instead, echograms were manually scrutinized to isolate anchovies from unwanted signals based on the intensity and the length of the echo-traces. There were two clearly distinguishable types of traces: intense (TS > −60 dB) and short (<4 consecutive pings, representing typical swimming velocities of small pelagic fish between 7 and 20 cm/s in most cases), attributed to anchovy, and less intense (<−60 dB) and long (typically between 20 and 100 pings, representing velocities between 0.2 and 2 cm/s) were attributed to slow-swimming (presumably planktonic) scatterers. Multiple targets To prevent the inclusion of unwanted multiple targets, two different procedures were followed depending on the type of data collected. In situ To reduce the multiple target bias associated with the in situ data, a high-density filter (Gauthier and Rose, 2001) was applied, for which the density threshold was empirically determined. The echogram of each haul was gridded in cells with a horizontal size of 5 pings (1 ping ≈ 1.8 m) and a vertical length of 5 m. The total number of fish per acoustic reverberation volume (Nv) (Sawada et al., 1993; Ona and Barange, 1999) was then plotted against the number of single targets per sample volume (Tv) in each cell. As in previous studies (Boyra et al., 2018, 2019), the maximum of Tv on Nv was used to determine the target density at which multiple target echoes are likely to be produced. However, in this study, rather than having a maximum, the Tv values were found to monotonously increase for small values of Nv until reaching stabilization at Nv values close to 1. This may have been due to the lower packing density observed, related to the typical dispersed night-time distribution of anchovy. The in situ data set was subsequently limited to cells with Nv values below 1. After filtering the high-density areas, a final quality check consisted on removing the hauls that presented either two similarly pronounced modes or fewer than five targets. Ex situ Measurements at the cage were performed at different pulse durations (Figure 2a). If the proximity between anchovies in the cage caused unresolved multiple targets, the probability of occurrence (and hence the mean TS) was expected to increase at higher pulse durations. In addition, the single target detection performance was tested for values of maximum standard phase deviation between 0.2° and 0.9° (Ona and Barange, 1999). Table 3 summarizes the different procedures performed during the in situ and ex situ data collection and processing. Table 3. Methodological differences between the in situ and ex situ data collection and processing. . In situ . Ex situ .  Sampling period 2010–2017 2012–2013  Season Autumn, spring Winter  Age Juvenile (5.8–16.3 cm) Adult (10–15.94 cm) Juvenile (10.1–10.9 cm)  Survey metdodology Daily Egg Production Metdod and acoustics Acoustics Data processing  Target selection Bi-frequency algoritdm (Ballón et al., 2011; Lezama-Ochoa et al., 2011) Manual selection of anchovy echo signals  Single target detection algoritdm parameters: standard deviation of minor/major axis 0.6° 0.2°  Extra multiple targets filtering High-density filter (Gautdier and Rose, 2001) Measurements at different sampling volumes (pulse durations) Data collection . In situ . Ex situ .  Sampling period 2010–2017 2012–2013  Season Autumn, spring Winter  Age Juvenile (5.8–16.3 cm) Adult (10–15.94 cm) Juvenile (10.1–10.9 cm)  Survey metdodology Daily Egg Production Metdod and acoustics Acoustics Data processing  Target selection Bi-frequency algoritdm (Ballón et al., 2011; Lezama-Ochoa et al., 2011) Manual selection of anchovy echo signals  Single target detection algoritdm parameters: standard deviation of minor/major axis 0.6° 0.2°  Extra multiple targets filtering High-density filter (Gautdier and Rose, 2001) Measurements at different sampling volumes (pulse durations) Data collection Open in new tab Table 3. Methodological differences between the in situ and ex situ data collection and processing. . In situ . Ex situ .  Sampling period 2010–2017 2012–2013  Season Autumn, spring Winter  Age Juvenile (5.8–16.3 cm) Adult (10–15.94 cm) Juvenile (10.1–10.9 cm)  Survey metdodology Daily Egg Production Metdod and acoustics Acoustics Data processing  Target selection Bi-frequency algoritdm (Ballón et al., 2011; Lezama-Ochoa et al., 2011) Manual selection of anchovy echo signals  Single target detection algoritdm parameters: standard deviation of minor/major axis 0.6° 0.2°  Extra multiple targets filtering High-density filter (Gautdier and Rose, 2001) Measurements at different sampling volumes (pulse durations) Data collection . In situ . Ex situ .  Sampling period 2010–2017 2012–2013  Season Autumn, spring Winter  Age Juvenile (5.8–16.3 cm) Adult (10–15.94 cm) Juvenile (10.1–10.9 cm)  Survey metdodology Daily Egg Production Metdod and acoustics Acoustics Data processing  Target selection Bi-frequency algoritdm (Ballón et al., 2011; Lezama-Ochoa et al., 2011) Manual selection of anchovy echo signals  Single target detection algoritdm parameters: standard deviation of minor/major axis 0.6° 0.2°  Extra multiple targets filtering High-density filter (Gautdier and Rose, 2001) Measurements at different sampling volumes (pulse durations) Data collection Open in new tab TS–length and TS–depth relationships Averaged TS values from each haul or cage experiment were plotted against mean length in the logarithmic scale to determine the TS–L relationship from (1) and (2) at the three frequencies. Linear regression models were fit at each frequency, providing probability values and coefficients of determination to measure the goodness of the relationships (p-values <0.05 were considered statistically significant). Obtained TS–length relationships were also compared with those predicted by the MFS model (see below), analysing the agreement between empirical and modelled data in terms of the pairs of mean and standard deviation of fish tilt angle used in the model as a proxy of anchovy behaviour. Finally, TS vs. length and depth relationships were investigated by means of linear models. To avoid the violation of independence of the explanatory variables we studied possible collinearity between depth and length. Acoustic scattering model To interpret the measured TS values, a numerical model of acoustic scattering for European anchovy was applied using the MFS (Fairweather et al., 2003; Pérez-Arjona et al., 2018). Anchovy is a physostomous fish with a dual-chambered swimbladder, which the simulation simplified as two-chambered prolate spheroids (Andreeva, 1974; Weston, 1966; Love, 1978; Furusawa, 1988; Ye, 1997) (PS1 and PS2). The major axis of the PS1 was considered orthogonal with respect to the incident acoustic pulse when fish was swimming horizontally and the major axis of PS2 was tilted α with respect to the major axis of PS1 (Figure 4). The model also considered the backbone’s contribution, which was expected to attenuate the swimbladder signal and/or modify the far-field distance achievement, while discarding the flesh’s contribution which is usually less significant for swimming directions close to the horizontal (Pérez-Arjona et al., 2020b). The prolate spheroids’ dimensions and tilt angle (Figure 4) were based on soft X-ray images (IntechForView CR system). Figure 4. Open in new tabDownload slide Lateral radiograph of a specimen of Engraulis encrasicolus showing the two connected swimbladder chambers (PS1 and PS2) and the backbone. Fish length = 10.1 cm, fish height = 1.04 cm. Figure 4. Open in new tabDownload slide Lateral radiograph of a specimen of Engraulis encrasicolus showing the two connected swimbladder chambers (PS1 and PS2) and the backbone. Fish length = 10.1 cm, fish height = 1.04 cm. Twelve good quality, freshly thawed individuals were radiographed and their swimbladder condition was visually examined from the digitized images. Individuals with ruptured or disfigured bladders were discarded, with the four individuals that appeared to be undamaged ultimately used for the morphological measurements (Figure 4). Calculations were carried out for mean SL = 10.5 cm, with corresponding PS1 and PS2 dimensions given by length (semi-major axes, a1 = 0.625 cm and a2 =0.5 cm), height (semi-minor axes, b1 = 0.2 cm and b2 = 0.2 cm), width (semi-minor axes c1 = b1 and c2 = b2), and relative angle β = 12 degrees, being β the tilt angle with respect to the fish body axis (Figure 5). The two-chambered swimbladder was deemed a pressure-release surface. The fish backbone was modelled as a fluid-filled (ρ = 1100 kg/m3, c = 2270 m/s) (Gorska et al., 2005; Pérez-Arjona et al., 2018) straight cylinder (length = 9 cm and radius = 1 mm) with smooth edges surrounded by a homogeneous host medium (seawater with acoustical properties ρ=1026 kg/m3, c = 1490 m/s). Figure 5. Open in new tabDownload slide Scheme of the TS directivity TS(α), being α the tilt angle of fish body axis with respect to the horizontal. The emitting and receiving echosounder position are fixed and depicted in the diagram. The echosounder’s emitted beam axis is orthogonal to the fish body axis for α = 0°. Figure 5. Open in new tabDownload slide Scheme of the TS directivity TS(α), being α the tilt angle of fish body axis with respect to the horizontal. The emitting and receiving echosounder position are fixed and depicted in the diagram. The echosounder’s emitted beam axis is orthogonal to the fish body axis for α = 0°. MFS was used to solve the three-dimensional Helmholtz equation in the frequency domain (Fairweather et al., 2003). For the 3D case, assuming a point source placed within the propagation domain, at point x0, it is possible to establish fundamental solutions of the Helmholtz equation G, for the sound pressure, and H, for the particle velocity, at a point x, which can be written respectively as: G3D(x,x0,k)=e-ikrr,(3) H3D(x,x0,k,n→)=1−iρω(−ikr−1)e−ikrr2∂r∂n→.(4) In these equations, r corresponds to the distance between the source point and the domain point, given; n→ represents the direction along which the particle velocity is calculated, k=ω/c the wave number, ω=2πf the angular frequency, f the frequency and c the sound propagation velocity within the acoustic medium and the medium density. The basic principle of the MFS (Fairweather et al., 2003; Pérez-Arjona et al., 2018) is that the sound field in a homogeneous region can be simulated by the linear superposition of the effects of a number of virtual sources, each one with its own amplitude. To define the formulation of this problem, let us first consider three sets of virtual sources located: the first within the swimbladder, with NS1 sources; the second within the spine, with NS2 sources; the third in the water, distributed around the spine, also with NS2 sources. The first two sets will allow the simulation of the sound field in the host medium, which, in that case, can be written as p(x,k)Ω1=∑j=1NS1PjG3D(x,x1,j,k1)+∑j=1NS2QjG3D(x,x2,j,k1)+pinc(x,xsource,k1) for x in Ω1,(5) while within the spine the acoustic pressure is given as p(x,k2)Ω2=∑j=1NS3RjG3D(x,x3,j,k2) for x in Ω2.(6) In (2) and (3), Pj, Qj, and Rj are unknown amplitudes of the virtual sources, pinc(x,xsource,k) represents the incident field generated by a source located at xsource and k1 and k2 represent the wavenumber in the host medium (water) and in the spine. To determine the relevant amplitudes, a system of equations must be established by imposing the necessary interface and boundary conditions. A complete description of the MFS application to TS calculation, including the mathematical description of the complete equations system can be found in Pérez-Arjona et al. (2018). To accurately resolve the acoustic wave the MFS virtual sources were considered and located in a number of at least six sources per wavelength. For the sake of comparison with standard experimental measurements, the simulated emitted acoustic field was assumed to be the corresponding to the analytical solution of the pressure created by a circular piston in its far field (Medwin and Clay, 1998), as an idealization of a scientific echosounder transducer. The transducer size was chosen to produce a half-beam angle at −3 dB of 3.5°, following the specifications of Simrad EK60 scientific echosounders at working frequencies. The MFS model was solved at the three frequencies used for the measurements: 38, 120, and 200 kHz. Moreover, convergence tests were carried out for each frequency to guarantee the proper density mesh. The TS directivity was calculated for dorsal incidence with fish tilt angle from α = −90 to α = 90° (Figure 5). To study possible fish swimming orientation, an optimization procedure was applied by running the model with tilt angles following sequences of normal distributions. Simulated TS values obtained using mean tilt angles from −10° to 10° and standard deviation from −20° to 20° (in increments of 1°) were compared to the experimental TS–L relationships, analysing the tilt angles that produced the best agreement. Results Near-to-far field experiment The near-to-far field calibration experiment showed that, when using a gain value obtained by calibrating at the far field, the mean TS measured in the far field were the correct nominal TS value of the sphere (−42.3 dB). However, the mean TS measured at ranges closer than four times the critical range were positively biased by different margins. When using gain values obtained by calibration at each range (Table 4), we were able to measure the same mean unbiased TS values of the target at all ranges. This showed that it was possible to correctly measure the TS of targets (for example anchovies) in the transition from the near to the far field, by applying calibration values at proper ranges (Figure 6). The gain differences obtained between near-to-far and far field were compared against the analytical solution for the circular piston and were then used to apply the corrections to the ex situ anchovy TS measurements. Figure 6. Open in new tabDownload slide Result of the near-to-far field calibration experiment, using a pulse duration of 256 μs. The upper figure shows a scatterplot (grey circles) of measured TS values of the sphere at 38 kHz at different horizontal distances from the transducer face. Black circles mark the mean TS value at each range and the grey dotted line, the nominal TS value of the tungsten carbide sphere at this frequency (−42.3 dB). (a) The values obtained when using for all distances a single set of calibration parameters obtained at 12 m (i.e. well inside the far field). (b) The measured TS values of the sphere using for each distance the calibration parameters obtained at each range (i.e. corrected for the near-to-far field effect). In the figure below, calibrated gain values against range applied in cases a and b are shown. Figure 6. Open in new tabDownload slide Result of the near-to-far field calibration experiment, using a pulse duration of 256 μs. The upper figure shows a scatterplot (grey circles) of measured TS values of the sphere at 38 kHz at different horizontal distances from the transducer face. Black circles mark the mean TS value at each range and the grey dotted line, the nominal TS value of the tungsten carbide sphere at this frequency (−42.3 dB). (a) The values obtained when using for all distances a single set of calibration parameters obtained at 12 m (i.e. well inside the far field). (b) The measured TS values of the sphere using for each distance the calibration parameters obtained at each range (i.e. corrected for the near-to-far field effect). In the figure below, calibrated gain values against range applied in cases a and b are shown. Table 4. Calibration parameters obtained from the TS measurements of the sphere in the near-to-far-field (ntff) at 38 kHz. Depth . Gain_ntff . SACorr_ntff . (m) . (dB) . (dB) . 2 22.4 −0.2 3 22.7 −0.2 4 22.4 −0.2 6 22.3 −0.2 12 22.2 −0.2 Depth . Gain_ntff . SACorr_ntff . (m) . (dB) . (dB) . 2 22.4 −0.2 3 22.7 −0.2 4 22.4 −0.2 6 22.3 −0.2 12 22.2 −0.2 Power = 800 W, pulse duration = 256 s. Open in new tab Table 4. Calibration parameters obtained from the TS measurements of the sphere in the near-to-far-field (ntff) at 38 kHz. Depth . Gain_ntff . SACorr_ntff . (m) . (dB) . (dB) . 2 22.4 −0.2 3 22.7 −0.2 4 22.4 −0.2 6 22.3 −0.2 12 22.2 −0.2 Depth . Gain_ntff . SACorr_ntff . (m) . (dB) . (dB) . 2 22.4 −0.2 3 22.7 −0.2 4 22.4 −0.2 6 22.3 −0.2 12 22.2 −0.2 Power = 800 W, pulse duration = 256 s. Open in new tab These calibration results were in general agreement with the analytical solution of a circular flat piston pressure radiation. As explained below, the analytical solution showed an oscillating pattern at the near field, decaying asymptotically at larger distances (Figure 7a). The exact solution and the analytical far field approximation converged at ∼4.5 m from the source (Figure 7a), the deviation between both curves also describing a decreasing pattern with distance (Figure 7b). The comparison of the far vs. near-to-far field deviation between the model and the empirical measurements of the sphere (Figure 7b) showed good agreement for ranges ≥3 m, but the deviation between model prediction and experimental values was important at 2 m. These results pointed to a larger critical range for the 38 kHz transducer than predicted by the model (1.14 m). According to these results, ex situ TS measurements of anchovy at depths <2.5 m were discarded at 38 kHz for being less reliable: worse agreement between model and measurements and expected faster TS changes with range (Figure 7a). For the rest, we applied bias correction factors according to the interpolated results of the near field experiment. Figure 7. Open in new tabDownload slide (a) Analytical exact on-axis solution for a circular piston simulating the Simrad ES38B transducer pressure (solid line) and far-field approximation (segmented line) relative to the maximum on-axis radiated pressure (pmax) against distance from the source. (b) TS deviation between on-axis pressure and far field analytical approximations (solid curve) vs. distance. The circles represent the mean experimental TS difference of the sphere between measurements done using near-to-far and far field calibration parameters. Deviation between model prediction and experimental values was important only at 2 m (empty circle). Figure 7. Open in new tabDownload slide (a) Analytical exact on-axis solution for a circular piston simulating the Simrad ES38B transducer pressure (solid line) and far-field approximation (segmented line) relative to the maximum on-axis radiated pressure (pmax) against distance from the source. (b) TS deviation between on-axis pressure and far field analytical approximations (solid curve) vs. distance. The circles represent the mean experimental TS difference of the sphere between measurements done using near-to-far and far field calibration parameters. Deviation between model prediction and experimental values was important only at 2 m (empty circle). Measured TS values Following the plankton filtering (Figure 3), the application of the high-density filter to the in situ data retained 30, 52, and 74% of the targets at 38, 12, and 200 kHz, respectively. Concerning ex situ measurements, no significant differences were observed among the mean TS values measured at different pulse durations (Figure 8). This was intended to be a quality control for the single target detection reliability, thus the lack of differences suggested that the single target detection was correctly achieved at all resolutions. Consequently, all three experiments using the ex situ measurement sets were included in the analysis. Figure 8. Open in new tabDownload slide Boxplots summarizing TS distributions against pulse duration used in the ex situ experiments. Dashed boxplots represent the TS distributions of the measurements from set 2, contaminated with 2% of horse mackerel. Pairwise t-test produced p-values >0.05 within pulse durations considering both contaminated and non-contaminated samples. The boxplots are bounded by first and third quartiles, the central line is the median, and the whiskers show the range of values. Figure 8. Open in new tabDownload slide Boxplots summarizing TS distributions against pulse duration used in the ex situ experiments. Dashed boxplots represent the TS distributions of the measurements from set 2, contaminated with 2% of horse mackerel. Pairwise t-test produced p-values >0.05 within pulse durations considering both contaminated and non-contaminated samples. The boxplots are bounded by first and third quartiles, the central line is the median, and the whiskers show the range of values. Combining both in situ and ex situ measurements, a total of 6388, 15,695, and 19,012 targets were finally used for the TS estimates at 38, 120, and 200 kHz, respectively. The filtered data set covered a depth range of 2.5–27.5 m. The fish length values measured from the different experiments ranged from 3.5 to 19.5 cm, with the highest mean (±standard deviation) value obtained in spring 13.4 (±1.5) cm, and smaller mean values in the cage experiments and autumn survey of 10.4 (±1) and 10 (±3.4) cm, respectively. The highest size variability was observed in the autumn JUVENA survey, with a wider distribution than in the spring BIOMAN survey and cage measurements (Figure 9a) caused by presence in this season of a mixture of juveniles and adults. The filtered TS distributions (Figure 9b) were clearly monomodal, except for the 120- and 200-kHz measurements from the cage, where the mode was less pure. Mean TS values obtained from the filtered TS distributions are presented in Table 5. Figure 9. Open in new tabDownload slide Length (a) and filtered TS (b) histograms grouped by in situ (BIOMAN, JUVENA) and ex situ (cage) measurements. Figure 9. Open in new tabDownload slide Length (a) and filtered TS (b) histograms grouped by in situ (BIOMAN, JUVENA) and ex situ (cage) measurements. Table 5. Mean TS values (± SD) obtained from the filtered TS distributions. . 38 kHz . 120 kHz . 200 kHz . BIOMAN −43.3(±1.5) dB −45.4((±2.3) dB −47.1(±2.1) dB CAGE −45.4(±0.9) dB −45.5(±2.3) dB −47.7(±2.3) dB JUVENA −46.2(±2.7) dB −49.4(±2.1) dB −50.3(±2.5) dB . 38 kHz . 120 kHz . 200 kHz . BIOMAN −43.3(±1.5) dB −45.4((±2.3) dB −47.1(±2.1) dB CAGE −45.4(±0.9) dB −45.5(±2.3) dB −47.7(±2.3) dB JUVENA −46.2(±2.7) dB −49.4(±2.1) dB −50.3(±2.5) dB Open in new tab Table 5. Mean TS values (± SD) obtained from the filtered TS distributions. . 38 kHz . 120 kHz . 200 kHz . BIOMAN −43.3(±1.5) dB −45.4((±2.3) dB −47.1(±2.1) dB CAGE −45.4(±0.9) dB −45.5(±2.3) dB −47.7(±2.3) dB JUVENA −46.2(±2.7) dB −49.4(±2.1) dB −50.3(±2.5) dB . 38 kHz . 120 kHz . 200 kHz . BIOMAN −43.3(±1.5) dB −45.4((±2.3) dB −47.1(±2.1) dB CAGE −45.4(±0.9) dB −45.5(±2.3) dB −47.7(±2.3) dB JUVENA −46.2(±2.7) dB −49.4(±2.1) dB −50.3(±2.5) dB Open in new tab TS–length relationships Overall, the measured TS values increased linearly with the logarithm of the fish length. In agreement with the corresponding mean body lengths, the TS values were generally higher for the spring hauls than the autumn hauls and the experimental cage results (Figure 9). In sum, the three types of measurements (ex situ and in situ from both surveys) closely fitted the same TS vs. log-length regression. When the slope of the regression was forced to 20, the b20 values for the in situ-measured data were −66.5, −68.9, and −70.5 dB at 38, 120, and 200 kHz, respectively, and −65.8, −66.4, and −68.7 dB for the ex situ measurements (Table 6). When considering the whole data set, these values were −66.4, −68.7, and −70.4 dB. The free-fitting linear model produced significant in situ regressions, with slopes of 22.5, 20.5, and 22.5 at the three frequencies, respectively. The non-significant TS–L relationship derived from the ex situ data was not considered relevant, instead being attributed to the small number of points (only three) available. When considering the entire data set, significant slopes slightly over 20 at the three operative frequencies, with intercepts at −68.7, −69, and −72.9 dB, respectively. Table 6. Statistics of the empirical and theoretical (by means of the MFS backscattering model) TS–L linear regression parameters. . Frequency (kHz) . a . B . b20 . N . In situ 38 22.2 −68.8 −66.5 53 120 20.5 −69.5 −68.9 53 200 22.5 −73.2 −70.5 53 Ex situ 38 – – −65.8 3 120 – – −66.4 3 200 – – −68.7 3 In situ + ex situ 38 22.2 −68.7 (r2 = 0.7) −66.4 (r2 = 0.8) 56 120 20.2 −69 (r2 = 0.7) −68.7 (r2 = 0.9) 56 200 22.4 −72.9 (r2 = 0.5) −70.4 (r2 = 0.9) 56 Theoretical model 38 15.2 −61.8 (r2 = 0.9) −67.1 (r2 = 0.8) – 120 21.6 −71 (r2 = 0.9) −69.7 (r2 = 0.9) – 200 20.2 −70.1 (r2 = 1) −70.4 (r2 = 1) – . Frequency (kHz) . a . B . b20 . N . In situ 38 22.2 −68.8 −66.5 53 120 20.5 −69.5 −68.9 53 200 22.5 −73.2 −70.5 53 Ex situ 38 – – −65.8 3 120 – – −66.4 3 200 – – −68.7 3 In situ + ex situ 38 22.2 −68.7 (r2 = 0.7) −66.4 (r2 = 0.8) 56 120 20.2 −69 (r2 = 0.7) −68.7 (r2 = 0.9) 56 200 22.4 −72.9 (r2 = 0.5) −70.4 (r2 = 0.9) 56 Theoretical model 38 15.2 −61.8 (r2 = 0.9) −67.1 (r2 = 0.8) – 120 21.6 −71 (r2 = 0.9) −69.7 (r2 = 0.9) – 200 20.2 −70.1 (r2 = 1) −70.4 (r2 = 1) – The number of points available from the ex situ measurements was too small, and the length range too narrow to be considered for estimating the a and b values for ex situ data. Results of the whole dataset (including both in situ and ex situ data) in bold. Open in new tab Table 6. Statistics of the empirical and theoretical (by means of the MFS backscattering model) TS–L linear regression parameters. . Frequency (kHz) . a . B . b20 . N . In situ 38 22.2 −68.8 −66.5 53 120 20.5 −69.5 −68.9 53 200 22.5 −73.2 −70.5 53 Ex situ 38 – – −65.8 3 120 – – −66.4 3 200 – – −68.7 3 In situ + ex situ 38 22.2 −68.7 (r2 = 0.7) −66.4 (r2 = 0.8) 56 120 20.2 −69 (r2 = 0.7) −68.7 (r2 = 0.9) 56 200 22.4 −72.9 (r2 = 0.5) −70.4 (r2 = 0.9) 56 Theoretical model 38 15.2 −61.8 (r2 = 0.9) −67.1 (r2 = 0.8) – 120 21.6 −71 (r2 = 0.9) −69.7 (r2 = 0.9) – 200 20.2 −70.1 (r2 = 1) −70.4 (r2 = 1) – . Frequency (kHz) . a . B . b20 . N . In situ 38 22.2 −68.8 −66.5 53 120 20.5 −69.5 −68.9 53 200 22.5 −73.2 −70.5 53 Ex situ 38 – – −65.8 3 120 – – −66.4 3 200 – – −68.7 3 In situ + ex situ 38 22.2 −68.7 (r2 = 0.7) −66.4 (r2 = 0.8) 56 120 20.2 −69 (r2 = 0.7) −68.7 (r2 = 0.9) 56 200 22.4 −72.9 (r2 = 0.5) −70.4 (r2 = 0.9) 56 Theoretical model 38 15.2 −61.8 (r2 = 0.9) −67.1 (r2 = 0.8) – 120 21.6 −71 (r2 = 0.9) −69.7 (r2 = 0.9) – 200 20.2 −70.1 (r2 = 1) −70.4 (r2 = 1) – The number of points available from the ex situ measurements was too small, and the length range too narrow to be considered for estimating the a and b values for ex situ data. Results of the whole dataset (including both in situ and ex situ data) in bold. Open in new tab Preliminary linear regression results indicated a TS increase with depth at all frequencies (Figure 10), contrarily to expectations of swimbladder compression with pressure increase for a physostomous fish such as anchovy. When building linear models of length against depth, we found a significant (p < 0.05 for adults and p < 0.005 for juveniles) increase of length also with depth, which, as the TS generally increases with fish body length, might explain the unexpected TS–depth pattern observed during the night trawls. However, concurrent multiple lineal analyses of TS against depth and length were not reliable due to the collinearity found between body length and depth of anchovy. To test whether the unexpected TS–depth relationship observed was caused by the increase of length of anchovy, further analyses were conducted: predicted TS values were combined with the observed TS–length relationship and the expected TS–depth relationship from theoretical swimbladder compression following Boylés law (explained in more detail in the corresponding section of the Supplementary material). Figure 10. Open in new tabDownload slide Top panel: relation between anchovy body length and mean depth per haul. Bottom panel: relation between mean TS per haul and mean depth. Both graphs distinguish between age groups and seasons. Figure 10. Open in new tabDownload slide Top panel: relation between anchovy body length and mean depth per haul. Bottom panel: relation between mean TS per haul and mean depth. Both graphs distinguish between age groups and seasons. Theoretical backscattering of anchovy The two-chambered swimbladder and backbone simulations (Figure 11) predicted more directive patterns of TS vs. swimming tilt angle for increasing frequencies, leading to a steeper reduction in TS values with angles for the first ±15°. Thus, although the maximum TS values were similar at all frequencies, the mean values were considerably higher for lower frequencies, due to the narrower directivity pattern decreasing the averages when frequency is increased. This agreed with the empirically-measured TS frequency response of anchovy at the three experimental frequencies. Figure 11. Open in new tabDownload slide Beam directivity patterns obtained with the backscattering simulation of the two-chambered swimbladder plus backbone at the three frequencies of study. The maximum TS values are obtained for a swimming tilt angle of 5° at higher frequencies (120 and 200 kHz), corresponding to more directive TS patterns Although these maximum values are similar at all frequencies, the mean TS values, for a normal distribution of swimming tilt angles around the horizontal (α = 0°), are lower for higher frequencies. Figure 11. Open in new tabDownload slide Beam directivity patterns obtained with the backscattering simulation of the two-chambered swimbladder plus backbone at the three frequencies of study. The maximum TS values are obtained for a swimming tilt angle of 5° at higher frequencies (120 and 200 kHz), corresponding to more directive TS patterns Although these maximum values are similar at all frequencies, the mean TS values, for a normal distribution of swimming tilt angles around the horizontal (α = 0°), are lower for higher frequencies. The model optimization produced the best values at mean tilt angles of 3 ± 9° (mean value ± standard deviation; r2 = 0.9) for the 38- and 120-kHz frequencies (i.e. swimming slightly oriented downwards). However, at 200 kHz, the best fit was obtained at −5 ± 9° (r2 = 0.99) (slightly oriented upwards (Figure 5)). When considering the three frequencies together, the mean tilt angle was practically 0° (i.e. swimming horizontally). Hence, given the discrepancy observed between the individual fits at the different frequencies (mainly due to the highest directivity of the 200-kHz frequency), a mean tilt angle of 0° was chosen. After fixing the mean tilt value at 0°, the optimization process produced the best standard deviation at 9° for 38- and 120-kHz frequencies (r2 = 0.9); 11° (r2 = 0.99) for 200 kHz; and 10° (r2 = 0.9) when considering the three frequencies together. According to these results, the a and b values used in the modelled TS–L relationships (Figure 12) were obtained considering a compromise “optimum” with a mean tilt angle of 0 ± 10° (Table 6). Figure 12. Open in new tabDownload slide Mean TS against total log-length (L) relationship. Dashed line = experimental forced fitting (b20); dotted line = experimental free fitting, bold solid line = theoretical model (b20); and grey solid line = theoretical model free fitting. Red circles correspond to BIOMAN hauls, blue squares to JUVENA hauls, and green triangles to the cage experiments. Figure 12. Open in new tabDownload slide Mean TS against total log-length (L) relationship. Dashed line = experimental forced fitting (b20); dotted line = experimental free fitting, bold solid line = theoretical model (b20); and grey solid line = theoretical model free fitting. Red circles correspond to BIOMAN hauls, blue squares to JUVENA hauls, and green triangles to the cage experiments. The modelled TS–L linear regression agreed better with the experimental results at the highest frequency (Table 6). At 38 and 120 kHz, the modelled b20 value was 0.7–1 dB lower than the experimental value. At 200 kHz instead, there was practically no difference between the modelled and experimental values. Discussion This work presents a comprehensive study of European anchovy TS in the Bay of Biscay. A combination of in situ, ex situ and theoretical modelling results are presented of the species-specific TS–L relationships to be used for biomass estimation and stock assessment of this species. Data were collected during spring and autumn for 7 years, covering a wide range of fish lengths (3.5–19.5 cm) and physiological conditions. The size and TS distributions (Figure 9) at the different sampling periods reflect this seasonality, as the largest specimens were measured during the spring spawning peak (when only adults are present in the area) and the smallest during the autumn survey (when part of the adult stock is absent due to trophic migrations, whereas juveniles predominate) (Boyra et al., 2013). The range of sizes of anchovy used in this experiment practically cover the full range of sizes observed during the acoustic surveys of anchovy in the Bay of Biscay (4–20 cm). The depth range of the anchovies studied (2.5–27.5 m) is in the upper range of the typical 5–120 m of the acoustically sampled depth for assessment (Boyra et al., 2013). When measuring TS values for small pelagic species such as anchovy, two of the key difficulties are to apply a correct lower threshold (Weimer and Ehrenberg, 1975) and to avoid bias by unresolved multiple targets (Soule et al., 1995). The application of an incorrect threshold can affect the lower end of the TS distribution by including small targets such as plankton. In the in situ measurements, this was avoided by applying a bi-frequency mask to distinguish fish from plankton, as has been successfully demonstrated in other studies (Lezama-Ochoa et al., 2011, 2014; Albaina et al., 2015; Gastauer et al., 2017), whereas in the ex situ measurements, this distinction was based on the observed difference between echo-traces according to expected differences in swimming velocity between fish and plankton. Potential bias due to unresolved multiple echoes was mitigated following different procedures. First, all measurements were made at night to facilitate the detection of single fish targets. Second, the high-density filter (Gauthier and Rose, 2001) applied to the in situ measurements further decreased the probability of erroneously assigning to a single fish the echo of multiple targets. As in a previous study (Boyra et al., 2018), the empirically determined Nv value was shown to be independent of the horizontal scale at which it was calculated as well as the initial TS value used, thus avoiding the well-known circularity issue of calculating a TS value based on a previous one. Finally, the ex situ measurement results were similar at different effective resolutions (pulse durations) of the acoustic sensors, supporting the notion that the relatively low fish density at the cage was sufficient to avoid multiple echoes. An additional difficulty encountered was that a large proportion of the ex situ measurements done at 38 kHz were made at distances inside the transition from the near to the far field of the transducer, which, due to the inherent divergence from spherical decay of acoustic beams in this area, were likely to be biased. The ad hoc experiment conducted to address this issue was in agreement with predictions by the analytical model of the circular piston (Figure 7b), thereby showing that it is possible to perform unbiased measurements at distances inside the near-to-far field, provided that calibrations are done at those distances. Moreover, the analytical model also showed that we can safely interpolate gain values at intermediate distances, provided that they are farther than the critical range. The observed deviations between the theoretical and empirical location range of critical range (empirical RC > 1.14 m, Figure 7b) can be attributable to an inaccurate knowledge of the specific geometry of the actual transducers. A better knowledge of the construction characteristics of transducers should improve the application of the proposed method. In future studies, this procedure is recommended when making ex situ acoustic experiments at short ranges, as it may be helpful to increase the range of valid measurements. This study has shown significant positive correlations between TS and fish length and a good consistency both between in situ results from different seasons and between in situ and ex situ results. The optimal slope obtained for the whole data set at all frequencies was consistently close to 20, meaning that the horizontal cross-sectional area of the swimbladder changes proportionally to the square of the fish length, as according to expectations (Simmonds and Maclennan, 2005). Assuming the correct performance of the model and based on the agreement with the experimental results, anchovy swimming orientation tendency should be, on average, not too far from horizontal, but with a rather high variability (standard deviation around 10°). These results should however be taken with care, since the optimization outputs for the different frequencies pointed towards opposite swimming directions, hence making our results not particularly revealing in this matter. Previous studies on similar engraulid species (Aoki and Inagaki, 1988; Madirolas et al., 2016) and other clupeoids (Huse and Ona, 1996) at the night scattering layer described a slightly head up swimming orientation. These results were justified by the need of the fish to compensate a negative buoyancy (Huse and Ona, 1996; Madirolas et al., 2016). Given that anchovy is a physostomous species and thus unable to compensate its swimbladder volume against pressure changes, it can be expected that TS decreases with depth according to Boyle’s law, as has been observed in previous measurements of anchovy (Zhao et al., 2008) and other physostomous species (Ona, 2003). In spite of this, we were unable to find evidence of the swimbladder compression on the observed TS–length relationship. On the contrary, the unexpected increase of TS with depth (Figure 10) was caused by a general depth stratification pattern of anchovies according to body size during night hauls. Our analysis showed that the TS increase produced by this length stratification was able (although not completely, Supplementary Figure S1) to justify the observed TS–depth increase, prevailing over and masking the expected decrease of TS with depth due to swimbladder compression. The failure to find evidence of swimbladder compression in our TS–depth relationships might have owed to the small depth span of our measurements, due to the typical shallower distribution of anchovy at night. An additional factor could be the lack of depth resolution of the sampling collection in the in situ experiments. The vertical opening of the haul was 15–20 m for near-surface trawls, hence the same magnitude of the whole depth range of the study. Consequently, we might be losing part of the length stratification inside each haul, which would explain the smaller predicted slopes. Other factors that might have yield to such inverted TS–depth slope could be for example a higher probability of failure of the single target detection filters with increasing depth or a change of behaviour of anchovy with depth. Consequently, further research is necessary to supplement the measurements obtained in this work at different depth ranges. This is a difficult objective to achieve using echosounders installed on a vessel, because it implies measuring the TS during the day, when anchovies operate near the sea bottom (according to their nycthemeral migrations) and aggregate in schools, precluding the identification of single targets. One possible solution would be to use submersible echosounders inside the trawls at different trawl depths to make these measurements. With respect to the observed increase of anchovy length with depth in night hauls, we consider this an interesting and completely unexpected result in itself. What appears to be happening is that, after the nocturnal migration there occurs a spontaneous stratification of the anchovies by size, perhaps due to change of swimming velocity with body length. Its remarkable to have been able to detect this stratification, given the poor resolution of the pelagic trawl with respect to the full extension of the sampled layer. Therefore, we expect that if this phenomenon is further studied in the future with higher sampling resolution, the observed stratification will be stronger. Our values lie within the range of the latest published TS values obtained for engraulid species. At the most commonly used frequency in fisheries acoustics (38 kHz) (Simmonds and Maclennan, 2005), recent studies have provided b20 = −68.6 dB on similar species Engraulis anchoita (Madirolas et al., 2016), obtained from 11 to 17 cm specimens during night-time in situ TS measurements. Other experiments on Engraulis japonicus (Zhao et al., 2008) have yielded TS–length relationships that have predicted TS values of −65.8 dB (Kang et al., 2009) for lengths ranging 4.8–12.2 cm and −66.5 dB (Sawada et al., 2009) for 10.6 cm at 70 kHz. The only previous work to have examined the TS of European anchovy funnelled the targets through a net with an open cod end, obtaining a b20 of −65.2 dB from 12.5-cm anchovies at ∼60-m depth at 70 kHz (Doray et al., 2016). This methodology avoided multiple target bias, but at the expense of forcing the anchovies to swim almost horizontally (i.e. with a narrower distribution of tilt angles than expected to be their natural behaviour) towards the net mouth. Although the results of that work are not directly comparable with our own due to differences in frequency and behaviour, they can be considered qualitatively consistent. However, these authors predicted a 2-dB decrease at 38 kHz, when increasing the fish tilt angle towards a more “natural” swimming behaviour. Hereby, a b20 value of −67.2 dB was reported. This value is consistent with a TS reduction with the increased range of tilt angles illustrated in Figure 11. The obtained TS trend with frequency, with higher responses at lower frequencies (Figure 9b), was typical of bladder-bearing fish species (Fernandes et al., 2006). This pattern may prove useful in developing multi-frequency masks to discriminate anchovy from plankton and other pelagic species (Lezama-Ochoa et al., 2011). The backscattering model provided some rough explanation of the TS frequency response. According to the model, despite the similarity between the highest TS values across frequencies, the greater directivity of the higher frequencies (Figure 11) produced lower mean TS values when averaged over a range of tilt angles. In general terms, a rather good general agreement was obtained between the simulations and the empirical results (Table 6). Implications for assessment Despite the need for a precise TS value in the acoustic assessment of fish abundance, alongside the recommendation that an empirical TS–L data relationship be established whenever new data are collected (McClatchie, 2003), biomass estimates of European anchovy in the Bay of Biscay have long been obtained with herring TS values published more than three decades ago (ICES, 1982; Degnbol et al., 1985). In response, this study has presented the first TS measurements for European anchovy at the frequency of 38 kHz used for assessment. The obtained b20 values at 38 kHz were 5–6 dB higher than those currently used by acoustic surveys in the assessment of European anchovy in the Bay of Biscay (Boyra et al., 2013). Such values would represent a more than twofold decrease if applied to estimate the acoustic-based biomass of anchovy. Nevertheless, the TS values were derived at a lower depth (z ≈ 13 m) than is typical for anchovy during the daytime (i.e. the period at which acoustic surveys are conducted), especially for adults and larger juveniles that are subjected to nycthemeral migrations. Thus, given the expected decrease of TS with depth for anchovy and other physostomous species (Ona, 2003; Zhao et al., 2008; Fässler et al., 2009; Madirolas et al., 2016), it is likely that the reduction in acoustic-based biomass will be somewhat lower than that inferred solely from this work. Therefore, it is recommended that further research is conducted to determine the TS–depth relationship for anchovy by lowering echosounders at different depths during day and night hauls. The use of this larger depth range should help us avoiding the artefact TS–depth relation observed in this work (Figure 10). The findings could then be combined with the present results to produce a thorough TS–length–depth relationship to update the acoustic-based assessment of this important species. Conclusion This study has provided the first TS–length relationship for European anchovy at the frequency used by assessment acoustic surveys (38 kHz) as well as the frequency response at typical frequencies (38, 120, and 200 kHz), which may be useful for building species discrimination masks. The measurements were done targeting anchovies both in their natural environment and in a harbour cage. Special care was taken to reduce the potential bias associated with measuring anchovies in situ and to guarantee that ex situ measurements are free from bias caused by range. The fact that the linear TS–L regressions using the entire ex situ and in situ data set produced such good fit adds robustness to the estimated values. The values obtained (−66.5, −68.7, and −70.4 dB at 38, 120, and 200 kHz, respectively) are in accordance with recently published values for the TS of anchovy in other areas as well as with backscattering models for bladdered fish. This points towards a general overestimation of current acoustic surveys for assessment, although it is necessary to extend the depth range of measurements beyond the ∼13 m of this study before it will be possible to update the assessment of this species in the Bay of Biscay. Supplementary data Supplementary material is available at the ICESJMS online version of the manuscript. Sobradillo, B., Boyra, G., Pérez-Arjona, I., Martinez, U., and Espinosa, V. 2021. Ex situ and in situ target strength measurements of European anchovy in the Bay of Biscay. – ICES Journal of Marine Science, 00:000–000. Acknowledgements This study was funded by AZTI-Tecnalia and supported by the research projects JAULA for the ex situ measurements and JUVENA for the in situ ones. The project JAULA was funded by the Department of Agriculture, Fisheries and Food of the Basque Country Government; we thank Unai Cotano, the leader of the project, for giving us access to the cage and for providing biological sampling information. We would like to thank the crew from the Itsas Lagunak for providing us with the anchovy specimens, the Aquaculture School from Mutriku for keeping the specimens during the acclimatization period and the technician and diver Gaizka Bidegain from AZTI for the maintenance and underwater inspection of the specimens used throughout the ex situ measurements. Yolanda Lacalle is thanked for the illustration in Figure 2. The project JUVENA was co-funded by the Dirección de Innovación y Desarrollo Tecnológico, Viceconsejería de Política e Industria Alimentaria, Dpto. Agricultura, Pesca y Alimentación of the Basque Government and the Secretaría General del Mar, Ministerio de Agricultura, Alimentación y Medio Ambiente of the Spanish Government. Thanks also to Andrés Uriarte (Azti) for improving this work with his valuable comments and to Mathieu Doray and Laurent Berger (Ifremer) for their helpful suggestions regarding the analysis and the interpretation of the results. Finally, thanks to the anonymous contribution of the reviewers for helping improve the quality and readability of this work. Author contributions BS and GB wrote the main manuscript text and were involved in the acquisition, analysis, and interpretation of data. IP-A and VE contributed to the theoretical modelling application and interpretation. UM collected data for the in situ analysis. All authors have substantially revised the manuscript. Data availability Datasets generated and/or analysed during this study are available from the corresponding author upon reasonable request. References Albaina A. , Irigoien X., Aldalur U., Boyra G., Santos M., Estonba A. 2015 . Macrozooplankton predation impact on anchovy (Engraulis encrasicolus) eggs mortality at the Bay of Biscay shelf break spawning centre . ICES Journal of Marine Science , 72 : 1370 – 1379 . Google Scholar Crossref Search ADS WorldCat Andreeva I. B. 1974 . Scattering of sound by air bladders of fish in deep sound-scattering ocean layers. Soviet Physics Acoustics, 10 : 17 – 20 . Aoki I. , Inagaki T. 1988 . Photographic observations on the behaviour of Japanese anchovy Engraulis japonica at night in the sea . Marine Ecology Progress Series , 43 : 213 – 221 . Google Scholar Crossref Search ADS WorldCat 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 : 360 – 381 . Google Scholar Crossref Search ADS WorldCat Barange M. , Hampton I., Soule M. 1996 . Empirical determination of in situ target strengths of three loosely aggregated pelagic fish species . ICES Journal of Marine Science , 53 : 225 – 232 . Google Scholar Crossref Search ADS WorldCat Boyra G. , Martínez U., Cotano U., Santos M., Irigoien X., Uriarte A. 2013 . Acoustic surveys for juvenile anchovy in the Bay of Biscay: abundance estimate as an indicator of the next year’s recruitment and spatial distribution patterns . ICES Journal of Marine Science , 70 . Google Scholar OpenURL Placeholder Text WorldCat Boyra G. , Moreno G., Orue B., Sobradillo B., Sancristobal I. 2019 . In situ target strength of bigeye tuna (Thunnus obesus) associated with fish aggregating devices . ICES Journal of Marine Science , 76 : 2446 – 2458 . Google Scholar OpenURL Placeholder Text WorldCat Boyra G. , Moreno G., Sobradillo B., Pérez-Arjona I., Sancristobal I., Demer D. A. 2018 . and Handling editor: purnima Ratilal Target strength of skipjack tuna (Katsuwanus pelamis) associated with fish aggregating devices (FADs) . ICES Journal of Marine Science , 75 : 1790 – 1802 . Google Scholar Crossref Search ADS WorldCat Chu D. , Eastland G. C. 2015 . Calibration of a broadband acoustic transducer with a standard spherical target in the near field . The Journal of the Acoustical Society of America , 137 : 2148 – 2157 . Google Scholar Crossref Search ADS PubMed WorldCat Degnbol P. , Lassen H., Staehr K. J. 1985 . In-situ determination of target strength of herring and sprat at 38 and 120 kHz . Dana , 5 : 45 – 54 . Google Scholar OpenURL Placeholder Text WorldCat Demer D. a. , Berger L., Bernasconi M., Bethke E., Boswell K. M., Chu D., Domokos R., et al. 2015 . Calibration of acoustic instruments . ICES Cooperative Research Report , 326 : 133 . Google Scholar OpenURL Placeholder Text WorldCat Doray M. , Berger L., Le Bouffant N., Coail J. Y., Vacherot J. P., de La Bernardie X., Morinière P., et al. 2016 . A method for controlled target strength measurements of pelagic fish, with application to European anchovy (Engraulis encrasicolus) . ICES Journal of Marine Science , 73 : 1987 – 1997 . Google Scholar Crossref Search ADS WorldCat Echoview Software . 2013 . Echoview Software Pty Lt , Hobart, Tasmania, Australia . Fairweather G. , Karageorghis A., Martin P. A. 2003 . The method of fundamental solutions for scattering and radiation problems . Engineering Analysis with Boundary Elements , 27 : 759 – 769 . Google Scholar Crossref Search ADS WorldCat Fässler S. M. M. , Fernandes P. G., Semple S. I. K., Brierley A. S. 2009 . Depth-dependent swimbladder compression in herring Clupea harengus observed using magnetic resonance imaging . Journal of Fish Biology , 74 : 296 – 303 . Google Scholar Crossref Search ADS PubMed WorldCat Fernandes A. , Masse J., Iglesias M., Diner N., Ona E. 2006 . The SIMFAMI Project: Species Identification Methods from Acoustic Multi-Frequency Information . Aberdeen , UK . Final Report to the EC Number Q5RS-2001-02054, 486 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Fernandes P. G. , Copland P., Garcia R., Nicosevici T., Scoulding B. 2016 . Additional evidence for fisheries acoustics: small cameras and angling gear provide tilt angle distributions and other relevant data for mackerel surveys . ICES Journal of Marine Science , 73 : 2009 – 2019 . Google Scholar Crossref Search ADS WorldCat Foote K. G. 2014 . Discriminating between the nearfield and the farfield of acoustic transducers . The Journal of the Acoustical Society of America , 136 : 1511 – 1517 . Google Scholar Crossref Search ADS PubMed WorldCat Foote K. G. 1980 . Importance of the swimbladder in acoustic scattering by fish: a comparison of Gadoid and mackerel target strengths . The Journal of the Acoustical Society of America, 67 : 2084 – 2089 . Foote K. G. , Francis D. T. 2002 . Comparing Kirchhoff-approximation and boundary-element models for computing gadoid target strengths . The Journal of the Acoustical Society of America , 111 : 1644 – 1654 . Google Scholar Crossref Search ADS PubMed WorldCat Foote K. G. , Knudsen H. P., Vestnes G., MacLennan D. N., Simmonds E. J. 1987 . Calibraton of acoustic instruments for fish density estimation: a practical guide. ICES Cooperative Research Report 144. Copenhagen. Fujino T. , Sadayasu K., Abe K., Kidokoro H., Tian Y., Yasuma H., Miyashita K. 2009 . Swimbladder morphology and target strength of a mesopelagic fish, Maurolicus japonicus . Journal of Marine Acoustics Society of Japan , 36 : 241 – 249 . Google Scholar Crossref Search ADS WorldCat Furusawa M. 1988 . Prolate spheroidal models for predicting general trends of fish target strength . Journal of the Acoustic Society of Japan , 9 : 13–24. Google Scholar OpenURL Placeholder Text WorldCat Gastauer S. , Scoulding B., Parsons M. 2017 . Estimates of variability of goldband snapper target strength and biomass in three fishing regions within the Northern Demersal Scalefish Fishery (Western Australia) . Fisheries Research , 193 : 250 – 262 . Google Scholar Crossref Search ADS WorldCat Gauthier S. , Rose G. A. 2001 . Diagnostic tools for unbiased in situ target strength estimation . Canadian Journal of Fisheries and Aquatic Sciences , 58 : 2149 – 2155 . Google Scholar Crossref Search ADS WorldCat Glass C. 2000 . Dynamics of Pelagic Fish Distribution and Behaviour: Effects on Fisheries and Stock Assessment. Pierre Fréeon and Ole Arve Misund. Reviews in Fish Biology and Fisheries , 10 : 124 – 124 . Crossref Search ADS Godinho L. M. C. , Costa E., Pereira A., Santiago J. 2012 . Some observations on the behaviour of the method of fundamental solutions in 3D acoustic problems . International Journal of Computational Methods , 09 : 04: 1250049. Google Scholar OpenURL Placeholder Text WorldCat Gorska N. , Ona E., Korneliussen R. 2005 . Acoustic backscattering by Atlantic mackerel as being representative of fish that lack a swimbladder. Backscattering by individual fish . ICES Journal of Marine Science , 62 : 984 – 995 . Google Scholar Crossref Search ADS WorldCat Henderson M. J. , Horne J. K. 2007 . Comparison of in situ, ex situ, and backscatter model estimates of Pacific hake (Merluccius productus) target strength . Canadian Journal of Fisheries and Aquatic Sciences , 64 : 1781 – 1794 . Google Scholar Crossref Search ADS WorldCat Huse I. , Ona E. 1996 . Tilt angle distribution and swimming speed of overwintering Norwegian spring spawning herring . ICES Journal of Marine Science , 53 : 863 – 873 . Google Scholar Crossref Search ADS WorldCat Ibaibarriaga L. , Fernández C., Uriarte A., Roel B. A. 2008 . A two-stage biomass dynamic model for Bay of Biscay anchovy: a Bayesian approach . ICES Journal of Marine Science , 65 : 191 – 205 . Google Scholar Crossref Search ADS WorldCat ICES . 1982 . Report of the 1982 Planning Group on ICES-Coordinated Herring and Sprat Acoustic Surveys. ICES Document CM 1982/H: 04. ICES . 2013 . ICES WGACEGG REPORT 2013 Working Group on Acoustic and Egg Surveys for Sardine and Anchovy in ICES Areas VIII and IX (WGACEGG) By Correspondence and 25–29 November 2013. ICES CM, SSGESST:20. 131 pp. ICES . 2015 . Report of the Working Group on Southern Horse Mackerel, Anchovy and Sardine (WGHANSA). ICES CM, ACOM: 16. 580 pp. ICES . 2017 . Report of the Working Group on Southern Horse Mackerel, Anchovy and Sardine (WGHANSA), 24–29 June 2017, Bilbao, Spain. Jech J. M. , Horne J. K. 2001 . Effects of in situ target spatial distributions on acoustic density estimates . ICES Journal of Marine Science , 58 : 123 – 136 . Google Scholar Crossref Search ADS WorldCat Jech J. M. , Horne J. K., Chu D., Demer D. A., Francis D. T. I., Gorska N., Jones B., et al. 2015 . Comparisons among ten models of acoustic backscattering used in aquatic ecosystem research . The Journal of the Acoustical Society of America , 138 : 3742 – 3764 . Google Scholar Crossref Search ADS PubMed WorldCat Kang D. , Cho S., Lee C., Myoung J.-G., Na J. 2009 . Ex situ target-strength measurements of Japanese anchovy (Engraulis japonicus) in the coastal Northwest Pacific . ICES Journal of Marine Science , 66 : 1219 – 1224 . Google Scholar Crossref Search ADS WorldCat Kang D. , Hwang D. 2003 . Ex situ target strength of rockfish (Sebastes schlegeli) and red sea bream (Pagrus major) in the Northwest Pacific . ICES Journal of Marine Science , 60 : 538 – 543 . Google Scholar Crossref Search ADS WorldCat Kieser R. , Mulligan T., Ehrenberg J. 2000 . Observation and explanation of systematic split-beam angle measurement errors. Aquat. Living Resour. 7. Kinsler L. E. , Frey A. R., Coppens A. B., Sanders J. V. 1999 . Fundamentals of acoustics. In Fundamentals of Acoustics , 4th edn, p. 560 . Ed. by Kinsler L. E., Frey A. R., Coppens A. B., Sanders J V.. ISBN: 0-471-84789-5. Wiley-VCH, USA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lezama-Ochoa A. , Ballón M., Woillez M., Grados D., Irigoien X., Bertrand A. 2011 . Spatial patterns and scale-dependent relationships between macrozooplankton and fish in the Bay of Biscay: an acoustic study . Marine Ecology Progress Series , 439 : 151 – 168 . Google Scholar Crossref Search ADS WorldCat Lezama-Ochoa A. , Irigoien X., Chaigneau A., Quiroz Z., Lebourges-Dhaussy A., Bertrand A. 2014 . Acoustics reveals the presence of a macrozooplankton biocline in the Bay of Biscay in response to hydrological conditions and predator-prey relationships . PLoS One , 9 : e88054 . Google Scholar Crossref Search ADS PubMed WorldCat Lilja J. , Marjomäki T. J., Jurvelius J., Rossi T., Heikkola E. 2004 . Simulation and experimental measurement of side-aspect target strength of Atlantic salmon (Salmo salar) at high frequency . Canadian Journal of Fisheries and Aquatic Sciences , 61 : 2227 – 2236 . Google Scholar Crossref Search ADS WorldCat Love R. H. 1977 . Target strength of an individual fish at any aspect . The Journal of the Acoustical Society of America , 62 : 1397 . Google Scholar Crossref Search ADS WorldCat Love R. H. 1978 . Resonant acoustic scattering by swimbladder‐bearing fish . The Journal of the Acoustical Society of America , 64 : 571 – 580 . Google Scholar Crossref Search ADS WorldCat MacLennan D. , Armstrong F., Simmonds E. 1990 . Further observations on the attenuation of sound by aggregations of fish. Proceedings of the Institute of Acoustics (UK), 12 : 99 – 106 . Maclennan D. N. , Fernandes P. G., Dalen J. 2002 . A consistent approach to definitions and symbols in fisheries acoustics . ICES Journal of Marine Science , 59 : 365 – 369 . Google Scholar Crossref Search ADS WorldCat Madirolas A. , Membiela F. a., Gonzalez J. D., Cabreira A. G., dell’Erba M., Prario I. S., Blanc S. 2016 . Acoustic target strength (TS) of argentine anchovy (Engraulis anchoita): the nighttime scattering layer . ICES Journal of Marine Science , 74 : 1408 – 1420 . Google Scholar Crossref Search ADS WorldCat Massé J. , Duhamel E., Petitgas P., Doray M., Huret M. 2018 . Pelagic Survey Series for Sardine and Anchovy in ICES Subareas 8 and 9—Towards an Ecosystem Approach. Ed. by J. Massé, A. Uriarte, M. Manuel Angélico, and P. Carrera. http://www.ices.dk/sites/pub/Publication%20Reports/Cooperative%20Research%20Report%20(CRR)/CRR%20332.pdf (last accessed 29 October 2019). McClatchie S. , Macaulay G. J., Coombs R. F. 2003 . A requiem for the use of 20log10Length for acoustic target strength with special reference to deep-sea fishes . ICES Journal of Marine Science , 60 : 419 – 428 . Google Scholar Crossref Search ADS WorldCat Medwin H. , Clay C. S. 1998 . Fundamentals of Acoustical Oceanography. Applications of Modern Acoustics . Academic Press , Boston . 712 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Murase H. , Kawabata A., Kubota H., Nakagami M., Amakasu K., Abe K., Miyashita K. 2011 . Effect of depth-dependent target strength on biomass estimation of japanese anchovy . Journal of Marine Science and Technology , 19 : 267 – 272 . Google Scholar OpenURL Placeholder Text WorldCat Ona E. , Barange M. 1999 . Single target recognition. CRR, Methodology for Target Strength Measurement, 235. http://www.ices.dk/sites/pub/Publication%20Reports/Cooperative%20Research%20Report%20(CRR)/crr235/CRR235.pdf (last accessed 28 February 2017). Ona E. 2003 . An expanded target-strength relationship for herring . ICES Journal of Marine Science , 60 : 493 – 499 . Google Scholar Crossref Search ADS WorldCat Peltonen H. , Balk H. 2005 . The acoustic target strength of herring (Clupea harengus L.) in the northern Baltic Sea . ICES Journal of Marine Science , 62 : 803 – 808 . Google Scholar Crossref Search ADS WorldCat Pérez-Arjona I. , Godinho L., Espinosa V. 2018 . Numerical simulation of target strength measurements from near to far field of fish using the method of fundamental solutions . Acta Acustica United with Acustica , 104 : 25 – 38 . Google Scholar Crossref Search ADS WorldCat Pérez-Arjona I. , Godinho L. M. C., Espinosa V. 2020a . Numerical simulation of target strength measurements from near to far field of fish using the method of fundamental solutions . ICES Journal of Marine Science . Google Scholar OpenURL Placeholder Text WorldCat Pérez-Arjona I. , Godinho L., Espinosa V. 2020b . Influence of fish backbone model geometrical features on the numerical target strength of swimbladdered fish . ICES Journal of Marine Science . Google Scholar OpenURL Placeholder Text WorldCat Santos M. , Uriarte A., Boyra G., Ibaibarriaga L. 2016 . Anchovy DEPM Surveys 2003-2012 in the Bay of Biscay (Subarea VIII) BIOMAN. ICES Cooperative Research Report, Pelagic Surveys Series for Sardine and Anchovy in ICES Areas VIII and IX (WGACEGG), 332. International Council for the Exploration of the Sea (ICES), Copenhagen. Sawada K. , Furusawa M., Williamson N. J. 1993 . Conditions for the precise measurement of fish target strength in situ . The Journal of the Marine Acoustics Society of Japan , 20 : 73 – 79 . Google Scholar Crossref Search ADS WorldCat Sawada K. , Takahashi H., Abe K., Ichii T., Watanabe K., Takao Y. 2009 . Target-strength, length, and tilt-angle measurements of Pacific saury (Cololabis saira) and Japanese anchovy (Engraulis japonicus) using an acoustic-optical system . ICES Journal of Marine Science , 66 : 1212 – 1218 . Google Scholar Crossref Search ADS WorldCat Sawada K. , Uchikawa K., Matsuura T., Sugisaki H., Amakasu K., Abe K. 2011 . In situ and ex situ target strength measurement of mesopelagic lanternfish, Diaphus Theta (Family Myctophidae) . Journal of Marine Science and Technology , 19 : 302 – 311 . Google Scholar OpenURL Placeholder Text WorldCat Simmonds J. E. , Maclennan D. N. 2005 . Fisheries Acoustics Theory and Practice . 2nd edn. Chapman & Hall , New York . 472 pp. Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Soule M. , Barange M., Hampton I. 1995 . Evidence of bias in estimates of target strength obtained with a split-beam echo-sounder . ICES Journal of Marine Science , 52 : 139 – 144 . Google Scholar Crossref Search ADS WorldCat Soule M. 1997 . Performance of a new phase algorithm for discriminating between single and overlapping echoes in a split-beam echosounder . ICES Journal of Marine Science , 54 : 934 – 938 . Google Scholar Crossref Search ADS WorldCat Torgersen T. , Kaartvedt S. 2001 . In situ swimming behaviour of individual mesopelagic fish studied by split-beam echo target tracking . ICES Journal of Marine Science , 58 : 346 – 354 . Google Scholar Crossref Search ADS WorldCat Weimer R. , Ehrenberg J. 1975 . Analysis of threshold-induced bias inherent in acoustic scattering cross-section estimates of individual fish . Journal of the Fisheries Board of Canada , 32 : 2547 – 2551 . Google Scholar Crossref Search ADS WorldCat Weston D. E. 1966 . Sound propagation in the presence of bladder fish. In Underwater Acoustics, pp. 55–58. Ed. by V. M. Albers. Plenum, New York. Ye Z. 1997 . Low-frequency acoustic scattering by gas-filled prolate spheroids in liquids . The Journal of the Acoustical Society of America , 101 : 1945 – 1952 . Google Scholar Crossref Search ADS WorldCat Zare P. , Kasatkina S. M., Shibaev S. V., Fazli H. 2017 . In situ acoustic target strength of anchovy kilka (Clupeonella engrauliformis) in the Caspian Sea (Iran) . Fisheries Research , 186 : 311 – 318 . Google Scholar Crossref Search ADS WorldCat Zhao X. , Wang Y., Dai F. 2008 . Depth-dependent target strength of anchovy (Engraulis japonicus) measured in situ . ICES Journal of Marine Science , 65 : 882 – 888 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2021. 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)

Caiger, Paul E; Lefebve, Lyndsey S; Llopiz, Joel K

doi: 10.1093/icesjms/fsaa247pmid: N/A

Abstract The mesopelagic zone covers a vast expanse of the World’s oceans and contains some of the most abundant vertebrates on the planet. This midwater region is central to the transfer of energy and carbon between the atmosphere and the deep, yet there are large knowledge gaps in our understanding of the life history of its animals. Here we synthesize the current state of knowledge of research on age, growth, and reproduction of mesopelagic fishes, the basic biological information fundamental to understanding the population dynamics of species in this ecosystem. Collectively, two-thirds of life history research on mesopelagic fishes has been undertaken on myctophids, yet many other abundant and important groups are lacking research. There are generally hotspots of mesopelagic fish research mostly centred in the northern hemisphere, with little to no coverage in the Indo-Pacific region nor the poles. Furthermore, the effects of some anthropogenic stressors—chiefly climate change and resource extraction—on the life history of the animals in this zone is uncertain and needs to be considered. Knowledge of growth and reproduction are key traits required for a holistic assessment and understanding of this ecosystem, and hopefully this synthesis will provide a springboard for greater focus in this area. Introduction The mesopelagic realm is a vast expanse of ocean covering an estimated quarter-billion cubic km globally. As a corollary, mesopelagic fishes are the most abundant vertebrates on Earth, with an estimated 5–15 billion metric tons (Kaartvedt et al., 2012; Irigoien et al., 2014; Proud et al., 2018), an order of magnitude higher than previous estimates (Gjøsæter and Kawaguchi, 1980). The variability in modern acoustic estimates is largely caused by uncertainty around deciphering the percentages of fishes to invertebrates, and the size and proportion of swimbladders in fishes (Proud et al., 2018). These midwater fishes form an important link between surface production and the deep sea, due in large part to the prevalence of diel vertical migration (DVM). Moreover, these largely zooplanktivorous and micronektivorous fishes are generally a link between lower and upper trophic levels, providing food for larger deeper-living fishes and also apex predators such as sharks, billfishes, seabirds, and marine mammals (Choy et al., 2013; Young et al., 2015). In addition to being a conduit for the active transport of carbon to the deep ocean (Davison et al., 2013), fishes in the upper 1000 m are also thought to contribute up to 26% of the oceanic carbonate production (Wilson et al., 2009). This contribution is thought to increase with rising CO2, thus becoming an increasingly important component in the inorganic carbon cycle (Wilson et al., 2009). Even though much of the role of mesopelagic fishes in global marine food webs and biogeochemical cycling remains poorly understood, it is clear their importance cannot and should not be understated. In order to fully understand mesopelagic ecosystems and the role fishes play in them, knowledge of the basic biology of populations is key. Relative to epipelagic, benthic, and coastal fish species, very little is known about the life history of mesopelagic fishes. This is largely due to the inaccessibility of this region—it is expensive to study, requiring oceanographic vessels and modern technologies capable of sampling at great depths. Furthermore, this lack of biological information is also due in part to the previous lack of commercial interest in these species (but see section on anthropogenic effects below). Additionally, there are likely considerable inter-regional variations in life histories, particularly among broadly distributed species that span equatorial to subtropical regions (e.g. Badcock and Araujo, 1988), meaning that even if there is some information for a species, it will often need to be obtained on a regional basis. Ultimately, these rudimentary biological data are crucial to understanding how quickly populations replenish, and thus how resilient they are. Information garnered from age and growth (e.g. age structure, growth rates, mortality) and reproductive analyses (e.g. maturity, spawning seasonality, fecundity) are fundamental for understanding the population dynamics of fishes in this ecosystem, and in many cases, the two are inherently linked. Our goal here is to synthesize life history research to date via an extensive literature search on age and growth and reproduction in mesopelagic fishes. Furthermore, we aim to highlight major gaps that remain in order to better direct future research. Lastly, anthropogenic impacts in the mesopelagic are increasing, from climate change to targeted harvesting, and as such, the impetus and timeliness for such research are also discussed in detail. Age and growth Age information forms the basis for calculations of growth rate, mortality rate, and productivity, ranking it among the most influential of biological variables (Campana, 2001). Therefore, in order to understand a biologically driven ecosystem such as the mesopelagic, a comprehensive understanding of age and growth is fundamental. A combination of literature databases, scholarly search engines, citations and reference lists was searched for all studies pertaining to reproduction in mesopelagic fishes. We define the term “investigation” each time a species was examined. Many species were examined more than once, and some studies/papers researched several species, thus containing multiple investigations. In all, 50 studies revealed 88 investigations on ageing, growth, and mortality in mesopelagic fishes from the literature, covering 62 species from ten families (Table 1). Table 1. Summary information for growth investigations on mesopelagic fishes. Family . Species . Location . DVM . Growth method . Max age . Growth rates . Life stage . Reference . Alepocephalidae Bajacalifornia burragei E North Pacific N O (a) 4 yr Li J, A Childress et al. (1980) Bathylagidae Pseudobathylagus milleri E North Pacific N O (a) 5 yr Asy J, A Childress et al. (1980) Leuroglossus stilbius E North Pacific Y O (a) 6 yr Semi-Asy J, A Childress et al. (1980) Lipolagus ochotensis E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Gonostomatidae Cyclothone alba Japan N LF 2 yr Asy J, A Miya and Nemoto (1986) Cyclothone atraria Japan N LF 7 yr Li J, A Miya and Nemoto (1987a) Sigmops elongatus E Gulf of Mexico Y O (d) 21 mo Li J, A Lancraft et al. (1988) Sigmops gracilis E China Sea (Japan) ? O (d) – – L Sassa and Takahashi (2018) Melamphaeidae Poromitra crassiceps E North Pacific N O (a) 8.5 Asy J, A Childress et al. (1980) Myctophidae Benthosema fibulatum Arabian Sea Y O (d) – Asy L, J Gjøsæter (1987) Benthosema glaciale W North Atlantic (Nova Scotia) Y O (a); LF 4.5 yr Asy J, A Halliday (1970) Benthosema glaciale E North Atlantic (Rockall Trough) Y LF – Asy L, J, A Kawaguchi and Mauchline (1982) Benthosema glaciale Norwegian Fjords Y O (a); LF 5 yr Asy J, A Gjøsæter (1973) Benthosema glaciale Norwegian Fjords Y O (a) 7 yr Asy Kristoffersen and Salvanes (2009) Benthosema glaciale W North Atlantic (Flemish Cap) Y O (a) 7 yr Asy J, A Garcia-Seoane et al. (2015) Benthosema pterotum Arabian Sea Y O (d) ∼1 yr Asy J, A Gjøsæter (1984) Benthosema pterotum Arabian Sea Y O (d) – Asy L, J Gjøsæter (1987) Benthosema pterotum E China Sea (Japan) Y O (d) – Asy L, J, A Ozawa and Peñaflor (1990) Benthosema pterotum E China Sea Y O (d) – Asy L Sassa et al. (2015) Benthosema suborbitale Gulf of Mexico Y O (d) 325 d Asy L, J, A Gartner (1991b) Benthosema suborbitale E Gulf of Mexico Y O (d) – Asy L Conley and Gartner (2009) Ceratoscopelus maderensis E North Atlantic Y O (d) ∼2 yr Asy J, A Linkowski et al. (1993) Ceratoscopelus townsendi E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Ceratoscopelus warmingii E North Atlantic Y O (d) 400 d Asy J, A Linkowski et al. (1993) Ceratoscopelus warmingii Indian Ocean Y O (d) – Asy J Tsarin (1994) Ceratoscopelus warmingii W North Pacific Y O (d) 416 d Asy L, J, A Takagi et al. (2006) Diaphus diademophilus Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Diaphus dumerilli Gulf of Mexico Y O (d) 362 d Asy L, J, A Gartner (1991b) Diaphus kapalae Coral Sea Y O (d) – – J Suthers (1996) Diaphus suborbitalis W North Pacific (Japan) ? O (a) 2.5 yr Asy J, A Go et al. (1977a) Diaphus suborbitalis W North Pacific (Japan) ? LF – Asy J, A Go et al. (1977b) Diaphus theta W North Pacific Y O (d) – – L, J Moku et al. (2001) Diaphus slender type E China Sea ? – – L Moku et al. (2005) Diaphus sp. E China Sea (Japan) ? O (d) – – L Sassa and Takahashi (2018) Diogenichthys laternatus E South Pacific (coastal Chile) Y O (d) – – L Landaeta et al. (2015b) Electrona antarctica Antarctica Y O (a) – – J, A Rowedder (1979) Electrona antarctica Southern Ocean Y O (a) 5 yr – J, A Shust and Kochkin (1985) Electrona antarctica Antarctica Y O (d) 3.5 yr Li J, A Greely et al. (1999) Electrona carlbergi South Atlantic Y 5 yr – J, A Konstantinova (1987) Electrona carlbergi Southern Ocean Y O (a) – – J, A Shust and Kochkin (1985) Gymnoscopelus braueri Scotia Sea Y O (a); LF 6 yr Asy A Saunders et al. (2020) Gymnoscopelus nicholsi South Atlantic Y O (a) 7 yrs Asy J, A Linkowski (1985) Gymnoscopelus nicholsi South Atlantic Y 5 yrs Asy J, A Konstantinova (1987) Hygophum benoitti North Atlantic ? IGS – – J, A Linkowski (1996) Hygophum hygomii North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum macrochir North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum reinhardtii North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum taaningi North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum taaningi E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Kreffichthys anderssoni Scotia Sea Y O (a); LF 2 yr Asy J, A Saunders et al. (2020) Lampanyctodes hectoris W Tasman Sea (Tasmania) Y O (d); LF 3 yr Asy L, J, A Young et al. (1988) Lampanyctus macdonaldi E North Atlantic (Rockall Trough) ? LF 6 yr Asy L, J, A Kawaguchi and Mauchline (1982) Lampanyctus regalis E North Pacific N O (a) 4.5 yr Li J, A Childress et al. (1980) Lampanyctus ritteri E North Pacific Y O (a) 5.5 yr Asy J, A Childress et al. (1980) Lampanyctus sp. Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Lepidophanes guentheri Gulf of Mexico Y O (d) 439 d Asy L, J, A Gartner (1991b) Myctophum affine W South Atlantic (Brazil) Y O (d) – – L Namiki et al. (2015) Myctophum asperum North Pacific Y O (d) – – J, A Hayashi et al. (2001) Myctophum asperum E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Myctophum asperum South China Sea Y O (d) 440 d Asy J, A Wang et al. (2019) Myctophum nitidulum Tropical Atlantic Y O (d) 1 yr – L, J, A Giragosov and Ovcharov (1992) Myctophum nitidulum E South Pacific (coastal Chile) Y O (d) – – L Landaeta et al. (2015b) Myctophum selenops E Gulf of Mexico ? O (d) – – L Conley and Gartner (2009) Myctophum spinosum Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Notolychnus valdiviae E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Protomyctophum arcticum E North Atlantic (Rockall Trough) M LF 1.25 yr Li L, J, A Kawaguchi and Mauchline (1982) Stenobrachius leucopsaurus E North Pacific Y O (a); LF 8 yr Asy L, J, A Smoker and Pearcy (1970) Stenobrachius leucopsaurus E North Pacific Y O (a) 7.5 yr Asy J, A Childress et al. (1980) Stenobrachius leucopsaurus W North Pacific Y O (d) – – L Methot (1981) Symbolophorus californiensis W North Pacific Y O (d) 541 d Asy L, J, A Takagi et al. (2006) Symbolophorus evermani Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Tarletonbeani crenularis E North Pacific Y O (d) 504 d Asy L, J, A Bystydzienska et al. (2010) Triphoturus mexicanus E North Pacific Y O (a) 5 yr Asy J, A Childress et al. (1980) Notoscopelidae Notoscopelus elongatus Norway and W British Isles Y O (d, a) 6 yr Asy J, A Gjøsæter (1981 b) Notoscopelus japonicus E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Notoscopelus resplendens E North Atlantic (Canary Islands) Y O (d, a) 4 yr Asy J, A Sarmiento-Lezcano et al. (2018) Phosichthyidae Vinciguerria nimbaria E China Sea (Japan) M O (d) – – L Sassa and Takahashi (2018) Vinciguerria nimbaria E tropical Atlantic M O (d) 240 d Asy L, J, A Tomas and Panfili (2000) Sternoptychidae Maurolicus muelleri Norwegian fjords and coast Y O (d); LF 3.5 yr Asy J, A Gjøsæter (1981a) Maurolicus muelleri Norwegian fjords Y LF – Asy J, A Goodson et al. (1995) Maurolicus muelleri Norwegian fjords Y LF – – J, A Salvanes and Stockley (1996) Maurolicus muelleri Norwegian fjords Y O (d) – – L Folkvord et al. (2016) Maurolicus parvipinnis Patagonian fjords Y O (d) – – L Zenteno et al. (2014) Maurolicus parvipinnis Patagonian fjords Y O (d) – – L Landaeta et al. (2015a) Valenciennellus tripunctulatus Caribbean, Gulf of Mexico M Model – – A Baird and Hopkins (1981) Stomiidae Borostomias panamensis E North Pacific N O (a) 5 yr Li J, A Childress et al. (1980) Tactostoma macropus E North Pacific Y O (a); LF 5 yr Asy L, J, A Fisher and Pearcy (1983) Trichiuridae Lepidopus caudatus NW Mediterranean N O (a) 8 yr Asy J, A Demestre et al. (1993) Family . Species . Location . DVM . Growth method . Max age . Growth rates . Life stage . Reference . Alepocephalidae Bajacalifornia burragei E North Pacific N O (a) 4 yr Li J, A Childress et al. (1980) Bathylagidae Pseudobathylagus milleri E North Pacific N O (a) 5 yr Asy J, A Childress et al. (1980) Leuroglossus stilbius E North Pacific Y O (a) 6 yr Semi-Asy J, A Childress et al. (1980) Lipolagus ochotensis E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Gonostomatidae Cyclothone alba Japan N LF 2 yr Asy J, A Miya and Nemoto (1986) Cyclothone atraria Japan N LF 7 yr Li J, A Miya and Nemoto (1987a) Sigmops elongatus E Gulf of Mexico Y O (d) 21 mo Li J, A Lancraft et al. (1988) Sigmops gracilis E China Sea (Japan) ? O (d) – – L Sassa and Takahashi (2018) Melamphaeidae Poromitra crassiceps E North Pacific N O (a) 8.5 Asy J, A Childress et al. (1980) Myctophidae Benthosema fibulatum Arabian Sea Y O (d) – Asy L, J Gjøsæter (1987) Benthosema glaciale W North Atlantic (Nova Scotia) Y O (a); LF 4.5 yr Asy J, A Halliday (1970) Benthosema glaciale E North Atlantic (Rockall Trough) Y LF – Asy L, J, A Kawaguchi and Mauchline (1982) Benthosema glaciale Norwegian Fjords Y O (a); LF 5 yr Asy J, A Gjøsæter (1973) Benthosema glaciale Norwegian Fjords Y O (a) 7 yr Asy Kristoffersen and Salvanes (2009) Benthosema glaciale W North Atlantic (Flemish Cap) Y O (a) 7 yr Asy J, A Garcia-Seoane et al. (2015) Benthosema pterotum Arabian Sea Y O (d) ∼1 yr Asy J, A Gjøsæter (1984) Benthosema pterotum Arabian Sea Y O (d) – Asy L, J Gjøsæter (1987) Benthosema pterotum E China Sea (Japan) Y O (d) – Asy L, J, A Ozawa and Peñaflor (1990) Benthosema pterotum E China Sea Y O (d) – Asy L Sassa et al. (2015) Benthosema suborbitale Gulf of Mexico Y O (d) 325 d Asy L, J, A Gartner (1991b) Benthosema suborbitale E Gulf of Mexico Y O (d) – Asy L Conley and Gartner (2009) Ceratoscopelus maderensis E North Atlantic Y O (d) ∼2 yr Asy J, A Linkowski et al. (1993) Ceratoscopelus townsendi E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Ceratoscopelus warmingii E North Atlantic Y O (d) 400 d Asy J, A Linkowski et al. (1993) Ceratoscopelus warmingii Indian Ocean Y O (d) – Asy J Tsarin (1994) Ceratoscopelus warmingii W North Pacific Y O (d) 416 d Asy L, J, A Takagi et al. (2006) Diaphus diademophilus Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Diaphus dumerilli Gulf of Mexico Y O (d) 362 d Asy L, J, A Gartner (1991b) Diaphus kapalae Coral Sea Y O (d) – – J Suthers (1996) Diaphus suborbitalis W North Pacific (Japan) ? O (a) 2.5 yr Asy J, A Go et al. (1977a) Diaphus suborbitalis W North Pacific (Japan) ? LF – Asy J, A Go et al. (1977b) Diaphus theta W North Pacific Y O (d) – – L, J Moku et al. (2001) Diaphus slender type E China Sea ? – – L Moku et al. (2005) Diaphus sp. E China Sea (Japan) ? O (d) – – L Sassa and Takahashi (2018) Diogenichthys laternatus E South Pacific (coastal Chile) Y O (d) – – L Landaeta et al. (2015b) Electrona antarctica Antarctica Y O (a) – – J, A Rowedder (1979) Electrona antarctica Southern Ocean Y O (a) 5 yr – J, A Shust and Kochkin (1985) Electrona antarctica Antarctica Y O (d) 3.5 yr Li J, A Greely et al. (1999) Electrona carlbergi South Atlantic Y 5 yr – J, A Konstantinova (1987) Electrona carlbergi Southern Ocean Y O (a) – – J, A Shust and Kochkin (1985) Gymnoscopelus braueri Scotia Sea Y O (a); LF 6 yr Asy A Saunders et al. (2020) Gymnoscopelus nicholsi South Atlantic Y O (a) 7 yrs Asy J, A Linkowski (1985) Gymnoscopelus nicholsi South Atlantic Y 5 yrs Asy J, A Konstantinova (1987) Hygophum benoitti North Atlantic ? IGS – – J, A Linkowski (1996) Hygophum hygomii North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum macrochir North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum reinhardtii North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum taaningi North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum taaningi E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Kreffichthys anderssoni Scotia Sea Y O (a); LF 2 yr Asy J, A Saunders et al. (2020) Lampanyctodes hectoris W Tasman Sea (Tasmania) Y O (d); LF 3 yr Asy L, J, A Young et al. (1988) Lampanyctus macdonaldi E North Atlantic (Rockall Trough) ? LF 6 yr Asy L, J, A Kawaguchi and Mauchline (1982) Lampanyctus regalis E North Pacific N O (a) 4.5 yr Li J, A Childress et al. (1980) Lampanyctus ritteri E North Pacific Y O (a) 5.5 yr Asy J, A Childress et al. (1980) Lampanyctus sp. Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Lepidophanes guentheri Gulf of Mexico Y O (d) 439 d Asy L, J, A Gartner (1991b) Myctophum affine W South Atlantic (Brazil) Y O (d) – – L Namiki et al. (2015) Myctophum asperum North Pacific Y O (d) – – J, A Hayashi et al. (2001) Myctophum asperum E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Myctophum asperum South China Sea Y O (d) 440 d Asy J, A Wang et al. (2019) Myctophum nitidulum Tropical Atlantic Y O (d) 1 yr – L, J, A Giragosov and Ovcharov (1992) Myctophum nitidulum E South Pacific (coastal Chile) Y O (d) – – L Landaeta et al. (2015b) Myctophum selenops E Gulf of Mexico ? O (d) – – L Conley and Gartner (2009) Myctophum spinosum Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Notolychnus valdiviae E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Protomyctophum arcticum E North Atlantic (Rockall Trough) M LF 1.25 yr Li L, J, A Kawaguchi and Mauchline (1982) Stenobrachius leucopsaurus E North Pacific Y O (a); LF 8 yr Asy L, J, A Smoker and Pearcy (1970) Stenobrachius leucopsaurus E North Pacific Y O (a) 7.5 yr Asy J, A Childress et al. (1980) Stenobrachius leucopsaurus W North Pacific Y O (d) – – L Methot (1981) Symbolophorus californiensis W North Pacific Y O (d) 541 d Asy L, J, A Takagi et al. (2006) Symbolophorus evermani Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Tarletonbeani crenularis E North Pacific Y O (d) 504 d Asy L, J, A Bystydzienska et al. (2010) Triphoturus mexicanus E North Pacific Y O (a) 5 yr Asy J, A Childress et al. (1980) Notoscopelidae Notoscopelus elongatus Norway and W British Isles Y O (d, a) 6 yr Asy J, A Gjøsæter (1981 b) Notoscopelus japonicus E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Notoscopelus resplendens E North Atlantic (Canary Islands) Y O (d, a) 4 yr Asy J, A Sarmiento-Lezcano et al. (2018) Phosichthyidae Vinciguerria nimbaria E China Sea (Japan) M O (d) – – L Sassa and Takahashi (2018) Vinciguerria nimbaria E tropical Atlantic M O (d) 240 d Asy L, J, A Tomas and Panfili (2000) Sternoptychidae Maurolicus muelleri Norwegian fjords and coast Y O (d); LF 3.5 yr Asy J, A Gjøsæter (1981a) Maurolicus muelleri Norwegian fjords Y LF – Asy J, A Goodson et al. (1995) Maurolicus muelleri Norwegian fjords Y LF – – J, A Salvanes and Stockley (1996) Maurolicus muelleri Norwegian fjords Y O (d) – – L Folkvord et al. (2016) Maurolicus parvipinnis Patagonian fjords Y O (d) – – L Zenteno et al. (2014) Maurolicus parvipinnis Patagonian fjords Y O (d) – – L Landaeta et al. (2015a) Valenciennellus tripunctulatus Caribbean, Gulf of Mexico M Model – – A Baird and Hopkins (1981) Stomiidae Borostomias panamensis E North Pacific N O (a) 5 yr Li J, A Childress et al. (1980) Tactostoma macropus E North Pacific Y O (a); LF 5 yr Asy L, J, A Fisher and Pearcy (1983) Trichiuridae Lepidopus caudatus NW Mediterranean N O (a) 8 yr Asy J, A Demestre et al. (1993) DVM information was also sourced from additional literature, and species names are those currently accepted by the California Academy of Sciences Eschmeyer’s Catalog of Fishes. DVM, diel vertical migration (Y, yes; N, no;?, unknown; M, mixed); O (d), otolith daily growth rings; O (a), otolith annual growth rings; IGS, incremental growth sequence; LF, length-frequency growth analyses; –, unknown or insufficient data; Asy, asymptotic; Li, linear; ID, insufficient data; L, larvae; J, juveniles; A, adults. Open in new tab Table 1. Summary information for growth investigations on mesopelagic fishes. Family . Species . Location . DVM . Growth method . Max age . Growth rates . Life stage . Reference . Alepocephalidae Bajacalifornia burragei E North Pacific N O (a) 4 yr Li J, A Childress et al. (1980) Bathylagidae Pseudobathylagus milleri E North Pacific N O (a) 5 yr Asy J, A Childress et al. (1980) Leuroglossus stilbius E North Pacific Y O (a) 6 yr Semi-Asy J, A Childress et al. (1980) Lipolagus ochotensis E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Gonostomatidae Cyclothone alba Japan N LF 2 yr Asy J, A Miya and Nemoto (1986) Cyclothone atraria Japan N LF 7 yr Li J, A Miya and Nemoto (1987a) Sigmops elongatus E Gulf of Mexico Y O (d) 21 mo Li J, A Lancraft et al. (1988) Sigmops gracilis E China Sea (Japan) ? O (d) – – L Sassa and Takahashi (2018) Melamphaeidae Poromitra crassiceps E North Pacific N O (a) 8.5 Asy J, A Childress et al. (1980) Myctophidae Benthosema fibulatum Arabian Sea Y O (d) – Asy L, J Gjøsæter (1987) Benthosema glaciale W North Atlantic (Nova Scotia) Y O (a); LF 4.5 yr Asy J, A Halliday (1970) Benthosema glaciale E North Atlantic (Rockall Trough) Y LF – Asy L, J, A Kawaguchi and Mauchline (1982) Benthosema glaciale Norwegian Fjords Y O (a); LF 5 yr Asy J, A Gjøsæter (1973) Benthosema glaciale Norwegian Fjords Y O (a) 7 yr Asy Kristoffersen and Salvanes (2009) Benthosema glaciale W North Atlantic (Flemish Cap) Y O (a) 7 yr Asy J, A Garcia-Seoane et al. (2015) Benthosema pterotum Arabian Sea Y O (d) ∼1 yr Asy J, A Gjøsæter (1984) Benthosema pterotum Arabian Sea Y O (d) – Asy L, J Gjøsæter (1987) Benthosema pterotum E China Sea (Japan) Y O (d) – Asy L, J, A Ozawa and Peñaflor (1990) Benthosema pterotum E China Sea Y O (d) – Asy L Sassa et al. (2015) Benthosema suborbitale Gulf of Mexico Y O (d) 325 d Asy L, J, A Gartner (1991b) Benthosema suborbitale E Gulf of Mexico Y O (d) – Asy L Conley and Gartner (2009) Ceratoscopelus maderensis E North Atlantic Y O (d) ∼2 yr Asy J, A Linkowski et al. (1993) Ceratoscopelus townsendi E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Ceratoscopelus warmingii E North Atlantic Y O (d) 400 d Asy J, A Linkowski et al. (1993) Ceratoscopelus warmingii Indian Ocean Y O (d) – Asy J Tsarin (1994) Ceratoscopelus warmingii W North Pacific Y O (d) 416 d Asy L, J, A Takagi et al. (2006) Diaphus diademophilus Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Diaphus dumerilli Gulf of Mexico Y O (d) 362 d Asy L, J, A Gartner (1991b) Diaphus kapalae Coral Sea Y O (d) – – J Suthers (1996) Diaphus suborbitalis W North Pacific (Japan) ? O (a) 2.5 yr Asy J, A Go et al. (1977a) Diaphus suborbitalis W North Pacific (Japan) ? LF – Asy J, A Go et al. (1977b) Diaphus theta W North Pacific Y O (d) – – L, J Moku et al. (2001) Diaphus slender type E China Sea ? – – L Moku et al. (2005) Diaphus sp. E China Sea (Japan) ? O (d) – – L Sassa and Takahashi (2018) Diogenichthys laternatus E South Pacific (coastal Chile) Y O (d) – – L Landaeta et al. (2015b) Electrona antarctica Antarctica Y O (a) – – J, A Rowedder (1979) Electrona antarctica Southern Ocean Y O (a) 5 yr – J, A Shust and Kochkin (1985) Electrona antarctica Antarctica Y O (d) 3.5 yr Li J, A Greely et al. (1999) Electrona carlbergi South Atlantic Y 5 yr – J, A Konstantinova (1987) Electrona carlbergi Southern Ocean Y O (a) – – J, A Shust and Kochkin (1985) Gymnoscopelus braueri Scotia Sea Y O (a); LF 6 yr Asy A Saunders et al. (2020) Gymnoscopelus nicholsi South Atlantic Y O (a) 7 yrs Asy J, A Linkowski (1985) Gymnoscopelus nicholsi South Atlantic Y 5 yrs Asy J, A Konstantinova (1987) Hygophum benoitti North Atlantic ? IGS – – J, A Linkowski (1996) Hygophum hygomii North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum macrochir North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum reinhardtii North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum taaningi North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum taaningi E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Kreffichthys anderssoni Scotia Sea Y O (a); LF 2 yr Asy J, A Saunders et al. (2020) Lampanyctodes hectoris W Tasman Sea (Tasmania) Y O (d); LF 3 yr Asy L, J, A Young et al. (1988) Lampanyctus macdonaldi E North Atlantic (Rockall Trough) ? LF 6 yr Asy L, J, A Kawaguchi and Mauchline (1982) Lampanyctus regalis E North Pacific N O (a) 4.5 yr Li J, A Childress et al. (1980) Lampanyctus ritteri E North Pacific Y O (a) 5.5 yr Asy J, A Childress et al. (1980) Lampanyctus sp. Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Lepidophanes guentheri Gulf of Mexico Y O (d) 439 d Asy L, J, A Gartner (1991b) Myctophum affine W South Atlantic (Brazil) Y O (d) – – L Namiki et al. (2015) Myctophum asperum North Pacific Y O (d) – – J, A Hayashi et al. (2001) Myctophum asperum E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Myctophum asperum South China Sea Y O (d) 440 d Asy J, A Wang et al. (2019) Myctophum nitidulum Tropical Atlantic Y O (d) 1 yr – L, J, A Giragosov and Ovcharov (1992) Myctophum nitidulum E South Pacific (coastal Chile) Y O (d) – – L Landaeta et al. (2015b) Myctophum selenops E Gulf of Mexico ? O (d) – – L Conley and Gartner (2009) Myctophum spinosum Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Notolychnus valdiviae E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Protomyctophum arcticum E North Atlantic (Rockall Trough) M LF 1.25 yr Li L, J, A Kawaguchi and Mauchline (1982) Stenobrachius leucopsaurus E North Pacific Y O (a); LF 8 yr Asy L, J, A Smoker and Pearcy (1970) Stenobrachius leucopsaurus E North Pacific Y O (a) 7.5 yr Asy J, A Childress et al. (1980) Stenobrachius leucopsaurus W North Pacific Y O (d) – – L Methot (1981) Symbolophorus californiensis W North Pacific Y O (d) 541 d Asy L, J, A Takagi et al. (2006) Symbolophorus evermani Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Tarletonbeani crenularis E North Pacific Y O (d) 504 d Asy L, J, A Bystydzienska et al. (2010) Triphoturus mexicanus E North Pacific Y O (a) 5 yr Asy J, A Childress et al. (1980) Notoscopelidae Notoscopelus elongatus Norway and W British Isles Y O (d, a) 6 yr Asy J, A Gjøsæter (1981 b) Notoscopelus japonicus E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Notoscopelus resplendens E North Atlantic (Canary Islands) Y O (d, a) 4 yr Asy J, A Sarmiento-Lezcano et al. (2018) Phosichthyidae Vinciguerria nimbaria E China Sea (Japan) M O (d) – – L Sassa and Takahashi (2018) Vinciguerria nimbaria E tropical Atlantic M O (d) 240 d Asy L, J, A Tomas and Panfili (2000) Sternoptychidae Maurolicus muelleri Norwegian fjords and coast Y O (d); LF 3.5 yr Asy J, A Gjøsæter (1981a) Maurolicus muelleri Norwegian fjords Y LF – Asy J, A Goodson et al. (1995) Maurolicus muelleri Norwegian fjords Y LF – – J, A Salvanes and Stockley (1996) Maurolicus muelleri Norwegian fjords Y O (d) – – L Folkvord et al. (2016) Maurolicus parvipinnis Patagonian fjords Y O (d) – – L Zenteno et al. (2014) Maurolicus parvipinnis Patagonian fjords Y O (d) – – L Landaeta et al. (2015a) Valenciennellus tripunctulatus Caribbean, Gulf of Mexico M Model – – A Baird and Hopkins (1981) Stomiidae Borostomias panamensis E North Pacific N O (a) 5 yr Li J, A Childress et al. (1980) Tactostoma macropus E North Pacific Y O (a); LF 5 yr Asy L, J, A Fisher and Pearcy (1983) Trichiuridae Lepidopus caudatus NW Mediterranean N O (a) 8 yr Asy J, A Demestre et al. (1993) Family . Species . Location . DVM . Growth method . Max age . Growth rates . Life stage . Reference . Alepocephalidae Bajacalifornia burragei E North Pacific N O (a) 4 yr Li J, A Childress et al. (1980) Bathylagidae Pseudobathylagus milleri E North Pacific N O (a) 5 yr Asy J, A Childress et al. (1980) Leuroglossus stilbius E North Pacific Y O (a) 6 yr Semi-Asy J, A Childress et al. (1980) Lipolagus ochotensis E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Gonostomatidae Cyclothone alba Japan N LF 2 yr Asy J, A Miya and Nemoto (1986) Cyclothone atraria Japan N LF 7 yr Li J, A Miya and Nemoto (1987a) Sigmops elongatus E Gulf of Mexico Y O (d) 21 mo Li J, A Lancraft et al. (1988) Sigmops gracilis E China Sea (Japan) ? O (d) – – L Sassa and Takahashi (2018) Melamphaeidae Poromitra crassiceps E North Pacific N O (a) 8.5 Asy J, A Childress et al. (1980) Myctophidae Benthosema fibulatum Arabian Sea Y O (d) – Asy L, J Gjøsæter (1987) Benthosema glaciale W North Atlantic (Nova Scotia) Y O (a); LF 4.5 yr Asy J, A Halliday (1970) Benthosema glaciale E North Atlantic (Rockall Trough) Y LF – Asy L, J, A Kawaguchi and Mauchline (1982) Benthosema glaciale Norwegian Fjords Y O (a); LF 5 yr Asy J, A Gjøsæter (1973) Benthosema glaciale Norwegian Fjords Y O (a) 7 yr Asy Kristoffersen and Salvanes (2009) Benthosema glaciale W North Atlantic (Flemish Cap) Y O (a) 7 yr Asy J, A Garcia-Seoane et al. (2015) Benthosema pterotum Arabian Sea Y O (d) ∼1 yr Asy J, A Gjøsæter (1984) Benthosema pterotum Arabian Sea Y O (d) – Asy L, J Gjøsæter (1987) Benthosema pterotum E China Sea (Japan) Y O (d) – Asy L, J, A Ozawa and Peñaflor (1990) Benthosema pterotum E China Sea Y O (d) – Asy L Sassa et al. (2015) Benthosema suborbitale Gulf of Mexico Y O (d) 325 d Asy L, J, A Gartner (1991b) Benthosema suborbitale E Gulf of Mexico Y O (d) – Asy L Conley and Gartner (2009) Ceratoscopelus maderensis E North Atlantic Y O (d) ∼2 yr Asy J, A Linkowski et al. (1993) Ceratoscopelus townsendi E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Ceratoscopelus warmingii E North Atlantic Y O (d) 400 d Asy J, A Linkowski et al. (1993) Ceratoscopelus warmingii Indian Ocean Y O (d) – Asy J Tsarin (1994) Ceratoscopelus warmingii W North Pacific Y O (d) 416 d Asy L, J, A Takagi et al. (2006) Diaphus diademophilus Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Diaphus dumerilli Gulf of Mexico Y O (d) 362 d Asy L, J, A Gartner (1991b) Diaphus kapalae Coral Sea Y O (d) – – J Suthers (1996) Diaphus suborbitalis W North Pacific (Japan) ? O (a) 2.5 yr Asy J, A Go et al. (1977a) Diaphus suborbitalis W North Pacific (Japan) ? LF – Asy J, A Go et al. (1977b) Diaphus theta W North Pacific Y O (d) – – L, J Moku et al. (2001) Diaphus slender type E China Sea ? – – L Moku et al. (2005) Diaphus sp. E China Sea (Japan) ? O (d) – – L Sassa and Takahashi (2018) Diogenichthys laternatus E South Pacific (coastal Chile) Y O (d) – – L Landaeta et al. (2015b) Electrona antarctica Antarctica Y O (a) – – J, A Rowedder (1979) Electrona antarctica Southern Ocean Y O (a) 5 yr – J, A Shust and Kochkin (1985) Electrona antarctica Antarctica Y O (d) 3.5 yr Li J, A Greely et al. (1999) Electrona carlbergi South Atlantic Y 5 yr – J, A Konstantinova (1987) Electrona carlbergi Southern Ocean Y O (a) – – J, A Shust and Kochkin (1985) Gymnoscopelus braueri Scotia Sea Y O (a); LF 6 yr Asy A Saunders et al. (2020) Gymnoscopelus nicholsi South Atlantic Y O (a) 7 yrs Asy J, A Linkowski (1985) Gymnoscopelus nicholsi South Atlantic Y 5 yrs Asy J, A Konstantinova (1987) Hygophum benoitti North Atlantic ? IGS – – J, A Linkowski (1996) Hygophum hygomii North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum macrochir North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum reinhardtii North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum taaningi North Atlantic Y IGS – – J, A Linkowski (1996) Hygophum taaningi E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Kreffichthys anderssoni Scotia Sea Y O (a); LF 2 yr Asy J, A Saunders et al. (2020) Lampanyctodes hectoris W Tasman Sea (Tasmania) Y O (d); LF 3 yr Asy L, J, A Young et al. (1988) Lampanyctus macdonaldi E North Atlantic (Rockall Trough) ? LF 6 yr Asy L, J, A Kawaguchi and Mauchline (1982) Lampanyctus regalis E North Pacific N O (a) 4.5 yr Li J, A Childress et al. (1980) Lampanyctus ritteri E North Pacific Y O (a) 5.5 yr Asy J, A Childress et al. (1980) Lampanyctus sp. Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Lepidophanes guentheri Gulf of Mexico Y O (d) 439 d Asy L, J, A Gartner (1991b) Myctophum affine W South Atlantic (Brazil) Y O (d) – – L Namiki et al. (2015) Myctophum asperum North Pacific Y O (d) – – J, A Hayashi et al. (2001) Myctophum asperum E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Myctophum asperum South China Sea Y O (d) 440 d Asy J, A Wang et al. (2019) Myctophum nitidulum Tropical Atlantic Y O (d) 1 yr – L, J, A Giragosov and Ovcharov (1992) Myctophum nitidulum E South Pacific (coastal Chile) Y O (d) – – L Landaeta et al. (2015b) Myctophum selenops E Gulf of Mexico ? O (d) – – L Conley and Gartner (2009) Myctophum spinosum Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Notolychnus valdiviae E Gulf of Mexico Y O (d) – – L Conley and Gartner (2009) Protomyctophum arcticum E North Atlantic (Rockall Trough) M LF 1.25 yr Li L, J, A Kawaguchi and Mauchline (1982) Stenobrachius leucopsaurus E North Pacific Y O (a); LF 8 yr Asy L, J, A Smoker and Pearcy (1970) Stenobrachius leucopsaurus E North Pacific Y O (a) 7.5 yr Asy J, A Childress et al. (1980) Stenobrachius leucopsaurus W North Pacific Y O (d) – – L Methot (1981) Symbolophorus californiensis W North Pacific Y O (d) 541 d Asy L, J, A Takagi et al. (2006) Symbolophorus evermani Arabian Sea Y O (d) – – L, J Gjøsæter (1987) Tarletonbeani crenularis E North Pacific Y O (d) 504 d Asy L, J, A Bystydzienska et al. (2010) Triphoturus mexicanus E North Pacific Y O (a) 5 yr Asy J, A Childress et al. (1980) Notoscopelidae Notoscopelus elongatus Norway and W British Isles Y O (d, a) 6 yr Asy J, A Gjøsæter (1981 b) Notoscopelus japonicus E China Sea (Japan) Y O (d) – – L Sassa and Takahashi (2018) Notoscopelus resplendens E North Atlantic (Canary Islands) Y O (d, a) 4 yr Asy J, A Sarmiento-Lezcano et al. (2018) Phosichthyidae Vinciguerria nimbaria E China Sea (Japan) M O (d) – – L Sassa and Takahashi (2018) Vinciguerria nimbaria E tropical Atlantic M O (d) 240 d Asy L, J, A Tomas and Panfili (2000) Sternoptychidae Maurolicus muelleri Norwegian fjords and coast Y O (d); LF 3.5 yr Asy J, A Gjøsæter (1981a) Maurolicus muelleri Norwegian fjords Y LF – Asy J, A Goodson et al. (1995) Maurolicus muelleri Norwegian fjords Y LF – – J, A Salvanes and Stockley (1996) Maurolicus muelleri Norwegian fjords Y O (d) – – L Folkvord et al. (2016) Maurolicus parvipinnis Patagonian fjords Y O (d) – – L Zenteno et al. (2014) Maurolicus parvipinnis Patagonian fjords Y O (d) – – L Landaeta et al. (2015a) Valenciennellus tripunctulatus Caribbean, Gulf of Mexico M Model – – A Baird and Hopkins (1981) Stomiidae Borostomias panamensis E North Pacific N O (a) 5 yr Li J, A Childress et al. (1980) Tactostoma macropus E North Pacific Y O (a); LF 5 yr Asy L, J, A Fisher and Pearcy (1983) Trichiuridae Lepidopus caudatus NW Mediterranean N O (a) 8 yr Asy J, A Demestre et al. (1993) DVM information was also sourced from additional literature, and species names are those currently accepted by the California Academy of Sciences Eschmeyer’s Catalog of Fishes. DVM, diel vertical migration (Y, yes; N, no;?, unknown; M, mixed); O (d), otolith daily growth rings; O (a), otolith annual growth rings; IGS, incremental growth sequence; LF, length-frequency growth analyses; –, unknown or insufficient data; Asy, asymptotic; Li, linear; ID, insufficient data; L, larvae; J, juveniles; A, adults. Open in new tab Species of focus There has been a strong bias towards myctophids in studies of age and growth, with almost three quarters of the investigations on species from this family. This is no surprise given the role they play in the midwater ecosystems (Catul et al., 2011). They are the most diverse family of fishes in the mesopelagic, numbering 252 extant species (Fricke et al., 2020), and behind gonostomatids are the most abundant fishes in midwater assemblages (Sutton et al., 2008; Davison et al., 2015; Olivar et al., 2017). Myctophids are also one of the major components of the deep scattering layer, and the most abundant of the vertically migrating mesopelagic fishes (Gartner et al., 1987). On the contrary, we found only two investigations that studied growth in the genus Cyclothone of the family Gonostomatidae, reported as being the most abundant family of fishes in the mesopelagic. For instance, in the Sargasso Sea Cyclothone spp. accounted for 74.5% of fish abundance between 0 and 1000 m (Sutton et al., 2010). However, thus far age and growth research on this genus have been limited to studies in a semi-enclosed bay on the Japanese coast (Miya and Nemoto, 1986, 1987a, 1991). Whether these results hold true for oceanic populations is unknown. Certainly, for species that are so abundant, information of their basic biology is paramount, even more so given the wide range in variability shown in this genus in these studies (i.e. semelparity and iteroparity, maximum ages of 1–2 years and 6–7 years; Miya and Nemoto, 1986 , 1987a ). There are several other particularly common and/or cosmopolitan mesopelagic fishes that appear to lack age and growth research that could be important to understand. There were no studies on hatchetfishes (Sternoptychidae), an important component of the mesopelagic including migrators and non-migrators, with the only effort on sternoptychids confined to the pearlisdes Maurolicus spp. (Table 1). Other research gaps include but are not limited to members of the families Melamphaidae, Nemichthyidae, Serrivomeridae, and Stomiidae. Furthermore, taxonomic research on mesopelagic fishes indicates some inter-regional variations in morphologies and life histories, particularly among broadly distributed species. Therefore, common and/or important species that have been investigated could warrant studying again, but in regions with a particular focus of interest, for instance, where broader studies on carbon budgets or food webs are being undertaken, or where harvesting is expected to occur. More generally, the predominance of studies has been on vertical migrators, which is little surprise given their abundance and important vertical transport of carbon. However, only seven known mesopelagic non-migrators have been investigated for age and growth. Furthermore, robust data on DVM are important, especially those that are reported for partial, gender-based or ontogenetic variation in DVM, as this also has implications for energy/carbon transport and budgets. Age and growth techniques Otolith increment analyses are by far the most common method of estimating age and growth in mesopelagic fishes (Figure 1a and b). Otoliths are the standard method of ageing in fish demography, with a considerable amount of published information on the subject, making it an invaluable and comparable technique for estimating age and growth (Stevenson and Campana, 1992). Furthermore, otolith-based ageing is valuable because the samples can be used concurrently for other analyses such as stomach contents, reproduction and stable isotopes (Figure 1c). Figure 1. Open in new tabDownload slide (a) Extracted otolith from Sigmops elongatus, (b) whole otolith of Poromitra megalops, (c) dissected gonads, stomachs, and other contents of the peritoneal cavity of Melanostomias bartonbeani, and (d) histological micrograph of an ovary from Sebastes fasciatus. Image credits: Helena McMonagle (a, c), Lucinda Quigley (b), Lyndsey Lefebvre (d). Figure 1. Open in new tabDownload slide (a) Extracted otolith from Sigmops elongatus, (b) whole otolith of Poromitra megalops, (c) dissected gonads, stomachs, and other contents of the peritoneal cavity of Melanostomias bartonbeani, and (d) histological micrograph of an ovary from Sebastes fasciatus. Image credits: Helena McMonagle (a, c), Lucinda Quigley (b), Lyndsey Lefebvre (d). Complete growth curves (larvae/juveniles through adults) have been estimated using either daily or annual increment methods. The use of annuli is the most common method of estimating full growth curves in fish ageing studies because daily rings are only consistently readable for larvae and juveniles, and become increasingly difficult to read in fish more than a few months old where increment widths become too narrow (e.g. <1 µm), and/or otoliths exhibit non-planar increment formation (Brothers, 1979; Campana and Neilson, 1985). However, mesopelagic fishes had an unusually high proportion of studies providing full growth curves with daily increment analysis (14 using daily rings and 21 using annuli; Table 1). This is likely because of the short lifespan in many of these species. Supporting this, of the 14 investigations that provided full growth curves (i.e. with maximum age) utilizing daily increment counts, all but one was a species that lived <2 years (Table 1). The other was a polar species, Electrona antarctica, that lives ∼3.5 years (Greely et al., 1999). There were many other investigations that utilized daily increment analyses when investigating early life history in mesopelagic fishes (e.g. larval growth), a much more common use of daily increments in fish demography in general (Table 1). One further use of daily growth increments has been to confirm the cyclical nature of DVM. Linkowski (1996) correlated the abrupt changes in increment deposition in myctophids (Hygophum spp.) with the varying light phases of the moon, which in turn influences migration depth and thus growth rates. The main caveat with using otoliths to age mesopelagic fishes is the lack of validation of ring deposition periodicity (Geffen, 1992). Due to the nature of sampling fish from the deep-pelagic environments, validating otolith increments in these species is very difficult. There are some methods that are commonly used in coastal fishes, including identifying an environmentally driven mark on the otolith with a known date (e.g. a radiation spill), or marking the fish in some way (e.g. tags, chemicals) and then either releasing or rearing the fish for a known time period (Campana, 2001). The latter is practically impossible for deep midwater fishes. The most feasible method is marginal increment analysis (assessing the temporal progression of the distance from the last ring to the otolith’s margin), requiring high-resolution sampling, either over 24 h periods for validating daily rings (Moku et al., 2001) or throughout the year for validating annuli (Gartner, 1991a). To date, this has only been performed on a few species of mesopelagic fishes, almost all of which are myctophids (Gartner, 1991a; Hayashi et al., 2001; Moku et al., 2001; Sarmiento-Lezcano et al., 2018). Without validation, growth curves are based on assumptions of increment periodicity, which for annuli in seasonal oceans (i.e. temperate) at least, are usually deemed acceptable (Salvanes and Kristoffersen, 2001). Other widely used techniques for ageing mesopelagic fishes included length-frequency analysis. This also involves high-resolution sample collection (e.g. monthly/bi-monthly) and assesses growth by tracking cohorts throughout the year(s). This technique is particularly useful for species with difficult to read otoliths, plus there is no otolith validation and/or assumption issues. Length-frequency analysis has been used as a measure of growth in 12 investigations (Table 1). In an ideal situation, multiple methods can be utilized to substantiate growth rates. For example, an investigation on Tactostoma macropus in the North Pacific (Fisher and Pearcy, 1983) created growth curves using three techniques—length-frequency analysis, otolith ageing, and back calculation from otoliths—and all showed similar results, verifying the growth curves. Growth-related patterns in life history strategies There are several different growth and life-span related strategies in mesopelagic fishes (e.g. see Figure 2). In general, mesopelagic fishes are short-lived relative to many coastal and deep-sea benthic species (e.g. Tracey and Horn, 1999; Munk, 2001). However, there is some variation, ranging from those that only live one or 2 years (e.g. several myctophids) to others that live several years, seemingly peaking at about 7 or 8 years (e.g. Poromitra crassiceps and Stenobrachius leucopsaurus). There are also differences in lifespan between congeneric species. For instance, Benthosema glaciale has been shown to live up to 7 years (Garcia-Seoane et al., 2015) and Benthosema suborbitale only 1 year (Gartner, 1991b). Whether these maximum ages are consistent in these species throughout their entire range, or are a regional or habitat difference is not known. Only a small percentage of mesopelagic fish species has been aged, and knowledge of which strategy is most successful in the mesopelagic or in vertically migrating species is largely unknown. This information is important to know when evaluating the potential effects of fishing pressure and a rapidly changing climate on these fishes and mesopelagic ecosystems as a whole. Figure 2. Open in new tabDownload slide Example growth curves for mesopelagic fishes. Sources of data: Lampanyctus regalis, Poromitra crassiceps (Childress et al., 1980); Benthosema suborbitale (Gartner, 1991a); Sigmops elongatus (Lancraft et al., 1998); and Benthosema glaciale (Garcia-Seoane et al., 2015). Figure 2. Open in new tabDownload slide Example growth curves for mesopelagic fishes. Sources of data: Lampanyctus regalis, Poromitra crassiceps (Childress et al., 1980); Benthosema suborbitale (Gartner, 1991a); Sigmops elongatus (Lancraft et al., 1998); and Benthosema glaciale (Garcia-Seoane et al., 2015). The synthesized results also highlight how growth-rate strategies have a strong connection to vertical migration behaviour. Regularly, the daily vertically migrating species have relatively slower growth than many of the non-migrators (e.g. Figure 2). Many myctophids, almost all of which are migrators, have typical asymptotic growth curves. This is a general strategy in fishes where, approaching maturity, growth is typically slowed and energy is shifted towards gonad development (Kozłowski, 1992). In contrast, some deeper/non-migrating mesopelagic species exhibit more rapid linear indeterminate growth patterns (e.g. Bajacalifornia burragei, Borostomias panamensis, and Lampanyctus regalis). A study that incorporated growth rates and energy budgets in several migrating and non-migrating species concluded that, in general, the migrators were giving priority to energy storage and usage over size, and the non-migrators priority to size (Childress et al., 1980). The reasoning is that the energy required to migrate each day is substantial, and thus storing energy for this is essential. And for the non-migrators, attaining sizes that minimize predation risk as early as possible is likely selected for. Furthermore, non-migrating species offset some of the reduced energy demands with lower metabolism through lack of movement (i.e. visual predation in sunlit waters generally drives higher locomotion—Siebel and Drazen, 2007) and higher water content in their tissues (Koslow, 1996). Where data were sufficient to produce full growth curves in the studies we examined, this generalization held true ∼¾ of the time. That is, for asymptotic growth rates, 78% of the time (21/27) these were migrating species, and growth rates were linear indeterminate 71% of the time (5/7) they were for non-migrators (Table 1). However, these conclusions are based on a fairly small sample of species, so growth data on more species would be informative in this regard. Little is known about how much migration ceases or slows down later in life for species that exhibit asymptotic growth. There are exceptions to the above. For instance, Sigmops elongatus is one of the more common larger species, reported as only living around 21 months (Lancraft et al., 1988) (Figure 2). Furthermore, for a species that is known to migrate, it has a fast linear growth rate and does not exhibit a pronounced slowing in growth when a certain size is attained (there is, however, some evidence of ontogenetic variation in migration depths in this species—Lancraft et al., 1988; Hopkins and Sutton, 1998). An even more uncommon strategy was seen in P, crassiceps, which exhibits an exponential growth curve, increasing in growth rate throughout its 7 or 8 years of life (Childress et al., 1980) (Figure 2). These two examples are likely explained by their semelparous reproductive strategies, whereby investing in body growth to maximize fecundity for a single spawning event. This highlights that age and growth are often intertwined with reproduction and benefit from being considered together. Furthermore, more robust data are needed on these species with different or in-between strategies, especially ones as common and cosmopolitan as S. elongatus. Reproduction Information on the dynamics of reproduction is vital for estimating the rates at which mesopelagic fishes can replenish themselves, and thus how resilient they are to impacts on midwater ecosystems. Assessing the reproductive dynamics of a species involves understanding the basic reproductive biology, including the physiology of the reproductive anatomy, sex ratios, sex pattern (gonochoristic or hermaphroditic), and lifetime reproductive opportunities (semelparous or iteroparous). Additionally, an understanding of the more complex aspects of reproduction—age or size at maturity, fecundity, spawning seasonality and phenology, spawning frequency, and spawning migration behaviour—addresses the questions of where, when, how, and how many offspring are produced, with the first two perhaps the most influential parameters. A combination of literature databases, scholarly search engines, citations, and reference lists were searched for all studies pertaining to reproduction in mesopelagic fishes. In all, 36 studies revealed 66 investigations representing 50 species from nine families (Table 2). An additional five studies with a broad focus on the general biology of mesopelagic fish families provided limited reproductive data on a large number of species but were not included in Table 2 for brevity (Clarke, 1973, 1983; Howell and Krueger, 1987; Karnella, 1987; Keene et al., 1987). Additionally, several studies describing gonadal morphology (Fishelson, 1994; Forsgren et al., 2017) and inferring spawning timing or location of adults based on egg and larval studies (Gjøsæter and Tilseth, 1988; Landaeta and Castro, 2002) are worth noting but were not included in Table 2. Table 2. Summary information for investigations of reproduction in mesopelagic fishes. Family . Species . Location . DVM . Seas . SR . Mat . Hist . Fec . Reference . Alepocephalidae Bajacalifornia burragei E North Pacific N N N D N N Childress et al. (1980) Bathylagidae Leuroglossus stilbius E North Pacific Y Y N D N N Childress et al. (1980) Pseudobathylagus milleri E North Pacific N Y N D N N Childress et al. (1980) Gonostomatidae Cyclothone alba Japan N Y N G, D N T Miya and Nemoto (1986, 1991) Cyclothone atraria Japan N Y N G, D N T Miya and Nemoto (1987a, 1991) Cyclothone pseudopallida Japan N Y N G, D N T Miya and Nemoto (1987b, 1991) Sigmops elongatus E Gulf of Mexico Y N N H Y N Fisher (1983) Sigmops elongatus E Gulf of Mexico Y Y N G N N Lancraft et al. (1988) Melamphaeidae Poromitra crassiceps E North Pacific N N N D N N Childress et al. (1980) Myctophidae Benthosema fibulatum Arabian Sea Y Y Y H Y N Hussain (1992) Benthosema fibulatum Arabian Sea Y Y N N N T Hussain and Ali-Khan (1987) Benthosema glaciale E North Atlantic Y Y N H* Y S, R García-Seoane et al. (2014) Benthosema glaciale W North Atlantic (Nova Scotia) Y Y N G N N Halliday (1970) Benthosema glaciale E North Atlantic (Rockall Trough) Y Y Y G N T Kawaguchi and Mauchline (1982) Benthosema pterotum W Indian Ocean and Bay of Bengal Y Y N D Y T, R Dalpadado (1988) Benthosema pterotum Red Sea and Gulf of Aden Y N Y G N T, R Dalpadado and Gjøsaeter, 1987) Benthosema pterotum Arabian Sea Y Y N N N T Hussain and Ali-Khan (1987) Benthosema pterotum E China Sea Y Y Y I Y T, R Sassa et al. (2014) Benthosema suborbitale Gulf of Mexico Y Y Y G* Y T Gartner (1993) Ceratoscopelus warmingii Gulf of Mexico Y Y Y G* Y T Gartner (1993) Diaphus chrysorhynchus E China Sea Y Y Y I Y Gr, R Sassa et al. (2016) Diaphus coeruleus Arabian Sea N N N N N T Meera et al. (2019) Diaphus dumerili Gulf of Mexico ? N Y G* Y N Gartner (1993) Diaphus garmani E China Sea Y Y N I Y Gr, R Sassa et al. (2016) Diaphus suborbitalis Indian Ocean Y N Y G, D Y T(Gr) Lisovenko and Prut’ko (1987a, b) Diaphus watasei E China Sea N Y Y I Y Gr, R Sassa et al. (2016) Lampanyctodes hectoris W Tasman Sea (Tasmania) ? Y N H Y T Young et al. (1987) Electrona antarctica Antarctica Y Y Y Y N N Lisovenko and Efremenko (1983) Electrona carlsbergi South Atlantic Y Y ? Y Y N Konstantinova (1987) Electrona carlsbergi Antarctica Y Y ? Y ? Y Mazhiryna and Poletayev (1990) Lampanyctodes hectoris E South Atlantic (South Africa) ? Y N G N T Prosch (1991) Lampanyctus alatus Gulf of Mexico ? Y Y G* Y T Gartner (1993) Lampanyctus crocodilus W Mediterranean Y Y N I N N Fanelli et al. (2014) Lampanyctus macdonaldi E North Atlantic (Rockall Trough) ? Y N N N T Kawaguchi and Mauchline (1982) Lampanyctus regalis E North Pacific N N N D N N Childress et al. (1980) Lampanyctus ritteri E North Pacific Y Y N D N N Childress et al. (1980) Lepidophanes guentheri Gulf of Mexico Y Y Y G* Y T Gartner (1993) Myctophum affine Gulf of Mexico ? N N G Y T Gartner (1993) Notolychnus valdiviae Gulf of Mexico Y Y Y G* Y T Gartner (1993) Protomyctophum arcticum E North Atlantic (Rockall Trough) N Y N N N T Kawaguchi and Mauchline (1982) Stenobrachius leucopsarus E North Pacific Y Y N D N N Childress et al. (1980) Stenobrachius leucopsarus E North Pacific Y Y N D N N Smoker and Pearcy (1970) Phosichthyidae Vinciguerria nimbaria E Atlantic (equatorial) M Y N H* Y T, R Stequert et al. (2003) Sternoptychidae Maurolicus muelleri Red Sea, Gulf of Aden Y N Y G N T, R Dalpadado and Gjøsæter (1987) Maurolicus muelleri Norwegian fjords Y Y N G Y T Gjøsæter (1981a) Maurolicus muelleri Norwegian fjords Y Y N G N T Goodson et al. (1995) Maurolicus muelleri Japan Sea Y Y N N N T Ikeda (1994) Maurolicus muelleri E South Atlantic (South Africa) Y Y N G N T Prosch (1991) Maurolicus muelleri Norwegian fjords Y N N N N Gr Salvanes and Stockley (1996) Maurolicus muelleri W Tasman Sea (Tasmania) Y Y N H Y T Young et al. (1987) Maurolicus stehmanni W South Atlantic (coastal Brazil) ? Y Y G* Y N de Almeida and Rossi-Wongtschowski (2007) Stomiidae Aristostomias xenostoma Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes atlanticus Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes richardsoni Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes similus Gulf of Mexico N Y Y H Y N Marks et al. (2020) Borostomias panamensis E North Pacific N N N D N N Childress et al. (1980) Chauliodus sloani Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Echiostoma barbatum Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias fissibarbis Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias hypopsilus Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias schmidti Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Malacosteus niger Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Melanostomias melanops Gulf of Mexico N Y Y H Y N Marks et al. (2020) Photostomias guernei Gulf of Mexico N Y Y H Y N Marks et al. (2020) Tactostoma macropus E North Pacific Y Y N D N V Fisher and Pearcy (1983) Trichiuridae Lepidopus caudatus W Mediterranean N Y N G* N N Demestre et al. (1993) Family . Species . Location . DVM . Seas . SR . Mat . Hist . Fec . Reference . Alepocephalidae Bajacalifornia burragei E North Pacific N N N D N N Childress et al. (1980) Bathylagidae Leuroglossus stilbius E North Pacific Y Y N D N N Childress et al. (1980) Pseudobathylagus milleri E North Pacific N Y N D N N Childress et al. (1980) Gonostomatidae Cyclothone alba Japan N Y N G, D N T Miya and Nemoto (1986, 1991) Cyclothone atraria Japan N Y N G, D N T Miya and Nemoto (1987a, 1991) Cyclothone pseudopallida Japan N Y N G, D N T Miya and Nemoto (1987b, 1991) Sigmops elongatus E Gulf of Mexico Y N N H Y N Fisher (1983) Sigmops elongatus E Gulf of Mexico Y Y N G N N Lancraft et al. (1988) Melamphaeidae Poromitra crassiceps E North Pacific N N N D N N Childress et al. (1980) Myctophidae Benthosema fibulatum Arabian Sea Y Y Y H Y N Hussain (1992) Benthosema fibulatum Arabian Sea Y Y N N N T Hussain and Ali-Khan (1987) Benthosema glaciale E North Atlantic Y Y N H* Y S, R García-Seoane et al. (2014) Benthosema glaciale W North Atlantic (Nova Scotia) Y Y N G N N Halliday (1970) Benthosema glaciale E North Atlantic (Rockall Trough) Y Y Y G N T Kawaguchi and Mauchline (1982) Benthosema pterotum W Indian Ocean and Bay of Bengal Y Y N D Y T, R Dalpadado (1988) Benthosema pterotum Red Sea and Gulf of Aden Y N Y G N T, R Dalpadado and Gjøsaeter, 1987) Benthosema pterotum Arabian Sea Y Y N N N T Hussain and Ali-Khan (1987) Benthosema pterotum E China Sea Y Y Y I Y T, R Sassa et al. (2014) Benthosema suborbitale Gulf of Mexico Y Y Y G* Y T Gartner (1993) Ceratoscopelus warmingii Gulf of Mexico Y Y Y G* Y T Gartner (1993) Diaphus chrysorhynchus E China Sea Y Y Y I Y Gr, R Sassa et al. (2016) Diaphus coeruleus Arabian Sea N N N N N T Meera et al. (2019) Diaphus dumerili Gulf of Mexico ? N Y G* Y N Gartner (1993) Diaphus garmani E China Sea Y Y N I Y Gr, R Sassa et al. (2016) Diaphus suborbitalis Indian Ocean Y N Y G, D Y T(Gr) Lisovenko and Prut’ko (1987a, b) Diaphus watasei E China Sea N Y Y I Y Gr, R Sassa et al. (2016) Lampanyctodes hectoris W Tasman Sea (Tasmania) ? Y N H Y T Young et al. (1987) Electrona antarctica Antarctica Y Y Y Y N N Lisovenko and Efremenko (1983) Electrona carlsbergi South Atlantic Y Y ? Y Y N Konstantinova (1987) Electrona carlsbergi Antarctica Y Y ? Y ? Y Mazhiryna and Poletayev (1990) Lampanyctodes hectoris E South Atlantic (South Africa) ? Y N G N T Prosch (1991) Lampanyctus alatus Gulf of Mexico ? Y Y G* Y T Gartner (1993) Lampanyctus crocodilus W Mediterranean Y Y N I N N Fanelli et al. (2014) Lampanyctus macdonaldi E North Atlantic (Rockall Trough) ? Y N N N T Kawaguchi and Mauchline (1982) Lampanyctus regalis E North Pacific N N N D N N Childress et al. (1980) Lampanyctus ritteri E North Pacific Y Y N D N N Childress et al. (1980) Lepidophanes guentheri Gulf of Mexico Y Y Y G* Y T Gartner (1993) Myctophum affine Gulf of Mexico ? N N G Y T Gartner (1993) Notolychnus valdiviae Gulf of Mexico Y Y Y G* Y T Gartner (1993) Protomyctophum arcticum E North Atlantic (Rockall Trough) N Y N N N T Kawaguchi and Mauchline (1982) Stenobrachius leucopsarus E North Pacific Y Y N D N N Childress et al. (1980) Stenobrachius leucopsarus E North Pacific Y Y N D N N Smoker and Pearcy (1970) Phosichthyidae Vinciguerria nimbaria E Atlantic (equatorial) M Y N H* Y T, R Stequert et al. (2003) Sternoptychidae Maurolicus muelleri Red Sea, Gulf of Aden Y N Y G N T, R Dalpadado and Gjøsæter (1987) Maurolicus muelleri Norwegian fjords Y Y N G Y T Gjøsæter (1981a) Maurolicus muelleri Norwegian fjords Y Y N G N T Goodson et al. (1995) Maurolicus muelleri Japan Sea Y Y N N N T Ikeda (1994) Maurolicus muelleri E South Atlantic (South Africa) Y Y N G N T Prosch (1991) Maurolicus muelleri Norwegian fjords Y N N N N Gr Salvanes and Stockley (1996) Maurolicus muelleri W Tasman Sea (Tasmania) Y Y N H Y T Young et al. (1987) Maurolicus stehmanni W South Atlantic (coastal Brazil) ? Y Y G* Y N de Almeida and Rossi-Wongtschowski (2007) Stomiidae Aristostomias xenostoma Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes atlanticus Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes richardsoni Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes similus Gulf of Mexico N Y Y H Y N Marks et al. (2020) Borostomias panamensis E North Pacific N N N D N N Childress et al. (1980) Chauliodus sloani Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Echiostoma barbatum Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias fissibarbis Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias hypopsilus Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias schmidti Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Malacosteus niger Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Melanostomias melanops Gulf of Mexico N Y Y H Y N Marks et al. (2020) Photostomias guernei Gulf of Mexico N Y Y H Y N Marks et al. (2020) Tactostoma macropus E North Pacific Y Y N D N V Fisher and Pearcy (1983) Trichiuridae Lepidopus caudatus W Mediterranean N Y N G* N N Demestre et al. (1993) Methods of determining maturity (Mat) were based on the diameter of oocytes (D); gross macroscopic appearance of the gonad (G); histological examination of the gonad (H); or the gonadosomatic index (I). An asterisk (*) indicates a maturity ogive was estimated. Fecundity (Fec) was estimated as the total count (T) of either all oocytes, the most advanced oocyte stage, or all oocytes above a minimum size threshold within the ovary; or by using the gravimetric (Gr), stereometric (S), or volumetric (V) methods. DVM was also sourced from additional resources, and species names are those currently accepted by the California Academy of Sciences Eschmeyer’s Catalog of Fishes). DVM, daily vertical migration; Hist, histological analysis of gonad tissue; M, mixed; N, no; R, relative fecundity; Seas, spawning season; SR, sex ratio; ?, unknown. Open in new tab Table 2. Summary information for investigations of reproduction in mesopelagic fishes. Family . Species . Location . DVM . Seas . SR . Mat . Hist . Fec . Reference . Alepocephalidae Bajacalifornia burragei E North Pacific N N N D N N Childress et al. (1980) Bathylagidae Leuroglossus stilbius E North Pacific Y Y N D N N Childress et al. (1980) Pseudobathylagus milleri E North Pacific N Y N D N N Childress et al. (1980) Gonostomatidae Cyclothone alba Japan N Y N G, D N T Miya and Nemoto (1986, 1991) Cyclothone atraria Japan N Y N G, D N T Miya and Nemoto (1987a, 1991) Cyclothone pseudopallida Japan N Y N G, D N T Miya and Nemoto (1987b, 1991) Sigmops elongatus E Gulf of Mexico Y N N H Y N Fisher (1983) Sigmops elongatus E Gulf of Mexico Y Y N G N N Lancraft et al. (1988) Melamphaeidae Poromitra crassiceps E North Pacific N N N D N N Childress et al. (1980) Myctophidae Benthosema fibulatum Arabian Sea Y Y Y H Y N Hussain (1992) Benthosema fibulatum Arabian Sea Y Y N N N T Hussain and Ali-Khan (1987) Benthosema glaciale E North Atlantic Y Y N H* Y S, R García-Seoane et al. (2014) Benthosema glaciale W North Atlantic (Nova Scotia) Y Y N G N N Halliday (1970) Benthosema glaciale E North Atlantic (Rockall Trough) Y Y Y G N T Kawaguchi and Mauchline (1982) Benthosema pterotum W Indian Ocean and Bay of Bengal Y Y N D Y T, R Dalpadado (1988) Benthosema pterotum Red Sea and Gulf of Aden Y N Y G N T, R Dalpadado and Gjøsaeter, 1987) Benthosema pterotum Arabian Sea Y Y N N N T Hussain and Ali-Khan (1987) Benthosema pterotum E China Sea Y Y Y I Y T, R Sassa et al. (2014) Benthosema suborbitale Gulf of Mexico Y Y Y G* Y T Gartner (1993) Ceratoscopelus warmingii Gulf of Mexico Y Y Y G* Y T Gartner (1993) Diaphus chrysorhynchus E China Sea Y Y Y I Y Gr, R Sassa et al. (2016) Diaphus coeruleus Arabian Sea N N N N N T Meera et al. (2019) Diaphus dumerili Gulf of Mexico ? N Y G* Y N Gartner (1993) Diaphus garmani E China Sea Y Y N I Y Gr, R Sassa et al. (2016) Diaphus suborbitalis Indian Ocean Y N Y G, D Y T(Gr) Lisovenko and Prut’ko (1987a, b) Diaphus watasei E China Sea N Y Y I Y Gr, R Sassa et al. (2016) Lampanyctodes hectoris W Tasman Sea (Tasmania) ? Y N H Y T Young et al. (1987) Electrona antarctica Antarctica Y Y Y Y N N Lisovenko and Efremenko (1983) Electrona carlsbergi South Atlantic Y Y ? Y Y N Konstantinova (1987) Electrona carlsbergi Antarctica Y Y ? Y ? Y Mazhiryna and Poletayev (1990) Lampanyctodes hectoris E South Atlantic (South Africa) ? Y N G N T Prosch (1991) Lampanyctus alatus Gulf of Mexico ? Y Y G* Y T Gartner (1993) Lampanyctus crocodilus W Mediterranean Y Y N I N N Fanelli et al. (2014) Lampanyctus macdonaldi E North Atlantic (Rockall Trough) ? Y N N N T Kawaguchi and Mauchline (1982) Lampanyctus regalis E North Pacific N N N D N N Childress et al. (1980) Lampanyctus ritteri E North Pacific Y Y N D N N Childress et al. (1980) Lepidophanes guentheri Gulf of Mexico Y Y Y G* Y T Gartner (1993) Myctophum affine Gulf of Mexico ? N N G Y T Gartner (1993) Notolychnus valdiviae Gulf of Mexico Y Y Y G* Y T Gartner (1993) Protomyctophum arcticum E North Atlantic (Rockall Trough) N Y N N N T Kawaguchi and Mauchline (1982) Stenobrachius leucopsarus E North Pacific Y Y N D N N Childress et al. (1980) Stenobrachius leucopsarus E North Pacific Y Y N D N N Smoker and Pearcy (1970) Phosichthyidae Vinciguerria nimbaria E Atlantic (equatorial) M Y N H* Y T, R Stequert et al. (2003) Sternoptychidae Maurolicus muelleri Red Sea, Gulf of Aden Y N Y G N T, R Dalpadado and Gjøsæter (1987) Maurolicus muelleri Norwegian fjords Y Y N G Y T Gjøsæter (1981a) Maurolicus muelleri Norwegian fjords Y Y N G N T Goodson et al. (1995) Maurolicus muelleri Japan Sea Y Y N N N T Ikeda (1994) Maurolicus muelleri E South Atlantic (South Africa) Y Y N G N T Prosch (1991) Maurolicus muelleri Norwegian fjords Y N N N N Gr Salvanes and Stockley (1996) Maurolicus muelleri W Tasman Sea (Tasmania) Y Y N H Y T Young et al. (1987) Maurolicus stehmanni W South Atlantic (coastal Brazil) ? Y Y G* Y N de Almeida and Rossi-Wongtschowski (2007) Stomiidae Aristostomias xenostoma Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes atlanticus Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes richardsoni Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes similus Gulf of Mexico N Y Y H Y N Marks et al. (2020) Borostomias panamensis E North Pacific N N N D N N Childress et al. (1980) Chauliodus sloani Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Echiostoma barbatum Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias fissibarbis Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias hypopsilus Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias schmidti Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Malacosteus niger Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Melanostomias melanops Gulf of Mexico N Y Y H Y N Marks et al. (2020) Photostomias guernei Gulf of Mexico N Y Y H Y N Marks et al. (2020) Tactostoma macropus E North Pacific Y Y N D N V Fisher and Pearcy (1983) Trichiuridae Lepidopus caudatus W Mediterranean N Y N G* N N Demestre et al. (1993) Family . Species . Location . DVM . Seas . SR . Mat . Hist . Fec . Reference . Alepocephalidae Bajacalifornia burragei E North Pacific N N N D N N Childress et al. (1980) Bathylagidae Leuroglossus stilbius E North Pacific Y Y N D N N Childress et al. (1980) Pseudobathylagus milleri E North Pacific N Y N D N N Childress et al. (1980) Gonostomatidae Cyclothone alba Japan N Y N G, D N T Miya and Nemoto (1986, 1991) Cyclothone atraria Japan N Y N G, D N T Miya and Nemoto (1987a, 1991) Cyclothone pseudopallida Japan N Y N G, D N T Miya and Nemoto (1987b, 1991) Sigmops elongatus E Gulf of Mexico Y N N H Y N Fisher (1983) Sigmops elongatus E Gulf of Mexico Y Y N G N N Lancraft et al. (1988) Melamphaeidae Poromitra crassiceps E North Pacific N N N D N N Childress et al. (1980) Myctophidae Benthosema fibulatum Arabian Sea Y Y Y H Y N Hussain (1992) Benthosema fibulatum Arabian Sea Y Y N N N T Hussain and Ali-Khan (1987) Benthosema glaciale E North Atlantic Y Y N H* Y S, R García-Seoane et al. (2014) Benthosema glaciale W North Atlantic (Nova Scotia) Y Y N G N N Halliday (1970) Benthosema glaciale E North Atlantic (Rockall Trough) Y Y Y G N T Kawaguchi and Mauchline (1982) Benthosema pterotum W Indian Ocean and Bay of Bengal Y Y N D Y T, R Dalpadado (1988) Benthosema pterotum Red Sea and Gulf of Aden Y N Y G N T, R Dalpadado and Gjøsaeter, 1987) Benthosema pterotum Arabian Sea Y Y N N N T Hussain and Ali-Khan (1987) Benthosema pterotum E China Sea Y Y Y I Y T, R Sassa et al. (2014) Benthosema suborbitale Gulf of Mexico Y Y Y G* Y T Gartner (1993) Ceratoscopelus warmingii Gulf of Mexico Y Y Y G* Y T Gartner (1993) Diaphus chrysorhynchus E China Sea Y Y Y I Y Gr, R Sassa et al. (2016) Diaphus coeruleus Arabian Sea N N N N N T Meera et al. (2019) Diaphus dumerili Gulf of Mexico ? N Y G* Y N Gartner (1993) Diaphus garmani E China Sea Y Y N I Y Gr, R Sassa et al. (2016) Diaphus suborbitalis Indian Ocean Y N Y G, D Y T(Gr) Lisovenko and Prut’ko (1987a, b) Diaphus watasei E China Sea N Y Y I Y Gr, R Sassa et al. (2016) Lampanyctodes hectoris W Tasman Sea (Tasmania) ? Y N H Y T Young et al. (1987) Electrona antarctica Antarctica Y Y Y Y N N Lisovenko and Efremenko (1983) Electrona carlsbergi South Atlantic Y Y ? Y Y N Konstantinova (1987) Electrona carlsbergi Antarctica Y Y ? Y ? Y Mazhiryna and Poletayev (1990) Lampanyctodes hectoris E South Atlantic (South Africa) ? Y N G N T Prosch (1991) Lampanyctus alatus Gulf of Mexico ? Y Y G* Y T Gartner (1993) Lampanyctus crocodilus W Mediterranean Y Y N I N N Fanelli et al. (2014) Lampanyctus macdonaldi E North Atlantic (Rockall Trough) ? Y N N N T Kawaguchi and Mauchline (1982) Lampanyctus regalis E North Pacific N N N D N N Childress et al. (1980) Lampanyctus ritteri E North Pacific Y Y N D N N Childress et al. (1980) Lepidophanes guentheri Gulf of Mexico Y Y Y G* Y T Gartner (1993) Myctophum affine Gulf of Mexico ? N N G Y T Gartner (1993) Notolychnus valdiviae Gulf of Mexico Y Y Y G* Y T Gartner (1993) Protomyctophum arcticum E North Atlantic (Rockall Trough) N Y N N N T Kawaguchi and Mauchline (1982) Stenobrachius leucopsarus E North Pacific Y Y N D N N Childress et al. (1980) Stenobrachius leucopsarus E North Pacific Y Y N D N N Smoker and Pearcy (1970) Phosichthyidae Vinciguerria nimbaria E Atlantic (equatorial) M Y N H* Y T, R Stequert et al. (2003) Sternoptychidae Maurolicus muelleri Red Sea, Gulf of Aden Y N Y G N T, R Dalpadado and Gjøsæter (1987) Maurolicus muelleri Norwegian fjords Y Y N G Y T Gjøsæter (1981a) Maurolicus muelleri Norwegian fjords Y Y N G N T Goodson et al. (1995) Maurolicus muelleri Japan Sea Y Y N N N T Ikeda (1994) Maurolicus muelleri E South Atlantic (South Africa) Y Y N G N T Prosch (1991) Maurolicus muelleri Norwegian fjords Y N N N N Gr Salvanes and Stockley (1996) Maurolicus muelleri W Tasman Sea (Tasmania) Y Y N H Y T Young et al. (1987) Maurolicus stehmanni W South Atlantic (coastal Brazil) ? Y Y G* Y N de Almeida and Rossi-Wongtschowski (2007) Stomiidae Aristostomias xenostoma Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes atlanticus Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes richardsoni Gulf of Mexico N Y Y H Y N Marks et al. (2020) Astronesthes similus Gulf of Mexico N Y Y H Y N Marks et al. (2020) Borostomias panamensis E North Pacific N N N D N N Childress et al. (1980) Chauliodus sloani Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Echiostoma barbatum Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias fissibarbis Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias hypopsilus Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Eustomias schmidti Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Malacosteus niger Gulf of Mexico N Y Y H* Y N Marks et al. (2020) Melanostomias melanops Gulf of Mexico N Y Y H Y N Marks et al. (2020) Photostomias guernei Gulf of Mexico N Y Y H Y N Marks et al. (2020) Tactostoma macropus E North Pacific Y Y N D N V Fisher and Pearcy (1983) Trichiuridae Lepidopus caudatus W Mediterranean N Y N G* N N Demestre et al. (1993) Methods of determining maturity (Mat) were based on the diameter of oocytes (D); gross macroscopic appearance of the gonad (G); histological examination of the gonad (H); or the gonadosomatic index (I). An asterisk (*) indicates a maturity ogive was estimated. Fecundity (Fec) was estimated as the total count (T) of either all oocytes, the most advanced oocyte stage, or all oocytes above a minimum size threshold within the ovary; or by using the gravimetric (Gr), stereometric (S), or volumetric (V) methods. DVM was also sourced from additional resources, and species names are those currently accepted by the California Academy of Sciences Eschmeyer’s Catalog of Fishes). DVM, daily vertical migration; Hist, histological analysis of gonad tissue; M, mixed; N, no; R, relative fecundity; Seas, spawning season; SR, sex ratio; ?, unknown. Open in new tab Species of focus There have been fewer reproductive studies on mesopelagic fishes than age and growth studies (Table 2). Myctophids were again the most commonly targeted study species and comprised nearly half of all investigations (33 investigations). There were four investigations into the most abundant group of fishes in the mesopelagic (Cyclothone spp.); however, these were only from a semi-enclosed bay with deep-water. Also, effort of reproductive studies was predominantly aimed at species that performed daily vertical migrations and those lower in the food web (i.e. secondary consumers/zooplanktivores). However, one recent study investigated 12 species of stomiiformes, the most comprehensive study of reproductive ecology in higher trophic level species (larger micronektivore/piscivore predators) in the mesopelagic (Marks et al., 2020). There are several other particularly common and/or cosmopolitan mesopelagic fishes that appear to lack detailed reproductive studies. These include but are not limited to: Hygophum spp. and Stomias spp. and members of Melamphaidae, Nemichthyidae, and Serrivomeridae. Methods of assessing reproductive parameters Direct observation of gonads is the most common method of determining sex and assessing parameters such as maturity, reproductive phenology, and reproductive output of a species. Knowing the ratio of sexes in a population, as well as at what size or age fish are reproductively mature are key components to understanding reproductive dynamics. Reproductive studies generally focus on females, partly because the number of eggs/larvae produced by a population is typically limited by egg rather than sperm production. Furthermore, nuanced seasonal and diel patterns used to identify reproductive and spawning strategies are more evident in females (Murua and Saborido-Rey, 2003; Lowerre-Barbieri et al., 2011a). Methods such as macroscopic evaluation of whole gonads and calculating the gonadosomatic index (the ratio of gonad to body mass) may be sufficient to determine sex and reproductive status in some cases (West, 1990). Microscopic methods, including examination of “whole mounts” of gonadal tissue, measurement of oocytes, and histological techniques, allow for detailed examination of tissue (West, 1990). Fewer than half of the studies of reproduction in mesopelagic species used histology (Table 2). While more expensive and time-consuming, histological methods are generally considered the most accurate method to assess maturity and spawning status of an individual. Histology can be particularly useful in determining sex of small, immature fish; identifying and determining the size/age at which sexual transition occurs in hermaphrodites; identifying phenomena such as skip spawning; and assessing maturity outside of the reproductive season (West, 1990; Lowerre-Barbieri et al., 2011a). Ultimately, a combination of techniques may be the most comprehensive approach to reproductive studies (Kjesbu et al., 2003; Vitale et al., 2006). The most common maturity metric reported for mesopelagic fishes was the observed minimum size or age of mature fish. Maturity ogives, which estimate the proportion of fish mature at given age and length classes and provide important metrics for stock assessments, were reported in few investigations (Table 2). Population-level spawning seasonality, which is often assessed by similar methods to that of maturity, was reported in the majority of investigations (Table 2). However, the more nuanced timing aspects of spawning interval and diel periodicity (the number of days between spawning events and time of day at which spawning occurs, respectively), depend on capturing and identifying females with spawning markers that can be short-lived and were reported in few studies (Lisovenko and Prut’ko, 1987b; Gartner, 1993; Stequert et al., 2003; Sassa et al., 2016). Other methods of determining seasonality and diel periodicity include egg and/or larval abundance surveys; however, mesopelagic eggs are delicate (Catul et al., 2011) and have only been described for a few species (Robertson, 1977; Gjøsæter and Tilseth, 1988; Landaeta and Castro, 2002). Fecundity is another key parameter of estimating reproductive potential of a species and was estimated in 33 investigations from 22 species (Table 2). Most of the investigations estimated fecundity by enumerating all oocytes or a group of oocytes (e.g. the largest mode) within the ovary, presumably due to the small nature of the fishes (Table 2). Maternal size and age have been shown to affect the quality and quantity of eggs in some taxa (see Hixon et al., 2014). However, only seven studies (representing nine species) presented estimates of weight-specific fecundity, an important metric to determine whether reproductive output is proportional to female size. Accurate estimates of reproductive output (e.g. total number of eggs produced) require knowledge of a species’ reproductive strategy. In particular, the fecundity pattern (determinate or indeterminate) and spawning strategy (batch or total spawning) dictate when samples should be collected and how estimates should be made (Hunter and Macewicz, 2003; Murua et al., 2003; Murua and Saborido-Rey, 2003). Among the studies reporting fecundity values, most, but not all, explicitly report which oocytes were counted (e.g. all or only those of a particular mode or size class) or what fecundity metric was reported (e.g. batch, annual, or lifetime). Furthermore, while spawning strategy is stated or could be inferred from the majority of studies, few addressed fecundity patterns directly—Lisovenko and Prut’ko (1987a) and García-Seoane et al. (2014) are the two with the most detailed descriptions. Patterns in reproductive strategies Fishes have evolved a complex assortment of reproductive strategies (Stearns, 1992) that include oviparity, viviparity, hermaphroditism, and parthenogenesis. However, much less is known about deep-living midwater fishes relative to coastal and epipelagic species. It is thought the large majority of mesopelagic fishes, similar to the majority of other marine fishes, are oviparous broadcast spawners with planktonic eggs and larvae (Blackburn, 1999). This r-selected strategy is well suited to a relatively low-density pelagic environment, where the strategy is suggested to have evolved as a means to disperse siblings in order to spread the risk of failure among members of a cohort (Doherty et al., 1985), and to utilize the productive and often patchy plankton in the surface waters as a food source (Hunter, 1980). There are, however, some exceptions, such as the unusual sexual parasitism exhibited in some ceratioid anglerfishes (Pietsch, 1975). While most of the investigations on mesopelagic fishes suggested species are gonochoristic, several gonostomatid species were shown to be protandrous hermaphrodites (Fisher, 1983; Miya and Nemoto, 1987a; Lancraft et al., 1988). This strategy can increase an individual’s fitness, particularly when there is a size-advantage such that reproductive potential increases with increasing female size (see Benvenuto et al., 2017). Sex ratios differed from parity in several investigations of mesopelagic fish populations. Often sex ratios were skewed seasonally (e.g. Kawaguchi and Mauchline, 1982) or within particular size ranges (e.g. Gartner, 1993), typically favouring males at smaller sizes and females at larger sizes. The sex ratio of a population can be innately tied to growth: sexual differences in growth rate and lifespan can lead to more of one sex, as is often the case for females that grow fastest and live longest (e.g. Clarke, 1983; Linkowski et al., 1993; Greely et al., 1999). The greater biomass of females is generally considered an adaptation to maximize the egg-producing biomass (Clarke, 1983). Sampling design is also a consideration, as the observed size classes and sex ratios may differ from the true values of those of the population due to unequal sampling of depths or through size- and sex-related net avoidance or escapement (Young et al., 1987; Gartner, 1993). Using different gear types that ensure sampling of the entire size range and sampling all depths throughout day and night can mitigate this. Trends in size and maturity of mesopelagic fishes showed a continuum from short-lived, small species that matured early to longer-lived fishes that grew larger and matured at later ages or larger sizes. For example, Vinciguerria nimbaria (Stequert et al., 2003) were estimated to reach sizes <5 cm standard length (SL), live <7 months, and mature half-way through their lives, spawning continuously until dying. In contrast, Poromitra crassiceps (Childress et al., 1980) reached sizes up to 14 cm SL, had longevities up to 9 years, and reached sexual maturity in the last 1–2 years of their lives. While larger females often contribute disproportionately in terms of quality and quantity of offspring in a variety of fishes (Green, 2008; Hixon et al., 2014), the adult mortality rate for mesopelagic fishes in general may be high, favouring r-selected life histories. It remains to be seen whether this strategy is representative of mesopelagic fishes generally or is a result of the overrepresentation of myctophids. The reproductive potential of a species depends not only on when sexual maturity is attained, but also on the number of reproductive events participated in, which represents a continuum ranging from one (semelparity) to many (iteroparity). Semelparity was rare in mesopelagic species but was reported in several gonostomatids and sternoptychids (Miya and Nemoto, 1986; Lancraft et al., 1988; Miya and Nemoto, 1991). Within a spawning season, most mesopelagic species were reportedly batch spawners (27 of 33 investigations). Interestingly, approximately half of those species could be considered semelparous on an annual scale, meaning an individual only participated in a single spawning season but reproduced multiple times within that season (Lowerre-Barbieri et al., 2011b). This strategy was common among myctophids (e.g. B. suborbitale, Gartner, 1993). Iteroparous total spawners included gonostomatids and stomiids (Fisher and Pearcy, 1983; Miya and Nemoto, 1986, 1987a, b, 1991). Batch fecundity was the most commonly reported fecundity metric among the investigations of mesopelagic fishes. In addition to being a bet-hedging strategy (Lambert and Ware, 1984), batch spawning may also be necessitated by physiological constraints associated with small body size (Wootton, 1992). In small fishes such as myctophids, batch spawning might allow them to increase their reproductive output compared to total spawning. For example, B. suborbitale was estimated to spawn 84 times, resulting in an estimated lifetime fecundity of nearly 20 000 eggs (Gartner, 1993). On the contrary, the larger, semelparous gonostomatid, S. elongatus, was estimated to spawn ∼22 000 eggs in a single reproductive event (Fisher, 1980). In general, reported fecundity values of mesopelagic fishes were low compared to epipelagic species. Batch fecundity estimates ranged from hundreds of eggs per batch in Maurolicus muelleri and Benthosema pterotum (Hussain and Ali-Khan, 1987; Sassa et al., 2014) to upwards of 12 000 in Ceratoscopelus spp. and Diaphus spp. (Gartner, 1993; Sassa et al., 2014; Sassa et al., 2016). Annual fecundity estimates were as low as 1500 in Cyclothone atraria and as high as 66 000 to 125 000 in T. macropus and Diaphus suborbitalis, respectively (Fisher and Pearcy, 1983; Lisovenko and Prut’ko, 1987b; Miya and Nemoto, 1987a). Lifetime fecundity estimates of <1000 were reported in some semelparous species (Kawaguchi and Mauchline, 1982; Miya and Nemoto, 1986). In contrast to non-mesopelagic species, both semelparous and iteroparous populations of American shad, Alossa sapidissima had batch fecundities ranging from 6000 to nearly 100 000 (Olney and McBride, 2003). Furthermore, species such as the epipelagic clupeiform Engraulis mordax reportedly spawn 7400 to 10 900 eggs as many as 20 times in a reproductive season: thus, some individuals may spawn hundreds of thousands of eggs annually and as many as 1 million eggs in a lifetime (Hunter and Leong, 1981). The number of offspring an individual is able to produce is intrinsically related to energetic reserves or availability (Rideout and Morgan, 2010); it is possible that energetic costs related to vertical migration or low productivity in the environment limit the reproductive potential of mesopelagic fishes. However, the myctophid D. suborbitalis was estimated to release 400–4600 eggs per batch in at least 8–24 (size-dependent), and as many as 70 spawning events (Lisovenko and Prut’ko, 1987b), suggesting some mesopelagic fishes may have an annual or lifetime reproductive output comparable to shallow-water species. Spawning season durations among mesopelagic fishes were variable, with some having narrow, well-defined temporal peaks and others exhibiting protracted or year-round spawning. Timing strategies within groups varied as well. For instance, among myctophids, B. pterotum in the East China Sea showed a clear spawning peak of August to September for females (Sassa et al., 2014); B. suborbitale exhibited a protracted spawning up to 6 months in duration in the Gulf of Mexico (Gartner, 1993); and D. suborbitalis spawned year round in the Indian Ocean (Lisovenko and Prut’ko, 1987a). These taxonomically related species were assessed in different locations, where ocean conditions and day lengths differ. Therefore, the results highlight the regional differences in the life history traits of mesopelagic species, even in those closely related, underscoring the need for inter-regional studies. Knowledge of mesopelagic fish spawning behaviour, including spawning depth, timing, and schooling behaviour is very limited (Brodeur and Yamamura, 2005). For vertically migrating species, a common strategy is to release eggs during the night near the surface, indicated by the collection of females with oocytes at the germinal vesicle breakdown or hydrated stage (Gjøsæter and Tilseth, 1988; Gartner, 1993; Flynn and Paxton, 2012). However, some Diaphus spp. likely spawn during the day at mesopelagic depths (Sassa et al., 2016). Furthermore, DVM behaviour can be altered to suit oceanographic conditions related to reproduction. An example is V. nimbaria, which is normally a daily migrator but stays in surface waters for several days when waters are strongly stratified to maximize food reserves during gonad development (Stequert et al., 2003). This atypical behaviour linked to hydrographic conditions makes this population more vulnerable to daytime predation in the sunlit waters, illustrating a trade-off between feeding conditions that favour enhanced egg quality and quantity and an increased predation risk. Schooling in mesopelagic fishes is also a risky behaviour, and very little is known about schooling in these fishes. However, there is some evidence that migrating species, like myctophids, aggregate near the surface (Flynn and Paxton, 2012) or migrate horizontally (Sassa, 2019) to spawn. These examples highlight the need to understand the behaviour of mesopelagic fishes, particularly through high-resolution sampling efforts at varying depths over a 24-h period. Latitudinal variation in reproductive traits is well documented in shallower water species (McBride et al., 2013; Lefebvre et al., 2019); however, it remains largely unknown how or if maturity metrics and fecundity differ geographically among mesopelagic fishes. Geographic variability in reproductive traits of mesopelagic fishes is difficult to address due to the paucity of research. Despite the broad geographic ranges of many mesopelagic species, there are few examples of multiple reproductive studies for the same species in different regions. Furthermore, maturity metrics and fecundity values are not directly comparable in the few species for which studies have been conducted in multiple regions due to the differences in the methods applied (e.g. each of the three studies in Table 2 reporting maturity for B. pterotum used different methods to define maturity). However, in a single study, García-Seoane et al. (2014) reported differences in the length at 50% maturity for the Flemish Cap and Baeleric Sea populations of B. glaciale, suggesting geographic variability in biological processes is an important consideration for mesopelagic fishes. Obtaining a thorough understanding of the reproductive ecology of any species requires considerable effort, though researching species with broad geographic ranges in the deep ocean and far from land at regular schedules presents logistical and economic difficulties (Prellezo, 2019). Moreover, sampling throughout the year, across the geographic range of a population, and using methods that capture the entirety of the size/age range of a population are all necessary. As a corollary, studies that examine and incorporate data on both growth and reproduction are very informative; for instance, quantitative estimates of the lifetime oocyte production can be calculated for those populations where validated growth data is available (e.g. B. suborbitale and Lepidophanes guentheri; Gartner, 1991a, b, 1993). Several detailed studies on mesopelagic fishes addressed the basic reproductive and spawning characteristics necessary to evaluate population reproductive dynamics (Hussain and Ali-Khan, 1987; Lisovenko and Prut’ko, 1987a, b; Dalpadado, 1988; Gartner, 1993; de Almeida and Rossi-Wongtschowski, 2007; Garcia-Seoane et al., 2015; Marks et al., 2020). Many other studies provide the basis for more detailed research, as gaining a complete understanding of the reproductive ecology of a species is often difficult to accomplish in a single study due to some aspects requiring knowledge of others. All the studies provide valuable insights for a region of the ocean greatly lacking research. Anthropogenic impacts: why more life history research is needed Mesopelagic ecosystems are increasingly affected by anthropogenic activities. One of the most prescient issues is resource extraction. To date, harvesting in the mesopelagic has mostly been exploratory or financially unsuccessful, especially when specifically regarding fishes (Figure 3). These commercial and exploratory fisheries have focused primarily on myctophids and phosichthyids. However, the depletion of many coastal and demersal fisheries due to the demand to feed the rapidly growing human population may precipitate a marked shift to harvesting the mesopelagic zone. Humans will be unlikely to consume these harvests directly, and the small oily and/or watery mesopelagic animals are expected to provide fish meal or fish oil for agriculture and aquaculture, in addition to nutraceuticals (Institute of Marine Research 2017). There are high levels of uncertainty in current biomass estimates (Proud et al., 2018), and over-harvesting has occurred when industrial fleets target areas beyond national jurisdiction (Cullis-Suzuki and Pauly, 2010; FAO, 2014). Therefore, the need for vital biological data—such as outlined in this review—to inform and help regulate these relatively intact resources in predominantly jurisdiction-free waters is paramount. Figure 3. Open in new tabDownload slide Global areas of life history research on mesopelagic fishes (green areas), and locations of commercial and exploratory (*) mesopelagic fish harvesting (red markers), with country and years of harvest in parentheses. Figure 3. Open in new tabDownload slide Global areas of life history research on mesopelagic fishes (green areas), and locations of commercial and exploratory (*) mesopelagic fish harvesting (red markers), with country and years of harvest in parentheses. The second central anthropogenic impact is that of climate change. Warming temperatures will increase stratification in the open ocean (Coma et al., 2009), and acidification will result in reductions in the ability of plankton to produce calcified structures (Hofmann et al., 2010), both having impacts on pelagic ecology. Furthermore, midwater ecosystems help control the rate at which the ocean can uptake atmospheric carbon dioxide, and DVM in particular accelerates the sequestration process to the deep ocean (Siegel et al., 2014; Buesseler et al., 2020). However, the exact scale at which these carbon and nutrient cycles operate via the mesopelagic is poorly understood (St. John et al., 2016). With increasing oceanic warming and acidification, it is uncertain how much the ecosystem services that the mesopelagic zone provides (e.g. the biological pump) will be disturbed. There is a multitude of other anthropogenic effects that may impact mesopelagic ecosystems. These include plastic pollution in the ocean, in particular microplastics, which can easily be ingested and passed through the food chain (Lusher et al., 2016; Wieczorek et al., 2018; Choy et al., 2019). Many pelagic animals across varying taxa are well adapted to filtering out small particles in the water column, and the DVM that is so effective at transporting energy is precisely suited to also transporting microplastics from the surface into deeper habitats. In fact, mesopelagic fishes have been shown to ingest more plastic than surface-dwelling fishes (Choy and Drazen, 2013). The oxygen content of the ocean has been declining for the past half century due to global warming and nutrient discharge (Breitburg et al., 2018). Some taxa are more suited to hypoxic conditions; for example, the ctenophore Mnemiopsis leidyi is more tolerant of low oxygen than trophically equivalent fishes (Kolesar et al., 2017). Therefore, regions of the ocean experiencing deoxygenation could lead to few species outcompeting the many, resulting in biodiversity loss. Additionally, there is a particular interest currently in mining deep-sea minerals. The spatial scale of these activities may be very large—for instance, the area claimed for mining in the Clarion-Clipperton Fracture Zone is around 7–8 million km2 (Wedding et al., 2015). As mining of this nature is still in development, the effects on midwater ecosystems for instance from sediment plumes, noise, and spills are impossible to predict at this point in time; however, the environmental risks should be evaluated (Halfar and Fujita, 2007; Drazen et al., 2020). The effects of each of these impacts on pelagic ecosystems as a whole and the life history of mesopelagic fishes in particular are only starting to be understood, but warrant in depth investigation moving forward (Drazen et al., 2020). Concluding remarks Mesopelagic ecosystems are understudied globally, especially fish life history research. For the data that are available, there are several knowledge gaps. Collectively, two-thirds of life history research on mesopelagic fishes has been undertaken on myctophids, and therefore there is a dearth of knowledge on most other families, including several of great importance. Furthermore, as DVM is so influential to the movement of energy in the oceans, a more robust understanding of each species’ migration behaviour is essential. A life history focus on those that do not migrate is also lacking. There are generally global hotspots of mesopelagic fish research, which are mostly centred in the northern hemisphere. Regions with little to no coverage include the South Atlantic, large parts of Indo-Pacific region and some polar environments. Thus, a focus on a wider suite of fish taxa, a more comprehensive understanding of DVM, and studying regions of the world that have received little attention would add greatly to our ecological understanding of the mesopelagic. Subsequently, there is still much that remains unknown about mesopelagic fishes’ role in global ocean ecosystems. Such an understanding requires a holistic assessment and understanding of the ecosystem and its inhabitants. In this regard, some of the foremost characteristics are information on population vital rates—knowledge of growth and reproduction. This review synthesizes the current state of knowledge on these key traits, highlighting the need for greater focus in this area. Data availability There are no new data associated with this article. Funding This work was part of the Woods Hole Oceanographic Institution’s Ocean Twilight Zone Project, funded as part of the Audacious Project housed at TED. Acknowledgements We thank Porter Hoagland for useful discussions and valuable input into Figure 3, and we thank the reviewers for their valuable input into this manuscript. References Badcock J. , Araujo T. M. H. 1988 . On the significance of variation in a warm water cosmopolitan species, nominally Ceratoscopelus warrningii (Pisces, Myctophidae) . Bulletin of Marine Science , 42 : 16 – 43 . Google Scholar OpenURL Placeholder Text WorldCat Baird R. C. , Hopkins T. L. 1981 . Trophodynamics of the fish Valenciennellus tripunctulatus. III. Energetics, resources and feeding strategy . Marine Ecology Progress Series , 5 : 21 – 28 . Google Scholar Crossref Search ADS WorldCat Benvenuto C. , Coscia I. , Chopelet J. , Sala-Bozano M. , Mariani S. 2017 . Ecological and evolutionary consequences of alternative sex- change pathways in fish . Scientific Reports , 7 : 9084 . Google Scholar Crossref Search ADS PubMed WorldCat Blackburn D. G. 1999 . Viviparity and oviparity: evolution and reproductive strategies . Encyclopedia of Reproduction , 4 : 994 – 1003 . Google Scholar OpenURL Placeholder Text WorldCat Breitburg D. , Levin L. A. , Oschlies A. , Grégoire M. , Chavez F. P. , Conley D. J. , Garçon V. et al. 2018 . Declining oxygen in the global ocean and coastal waters . Science , 359 : eaam7240 . Google Scholar Crossref Search ADS PubMed WorldCat Brodeur R. D. , Yamamura O. 2005 . Micronekton of the North Pacific . PICES Scientific Report , 30 : 1 – 115 . Google Scholar OpenURL Placeholder Text WorldCat Brothers E. B. 1979 . Age and growth studies on tropical fishes. In Stock Assesment for Tropical Small-Scale Fisheries , pp. 119–136. Ed. by Saila S. B. , Roedel P. M. . University of Rhode Island , Kingston, RI . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Buesseler K. O. , Boyd P. W. , Black E. E. , Siegel D. A. 2020 . Metrics that matter for assessing the ocean biological carbon pump . Proceedings of the National Academy of Sciences of the United States of America , 117 : 9679 – 9687 . Google Scholar Crossref Search ADS PubMed WorldCat Bystydzienska Z. E. , Phillips A. J. , Linkowski T. B. 2010 . Larval stage duration, age and growth of blue lanternfish Tarletonbeania crenularis (Jordan and Gilbert, 1880) derived from otolith microstructure . Environmental Biology of Fishes , 89 : 493 – 503 . Google Scholar Crossref Search ADS WorldCat Campana S. E. 2001 . Accuracy, precision and quality control in age determination, including a review of the use and abuse of age validation methods . Journal of Fish Biology , 59 : 197 – 242 . Google Scholar Crossref Search ADS WorldCat Campana S. E. , Neilson J. D. 1985 . Microstructure of fish otoliths . Canadian Journal of Fisheries and Aquatic Sciences , 42 : 1014 – 1032 . Google Scholar Crossref Search ADS WorldCat Catul V. , Gauns M. , Karuppasamy P. K. 2011 . A review on mesopelagic fishes belonging to family Myctophidae . Reviews in Fish Biology and Fisheries , 21 : 339 – 354 . Google Scholar Crossref Search ADS WorldCat Childress J. J. , Taylor S. M. , Cailliet G. M. , Price M. H. 1980 . Patterns of growth, energy-utilization and reproduction in some mesopelagic and bathypelagic fishes off southern California . Marine Biology , 61 : 27 – 40 . Google Scholar Crossref Search ADS WorldCat Choy C. A. , Drazen J. C. 2013 . Plastic for dinner? Observations of frequent debris ingestion by pelagic predatory fishes from the central North Pacific . Marine Ecology Progress Series , 485 : 155 – 163 . Google Scholar Crossref Search ADS WorldCat Choy C. A. , Portner E. , Iwane M. , Drazen J. C. 2013 . Diets of five important predatory mesopelagic fishes of the central North Pacific . Marine Ecology Progress Series , 492 : 169 – 184 . Google Scholar Crossref Search ADS WorldCat Choy C. A. , Robison B. H. , Gagne T. O. , Erwin B. E. , Firl E. , Halden R. U. , Hamilton J. A. et al. 2019 . The vertical distribution and biological transport of marine microplastics across the epipelagic and mesopelagic water column . Scientific Reports , 9 : 7843 . Google Scholar Crossref Search ADS PubMed WorldCat Clarke T. A. 1973 . Some aspects of the ecology of laternfishes (Myctophidae) in the Pacific Ocean near Hawaii . Fisheries Bulletin , 71 : 401 – 434 . Google Scholar OpenURL Placeholder Text WorldCat Clarke T. A. 1983 . Sex-ratios and sexual differences in size among mesopelagic fishes from the central Pacific ocean . Marine Biology , 73 : 203 – 209 . Google Scholar Crossref Search ADS WorldCat Coma R. , Ribes M. , Serrano E. , Jimenez E. , Salat J. , Pascual J. 2009 . Global warming-enhanced stratification and mass mortality events in the Mediterranean . Proceedings of the National Academy of Sciences of the United States of America , 106 : 6176 – 6181 . Google Scholar Crossref Search ADS PubMed WorldCat Conley W. J. , Gartner J. V. Jr. 2009 . Growth among larvae of lanternfishes (teleostei: myctophidae) from the eastern Gulf of Mexico . Bulletin of Marine Science , 84 : 123 – 135 . Google Scholar OpenURL Placeholder Text 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 Dalpadado P. , and Gjøsaeter J. 1987 . Observations on mesopelagic fish from the Red Sea . Marine Biology , 96 : 173 – 183 . Google Scholar Crossref Search ADS WorldCat Dalpadado P. 1988 . Reproductive biology of the lanternfish Benthosema pterotum from the Indian Ocean . Marine Biology , 98 : 307 – 316 . Google Scholar Crossref Search ADS WorldCat Davison P. C. , Checkley D. M. J. , Koslow J. A. , Barlow J. 2013 . Carbon export mediated by mesopelagic fishes in the northeast Pacific Ocean . Progress in Oceanography , 116 : 14 – 30 . Google Scholar Crossref Search ADS WorldCat Davison P. C. , Lara-Lopez A. , Koslow J. A. 2015 . Mesopelagic fish biomass in the southern California current ecosystem . Deep-Sea Research II , 112 : 129 – 142 . Google Scholar Crossref Search ADS WorldCat de Almeida E. M. , Rossi-Wongtschowski CLdB. 2007 . Maurolicus stehmanni (Sternoptychidae): length of first maturation, and spawning seasons in the south-southeast Brazilian region . Brazilian Journal of Oceanography , 55 : 309 – 322 . Google Scholar Crossref Search ADS WorldCat Demestre M. , Moli B. , Recasens L. , Sanchez P. 1993 . Life history and fishery of Lepidopus caudatus (Pisces: Trichiuridae) in the Catalan Sea (Northwestern Mediterranean) . Marine Biology , 115 : 23 – 32 . Google Scholar Crossref Search ADS WorldCat Doherty P. J. , Williams B. M. , Sale P. F. 1985 . The adaptive significance of larval dispersal in coral reef fishes . Environmental Biology of Fishes , 12 : 81 – 90 . Google Scholar Crossref Search ADS WorldCat Drazen J. C. , Smith C. R. , Gjerde K. M. , Haddock S. H. D. , Carter G. S. , Choy C. A. , Clark M. R. et al. 2020 . Midwater ecosystems must be considered when evaluating environmental risks of deep-sea mining . Proceedings of the National Academy of Sciences of the United States of America , 117 : 17455 – 17460 . Google Scholar Crossref Search ADS PubMed WorldCat Fanelli E. , Papiol V. , Cartes J. E. , and Rodriguez-Romeu O. 2014 . Trophic ecology of Lampanyctus crocodilus on north-west Mediterranean Sea slopes in relation to reproductive cycle and environmental variables . Journal of Fish Biology , 84 : 1654 – 1688 . Google Scholar Crossref Search ADS PubMed WorldCat FAO . 2014 . United Nations Food and Agriculture Organization (FAO), The State of World Fisheries and Aquaculture 2014. http://www.fao.org/fishery/sofia/en (last accessed 1 March 2020). Fishelson L. 1994 . Comparative internal morphology of deep-sea eels, with particular emphasis on gonads and gut structure . Journal of Fish Biology , 44 : 75 – 101 . Google Scholar Crossref Search ADS WorldCat Fisher J. P. , Pearcy W. G. 1983 . Reproduction, growth and feeding of the mesopelagic fish Tactostoma macropus (Melanostomiatidae) . Marine Biology , 74 : 257 – 267 . Google Scholar Crossref Search ADS WorldCat Fisher R. A. 1980 . Aspects of reproduction in thc midwater fish families Gonostomatidae and Sternoptychidae. MSc thesis , University of South Florida , St. Petersburg, FL . 93 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Fisher R. A. 1983 . Protrandric sex reversal in Gonostoma elongatum (Pisces: Gonostomatidae) from the eastern Gulf of Mexico . Copeia , 1983 : 554 – 557 . Google Scholar Crossref Search ADS WorldCat Flynn A. J. , Paxton J. R. 2012 . Spawning aggregation of the lanternfish Diaphus danae (family Myctophidae) in the north-western Coral Sea and associations with tuna aggregations . Marine & Freshwater Research , 63 : 1255 – 1271 . Google Scholar Crossref Search ADS WorldCat Folkvord A. , Gundersen G. , Albretsen J. , Asplin L. , Kaartvedt S. , Giske J. 2016 . Impact of hatch date on early life growth and survival of Mueller's pearlside (Maurolicus muelleri) larvae and life-history consequences . Canadian Journal of Fisheries and Aquatic Sciences , 73 : 163 – 176 . Google Scholar Crossref Search ADS WorldCat Forsgren K. L. , Jamal H. , Barrios A. , Paig-Tran E. W. M. 2017 . Reproductive morphology of oarfish (Regalecus russellii) . Anatomical Record - Advances in Integrative Anatomy and Evolutionary Biology , 300 : 1695 – 1704 . Google Scholar Crossref Search ADS WorldCat Fricke R. , Eschmeyer W.N. , Van der Laan R. (Eds). 2020 . Eschmeyer's catalog of fishes: Genera, species, references. http://researcharchive.calacademy.org/research/ichthyology/catalog/fishcatmain.asp (last accessed 20 August 2020). García-Seoane E. , Bernal A. , Saborido-Rey F. 2014 . Reproductive ecology of the glacier lanternfish Benthosema glaciale . Hydrobiologia , 727 : 137 – 149 . Google Scholar Crossref Search ADS WorldCat Garcia-Seoane E. , Fabeiro M. , Silva A. , Meneses I. 2015 . Age-based demography of the glacier lanternfish (Benthosema glaciale) in the Flemish Cap . Marine and Freshwater Research , 66 : 78 – 85 . Google Scholar Crossref Search ADS WorldCat Gartner J. V. 1991a . 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 Gartner J. V. 1991b . Life histories of 3 species of lanternfishes (Pisces, Myctophidae) from the eastern Gulf of Mexico: 2. Age and growth-patterns . Marine Biology , 111 : 21 – 27 . Google Scholar Crossref Search ADS WorldCat Gartner J. V. 1993 . Patterns of reproduction in the dominant lanternfish species (Pisces, Myctophidae) of the eastern Gulf of Mexico, with a review of reproduction among tropical-subtropical Myctophidae . Bulletin of Marine Science , 52 : 721 – 750 . Google Scholar OpenURL Placeholder Text WorldCat Gartner J. V. J. , Hopkins T. L. , Baird R. C. , Milliken D. M. 1987 . The lanternfishes (Pisces: Myctophidae) of the eastern Gulf of Mexico . Fisheries Bulletin , 85 : 81 – 98 . Google Scholar OpenURL Placeholder Text WorldCat Geffen A. J. 1992 . Validation of otolith increment deposition rate. In Otolith Microstructure Examination and Analysis, p. 117 . Ed. by Stevenson D. K. , Campana S. E. . Canadian Special Publication on Fisheries and Aquatic Species . 192 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Giragosov V. , Ovcharov O. P. 1992 . Age and growth of the lantern fish Myctophum nitidulum (Myctophidae) from the tropical Atlantic . Journal of Ichthyology , 32 : 34 – 42 . Google Scholar OpenURL Placeholder Text WorldCat Gjøsæter H. 1981a . Life history and ecology of Maurolicus muelleri (Gonostomatidae) in Norwegian waters . Fiskeridirektoratets Skrifter Serie Havundersøkelser , 17 : 109 – 131 . Google Scholar OpenURL Placeholder Text WorldCat Gjøsæter H. 1987 . Primary growth increments in otoliths of six tropical myctophid species . Biological Oceanography , 4 : 359 – 382 . Google Scholar OpenURL Placeholder Text WorldCat Gjøsæter J. 1973 . Age, growth, and mortality of the myctophid fish, Benthosema glaciale (Reinhardt), from western Norway . Sarsia , 52 : 1 – 14 . Google Scholar Crossref Search ADS WorldCat Gjøsæter J. 1981b . Life history and ecology of the myctophid fish Notoscopelus elongatus kroeyeri from the northeast Atlantic . Fiskdir Skr Ser Havunders , 17 : 133 – 152 . Google Scholar OpenURL Placeholder Text WorldCat Gjøsæter J. 1984 . Mesopelagic fish, a large potential resource in the Arabian Sea . Deep-Sea Research Part A—Oceanographic Research Papers , 31 : 1019 – 1035 . Google Scholar Crossref Search ADS WorldCat Gjøsæter J. , Kawaguchi K. 1980 . A Review of the World Resources of Mesopelagic Fish. 193, F.T.P. UN, FAO, Rome. Gjøsæter J. , Tilseth S. 1988 . Spawning behavior, egg and larval development of the myctophid fish Benthosema pterotum . Marine Biology , 98 : 1 – 6 . Google Scholar Crossref Search ADS WorldCat Go Y. B. , Kawaguchi K. , Kusaka T. 1977a . Ecologic study on Diaphus suborbitalis (Pisces, Myctophidae) in Suruga Bay, Japan. I. Method of aging and its lifespan . Bulletin of the Japanese Society of Scientific Fisheries , 43 : 913 – 919 . Google Scholar Crossref Search ADS WorldCat Go Y. B. , Kawaguchi K. , Kusaka T. 1977b . Ecologic study on Diaphus suborbitalis (Pisces, Myctophidae) in Suruga Bay, Japan. II. Growth Pattern . Bulletin of the Japanese Society of Scientific Fisheries , 43 : 1411 – 1416 . Google Scholar Crossref Search ADS WorldCat Goodson M. S. , Giske J. , and Rosland R. 1995 . Growth and ovarian development of Maurolicus muelleri during spring . Marine Biology , 124 : 185 – 195 . Google Scholar Crossref Search ADS WorldCat Greely T. M. , Gartner J. V. , Torres J. J. 1999 . Age and growth of Electrona antarctica (Pisces: Myctophidae), the dominant mesopelagic fish of the Southern Ocean . Marine Biology , 133 : 145 – 158 . Google Scholar Crossref Search ADS WorldCat Green B. S. 2008 . Maternal effects in fish populations. In Advances in Marine Biology . Ed. by Sims D. W. . Elsevier Academic Press . 54 : 1 – 105 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Halfar J. , Fujita R. M. 2007 . Danger of deep-sea mining . Science , 316 : 987 – 987 . Google Scholar Crossref Search ADS PubMed WorldCat Halliday R. G. 1970 . Growth and vertical distribution of the glacier lanternfish, Benthosema glaciale, in the northwestern Atlantic . Journal of Fisheries Research Board of Canada , 27 : 105 – 116 . Google Scholar Crossref Search ADS WorldCat Hayashi A. , Kawaguchi K. , Watanabe H. , Ishida M. 2001 . Daily growth increment formation and its lunar periodicity in otoliths of the myctophid fish Myctophum asperum (Pisces: Myctophidae) . Fisheries Science , 67 : 811 – 817 . Google Scholar Crossref Search ADS WorldCat 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 Hofmann G. E. , Barry J. P. , Edmunds P. J. , Gates R. D. , Hutchins D. A. , Klinger T. , Sewell M. A. 2010 . The effect of ocean acidification on calcifying organisms in marine ecosystems: an organismto-ecosystem perspective . Annual Review of Ecology, Evolution, and Systematics , 41 : 127 – 147 . Google Scholar Crossref Search ADS WorldCat Hopkins T. L. , Sutton T. T. 1998 . Midwater fishes and shrimps as competitors and resource partitioning in low latitude oligotrophic ecosystems . Marine Ecology Progress Series , 164 : 37 – 45 . Google Scholar Crossref Search ADS WorldCat Howell W. H. , Krueger W. H. 1987 . Family sternoptychidae, marine hatchetfishes and related species. In Biology of Midwater Fishes of the Bermuda Ocean Acre. Ed. by Gibbs R. H. J. , Krueger W. H. . Smithsonian Contributions to Zoology . 452 : 32 – 50 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hunter J. R. 1980 . The feeding behavior and ecology of marine fish larvae. In Fish Behavior and Its Use in the Capture and Culture of Fishes. ICLARM Conference Proceedings, pp. 287 – 330 . Ed. by Bardach J. E. , Magnuson J. J. , May R. C. , Reinhart J. M. . International Centre for Living Aquatic Resources Management , Manila, Phillipines . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hunter J. R. , Leong R. 1981 . The spawning energetics of female northern anchovy, Engraulis mordax . Fishery Bulletin , 79 : 215 – 229 . Google Scholar OpenURL Placeholder Text WorldCat Hunter J. R. , Macewicz B. J. 2003 . Improving the accuracy and precision of reproductive information used in fisheries. In Report of the Working Group on Modern Approaches to Assess Maturity and Fecundity of Warm- and Cold-Water Fish and Squids, pp. 57 – 68 . Ed. by Kjesbu O. S. , Hunter J. R. , Witthames P. R. . Institute of Marine Research , Bergen, Norway . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hussain S. M. 1992 . The reproductive biology of the lantern fish Benthosema fibulatum from the northern Arabian Sea . Fisheries Research , 13 : 381 – 393 . Google Scholar Crossref Search ADS WorldCat Hussain S. M. , Ali-Khan J. 1987 . Fecundity of Benthosema fibulatum and Benthosema pterotum from the northern Arabian Sea . Deep-Sea Research , 34 : 1293 – 1299 . Google Scholar Crossref Search ADS WorldCat Ikeda T. 1994 . Growth and life cycle of the mesopelagic fish Maurolicus muelleri (Sternoptychidae) in Toyama Bay, southern Japan Sea . Bulletin of Plankton Society of Japan , 40 : 127 – 138 . Google Scholar OpenURL Placeholder Text WorldCat Institute of Marine Research (2017). Mesopelagic Initiative: Unleashing new marine resources for a growing human population. Norwegian Institute of Food Fisheries and Aquaculture Research, University of Bergen and NIFES . https://www.hi.no/filarkiv/2017/rad-bestander_og_ressurser-mesopelagic_initiative-unleashing_ new_marine_resources_for_a_growing_human_population.pdf/nb-no. (last accessed 9 July 2019). Irigoien X. , Klevjer T. A. , Røstad A. , Martinez U. , Boyra G. , Acuña J. L. , Bode A. , et al. 2014 . Large mesopelagic fishes biomass and trophic efficiency in the open ocean . Nature Communications , 5 : 3271 – 3280 . Google Scholar Crossref Search ADS PubMed WorldCat Kaartvedt S. , Staby A. , Aksnes D. L. 2012 . Efficient trawl avoidance by mesopelagic fishes causes large underestimation of their biomass . Marine Ecology Progress Series , 456 : 1 – 6 . Google Scholar Crossref Search ADS WorldCat Karnella C. 1987 . Family Myctophidae, lanternfishes . In Biology of Midwater Fishes of the Bermuda Ocean Acre, pp. 51 – 168 . Ed. by Gibbs R. H. J. , Krueger W. H. . Smithsonian Institution Press , Washington, DC . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kawaguchi K. , Mauchline J. 1982 . Biology of myctophid fishes (Family Myctophidae) in the Rockall Trough, Northeastern Atlantic Ocean . Biological Oceanography , 1 : 337 – 373 . Google Scholar OpenURL Placeholder Text WorldCat Keene M. J. , Gibbs R. H. J. , Krueger W. H. 1987 . Family Melamphaidae, big scales . Smithsonian Contributions to Zoology , 452 : 169 – 185 . Google Scholar OpenURL Placeholder Text WorldCat Kjesbu O. S. , Hunter J. R. , Witthames P. R. 2003 . Report of the Working Group on Modern Methods to Assess Maturiy and Fecundity in Warm- and Cold-Water Fish and Squids . Institute of Marine Science , Bergen . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kolesar S. E. , Rose K. A. , Breitburg D. L. 2017 . Hypoxia Effects within an Intra-Guild Predation Food Web of Mnemiopsis leidyi Ctenophores, Larval Fish, and Copepods , pp. 279 – 317 . In Modeling Coastal Hypoxia. Ed. by Justic D. , Rose K. , Hetland R. , Fennel K. . Springer , Cham . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Konstantinova M. P. 1987 . Growth and natural mortality rates of three species of myctophids from the Southern Atlantic. All-Union Conference: Resources of the Southern Ocean and Problems of Their Rational Expolitation. USSR Ministry of Fisheries , Kerch , USSR. pp. 117 – 118 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Koslow J. A. 1996 . Energetic and life-history patterns of deep-sea benthic, benthopelagic and seamount-associated fish . Journal of Fish Biology , 49 : 54 – 74 . Google Scholar Crossref Search ADS WorldCat Kozłowski J. 1992 . Optimal allocation of resources to growth and reproduction: implications for age and size at maturity . Trends in Ecology & Evolution , 7 : 15 – 19 . Google Scholar Crossref Search ADS PubMed WorldCat Kristoffersen J. B. , Salvanes A. G. V. 2009 . Distribution, growth, and population genetics of the glacier lanternfish (Benthosema glaciale) in Norwegian waters: contrasting patterns in fjords and the ocean . Marine Biology Research , 5 : 596 – 604 . Google Scholar Crossref Search ADS WorldCat Lambert T. C. , and Ware D. M. 1984 . Reproductive strategies of demersal and pelagic spawning fish . Canadian Journal of Fisheries and Aquatic Sciences , 41 : 1565 – 1569 . Google Scholar Crossref Search ADS WorldCat Lancraft T. M. , Hopkins T. L. , Torres J. J. 1988 . Aspects of the ecology of the mesopelagic fish Gonostoma elongatum (Gonostomatidae, stomiiformes) in the eastern Gulf of Mexico . Marine Ecology Progress Series , 49 : 27 – 40 . Google Scholar Crossref Search ADS WorldCat Landaeta M. F. , Bustos C. A. , Contreras J. E. , Salas-Berrios F. , Palacios-Fuentes P. , Alvarado-Nino M. , Letelier J. et al. 2015a . Larval fish feeding ecology, growth and mortality from two basins with contrasting environmental conditions of an inner sea of northern Patagonia, Chile . Marine Environmental Research , 106 : 19 – 29 . Google Scholar Crossref Search ADS WorldCat Landaeta M. F. , Castro L. R. 2002 . Spring spawning and early nursery zone of the mesopelagic fish Maurolicus parvipinnis at the coastal upwelling zone off Talcahuano, central Chile . Marine Ecology Progress Series , 226 : 179 – 191 . Google Scholar Crossref Search ADS WorldCat Landaeta M. F. , Contreras J. E. , Bustos C. A. , Munoz G. 2015b . Larval growth of two species of lanternfish at nearshore waters from an upwelling zone based on otolith microstructure analyses . Journal of Applied Ichthyology , 31 : 106 – 113 . Google Scholar Crossref Search ADS WorldCat Lefebvre L. S. , Friedlander C. L. , Field J. C. 2019 . Reproductive ecology and size-dependent fecundity in the petrale sole (Eopsetta jordani) in waters of California, Oregon, and Washington . Fishery Bulletin , 117 : 291 – 302 . Google Scholar Crossref Search ADS WorldCat Linkowski T. B. 1985 . Population biology of the myctophid fish Gymnoscopelus nicholsi (Gillbert, 1911) from the western South Atlantic . Journal of Fish Biology , 27 : 683 – 698 . Google Scholar Crossref Search ADS WorldCat Linkowski T. B. 1996 . Lunar rhythms of vertical migrations coded in otolith microstructure of North Atlantic lanternfishes, genus Hygophum (Myctophidae) . Marine Biology , 124 : 495 – 508 . Google Scholar Crossref Search ADS WorldCat Linkowski T. B. , Radtke R. L. , Lenz P. H. 1993 . Otolith microstructure, age and growth of 2 species of Ceratoscopelus (Osteichthyes, Myctophidae) from the eastern North Atlantic . Journal of Experimental Marine Biology and Ecology , 167 : 237 – 260 . Google Scholar Crossref Search ADS WorldCat Lisovenko L. A. , Efremenko V. N. 1983 . Characteristic Features of Reproduction of Abundant Mesopelagic Fish of the West Antarctic . VNIRO Research Report , Moscow . 36 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Lisovenko L. A. , Prut’ko V. G. 1987a . Reproductive biology of Diaphus suborbitalis (Myctophidae) in the equatorial part of the Indian Ocean. 1. Nature of oogenesis and type of spawning . Journal of Ichthyology , 26 : 47 – 58 . Google Scholar OpenURL Placeholder Text WorldCat Lisovenko L. A. , Prut’ko V. G. 1987b . Reproductive biology of Diaphus suborbitalis (Myctophidae) in the equatorial part of the Indian Ocean. 2. Fecundity and reproductive potential . Journal of Ichthyology , 27 : 1 – 12 . Google Scholar OpenURL Placeholder Text WorldCat Lowerre-Barbieri S. K. , Brown-Peterson N. J. , Murua H. , Tomkiewicz J. , Wyanski D. M. , Saborido-Rey F. 2011a . Emerging issues and methodological advances in fisheries reproductive biology . Marine and Coastal Fisheries: Dynamics, Management, and Ecosystem Science , 3 : 32 – 51 . Google Scholar Crossref Search ADS WorldCat Lowerre-Barbieri S. K. , Ganias K. , Saborido-Rey F. , Murua H. , Hunter J. R. 2011b . Reproductive timing in marine fishes: variability, temporal scales, and methods . Marine and Coastal Fisheries , 3 : 71 – 91 . Google Scholar Crossref Search ADS WorldCat Lusher A. L. , O'Donnell C. , Officer R. , O'Connor I. 2016 . Microplastic interactions with North Atlantic mesopelagic fish . ICES Journal of Marine Science , 73 : 1214 – 1225 . Google Scholar Crossref Search ADS WorldCat Marks A. D. , Kerstetter D. W. , Wyanski D. M. , Sutton T. T. 2020 . Reproductive ecology of dragonfishes (Stomiiformes: Stomiidae) in the Gulf of Mexico . Frontiers in Marine Science , 7 : 1 – 17 . Google Scholar Crossref Search ADS PubMed WorldCat Mazhiryna G. P. , Poletayev V. A. 1990 . Distribution and some features of the biology of E. carlsbergi (Myctophidae) of the Atlantic Antarctic. All-Union Conference on Reserves of Biological Food Resources of the Ocean and the USSR Coastal Seas . Kaliningrad , Moscow . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC McBride R. S. , Wuenschel M. J. , Nitschke P. , Thornton G. , King J. R. 2013 . Latitudinal and stock-specific variation in size- and age-at-maturity of female winter flounder, Pseudopleuronectes americanus, as determined with gonad histology . Journal of Sea Research , 75 : 41 – 51 . Google Scholar Crossref Search ADS WorldCat Meera K. M. , Hashim M. , Sanjeevan V. N. , Jayasankar J. , Ambrose T. V. , and Sudhakar M. 2019 . Systematics and biology of the blue lanternfish, Diaphus coeruleus from the south-eastern Arabian Sea . Journal of the Marine Biological Association of the United Kingdom , 99 : 239 – 248 . Google Scholar Crossref Search ADS WorldCat Methot R. D. 1981 . Spatial covariation of daily growth rates of larval northern anchovy, Engraulis mordax and northern lampfish, Stenobrachius leucopsaurus. Rapports et Proces-verbaux des Réunions . Conseil International Pour L'Éxploration de la Mer , 178 : 424 – 431 . Google Scholar OpenURL Placeholder Text WorldCat Miya M. , Nemoto T. 1986 . Life-history and vertical-distribution of the mesopelagic fish Cyclothone alba (family Gonostomatidae) in Sagami Bay, central Japan . Deep-Sea Research Part A-Oceanographic Research Papers , 33 : 1053 – 1068 . Google Scholar Crossref Search ADS WorldCat Miya M. , Nemoto T. 1987a . Reproduction, growth and vertical-distribution of the mesopelagic and bathypelagic fish Cyclothone atraria (Pisces, Gonostomatidae) in Sagami Bay, Central Japan . Deep-Sea Research Part A-Oceanographic Research Papers , 34 : 1565 – 1577 . Google Scholar Crossref Search ADS WorldCat Miya M. , Nemoto T. 1987b . Some aspects of the biology of file micronektonic fish Cyclothone pallida and Cyclothone acclinidens (Pisces: Gonostomalidae) in Sagami Bay, Central Japan . Journal of the Oceanographical Society of Japan , 42 : 473 – 480 . Google Scholar Crossref Search ADS WorldCat Miya M. , Nemoto T. 1991 . Comparative life histories of the mesopelagic and bathypelagic fishes of the genus Cyclothone (Pisces, Gonostomatidae) in Sagami Bay . Central Japan. Deep-Sea Research Part A-Oceanographic Research Papers , 38 : 67 – 89 . Google Scholar Crossref Search ADS WorldCat Moku M. , Hayashi A. , Mori K. , Watanabe Y. 2005 . Validation of daily otolith increment formation in the larval myctophid fish Diaphus slender-type spp . Journal of Fish Biology , 67 : 1481 – 1485 . Google Scholar Crossref Search ADS WorldCat Moku M. , Ishimaru K. , Kawaguchi K. 2001 . Growth of larval and juvenile Diaphus theta (Pisces: Myctophidae) in the transitional waters of the western North Pacific . Ichthyological Research , 48 : 385 – 390 . Google Scholar Crossref Search ADS WorldCat Munk K. M. 2001 . Maximum ages of groundfishes in waters off Alaska and British Columbia and considerations of age determination . Alaska Fisheries Research Bulletin , 8 : 12 – 21 . Google Scholar OpenURL Placeholder Text WorldCat Murua H. , Kraus G. , Saborido-Rey F. , Wittames P. R. , Thorsen A. , Junquera S. 2003 . Procedures to estimate fecundity of marine fishes in relation to their reproductive strategy . Journal of Northwest Atlantic Fishery Science , 33 : 33 – 54 . Google Scholar Crossref Search ADS WorldCat Murua H. , Saborido-Rey F. 2003 . Female reproductive strategies of marine fish species of the North Atlantic . Journal of Northwest Atlantic Fishery Science , 33 : 23 – 31 . Google Scholar Crossref Search ADS WorldCat Namiki C. , Katsuragawa M. , Zani-Teixeira M. L. 2015 . Growth and mortality of larval Myctophum affine (Myctophidae, Teleostei) . Journal of Fish Biology , 86 : 1335 – 1347 . Google Scholar Crossref Search ADS PubMed WorldCat Olivar M. P. , Hulley P. A. , Castellón A. , Emelianov M. , López C. , Tuset V. M. , Contreras T. et al. 2017 . Mesopelagic fishes across the tropical and equatorial Atlantic: biogeographical and vertical patterns . Progress in Oceanography , 151 : 116 – 137 . Google Scholar Crossref Search ADS WorldCat Olney J. E. , McBride R. S. 2003 . Intraspecific variation in batch fecundity of American shad: revisiting the paradigm of reciprocal latitudinal trends in reproductive traits . American Fisheries Society Symposium , 35 : 185 – 192 . Google Scholar OpenURL Placeholder Text WorldCat Ozawa T. , Peñaflor G. C. 1990 . Otolith microstructure and early ontogeny of a Myctophid species Benthosema pterotum . Nippon Suisan Gakkaishi , 56 : 1987 – 1995 . Google Scholar Crossref Search ADS WorldCat Pietsch T. W. 1975 . Precocious sexual parasitism in the deep sea ceratioid anglerfish, Cryptopsaras couesi Gill . Nature , 256 : 38 – 40 . Google Scholar Crossref Search ADS WorldCat Prellezo R. 2019 . Exploring the economic viability of a mesopelagic fishery in the Bay of Biscay . ICES Journal of Marine Science , 76 : 771 – 779 . Google Scholar Crossref Search ADS WorldCat Prosch R. M. 1991 . Reproductive biology and spawning of the myctophid Lampanyctodes hectoris and the sternoptychid Maurolicus muelleri in the southern Benguela Ecosystem . South African Journal of Marine Science , 10 : 241 – 252 . Google Scholar Crossref Search ADS WorldCat Proud R. , Handegard N. O. , Kloser R. J. , Cox M. J. , Brierley A. S. 2018 . From siphonophores to deep scattering layers: uncertainty ranges for the estimation of global mesopelagic fish biomass . ICES Journal of Marine Science , 76 : 718 – 733 . Google Scholar Crossref Search ADS WorldCat Rideout R. M. , Morgan M. J. 2010 . Relationships between maternal body size, condition and potential fecundity of four north-west Atlantic demersal fishes . Journal of Fish Biology , 76 : 1379 – 1395 . Google Scholar Crossref Search ADS PubMed WorldCat Robertson D. A. 1977 . Planktonic eggs of the lanternfish, Lampanyctodes hectoris (family Myctophidae) . Deep Sea Research , 24 : 849 – 852 . Google Scholar Crossref Search ADS WorldCat Rowedder U. 1979 . Some aspects of the biology of Electrona antarctica (Gunther, 1878) (Family Myctophidae) . Meeresforschung , 27 : 244 – 251 . Google Scholar OpenURL Placeholder Text WorldCat Salvanes A. G. V. , Kristoffersen J. B. 2001 . Mesopelagic fishes . In: Encyclopedia of Ocean Sciences, Vol. 3, pp. 1711 – 1717 . Ed. by Steele J. H. , Turekian K. K. , Thorpe S. A. . Academic Press, University of Bergen, Bergen, Norway . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Salvanes A. G. V. , Stockley B. M. 1996 . Spatial variation of growth and gonadal developments of Maurolicus muelleri in the Norwegian sea and in a Norwegian fjord . Marine Biology , 126 : 321 – 332 . Google Scholar Crossref Search ADS WorldCat Sarmiento-Lezcano A.-N. , Triay-Portella R. , Juan Castro J. , Rubio-Rodriguez U. , Pajuelo J. G. 2018 . Age-based life-history parameters of the mesopelagic fish Notoscopelus resplendens (Richardson, 1845) in the Central Eastern Atlantic . Fisheries Research , 204 : 412 – 423 . Google Scholar Crossref Search ADS WorldCat Sassa C. 2019 . Reproduction and early life history of mesopelagic fishes in the kuroshio region: a review of recent advances. In Kuroshio Current: Physical, Biogeochemical, and Ecosystem Dynamics, Geophysical Monograph 243 , pp. 273–294. 1st edn. Ed. by Nagai T. , Saito H. , Suzuki K. , Takahashi M. . John Wiley & Sons, Inc . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Sassa C. , Ohshimo S. , Tanaka H. , Tsukamoto Y. 2014 . Reproductive biology of Benthosema pterotum (Teleostei: Myctophidae) in the shelf region of the East China Sea . Journal of the Marine Biological Association of the United Kingdom , 94 : 423 – 433 . Google Scholar Crossref Search ADS WorldCat Sassa C. , Takahashi M. 2018 . Comparative larval growth and mortality of mesopelagic fishes and their predatory impact on zooplankton in the Kuroshio region . Deep-Sea Research Part I - Oceanographic Research Papers , 131 : 121 – 132 . Google Scholar Crossref Search ADS WorldCat Sassa C. , Takahashi M. , Tsukamoto Y. 2015 . Distribution, hatch-date, growth, and mortality of larval Benthosema pterotum (Pisces: Myctophidae) in the shelf region of the East China Sea . Journal of the Marine Biological Association of the United Kingdom , 95 : 161 – 174 . Google Scholar Crossref Search ADS WorldCat Sassa C. , Tanaka H. , Ohshimo S. 2016 . Comparative reproductive biology of three dominant myctophids of the genus Diaphus on the slope region of the East China Sea . Deep-Sea Research Part I—Oceanographic Research Papers , 115 : 145 – 158 . Google Scholar Crossref Search ADS WorldCat Saunders R. A. , Lourenco S. , Vieira R. P. , Collins M. A. , Assis C. A. , Xavier J. C. 2020 . Age and growth of Brauer's lanternfish Gymnoscopelus braueri and rhombic lanternfish Krefftichthys anderssoni (Family Myctophidae) in the Scotia Sea, Southern Ocean . Journal of Fish Biology , 96 : 364 – 377 . Google Scholar Crossref Search ADS PubMed WorldCat Shust K. V. , Kochkin N. P. 1985 . Age, Growth Rate and Age/Length Composition of Populations of Abundant Neritic and Mesopelagic Fish of the Southern Ocean . VNIRO, Moscow . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Siebel B. A. , Drazen J. C. 2007 . The rate of metabolism in marine animals: environmental constraints, ecological demands and energetic opportunities . Philosophical Transactions of the Royal Society B—Biological Sciences , 362 : 2061 – 2078 . Google Scholar Crossref Search ADS WorldCat Siegel D. A. , Buesseler K. O. , Doney S. C. , Sailley S. , Behrenfeld M. J. , Boyd P. W. 2014 . Global assessment of ocean carbon export by combining satellite observations and food-web models . Global Biogeochemical Cycles , 28 : 181 – 196 . Google Scholar Crossref Search ADS WorldCat Smoker W. , Pearcy W. G. 1970 . Growth and reproduction of the lanternfish Stenobrachius leucopsarus . Journal of the Fisheries Research Board of Canada , 27 : 1265 – 1275 . Google Scholar Crossref Search ADS WorldCat St. John M. A. , Borja A. , Chust G. , Heath M. , Grigorov I. , Mariani P. , Martin A. P. et al. 2016 . A dark hole in our understanding of marine ecosystems and their services: perspectives from the mesopelagic community . Frontiers in Marine Science , 3 : 1 – 6 . Google Scholar Crossref Search ADS WorldCat Stearns S. C. 1992 . The Evolution of Life Histories . Oxford University Press , New York . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Stequert B. , Menard F. , Marchal E. 2003 . Reproductive biology of Vinciguerria nimbaria in the equatorial waters of the eastern Atlantic Ocean . Journal of Fish Biology , 62 : 1116 – 1136 . Google Scholar Crossref Search ADS WorldCat Stevenson D. K. , Campana S. E. 1992 . Otolith microstructure examination and analysis . Canadian Special Publication of Fisheries and Aquatic Sciences , 117 : 126 p. Google Scholar OpenURL Placeholder Text WorldCat Suthers I. M. 1996 . Spatial variability of recent otolith growth and RNA indices in pelagic juvenile Diaphus kapalae (Myctophidae): an effect of flow disturbance near an island? Marine and Freshwater Research , 47 : 273 – 282 . Google Scholar Crossref Search ADS WorldCat Sutton T. T. , Porteiro F. M. , Heino M. , Byrkjedal I. , Langhelle G. , Anderson C. I. H. , Horne J. et al. 2008 . Vertical structure, biomass and topographic association of deep-pelagic fishes in relation to a mid-ocean ridge system . Deep-Sea Research II , 55 : 161 – 184 . Google Scholar Crossref Search ADS WorldCat Sutton T. T. , Wiebe P. H. , Madin L. , Bucklin A. 2010 . Diversity and community structure of pelagic fishes to 5000 m depth in the Sargasso Sea . Deep-Sea Research Part II—Topical Studies in Oceanography , 57 : 2220 – 2233 . Google Scholar Crossref Search ADS WorldCat Takagi K. , Yatsu A. , Moku M. , Sassa C. 2006 . Age and growth of lanternfishes, Symbolophorus californiensis and Ceratoscopelus warmingii (Myctophidae), in the Kuroshio-Oyashio Transition Zone . Ichthyological Research , 53 : 281 – 289 . Google Scholar Crossref Search ADS WorldCat Tomas J. , Panfili J. 2000 . Otolith microstructure examination and growth patterns of Vinciguerria nimbaria (Photichthyidae) in the tropical Atlantic Ocean . Fisheries Research , 46 : 131 – 145 . Google Scholar Crossref Search ADS WorldCat Tracey D. M. , Horn P. L. 1999 . Background and review of ageing orange roughy (Hoplostethus atlanticus, Trachichthyidae) from New Zealand and elsewhere . New Zealand Journal of Marine and Freshwater Research , 33 : 67 – 86 . Google Scholar Crossref Search ADS WorldCat Tsarin S. A. 1994 . Age, growth, and some production characteristics of Ceratoscopelus warmingii (Myctophidae) in the tropical zone of the Indian Ocean . Journal of Ichthyology , 34 : 59 – 72 . Google Scholar OpenURL Placeholder Text WorldCat Vitale F. , Svedäng H. , Cardinale M. 2006 . Histological analysis invalidates macroscopically determined maturity ogives of the Kattegat cod (Gadus morhua) and suggests new proxies for estimating maturity status of individual fish . ICES Journal of Marine Science , 63 : 485 – 492 . Google Scholar Crossref Search ADS WorldCat Wang Y. , Zhang J. , Chen Z. , Jiang Y. , Xu S. , Li Z. , Wang X. , Ying Y. , Zhao X. , Zhou M. 2019 . Age and growth of Myctophum asperum in the South China Sea based on otolith microstructure analysis. Deep-Sea Research Part II, 167 : 121 – 127 . Wedding L. M. , Reiter S. M. , Smith C. R. , Gjerde K. M. , Kittinger J. N. , Friedlander A. M. , Gaines S. D. et al. 2015 . Managing mining of the deep seabed . Science , 349 : 144 – 145 . Google Scholar Crossref Search ADS PubMed WorldCat West G. 1990 . Methods of assessing ovarian development in fishes: a review . Australian Journal of Marine and Freshwater Research , 41 : 199 – 222 . Google Scholar Crossref Search ADS WorldCat Wieczorek A. M. , Morrison L. , Croot P. L. , Allcock A. L. , MacLoughlin E. , Savard O. , Brownlow H. et al. 2018 . Frequency of microplastics in mesopelagic fishes from the Northwest Atlantic . Frontiers in Marine Science , 5 : 39 . Google Scholar Crossref Search ADS WorldCat Wilson R. W. , Millero F. J. , Taylor J. R. , Walsh P. J. , Christensen V. , Jennings S. , Grosell M. 2009 . Contribution of fish to the marine inorganic carbon cycle . Science , 323 : 359 – 362 . Google Scholar Crossref Search ADS PubMed WorldCat Wootton R. J. 1992 . Constraints in the evolution of fish life histories . Netherlands Journal of Zoology , 42 : 291 – 303 . Google Scholar Crossref Search ADS WorldCat Young J. W. , Blaber S. J. M. , Rose R. 1987 . Reproductive biology of three species of midwater fishes associated with the continental slope of eastern Tasmania, Australia . Marine Biology , 95 : 323 – 332 . Google Scholar Crossref Search ADS WorldCat Young J. W. , Bulman C. M. , Blaber S. J. M. , Wayte S. E. 1988 . Age and growth of the lanternfish Lampanyctodes hectoris (Myctophidae) from eastern Tasmania, Australia . Marine Biology , 99 : 569 – 576 . Google Scholar Crossref Search ADS WorldCat Young J. W. , Hunt B. P. V. , Cook T. R. , Llopiz J. K. , Hazen E. L. , Pethybridge H. R. , Ceccarelli D. et al. 2015 . The trophodynamics of marine top predators: current knowledge, recent advances and challenges . Deep-Sea Research Part II - Topical Studies in Oceanography , 113 : 170 – 187 . Google Scholar Crossref Search ADS WorldCat Zenteno J. I. , Bustos C. A. , Landaeta M. F. 2014 . Larval growth, condition and fluctuating asymmetry in the otoliths of a mesopelagic fish in an area influenced by a large Patagonian glacier . Marine Biology Research , 10 : 504 – 514 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2021. 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)