ICES Journal of Marine Science: Journal du Conseil

Lynch, T, P;Smallwood, C, B;Ochwada-Doyle, F, A;Lyle,, J;Williams,, J;Ryan, K, L;Devine,, C;Gibson,, B;Jordan,, A

doi: 10.1093/icesjms/fsz092pmid: N/A

Abstract Recreational fishing is popular in Australia and is managed by individual states in consultation with the Commonwealth for those fisheries that they regulate and also for Australian Marine Parks (AMPs). Fishers regularly access both state and offshore Commonwealth waters but this offshore component of the recreational fishery is poorly understood. Our study tested the functionality of existing state-based surveys in Western Australia (WA) and New South Wales (NSW) to better inform Commonwealth fisheries and AMP managers about recreational fishing in their jurisdictions. Catch estimates for nine species of interest to the Commonwealth were developed and two case study AMPs [Ningaloo (WA) and The Hunter (NSW)] were also chosen to test the ability of the state survey data to be disaggregated to the park scale. As each state’s fishery survey designs were contextual to their own management needs, the application of the data to Commonwealth jurisdictions were limited by their statistical power, however aspects of each states surveys still provided useful information. Continued evolution of state-wide survey methods, including collection of precise spatial data, and regional over-sampling would be beneficial, particularly where there are multiple stakeholder and jurisdictional interests. National coordination, to temporally align state surveys, would also add value to the existing approaches. Introduction Unlike commercial fisheries, which often have well-defined areas of operations and mandatory reporting, open-access marine recreational fisheries (MRFs) can cross jurisdictional boundaries and require sampling to estimate metrics such as participation, catch, and effort (McCluskey and Lewison, 2008). Surveys of MRF to understand their impact on fish stocks as well as their socio-economic characteristics have grown in importance globally over the past few decades (McPhee et al., 2002; Ihde et al., 2011; Venerus and Cedrola, 2017; Hyder et al., 2018). Many jurisdictions undertake coordinated and consistent national recreational fishing surveys to provide these data, and they are especially common in Europe, North America, and Oceania, although survey design and data quality vary (Hyder et al., 2018). Since 1981, the National Oceans and Atmospheric Administration has run large-scale phone and now mail offsite surveys as well as on-site intercept survey for catch rates (Coleman et al., 2004; The National Academies of Sciences and Medicine, 2017; Camp et al., 2018). Fisheries and Oceans Canada also use a mail recall design for national surveys of recreational fishing, which have been undertaken every 5 years since 1975 (Brownscombe et al., 2014). While New Zealand’s National Institute of Water and Atmospheric Research has undertaken a national survey that involved face-to-face recruitment and then telephone–diary interview follow-up to determine fishing activity (Holdsworth et al., 2018). The French Research Institute for Exploitation of the Sea has also undertaken a hybrid off-site random dial telephone survey in combination with on-site interviews targeting fishers (Herfaut et al., 2013). Recreational fishing is a very popular activity in Australia when compared to global norms, with an estimated national participation rate of 19.5% (in 2000/2001) (McPhee et al., 2002; Henry and Lyle, 2003; Cooke and Cowx, 2004; Lewin et al., 2006; Arlinghaus et al., 2015; Hyder et al., 2018). Regardless of this popularity, Australia does not have a time-series of coordinated national recreational fisheries statistics, with only single national survey conducted in 2000/2001 (Henry and Lyle, 2003). Since this baseline was established most states have continued to undertake state-wide or regional surveys but with little coordination and diverging methodologies between states (Lyle et al., 2014; Giri and Hall, 2015; Moore et al., 2015; West et al., 2015). While licencing systems could provide an effective sample frame for off-site surveys (Productivity Commission, 2016; Teixeira et al., 2016), and numbers of licence holders are often reported by fisheries departments, these systems are not consistent across Australian states, with some jurisdictions not requiring recreational licences or, where licences are present, many exemptions apply e.g. pensioners, children, veterans, and indigenous people (Table 1). Table 1. Licensing arrangements for marine recreational fishing in each Australian state. TAS VIC NSW QLD NT WA SA General saltwater licence (rod and line) No Yes Yes No No No No Boat fish licence Yesg Gear/species specific licences Yesa No No No No Yesh Yesh Indigenous exemption Yes Yesb Yes – – Yes – Pension exemption No Yes Yesi – – No – Child exemption No Yes Yese – – No – Age exemption No Yesc No – – No – Veterans exemption No Yesd Yesf – – No – Charter boat and fishing guide exemption No No Yes – – Yes – Fishing with someone who has a licence exception – – – Yes – TAS VIC NSW QLD NT WA SA General saltwater licence (rod and line) No Yes Yes No No No No Boat fish licence Yesg Gear/species specific licences Yesa No No No No Yesh Yesh Indigenous exemption Yes Yesb Yes – – Yes – Pension exemption No Yes Yesi – – No – Child exemption No Yes Yese – – No – Age exemption No Yesc No – – No – Veterans exemption No Yesd Yesf – – No – Charter boat and fishing guide exemption No No Yes – – Yes – Fishing with someone who has a licence exception – – – Yes – Data extracted from recreational fishing websites in each state. State codes are TAS, Tasmania; VIC, Victoria; NSW, New South Wales; QLD, Queensland; NT, Northern Territory; WA, Western Australia; SA, South Australia. a Licence is required for Rock Lobster pot, dive and ring, Abalone, Scallop, Graball Net, Mullet Net, Beach Seine Net, Set line. b No licence is required for a member of a traditional owner group fishing within an area subject to a natural resource agreement relevant to that traditional owner group. c No licence if over 70 years of age. d No licence for veterans with healthcare card TPI. e Exempt from a licence are the adult assisting a person under the age of 18 to take a fish using a single rod or to take prawns using a single dip or scoop net. f Licence exceptions for Veteran healthcare cards holders TPI, EDA, or letter from Veteran's affairs minister. g Licence exempt for fishing from a boat without a motor, such as a kayak. h Rock Lobster, Abalone, Net fishing (set, haul, and throw). i Concessions for Commonwealth Pension, DVA Concession, and Government Seniors card holders and persons under 16. View Large Table 1. Licensing arrangements for marine recreational fishing in each Australian state. TAS VIC NSW QLD NT WA SA General saltwater licence (rod and line) No Yes Yes No No No No Boat fish licence Yesg Gear/species specific licences Yesa No No No No Yesh Yesh Indigenous exemption Yes Yesb Yes – – Yes – Pension exemption No Yes Yesi – – No – Child exemption No Yes Yese – – No – Age exemption No Yesc No – – No – Veterans exemption No Yesd Yesf – – No – Charter boat and fishing guide exemption No No Yes – – Yes – Fishing with someone who has a licence exception – – – Yes – TAS VIC NSW QLD NT WA SA General saltwater licence (rod and line) No Yes Yes No No No No Boat fish licence Yesg Gear/species specific licences Yesa No No No No Yesh Yesh Indigenous exemption Yes Yesb Yes – – Yes – Pension exemption No Yes Yesi – – No – Child exemption No Yes Yese – – No – Age exemption No Yesc No – – No – Veterans exemption No Yesd Yesf – – No – Charter boat and fishing guide exemption No No Yes – – Yes – Fishing with someone who has a licence exception – – – Yes – Data extracted from recreational fishing websites in each state. State codes are TAS, Tasmania; VIC, Victoria; NSW, New South Wales; QLD, Queensland; NT, Northern Territory; WA, Western Australia; SA, South Australia. a Licence is required for Rock Lobster pot, dive and ring, Abalone, Scallop, Graball Net, Mullet Net, Beach Seine Net, Set line. b No licence is required for a member of a traditional owner group fishing within an area subject to a natural resource agreement relevant to that traditional owner group. c No licence if over 70 years of age. d No licence for veterans with healthcare card TPI. e Exempt from a licence are the adult assisting a person under the age of 18 to take a fish using a single rod or to take prawns using a single dip or scoop net. f Licence exceptions for Veteran healthcare cards holders TPI, EDA, or letter from Veteran's affairs minister. g Licence exempt for fishing from a boat without a motor, such as a kayak. h Rock Lobster, Abalone, Net fishing (set, haul, and throw). i Concessions for Commonwealth Pension, DVA Concession, and Government Seniors card holders and persons under 16. View Large Regulatory responsibilities for Australian fisheries are shared between the Australian Commonwealth Government (herein referred to as Commonwealth) and the state governments based on agreements made under the Offshore Constitutional Settlement. Generally, the demarcation between state and Commonwealth waters occurs at 3 nautical miles (nm) out to sea, with Commonwealth waters then extending to 200 nm offshore. Commercial fisheries are managed by the Commonwealth through the Australian Fisheries Management Authority (AFMA) under the Fisheries Management Act 1991, although some fisheries are managed by the relevant states under agreements with the Commonwealth, often out to 80 nm. All MRFs are managed by the states, are open access and fishers regulated with a combination of input and output controls such as bag, gear, and size limits, licensing and spatial closures (Kearney, 2001). It is thought that recent expansion of recreational fishers into offshore waters has been facilitated by the increased affordability of marine technology (i.e. GPS, echo sounders, electric reels, vessels) (West et al., 2015; Evans et al., 2017). The Fisheries Management Act 1991 has recently been amended (2017) such that AFMA is now required to consider the interests of the recreational sector as well as all sources of mortality when setting sustainable catch rates (Agriculture and Water Resources, 2018). There is, however, a paucity of data on the offshore MRF and generally recreational catch is not incorporated into Commonwealth harvest strategies. One exception to this is southern bluefin tuna (Thunnus maccoyii), where 250 tonne of Australia’s 2017 national catch allocation was set aside for non-commercial mortality (AFMA, 2018). Recreational fisheries are a social activity that are not driven by the economics of the activity, and are therefore difficult to manage within objectives that are normal to commercial fisheries such as quotas, production, and profit. Maximizing social utility and non-market value of these public resources is an active but relatively new area of research (Brownscombe et al., 2014, 2019; Southwick et al., 2018). Also, niche fisheries, such as those occurring offshore are difficult to assess using broad scale state-wide surveys due to lack of sufficient statistical power (Griffiths et al., 2010,, 2017). All of these issues contribute to the complexity of consideration of MRF for offshore fisheries. Marine Parks are located both within the states inshore waters and the Commonwealth's offshore waters, though there are some complications to this general rule around state-controlled islands (Figure 1). Planning for the implementation of the Commonwealth’s Australian Marine Parks (AMPs) commenced in 1998 (Nevill and Ward, 2009) and as of July 2018 there are 58 AMP managed by one Commonwealth agency (Parks Australia) under the Environmental Protection and Biodiversity Conservation Act 1999. The main objectives of AMPs are (i) protection and conservation of biodiversity and other natural, cultural, and heritage values and (ii) ecologically sustainable use and enjoyment of the natural resources within marine parks where this is consistent with objective (i). In addition to the AMPs, other Commonwealth parks include the Great Barrier Reef Marine Park, which is managed by the Great Barrier Reef Marine Park Authority and the Heard Island and McDonald Islands Marine Park, which are managed by the Australian Antarctic Division. Figure 1. View largeDownload slide Network of AMPs in Commonwealth waters showing details of (a) NMP in WA and (b) HMP in NSW. Figure 1. View largeDownload slide Network of AMPs in Commonwealth waters showing details of (a) NMP in WA and (b) HMP in NSW. Systematic data collection has been identified as critical for ongoing planning, research, and monitoring of management plans for marine park networks (Day, 2008; Lynch et al., 2014; Emslie et al., 2015; Horigue et al., 2015). There are, however, considerable challenges in undertaking data collection for recreational fishing in offshore waters. State-based recreational fishing surveys are designed to enable state management agencies to make informed decisions on the sustainable management of fisheries. Although such surveys target all recreational fishing occurring across all ecosystems, reporting is generally at broad spatial (bioregional) and temporal (seasonal) scales due to the high cost of implementation (West et al., 2015; Ryan et al., 2017). Smaller-scale, targeted research may also be undertaken to meet legislated requirements for fishery performance and resource allocation, to develop new survey approaches or address specific research questions (Smallwood et al., 2012; Wise et al., 2012; Crowe et al., 2013; Lynch, 2014; Wood et al., 2016; Newman et al., 2018). AMP management is mainly concerned with the performance of zoning and management plans in achieving conservation of biodiversity and other natural, cultural, socio-economic, and heritage values. At the coarsest level, there is a need for gross numbers and activity types of park users, which are important for targeting outreach, compliance, and infrastructure for parks. At the other end of the scale is an understanding of detailed levels of usage and catch by park users within different park zones, which are needed to ensure bio-diversity is being conserved through the management plan. In the absence of data collection on recreational fishing in offshore waters by the Commonwealth, this paper examines if two state-wide MRF surveys, conducted throughout Western Australia (WA) and New South Wales (NSW), could meet their information needs. The specific aims included; (i) a comparison of state-based approaches for data collection in WA and NSW, (ii) estimates (with associated uncertainty) of catch occurring state-wide for nine species of interest to AFMA, and (iii) estimates (with associated uncertainty) of fishing effort and catch (all species) occurring within two AMP: Ningaloo Marine Park (NMP) in WA and the Hunter Marine Park (HMP) in NSW. Methods Australia has a continental coastline of 35 877 km (Short and Woodroffe, 2009), of which WA has the largest coastal extent (12 880 km or 35.9%) (Figure 1). The state is sparsely populated with 73% of the state’s population of 2.5 million living in the capital city, Perth (ABS, 2018). NSW has the smallest coastline of all the Australian states of 2007 km (Short and Woodroffe, 2009), however, NSW has the largest population with 7.9 million residents, representing 32% of Australia’s population and of which 62% reside in the capital city of Sydney (ABS, 2018). In WA, the participation in recreational fishing has changed from 19% in 1989/1990 (Lindner and McLeod, 1991), to 31.1% in 2015/2016 (Ryan et al., 2017). Recreational fishing is less popular in the heavily urbanized NSW, with 11.9% of the population participating in recreational fishing. However, due to the large overall state population, NSW has more recreational fishers compared to WA with an estimated 849 249 fishing annually (West et al., 2015). NMP is located 1200 km north of Perth and, with the associated nearshore state Marine Park, includes one of the largest fringing coral reef systems in the world (Director of National Parks, 2018a) (Figure 1). While the area is sparsely populated, with a residential population of ∼10 000 people, there are ∼250 000 visitors to the area annually (MPRA and CALM, 2005; Smallwood et al., 2012). NMP covers an area of ∼2435 km2, with depths ranging from 30 to 500 m, and has been assigned as an IUCN category IV which includes two zones; a National Park Zone (II) and Recreational Use Zone (IV) (Director of National Parks, 2018a). The HMP is located off the NSW coastline ∼280 km north of Sydney. Similar to Ningaloo it adjoins the state’s Port Stephens–Great Lakes Marine Park and covers an area of ∼6257 km2, stretching from NSW state waters to ∼100 km offshore (Buxton and Cochrane, 2015). Water depths within the HMP range from 15 to 6000 m and covers the area on the continental shelf outside state waters. It is assigned IUCN category IV and includes two zones; a Habitat Protection Zone (IV) and Special Purpose Zone (Trawl) (VI) (Director of National Parks, 2018b). The nine species of interest to AFMA (from a recreational fishing perspective) included in this paper are gummy sharks (Mustelus antarcticus and M. stevensi), school sharks (Galeorhinus galeus), southern bluefin tuna (T. maccoyii), yellowfin tuna (Thunnus albacares), striped marlin (Kajikia audax), broadbill swordfish (Xiphias gladius), blue-eye trevalla (Hyperoglyphe antarctica), pink Ling (Genypterus blacodes), gemfish (Rexea solandri), bluespotted flathead (Platycephalus caeruleopunctatus), and deepwater flathead (Neoplatycephalus conatus). Western Australia A telephone–diary survey of recreational fishing is used to provide catch and effort estimates at broad spatial (state-wide, regional) and temporal (annual, seasonal) scales. The sampling frame for this survey was the database of Recreational Boat Fishing Licence (RBFL) holders. An RBFL is a mandatory licence that needs to be held by at least one member of the party when fishing from a motorized vessel (Table 1). Surveys have been completed in 2011/2012, 2013/2014, and 2015/2016 (Ryan et al., 2013, 2015, 2017). The main survey elements and output specifications are presented in Table 2. Table 2. Survey elements and output specifications for telephone–diary surveys in WA and NSW. Survey element Item WA NSW Survey design Sampling frame RBFL White Pages telephone directory Sampling design Stratified random sample Stratified random sample Primary sample RBFL (person based reporting on boat catch) White Pages Listed Number (household based) Data collection Sample selection and stratification Random selection of RBFL holders within 11 statistical regions of different population size, of which nine are Regional Development Commission boundaries Random selection of households from ten Statistical Areas (SA4) defined by the Australian Bureau of Statistics Recruitment Telephone interview Telephone interview Data collection Telephone–diary interview Telephone–diary interview Persons in scope Residency WA and interstate residents NSW/ACT residents only (interstate fishing participation and effort by NSW/ACT residents was assessed separately) Age <5 years excluded <5 years excluded Activities Sectors Recreational fishing only Recreational fishing only Platform Boat Shore and boat Boat type Private and for-hire vessels (excluding charter) Private, for-hire, and charter vessels (though separate charter log book database is also kept) Methods All methods including line fishing, diving, hand collection, netting, potting, and spearing All methods including line fishing, diving, hand collection netting, potting, and spearing Species Species All aquatic (animal) species All aquatic (animal) species Catch Retained and released Retained and released Geographic scope Fishing activity State-wide State-wide Marine bioregions (4) Regions/fishing zones (10) 10 × 10 nm block Waterbody (coastal fishing separated as estuarine, inshore <5 km from coastline and offshore >5 km from coastline) Fishing sites (some GPS) Fishing sites (classified using GIS coding system) Fishing access Boat ramps (public and private), moorings, and marinas Boat ramps (public and private), moorings, and marinas Shore fishing from ocean beach, ocean rocks, manmade and natural structures, natural shore Temporal scope Duration 12-month longitudinal survey 12-month longitudinal survey Coverage 24 h 24 h Survey periods 1 March 2011–29 February 2012 1 June 2013–31 May 2014 1 May 2013–30 April 2014 1 September 2015–31 August 2016 Survey outputs Expansion RBFL population ABS population and non-response adjustments Fishing effort Boat days Fisher days Total catch By number (for key species) By number (for key species) Survey element Item WA NSW Survey design Sampling frame RBFL White Pages telephone directory Sampling design Stratified random sample Stratified random sample Primary sample RBFL (person based reporting on boat catch) White Pages Listed Number (household based) Data collection Sample selection and stratification Random selection of RBFL holders within 11 statistical regions of different population size, of which nine are Regional Development Commission boundaries Random selection of households from ten Statistical Areas (SA4) defined by the Australian Bureau of Statistics Recruitment Telephone interview Telephone interview Data collection Telephone–diary interview Telephone–diary interview Persons in scope Residency WA and interstate residents NSW/ACT residents only (interstate fishing participation and effort by NSW/ACT residents was assessed separately) Age <5 years excluded <5 years excluded Activities Sectors Recreational fishing only Recreational fishing only Platform Boat Shore and boat Boat type Private and for-hire vessels (excluding charter) Private, for-hire, and charter vessels (though separate charter log book database is also kept) Methods All methods including line fishing, diving, hand collection, netting, potting, and spearing All methods including line fishing, diving, hand collection netting, potting, and spearing Species Species All aquatic (animal) species All aquatic (animal) species Catch Retained and released Retained and released Geographic scope Fishing activity State-wide State-wide Marine bioregions (4) Regions/fishing zones (10) 10 × 10 nm block Waterbody (coastal fishing separated as estuarine, inshore <5 km from coastline and offshore >5 km from coastline) Fishing sites (some GPS) Fishing sites (classified using GIS coding system) Fishing access Boat ramps (public and private), moorings, and marinas Boat ramps (public and private), moorings, and marinas Shore fishing from ocean beach, ocean rocks, manmade and natural structures, natural shore Temporal scope Duration 12-month longitudinal survey 12-month longitudinal survey Coverage 24 h 24 h Survey periods 1 March 2011–29 February 2012 1 June 2013–31 May 2014 1 May 2013–30 April 2014 1 September 2015–31 August 2016 Survey outputs Expansion RBFL population ABS population and non-response adjustments Fishing effort Boat days Fisher days Total catch By number (for key species) By number (for key species) View Large Table 2. Survey elements and output specifications for telephone–diary surveys in WA and NSW. Survey element Item WA NSW Survey design Sampling frame RBFL White Pages telephone directory Sampling design Stratified random sample Stratified random sample Primary sample RBFL (person based reporting on boat catch) White Pages Listed Number (household based) Data collection Sample selection and stratification Random selection of RBFL holders within 11 statistical regions of different population size, of which nine are Regional Development Commission boundaries Random selection of households from ten Statistical Areas (SA4) defined by the Australian Bureau of Statistics Recruitment Telephone interview Telephone interview Data collection Telephone–diary interview Telephone–diary interview Persons in scope Residency WA and interstate residents NSW/ACT residents only (interstate fishing participation and effort by NSW/ACT residents was assessed separately) Age <5 years excluded <5 years excluded Activities Sectors Recreational fishing only Recreational fishing only Platform Boat Shore and boat Boat type Private and for-hire vessels (excluding charter) Private, for-hire, and charter vessels (though separate charter log book database is also kept) Methods All methods including line fishing, diving, hand collection, netting, potting, and spearing All methods including line fishing, diving, hand collection netting, potting, and spearing Species Species All aquatic (animal) species All aquatic (animal) species Catch Retained and released Retained and released Geographic scope Fishing activity State-wide State-wide Marine bioregions (4) Regions/fishing zones (10) 10 × 10 nm block Waterbody (coastal fishing separated as estuarine, inshore <5 km from coastline and offshore >5 km from coastline) Fishing sites (some GPS) Fishing sites (classified using GIS coding system) Fishing access Boat ramps (public and private), moorings, and marinas Boat ramps (public and private), moorings, and marinas Shore fishing from ocean beach, ocean rocks, manmade and natural structures, natural shore Temporal scope Duration 12-month longitudinal survey 12-month longitudinal survey Coverage 24 h 24 h Survey periods 1 March 2011–29 February 2012 1 June 2013–31 May 2014 1 May 2013–30 April 2014 1 September 2015–31 August 2016 Survey outputs Expansion RBFL population ABS population and non-response adjustments Fishing effort Boat days Fisher days Total catch By number (for key species) By number (for key species) Survey element Item WA NSW Survey design Sampling frame RBFL White Pages telephone directory Sampling design Stratified random sample Stratified random sample Primary sample RBFL (person based reporting on boat catch) White Pages Listed Number (household based) Data collection Sample selection and stratification Random selection of RBFL holders within 11 statistical regions of different population size, of which nine are Regional Development Commission boundaries Random selection of households from ten Statistical Areas (SA4) defined by the Australian Bureau of Statistics Recruitment Telephone interview Telephone interview Data collection Telephone–diary interview Telephone–diary interview Persons in scope Residency WA and interstate residents NSW/ACT residents only (interstate fishing participation and effort by NSW/ACT residents was assessed separately) Age <5 years excluded <5 years excluded Activities Sectors Recreational fishing only Recreational fishing only Platform Boat Shore and boat Boat type Private and for-hire vessels (excluding charter) Private, for-hire, and charter vessels (though separate charter log book database is also kept) Methods All methods including line fishing, diving, hand collection, netting, potting, and spearing All methods including line fishing, diving, hand collection netting, potting, and spearing Species Species All aquatic (animal) species All aquatic (animal) species Catch Retained and released Retained and released Geographic scope Fishing activity State-wide State-wide Marine bioregions (4) Regions/fishing zones (10) 10 × 10 nm block Waterbody (coastal fishing separated as estuarine, inshore <5 km from coastline and offshore >5 km from coastline) Fishing sites (some GPS) Fishing sites (classified using GIS coding system) Fishing access Boat ramps (public and private), moorings, and marinas Boat ramps (public and private), moorings, and marinas Shore fishing from ocean beach, ocean rocks, manmade and natural structures, natural shore Temporal scope Duration 12-month longitudinal survey 12-month longitudinal survey Coverage 24 h 24 h Survey periods 1 March 2011–29 February 2012 1 June 2013–31 May 2014 1 May 2013–30 April 2014 1 September 2015–31 August 2016 Survey outputs Expansion RBFL population ABS population and non-response adjustments Fishing effort Boat days Fisher days Total catch By number (for key species) By number (for key species) View Large A screening survey of RBFL holders is completed in the 3 months prior to each 12-month longitudinal telephone–diary survey (Table 3). In 2015/2016, 4261 RBFL holders completed the screening survey and 2931 completed the diary survey, with response rates of >90 and >80% for each, respectively (Ryan et al., 2017). The surveys residential strata were based on nine Regional Development Commissions areas, in addition to the Perth Metropolitan Area (∼50% of licence holders in each survey year) and interstate populations (<2% of licence holders in each survey year) (Ryan et al., 2013, 2015, 2017). As a stratified random sampling design the samples in each stratum were proportionally allocated to the RBFL population and were divided into homogeneous units to reduce variance (Cochran, 1977; Pollock et al., 1994). Over-sampling for strata with low residential populations (i.e. Gascoyne, Kimberley) increased the number of survey participants for these strata and ensures that fishing activity in regional areas was reported with sufficient sample sizes to produce robust estimates. Table 3. Sample size (where the primary sampling unit is RBFL holders reporting on all catch to the boat) for screening and telephone–diary survey for each stratum for the WA 2015/2016 survey year. Stratum Total population [ABS census 2016] Total RBFL holders (sampling frame) Number RBFL holders completed screening survey Number RBFL holders competed telephone–diary survey Kimberleya 36 392 3 612 212 163 Pilbaraa 61 435 6 513 202 145 Gascoynea,b 9 757 2 331 212 137 Mid-Westa 55 127 7 578 222 149 Wheatbelta 74 530 5 645 209 142 Perth Metropolitan 1 894 943 68 028 1 706 1 189 Peela 133 938 14 146 344 243 South Westa 175 904 18 682 484 363 Great Southerna 60 319 5 475 215 170 Goldfieldsa 56 606 2 399 224 159 Interstate 21 568 249 2 979 231 71 Total 24 127 200 137 388 4 261 2 931 Stratum Total population [ABS census 2016] Total RBFL holders (sampling frame) Number RBFL holders completed screening survey Number RBFL holders competed telephone–diary survey Kimberleya 36 392 3 612 212 163 Pilbaraa 61 435 6 513 202 145 Gascoynea,b 9 757 2 331 212 137 Mid-Westa 55 127 7 578 222 149 Wheatbelta 74 530 5 645 209 142 Perth Metropolitan 1 894 943 68 028 1 706 1 189 Peela 133 938 14 146 344 243 South Westa 175 904 18 682 484 363 Great Southerna 60 319 5 475 215 170 Goldfieldsa 56 606 2 399 224 159 Interstate 21 568 249 2 979 231 71 Total 24 127 200 137 388 4 261 2 931 a Based on Regional Development Commissions. b Ningaloo AMP located offshore from this stratum. View Large Table 3. Sample size (where the primary sampling unit is RBFL holders reporting on all catch to the boat) for screening and telephone–diary survey for each stratum for the WA 2015/2016 survey year. Stratum Total population [ABS census 2016] Total RBFL holders (sampling frame) Number RBFL holders completed screening survey Number RBFL holders competed telephone–diary survey Kimberleya 36 392 3 612 212 163 Pilbaraa 61 435 6 513 202 145 Gascoynea,b 9 757 2 331 212 137 Mid-Westa 55 127 7 578 222 149 Wheatbelta 74 530 5 645 209 142 Perth Metropolitan 1 894 943 68 028 1 706 1 189 Peela 133 938 14 146 344 243 South Westa 175 904 18 682 484 363 Great Southerna 60 319 5 475 215 170 Goldfieldsa 56 606 2 399 224 159 Interstate 21 568 249 2 979 231 71 Total 24 127 200 137 388 4 261 2 931 Stratum Total population [ABS census 2016] Total RBFL holders (sampling frame) Number RBFL holders completed screening survey Number RBFL holders competed telephone–diary survey Kimberleya 36 392 3 612 212 163 Pilbaraa 61 435 6 513 202 145 Gascoynea,b 9 757 2 331 212 137 Mid-Westa 55 127 7 578 222 149 Wheatbelta 74 530 5 645 209 142 Perth Metropolitan 1 894 943 68 028 1 706 1 189 Peela 133 938 14 146 344 243 South Westa 175 904 18 682 484 363 Great Southerna 60 319 5 475 215 170 Goldfieldsa 56 606 2 399 224 159 Interstate 21 568 249 2 979 231 71 Total 24 127 200 137 388 4 261 2 931 a Based on Regional Development Commissions. b Ningaloo AMP located offshore from this stratum. View Large Data from diarists were collected via regular Computer-Assisted Telephone Interviewing with responses entered directly into electronic survey databases. Training was provided to interviewers and diary participants were sent kits containing species identification guides (Department of Fisheries, 2017), fishing location guide (Department of Fisheries, 2011) and diary cards to record key fishing data. Data from the telephone–diary survey were expanded to the RBFL population by using the total number of RBFL holders in each residential stratum divided by the number of RBFL holders sampled from that stratum. This process was completed using the survey (Lumley, 2010) package in the statistical computing package R (Lyle et al., 2010; R Core Team, 2017). Estimates of fishing effort were calculated as boat days (with each separate day of fishing representing 1 day of effort) and is non-directed, as the species being targeted or gear type is not taken into account during analysis. Total catch was estimated as number of fish. This was based on reported catches by the diarist of individual species retained or released by all fishers on the boat (regardless of whether or not they held a RBFL). Each of these estimates also had an associated level of uncertainty [standard error, relative standard error (RSE), and 95% confidence interval]. Overlapping 95% confidence intervals were used to ascertain statistical differences in estimates between survey years. Samples of <30 diarists and RSE >40% were used as cut off points to indicate that estimates may not be robust and were excluded (West et al., 2012; Lyle et al., 2014; Webley et al., 2015). The location of each fishing event was reported using a 10 × 10 nm block. Expanding raw data to population estimates at the finer, AMP scale, followed the same process as for the broader, state-wide and bioregional scales. However, all 10 × 10 nm blocks only partially intersect the NMP due to the parks elongated shape (Figure 1). A proportional approach was therefore used to adjust catch estimates based on the area (% coverage) of each block situated within its boundary. Proportional allocation was applied to retained and released catches within each individual fishing event. Fishing effort was calculated from boat days with each fishing event representing 1 day, it was therefore not possible to proportion boat days to different spatial strata (within blocks) without over-representing the number of boat days. New South Wales NSW also uses a telephone–diary survey of recreational fishing to provide catch and effort estimates at broad spatial (state-wide, regional) and temporal (annual, seasonal) scales. However, while NSW does have a general saltwater licence, the sampling frame for this survey was drawn from the White Pages telephone directories (West et al., 2015) (Tables 1 and 2). The White Pages sample frame was used for this survey to ensure direct comparability with NSW results from the 2000/2001 national survey, which had also used the White Pages as the primary sample frame. The data analysed for this study was from the survey completed in 2013/2014. A screening survey was conducted during the 3 months prior to the 12-month longitudinal survey. In 2013/14, 12 461 households were contacted, of which 9412 households fully responded to the screening survey (West et al., 2015) (Table 4). Households were identified as eligible for telephone–diary survey if any household member aged 5 years or older indicated an intention to fish during the upcoming 12 months. Of the eligible households, 1681 (representing 4433 residents) completed the survey (West et al., 2015). The response rate for the 9412 households that completed the screening survey was >75%, and the rate for completion of the 1681 diary surveys was >80%. A stratified random sample of households was selected from this frame, with each selected listing being assigned to one of ten residential strata based on the Australian Bureau of Statistics’ regional classifications (Statistical Area, Level 4), the sampling rate within each stratum being inversely proportional to a stratum’s resident population size. Table 4. Sample size (where the primary sampling unit is household) by region (ABS strata) for the NSW/ACT recreational fishing survey of 2013/2014. Region (ABS residential stratum) Total number of people (>5 years old) in population Total households in population (sampling frame) Number of households completed screening survey Number of households completed telephone–diary survey Sydney 4 358 514 1 713 988 2 652 298 Hunter 571 626 242 864 1 003 192 Illawarra 403 161 170 498 764 173 Richmond/Tweed 221 026 98 349 703 137 Mid North Coast 319 949 143 945 734 164 Central West/North 358 731 154 988 773 152 North West 108 051 46 963 702 139 South East 202 064 88 608 562 140 South West 248 339 107 975 721 159 ACT 344 060 145 347 798 127 Total 7 135 521 2 913 525 9 412 1 681 Region (ABS residential stratum) Total number of people (>5 years old) in population Total households in population (sampling frame) Number of households completed screening survey Number of households completed telephone–diary survey Sydney 4 358 514 1 713 988 2 652 298 Hunter 571 626 242 864 1 003 192 Illawarra 403 161 170 498 764 173 Richmond/Tweed 221 026 98 349 703 137 Mid North Coast 319 949 143 945 734 164 Central West/North 358 731 154 988 773 152 North West 108 051 46 963 702 139 South East 202 064 88 608 562 140 South West 248 339 107 975 721 159 ACT 344 060 145 347 798 127 Total 7 135 521 2 913 525 9 412 1 681 View Large Table 4. Sample size (where the primary sampling unit is household) by region (ABS strata) for the NSW/ACT recreational fishing survey of 2013/2014. Region (ABS residential stratum) Total number of people (>5 years old) in population Total households in population (sampling frame) Number of households completed screening survey Number of households completed telephone–diary survey Sydney 4 358 514 1 713 988 2 652 298 Hunter 571 626 242 864 1 003 192 Illawarra 403 161 170 498 764 173 Richmond/Tweed 221 026 98 349 703 137 Mid North Coast 319 949 143 945 734 164 Central West/North 358 731 154 988 773 152 North West 108 051 46 963 702 139 South East 202 064 88 608 562 140 South West 248 339 107 975 721 159 ACT 344 060 145 347 798 127 Total 7 135 521 2 913 525 9 412 1 681 Region (ABS residential stratum) Total number of people (>5 years old) in population Total households in population (sampling frame) Number of households completed screening survey Number of households completed telephone–diary survey Sydney 4 358 514 1 713 988 2 652 298 Hunter 571 626 242 864 1 003 192 Illawarra 403 161 170 498 764 173 Richmond/Tweed 221 026 98 349 703 137 Mid North Coast 319 949 143 945 734 164 Central West/North 358 731 154 988 773 152 North West 108 051 46 963 702 139 South East 202 064 88 608 562 140 South West 248 339 107 975 721 159 ACT 344 060 145 347 798 127 Total 7 135 521 2 913 525 9 412 1 681 View Large Fishing activity was monitored via diary entries completed by the survey participants as well as by follow up telephone interviews by trained interviewers and, like WA, all participants were sent a survey kit. Interviewers collected detailed information about each fishing activity (event) to enable classification of the fishing site using a GIS coding system (i.e. latitude and longitude). Depending on the types of fishing location, different information was obtained by interviewers to determine if the fishing was estuarine (within bays and rivers), inshore (<5 km; 2.7 nm from the coast) or offshore (>5 km; 2.7 nm from the coast) (West et al., 2015). This inshore/offshore classification, along with GIS coding, enabled approximate identification of fishing events that occurred within the HMP and, more broadly, within Commonwealth waters. For HMP, fishing events were approximated to have occurred if they took place within the offshore waters west 153°42′E and within 32°01′S 32°41′S. Expansion of samples to population estimates was undertaken by calibrating against ABS population bench marks and was implemented for residents in each residential stratum, taking account of household and person-based demographics and various biases such as avidity and “drop-in” and “drop-outs” to the fishery (West et al., 2015). Like WA, this expansion process was completed using the statistical computing package R (R Core Team, 2017) using the survey (Lumley, 2010) and resurvey (Lyle et al., 2010) packages. Disaggregation of the NSW telephone–diary survey to provide fine-scale estimates of fishing effort and catch at the scale of the HMP followed the same process of expanding the raw data to population estimates of fishing effort (in fisher days) and total catch (in number of individuals). Fisher days are defined as the total number of person days spent fishing. Each of these estimates also had an associated level of uncertainty (standard error). Results Western Australia Australian Fisheries Management Authority Three of the nine species of interest to AFMA had catches which met RSE and sample size reporting criteria used in WA including gummy sharks (M. antarcticus and M. stevensi), southern bluefin tuna (T. maccoyii), and yellowfin tuna (T. albacares) (Figure 2). Retained catch for gummy sharks was highest in 2011/2012 (1734 ± 639) although there was no significant difference in the catches between survey years. Retained catches for southern bluefin tuna were significantly different between all survey years, with the highest retained catches occurring in 2015/2016 (2009 ± 344). Retained catches for yellowfin tuna were also significantly higher in 2011/2012 (1500 ± 282) when compared to 2015/2016 (442 ± 101). Figure 2. View largeDownload slide Estimated retained, released, and total catch (by numbers and with associated standard errors) from boat-based recreational fishers for the AFMA species of interest state-wide (WA) in each survey year. Estimates were not reported for some survey years due to sample size <=30 and/or RSE >40%. Figure 2. View largeDownload slide Estimated retained, released, and total catch (by numbers and with associated standard errors) from boat-based recreational fishers for the AFMA species of interest state-wide (WA) in each survey year. Estimates were not reported for some survey years due to sample size <=30 and/or RSE >40%. Ningaloo Marine Park Estimates of fishing effort (boat days) in NMP were highest in 2011/2012 (21 160; ± 2174) and lowest in 2015/2016 (14 245; ± 1831) (Table 5). There was no significant difference between survey years. The number of different species caught by boat-based recreational fishers in each survey year ranged from a maximum of 111 in 2011/2012 to a minimum of 99 in 2015/2016. Table 5. Total estimated fishing effort (boat days) and standard error for NMP (WA) in each survey year as well as number of species caught. Survey year 2011/2012 2013/2014 2015/2016 Sample size (n) 154 150 127 Boat days 21 160 (2 174) 17 379 (2 041) 14 245 (1 831) Number of species 111 102 99 Survey year 2011/2012 2013/2014 2015/2016 Sample size (n) 154 150 127 Boat days 21 160 (2 174) 17 379 (2 041) 14 245 (1 831) Number of species 111 102 99 RSE <40% for all measures of fishing effort. View Large Table 5. Total estimated fishing effort (boat days) and standard error for NMP (WA) in each survey year as well as number of species caught. Survey year 2011/2012 2013/2014 2015/2016 Sample size (n) 154 150 127 Boat days 21 160 (2 174) 17 379 (2 041) 14 245 (1 831) Number of species 111 102 99 Survey year 2011/2012 2013/2014 2015/2016 Sample size (n) 154 150 127 Boat days 21 160 (2 174) 17 379 (2 041) 14 245 (1 831) Number of species 111 102 99 RSE <40% for all measures of fishing effort. View Large Estimated total catch (all species) was highest in 2011/2012 (28 632 ± 3837) (Figure 3). The numbers of fish retained and released were also highest in 2011/2012 with 12 941 (±1867) and 15 692 (±2359), respectively. There was no significant difference in total, retained or released catches between survey years except the numbers of fish released in 2011/2012, which were significantly higher than in 2015/2016. The percentage of catch released by fishers was greater than those retained in 2011/2012 (54%) and 2013/2014 (57%). Figure 3. View largeDownload slide Estimated retained, released, and total catch (by numbers and with associated standard errors) from boat-based recreational fishers for all species in the NMP (WA) in each survey year. Sample size >30 and RSE <40% for all estimates. Figure 3. View largeDownload slide Estimated retained, released, and total catch (by numbers and with associated standard errors) from boat-based recreational fishers for all species in the NMP (WA) in each survey year. Sample size >30 and RSE <40% for all estimates. Only six species caught in NMP met the reporting criteria. Catches of spangled emperor (Lethrinus nebulosus), chinaman rockcod (Epinephelus rivulatus), and redthroat emperor (Lethrinus miniatus) were the highest in each survey year, with total catches exceeding 2000 fish in the majority of survey years (Figure 4). Rankin cod (Epinephelus multinotatus), goldband snapper (Pristipomoides multidens), Spanish mackerel (Scomberomorus commerson), and red emperor (Lutjanus sebae) all had total catches of <900 fish in each survey year. Figure 4. View largeDownload slide Estimated retained, released, and total catch (by numbers and with associated standard errors) from boat-based recreational fishers for key species in the NMP (WA) in each survey year. Estimates were not reported for some survey years due to sample size <=30 and/or RSE >40%. Figure 4. View largeDownload slide Estimated retained, released, and total catch (by numbers and with associated standard errors) from boat-based recreational fishers for key species in the NMP (WA) in each survey year. Estimates were not reported for some survey years due to sample size <=30 and/or RSE >40%. Comparisons of catch between survey years were possible for all species except goldband snapper (P. multidens). There was no significant difference between the estimated catch retained by recreational fishers in each survey year, except for spangled emperor (2011/2012 was significantly higher than 2015/2016) and Rankin cod (E. multinotatus) (2011/2012 was significantly higher than 2013/2014). There was no significant difference in released and total catches between survey years for each species. New South Wales Australian Fisheries Management Authority Five of the nine species of interest to AFMA had catches recorded within NSW waters: bluespotted and sand flathead species grouping (P. caeruleopunctatus and P. bassensis) - which will be referred to as bluespotted flathead hereafter, gummy shark (M. antarcticus), striped marlin (T. audax), school shark (G. galeus), and yellowfin tuna (T. albacares) (Figure 5). Among these, the bluespotted and sand flathead grouping had the highest total catch (962 892 ± 181 433 fish caught) and striped marlin had the lowest catch (163 ± 162). The spatially explicit nature of the data allowed for easy differentiation among estuarine, nearshore, and offshore waters. In particular flatheads were caught in estuarine waters in large numbers. Across NSW, 60 households were sampled who fished in Commonwealth waters (i.e. offshore) compared to 436 households sampled who fished within inshore waters. Figure 5. View largeDownload slide Total estimated retained, released, and total recreational state-wide (NSW/ACT) catch (by numbers) and associated standard errors for the AFMA species of interest during 2013/2014. Figure 5. View largeDownload slide Total estimated retained, released, and total recreational state-wide (NSW/ACT) catch (by numbers) and associated standard errors for the AFMA species of interest during 2013/2014. Hunter Marine Park Estimates of fishing effort (fisher days) were higher in the inshore waters [37 426 (±8557)] adjacent to the HMP compared to fishing effort within the approximate bounds of the HMP itself [1901 (±1442)] (Table 6). The total number of species caught by fishers within the HMP was 9 (Table 7). Table 6. Total estimated recreational fishing effort and standard error for Hunter Commonwealth Marine Reserve in NSW/ACT (HCMR) during 2013/2014. Waterbody Inshore waters HMP (offshore waters) Sample size (households) 38 3 Fisher days 37 426 (8 557) 1 901 (1 442) Waterbody Inshore waters HMP (offshore waters) Sample size (households) 38 3 Fisher days 37 426 (8 557) 1 901 (1 442) The HCMR is approximately located within offshore waters (>5 km from coastline/mainland). The relative inshore (<5 km from coastline/mainland) recreational fishing effort proximal to the HCMR is also depicted. View Large Table 6. Total estimated recreational fishing effort and standard error for Hunter Commonwealth Marine Reserve in NSW/ACT (HCMR) during 2013/2014. Waterbody Inshore waters HMP (offshore waters) Sample size (households) 38 3 Fisher days 37 426 (8 557) 1 901 (1 442) Waterbody Inshore waters HMP (offshore waters) Sample size (households) 38 3 Fisher days 37 426 (8 557) 1 901 (1 442) The HCMR is approximately located within offshore waters (>5 km from coastline/mainland). The relative inshore (<5 km from coastline/mainland) recreational fishing effort proximal to the HCMR is also depicted. View Large Table 7. Sample size of fishing parties that were successful in capturing fish and number of species caught by recreational fishers in the Hunter Commonwealth Marine Reserve in NSW/ACT (HCMR) during 2013–2014. Waterbody Inshore waters HMP (offshore waters) Sample size (households) 26 3 Number species 34 9 Waterbody Inshore waters HMP (offshore waters) Sample size (households) 26 3 Number species 34 9 The HCMR is approximately located within offshore waters (>5 km from coastline/mainland). The relative sample size and number of species caught within inshore waters proximal to the HCMR (<5 km from coastline/mainland) is also depicted. While 38 households were sampled inshore (see Table 6) only 26 households caught any animals. View Large Table 7. Sample size of fishing parties that were successful in capturing fish and number of species caught by recreational fishers in the Hunter Commonwealth Marine Reserve in NSW/ACT (HCMR) during 2013–2014. Waterbody Inshore waters HMP (offshore waters) Sample size (households) 26 3 Number species 34 9 Waterbody Inshore waters HMP (offshore waters) Sample size (households) 26 3 Number species 34 9 The HCMR is approximately located within offshore waters (>5 km from coastline/mainland). The relative sample size and number of species caught within inshore waters proximal to the HCMR (<5 km from coastline/mainland) is also depicted. While 38 households were sampled inshore (see Table 6) only 26 households caught any animals. View Large Total estimated catch across species within the HMP is shown in Figure 6. Catches of bluespotted flathead species grouping (P. caeruleopunctatus and P. bassensis), red rockcod (Scorpaena jacksoniensis), and blue mackerel (Scomber australasicus) were the highest within the HMP exceeding 4000 fish (Figure 6). Figure 6. View largeDownload slide Total estimated retained, released, and total catch (by numbers and with associated standard errors) from boat-based recreational fishers for all key species in the Hunter Commonwealth Marine Reserve (NSW/ACT) during 2013/2014. Figure 6. View largeDownload slide Total estimated retained, released, and total catch (by numbers and with associated standard errors) from boat-based recreational fishers for all key species in the Hunter Commonwealth Marine Reserve (NSW/ACT) during 2013/2014. Discussion Since the first national survey on recreational fishing in 2000/2001 (Lyle et al., 2002; Henry and Lyle, 2003) both WA and NSW have maintained the basic methodology with a screening survey followed by 12-month longitudinal telephone–diary survey. A key difference in methodology is that WA uses a RBFL database as the sampling frame, while NSW uses the White Pages telephone directory. NSW has started supplementing this data frame with their general saltwater fishing licence holders (Table 1) but this survey design is still being developed. Post hoc analysis of the existing WA and NSW survey databases provided some useful information on both fishing effort and catch for Commonwealth waters. For the nine species of interest to AFMA, robust catch estimates for six were possible, with estimates for two species, gummy sharks (M. antarcticus) and yellowfin tuna (T. albacares) available in both states. The spatially explicit classification used in NSW (i.e. estuarine, nearshore, and offshore) approximates the boundary between state and Commonwealth waters and facilitated an additional breakdown of catches between state and Commonwealth jurisdictions. In WA, the 10 × 10 nm blocks overlayed both state and Commonwealth waters and we could only provide catch estimates from combined jurisdictions. Disaggregation of state-based data to our case study AMP scale showed some potential—with various caveats—for providing park scale estimates of fishing effort and total catch. Only the WA survey was of sufficient statistical power to provide robust estimations of effort and catch for commonly caught species for NMP. In some instances, such as for numbers of fish released, these results were sensitive enough to show some significant differences between (annual) survey periods. Our analysis showed no significant difference between years for effort and catch at Ningaloo, but in context to the regional (Gascoyne Coast) and state-wide scales, similar patterns emerged. For the Gascoyne Coast, the highest yearly number of boat days occurred in 2011/2012 (58 123 boat days ± 3672), followed by 2013/2014 (53 832 ± 3603) and 2015/2016 (43 237 ± 3152). The relative percentage of effort in the Gascoyne Coast compared to the state-wide assessment was also highly consistent among years, with 13, 14, and 12% of the total state effort. State wide, the highest year of effort was again 2011/2012 (439 029 boat days ± 11 160) which differed to both 2013/2014 (383 107 ± 12 385) and 2015/2016 (370 368 ± 11 567) (Ryan et al., 2017). A similar temporal pattern occurred with harvest of the top 10 demersal species (or groupings) in the Gascoyne Coast which was highest in 2011/2012 at 127–159 (95% CI) tonnes before declining to 88–115t in 2013/2014 and then remaining steady at 87–118t in 2015/2016. Estimates produced from seasonal disaggregation of data were not robust. However, seasonal patterns of fishing effort for NMP are clearly evident at a broader bioregional scale (Ryan et al., 2017). For the HMP the large variances of estimates meant the data was un-reliable above presence/absence levels. This difference in granularity was probably due to regional oversampling applied by the Western Australian’s to their survey. This resulted in a much larger sample as a proportion of the population collected within the Regional Commission Boundary (Gascoyne) for NMP compared to the Hunter sampling bio-region which contained the HMP. However, even in WA, data could only be disaggregated to the park scale, with distributions within the park (i.e. to specific zones) unable to be calculated. This fine-scale investigation of spatial use by fishers of zoning plans within marine parks would probably be better served by dedicated on-site studies. The recently revised Commonwealth Harvest Strategy Policy states that harvest strategies will account for all known sources of fishing mortality on a stock, including recreational and Indigenous fishing; and fishing under the management of another jurisdiction (Agriculture and Water Resources, 2018). This is challenging due to cross-jurisdictional movements of fish and fishers as well as the need to survey for recreational fishers, however consideration of non-commercial mortality may be important for some species. Many recreational fishers in Australia have consumption as an important motivation and for some species catch can exceed the take of the commercial fishery (Zischke et al., 2012; Lyle et al., 2014; Giri and Hall, 2015). It is interesting to note that both states develop metrics for catch, which combines harvest with released animals, as release mortality can be variable based on fisher skill, gear type, species, and depth of capture (Muoneke and Childress, 1994; Cooke and Philipp, 2004; Skomal, 2007; Brownscombe et al., 2014,, 2017; Shertzer et al., 2018). For consideration of total non-commercial mortality both species and/or regional MRF estimates may require adjustments to account for the proportion of released fish that die. Long-lived and historically overfished shark species (Last and Stevens, 2009) such as gummy shark (M. antarcticus) were captured in both states. Both WA and NSW border the centre of the shark’s distribution and the associated recreational fishery for this species. Detection of fish caught within RSE limits by state-wide assessment methods—which are a “broad brush” approach—suggest considerable catches. When combined with catches from other states (i.e. South Australia, Victoria, and Tasmania), and the relatively small tonnage considered sustainable (1774t) (AFMA, 2019), the recreational take may be a considerable. Recreational catches of gummy sharks may be a fishery that requires targeted cross-jurisdictional studies to estimate catch across the full distribution of this species. Other shark species in Australia have recognized cross-jurisdictional management requirements, for instance the grey nurse shark (Carcharias taurus), which is the subject of specific MPA and fisheries regulations across multiple states on Australia’s east coast (Lynch et al., 2013). Southern bluefin tuna (T. maccoyii) is another species moving across jurisdictions that is targeted by both commercial and recreational fishers, and provides an example of a collaboration between the Commonwealth and states to obtain catch data (Moore et al., 2015). Southern bluefin tuna caught in WA are small animals (2–3 kg) compared to those captured in the eastern states (up to 160 kg and commonly to 100 kg), hence they will only contribute marginally to the 250 tonne set aside to non-commercial mortality. In jurisdictions such as Victoria and Tasmania the take of this species by the recreational sector may be considerable, with Victoria’s catch estimated at 240 tonne (±31) over a 5-month season in 2011 (Green et al., 2012). In both the cases of shark and tuna MRF the jurisdictional extent of catch in Commonwealth waters may be relevant to issues such as resource allocation with the commercial sector, in particular for locations where the combined multi-sectorial catch might be substantial. For species subject to international management, recreational catch estimates must also be consistent, accurate, and reliable to ensure that recreational fishers do not exceed sustainable limits, especially if these species become popular targets with increased fishing effort (Moore et al., 2015; Kristianson, 2018). How to target any specific recreational fishing surveys is somewhat problematic in Australia where there are highly diverse eco-systems and landed catches tend to be small, but comprise many species (Newman et al., 2018). Recent surveys in WA have recorded a diverse range of species/taxa being caught including scale-fish (182 species/taxa), elasmobranchs (18), crustaceans (7), and molluscs (5) (Ryan et al., 2017). Sample sizes for state-based assessments are determined to ensure robust estimates are obtained for key species, thus statistical power is not equivalent among species. Consequently, there are difficulties in providing estimates for those species that are less commonly caught, though grouping of catches by family may provide some opportunities for catch estimates at a coarser level. This also poses a challenge for agencies that rely on external data, where their information requirements exceed the objectives, and available budget, of the state-based survey. However, this study has revealed that state datasets are sensitive enough to provide robust estimates for some species of interest to AFMA, and those species not reported (swordfish, blue eye trevalla, pink ling, and eastern gemfish) are either generally caught in areas outside of the scoping project states, are deep sea species or capture may be just be a rare event. If species are rarely caught and true catch is low the question then can be posed: does the recreational catch really matter if it is an insignificant component of total mortality of a stock? State-based surveys, if well-resourced, may act as a form of sensitivity analysis for recreational catch in Commonwealth waters for some species, thereby illustrating those that may currently be caught in numbers significant enough to be of concern from a sustainability perspective. As recreational effort and catch is also variable over time, repeat surveys could also detect those species that emerge or decline as targets for the recreational fishery in the future. Statistical power in the design also played an important role in the functionality of the results obtained for the AMPs. In those WA regional strata with low residential populations (i.e. Gascoyne, Kimberley) over-sampling ensured that fishing activity was reported with sufficient sample sizes to produce robust estimates at the bioregional scale. This differed from the NSW approach which was proportional to the strata region’s population size. Both in WA and more generally in Australia recreational fishing rates in regional, low-population areas are high when compared to state-wide averages (Henry and Lyle, 2003; Lyle et al., 2014; Ryan et al., 2017). The collection of more samples than proportional representation to population in the Gascoyne may help explain the difference in statistical power after the disaggregation of data at the NMP when compared to HMP. For the Gascoyne Regional Development Commission area, of which NMP sits offshore, 2331 residents held RBFL, which is 23.6% of the total population with an estimated 1914 (83%) fishing at least once in 2015/2016 (Table 3). Of this population of 2331 Gascoyne RBFL holders 137 completed the telephone–diary survey or nearly 6% of the sample frame. This resulted in between 127 and 154 samples collected for NMP, compared to only three for the HMP in NSW (or 41 if the inshore data was added). More generally outside of the Metropolitan strata around Perth, WA screened and sampled ∼10% of all RBFL holders. Estimates of spatial use (i.e. bioregion, marine park) can include fishers from any residential strata, and although RBFL holders do travel throughout WA to fish, >80% of RBFL holders that fished in the Gascoyne bioregion resided within the Gascoyne Regional Development Commission boundary (Ryan et al., 2013,, 2015). Another potential reason for the ability of the WA data to provide more robust estimates of the NMP compared to the NSW estimates at HMP may be the unusual nature of the RFBL, which was used to generate the WA sampling frame. Unlike all other licencing in Australia the RFBL is a form of communal licence where multiple, non-licenced fishers on board a vessel can fish to the boat bag limits of the persons which hold RFBLs. The RFBL holder involved in the survey thus reports on the whole of boat catch and not just their individual success or otherwise. In contrast to WA, very few of the estimates generated for the HMP were of adequate resolution to provide robust information. The high uncertainty associated with most HMP estimates indicate either a need for more targeted sampling effort within this area, relatively low levels of recreational fishing or fishing occurring outside of the sampling frame. Of these the first and last options seem most probable. Across NSW, samples that included inshore fishing activity were generally much greater than offshore (>5 km off coastline) (West et al., 2015) but as a proportion of population fewer samples were taken in NSW, particularly in non-metropolitan areas, compared to WA. Other aspects of the sampling frames may have also contributed, as anecdotally, southern bluefin tuna are caught recreationally in NSW but were not recorded in our case study. In NSW, access to this popular game fishing species can be limited to larger vessels due to the long distance offshore of their preferred oceanographic currents and short fishing season (Moore et al., 2015). These larger vessels are often charter or private game fishing vessels involved in tournaments and data from these fishers are collected separately to the state-wide surveys via logbooks and offshore game fishing tournament catch records. Technically, both charter and game fishers are included in the NSW state survey if a respondent participated in a tournament or charter fishing. However, tournaments are a localized concentration of fishing effort (Flynn et al., 2018) by a small subset of the general fishing population (Griffiths et al., 2010) which like charter fishing have low probability of being captured in state-wide surveys. For HMP, tournament data collected between 1994 and 2013 (Ghosn et al., 2015) showed extensive effort within the park and charter logbooks also contain more information of the offshore recreational fisheries (Lowry and Murphy, 2003). A limitation of disaggregating data for NMP was that spatial heterogeneity within the 10 × 10 nm blocks used in the WA survey could not be accounted for while proportioning catches. In our proportional analysis of these blocks, species distribution and fishing are assumed to be evenly distributed, even though the heterogeneous distribution of recreational fishers are well-documented (Lynch, 2006,, 2008; Rufino et al., 2006; Flynn et al., 2018). Species distributions are also not uniform (i.e. different species respond to different habitats) and fisher behaviour can reflect their understanding of where fish occur as well as access to fishing locations. As expected, estimated catches for all species in the NMP were lower when proportional allocation was applied based on the area (% coverage) of each block situated within its boundary. The distribution of species is also important to consider and may also affect the number of species recorded in NMP during each survey period. For example, chinaman rockcod (E. rivulatus) is a shallow water species which is found in water depths only up to 150 m and therefore are not likely to be caught in the deeper waters of the NMP which extend to 500 m. Accurate mapping of effort and catch rates for defining spatial “hotspots” is ideal for identifying high priority areas for fishery management (Aidoo et al., 2015), conservation (Stelzenmüller et al., 2004; Lynch, 2006), and shifts in distribution of fishing effort over space and time (Ciannelli et al., 2008; Lynch, 2014; Aidoo et al., 2016). The analysis undertaken in this paper provided a pilot study for a larger WA state-wide application of disaggregation techniques to investigate small-scale patterns of marine recreational fishing using one of several available methods. As NMP is 1 of 22 AMP in Commonwealth waters surrounding WA, the exploration of new methods and support for this state-wide survey may be an effective way to provide more detailed information on a wider suite of AMPs. Generally methods for determining catch and effort were based on the previous national baseline (Henry and Lyle, 2003) and were similar between states in their use of telephone interviews supported by fishers keeping diaries of their experiences. One area where they strongly differed was in the reporting of spatial data for each fishing event; 10 nm blocks in WA compared to a classification of the fishing sites using a GIS coding system (i.e. latitude and longitude) in NSW. An increased focus on collection of data on fishing location, facilitated by increased uptake of GPS technology, could assist future disaggregation of survey data to spatial scales relevant for both Commonwealth fisheries and AMP managers. Collection of data on broad ecosystem or habitat type and depth fished may also help to identify locations fished. However, the increased burden on respondents must be carefully considered against the benefits of obtaining these additional data. It can also be difficult for all fishers to understand habitat categories and collect this data consistently. Fishers in the NSW survey identified broad ecosystems in which they fished (i.e. estuarine, inshore, and offshore), as did fishers in the WA survey (i.e. pelagic, offshore, inshore, nearshore, estuary, and freshwater). In WA however these definitions did not, however, align with demarcate between state and Commonwealth waters. State-based surveys are primarily designed to provide robust estimates (with acceptable sample size and precision) at broad spatial and temporal scales. For species caught less frequently, fewer diarists will reported data and hence estimates of catch will often have high uncertainty; similarly disaggregated data will lose resolution if sample sizes are set to only have sufficient power to detect changes at regional scales. The state-wide surveys may however provide a sensitivity test for identifying species or locations where MRF may be of interest to the Commonwealth. However, increased or more targeted resourcing within bio-regional strata of interest may be needed to raise sample sizes and achieve adequate statistical power. Over-sampling of regional areas in the WA state-based survey shows the success of this approach. Further harmonization of state-wide survey datasets with charter-fishing logbook data, tournament data and size frequency information would also allow for better understandings of fish mortality and effort distributions. Alternatively, on-site interviews can be used as another mechanism to answer specific management questions on AMP. Recreational fisheries are of great importance to the states, with 16 state-wide assessments having occurred or in progress since the national survey (Table 8) and in most states and all territories multiple surveys have been held in an effort to commence or develop time-series data. An issue for Commonwealth fisheries from the existing data sources is that state-based recreational fishing surveys are not aligned temporally between states and have varying years between repeat surveys. Catch estimates for species that straddle multiple jurisdictions are more readily compared when simultaneous data collection occurs across the full distributional range. Surveys that are not synchronized complicate any inferences for management as catch and subsequent mortality of cohorts will be temporally and, if the species migrate, spatially confounded. The lack of a coordinated programme of state-wide surveys makes it very difficult to provide reliable catch estimates for stocks or species. Table 8. Completed and in progress state-wide assessments for all states and territories—with the exception of Victoria which conducts bio-regional assessments—since the national survey in 2000/2001 (Henry and Lyle, 2003). State Period of survey Reference South Australia 2007/2008 (November 2007–October 2008) Jones (2009) South Australia 2013/2014 (December 2013–November 2014) Giri and Hall (2015) Tasmania 2007/2008 (December 2007–November 2008) Lyle et al. (2009) Tasmania 2012/2013 (November 2012–October 2013) Lyle et al. (2014) Tasmania 2017/2018 In progress NSW 2013/2014 (June 2013–May 2014) West et al. (2015) NSW 2017/2018 In progress Northern Territory 2009/2010 (April 2009–March 2010) West et al. (2012) Northern Territory 2017/2018 In progress Queensland 2010/2011 (October 2010–September 2011) Taylor et al. (2012) Queensland 2013/2014 (November 2013–October 2014) Webley et al. (2015) Queensland 2019/2020 In progress WA 2011/2012 (March 2011–February 2012) Ryan et al. (2013) WA 2013/2014 (May 2013–April 2014) Ryan et al. (2015) WA 2015/2016 (September 2015–August 2016) Ryan et al. (2017) WA 2017/2018 In progress State Period of survey Reference South Australia 2007/2008 (November 2007–October 2008) Jones (2009) South Australia 2013/2014 (December 2013–November 2014) Giri and Hall (2015) Tasmania 2007/2008 (December 2007–November 2008) Lyle et al. (2009) Tasmania 2012/2013 (November 2012–October 2013) Lyle et al. (2014) Tasmania 2017/2018 In progress NSW 2013/2014 (June 2013–May 2014) West et al. (2015) NSW 2017/2018 In progress Northern Territory 2009/2010 (April 2009–March 2010) West et al. (2012) Northern Territory 2017/2018 In progress Queensland 2010/2011 (October 2010–September 2011) Taylor et al. (2012) Queensland 2013/2014 (November 2013–October 2014) Webley et al. (2015) Queensland 2019/2020 In progress WA 2011/2012 (March 2011–February 2012) Ryan et al. (2013) WA 2013/2014 (May 2013–April 2014) Ryan et al. (2015) WA 2015/2016 (September 2015–August 2016) Ryan et al. (2017) WA 2017/2018 In progress Months are all inclusive (1–30/1). The Australian Capital Territory is included in statistics reported for NSW and WA reports are only for boat based fishing. View Large Table 8. Completed and in progress state-wide assessments for all states and territories—with the exception of Victoria which conducts bio-regional assessments—since the national survey in 2000/2001 (Henry and Lyle, 2003). State Period of survey Reference South Australia 2007/2008 (November 2007–October 2008) Jones (2009) South Australia 2013/2014 (December 2013–November 2014) Giri and Hall (2015) Tasmania 2007/2008 (December 2007–November 2008) Lyle et al. (2009) Tasmania 2012/2013 (November 2012–October 2013) Lyle et al. (2014) Tasmania 2017/2018 In progress NSW 2013/2014 (June 2013–May 2014) West et al. (2015) NSW 2017/2018 In progress Northern Territory 2009/2010 (April 2009–March 2010) West et al. (2012) Northern Territory 2017/2018 In progress Queensland 2010/2011 (October 2010–September 2011) Taylor et al. (2012) Queensland 2013/2014 (November 2013–October 2014) Webley et al. (2015) Queensland 2019/2020 In progress WA 2011/2012 (March 2011–February 2012) Ryan et al. (2013) WA 2013/2014 (May 2013–April 2014) Ryan et al. (2015) WA 2015/2016 (September 2015–August 2016) Ryan et al. (2017) WA 2017/2018 In progress State Period of survey Reference South Australia 2007/2008 (November 2007–October 2008) Jones (2009) South Australia 2013/2014 (December 2013–November 2014) Giri and Hall (2015) Tasmania 2007/2008 (December 2007–November 2008) Lyle et al. (2009) Tasmania 2012/2013 (November 2012–October 2013) Lyle et al. (2014) Tasmania 2017/2018 In progress NSW 2013/2014 (June 2013–May 2014) West et al. (2015) NSW 2017/2018 In progress Northern Territory 2009/2010 (April 2009–March 2010) West et al. (2012) Northern Territory 2017/2018 In progress Queensland 2010/2011 (October 2010–September 2011) Taylor et al. (2012) Queensland 2013/2014 (November 2013–October 2014) Webley et al. (2015) Queensland 2019/2020 In progress WA 2011/2012 (March 2011–February 2012) Ryan et al. (2013) WA 2013/2014 (May 2013–April 2014) Ryan et al. (2015) WA 2015/2016 (September 2015–August 2016) Ryan et al. (2017) WA 2017/2018 In progress Months are all inclusive (1–30/1). The Australian Capital Territory is included in statistics reported for NSW and WA reports are only for boat based fishing. View Large Another issue for alignment is an understanding of cross-jurisdictional fishing. For example, in jurisdictions where recreational fishing is predominately inter-state visitors, surveys will omit a large proportion of the catch unless other jurisdictions collect data at similar times. Recreational charter boat fishing data is also not well included in the state-wide assessments, with separate log book programmes in WA, SA, and NSW and voluntary programmes in the past for TAS and VIC (Moore et al., 2015). MRFs in Commonwealth waters share resources with many other sectors and industries, such as commercial fishers, charter boat, and tourism operators (Kearney, 2001; Campbell et al., 2003; Collins, 2008) while individual states separately manage and monitor MRF on behalf of the Commonwealth. In this complex, cross-jurisdictional and contested setting there is a need for accurate, precise, and consistent information on MRF (Kristianson, 2018). State-based surveys, while not perfect in addressing the research needs for the Commonwealth, still go a long way in providing the required information. They also demonstrate a well-established framework of expertise, data collection, sampling design, analysis, and innovation across Australian states and particularly for commercial fisheries a strong partnership approach already exists between the Commonwealth and the states. Building a stand-alone national project to assess offshore recreational fisheries from scratch may not be necessary, with all states already conducting surveys. A number of barriers remain, however, to coordination which include: (i) the lack of a common, easily targeted data frame—such as licencing—which has high response rates especially as White Pages data frames are declining in usefulness, (ii) the wide range of agencies and decisions makers to coordinate to reach design and other agreements, and (iii) resourcing. One significant advantage of surveying fishers in Commonwealth waters, however, is the lack of land-based catch so there may be opportunities of using proxy sample frames such as the national database of recreational boat registrations, which could include regional oversampling in areas of interest. Though in experimental work issues with boat ownership and identification of fishing vs. non-fishing trips still made it difficult to use registrations as a proxy data frame (Wise and Fletcher, 2013), and oversampling may still not capture rare events and will increase costs. One aspect of the work that quickly became evident was the ongoing usefulness and exceptional influence of the only nationally coordinated recreational and indigenous fishery assessment conducted to date, almost 20 years ago (Henry and Lyle, 2003). Snapshots of stock demographics can be useful, but management benefits most from continual data collection and monitoring and a regular periodic repeat of a national survey, or at least simultaneous state-wide assessments, would be invaluable for both management of fisheries and the now extensive AMP network. Further development and coordination of recreational research would be highly beneficial, recognizing that the majority of recreational fishers will travel through state waters, that state managers will be asking similar questions to their Commonwealth colleagues and knowing that activities are often governed by complementary management arrangements. The Commonwealth also now has specific requirements the new National Harvest Strategy for better estimates of harvest by MRF catch. One current programme that looks to coordinate activity across at least the southern states of Australia to survey recreational fisher effort, catch, release, and harvest of southern bluefin tuna and other large tuna and billfish may provide a template for both coordination and resourcing (Moore et al., 2015). More generally for all MRF harvested species, national investment in coordination and capability development and adoption of best practise innovations from across the states to conduct a simultaneous national survey on a long cycle (i.e. 5 yearly) may be a way forward to align state surveys, while allowing states to continue to innovate. The commencement of a revitalized national survey in 2020/2021 would be appropriate, being exactly 20 years from the original baseline. Acknowledgements The authors are grateful to Agata Zabolotny, Shane Walters, Brent Wise, and Gary Jackson at the Department of Primary Industries and Regional Development in Western Australia for their assistance and review of the manuscript. The authors would also like to thank Jeff Murphy from the Department of Primary Industries in NSW for comments. The authors also wish to thank the relevant staff at Parks Australia and AFMA, with particular mention to Amanda Richley, David Logan, and Cath Samson. The findings and views expressed are those of the authors and do not necessarily represent the view of Parks Australia, the Director of National Parks, AFMA or the Australian Government. Funding This work was undertaken for the Marine Biodiversity Hub, a collaborative partnership supported through funding from the Australian Government's National Environmental Science Program (Grant E4 Valuing recreational fishing in Commonwealth waters). References ABS. 2018 . Australian Demographic Statistics . Australian Bureau of Statistics , Canberra, Australia. AFMA. 2018 . Commonwealth Southern Bluefin Tuna Catch Limit Set for 2018 Season. https://www.afma.gov.au/commonwealth-southern-bluefin-tuna-catch-limit-set-2018-season (last accessed 19 May 2019). AFMA. 2019 . Gummy shark. https://www.afma.gov.au/fisheries-management/species/gummy-shark (last accessed 20 May 2019). Agriculture and Water Resources. 2018 . Commonwealth Fisheries Harvest Strategy Policy. Agriculture and Water Resources, 21 pp. Aidoo E. N. , Mueller U. , Goovaerts P. , Hyndes G. A. 2015 . Evaluation of geostatistical estimators and their applicability to characterise the spatial patterns of recreational fishing catch rates . Fisheries Research , 168 : 20 – 32 . Google Scholar Crossref Search ADS PubMed Aidoo E. N. , Mueller U. , Hyndes G. A. , Ryan K. L. 2016 . The effects of measurement uncertainty on spatial characterisation of recreational fishing catch rates . Fisheries Research , 181 : 1 – 13 . Google Scholar Crossref Search ADS Arlinghaus R. , Tillner R. , Bork M. 2015 . Explaining participation rates in recreational fishing across industrialised countries . Fisheries Management and Ecology , 22 : 45 – 55 . Google Scholar Crossref Search ADS Brownscombe J. W. , Bower S. D. , Bowden W. , Nowell L. , Midwood J. D. , Johnson N. , Cooke S. J. 2014 . Canadian recreational fisheries: 35 years of social, biological, and economic dynamics from a national survey . Fisheries , 39 : 251 – 260 . Google Scholar Crossref Search ADS Brownscombe J. W. , Danylchuk A. J. , Chapman J. M. , Gutowsky L. F. G. , Cooke S. J. 2017 . Best practices for catch-and-release recreational fisheries—angling tools and tactics . Fisheries Research , 186 : 693 – 705 . Google Scholar Crossref Search ADS Brownscombe J. W. , Hyder K. , Potts W. , Wilson K. L. , Pope K. L. , Danylchuk A. J. , Cooke S. J. et al. 2019 . The future of recreational fisheries: advances in science, monitoring, management, and practice . Fisheries Research , 211 : 247 – 255 . Google Scholar Crossref Search ADS Buxton C. D. , Cochrane P. 2015 . Commonwealth Marine Reserves Review; Report of the Bioregional Advisory Panel. Department of the Environment, Canberra. 341 pp. Camp E. V. , Ahrens R. N. M. , Crandall C. , Lorenzen K. 2018 . Angler travel distances: implications for spatial approaches to marine recreational fisheries governance . Marine Policy , 87 : 263 – 274 . Google Scholar Crossref Search ADS Campbell R. A. , Pepperell J. G. , Davis T. L. O. 2003 . Use of charter boat data to infer the annual availability of black marlin,Makaira indica, to the recreational fishery off Cairns, Australia . Marine and Freshwater Research , 54 : 447 – 457 . Google Scholar Crossref Search ADS Ciannelli L. , Fauchald P. , Chan K. S. , Agostini V. N. , Dingsør G. E. 2008 . Spatial fisheries ecology: recent progress and future prospects . Journal of Marine Systems , 71 : 223 – 236 . Google Scholar Crossref Search ADS Cochran W. G. 1977 . Sampling Techniques . John Wiley & Sons , New York, NY . 428 pp. Coleman F. C. , Figueira W. F. , Ueland J. S. , Crowder L. B. 2004 . The impact of United States recreational fisheries on marine fish populations . Science , 305 : 1958 – 1960 . Google Scholar Crossref Search ADS PubMed Collins J. H. 2008 . Marine tourism in the Kimberley region of Western Australia . Geographical Research , 46 : 111 – 123 . Google Scholar Crossref Search ADS Cooke S. J. , Cowx I. G. 2004 . The role of recreational fishing in global fish crises . BioScience , 54 : 857 – 859 . Google Scholar Crossref Search ADS Cooke S. J. , Philipp D. P. 2004 . Behavior and mortality of caught-and-released bonefish (Albula spp.) in Bahamian waters with implications for a sustainable recreational fishery . Biological Conservation , 118 : 599 – 607 . Google Scholar Crossref Search ADS Crowe F. M. , Longson I. G. , Joll L. M. 2013 . Development and implementation of allocation arrangements for recreational and commercial fishing sectors in Western Australia . Fisheries Management and Ecology , 20 : 201 – 210 . Google Scholar Crossref Search ADS Day J. 2008 . The need and practice of monitoring, evaluating and adapting marine planning and management—lessons from the Great Barrier Reef . Marine Policy , 32 : 823 – 831 . Google Scholar Crossref Search ADS Department of Fisheries. 2011. Recreational Fishing in Western Australia: Fishing Location Guide. Department of Fisheries, Perth, Western Australia. Number 92. 19 pp. Department of Fisheries. 2017. Recreational fishing in Western Australia - fish identification guide. Fisheries Occasional Publication No. 103. Western Australia, Perth. 33 pp. Director of National Parks. 2018a. North-west Marine Parks Network Management Plan 2018, Director of National Parks, Canberra. Australian Capital Territory. 159 pp. Director of National Parks 2018b. Temperate East Marine Parks Network Management Plan 2018, Director of National Parks, Canberra. Australian Capital Territory. 137 pp. Emslie M. J. , Logan M. , Williamson D. H. , Ayling A. M. , MacNeil M. A. , Ceccarelli D. , Cheal A. J. et al. 2015 . Expectations and outcomes of reserve network performance following re-zoning of the Great Barrier Reef Marine Park . Current Biology , 25 : 983 – 992 . Google Scholar Crossref Search ADS PubMed Evans K. , Bax N. J. , Smith D. C. 2017 . Australia State of the Environment 2016: Marine Environment, Independent Report to the Australian Government Minister for the Environment and Energy, Australian Government Department of the Environment and Energy, Canberra. 226 pp. Flynn D. J. H. , Lynch T. P. , Barrett N. S. , Wong L. S. C. , Devine C. , Hughes D. 2018 . Gigapixel big data movies provide cost-effective seascape scale direct measurements of open-access coastal human use such as recreational fisheries . Ecology and Evolution , 8 : 9372 – 9383 . Google Scholar Crossref Search ADS PubMed Ghosn D. L. , Collins D. P. , Gould A. 2015 . The NSW Game Fish Tournament Monitoring Program 1994 to 2013: a summary of data and assessment of the role and design. NSW Department of Primary Industries - Fisheries Final Report Series No. 144. 162 pp. Giri K. , Hall K. 2015 . South Australian Recreational Fishing Survey . Fisheries Victoria Internal Report Series No. 62. Victorian Government, Department of Economic Development, Jobs, Transport and Resources. 66 pp. Green C. , Brown P. , Giri K. , Bell J. , Conron S. 2012 . Quantifying the Recreational Catch of Southern Bluefin Tuna off the Victorian Coast. ICES Document R09/10/03. 7 pp. Griffiths S. , Lynch T. P. , Lyle J. , Wotherspoon S. , Wong L. , Devine C. , Pollock K. et al. 2017 . Trial and validation of Respondent-Driven Sampling as a cost-effective method for obtaining representative catch, effort, social and economic data from recreational fisheries. FRDC Project No 2012/021. 145 pp. Griffiths S. P. , Pollock K. H. , Lyle J. M. , Pepperell J. G. , Tonks M. L. , Sawynok W. 2010 . Following the chain to elusive anglers . Fish and Fisheries , 11 : 220 – 228 . Google Scholar Crossref Search ADS Henry G. W. , Lyle J. M. 2003 . National Recreational and Indigenous Fishing Survey. ICES Document Project, 1999/158. 188 pp. Herfaut J. , Levrel H. , Thébaud O. , Véron G. 2013 . The nationwide assessment of marine recreational fishing: a French example . Ocean & Coastal Management , 78 : 121 – 131 . Google Scholar Crossref Search ADS Holdsworth J. C. , Hartill B. W. , Heinemann A. , Wynne-Jones J. 2018 . Integrated survey methods to estimate harvest by marine recreational fishers in New Zealand . Fisheries Research , 204 : 424 – 432 . Google Scholar Crossref Search ADS Horigue V. , Pressey R. L. , Mills M. , Brotánková J. , Cabral R. , Andréfouët S. 2015 . Benefits and challenges of scaling up expansion of marine protected area networks in the Verde Island Passage, Central Philippines . PLoS One , 10 : e0135789. Google Scholar Crossref Search ADS PubMed Hyder K. , Weltersbach M. S. , Armstrong M. , Ferter K. , Townhill B. , Ahvonen A. , Arlinghaus R. et al. 2018 . Recreational sea fishing in Europe in a global context—participation rates, fishing effort, expenditure, and implications for monitoring and assessment . Fish and Fisheries , 19 : 225 – 243 . Google Scholar Crossref Search ADS Ihde T. F. , Wilberg M. J. , Loewensteiner D. A. , Secor D. H. , Miller T. J. 2011 . The increasing importance of marine recreational fishing in the US: challenges for management . Fisheries Research , 108 : 268 – 276 . Google Scholar Crossref Search ADS Jones K. 2009 . South Australian Recreational Fishing Survey. ICES Document Paper, 54. 84 pp. Kearney R. E. 2001 . Fisheries property rights and recreational/commercial conflict: implications of policy developments in Australia and New Zealand . Marine Policy , 25 : 49 – 59 . Google Scholar Crossref Search ADS Kristianson G. 2018 . Lessons in fisheries politics . Fisheries Research , 205 : 32 – 34 . Google Scholar Crossref Search ADS Last P. R. , Stevens J. D. 2009 . Sharks and Rays of Australia . CSIRO Publishing, Collingwood, VIC . Lewin W. C. , Arlinghaus R. , Mehner T. 2006 . Documented and potential biological impacts of recreational fishing: insights for management and conservation . Reviews in Fisheries Science , 14 : 305 – 367 . Google Scholar Crossref Search ADS Lindner, R. K., and McLeod, P. B. 1991. The Economic Impact of Recreational Fin Western Australia. Western Australia, Perth. Lowry M. , Murphy J. 2003 . Monitoring the recreational gamefish fishery off south-eastern Australia . Marine and Freshwater Research , 54 : 425 – 434 . Google Scholar Crossref Search ADS Lumley T. 2010 . Complex Surveys: A Guide to Analysis Using R . John Wiley & Sons , New Jersey . Lyle J. M. , Coleman A. P. M. , West L. , Campbell D. , Henry G. W. 2002 . New large-scale methods for evaluating sport fisheries. In Recreational Fisheries: Ecological, Economic and Social Evaluation. Fish and Aquatic Resources Series 8 , pp. 207 – 226 . Ed. by Pitcher T. J. , Hollingworth C. . Blackwell Science , London . Lyle, J. M., Tracey, S. R., Stark, K. E., and Wotherspoon, S. 2014. 2012–2013 survey of recreational fishing in Tasmania. IMAS, Hobart, Tasmania. 124 pp. Lyle, J. M., Tracey, S. R., Stark, K. E., and Wotherspoon, S. 2009. 2007–08 Survey of Recreational Fishing in Tasmania. Tasmanian Aquaculture and Fisheries Institute, Hobart, Tasmania. 97 pp. Lyle J. M. , Wotherspoon S. , Stark K. E. 2010 . Developing an analytical module for large-scale recreational fishery data based on phone–diary survey methodology. FRDC Project no. 2007/064. Tasmanian Aquaculture and Fisheries Institute, 2010. 99 pp. Lynch T. P. 2006 . Incorporation of recreational fishing effort into design of marine protected areas . Conservation Biology , 20 : 1466 – 1476 . Google Scholar Crossref Search ADS PubMed Lynch T. P. 2008 . The difference between spatial and temporal variation in recreational fisheries for planning of marine protected areas: response to Steffe . Conservation Biology , 22 : 486 – 491 . Google Scholar Crossref Search ADS Lynch T. P. 2014 . A decadal time-series of recreational fishing effort collected during and after implementation of a multiple use marine park shows high inter-annual but low spatial variability . Fisheries Research , 151 : 85 – 90 . Google Scholar Crossref Search ADS Lynch T. P. , Harcourt R. , Edgar G. , Barrett N. 2013 . Conservation of the critically endangered Eastern Australian Population of the Grey Nurse Shark (Carcharias taurus) through cross-jurisdictional management of a network of marine-protected areas . Environmental Management , 52 : 1341 – 1354 . Google Scholar Crossref Search ADS PubMed Lynch T. P. , Morello E. B. , Evans K. , Richardson A. J. , Rochester W. , Steinberg C. R. , Roughan M. et al. 2014 . IMOS National Reference Stations: a continental-wide physical, chemical and biological coastal observing system . PLoS One , 9 : e113652. Google Scholar Crossref Search ADS PubMed McCluskey S. M. , Lewison R. L. 2008 . Quantifying fishing effort: a synthesis of current methods and their applications . Fish and Fisheries , 9 : 188 – 200 . Google Scholar Crossref Search ADS McPhee D. , Leadbitter D. , A. Skilleter G. 2002 . Swallowing the bait: is recreational fishing in Australia ecologically sustainable? Pacific Conservation Biology , 8 : 40. Google Scholar Crossref Search ADS Moore A. , Hall K. , Giri K. , Tracey S. , Penrose L. , Hansen S. , Ward P. et al. 2015 . Developing Robust and Cost-effective Methods for Estimating the National Recreational Catch of Southern Bluefin Tuna in Australia. ICES Document 2012/022.20. 123 pp. MPRA and CALM. 2005. Management Plan for the Ningaloo Marine Park and Muiron Islands Marine Management Area; 2005–2015. Perth, Western Australia. Number 52. 115 pp. Muoneke M. I. , Childress W. M. 1994 . Hooking mortality: a review for recreational fisheries . Reviews in Fisheries Science , 2 : 123 – 156 . Google Scholar Crossref Search ADS Nevill J. , Ward T. 2009 . The National Representative System of Marine Protected Areas: comment on recent progress . Ecological Management & Restoration , 10 : 228 – 231 . Google Scholar Crossref Search ADS Newman S. J. , Brown J. I. , Fairclough D. V. , Wise B. S. , Bellchambers L. M. , Molony B. W. , Lenanton R. C. J. et al. 2018 . A risk assessment and prioritisation approach to the selection of indicator species for the assessment of multi-species, multi-gear, multi-sector fishery resources . Marine Policy , 88 : 11 – 22 . Google Scholar Crossref Search ADS Pollock K. H. , Jones C. M. , Brown T. L. 1994 . Angler survey methods and their applications in fisheries management. American Fisheries Society, 25: 371. Productivity Commission. 2016 . Marine fisheries and aquaculture . Final Report , 81 . 43 pp. R Core Team. 2017. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Rufino M. M. , Stelzenmüller V. , Maynou F. , Zauke G. P. 2006 . Assessing the performance of linear geostatistical tools applied to artificial fisheries data . Fisheries Research , 82 : 263 – 279 . Google Scholar Crossref Search ADS Ryan K. L. , Hall H. G. , Lai E. K. , Smallwood C. B. , Taylor S. M. , Wise B. S. 2015 . State-wide survey of boat-based recreational fishing in Western Australia 2013/14. Department of Fisheries, Perth, Western Australia. Number 268. 200 pp. Ryan K. L. , Hall N. G. , Lai E. K. , Smallwood C. B. , Taylor S. M. , Wise B. S. 2017 . Statewide Survey of Boat-based Recreational Fishing in Western Australia 2015/2016. ICES Document 287. 205 pp. Ryan K. L. , Wise B. S. , Hall N. G. , Pollock K. H. , Sulin E. H. , Gaughan D. J. 2013 . An integrated system to survey boat-based recreational fishing in Western Australia 2011/2012. Department of Fisheries, Perth, Western Australia. Number 249. 249 pp. Shertzer K. W. , Bacheler N. M. , Kellison G. T. , Fieberg J. , Wiggers R. K. 2018 . Release mortality of endangered Warsaw grouper Hyporthodus nigritus: a state-space model applied to capture-recapture data . Endangered Species Research , 35 : 15 – 22 . Google Scholar Crossref Search ADS Short A. D. , Woodroffe C. D. 2009 . The Coast of Australia . Cambridge University Press , Cambridge . Skomal G. B. 2007 . Evaluating the physiological and physical consequences of capture on post-release survivorship in large pelagic fishes . Fisheries Management and Ecology , 14 : 81 – 89 . Google Scholar Crossref Search ADS Smallwood C. B. , Beckley L. E. , Moore S. A. 2012 . Influence of zoning and habitats on the spatial distribution of recreational activities in a multiple-use Marine Park . Coastal Management , 40 : 381 – 400 . Google Scholar Crossref Search ADS Smallwood C. B. , Pollock K. H. , Wise B. S. , Hall N. G. , Gaughan D. J. 2012 . Expanding aerial-roving surveys to include counts of shore-based recreational fishers from remotely operated cameras: benefits, limitations, and cost effectiveness . North American Journal of Fisheries Management , 32 : 1265 – 1276 . Google Scholar Crossref Search ADS Southwick R. , Holdsworth J. C. , Rea T. , Bragg L. , Allen T. 2018 . Estimating marine recreational fishing’s economic contributions in New Zealand . Fisheries Research , 208 : 116 – 123 . Google Scholar Crossref Search ADS Stelzenmüller V. , Maynou F. , Ehrich S. , Zauke G.-P. 2004 . Spatial analysis of twaite shad, Alosa fallax (Lacepède, 1803), in the Southern North Sea: application of non-linear geostatistics as a tool to search for special areas of conservation . International Review of Hydrobiology , 89 : 337 – 351 . Google Scholar Crossref Search ADS Taylor S. , Webley J. , McInnes K. 2012 . 2010 Statewide Recreational Fishing Survey. Fisheries Queensland. 93 pp. Teixeira D. , Zischke M. T. , Webley J. A. C. 2016 . Investigating bias in recreational fishing surveys: fishers listed in public telephone directories fish similarly to their unlisted counterparts . Fisheries Research , 181 : 127 – 136 . Google Scholar Crossref Search ADS The National Academies of Sciences, Engineering, and Medicine. 2017 . Review of the Marine Recreational Information Program . The National Academies Press , Washington, DC . Venerus L. A. , Cedrola P. V. 2017 . Review of marine recreational fisheries regulations in Argentina . Marine Policy , 81 : 202 – 210 . Google Scholar Crossref Search ADS Webley J. , McInnes K. , Teixeira D. , Lawson A. , Quinn R. 2015 . Statewide Recreational Fishing Survey 2013–2014. The Queensland Government, Brisbane, Queensland. 127 pp. West L. D. , Lyle J. M. , Matthews S. R. , Stark K. E. , Steffe A. S. 2012 . Survey of recreational fishing in the Northern Territory, 2009–2010. Northern Territory Government, Australia. Fishery Report No. 109. 125 pp. West L. D. , Stark K. E. , Murphy J. J. , Lyle J. M. , Ochwada-Doyle F. A. 2015 . Survey of Recreational Fishing in New South Wales and the ACT, 2013/2014 . ICES Document , 149 . 150 pp. Wise B. S. , Fletcher W. J. 2013 . Determination and development of cost effective techniques to monitor recreational catch and effort in Western Australian demersal finfish fisheries. Department of Fisheries. Perth, Western Australia. No. 245. 168 pp. Wise B. S. , Telfer C. F. , Lai E. K. M. , Hall N. G. , Jackson G. 2012 . Long-term monitoring of boat-based recreational fishing in Shark Bay, Western Australia: providing scientific advice for sustainable management in a World Heritage Area . Marine and Freshwater Research , 63 : 1129 – 1141 . Google Scholar Crossref Search ADS Wood G. , Lynch T. P. , Devine C. , Keller K. , Figueira W. 2016 . High-resolution photo-mosaic time-series imagery for monitoring human use of an artificial reef . Ecology and Evolution , 6 : 6963 – 6968 . Google Scholar Crossref Search ADS PubMed Zischke M. T. , Griffiths S. P. , Tibbetts I. R. 2012 . Catch and effort from a specialised recreational pelagic sport fishery off eastern Australia . Fisheries Research , 127–128 : 61 – 72 . Google Scholar Crossref Search ADS © International Council for the Exploration of the Sea 2019. 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)

Gíslason, Sindri; Pálsson, Snæbjörn; Jónasson, Jónas P; Guls, Hermann Dreki; Svavarsson, Jörundur; Halldórsson, Halldór P

doi: 10.1093/icesjms/fsaa059pmid: N/A

Abstract The Atlantic rock crab (Cancer irroratus) was first found in Icelandic waters in 2006. Since then, the species has dispersed rapidly and is currently found clockwise from the southwest coast of Iceland to the east, corresponding to >70% of the coastline. Here, we present a monitoring study on this non-indigenous crab species in Iceland from 2007 to 2019. The study shows that the rock crab is now the most abundant brachyuran crab species on soft substrate bottoms in Southwest Iceland, both as adults and planktonic larvae, indicating that it is outcompeting its rival native species, the European green crab (Carcinus maenas) and the spider crab (Hyas araneus). The average size of the rock crab was similar over time (2007–2019), although it fluctuated between years in a pattern similar to that for the green crab, while significant reduction in size was observed for male spider crabs. The rock crab population is still in a growth phase in Icelandic waters, as seen in increasing distributional range, and can be found in densities comparable to the highest reported for the species in its native range in North America. Introduction Marine non-indigenous species are a worldwide problem, which has increased enormously in recent decades due to increasing world oceanic trade and travel by humans (Cohen and Carlton, 1998; Brickman, 2006). They are considered important drivers of ecological changes, as their impact can lead to habitat changes, displacement of native species through predation and/or competition, spread of diseases, and reduction in biodiversity (Bax et al., 2003). Crustaceans (Arthropoda) are among the most successful marine invaders, and one of their largest and most widespread groups is decapods (Hänfling et al., 2011). Decapods play an important role in benthic communities, ranging from intertidal to deep waters. They are successful and versatile predators that prey on more than one trophic level and interact in various ways with their habitat and its inhabitants, as well as being prey for a large range of both vertebrates and invertebrates (Boudreau and Worm, 2012). Examples of the invasive decapod species include the European green crab (Carcinus maenas) (Klassen and Locke, 2007), the Harris mud crab (Rhithropanopeus harrisii) (Roche and Torchin, 2007), the Chinese mitten crab (Eriochier sinensis) (Schrimpf et al., 2014), the red king crab (Paralithodes camtschaticus) (Dvoretsky and Dvoretsky, 2013), the Asian shore crab (Hemigrapsus sanguineus) (Jungblut et al., 2017), and the Asian brush-clawed shore crab (Hemigrapsus takanoi) (Makino et al., 2018). Despite extensive shipping in Nordic waters, relatively few anthropogenic oceanic introductions have been reported at high northern latitudes in the Atlantic Ocean. That might be explained by a rapid dispersal following the retreat of glaciers about 10 000 years ago, as has been demonstrated by Ingólfsson (1992) that the faunas of the northernmost Atlantic regions (Canadian Maritimes, Iceland, Norway) are closely related despite long distances between them. As current maritime traffic is most often between low and high latitudes, the risk of invasion might be diminished as temperature may affect the establishment of species from lower latitudes to colder regions (Seebens et al., 2013), although many historical introductions may have been overlooked (Carlton, 2003, 2009). One of the most recent members of the invasive alien crustaceans in Icelandic coastal waters is the Atlantic rock crab (Cancer irroratus), which was first reported in Icelandic coastal waters in 2006 (Gíslason et al., 2014) (Figure 1). Its occurrence in Iceland is the first discovery of the species outside its native range in North America. Its colonization may possibly be aided by the large-scale changes that started in the North Atlantic in 1996 that resulted in warmer waters around Iceland and led to noticeable changes in the Icelandic marine ecosystem (Anonymous, 2004; Astthorsson and Palsson, 2006; Astthorsson et al., 2007, 2012; Stefansdottir et al., 2010; Jochumsen et al., 2016) and also to increased shipping in past decades. The apparent lack of founder effects, high genetic variation (Gíslason et al., 2013), and apparently favourable environmental conditions may have led to the rapid spread of the rock crab since 2006 (Gíslason et al., 2014). The success of the rock crab and its ability to thrive in Iceland is further demonstrated by the fact that it has been found in high abundance, comparable to the highest records in its native range (Gíslason et al., 2017). In its new habitat in Icelandic coastal waters, the rock crab has few competitors for food and shelter, as there are only two native brachyuran decapod crab species commonly found in shallow coastal waters, i.e. the European green crab (C. maenas) and the great spider crab (Hyas araneus) (Gíslason et al., 2014). All three species are known to display different life history traits (Supplementary Table S1), but very limited information is available for Iceland. Figure 1. Open in new tabDownload slide Known distribution of the Atlantic rock crab (Cancer irroratus) around Iceland in 2019 (grey shaded area). The map divides Iceland into six areas showing the year when specimens where first found. Asterisk shows the location where rock crab was first found in Hvalfjörður Southwest Iceland. Figure 1. Open in new tabDownload slide Known distribution of the Atlantic rock crab (Cancer irroratus) around Iceland in 2019 (grey shaded area). The map divides Iceland into six areas showing the year when specimens where first found. Asterisk shows the location where rock crab was first found in Hvalfjörður Southwest Iceland. Our aims were to evaluate (i) population changes in the Atlantic rock crab in Southwest Icelandic waters during the latter stages of the colonization and (ii) the interaction of the rock crab and the two native brachyuran crab species, the European green crab and the spider crab. This was done by sampling larvae and adults of all species, comparing changes in abundance, and seeing if sampling areas differed in abundance, species composition, and size distributions of the three crab species. Material and methods Sampling Trap fishing and plankton sampling were carried out in four areas in the inner parts of Faxaflói Bay, Southwest Iceland: Hvalfjörður, Kollafjörður, Skerjafjörður, and Borgarfjörður (Figure 2). Rock crabs and other decapod crabs were captured with commercial crab traps (depth range 7–80 m) at various times during 2007–2019 (Table 1). Twenty to 30 traps were used on each sampling trip (height 30 cm, length 80 cm, width 40 cm, mesh size 4.8 cm, escape opening for juveniles closed). Traps were baited with fish, always containing some gadoids (Gadus morhua, Pollachius virens, Melanogrammus aeglefinus, Merlangius merlangus). Mixed bait was placed in mesh bags hanging in the traps, ca. 250 g per trap. Baited traps remained in place for about 48 h before retrieval. Figure 2. Open in new tabDownload slide The sampling areas in Faxaflói Bay, Southwest Iceland: Skerjafjörður (S), Kollafjörður (K), Hvalfjörður (H), and Borgarfjörður (B). Trap fisheries were carried out in areas shaded with grey. Plankton sampling was carried out at six stations marked with black dots (B1–B6). Figure 2. Open in new tabDownload slide The sampling areas in Faxaflói Bay, Southwest Iceland: Skerjafjörður (S), Kollafjörður (K), Hvalfjörður (H), and Borgarfjörður (B). Trap fisheries were carried out in areas shaded with grey. Plankton sampling was carried out at six stations marked with black dots (B1–B6). Table 1. Length and weight characteristics (mean, range, and standard error) for the Atlantic rock crab (Cancer irroratus), European green crab (Carcinus maenas), and spider crab (Hyas araneus) in four areas in Faxaflói Bay, Southwest Iceland. Species . Area . Sampling years . No. trips . N . Carapace (cm) . Wet weight (g) . Mean . Range . s.e. . Mean . Range . s.e. . C. irroratus Hvalfjörður  Males 2007–2019 42 9 361 11.7 6.5–15.1 0.012 256.2 60–487 1.235  Females 1 347 8.8 5.8–11.7 0.020 116.2 59–254 1.574  All 10 708 11.3 5.8–15.1 0.014 240.4 59–487 1.356 Kollafjörður  Males 2011–2019 27 8 140 10.4 5.3–13.8 0.014 185.0 40–452 1.015  Females 1 491 8.3 5.6–11.8 0.017 92.0 46–220 0.947  All 9 631 10.1 5.3–13.8 0.015 174.1 40–452 0.997 Skerjafjörður  Males 2013, 2016–2017 5 1 810 11.3 7.0–14.2 0.032 – – –  Females 66 8.9 7.5-10.2 0.079 – – –  All 1 876 11.2 7.0–14.2 0.031 – – – Borgarfjörður  Males 2013, 2017 2 994 12.8 8.2–15.3 0.033 – – –  Females 968 8.9 7.0–11.1 0.019 – – –  All 1 962 10,9 7.0–15.3 0.048 – – – C. maenas Hvalfjörður  Males 2007–2019 42 730 7.3 5.0–10.2 0.027 99.1 35–161 4.559  Females 139 6.3 4.7–8.5 0.056 54.9 29–96 3.832  All 869 7.1 4.7–10.2 0.027 86.4 29–161 4.221 Kollafjörður  Males 2011–2019 27 4 741 7.1 5.1–9.9 0.008 100.1 40–196 1.055  Females 306 6.5 4.5–7.9 0.030 63.5 20–120 1.331  All 5 047 7.1 4.5–9.9 0.008 93.3 20–196 1.024 Skerjafjörður  Males 2013, 2016–2017 5 467 7.9 6.2–9.6 0.030 –- – –  Females 32 6.8 5.4–8.2 0.113 – – –  All 499 7.8 5.4–9.6 0.033 – – – Borgarfjörður  Males 2013, 2017 2 4 6.8 5.8–7.5 0.382 – – –  Females 5 6.3 5.9–7.2 0.231 – – –  All 9 6.5 5.8–7.5 0.231 – – – H. araneus Hvalfjörður  Males 2007–2019 42 1 657 8.8 4.7–12.2 0.028 174.6 42–403 3.338  Females 501 7.7 4.5–10.2 0.031 100.2 30–234 2.519  All 2 158 8.6 4.5–12.2 0.025 156.6 30–403 2.917 Kollafjörður  Males 2011–2019 27 597 8.8 4.5–11.5 0.050 177.3 36–310 5.389  Females 78 6.8 5.4–9.9 0.094 61.0 30–136 3.446  All 675 8.6 4.5–11.5 0.052 148.9 30–310 5.431 Skerjafjörður  Males 2013, 2016–2017 5 189 8.1 5.5–10.9 0.082 – – –  Females 29 7.2 5.7–8.3 0.134 – – –  All 218 7.9 5.5–10.9 0.075 – – – Borgarfjörður  Males 2013, 2017 2 6 9.3 8.2–10.1 0.251 – – –  Females 2 6.3 5.4–7.1 0.850 – – –  All 8 8.5 5.4–10.1 0.548 – – – Species . Area . Sampling years . No. trips . N . Carapace (cm) . Wet weight (g) . Mean . Range . s.e. . Mean . Range . s.e. . C. irroratus Hvalfjörður  Males 2007–2019 42 9 361 11.7 6.5–15.1 0.012 256.2 60–487 1.235  Females 1 347 8.8 5.8–11.7 0.020 116.2 59–254 1.574  All 10 708 11.3 5.8–15.1 0.014 240.4 59–487 1.356 Kollafjörður  Males 2011–2019 27 8 140 10.4 5.3–13.8 0.014 185.0 40–452 1.015  Females 1 491 8.3 5.6–11.8 0.017 92.0 46–220 0.947  All 9 631 10.1 5.3–13.8 0.015 174.1 40–452 0.997 Skerjafjörður  Males 2013, 2016–2017 5 1 810 11.3 7.0–14.2 0.032 – – –  Females 66 8.9 7.5-10.2 0.079 – – –  All 1 876 11.2 7.0–14.2 0.031 – – – Borgarfjörður  Males 2013, 2017 2 994 12.8 8.2–15.3 0.033 – – –  Females 968 8.9 7.0–11.1 0.019 – – –  All 1 962 10,9 7.0–15.3 0.048 – – – C. maenas Hvalfjörður  Males 2007–2019 42 730 7.3 5.0–10.2 0.027 99.1 35–161 4.559  Females 139 6.3 4.7–8.5 0.056 54.9 29–96 3.832  All 869 7.1 4.7–10.2 0.027 86.4 29–161 4.221 Kollafjörður  Males 2011–2019 27 4 741 7.1 5.1–9.9 0.008 100.1 40–196 1.055  Females 306 6.5 4.5–7.9 0.030 63.5 20–120 1.331  All 5 047 7.1 4.5–9.9 0.008 93.3 20–196 1.024 Skerjafjörður  Males 2013, 2016–2017 5 467 7.9 6.2–9.6 0.030 –- – –  Females 32 6.8 5.4–8.2 0.113 – – –  All 499 7.8 5.4–9.6 0.033 – – – Borgarfjörður  Males 2013, 2017 2 4 6.8 5.8–7.5 0.382 – – –  Females 5 6.3 5.9–7.2 0.231 – – –  All 9 6.5 5.8–7.5 0.231 – – – H. araneus Hvalfjörður  Males 2007–2019 42 1 657 8.8 4.7–12.2 0.028 174.6 42–403 3.338  Females 501 7.7 4.5–10.2 0.031 100.2 30–234 2.519  All 2 158 8.6 4.5–12.2 0.025 156.6 30–403 2.917 Kollafjörður  Males 2011–2019 27 597 8.8 4.5–11.5 0.050 177.3 36–310 5.389  Females 78 6.8 5.4–9.9 0.094 61.0 30–136 3.446  All 675 8.6 4.5–11.5 0.052 148.9 30–310 5.431 Skerjafjörður  Males 2013, 2016–2017 5 189 8.1 5.5–10.9 0.082 – – –  Females 29 7.2 5.7–8.3 0.134 – – –  All 218 7.9 5.5–10.9 0.075 – – – Borgarfjörður  Males 2013, 2017 2 6 9.3 8.2–10.1 0.251 – – –  Females 2 6.3 5.4–7.1 0.850 – – –  All 8 8.5 5.4–10.1 0.548 – – – Minimum catch size of crabs was 4.5 cm (carapace width/length). Trap catches were carried out in 2013 and 2017 in Borgarfjörður (September, October), in 2007–2019 in Hvalfjörður (April–December), in 2011–2019 in Kollafjörður (August and September), and in 2013, 2016, and 2017 in Skerjafjörður (March, June, July, December). Number of traps varied from 20 to 30 per sampling trip. Open in new tab Table 1. Length and weight characteristics (mean, range, and standard error) for the Atlantic rock crab (Cancer irroratus), European green crab (Carcinus maenas), and spider crab (Hyas araneus) in four areas in Faxaflói Bay, Southwest Iceland. Species . Area . Sampling years . No. trips . N . Carapace (cm) . Wet weight (g) . Mean . Range . s.e. . Mean . Range . s.e. . C. irroratus Hvalfjörður  Males 2007–2019 42 9 361 11.7 6.5–15.1 0.012 256.2 60–487 1.235  Females 1 347 8.8 5.8–11.7 0.020 116.2 59–254 1.574  All 10 708 11.3 5.8–15.1 0.014 240.4 59–487 1.356 Kollafjörður  Males 2011–2019 27 8 140 10.4 5.3–13.8 0.014 185.0 40–452 1.015  Females 1 491 8.3 5.6–11.8 0.017 92.0 46–220 0.947  All 9 631 10.1 5.3–13.8 0.015 174.1 40–452 0.997 Skerjafjörður  Males 2013, 2016–2017 5 1 810 11.3 7.0–14.2 0.032 – – –  Females 66 8.9 7.5-10.2 0.079 – – –  All 1 876 11.2 7.0–14.2 0.031 – – – Borgarfjörður  Males 2013, 2017 2 994 12.8 8.2–15.3 0.033 – – –  Females 968 8.9 7.0–11.1 0.019 – – –  All 1 962 10,9 7.0–15.3 0.048 – – – C. maenas Hvalfjörður  Males 2007–2019 42 730 7.3 5.0–10.2 0.027 99.1 35–161 4.559  Females 139 6.3 4.7–8.5 0.056 54.9 29–96 3.832  All 869 7.1 4.7–10.2 0.027 86.4 29–161 4.221 Kollafjörður  Males 2011–2019 27 4 741 7.1 5.1–9.9 0.008 100.1 40–196 1.055  Females 306 6.5 4.5–7.9 0.030 63.5 20–120 1.331  All 5 047 7.1 4.5–9.9 0.008 93.3 20–196 1.024 Skerjafjörður  Males 2013, 2016–2017 5 467 7.9 6.2–9.6 0.030 –- – –  Females 32 6.8 5.4–8.2 0.113 – – –  All 499 7.8 5.4–9.6 0.033 – – – Borgarfjörður  Males 2013, 2017 2 4 6.8 5.8–7.5 0.382 – – –  Females 5 6.3 5.9–7.2 0.231 – – –  All 9 6.5 5.8–7.5 0.231 – – – H. araneus Hvalfjörður  Males 2007–2019 42 1 657 8.8 4.7–12.2 0.028 174.6 42–403 3.338  Females 501 7.7 4.5–10.2 0.031 100.2 30–234 2.519  All 2 158 8.6 4.5–12.2 0.025 156.6 30–403 2.917 Kollafjörður  Males 2011–2019 27 597 8.8 4.5–11.5 0.050 177.3 36–310 5.389  Females 78 6.8 5.4–9.9 0.094 61.0 30–136 3.446  All 675 8.6 4.5–11.5 0.052 148.9 30–310 5.431 Skerjafjörður  Males 2013, 2016–2017 5 189 8.1 5.5–10.9 0.082 – – –  Females 29 7.2 5.7–8.3 0.134 – – –  All 218 7.9 5.5–10.9 0.075 – – – Borgarfjörður  Males 2013, 2017 2 6 9.3 8.2–10.1 0.251 – – –  Females 2 6.3 5.4–7.1 0.850 – – –  All 8 8.5 5.4–10.1 0.548 – – – Species . Area . Sampling years . No. trips . N . Carapace (cm) . Wet weight (g) . Mean . Range . s.e. . Mean . Range . s.e. . C. irroratus Hvalfjörður  Males 2007–2019 42 9 361 11.7 6.5–15.1 0.012 256.2 60–487 1.235  Females 1 347 8.8 5.8–11.7 0.020 116.2 59–254 1.574  All 10 708 11.3 5.8–15.1 0.014 240.4 59–487 1.356 Kollafjörður  Males 2011–2019 27 8 140 10.4 5.3–13.8 0.014 185.0 40–452 1.015  Females 1 491 8.3 5.6–11.8 0.017 92.0 46–220 0.947  All 9 631 10.1 5.3–13.8 0.015 174.1 40–452 0.997 Skerjafjörður  Males 2013, 2016–2017 5 1 810 11.3 7.0–14.2 0.032 – – –  Females 66 8.9 7.5-10.2 0.079 – – –  All 1 876 11.2 7.0–14.2 0.031 – – – Borgarfjörður  Males 2013, 2017 2 994 12.8 8.2–15.3 0.033 – – –  Females 968 8.9 7.0–11.1 0.019 – – –  All 1 962 10,9 7.0–15.3 0.048 – – – C. maenas Hvalfjörður  Males 2007–2019 42 730 7.3 5.0–10.2 0.027 99.1 35–161 4.559  Females 139 6.3 4.7–8.5 0.056 54.9 29–96 3.832  All 869 7.1 4.7–10.2 0.027 86.4 29–161 4.221 Kollafjörður  Males 2011–2019 27 4 741 7.1 5.1–9.9 0.008 100.1 40–196 1.055  Females 306 6.5 4.5–7.9 0.030 63.5 20–120 1.331  All 5 047 7.1 4.5–9.9 0.008 93.3 20–196 1.024 Skerjafjörður  Males 2013, 2016–2017 5 467 7.9 6.2–9.6 0.030 –- – –  Females 32 6.8 5.4–8.2 0.113 – – –  All 499 7.8 5.4–9.6 0.033 – – – Borgarfjörður  Males 2013, 2017 2 4 6.8 5.8–7.5 0.382 – – –  Females 5 6.3 5.9–7.2 0.231 – – –  All 9 6.5 5.8–7.5 0.231 – – – H. araneus Hvalfjörður  Males 2007–2019 42 1 657 8.8 4.7–12.2 0.028 174.6 42–403 3.338  Females 501 7.7 4.5–10.2 0.031 100.2 30–234 2.519  All 2 158 8.6 4.5–12.2 0.025 156.6 30–403 2.917 Kollafjörður  Males 2011–2019 27 597 8.8 4.5–11.5 0.050 177.3 36–310 5.389  Females 78 6.8 5.4–9.9 0.094 61.0 30–136 3.446  All 675 8.6 4.5–11.5 0.052 148.9 30–310 5.431 Skerjafjörður  Males 2013, 2016–2017 5 189 8.1 5.5–10.9 0.082 – – –  Females 29 7.2 5.7–8.3 0.134 – – –  All 218 7.9 5.5–10.9 0.075 – – – Borgarfjörður  Males 2013, 2017 2 6 9.3 8.2–10.1 0.251 – – –  Females 2 6.3 5.4–7.1 0.850 – – –  All 8 8.5 5.4–10.1 0.548 – – – Minimum catch size of crabs was 4.5 cm (carapace width/length). Trap catches were carried out in 2013 and 2017 in Borgarfjörður (September, October), in 2007–2019 in Hvalfjörður (April–December), in 2011–2019 in Kollafjörður (August and September), and in 2013, 2016, and 2017 in Skerjafjörður (March, June, July, December). Number of traps varied from 20 to 30 per sampling trip. Open in new tab Hvalfjörður was sampled during 42 trips during 2007–2019; 2016 is the only year with no data from that area (Supplementary Table S2). Each year and on each occasion, ten traps were placed out on a transect at a depth gradient (10, 20, 30, 40, and 60 m) for comparison with the former study by Gíslason et al. (2014), except in 2012, when traps were only laid out at 10-m depth. Kollafjörður was sampled during 27 trips in 2011–2019 (Table 1) in August and September. Borgarfjörður was sampled during two trips in 2013 and 2017 (Table 1) in September and October, respectively. Skerjafjörður was sampled during five trips in 2013, 2016, and 2017 (Table 1) in June, July, December, and March, respectively. Most of the crabs were identified to species and gender. Presence or absence of egg remnants on females was determined, and the developmental stages of eggs in berried females were determined by their colour (bright orange = undeveloped; brown = developed). All three species are sexually dimorphic in size, with females considerably smaller than males. Size and weight of Atlantic rock crabs, green crabs, and spider crabs were measured. Total body weight was measured using an electronic scale, with an accuracy of ±1 g. The sizes of the rock crab and green crab, measured between the two most distant points on the carapace [maximum carapace width (CW)], and the lengths of the spider crab (maximum carapace length) were measured to the nearest 0.1 cm using a vernier calliper. Plankton samples were taken in Faxaflói Bay (Figure 2) in 2012–2014 at the same stations as in our previous study (Gíslason et al., 2014), when the peak in larval abundance was expected (July–August), with standard Bongo nets, 60-cm ring diameter, 250-cm net length, and 500-µm mesh size (Hydro-bios Apparatebau GmbH®, Germany). The nets were towed at 10-m depth at 1.3 m s−1 for 10 min at each sampling station. Filtered volume was estimated with a flow meter (Hydro-Bios Kiel, Model 438 110) fitted on one of the net’s opening. Samples were immediately preserved in 10% formalin in seawater and later washed and preserved in 96% ethanol. Statistical analysis Variation in proportions of Atlantic rock crabs in Hvalfjörður and Kollafjörður was analysed with a linear regression of the relative frequencies against time and transformed with the arcsin of their square root (Sokal and Rohlf, 2008). Heterogeneity of the counts of different species within and among the sites were tested with a chi-square test. Size distributions of the rock crabs were analysed in each sampling area with the Shapiro–Wilk test (Shapiro and Wilk, 1965) to estimate deviation from normality, i.e. skewness (g1) and kurtosis (g2) (Sokal and Rohlf, 2008). One-way analysis of variance was used to test variability in size between years. All calculations and graphical analyses were conducted in R (version 3.6.2; R Development Core Team, 2020). Results Distribution The spread of the Atlantic rock crab in Iceland has been monitored through various indirect and direct sampling methods. After its first discovery in Hvalfjörður in 2006 (Southwest Iceland), it was reported in Breiðafjörður in 2008 (West Iceland), in Eyjafjörður in 2013 (North Iceland) and in Borgarfjörður Eystri in 2017 (East Iceland) (Figure 1), corresponding to >70% of the coastline. Trap fishing In total, 24 245 specimens of the Atlantic rock crab were caught in the four areas (Hvalfjörður, Kollafjörður, Skerjafjörður, and Borgarfjörður) during 2007–2019. A total of 10 708 were size measured (Table 1) and 24 172 identified to gender (Supplementary Table S2). In addition, 6451 specimens of the European green crab (6424 measured and 6383 to gender) and 3797 specimens of the spider crab were collected (3059 measured and 3494 to gender). Rock crab was the most abundant species in the trap catches in all four sampling areas (Figure 3a). The proportion of rock crabs was significantly different between sampling areas in 2013 (p < 0.001) and ranged from 64% in Kollafjörður to 99% in Borgarfjörður. Figure 3. Open in new tabDownload slide Proportion of the Atlantic rock crab (Cancer irroratus), European green crab (Carcinus maenas), and spider crab (Hyas araneus) in trap catches in (a) the four sampling areas: Hvalfjörður (H), Kollafjörður (K), Skerjafjörður (S), and Borgarfjörður (B), with total catch in each area; (b) Hvalfjörður in 2007–2019; and (c) Kollafjörður in 2011–2019. No sampling was carried out in Hvalfjörður and Kollafjörður in 2016. Figure 3. Open in new tabDownload slide Proportion of the Atlantic rock crab (Cancer irroratus), European green crab (Carcinus maenas), and spider crab (Hyas araneus) in trap catches in (a) the four sampling areas: Hvalfjörður (H), Kollafjörður (K), Skerjafjörður (S), and Borgarfjörður (B), with total catch in each area; (b) Hvalfjörður in 2007–2019; and (c) Kollafjörður in 2011–2019. No sampling was carried out in Hvalfjörður and Kollafjörður in 2016. The average number of rock crabs per trap on the transect (10–60 m depth) in the inner part of Hvalfjörður in 2007–2015 (June–October) averaged 7.7 crabs/trap (±1.2 SD), showing no clear trend in the catch between years. The average catch has since increased significantly from an average catch of 13.8 crabs/trap in 2017 to 24 crabs/trap in 2019 (p < 0.001). However, the proportion of rock crabs in the total catch in Hvalfjörður increased significantly (p < 0.001) over the entire study period from 55% in 2007 to >95% in 2018 and 2019 (Figure 3b). In contrast, the proportional abundance of rock crab in Kollafjörður (sampling in 2011–2019) dropped significantly (p < 0.01) from 71% in 2011 to 52% in 2013 but increased steadily after that up to 87% in 2019 (Figure 3c). Size frequency distribution and average sizes The overall size frequency distribution and average size of the rock crab differed considerably among study areas (Table 1, Supplementary Figure S1). Significant size difference was observed among years for both males (p < 0.001) and females (p < 0.05), where the average CW fluctuated around the mean without clear trends over time (Figure 4). The variance in size was greatest in 2011 for both genders (males: 2.2 cm; females: 1 cm) but was ca. 1 and 0.5 cm for the males and females, respectively, in all other years (Figure 4). Significant size difference was observed for both males and females among sampling areas (p < 0.001, Figure 5). The average size of the males was lowest in Kollafjörður (10.4 cm CW), but highest in Borgarfjörður (12.8 cm CW). Exceptionally large rock crab males were caught in Borgarfjörður in 2013, the largest measuring 15.3 CW and 577 g in wet weight (Table 1). The male rock crab size frequency distribution in Skerjafjörður was similar to that in Hvalfjörður (Table 1). Figure 4. Open in new tabDownload slide Difference in CW (with the median size ± standard deviation) with years in Hvalfjörður, Southwest Iceland, for both genders of the Atlantic rock crab (Cancer irroratus), European green crab (Carcinus maenas), and CL of the spider crab (Hyas araneus). No sampling was carried out in 2016. CL, carapace length. Figure 4. Open in new tabDownload slide Difference in CW (with the median size ± standard deviation) with years in Hvalfjörður, Southwest Iceland, for both genders of the Atlantic rock crab (Cancer irroratus), European green crab (Carcinus maenas), and CL of the spider crab (Hyas araneus). No sampling was carried out in 2016. CL, carapace length. Figure 5. Open in new tabDownload slide Difference in CW (with the median size ± standard deviation) for both genders of the Atlantic rock crab (Cancer irroratus) at the four sampling areas: Borgarfjörður (B), Kollafjörður (K), Hvalfjörður (H), and Skerjafjörður (S). Figure 5. Open in new tabDownload slide Difference in CW (with the median size ± standard deviation) for both genders of the Atlantic rock crab (Cancer irroratus) at the four sampling areas: Borgarfjörður (B), Kollafjörður (K), Hvalfjörður (H), and Skerjafjörður (S). The size frequency distribution of rock crabs in Hvalfjörður from 2007 to 2019 fluctuated over years (p < 0.05) with no temporal trend. It showed differences among years, with a slight increase in average size from 2007 to 2009, a decline until 2012, and an increase again in 2013 and since then has been relatively stable (Figure 4, Supplementary Figure S2). The multimodality indicates the occurrence of several cohorts of different ages among the adult crabs. The average male size may be influenced by an apparently strong cohort entering the fisheries in 2011 (Supplementary Figure S2). The shape of the size frequency data of the rock crab also varied among regions. In all four sampling areas, samples deviated from a normal distribution for males (p < 0.001), being skewed to the left (Supplementary Table S3). When samples from each year were analysed separately, they were all skewed to the left for males, but differed from a normal distribution to skewness to either sides for females (Supplementary Table S3). Males in Hvalfjörður and Borgarfjörður also showed an overall significant leptokurtic distribution (p < 0.001), whereas males in Kollafjörður showed a platykurtic distribution (p < 0.001). Females in Hvalfjörður and Kollafjörður exhibited a significant leptokurtic distribution (p < 0.001) (Supplementary Table S3). The average male and female CW of green crabs fluctuated over years (p < 0.05), with no temporal trend (Figure 4). A strong cohort of green crabs was observed entering the fishery in 2011 (Supplementary Figure S2). However, a significant reduction in size with time was observed for male spider crabs (b = −0.03, p < 0.05), female spider crabs (b = −0.06, p < 0.05), and female rock crabs (b = −0.08, p < 0.001). Sex ratio Overall, males outnumbered females in trap catches for all three species except for the spider crab in 2007 (Table 1; Supplementary Table S2). The proportion of males per year ranged from 71 to 95% for the rock crab, 63–97% for the green crab, and 44–91% for the spider crab. Significant difference was observed in sex ratio of rock crab between both years and months (p < 0.001) (Figure 6). No significant difference in sex ratio was observed over months for the green crab, but they varied for the spider crab (p < 0.001), where females were most common in May (Supplementary Table S2). Figure 6. Open in new tabDownload slide Proportion of male (black bars) and females (grey bars) for the Atlantic rock crab (Cancer irroratus), European green crab (Carcinus maenas), and spider crab (Hyas araneus) per year in Hvalfjörður during 2007–2019. No sampling was carried out in 2016. Figure 6. Open in new tabDownload slide Proportion of male (black bars) and females (grey bars) for the Atlantic rock crab (Cancer irroratus), European green crab (Carcinus maenas), and spider crab (Hyas araneus) per year in Hvalfjörður during 2007–2019. No sampling was carried out in 2016. Berried females Of the 3870 rock crab females caught in 2007–2019, only 86 (2.2%) carried eggs. Berried females were caught from May to October (Supplementary Table S2). Fourteen of the berried females (16%) had undeveloped eggs, 26 had well-developed eggs (30%), but about half (41 individuals, 47%) had recently lost/hatched their eggs, only carrying egg remnants. Only 14 (3%) berried green crabs were caught, carrying either orange or brown eggs. More than half of all 641 captured female spider crabs caught from May to December were carrying eggs (71%) (Supplementary Table S2). Half of the berried spider crab females were carrying orange eggs (50%), 19% were carrying brown eggs, and 2% had egg remnants. Larval abundance Larvae of two of the three species, rock crab and green crab were present in plankton samples during the summer months (June–August). The mean abundance of rock crab larvae in July ranged from 0.5 to 2.3 larvae m−3 in 2007–2008 and from 7.6 to 10.7 larvae m−3 in 2012–2014, with the highest abundance in 2012 and 2014 (Figure 7). Proportion of Atlantic rock crab larvae was significantly higher (p < 0.05) at the sampling sites in Faxaflói Bay (B3–B6) than within Hvalfjörður (B1 and B2); however, abundance varied extensively (Table 2, Figure 8) and no significant differences in proportions of larval species were detected among years within sampling sites. Figure 7. Open in new tabDownload slide Mean density (larvae m − 3) of three brachyuran crab species: the Atlantic rock crab (Cancer irroratus), green crab (Carcinus maenas), and spider crab (Hyas araneus) in July in Faxaflói Bay, Southwest Iceland, in 2007–2008 and 2012–2014. Figure 7. Open in new tabDownload slide Mean density (larvae m − 3) of three brachyuran crab species: the Atlantic rock crab (Cancer irroratus), green crab (Carcinus maenas), and spider crab (Hyas araneus) in July in Faxaflói Bay, Southwest Iceland, in 2007–2008 and 2012–2014. Figure 8. Open in new tabDownload slide Proportion of Atlantic rock crab (Cancer irroratus) larvae in relation to other brachyuran decapod larvae per sampling station in July in 2007–2008 and 2012–2014 in Faxaflói Bay, Southwest Iceland. Figure 8. Open in new tabDownload slide Proportion of Atlantic rock crab (Cancer irroratus) larvae in relation to other brachyuran decapod larvae per sampling station in July in 2007–2008 and 2012–2014 in Faxaflói Bay, Southwest Iceland. Table 2. Abundance of the Atlantic rock crab (Cancer irroratus) and the European green crab (Carcinus maenas) larvae in Faxaflói Bay, Southwest Iceland, in July 2007–2008 and 2012–2014. Year . N . Station . Abundance (larvae m−3) . Proportion of C. irroratus (%) . C. irroratus . C. maenas . 2007 2 B1 0.3 1.7 16 B2 1.1 0.8 58 B3 0.3 1.7 14 B4 0.4 0.0 100 B5 – – – B6 – – – 2008 3 B1 0.6 4.1 12 B2 1.2 0.5 71 B3 5.9 0.9 87 B4 0.04 0.01 74 B5 2.7 1.3 67 B6 1.8 0.5 85 2012 1 B1 0.5 0.3 58 B2 2.9 0.1 98 B3 0.8 0.2 77 B4 – – – B5 39.5 1.9 95 B6 4.2 0.3 96 2013 1 B1 6.1 33.0 16 B2 1.8 7.3 20 B3 5.8 7.8 43 B4 14.2 3.6 80 B5 10.1 9.6 51 B6 – – – 2014 1 B1 1.6 1.5 52 B2 2.0 4.0 33 B3 2.8 1.4 67 B4 2.9 0.0 100 B5 41.6 12.6 77 B6 – – – Year . N . Station . Abundance (larvae m−3) . Proportion of C. irroratus (%) . C. irroratus . C. maenas . 2007 2 B1 0.3 1.7 16 B2 1.1 0.8 58 B3 0.3 1.7 14 B4 0.4 0.0 100 B5 – – – B6 – – – 2008 3 B1 0.6 4.1 12 B2 1.2 0.5 71 B3 5.9 0.9 87 B4 0.04 0.01 74 B5 2.7 1.3 67 B6 1.8 0.5 85 2012 1 B1 0.5 0.3 58 B2 2.9 0.1 98 B3 0.8 0.2 77 B4 – – – B5 39.5 1.9 95 B6 4.2 0.3 96 2013 1 B1 6.1 33.0 16 B2 1.8 7.3 20 B3 5.8 7.8 43 B4 14.2 3.6 80 B5 10.1 9.6 51 B6 – – – 2014 1 B1 1.6 1.5 52 B2 2.0 4.0 33 B3 2.8 1.4 67 B4 2.9 0.0 100 B5 41.6 12.6 77 B6 – – – N is the number of samples taken yearly in July at each station (see Figure 1). Open in new tab Table 2. Abundance of the Atlantic rock crab (Cancer irroratus) and the European green crab (Carcinus maenas) larvae in Faxaflói Bay, Southwest Iceland, in July 2007–2008 and 2012–2014. Year . N . Station . Abundance (larvae m−3) . Proportion of C. irroratus (%) . C. irroratus . C. maenas . 2007 2 B1 0.3 1.7 16 B2 1.1 0.8 58 B3 0.3 1.7 14 B4 0.4 0.0 100 B5 – – – B6 – – – 2008 3 B1 0.6 4.1 12 B2 1.2 0.5 71 B3 5.9 0.9 87 B4 0.04 0.01 74 B5 2.7 1.3 67 B6 1.8 0.5 85 2012 1 B1 0.5 0.3 58 B2 2.9 0.1 98 B3 0.8 0.2 77 B4 – – – B5 39.5 1.9 95 B6 4.2 0.3 96 2013 1 B1 6.1 33.0 16 B2 1.8 7.3 20 B3 5.8 7.8 43 B4 14.2 3.6 80 B5 10.1 9.6 51 B6 – – – 2014 1 B1 1.6 1.5 52 B2 2.0 4.0 33 B3 2.8 1.4 67 B4 2.9 0.0 100 B5 41.6 12.6 77 B6 – – – Year . N . Station . Abundance (larvae m−3) . Proportion of C. irroratus (%) . C. irroratus . C. maenas . 2007 2 B1 0.3 1.7 16 B2 1.1 0.8 58 B3 0.3 1.7 14 B4 0.4 0.0 100 B5 – – – B6 – – – 2008 3 B1 0.6 4.1 12 B2 1.2 0.5 71 B3 5.9 0.9 87 B4 0.04 0.01 74 B5 2.7 1.3 67 B6 1.8 0.5 85 2012 1 B1 0.5 0.3 58 B2 2.9 0.1 98 B3 0.8 0.2 77 B4 – – – B5 39.5 1.9 95 B6 4.2 0.3 96 2013 1 B1 6.1 33.0 16 B2 1.8 7.3 20 B3 5.8 7.8 43 B4 14.2 3.6 80 B5 10.1 9.6 51 B6 – – – 2014 1 B1 1.6 1.5 52 B2 2.0 4.0 33 B3 2.8 1.4 67 B4 2.9 0.0 100 B5 41.6 12.6 77 B6 – – – N is the number of samples taken yearly in July at each station (see Figure 1). Open in new tab Discussion The Atlantic rock crab is now well established in Icelandic waters. Since first recorded in 2006 (Gíslason et al., 2014), the rock crab has spread rapidly and is now found along >70% of the coastline, i.e. clockwise from Faxaflói in Southwest Iceland to East Iceland. The population is still in a growth phase, as seen in a rapid increase in distributional range, and is now found in densities that are among the highest reported for the species in its native range (Gíslason et al., 2017). According to the present study, the rock crab is the most abundant brachyuran crab species on soft-bottom habitats in Southwest Iceland. The proportion of rock crab in trap catches increased from 55% in Hvalfjörður in 2007 to >95% in 2018 and 2019, and the abundance of rock crab larvae in July, when the annual peak was seen in larval abundance (Gíslason et al. 2014), was sevenfold higher in 2014 than in 2007. Size distribution The mean size of rock crabs in Hvalfjörður fluctuated between years, with no clear trend over time, showing a pattern in fluctuations similar to that seen for green crab. This indicates that the same environmental factors are affecting both species. The size distribution of the rock crab was left-skewed and leptokurtic. Right skewness, however, applies in general for most species on a large geographical scale (Kozlowski and Gawelczyk, 2002). This might be due either to the crab still colonizing the local area or simply that the trap catches are not representing the smaller individuals in the population. It is known that crab traps are both size and sex selective (Workman et al., 2002; Smith et al., 2004; Hernaez et al., 2012), and trap fishing may also be affected by moulting period, reproductive status (berried females), and health condition of the animals. The much higher catchability of male rock crabs compared to females in trap fishing in Iceland, as for many decapods elsewhere, is presumably because the commercial traps used are more efficient at retaining larger crabs, which are predominantly males (Smith and Jamieson, 1991; Workman et al., 2002). Presence of large aggressive individuals in the traps has also been considered to affect the catchability of smaller animals such as juveniles and females by restricting their entrance into the traps (Fischer and Wolff, 2006). The different size frequency distributions of the rock crab, seen here partly in very large crabs in Borgarfjörður, may indicate that favourable local conditions in the absence of other crabs may lead either to larger sizes or that an additional moult would occur. The former explanation is probably more likely, as a small increase in size increments between moults under favourable conditions could easily lead to the ca. 1 cm increase in the size of males in Borgarfjörður compared to other areas. Furthermore, to our best knowledge, the maximum size of rock crab males and females in Iceland is substantially larger than documented in its native range (Bigford, 1979; Robichaud et al., 2000; Robichaud and Frail, 2006). Little information is available on the number of moults undertaken by the Atlantic rock crab, except for Reilly’s (1975) estimator on age and number of moults of the rock crab. Gíslason et al. (2017) found size increments between moults of adult rock crabs in Kollafjörður to be ca. 2 cm. Berried females Proportion of berried females varied among the three species. Berried rock crabs were caught from May to October, which is similar to what is seen in the native habitat of the crab in North America. In Canada and Maine, females with well-developed eggs or egg remnants are mainly seen from June to August (Krouse, 1972; DFO, 2008), but from March to June for more southern rock crab populations (Reilly, 1975; Reilly and Saila, 1978). The proportion of berried green crabs was low compared to rock crabs and was caught from June to October. Proportion of berried green crabs varies greatly between regions, where berried green crabs have been caught from April to August in Maine, United States (Berrill, 1982), from February to June in Swansea, United Kingdom (Naylor, 1962), and year-round in Portugal (Baeta et al., 2005). The proportion of berried spider crabs was much higher than for the other two species, or >60% in total, and they were captured from May to December, which is consistent with Einarsson (1988), where spider crab females were reported to carry undeveloped (orange) and developed (brown) eggs year-round in Iceland. Environmental conditions and larval abundance Environmental conditions in Southwest Iceland seem to be favourable for larval development of the rock crab. Gíslason et al. (2014) showed that rock crab larvae are present in surface waters from May to November, with a peak period in July when temperature is near maximum in Icelandic waters. Since 2007 and 2008, larval abundance has increased significantly up to sevenfold in 2014, as shown in the present study. Similar results were observed for larval density of the green crab, with significant increase in 2013 and 2014 from previous years (Gíslason et al., 2014). This may reflect the expansion phase of the colonization process, but as for the green crab, environmental conditions have been improving, which are likely linked to the large-scale changes occurring in the North Atlantic in recent years (Anonymous, 2004), which has led to noticeable changes in Icelandic marine ecosystems (Astthorsson and Palsson, 2006; Astthorsson et al., 2007, 2012; Jochumsen et al., 2016). Spider crab larvae were absent in plankton samples in July. This is in good agreement with what is known about spider crabs in the North Sea where hatching occurs mainly in late winter to early spring (Anger, 1983a; Kunisch and Anger, 1984) and to results previously reported by Gíslason et al. (2014) in Iceland. Impact and competition The rock crab expansion phase, which is marked by an increasing spread rate, fits well with the second classification of the invasion process by Arim et al. (2006), indicating that the rock crab has enough resources in its new habitat in Iceland and that both competition and predation are weak or lacking. In its native range, the rock crab has been shown to be a major player in structuring benthic communities by influencing species composition and abundance. In an experiment where the crab was excluded from a region where it was the most abundant and frequently encountered predator, the polychaete Pholoe tecta and the clam Macoma calcarea became the dominant benthic infauna and the overall species richness increased (Quijon and Snelgrove, 2005a, b). The effects of the rock crab colonization in the Icelandic ecosystem are still unforeseen, though it is highly likely that a large species, like the rock crab, found in such high densities has a significant impact on coastal biota. By outnumbering its rival native species, the spider crab (H. araneus) and the European green crab (C. maenas), on soft-bottom substrates, the rock crab shows its fitness. Comparative density studies on the rock crab and green crab in North America on various substrates have shown that both species are found in highest densities on sandy bottom substrates (Fogarty, 1976). Belair and Miron (2009) showed that predation rates and stomach contents of the rock crab and the green crab remained unaffected by the presence of each other and that the two species avoided each other passively and actively, which could enable them to coexist and reduce competition. The rock crab is both bigger and bulkier than both the spider crab and the green crab, which makes it a harder competitor for food and shelter. The size advantage is also likely to make the rock crab less vulnerable to predation. However, despite the smaller size of the green crab and that it is not as well adapted to low water temperatures (<10°C) as the other two species (Anger, 1983b; Belair and Miron, 2009), which may inhibit feeding (Berrill, 1982) and moulting (Audet et al., 2003), the green crab is considered to be the most aggressive and generally the fittest of the three species being listed among the most successful alien species in the world (Klassen and Locke, 2007). It has, for example, invaded North America (Cohen et al., 1995; Audet et al., 2003, 2008), where it is regarded as a potential threat to native species. Abiotic factors in Iceland, whether temperature or others, constrain the green crab distribution to the southwest and west coasts of Iceland (Ingólfsson, 1996). The rock crab, on the other hand, new in the ecosystem, has spread fast along the Icelandic coastline and has now established itself in the Westfjords and North Iceland, where the green crab is absent. Of the three species, coexistence has only been studied to some degree between the rock crab and the green crab. Matheson and Gagnon (2012) observed that medium-sized rock crabs (which were as big as large green crabs) won <20% of their contests with large green crabs, whereas large rock crabs won as many contests, demonstrating that large green crabs can outcompete rock crabs even if the latter is larger. Conclusion Whether it is fitness, size advantage over the native species, or preference for the soft-bottom substrates that make the rock crab more abundant in such habitats in Iceland, it is clear that the rock crab is thriving well. The successful establishment of the rock crab in Icelandic waters is of concern as the magnitude of environmental and ecological effects is not fully known at this stage. The recent record of a single specimen of rock crab in southern Kattegat, Sweden (Berggren, 2019), raises wariness of further spread and establishment of the species, and the possible role of Iceland as a stepping stone for the rock crab to mainland Europe. Acknowledgements We thank all the people who have assisted us with data collection through the years. Funding This work was supported by the University of Iceland Research Fund (HI11090118 and HI14090070); Suðurnes Regional Development Fund (31/2011 and 21/2012); and the Icelandic Ministry of Fisheries and Agriculture (F 12 012-12). References Anger K. 1983 a. Moult cycle and morphogenesis in Hyas araneus larvae (Decapoda, Majidae), reared in the laboratory . Helgoland Marine Research , 36 : 285 – 302 . Google Scholar OpenURL Placeholder Text WorldCat Anger K. 1983 b. Temperature and the larval development of Hyas araneus L. (Decapoda, Majidae); extrapolation of laboratory data to field conditions . Journal of Experimental Marine Biology and Ecology , 69 : 203 – 215 . Google Scholar Crossref Search ADS WorldCat Anonymous. 2004 . Hydrographic Status Report 2003. Report of the Working Group on Oceanic Hydrography (WGOH), 29 March–1 April 2004, Southampton, UK. ICES Document CM 2004/C: 06. 182 pp. Arim M. , Abades S. R., Neill P. E., Lima M., Marquet P. A. 2006 . Spread dynamics of invasive species . Proceedings of the National Academy of Sciences of the United States of America , 103 : 374 – 378 . Google Scholar Crossref Search ADS PubMed WorldCat Astthorsson O. S. , Gislason A., Jonsson S. 2007 . Climate variability and the Icelandic marine ecosystem . Deep Sea Research Part II: Topical Studies in Oceanography , 54 : 2456 – 2477 . Google Scholar Crossref Search ADS WorldCat Astthorsson O. S. , Palsson J. 2006 . New fish records and records of rare southern fish species in Icelandic waters in the warm period 1996–2005. ICES Document CM 2006/C: 20. 22 pp. Astthorsson O. S. , Valdimarsson H., Gudmundsdottir A., Óskarsson G. J. 2012 . Climate-related variations in the occurrence and distribution of mackerel (Scomber scombrus) in Icelandic waters . ICES Journal of Marine Science , 69 : 1289 – 1297 . Google Scholar Crossref Search ADS WorldCat Audet D. , Davis D. S., Miron G., Moriyasu M., Benhalima K., Campbell R. 2003 . Geographical expansion of a nonindigenous crab, Carcinus maenas (L.), along the Nova Scotian shore into the southeastern Gulf of St. Lawrence, Canada . Journal of Shellfish Research , 22 : 255 – 262 . Google Scholar OpenURL Placeholder Text WorldCat Audet D. , Miron G., Moriyasu M. 2008 . Biological characteristics of a newly established green crab (Carcinus maenas) population in the southern Gulf of St . Lawrence, Canada. Journal of Shellfish Research , 27 : 427 – 441 . Google Scholar Crossref Search ADS WorldCat Baeta A. , Cabral H. N., Neto J. M., Marques J. C., Pardal M. A. 2005 . Biology, population dynamics and secondary production of the green crab Carcinus maenas (L.) in a temperate estuary . Estuarine Coastal and Shelf Science , 65 : 43 – 52 . Google Scholar Crossref Search ADS WorldCat Bax N. , Williamson A., Aguero M., Gonzalez E., Geeves W. 2003 . Marine invasive alien species: a threat to global biodiversity . Marine Policy , 27 : 313 – 323 . Google Scholar Crossref Search ADS WorldCat Belair M. C. , Miron G. 2009 . Predation behaviour of Cancer irroratus and Carcinus maenas during conspecific and heterospecific challenges . Aquatic Biology , 6 : 41 – 49 . Google Scholar Crossref Search ADS WorldCat Berggren M. 2019 . Första fyndet av stenkrabba på den europeiska kontinenten . Fauna och Flora , 114 : 16 . Google Scholar OpenURL Placeholder Text WorldCat Berrill M. 1982 . The life cycle of the green crab Carcinus maenas at the northern end of its range . Journal of Crustacean Biology , 2 : 31 – 39 . Google Scholar Crossref Search ADS WorldCat Bigford T. E. 1979 . Synopsis of Biological Data on the Rock Crab , Cancer irroratus Say. NOAA Technical Report NMFS Circcular 426 . 26 pp. Google Scholar OpenURL Placeholder Text WorldCat Boudreau S. A. , Worm B. 2012 . Ecological role of large benthic decapods in marine ecosystems: a review . Marine Ecology Progress Series , 469 : 195 – 213 . Google Scholar Crossref Search ADS WorldCat Brickman D. 2006 . Risk assessment model for dispersion of ballast water organisms in shelf seas . Canadian Journal of Fisheries and Aquatic Sciences , 63 : 2748 – 2759 . Google Scholar Crossref Search ADS WorldCat Carlton J. T. 2003 . Community assembly and historical biogeography in the North Atlantic Ocean: the potential role of human-mediated dispersal vectors . Hydrobiologia , 503 : 1 – 8 . Google Scholar Crossref Search ADS WorldCat Carlton J. T. 2009 . Deep invasion ecology and the assembly of communities in historical time. In Biological Invasions in Marine Ecosystems , pp. 13 – 56 . Ed. by Rilov G., Crooks J. A.. Springer-Verlag , Berlin . 641 pp. Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Cohen A. N. , Carlton J. T. 1998 . Accelerating invasion rate in a highly invaded estuary . Science , 279 : 555 – 558 . Google Scholar Crossref Search ADS PubMed WorldCat Cohen A. N. , Carlton J. T., Fountain M. C. 1995 . Introduction, dispersal and potential impacts of the green crab Carcinus maenas in San-Francisco bay, California . Marine Biology , 122 : 225 – 237 . Google Scholar OpenURL Placeholder Text WorldCat DFO. 2008 . Assessment of the rock crab (Cancer irroratus) fishery in the Southern Gulf of St. Lawrence lobster fishing areas (LFA’s) 23, 24, 25, 26A & 26B for 2000 to 2006. DFO Canadian Science Advisory Secretariat Science Advisory Report 2008/022. Dvoretsky A. G. , Dvoretsky V. G. 2013 . Population dynamics of the invasive lithodid crab, Paralithodes camtschaticus, in a typical bay of the Barents Sea . ICES Journal of Marine Science , 70 : 1255 – 1262 . Google Scholar Crossref Search ADS WorldCat Einarsson S. T. 1988 . The distribution and density of the common spider crab (Hyas araneus) in Icelandic waters . ICES Document CM 1988/K : 28 . 25 pp. Google Scholar OpenURL Placeholder Text WorldCat Fischer S. , Wolff M. 2006 . Fisheries assessment of Callinectes arcuatus (Brachyura, Portunidae) in the Gulf of Nicoya, Costa Rica . Fisheries Research , 77 : 301 – 311 . Google Scholar Crossref Search ADS WorldCat Fogarty M. J. 1976 . Competition and resource partitioning in two species of Cancer (Crustacea, Brachyura). MSc thesis, University of Rhode Island, Kingston. 93 pp. Gíslason Ó. S. , Halldórsson H. P., Pálsson M. F., Pálsson S., Davíðsdóttir B., Svavarsson J. 2014 . Invasion of the Atlantic rock crab (Cancer irroratus) at high latitudes . Biological Invasions , 16 : 1865 – 1877 . Google Scholar Crossref Search ADS WorldCat Gíslason Ó. S. , Jónasson J. P., Pálsson S., Svavarsson J., Halldórsson H. P. 2017 . Population density and growth of the newly introduced Atlantic rock crab Cancer irroratus Say, 1817 (Decapoda, Brachyura) in Iceland: a four-year mark-recapture study . Marine Biology Research , 13 : 198 – 209 . Google Scholar Crossref Search ADS WorldCat Gíslason Ó. S. , Pálsson S., McKeown N. J., Halldórsson H. P., Shaw P. W., Svavarsson J. 2013 . Genetic variation in a newly established population of the Atlantic rock crab Cancer irroratus in Iceland . Marine Ecology Progress Series , 494 : 219 – 230 . Google Scholar Crossref Search ADS WorldCat Hänfling B. , Edwards F., Gherardi F. 2011 . Invasive alien Crustacea: dispersal, establishment, impact and control . Biocontrol , 56 : 573 – 595 . Google Scholar Crossref Search ADS WorldCat Hernaez P. , Rombenso A., Pinheiro M. A. A., Simoes N. 2012 . Population structure and sexual maturity of the calico box crab Hepatus epheliticus Linnaeus (Brachyura, Hepatidae) from Yucatan Peninsula, Mexico . Latin American Journal of Aquatic Research , 40 : 480 – 486 . Google Scholar Crossref Search ADS WorldCat Ingólfsson A. 1992 . The origin of the rocky shore fauna of Iceland and the Canadian Maritimes . Journal of Biogeography , 19 : 705 – 712 . Google Scholar Crossref Search ADS WorldCat Ingólfsson A. 1996 . The distribution of intertidal macrofauna on the coasts of Iceland in relation to temperature . Sarsia , 81 : 29 – 44 . Google Scholar Crossref Search ADS WorldCat Jochumsen K. , Schnurr S. M., Quadfasel D. 2016 . Bottom temperature and salinity distribution and its variability around Iceland . Deep-Sea Research Part I-Oceanographic Research Papers , 111 : 79 – 90 . Google Scholar Crossref Search ADS WorldCat Jungblut S. , Beermann J., Boos K., Saborowski R., Hagen W. 2017 . Population development of the invasive crab Hemigrapsus sanguineus (De Haan, 1853) and its potential native competitor Carcinus maenas (Linnaeus, 1758) at Helgoland (North Sea) between 2009 and 2014 . Aquatic Invasions , 12 : 85 – 96 . Google Scholar Crossref Search ADS WorldCat Klassen G. , Locke A. 2007 . A biological synopsis of the European green crab, Carcinus maenas. Canadian Manuscript Report Fisheries and Aquatic Science no. 2818. 75 pp. Kozlowski J. , Gawelczyk A. T. 2002 . Why are species’ body size distributions usually skewed to the right? Functional Ecology , 16 : 419 – 432 . Google Scholar Crossref Search ADS WorldCat Krouse J. S. 1972 . Some life-history aspects of rock crab, Cancer irroratus, in Gulf of Maine . Journal of the Fisheries Research Board of Canada , 29 : 1479 – 1482 . Google Scholar Crossref Search ADS WorldCat Kunisch M. , Anger K. 1984 . Variation in development and growth rates of larval and juvenile spider crabs Hyas araneus reared in the laboratory . Marine Ecology Progress Series , 15 : 293 – 301 . Google Scholar Crossref Search ADS WorldCat Makino W. , Miura O., Kaiser F., Geffray M., Katsube T., Urabe J. 2018 . Evidence of multiple introductions and genetic admixture of the Asian brush-clawed shore crab Hemigrapsus takanoi (Decapoda: Brachyura: Varunidae) along the Northern European coast . Biological Invasions , 20 : 1369 – 1369 . Google Scholar Crossref Search ADS WorldCat Matheson K. , Gagnon P. 2012 . Temperature mediates non-competitive foraging in indigenous rock (Cancer irroratus Say) and recently introduced green (Carcinus maenas L.) crabs from Newfoundland and Labrador . Journal of Experimental Marine Biology and Ecology , 414–415 : 6 – 18 . Google Scholar Crossref Search ADS WorldCat Naylor E. 1962 . Seasonal changes in a population of Carcinus maenas (L.) in the littoral zone . Journal of Animal Ecology , 31 : 601 – 609 . Google Scholar Crossref Search ADS WorldCat Quijon P. A. , Snelgrove P. V. R. 2005 a. Differential regulatory roles of crustacean predators in a sub-arctic, soft-sediment system . Marine Ecology Progress Series , 285 : 137 – 149 . Google Scholar Crossref Search ADS WorldCat Quijon P. A. , Snelgrove P. V. R. 2005 b. Predation regulation of sedimentary faunal structure: potential effects of a fishery-induced switch in predators in a Newfoundland sub-Arctic fjord . Oecologia , 144 : 125 – 136 . Google Scholar Crossref Search ADS PubMed WorldCat R Core Team. 2020 . R: a language and environment for statistical computing. R Foundation for Statistical Computing , Vienna , Austria. http://www.R-project.org/. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Reilly P. N. 1975 . The biology and ecology of juvenile and adult rock crabs, Cancer irroratus Say, in southern New England waters. MSc thesis, University of Rhode Island, Kingston. 146 pp. Reilly P. N. , Saila S. B. 1978 . Biology and ecology of rock crab, Cancer irroratus Say, 1817, in, Southern New England waters (Decapoda, Brachyura). Crustaceana , 34 : 121 – 140 . Google Scholar OpenURL Placeholder Text WorldCat Robichaud D. A. , Frail C. 2006 . Development of Jonah crab, Cancer borealis, and rock crab, Cancer irroratus, fisheries in the Bay of Fundy (LFAs 35-38) and off southwest Nova Scotia (LFA 34): from exploratory to commercial status (1995-2004). Canadian Manuscript Report of Fisheries and Aquatic Sciences 2775. 48 pp. Robichaud D. A. , Lawton P., Strong M. B. 2000 . Exploratory fisheries for rock crab, Cancer irroratus, and Jonah crab Cancer borealis, in Canadian Lobster Fishing Areas 34, 35, 36 and 38. Fisheries and Oceans Science, Canadian Stock Assessment Secretariat Research Document 2000/051. 27 pp. Roche D. , Torchin M. 2007 . Established population of the North American Harris mud crab, Rhithropanopeus harrisii (Gould 1841) (Crustacea: Brachyura: Xanthidae) in the Panama Canal . Aquatic Invasions , 2 : 155 – 161 . Google Scholar Crossref Search ADS WorldCat Schrimpf A. , Schmidt T., Schulz R. 2014 . Invasive Chinese mitten crab (Eriocheir sinensis) transmits crayfish plague pathogen (Aphanomyces astaci) . Aquatic Invasions , 9 : 203 – 209 . Google Scholar Crossref Search ADS WorldCat Seebens H. , Gastner M. T., Blasius B. 2013 . The risk of marine bioinvasion caused by global shipping . Ecology Letters , 16 : 782 – 790 . Google Scholar Crossref Search ADS PubMed WorldCat Shapiro S. S. , Wilk M. B. 1965 . An analysis of variance test for normality (complete samples) . Biometrika , 52 : 591 – 611 . Google Scholar Crossref Search ADS WorldCat Smith B. D. , Jamieson G. S. 1991 . Movement, spatial distribution, and mortality of male and female Dungeness crab Cancer magister near Tofino, British Columbia . Fishery Bulletin , 89 : 137 – 148 . Google Scholar OpenURL Placeholder Text WorldCat Smith K. D. , Hall N. G., de Lestang S., Potter I. C. 2004 . Potential bias in estimates of the size of maturity of crabs derived from trap samples . ICES Journal of Marine Science , 61 : 906 – 912 . Google Scholar Crossref Search ADS WorldCat Sokal R. R. , Rohlf F. J. 2008 . Biometry. W. H. Freeman and Company , New York . 937 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Stefansdottir L. , Solmundsson J., Marteinsdottir G., Kristinsson K., Jonasson J. P. 2010 . Groundfish species diversity and assemblage structure in Icelandic waters during recent years of warming . Fisheries Oceanography , 19 : 42 – 62 . Google Scholar Crossref Search ADS WorldCat Workman G. D. , Phillips A., Scurrah F. E., Boutillier J. A., 2002 . A new fishery for grooved Tanner crab (Chionoecetes tanneri) off the coast of British Columbia, Canada. In Crabs in Cold Water Regions: Biology, Management, and Economics , pp. 439 – 456 . Ed. by Paul A. J., Dawe E. G., Elner R., Jamieson G. S., Kruse G. H., Otto R. S., Sainte-Marie B.. et al. . University of Alaska Sea Grant , Fairbanks . 876 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC © International Council for the Exploration of the Sea 2020. 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)

O’Leary, Cecilia, A;Thorson, James, T;Miller, Timothy, J;Nye, Janet, A

doi: 10.1093/icesjms/fsz215pmid: N/A

Abstract Fisheries managers use biological reference points (BRPs) as targets or limits on fishing and biomass to maintain productive levels of fish stock biomass. There are multiple ways to calculate BRPs when biological parameters are time varying. Using summer flounder (Paralichthys dentatus) as a case study, we investigated time-varying approaches in concert with climate-linked population models to understand the impact of environmentally driven variability in natural mortality, recruitment, and size-at-age on two commonly used BRPs [B0(t) and F35%(⁠ t ⁠)]. We used the following two approaches to calculate time-varying BRPs: dynamic-BRP and moving-average-BRP. We quantified the variability and uncertainty of different climate dependencies and estimation approaches, attributed BRP variation to variation in life-history processes, and evaluated how using different approaches impacts estimates of stock status. Results indicate that the dynamic-BRP approach using the climate-linked natural mortality model produced the least variable reference points compared to others calculated. Summer flounder stock status depended on the estimation approach and climate model used. These results emphasize that understanding climate dependencies is important for summer flounder reference points and perhaps other species, and careful consideration is warranted when considering what time-varying approach to use, ideally based upon simulation studies within a proposed set of management procedures. Introduction In fisheries management, an estimate of the fish stock state is compared to a biological reference point (BRP) to define the stock status. The BRP is typically related to a biologically sustainable population size. This comparison is critical to determine if a stock is overfished and whether any changes are needed in current management to meet targets (Gabriel and Mace, 1999; Quinn and Deriso, 1999; Collie and Gislason, 2001; Hilborn, 2002; Haltuch et al., 2008). Harvest control rules are guidelines that determine how much fishing can occur based on the current state of the system relative to target and/or limit reference points for the stock size and catch/fishing effort (Deroba and Bence, 2008). Harvest control rules attempt to balance biological, economic, and social sustainability and often use BRPs to define the limit and target for catch. The goal of a BRP-based management framework is to regulate fishing by setting a quota such as a fishing mortality rate or biomass threshold that is based on BRPs. For example, the fishing quota produced using the spawning potential ratio (SPR), a common metric used in fisheries management, is based on a preset fixed ratio of fished and unfished spawning biomass per recruit (SBPR) (Gabriel et al., 1989; Goodyear, 1993; Williams and Shertzer, 2003). The BRPs are based on parameters that reflect the long-term productivity of the fish stock, such as growth, recruitment, and mortality (Williams and Shertzer, 2003; Maunder, 2012), all of which can vary in response to a variety of factors. Productivity can vary over time for many different reasons, such as changes in the environment, available food, predation mortality, or fishing pressure, and this directly affects the management process (Jackson et al., 2001; Pitchford et al., 2005; Vert-pre et al., 2013; Nye et al., 2014; Pershing et al., 2015; Koenigstein et al., 2016; Collie et al., 2017; Stock et al., 2017; Barrow et al., 2018). Climate can impact productivity directly through physiological effects in response to temperature conditions and a change in the allocation of energy between growth and reproduction (Buckley et al., 2004; Baudron et al., 2011, 2014). Climate can also impact productivity indirectly through predator–prey interactions (such as a change in community composition and overlap of a predator or prey), fish behaviour, or recruitment via reproductive potential, timing of spawning or migration, and larval growth (Blanchard et al., 2005; Laurel et al., 2007). Many studies have found that fish population fluctuations are associated with large-scale climate variability (Lehodey et al., 2006; Brander, 2007, 2010; Holsman et al., 2012; Barange et al., 2014; Free et al., 2019). Not accounting for productivity changes may lead to errors in BRP estimation (Whitten et al., 2013; Audzijonyte et al., 2016; Karp et al., 2019). One approach to deal with temporal variability is to calculate BRPs using the average stock dynamics over the most recent 3–5 years or across the entire period being modelled. More recently, as part of the move to include a broader set of considerations in stock assessments, fisheries scientists have attempted to incorporate environmental effects into fisheries models (Hare et al., 2016; Tommasi et al., 2017). There are uncertainties regarding the fish stock’s dynamics and interaction with its environment (Hilborn and Walters, 1992; Quinn and Deriso, 1999; Maunder, 2012). Given that environmental variability changes vital rates, such as recruitment and natural mortality, and progress has been made to incorporate these processes explicitly in stock assessments, understanding BRPs in the context of climate is crucial (Mantua and Hare, 2002; Amar et al., 2009; Thorson et al., 2015; O’Leary et al., 2018). Demographic changes due to variation in fish vital rates can change BRPs and, thus, stock status and catch quotas. The magnitude of the effect of these changes on BRPs depends on the BRP used (Gerber and Heppell, 2004; Frisk et al., 2005; Thorson et al., 2015). Miller et al. (2018) found that incorporating environmental indices into a stock-assessment model not only influenced Georges Bank Atlantic cod (Gadus morhua) demographic estimates and BRPs but also increased the uncertainty in BRP estimates. Therefore, tailoring BRPs to climate state is a particularly important management strategy if future population conditions differ from past conditions due to a changing climate and a consequent regime shift; that is, an abrupt change within the population (Amar et al., 2009; Punt et al., 2016). If past stock conditions are used to estimate future stock, these catch targets are often unsustainable (Haltuch et al., 2009; Punt et al., 2016). If the influence of climate on a fish stock is understood and successfully modelled, there is still a choice on how to incorporate these temporal dynamics into BRP calculations that likely will influence the BRP estimate (Berger, 2019). The following two main approaches can be used to incorporate temporal dynamics (and consequently climate influence) into BRP calculation: (i) dynamic-BRPs and (ii) moving-average-BRPs. The dynamic-BRP is a generalization of the dynamic-B0 from MacCall et al. (1985) that calculates the SBPR following each cohort through time at a set fishing pressure (F*) given estimated parameters for stock productivity from an unfished population (MacCall et al., 1985; Punt and Donovan, 2007; Haltuch et al., 2009). The dynamic approach is referred to as “dynamic” because it generalizes “dynamic-B0”, i.e. where it projects dynamics from t−1 to t in the absence of fishing (to calculate biomass targets from B0) or with different fishing mortality rates (to calculate exploitation targets from SPR). The moving-average-BRP is an equilibrium approach assuming that natural mortality, growth, and other biological processes in year t (or a defined time-interval) are held constant (at their value in a single year or average across years) to calculate stock productivity (Punt and Donovan, 2007; Cordue, 2012). More informed management decisions can be achieved by understanding the properties and assumptions that led to the BRPs used in management, and whether these summaries of temporal and environmental dynamics are appropriate for stock management (Walters and Parma, 1996; McAllister et al., 1999; Punt and Donovan, 2007; Kolody et al., 2008; Kurota et al., 2010; Punt et al., 2016). Effective BRP-based management can be compromised if we miss changes in fish productivity due to using stock summary values that smooth over important temporal dynamics or making inaccurate assumptions about the relationship between a stock and its environment. Therefore, a comparison of different time-varying approaches to BRP estimation with different forms of climate dependencies in those calculations will help illuminate how variability and climate influence BRP-based management. The purpose of this article was twofold: (i) to examine the effects of climate dependency on BRP variability and uncertainty used in harvest control rules and (ii) to determine how sensitive estimates of BRPs are to the choice of “dynamic” or “moving-average” BRP approaches. We use summer flounder, a demersal flatfish found in the Northwest Atlantic, as an empirical case study to investigate BRPs. To evaluate the implications of both climate assumptions and estimation approaches on the BRP value output, our objectives were to (1) quantify the variability over time of climate-dependent BRPs vs. climate-independent BRPs (henceforth “variability”); (2) quantify the standard error for different BRPs and approaches (henceforth “uncertainty”); (3) determine whether moving-average- or dynamic-BRPs are more variable and uncertain over time; (4) attribute trends and variation in BRPs to variation in mortality, recruitment, and growth processes; and (5) evaluate how different methods impact estimates of stock status for summer flounder (Paralichthys dentatus). Given the importance of the BRPs in the management framework and the likely influence of both types of temporal variability and method of incorporation, we sought to demonstrate the implications of both the estimation approach and climate dependency on the reference BRP value. Methods We use two different temporally varying methods to provide information regarding the impact of BRP calculation choices on management reference points through direct comparison of BRP uncertainty and temporal variability differences. In this study, we aim to fill the gap in BRP documentation and methods’ development by comparing the differences in BRPs calculated from different climate relationships with biological parameters and different approaches to accounting for temporal dynamics. Population models with climate dependencies Posterior distributions for parameters used in BRP calculations were drawn from previously constructed hierarchical population models used to understand changes in past summer flounder abundance (Figures 1 and 2; O’Leary et al., 2018). Summer flounder is a data-rich stock where fishing pressure and environmental variability were shown to impact population dynamics. Moving-average- and dynamic-BRPs were compared in three population models that differed in their link to the environment: a climate-dependent natural mortality relationship (CM), a climate-dependent recruitment relationship (CR), and no relationship with environmental conditions (CI) (O’Leary et al., 2018). In this study, the climate covariate or Tt represents the Gulf Stream Index, the index used to describe climate conditions in the Northwest Atlantic, which is available for every modelled year t. The natural mortality and recruitment estimates for each model type are shown in Figure 2. These relationships were established and tested in O’Leary et al. (2018) that found overall the Gulf Stream Index provided information to improve the estimation of natural mortality and subsequently fishing mortality. The Gulf Stream Index represents the position of the north wall of the Gulf Stream and provides an integrative representation of oceanographic conditions of the Northeast US shelf. The Gulf Stream Index was used to represent the emergent properties of the local environment to which the organisms are responding. We did not consider models with multiple types of climate effects so that we could first determine the impact of the mechanism by which climate affected population dynamics in isolation. The models considered age-specific population processes and included both process and observation error. We used the following two data sources from 1982 to 2015 in the population models: (i) fisheries-independent annual bottom trawl surveys by the Northeast Fisheries Science Center (NEFSC; Azarovitz, 1981; Clark et al., 1997) and (ii) fisheries-dependent commercial and recreational landings of summer flounder (Burns et al., 1983) from the NEFSC fisheries database. Parameters of Bayesian hierarchical models were estimated from empirical data using Just Another Gibbs Sampler (Plummer, 2003) integrated through R version 3.2.4 (R Core Team, 2017) using R package “R2Jags” (Su and Yajima, 2012). As a group, we refer to these models described in the following sections as estimation models. Figure 1. Open in new tabDownload slide Weight-at-age from 1982 to 2015 used in all three estimation models. Data were extracted from the summer flounder stock-assessment tables (Terceiro 2016). Figure 1. Open in new tabDownload slide Weight-at-age from 1982 to 2015 used in all three estimation models. Data were extracted from the summer flounder stock-assessment tables (Terceiro 2016). Figure 2. Open in new tabDownload slide Estimates of recruitment (a) and age 4 natural mortality (b) from 1982 to 2015 for the three climate models, CI, CM, and CR, surrounded by 95% credible intervals. Figure 2. Open in new tabDownload slide Estimates of recruitment (a) and age 4 natural mortality (b) from 1982 to 2015 for the three climate models, CI, CM, and CR, surrounded by 95% credible intervals. Including the effects of climate on population dynamics The three population models followed a general structure described below as a simpler version of the summer flounder stock-assessment model, with variations in either the natural mortality or recruitment equation. Summer flounder abundance (⁠ Na,t ⁠) was estimated across time (t) by age (a) from age-at-recruitment (age 0, a = 0) to age 7+ (any fish age 7 or older is treated as a part of a single “plus group”), where initial abundance is defined by recruitment Rt for year t ⁠: Na,t={Rta=0e-Za-1,t-1Na-1,t-11≤a≤6e-Za-1,t-1Na-1,t-1+e-Za,t-1Na,t-1a≥7. (1) where survival was specified as sa,t=e-Za,t and the total mortality Za,tconsisted of natural mortality Ma,tand fishing mortality Fa,t ⁠: Za,t=Ma,t+Fa,t ⁠. Recruitment was estimated by predicting log-recruitment given spawning biomass and a multiplicative lognormal residual variability. In this study, log() is used to indicate the natural log. Log-recruitment log⁡(Rt) (defined as abundance at a = 0) was parameterized as recruitment deviations: log(Rt)=fSBt⁡+εt, (2) where f(SBt) is the Beverton–Holt function predicting log-recruitment as a function of spawning stock biomass, fSBt=log⁡(SBt-1β+αSBt-1)ifusingCIorCMmodellog(SBt-1β+αSBt-1ecT) ifusingCRmodel, (3) and recruitment deviations are the normally distributed variable εt∼Normal(-σr22,σr2) [(2); Methot and Taylor, 2011; Terceiro, 2015, 2016). Climate covariate effects (c) on recruitment were allowed in the general model. The recruitment–environment relationship used in the CR model is controlling recruitment (as opposed to limiting or masking), where climate is expected to influence recruitment via the larval/young fish mortality rates (Iles and Beverton, 1998; O’Leary et al., 2018), while in the CI and CM model, it was the standard Beverton–Holt form. The recruitment estimates for each model type are shown in Figure 2. The R package “Fish Life” was used to provide a starting point for an informative prior for α to the nearest integer (log-normally distributed with a log-mean of 3 and a log-standard deviation of 1; Thorson et al., 2017). Fish Life was also used to create an informative prior for the standard deviation of recruitment deviations, σr2 (bounded between 0.1 and 0.9) to be used in the estimation of the variance εt (Thorson et al., 2017). Log-abundance, logNa,1 ⁠, for each age a in the first modelled year was assigned a uniform prior distribution with realistic biological bounds selected such that the prior distribution did not qualitatively affect model results. Spawning stock biomass (⁠ SBt) was depended on the abundance at age a at time t(⁠ Na,t ⁠), weight-at-age a at time t(⁠ wa,t ⁠; Figure 1), maturity at age a(⁠ ma ⁠) up to the final age class amax SBt=12∑a=1amaxwa,tmaNa,t,(4) where we specify that females represent 50% of total abundance. For all three estimation models, we modelled natural mortality (⁠ Ma,t ⁠) as a time- and age-varying process, with values drawn from a lognormal distribution with log-mean (⁠ Va,t ⁠; hyperparameter for natural mortality) and variance (⁠ σM2 ⁠): log⁡Ma,t∼NormalVm,a,t,σM2, (5) where the specification of Vm,a,t differs among models m. In the CM model, the log-mean of natural mortality (⁠ Vm,a,t ⁠) followed a quadratic function of climate, while it was constant for CI and CR models: Vm,a,t=x0if usingCIor CRmodelsx0+x1Tt+x2Tt2ifusingCMmodel. (6) We incorporated the estimate of natural mortality Ma,t for each age and year [(5) and (6)] into the survival equation. The log-quadratic relationship of the Gulf Stream Index–natural mortality is suggested to be related to both preferred warmer temperatures that occur at high Gulf Stream Index when the north wall of the Gulf Stream is pushed further north, and changes in available habitat (that consequently impacts mobility, predator and prey densities, and ontogenetic migration) (O’Leary et al., 2018). For full methods, the remaining equations, prior distributions, and equation definitions, see O’Leary et al. (2018). BRP calculation We calculated 12 time series of BRPs from 1990 to 2015, formed as the factorial cross of three estimation models (explained previously) and two approaches to estimating BRPs (dynamic and moving-average) for two BRPs (a fishing mortality and a spawning biomass BRP) that utilize the SPR, also defined below. The four paired estimation approaches and BRP calculations were dynamic- B0(t) ⁠, dynamic-F35%(⁠ t ⁠), moving-average- B0(t) ⁠, and moving-average-F35%(⁠ t ⁠). Complete dynamic-BRP calculations require information on population processes for the recorded lifespan of summer flounder. Our initial year of data available is 1982 and the full range of recorded age classes is 8 years (age class plus group), and so our BRP calculations begin in 1990 (i.e. 1982 + 8 = 1990). We calculated each time series as a posterior predictive distribution given the posterior distribution for parameters in the CI, CR, and CR models. For consistency with the National Oceanic and Atmospheric Administration stock assessment of summer flounder (Terceiro, 2016), we use the SPR-based BRP [F35%(⁠ t ⁠)] as a proxy for FMSY (Table 1). Previous research defined 0.35 as a sufficient ratio to maintain SBPR levels that meet management targets for New England groundfish (Clark, 1991), although Clark (1993) noted that, in the presence of randomly variable recruitment, 0.40 was a better ratio than 0.35. Table 1. BRP definitions and equations. BRP Definition Equation Moving-average Dynamic F35% (⁠ t ⁠) The fishing mortality value in year t for posterior sample θr at which spawning biomass per recruit is 35% of the unfished spawning biomass per recruit, given parameters defined in a single year t For a given F* in year t, 0.35=SBPRF*|r,t SBPRF*=0|r,t For a given F* in year t, 0.35=SBPRF*|θr,tSBPRF*=0|θr,t B0(t) The spawning biomass in year t for posterior sample r from past recruitment deviations in the absence of fishing given estimated parameters for stock productivity θr 1n(r)∑r=1n(r)SB(F*=0|r,t) The spawning biomass when there is no fishing pressure (F*) in year t, SB(F*=0|r,t) BRP Definition Equation Moving-average Dynamic F35% (⁠ t ⁠) The fishing mortality value in year t for posterior sample θr at which spawning biomass per recruit is 35% of the unfished spawning biomass per recruit, given parameters defined in a single year t For a given F* in year t, 0.35=SBPRF*|r,t SBPRF*=0|r,t For a given F* in year t, 0.35=SBPRF*|θr,tSBPRF*=0|θr,t B0(t) The spawning biomass in year t for posterior sample r from past recruitment deviations in the absence of fishing given estimated parameters for stock productivity θr 1n(r)∑r=1n(r)SB(F*=0|r,t) The spawning biomass when there is no fishing pressure (F*) in year t, SB(F*=0|r,t) F* is the instantaneous fishing mortality, t is the year, θr is a sample from the posterior distribution of parameters, nr is the total number of posterior samples, Rt,r is the recruitment in year t for posterior sample r, and SBPRF*|θr,t is the spawning biomass per recruit under F* moving-average on year t and posterior distribution θr ⁠. Open in new tab Table 1. BRP definitions and equations. BRP Definition Equation Moving-average Dynamic F35% (⁠ t ⁠) The fishing mortality value in year t for posterior sample θr at which spawning biomass per recruit is 35% of the unfished spawning biomass per recruit, given parameters defined in a single year t For a given F* in year t, 0.35=SBPRF*|r,t SBPRF*=0|r,t For a given F* in year t, 0.35=SBPRF*|θr,tSBPRF*=0|θr,t B0(t) The spawning biomass in year t for posterior sample r from past recruitment deviations in the absence of fishing given estimated parameters for stock productivity θr 1n(r)∑r=1n(r)SB(F*=0|r,t) The spawning biomass when there is no fishing pressure (F*) in year t, SB(F*=0|r,t) BRP Definition Equation Moving-average Dynamic F35% (⁠ t ⁠) The fishing mortality value in year t for posterior sample θr at which spawning biomass per recruit is 35% of the unfished spawning biomass per recruit, given parameters defined in a single year t For a given F* in year t, 0.35=SBPRF*|r,t SBPRF*=0|r,t For a given F* in year t, 0.35=SBPRF*|θr,tSBPRF*=0|θr,t B0(t) The spawning biomass in year t for posterior sample r from past recruitment deviations in the absence of fishing given estimated parameters for stock productivity θr 1n(r)∑r=1n(r)SB(F*=0|r,t) The spawning biomass when there is no fishing pressure (F*) in year t, SB(F*=0|r,t) F* is the instantaneous fishing mortality, t is the year, θr is a sample from the posterior distribution of parameters, nr is the total number of posterior samples, Rt,r is the recruitment in year t for posterior sample r, and SBPRF*|θr,t is the spawning biomass per recruit under F* moving-average on year t and posterior distribution θr ⁠. Open in new tab For each BRP, we used 2850 Markov Chain Monte Carlo samples (r) from the posterior distribution for parameters in each estimation model. For each posterior sample, we projected population dynamics for 201 levels of fishing (F*)ranging from 0 to 2 in increments of 0.01, resulting in 572 850 total projections. The following sections detail the calculations used for dynamic-BRP and moving-average-BRP approaches for the reference points F35%(⁠ t ⁠) and B0(t) ⁠. BRP calculation algorithm The main difference between the dynamic- and moving-average-F35%(⁠ t ⁠) and B0(t) is the method of incorporation of temporal variability. The dynamic approach incorporated the varying vital rates of a cohort through time to calculate productivity given estimated parameters for stock productivity from a dynamic B0estimated population. In the dynamic-BRP approach, the population was projected using values in year t−1 for each cohort. Therefore, the population dynamics reflect estimates of demographic parameters (recruitment, natural mortality, growth, and maturity) for preceding years. The dynamic approach is referred to as “dynamic” because it generalizes “dynamic-B0”, i.e. where it projects dynamics from t−1 to t in the absence of fishing (to calculate biomass targets from B0) or with different fishing mortality rates (to calculate exploitation targets from SPR). The moving-average approach is calculated assuming equilibrium conditions given a set of environmental conditions and demographic parameters in a given year (or their average over a window of years) at a specified level of fishing over the entire time period of the population. To calculate dynamic- B0(t) ⁠, dynamic-F35%(⁠ t ⁠), moving-average- B0(t) ⁠, and moving-average-F35%(⁠ t ⁠), we used the following general steps: (1) calculate unfished SBPR and SPR and (ii) calculate total numbers and unfished biomass, for 201 levels of F* ranging from 0 to 2 in increments of 0.01. Calculations used a sample θr for natural mortality, recruitment deviations, initial age structure, selectivity, and initial numbers-at-age from the estimation model posterior distribution r. Full algorithms with equations for each BRP calculation are presented in Supplementary Material 1. Evaluation of estimation approaches We calculated the mean, uncertainty, and variability (i.e. the temporal coefficient of variation due to biological variation) of F35%(⁠ t ⁠) and B0(t) over the entire period to address the first three objectives: (1) quantify the variability over time of climate-dependent vs. climate-independent BRPs over a 26-year period, (2) quantify the standard error for different BRPs and approaches, and (3) determine whether moving-average- or dynamic-BRPs are more variable and uncertain over time. We calculated uncertainty as the mean of the C.V. across L samples of the posterior xr for Y years, where C.V.y=1L∑r=1L(xr-μ)2∑r=1LxrL and average uncertainty=∑y=1YC.V.yY ⁠. Temporal variability was calculated as the C.V. across a total of Y years for a total of L posterior samples followed by a mean across samples. Here, C.V.r=1Y∑y=1Y(xy-μ)2∑y=1YxyY and temporal variability=∑r=1LC.V.rL ⁠. To address objective (4) to attribute trends and variation in BRPs to variation in mortality, recruitment, and growth processes, we tested the following three separate model fits: (i) natural mortality varied with all other BRP inputs fixed at their averages, (ii) weight-at-age (i.e. growth) varied with all other BRP inputs fixed at their averages, and (iii) recruitment varied with all other B0 inputs fixed at their averages (Miller et al., 2018). Each scenario describes the biological process that varied while holding all other biological parameters at their average conditions. In model fits (ii) and (iii) where natural mortality was constant, it was fixed at the average over time for each age. To address objective (5) to evaluate how different methods impact estimates of stock status for summer flounder, we compared the estimation model fishing rate and spawning stock biomass to estimated moving-average and dynamic F35%(⁠ t ⁠) and B35%(t) ⁠, or 35% of B0(t) ⁠, to determine if the stock was overfished or if overfishing was occurring. We determined “overfished” and “overfishing” status for summer flounder for each BRP estimation approach and climate model. Overfishing here is defined as when the current fishing rate is higher than the BRP fishing value. Overfished is defined as when the stock is unable to maintain biomass levels at or above B35%(t) ⁠. This is different from the biomass reference point used in the stock assessment for summer flounder, where overfishing is calculated by projection method using the fishing rate at F35% and average recruitment (Terceiro, 2016). Results Objective 1: how variable are climate-linked BRPs? Overall, the temporal variability in F35%(⁠ t ⁠) and B0 was greater for the CR and CI models than the CM model for both estimation approaches (Figure 3 and Table 2). This difference in variability indicated that the CM model provided the most stable BRPs. In addition, the mean F35%(⁠ t ⁠) was lowest for the CM model, indicating that the CM model estimated the most restrictive fishing mortality threshold. Figure 3. Open in new tabDownload slide Moving-average-F35%(⁠ t ⁠) (a), dynamic-F35%(⁠ t ⁠) (b), moving-average- B0(t) in mt (c), and dynamic- B0(t) in mt (d) for CI, CM, and CR models for t = 1990–2015 with ±50% credible intervals, ±75% credible intervals, and ±95% credible intervals from darkest to lightest grey. Figure 3. Open in new tabDownload slide Moving-average-F35%(⁠ t ⁠) (a), dynamic-F35%(⁠ t ⁠) (b), moving-average- B0(t) in mt (c), and dynamic- B0(t) in mt (d) for CI, CM, and CR models for t = 1990–2015 with ±50% credible intervals, ±75% credible intervals, and ±95% credible intervals from darkest to lightest grey. Table 2. BRPs using moving-average-BRP and dynamic-BRP approaches: mean F35%(⁠ t ⁠), F35%(⁠ t ⁠) uncertainty, F35%(⁠ t ⁠) temporal variability, mean B0(t) ⁠, B0t uncertainty, and B0(t) temporal variability across years 1990–2015 for the CR, CM, and CI models. Reference Point Moving-average Dynamic CR CM CI CR CM CI F35% (⁠ t ⁠)  Mean 0.35 0.25 0.37 0.34 0.29 0.35  Uncertainty 0.19 0.13 0.20 0.1 0.05 0.11  Variability 0.32 0.14 0.32 0.13 0.06 0.14 B0(t)  Mean (mt) 70 016 83 544 76 938 63 988 52 037 68 877  Uncertainty 0.46 0.26 0.53 0.25 0.16 0.3  Variability 0.49 0.40 0. 56 0. 28 0.24 0.32 Reference Point Moving-average Dynamic CR CM CI CR CM CI F35% (⁠ t ⁠)  Mean 0.35 0.25 0.37 0.34 0.29 0.35  Uncertainty 0.19 0.13 0.20 0.1 0.05 0.11  Variability 0.32 0.14 0.32 0.13 0.06 0.14 B0(t)  Mean (mt) 70 016 83 544 76 938 63 988 52 037 68 877  Uncertainty 0.46 0.26 0.53 0.25 0.16 0.3  Variability 0.49 0.40 0. 56 0. 28 0.24 0.32 Uncertainty in F35%(⁠ t ⁠) and B0(t) is expressed as the average coefficient of variation and variability in F35%(⁠ t ⁠) and B0(t) is expressed as the coefficient of variation over time. Open in new tab Table 2. BRPs using moving-average-BRP and dynamic-BRP approaches: mean F35%(⁠ t ⁠), F35%(⁠ t ⁠) uncertainty, F35%(⁠ t ⁠) temporal variability, mean B0(t) ⁠, B0t uncertainty, and B0(t) temporal variability across years 1990–2015 for the CR, CM, and CI models. Reference Point Moving-average Dynamic CR CM CI CR CM CI F35% (⁠ t ⁠)  Mean 0.35 0.25 0.37 0.34 0.29 0.35  Uncertainty 0.19 0.13 0.20 0.1 0.05 0.11  Variability 0.32 0.14 0.32 0.13 0.06 0.14 B0(t)  Mean (mt) 70 016 83 544 76 938 63 988 52 037 68 877  Uncertainty 0.46 0.26 0.53 0.25 0.16 0.3  Variability 0.49 0.40 0. 56 0. 28 0.24 0.32 Reference Point Moving-average Dynamic CR CM CI CR CM CI F35% (⁠ t ⁠)  Mean 0.35 0.25 0.37 0.34 0.29 0.35  Uncertainty 0.19 0.13 0.20 0.1 0.05 0.11  Variability 0.32 0.14 0.32 0.13 0.06 0.14 B0(t)  Mean (mt) 70 016 83 544 76 938 63 988 52 037 68 877  Uncertainty 0.46 0.26 0.53 0.25 0.16 0.3  Variability 0.49 0.40 0. 56 0. 28 0.24 0.32 Uncertainty in F35%(⁠ t ⁠) and B0(t) is expressed as the average coefficient of variation and variability in F35%(⁠ t ⁠) and B0(t) is expressed as the coefficient of variation over time. Open in new tab Objective 2: how uncertain are climate-linked BRPS? Overall, the uncertainty (i.e. C.V. for the posterior distribution) in both the F35%(⁠ t ⁠) and B0(t) was smaller for the CM model than for all other models (Figure 3 and Table 2). This likely occurs because the CM model has a different posterior distribution than the other two models, where the CM model posterior explains more of the population variance (Figure 2). Objective 3: are moving-average-BRPs or dynamic-BRPs more uncertain and variable over time? The temporal variability in F35%(⁠ t ⁠) and B0(t) was greater for the moving-average-BRP approach than for the dynamic-BRP approach for all tested models (Figure 3 and Table 2). These results suggest that tracking individual cohorts (i.e. dynamic-BRP approach) results in less variable fishing and biomass reference points than if the population achieves equilibrium given average demographic rates over a 1-year window (i.e. moving-average-BRP approach) because the dynamic-BRP approach is smoothing across cohorts in a given year. Variability in the moving-average-BRP approach also depends upon whether calculations are based on conditions in a single-year, 3-year, or 5-year window. As the window to calculate the F35% (⁠ t ⁠) increased to 3 and 5 years (both with t as the centre and terminal years), variability using the moving-average approach decreased relative to the 1-year window (see Supplementary Material 2). Overall, the moving-average approach F35%(⁠ t ⁠) was more variable than the dynamic approach F35%(⁠ t ⁠) for all windows (except for the 5-year window with t as the terminal year for the CM model). As the window to calculate B0(t) increased to 3 and 5 years (with t as the centre and terminal years), the moving-average approach variability decreased again relative to the 1-year window. Overall, the moving-average approach B0(t) was more variable than the dynamic approach for the 1- and 3-year windows and less variable than the dynamic approach for the 5-year window (see Supplementary Material 2 for the full results for the moving-average approach using 3- and 5-year window averages with t as the centre year and t as the terminal year). The uncertainty in the F35%(⁠ t ⁠) estimates (i.e. the coefficient of variation of the posterior distribution) was lower in the dynamic-BRP approach than in the moving-average-BRP approach (Table 2). The uncertainty in the B0t estimates was also lower in the dynamic-BRP approach than in the moving-average-BRP approach (Table 2). The overall temporal trend for F35%(⁠ t ⁠) was different between the moving-average and dynamic approach results (Figure 3). Both the moving-average- and dynamic-F35%(⁠ t ⁠) decreased over time for the CR and CI models and remained relatively constant for the CM model. In the final 5 years (2011–2015), the moving-average-F35%(⁠ t ⁠) increased whereas the dynamic-F35%(⁠ t ⁠) continued to decline for the CR and CI models (Figure 3). The CM model F35%(⁠ t ⁠) continued to remain relatively constant over time using both the moving-average-BRP and dynamic-BRP approaches. Objective 4: attributing variation in BRPs to changes in natural mortality, recruitment, or weight-at-age We explored the impact of each of the three time-varying biological processes in isolation on BRP estimation to determine the causes of observed variation in BRPs: (i) varying weight-at-age, (ii) varying natural mortality, and (iii) varying recruitment. We found that variation in natural mortality was the largest driver of temporal variation in summer flounder F35%(⁠ t ⁠), followed by changes in weight-at-age. For the CR and CI models, we note that temporal variation in estimated natural mortality drives the large decrease in B0 (⁠ t ⁠) from 2011 to 2015 in moving-average approach and, therefore, drives an associated increase in F35%(⁠ t ⁠) for those two models. Variation in natural mortality was the largest driver of temporal variation in summer flounder B0(t) for the moving-average approach. However, recruitment was the largest driver of variation in summer flounder B0(t) for the dynamic approach. This suggests that the contribution of temporal variation in natural mortality and recruitment to the BRP variability is dependent on the estimation approach (Figure 4 and Table 3). Figures 1 and 2 show the time-varying parameters for each of the three estimation models. Figure 4. Open in new tabDownload slide The fully varying BRPs compared to BRPs from three sensitivity analyses attributing change to varying weight-at-age, varying natural mortality, or varying recruitment in isolation while holding all other parameters at their average values. Figure 4. Open in new tabDownload slide The fully varying BRPs compared to BRPs from three sensitivity analyses attributing change to varying weight-at-age, varying natural mortality, or varying recruitment in isolation while holding all other parameters at their average values. Table 3. Variability in F35%(⁠ t ⁠) and B0t estimates using moving-average-BRP and dynamic-BRP approaches and varying natural mortality (natural mortality scenario), varying weight-at-age (weight-at-age scenario), or varying recruitment (recruitment scenario) across years 1990–2015 for the CR, CM, and CI models. Scenario Moving-average Dynamic CR CM CI CR CM CI F35% (⁠ t ⁠)  Natural mortality scenario variability 0.35 0.14 0.35 0.15 0.06 0.16  Weight-at-age scenario variability 0.06 0.06 0.07 0.05 0.05 0.05  Recruitment scenario variability 0.04 0.03 0.05 0.02 0.02 0.02 B0t  Natural mortality scenario variability 0.45 0.27 0.51 0.15 0.09 0.12  Weight-at-age scenario variability 0.15 0.11 0.16 0.11 0.11 0.09  Recruitment scenario variability 0.23 0.28 0.28 0.28 0.22 0.35 Scenario Moving-average Dynamic CR CM CI CR CM CI F35% (⁠ t ⁠)  Natural mortality scenario variability 0.35 0.14 0.35 0.15 0.06 0.16  Weight-at-age scenario variability 0.06 0.06 0.07 0.05 0.05 0.05  Recruitment scenario variability 0.04 0.03 0.05 0.02 0.02 0.02 B0t  Natural mortality scenario variability 0.45 0.27 0.51 0.15 0.09 0.12  Weight-at-age scenario variability 0.15 0.11 0.16 0.11 0.11 0.09  Recruitment scenario variability 0.23 0.28 0.28 0.28 0.22 0.35 Variability in F35% (⁠ t ⁠) and B0t is expressed as the coefficient of variation over time. Open in new tab Table 3. Variability in F35%(⁠ t ⁠) and B0t estimates using moving-average-BRP and dynamic-BRP approaches and varying natural mortality (natural mortality scenario), varying weight-at-age (weight-at-age scenario), or varying recruitment (recruitment scenario) across years 1990–2015 for the CR, CM, and CI models. Scenario Moving-average Dynamic CR CM CI CR CM CI F35% (⁠ t ⁠)  Natural mortality scenario variability 0.35 0.14 0.35 0.15 0.06 0.16  Weight-at-age scenario variability 0.06 0.06 0.07 0.05 0.05 0.05  Recruitment scenario variability 0.04 0.03 0.05 0.02 0.02 0.02 B0t  Natural mortality scenario variability 0.45 0.27 0.51 0.15 0.09 0.12  Weight-at-age scenario variability 0.15 0.11 0.16 0.11 0.11 0.09  Recruitment scenario variability 0.23 0.28 0.28 0.28 0.22 0.35 Scenario Moving-average Dynamic CR CM CI CR CM CI F35% (⁠ t ⁠)  Natural mortality scenario variability 0.35 0.14 0.35 0.15 0.06 0.16  Weight-at-age scenario variability 0.06 0.06 0.07 0.05 0.05 0.05  Recruitment scenario variability 0.04 0.03 0.05 0.02 0.02 0.02 B0t  Natural mortality scenario variability 0.45 0.27 0.51 0.15 0.09 0.12  Weight-at-age scenario variability 0.15 0.11 0.16 0.11 0.11 0.09  Recruitment scenario variability 0.23 0.28 0.28 0.28 0.22 0.35 Variability in F35% (⁠ t ⁠) and B0t is expressed as the coefficient of variation over time. Open in new tab Objective 5: how do these decisions affect estimates of stock status? To determine the “overfishing” (i.e. F(t)F35%(t)>1) and “overfished” (i.e. SB(t)B35%t<1) status, we calculated the ratio of fishing mortality to F35% (⁠ t ⁠) and the spawning stock biomass to 35% of B0t [typically called B35% t], for both BRP methods for each year. Stock status was highly dependent on the climate model and estimation approach. In the moving-average approach, the stock was classified as overfished up to 2000 (or 2002 for the CM model; Figure 5). From 2000 up to 2013 (or 2002–2005 and 2011 for the CM model), the stock classified as not overfished based upon the moving-average approach BRPs. In the final 2–3 years, the moving-average-BRP estimated the stock as overfished (Figure 5). In the dynamic approach, the stock was classified as overfished up until 1997 for the CR and CM models and up to 1996 for the CI model (Figure 5). From that year onwards, the stock was classified as not overfished for all models up until the final year 2015 based upon the dynamic approach BRPs. The exception to this was, in 2010, the CM model, where the stock drops back down to overfished. Overfishing was occurring on the summer flounder stock throughout the entire time period based upon BRP calculations from both the moving-average and dynamic approaches (Figure 5), except for, in years 1993 and 2002, the CR and CI models. Figure 5. Open in new tabDownload slide Moving-average-BRP (grey lines) and dynamic-BRP (black lines) stock status plots for the CI, CM, and CR models. Plots show (top row) estimated instantaneous fishing mortality (F) divided by F35% (⁠ t ⁠) across time and (bottom row) the estimated spawning stock biomass (SB) divided by B35%(t) across time. A fishing mortality ratio of >1 indicates overfishing of the stock and that of <1 indicates no overfishing. A spawning stock biomass ratio of <1 indicates that the stock is overfished and that of >1 indicates that the stock is not overfished. Note that y-axis is plotted using a log-scale. Figure 5. Open in new tabDownload slide Moving-average-BRP (grey lines) and dynamic-BRP (black lines) stock status plots for the CI, CM, and CR models. Plots show (top row) estimated instantaneous fishing mortality (F) divided by F35% (⁠ t ⁠) across time and (bottom row) the estimated spawning stock biomass (SB) divided by B35%(t) across time. A fishing mortality ratio of >1 indicates overfishing of the stock and that of <1 indicates no overfishing. A spawning stock biomass ratio of <1 indicates that the stock is overfished and that of >1 indicates that the stock is not overfished. Note that y-axis is plotted using a log-scale. For some years (e.g. 1993 and 2002), classification as “overfished” (or not) depended on the estimation approach and climate models used. However, it is worth noting how similar the overfishing status is between the two approaches for a given model. As well, the moving-average approach resulted in an overfished stock status for a slightly longer duration. From the management perspective, these small differences can be important because each stock status enacts a different series of management actions. As well, the outcome is affected by any management actions that occurred during this period. All BRP approaches and estimation models classified the stock as overfished with overfishing during early years, with a transition to not overfished with overfishing in later years. The largest difference in stock status between the two estimation approaches is in the final 6 years for the biomass reference, particularly for the CI and CR models (Figure 5). Discussion In this case study using summer flounder, both the mechanism of climate dependency in the underlying empirical model and the temporal variability used to calculate BRPs altered BRP uncertainty, variability, and thus stock status. Specifically, the dynamic-BRP and CM-linked model estimated lower fishing rates and BRPs for a 1-year window than the other models and BRPs tested. BRP variability stemmed principally from varying natural mortality regardless of the estimation approach and underlying climate-dependent model. It is worth noting that O’Leary et al. (2018) demonstrated that the CM model was the best-fitting model to capture past summer flounder abundances. However, because the underlying model used to calculate BRPs can always be incorrectly specified, a more stable or restrictive BRP does not always equate to a “better” reference point. The climate dependencies represented here demonstrate how variable fish stock dynamics can be and how the overfishing determination depends on these defined relationships. Previously developed models with time-varying vital rates used to estimate BRPs highlight the importance of correctly identifying climate dependencies when determining stock status, the large variability in BRP estimates that can result from using climate-dependent models, and the differences in BRP uncertainty depending on the time-varying properties used (i.e. current vs. a 3–5-year average). Mangel et al. (2013) discuss the false sense of precision that arises from using point estimates or posterior samples when using a two-parameter stock–recruit relationship or fixed natural mortality. Similarly, we suggest that basing reference points on fixed values of life-history parameters will often convey a false sense of precision, given that all populations have some degree of time-varying growth, mortality, maturity, or other processes. Time-varying BRPs inherently have much more variability, may be more difficult to understand, and be difficult for managers to use. We suggest developing time-varying BRPs more generally and comparing them to “static” BRPs to understand this uncertainty hidden by decisions to use fixed natural mortality. Empirical analyses, such as the one presented here, will help managers evaluate risks and priorities through understanding which changing vital rates impact scientific advice to management the most (Karp et al., 2019). In this study, time-varying natural mortality resulted in the greatest variability in BRPs in both the CR and CI models. The CM model estimated lower, less variable, and more precise natural mortality parameter estimates than the CR and CI models. This is likely due to the incorporation of the relationship between natural mortality and the Gulf Stream Index. In addition, natural mortality decreased slightly in the CM model as the Gulf Stream deviated from average conditions. As natural mortality decreased, peak SBPR at F35%(⁠ t ⁠) decreased and occurred at older ages in the CM model. The reduction in peak SBPR, in turn, reduced the BRP value and its associated uncertainty while also providing less variable mean BRPs over time. These changes in productivity are consistent with the theoretical rationale that if natural mortality decreases and all else remains the same, the vital rates describe a longer-lived fish species with productivity spanning over a longer time frame (i.e. the fecundity over the fish’s lifetime is greater). Therefore, because of the large influence that time-varying natural mortality has on per recruit BRPs, strong empirical evidence is needed to support the use of a time-varying natural mortality in BRP calculations (Legault and Palmer, 2016). The choice to include climate dependency or temporally varying vital rates in BRP calculations based on empirical evidence can have implications for stock status. Depending on a stock’s vulnerability to changing ocean conditions, these differences in BRPs due to assumed temporal dynamics will need to be considered in any risk assessment evaluation. The methods herein provide a tool for scientists and managers to consider when preparing their fishery for both near-term and long-term management under shifting oceanographic conditions. For summer flounder, the temporal variability in F35%(⁠ t ⁠) and B0(t)was greater for the moving-average-BRP approach than for the dynamic-BRP approach for all climate models using a 1-year window. The assumed temporal dynamics for each cohort can potentially explain the greater variability in F35%(⁠ t ⁠) and B0(t)in the 1-year window moving-average-BRP approach than in the dynamic-BRP approach. The dynamic-BRP approach tracks cohorts and uses the time-varying natural mortality from that year and all previous years to calculate F35%(⁠ t ⁠) and B0(t). The dynamic approach reflected the productivity at each age in the cohort’s lifespan using the natural mortality specific to that age giving each cohort a different productivity history. Because we calculated each cohort with age-specific natural mortality values, the productivity between cohorts within a year was more consistent because it represents a blend of productivities up to that year. The moving-average-BRP approach, on the other hand, was an equilibrium approach (i.e. assuming that environmental conditions continue for indefinitely long period such that the population achieves population equilibrium). The natural mortality was conditioned from a 1-year period (i.e. ignores the cohort history) and assumed equal to that value for the previous years of that cohort’s lifespan. Therefore, the productivity of each cohort in any year is calculated in the moving-average approach assuming its productivity at that point in time. This is likely why the moving-average BRP from year to year is more variable than the dynamic-BRP. Trends in both BRPs were primarily attributed to natural mortality (Table 2; varying natural mortality scenario). This finding supports the suggestions of Brodziak et al. (2011) and Thorson et al. (2015) that identifying changes in natural mortality should be a priority when expanding stock assessments to including time-varying biological processes because of their large influence over spawning biomass and catch management targets. As well, in the varying natural mortality scenario, the variability in both BRPs was lowest in the CM model. This suggests that including an environmental index as a covariate in the estimation of natural mortality successfully constrained variation in natural mortality in the CM model, whereas natural mortality was estimated as an unconstrained random process in the CR and CR models. Different methods of incorporating temporal variability and climate effects into the fish’s population dynamics influence the uncertainty and variability of BRPs. Therefore, careful consideration is warranted when considering which approach to use and how to incorporate climate effects, ideally based upon management strategy evaluation within a proposed set of management procedures (Karp et al., 2019). The variability in the moving-average approach BRPs was highly dependent not only on natural mortality values but also on the length of window used to calculate the moving-average-BRP. The assumed dynamics for each cohort over the BRP calculation period likely explain the different temporal variability in BRPs for each estimation approach. For the 1-year window, this may be due to the large increase in natural mortality in the final year of the models where climate and natural mortality are not linked. The large increase in time-varying natural mortality in the final year is due to insufficient information in the parameterization of these models relative to the climate-linked natural mortality models. This characteristic is not present in the summer flounder stock assessment due to a different parameterization of natural mortality. We can implement these developed approaches for an MSY reference point where the stock more closely follows a stock–recruit relationship. The models used in this study do contain a stock–recruit relationship. However, the stock–recruit relationship for summer flounder is weak and uncertain (Terceiro, 2016). The interpretation of an MSY-based reference point, therefore, was not meaningful or dependable and hence the use of F35%(⁠ t ⁠) both here and in the stock assessment (Maunder, 2012). Natural mortality has a considerable influence on FMSY similar to F35%(⁠ t ⁠) as demonstrated here and so would be interesting to look at for other stocks (Maunder, 2012). In the 2019 summer flounder stock assessment, the stock was determined not overfished and no overfishing was occurring. As well, based on the BRPs calculated here, there was no overfishing of the summer flounder stock despite differences in BRP uncertainty and variability in this study. Some of the calculated BRPs in this article did categorize summer flounder as overfished in recent years, which can potentially trigger a rebuilding plan if the fishing mortality is not low enough to allow the stock to be projected to be rebuilt in sufficient time. However, the purpose of this study was not to present the “correct” BRP values or BRP values for management but rather to present the consequences of considering multiple model productivity estimates and BRP calculation approaches within the same stock. The models used to explore the consequences of the estimation model and estimation approach are simpler than typical stock-assessment models so that we could incorporate time-varying natural mortality and other time-varying processes. For instance, fleets in this study combine landings and discards, likely influencing the magnitude of BRP values. As well, the climate-dependent models used here are too simple for determining stock status for management (O’Leary et al., 2018) but rather serve to demonstrate the relative ramifications of different climate dependencies when using time-varying reference points. Thus, values should be interpreted only in comparison to the other BRPs in this study, with the knowledge that the CM model was the best performing model for this particular stock. The differences in BRP uncertainty and variability for both approaches and climate dependencies may be greater if we calculate BRPs for a fish stock with more variable productivity, productivity closer to a threshold tipping point, or with a greater magnitude response to climate. This is particularly relevant for the current management process in many regions, where BRPs include temporal variation by conditioning them on information and stock-assessment estimates from the most recent years. Therefore, we suggest that there should be a consultative, iterative process with stakeholders to identify the method used to calculate BRPs. As well, plausible climate hypotheses should be developed and used to test climate dependencies relevant to the managed fish stock. Researchers can then optimize management procedures based upon the BRP input used for management practices. We suggest incorporating the following steps into management: Consider the method of incorporating time variation into the BRP calculation before harvesting control rules and management proceeds. Evaluate how both the dynamic- and moving-average-BRP approaches influence the BRP calculation based on the longevity and time-varying population dynamics of the fish stock. Comparing various BRPs calculated with different modes of temporal variability incorporation is central to understanding the ramifications of any model choice for management. Following these two BRP calculations, if a Bayesian assessment model is used, a direct comparison between the BRP posterior distributions can be used to determine the likelihood of overfishing a fish stock. We suggest performance measures to compare between BRP outcomes including average catch, revenue, and avoidance of fishery collapse. A simulation study can be used to determine the trade-off of these performance measures for each BRP decision and identify control rules that balance the competing objectives of the fishery, similar to the Wiedenmann et al. (2013) approach used for data-poor fisheries. Take into account the timeline and management goals of the various stakeholders to determine how to incorporate BRP uncertainty, variability, and probability of overfishing the stock. A “stable” reference point does not necessarily imply the “best” advice. For example, the assessment for summer flounder occurs every 5 years and the Acceptable Biological Catch is set every 3 years. In this case, the moving-average-BRPs are more uncertain. Therefore, we advise using these moving-average BRPs over a longer time frame to more cautiously approach 3–5-year management timeframe of a fish stock whose natural mortality temporally varies from year to year. Using this moving-average approach, a fish stock’s temporally varying life-history and productivity conditions are more likely captured by the greater uncertainty. These steps can be used to extend the time-varying BRP methods established here to provide a quantitative understanding of the risk associated with each BRP decision. Simulation studies to directly compare and evaluate the implications of BRP temporal variability incorporation and climate dependencies on time-varying BRP calculations for their stock of interest would greatly advance this study. The use of time-varying biological inputs can smooth the effects of population dynamics over adjacent years and across cohorts, making emergent trends challenging to interpret. Despite this limitation, this study provided evidence that temporally varying climate-inclusive BRP calculations resulted in changes in BRP values over time but did not cause such a large increase in uncertainty to make BRPs uninterpretable. We also successfully incorporated time-varying parameters using two different methods without making BRPs uninterpretable, providing a method to account for variability and evaluate the risk of multiple scenarios in management as climate conditions continue to change. Importantly, the uncertainty in the time-varying BRPs would likely increase if forecasted, particularly in cases where climate is incorporated into the time-varying population processes (Miller et al., 2016). Differences in stock status that depend on the climate model used highlight the need to account for the effects of changing climate conditions on stock productivity if present. Acknowledgements We would like to thank Mark Terceiro (NEFSC) for providing valuable comments, consultations, and additional data. We would also like to thank Phil McDowall (Google) for his consultation in code production. Finally, we would like to thank the NOAA-AFSC reviewers Melissa Karp and Martin Dorn for their thorough and thoughtful comments, as well as reviewer comments from K. Johnson and two anonymous reviewers. This work was supported by the NOAA, National Marine Fisheries Service Sea Grant Population and Ecosystem Dynamics Fellowship. The authors have no significant competing financial, professional, or personal interests that might have influenced the materials presented in this article. The findings and conclusions in the article are those of the author(s) and do not necessarily represent the views of the National Marine Fisheries Service, NOAA. References Amar Z. T. , Punt A. E. , Dorn M. W. 2009 . The evaluation of two management strategies for the Gulf of Alaska walleye pollock fishery under climate change . ICES Journal of Marine Science , 66 : 1614 – 1632 . Google Scholar Crossref Search ADS WorldCat Audzijonyte A. , Fulton E. , Haddon M. , Helidoniotis F. , Hobday A. J. , Kuparinen A. , Morrongiello J. 2016 . Trends and management implications of human-induced life history changes in marine ectotherms . Fish and Fisheries , 17 : 1005 – 10028 . Google Scholar Crossref Search ADS WorldCat Azarovitz T. R. 1981 . A brief historical review of the Woods Hole Laboratory trawl survey time series . Canadian Special Publication of Fisheries and Aquatic Sciences , 58 : 67 . WorldCat Barange M. , Merino G. , Blanchard J. L. , Scholtens J. , Harle J. , Allison E. H. , Allen J. I. , et al. 2014 . Impacts of climate change on marine ecosystem production in societies dependent on fisheries . Nature Climate Change , 4 : 211 – 216 . Google Scholar Crossref Search ADS WorldCat 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 Baudron A. R. , Needle C. L. , Marshall C. T. 2011 . Implications of a warming North Sea for the growth of haddock Melanogrammus aeglefinus . Journal of Fish Biology , 78 : 1874 – 1889 . Google Scholar Crossref Search ADS PubMed WorldCat Baudron A. R. , Needle C. L. , Rijnsdorp A. D. , Marshall C. T. 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 Berger A. M. 2019 . Character of temporal variability in stock productivity influences the utility of dynamic reference points . Fisheries Research , 217 : 185 – 197 . Google Scholar Crossref Search ADS WorldCat Blanchard J. L. , Mills C. , Jennings S. , Fox C. J. , Rackham B. D. , Eastwood P. D. , O’Brien C. M. 2005 . Distribution abundance relationships for North Sea Atlantic cod (Gadus morhua): observation versus theory . Canadian Journal of Fisheries and Aquatic Sciences , 62 : 2001 – 2009 . Google Scholar Crossref Search ADS WorldCat Brander K. M. 2007 . Global fish production and climate change . Proceedings of the National Academy of Sciences of the United States of America , 104 : 19709 – 19714 . Google Scholar Crossref Search ADS PubMed WorldCat Brander K. 2010 . Impacts of climate change on fisheries . Journal of Marine Systems , 79 : 389 – 402 . Google Scholar Crossref Search ADS WorldCat Brodziak J. , Ianelli J. , Lorenzen K. , Methot R. D. Jr . 2011 . Estimating natural mortality in stock assessment applications. US Department of Commerce, NOAA Technical Memorandum NMFS-F/SPO-119. 38 pp. Buckley L. J. , Caldarone E. M. , Lough R. G. 2004 . Optimum temperature and food‐limited growth of larval Atlantic cod (Gadus morhua) and haddock (Melanogrammus aeglefinus) on Georges Bank . Fisheries Oceanography , 13 : 134 – 140 . Google Scholar Crossref Search ADS WorldCat Burns T. S. , Schultz R. , Brown B. E. 1983 . The commercial catch sampling program in the northeastern United States . Canadian Special Publication of Fisheries and Aquatic Sciences , 66 : 82 – 95 . WorldCat Clark W. G. 1991 . Groundfish exploitation rates based on life history parameters . Canadian Journal of Fisheries and Aquatic Sciences , 48 : 734 – 750 . 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 Population Alaska Sea Grant College Program Report , pp. 233 – 246 . Ed. by Kruse G. , Eggers D. M. , Marasco R. J. , Pautzke C. , Quinn T. J. . International Pacific Halibut Commission Technical Reports, https://iphc.ipower.com/publications/techrep/tech0037.pdf. Google Preview WorldCat COPAC Clark W. G. , St-Pierre G. , Brown E. S. 1997 . Estimates of Halibut Abundance from NMFS Trawl Surveys. International Pacific Halibut Commission Technical Report. 37 pp. Collie J. S. , Gislason H. 2001 . Biological reference points for fish stocks in a multispecies context . Canadian Journal of Fisheries and Aquatic Sciences , 58 : 2167 – 2176 . Google Scholar Crossref Search ADS WorldCat Collie J. , Hiddink J. G. , van Kooten T. , Rijnsdorp A. D. , Kaiser M. J. , Jennings S. , Hilborn R. 2017 . Indirect effects of bottom fishing on the productivity of marine fish . Fish and Fisheries , 18 : 619 – 637 . Google Scholar Crossref Search ADS WorldCat Cordue P. L. 2012 . Fishing intensity metrics for use in overfishing determination . ICES Journal of Marine Science , 69 : 615 – 623 . Google Scholar Crossref Search ADS WorldCat Deroba J. J. , Bence J. R. 2008 . A review of harvest policies: understanding relative performance of control rules . Fisheries Research , 94 : 210 – 223 . Google Scholar Crossref Search ADS WorldCat 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 Frisk M. G. , Miller T. J. , Dulvy N. K. 2005 . Life histories and vulnerability to exploitation of elasmobranchs: inferences from elasticity, perturbation and phylogenetic analyses . Journal of Northwest Atlantic Fishery Science , 35 : 27 – 45 . Google Scholar Crossref Search ADS WorldCat Gabriel W. L. , Sissenwine M. P. , Overholtz W. J. 1989 . Analysis of spawning stock biomass per recruit: an example for Georges Bank haddock . North American Journal of Fisheries Management , 9 : 383 – 391 . Google Scholar Crossref Search ADS WorldCat Gabriel W. L. , Mace P. M. 1999 . A review of biological reference points in the context of the precautionary approach. In Proceedings of the Fifth National NMFS Stock Assessment Workshop: Providing Scientific Advice to Implement the Precautionary Approach under the Magnuson-Stevens Fishery Conservation and Management Act. US Department of Commerce, NOAA Technical Memorandum NMFS-F/SPO-40, pp. 34–45. Ed. by V. R. Restrepo. Gerber L. R. , Heppell S. S. 2004 . The use of demographic sensitivity analysis in marine species conservation planning . Biological Conservation , 120 : 121 – 128 . Google Scholar Crossref Search ADS WorldCat Goodyear C. P. 1993 . Spawning stock biomass per recruit in fisheries management: foundation and current use . Canadian Special Publication of Fisheries and Aquatic Sciences , 120 : 67 – 81 . WorldCat Haltuch M. A. , Punt A. E. , Dorn M. W. 2008 . Evaluating alternative estimators of fishery management reference points . Fisheries Research , 94 : 290 – 303 . Google Scholar Crossref Search ADS WorldCat Haltuch M. A. , Punt A. E. , Dorn M. W. 2009 . Evaluating the estimation of fishery management reference points in a variable environment . Fisheries Research , 100 : 42 – 56 . Google Scholar Crossref Search ADS WorldCat Hare J. A. , Morrison W. E. , Nelson M. W. , Stachura M. M. , Teeters E. J. , Griffis R. B. , Alexander M. A. , et al. 2016 . A vulnerability assessment of fish and invertebrates to climate change on the Northeast US Continental Shelf . PLoS One , 11 : e0146756 . Google Scholar Crossref Search ADS PubMed WorldCat Hilborn R. , Walters C. J. 1992 . Quantitative fisheries stock assessment: choice, dynamics and uncertainty . Reviews in Fish Biology and Fisheries , 2 : 177 – 178 . Google Scholar Crossref Search ADS WorldCat Hilborn R. 2002 . The dark side of reference points . Bulletin of Marine Science , 70 : 403 – 408 . WorldCat Holsman K. K. , Essington T. , Miller T. J. , Koen-Alonso M. , Stockhausen W. J. 2012 . Comparative analysis of cod and herring production dynamics across 13 northern marine ecosystems . Marine Ecology Progress Series , 459 : 231 – 246 . Google Scholar Crossref Search ADS WorldCat Iles T. C. , Beverton R. J. H. 1998 . Stock, recruitment and moderating processes in flatfish . Journal of Sea Research , 39 : 41 – 55 . Google Scholar Crossref Search ADS WorldCat Jackson J. B. , Kirby M. X. , Berger W. H. , Bjorndal K. A. , Botsford L. W. , Bourque B. J. , Bradbury R. H. , et al. 2001 . Historical overfishing and the recent collapse of coastal ecosystems . Science , 293 : 629 – 637 . Google Scholar Crossref Search ADS PubMed WorldCat Karp M. A. , Peterson J. O. , Lynch P. D. , Griffis R. B. , Adams C. F. , Arnold W. S. , Barnett L. , et al. 2019 . Accounting for shifting distributions and changing productivity in the development of scientific advice for fishery management . ICES Journal of Marine Science , 76 : 1305 – 1315 . WorldCat Koenigstein S. , Mark F. C. , Gößling‐Reisemann S. , Reuter H. , Poertner H. O. 2016 . Modelling climate change impacts on marine fish populations: process‐based integration of ocean warming, acidification and other environmental drivers . Fish and Fisheries , 17 : 972 – 1004 . Google Scholar Crossref Search ADS WorldCat Kolody D. , Polacheck T. , Basson M. , Davies C. 2008 . Salvaged pearls: lessons learned from a floundering attempt to develop a management procedure for Southern Bluefin Tuna . Fisheries Research , 94 : 339 – 350 . Google Scholar Crossref Search ADS WorldCat Kurota H. , Hiramatsu K. , Takahashi N. , Shono H. , Itoh T. , Tsuji S. 2010 . Developing a management procedure robust to uncertainty for southern bluefin tuna: a somewhat frustrating struggle to bridge the gap between ideals and reality . Population Ecology , 52 : 359 – 372 . Google Scholar Crossref Search ADS WorldCat Laurel B. J. , Stoner A. W. , Hurst T. P. 2007 . Density-dependent habitat selection in marine flatfish: the dynamic role of ontogeny and temperature . Marine Ecology Progress Series , 338 : 183 – 192 . Google Scholar Crossref Search ADS WorldCat Legault C. M. , Palmer M. C. 2016 . In what direction should the fishing mortality target change when natural mortality increases within an assessment? Canadian Journal of Fisheries and Aquatic Sciences , 73 : 349 – 357 . 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 MacCall A. D. , Klingbeil R. A. , Methot R. D. 1985 . Recent increased abundance and potential productivity of Pacific mackerel (Scomber japonicus) . California Cooperative Oceanic Fisheries Investigations Report , 26 : 119 – 129 . WorldCat Mangel M. , MacCall A. D. , Brodziak J. , Dick E. J. , Forrest R. E. , Pourzand R. , Ralston S. 2013 . A perspective on steepness, reference points, and stock assessment . Canadian Journal of Fisheries and Aquatic Sciences ,. 70 : 930 – 940 . Google Scholar Crossref Search ADS WorldCat Mantua N. J. , Hare S. R. 2002 . The Pacific decadal oscillation . Journal of Oceanography , 58 : 35 – 44 . Google Scholar Crossref Search ADS WorldCat Maunder M. N. 2012 . Evaluating the stock-recruitment relationship and management reference points: application to summer flounder (Paralichthys dentatus) in the U.S. mid-Atlantic . Fisheries Research , 125–126 : 20 – 26 . Google Scholar Crossref Search ADS WorldCat McAllister M. K. , Starr P. J. , Restrepo V. R. , Kirkwood G. P. 1999 . Formulating quantitative methods to evaluate fishery-management systems: what fishery processes should be modelled and what trade-offs should be made? ICES Journal of Marine Science , 56 : 900 – 916 . Google Scholar Crossref Search ADS WorldCat Methot R. D. , Taylor I. G. 2011 . Adjusting for bias due to variability of estimated recruitments in fishery assessment models . Canadian Journal of Fisheries and Aquatic Sciences , 68 : 1744 – 1760 . Google Scholar Crossref Search ADS WorldCat Miller T. J. , Hare J. A. , Alade L. A. 2016 . A state-space approach to incorporating environmental effects on recruitment in an age-structured assessment model with an application to southern New England yellowtail flounder . Canadian Journal of Fisheries and Aquatic Sciences , 73 : 1261 – 1270 . Google Scholar Crossref Search ADS WorldCat Miller T. J. , O’Brien L. , Fratantoni P. S. 2018 . Temporal and environmental variation in growth and maturity and effects on management reference points of Georges Bank Atlantic cod . Canadian Journal of Fisheries and Aquatic Sciences , 75 : 2159 – 2171 . Google Scholar Crossref Search ADS WorldCat Nye J. A. , Baker M. R. , Bell R. , Kenny A. , Kilbourne K. H. , Friedland K. D. , Martino E. , et al. 2014 . Ecosystem effects of the Atlantic multidecadal oscillation . Journal of Marine Systems , 133 : 103 – 116 . Google Scholar Crossref Search ADS WorldCat O’Leary C. A. , Miller T. J. , Thorson J. T. , Nye J. A. 2018 . Understanding historical Summer Flounder (Paralichthys dentatus) abundance patterns through the incorporation of oceanography-dependent vital rates in Bayesian hierarchical models . Canadian Journal of Fisheries and Aquatic Sciences , 999 : 1 – 20 . WorldCat Pershing A. J. , Alexander M. A. , Hernandez C. M. , Kerr L. A. , Le Bris A. , Mills K. E. , Nye J. A. , et al. 2015 . Slow adaptation in the face of rapid warming leads to collapse of the Gulf of Maine cod fishery . Science , 350 : 809 – 812 . Google Scholar Crossref Search ADS PubMed WorldCat Pitchford J. W. , James A. , Brindley J. 2005 . Quantifying the effects of individual and environmental variability in fish recruitment . Fisheries Oceanography , 14 : 156 – 160 . Google Scholar Crossref Search ADS WorldCat Plummer M. 2003 . JAGS: a program for analysis of Bayesian graphical models using Gibbs sampling. In Proceedings of the 3rd International Workshop on Distributed Statistical Computing, 124 , No. 125.10. Punt A. E. , Donovan G. P. 2007 . Developing management procedures that are robust to uncertainty: lessons from the International Whaling Commission . ICES Journal of Marine Science ,. 64 : 603 – 612 . Google Scholar Crossref Search ADS WorldCat Punt A. E. , Butterworth D. S. , de Moor C. L. , De Oliveira J. A. , Haddon M. 2016 . Management strategy evaluation: best practices . Fish and Fisheries , 17 : 303 – 334 . Google Scholar Crossref Search ADS WorldCat Quinn T. J. , Deriso R. B. 1999 . Quantitative Fish Dynamics . Oxford University Press , New York, NY. Google Preview WorldCat COPAC R Core Team. 2017 . R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/ (last accessed 1 December 2018). Stock C. A. , John J. G. , Rykaczewski R. R. , Asch R. G. , Cheung W. W. L. , Dunne J. P. , Friedland K. D. , et al. 2017 . Reconciling fisheries catch and ocean productivity . Proceedings of the National Academy of Sciences of the United of States , 114 : E1441 – E1449 . Google Scholar Crossref Search ADS WorldCat Su Y. S. , Yajima M. 2012 . R2jags: A Package for Running Jags from R. R Package Version 0.03-08. http://CRAN.R-project.org/package=R2jags (last accessed 1 December 2018). Terceiro M. 2015 . Stock Assessment of Summer Flounder for 2015. US Department of Commerce, National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Northeast Fisheries Science Center. Reference Document 15-13. Terceiro M. 2016 . Stock Assessment of Summer Flounder for 2016. US Department of Commerce, National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Northeast Fisheries Science Center. Reference Document 16-15. Thorson J. T. , Monnahan C. C. , Cope J. M. 2015 . The potential impact of time-variation in vital rates on fisheries management targets for marine fishes . Fisheries Research , 169 : 8 – 17 . Google Scholar Crossref Search ADS WorldCat Thorson J. T. , Munch S. B. , Cope J. M. , Gao J. 2017 . Predicting life history parameters for all fishes worldwide . Ecological Applications , 27 : 2262 – 2276 . Google Scholar Crossref Search ADS PubMed WorldCat Tommasi D. , Stock C. A. , Pegion K. , Vecchi G. A. , Methot R. D. , Alexander M. A. , Checkley D. M. 2017 . Improved management of small pelagic fisheries through seasonal climate prediction . Ecological Applications ,. 27 : 378 – 388 . Google Scholar Crossref Search ADS PubMed WorldCat Vert-pre K. A. , Amoroso R. O. , Jensen O. P. , Hilborn R. 2013 . Frequency and intensity of productivity regime shifts in marine fish stocks . Proceedings of the National Academy of Sciences of the United States of America , 110 : 1779 – 1784 . Google Scholar Crossref Search ADS PubMed WorldCat Walters C. , Parma A. M. 1996 . Fixed exploitation rate strategies for coping with effects of climate change . Canadian Journal of Fisheries and Aquatic Sciences , 53 : 148 – 158 . Google Scholar Crossref Search ADS WorldCat Wiedenmann J. , Wilberg M. J. , Miller T. J. 2013 . An evaluation of harvest control rules for data-poor fisheries . North American Journal of Fisheries Management , 33 : 845 – 860 . Google Scholar Crossref Search ADS WorldCat Williams E. H. , Shertzer K. W. 2003 . Implications of life-history invariants for biological reference points used in fishery management . Canadian Journal of Fisheries and Aquatic Sciences , 60 : 710 – 720 . Google Scholar Crossref Search ADS WorldCat Whitten A. R. , Klaer N. L. , Tuck G. N. , Day R. W. 2013 . Accounting for cohort-specific variable growth in fisheries stock assessments: a case study from south-eastern Australia . Fisheries Research , 142 : 27 – 36 . Google Scholar Crossref Search ADS WorldCat © International Council for the Exploration of the Sea 2019. 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)

Martin, Karen L, M;Pierce, Emily, A;Quach, Vincent, V;Studer,, Melissa

doi: 10.1093/icesjms/fsz086pmid: N/A

Abstract California Grunion Leuresthes tenuis (Atherinopsidae), an indigenous endemic marine fish, makes spectacular midnight spawning runs onto sandy beaches on the Pacific coast of California and Baja California. In a unique recreational fishery, people capture the fish out of water with bare hands. Grunion hunters are not required to report their catch, and there is no bag limit. California Grunion rarely appear in trawls and do not take a hook, so population status for this species is impossible to obtain by traditional fishery methods. With citizen scientists, the “Grunion Greeters,” we monitored spawning runs along most of their habitat range. California Grunion recently underwent a northward range extension, but runs appear to be declining broadly across the core habitat. Noisy activities of recreational grunion hunters on shore disrupt spawning runs, preventing fish from reproducing before capture. Leuresthes tenuis has been identified as a Key Indicator Species for the South and Central regions of California Marine Protected Areas, and as an indicator species for climate change on beaches. Gear restrictions, license requirements, and a two-month closed season are rarely enforced late at night. We recommend continued monitoring for L. tenuis in California and increased protections for this unique charismatic fish. Introduction California Grunion Leuresthes tenuis (Atherinopsidae) is an indigenous endemic marine fish on the Pacific coast of California. Famous for forming large assemblages that lead to massive runs, individual fish emerge fully out of waves onto beach sand to spawn (Martin, 2015). Runs may last for over an hour following full or new moons in spring and summer, and fish may cover the beach along the water line (see Supplementary Material). In the traditional habitat range of southern California, between Pt. Conception, California and Punto Abreojos, Mexico, spawning season starts in March and may extend into August, peaking between April and June (Clark, 1938; Walker, 1952). Females dig into the soft wet sand to deposit 1500–3000 eggs while surrounded by males providing milt for external fertilization. Males do not dig into the sand, and may outnumber females by 10 to 1 during the run. Multiple paternity of clutches is typical (Byrne and Avise, 2009), and each male may repeatedly return to shore during a single night’s run (Walker, 1949), providing milt for multiple females with a muscular genital papilla (Aryafar et al., 2019). Thus, multiple waves may carry hundreds of the same individuals over and over again. Females spawn once during a series but can spawn multiple times across the season (Clark, 1925; Walker, 1949). The number of fish on shore cannot be easily counted during a large run, but the density, duration, and extent of the fish are far greater during some runs than others (Walker, 1949; Martin et al., 2007). Leuresthes tenuis is targeted by a unique recreational fishery, solely during these spawning runs (Spratt, 1986; Sandrozinski, 2013). Because of their unusual life cycle, California Grunion are particularly vulnerable to overharvest. Less than 10 years after the first published scientific description of their spawning behaviour (Barnhart, 1918; Thompson,1919), the first regulations to protect them were enacted in 1927 (Clark, 1926, 1938) by the California Department of Fish and Game (now Wildlife), CDFW. At that time, people would line the shore and capture hundreds of grunion with improvised nets made of bed sheets (Andrew Olson, pers. comm.). Early protections included a closure with no take from April to June, the peak of the spawning season, and gear restrictions that specify no gear at all. Only bare hands were (and are) allowed for capturing the fish, presumably to give them a sporting chance while on shore. Those under the age of 16 did not (and still do not) need a fishing license to catch grunion during the open season. Walker (1949) observed grunion runs on Scripps Beach directly following World War II. On the basis of his recommendations, CDFW reduced the closed season to just April and May. Gear restrictions and license requirements remain in place. At that time California’s population was substantially smaller, around 10 million, than it is today, with >35 million people living along one of the most extensively populated and urbanized coasts in the world. During open season there is no bag limit and no requirement to report catch of this species. No commercial use of the species is permitted. Some anglers catch this species for bait, some people catch these small fish to consume whole, but most of those capturing the grunion report they are doing so for the sport, not for any particular use but because it is part of popular culture. In reality, regulations are rarely enforced, in part because spawning runs always occur in the dark of night. Although this endemic species enjoys some unique protections, regulations have not been changed since 1949. California Grunion runs are highlighted in public education programs of coastal public aquariums and California State Beaches, and for youth organizations such as the Boy Scouts. Because runs follow the highest spring tides of full or new moons, likely nights and times can be predicted with some success (Walker, 1952; Spratt, 1986). Especially during closed season, observation of runs can be dazzling, with thousands of fish moving out onto shore from waves for an hour or more. Runs may occur when tides are suitable, within a 2-h window following the highest nightly tide in four nights after full and new moons in spring and summer. However, often on nights when runs are forecast, no grunion are seen on shore. Sandy beaches are critical to L. tenuis as essential fish habitat for spawning (Robbins, 2006). However, beaches in California and worldwide are undergoing habitat loss by coastal squeeze (Defeo et al., 2009; Schoeman et al., 2014; Martin, 2015), with sea level rise and erosion encroaching on the beach from the seaward side, and coastal development and shoreline armouring preventing natural retreat of the beach on the landward side (Dugan et al., 2008). Exacerbated by climate change and increasing human population, California is predicted to lose 31–67% of its sandy beaches by the year 2100 under current predictions of sea level rise (Vitousek et al., 2017). Because of its beach-spawning habits, L. tenuis has been identified as a Key Indicator Species for the South and Central regions of California Marine Protected Area (Marine Protected Area Monitoring Action Plan, 2018), and as an indicator species for climate change on beaches in the Ventura County Coastal Resilience Plan (https://www.vcrma.org/vc-resilient-coastal-adaptation-project). However, monitoring for L. tenuis is problematic. This species has never been abundant (Gregory, 2001). Leuresthes tenuis is planktivorous (Higgins and Horn, 2014); this species does not take a hook. Adults are rarely caught in trawl surveys except within enclosed bays (Allen et al., 2002; Martin et al., 2013; Williams et al., 2016). Recreational fishers are not required to report catch of this species. Thus, traditional fishery methods cannot be used for stock assessments. The only time L. tenuis adults can reliably be observed is during their spawning runs. We developed a group of volunteer citizen scientists, the Grunion Greeters, to report observations of spawning runs on suitable nights all along the California Coast. This started as a way of addressing management issues on sandy beaches, particularly the ecological effects of raking or grooming of beach sand for aesthetic purposes (Martin et al., 2006; Defeo et al., 2009; Dugan and Hubbard, 2010). On the basis of observations and reports across the habitat range over two decades (Martin et al., 2007, 2011), we have become concerned about the status of the California Grunion population as a whole. We hypothesized that this long-term dataset from Grunion Greeter observations would enable us to discern broad trends in population size of this species along its habitat range, in order to guide conservation of this endemic species. Methods Metric for spawning run assessment Strength, duration, and extent of the spawning runs are assessed by a species-specific metric, the Walker Scale, developed in 1999 by the first author with Mike Schaadt and Suzanne Lawrenz-Miller of Cabrillo Marine Aquarium in San Pedro, CA (Table 1). Initially used to compare runs in Malibu with runs in San Pedro, this method was adopted for volunteers in the Grunion Greeter program starting in 2002 (Martin et al., 2007, 2011). The metric was named after Boyd Walker, in honour of his research on the timing of grunion spawning runs, mainly at Scripps Beach in La Jolla, CA. Walker also relied on volunteer observers to assess runs on two nights in 1947 from multiple different beach locations (Walker, 1949), although they used a different metric than ours. Table 1. The Walker Scale for assessment of grunion runs. Scale Number of Grunion on shore at the peak of the run Duration of peak Descriptor W0 No fish or only a few, little or no spawning Up to an hour Not a run W1 Up to 100 fish scattered over a wide area of the beach at a time, some spawning Up to an hour Light run W2 100–500 fish spawning over time, many fish ashore with many of the waves Up to an hour Good run W3 Hundreds of fish spawning at once on several areas of the beach, or thousands in one area Up to an hour or more Strong run W4 Thousands of fish together over a broad area, little sand visible between fish at peak of run Peak lasts minutes up to an hour Excellent run W5 Fish covering the beach several individuals deep, a silver lining of the surf over an extensive area, impossible to walk through run without stepping on fish Peak spawning continues longer than 1 h Incredible run Scale Number of Grunion on shore at the peak of the run Duration of peak Descriptor W0 No fish or only a few, little or no spawning Up to an hour Not a run W1 Up to 100 fish scattered over a wide area of the beach at a time, some spawning Up to an hour Light run W2 100–500 fish spawning over time, many fish ashore with many of the waves Up to an hour Good run W3 Hundreds of fish spawning at once on several areas of the beach, or thousands in one area Up to an hour or more Strong run W4 Thousands of fish together over a broad area, little sand visible between fish at peak of run Peak lasts minutes up to an hour Excellent run W5 Fish covering the beach several individuals deep, a silver lining of the surf over an extensive area, impossible to walk through run without stepping on fish Peak spawning continues longer than 1 h Incredible run Boyd Walker’s pioneering research on grunion provided the scientific basis for understanding the periodicity of the spawning runs in California. The Walker Scale, developed by K. Martin, M. Schaadt, and S. Lawrenz-Miller, is a way to assess the spawning run without actually counting the fish, for comparisons across space and time. Observations should start at or before the time of the highest tides on the four nights following a new or full moon, and continue for 2 h as the tide falls. The number of grunion should be assessed at the peak of the run; most runs start small but some may build up over time. At the peak of the run, how many fish are on shore at any given time? Are they on shore over a short or long period of time? Over a small area or a large extent of the beach? How long does the peak spawning aggregation last? (c) Grunion Greeters and Beach Ecology Coalition, used by permission. View Large Table 1. The Walker Scale for assessment of grunion runs. Scale Number of Grunion on shore at the peak of the run Duration of peak Descriptor W0 No fish or only a few, little or no spawning Up to an hour Not a run W1 Up to 100 fish scattered over a wide area of the beach at a time, some spawning Up to an hour Light run W2 100–500 fish spawning over time, many fish ashore with many of the waves Up to an hour Good run W3 Hundreds of fish spawning at once on several areas of the beach, or thousands in one area Up to an hour or more Strong run W4 Thousands of fish together over a broad area, little sand visible between fish at peak of run Peak lasts minutes up to an hour Excellent run W5 Fish covering the beach several individuals deep, a silver lining of the surf over an extensive area, impossible to walk through run without stepping on fish Peak spawning continues longer than 1 h Incredible run Scale Number of Grunion on shore at the peak of the run Duration of peak Descriptor W0 No fish or only a few, little or no spawning Up to an hour Not a run W1 Up to 100 fish scattered over a wide area of the beach at a time, some spawning Up to an hour Light run W2 100–500 fish spawning over time, many fish ashore with many of the waves Up to an hour Good run W3 Hundreds of fish spawning at once on several areas of the beach, or thousands in one area Up to an hour or more Strong run W4 Thousands of fish together over a broad area, little sand visible between fish at peak of run Peak lasts minutes up to an hour Excellent run W5 Fish covering the beach several individuals deep, a silver lining of the surf over an extensive area, impossible to walk through run without stepping on fish Peak spawning continues longer than 1 h Incredible run Boyd Walker’s pioneering research on grunion provided the scientific basis for understanding the periodicity of the spawning runs in California. The Walker Scale, developed by K. Martin, M. Schaadt, and S. Lawrenz-Miller, is a way to assess the spawning run without actually counting the fish, for comparisons across space and time. Observations should start at or before the time of the highest tides on the four nights following a new or full moon, and continue for 2 h as the tide falls. The number of grunion should be assessed at the peak of the run; most runs start small but some may build up over time. At the peak of the run, how many fish are on shore at any given time? Are they on shore over a short or long period of time? Over a small area or a large extent of the beach? How long does the peak spawning aggregation last? (c) Grunion Greeters and Beach Ecology Coalition, used by permission. View Large Grunion Greeters were trained in a series of short workshops from 2002 to 2018 to understand the Walker Scale categories and assess the number of fish on shore at the peak of the run, the duration of the peak of the run, and the extent of shoreline involved in the peak of the run. Greeters make other observations about the conditions during a night when a grunion run is forecast, including weather and presence of animal predators or grunion hunters. Observers use an online web portal to input their data, usually within 24 h. The data portal is open to the public, and the questionnaire includes an assessment of the experience of the observer and whether or not they attended previous training workshops. See www.Grunion.org for additional details. Grunion Greeter data focus on closed season, April and May, but also includes reports from open season before and after. Because the Greeters are volunteers, the locations and number of reports are not constant from year to year, however some beaches are more consistently observed, and may be considered sentinel beaches. Quality control for Grunion Greeter data All data were evaluated by scientists before use in analysis. Incomplete forms or forms with no identification from the observer were discarded. Forms from dates or times that were unlikely for grunion to run, or from unclear locations were discarded. Grunion Greeters generally work in pairs to provide internal validation. If multiple observer groups on the same run gave different scores, more credence was given to a more experienced, trained observer. Multiple observers on the same run may have different scores because they observed from different locations on the shore; this was evaluated in the reports. Unusual or atypical reports for a location or time are followed up with an e-mail or phone call for additional details. Reports were verified on subsequent days by sampling for presence and density of clutches of eggs in the sand in some but not all cases. For the purposes of this study and to avoid bias for data from certain beaches that have more frequent observations, we selected for each beach, only the highest Walker score reported from each spawning series (the four-day period following a new or full moon), from our verified data. Thus, a spawning series with few grunion on the first two nights after a full moon but a large run on the third would be represented only by the highest Walker score for that series. Data were compared by beach location, county, and year using non-parametric statistics. Data from within the primary habitat of southern California, containing over 90% of the species population (Martin et al., 2013; Martin, 2015), were analysed separately from much sparser data for the central coast that followed a northward range extension in 2002 (Roberts et al., 2007; Johnson et al., 2009). Results Since 2002, over 4500 Grunion Greeters have provided over 5000 reports. This Grunion Greeter compilation is the most complete dataset for spawning runs of this species in existence, both in terms of geographic coverage and duration of observations. Reports have come from the entire range of the species, over 50 beaches in California and Baja California, Mexico. A northern range extension for spawning runs was discovered in 2002 in San Francisco Bay (Johnson et al., 2009), followed by a northward range extension to Tomales Bay in 2005 (Roberts et al., 2007). Many Grunion Greeters provided multiple observations over several years. Verified data from professional biologists using our methods to observe California Grunion as part of their monitoring efforts for coastal construction projects are also included. Grunion Greeters reliably report the location of a run and its strength, based on both multiple independent observations of the same run, and on sporadic post-run sampling of beaches for clutches. In 445 runs with multiple observers, there is 87.6% agreement on the ranking of the Walker Scale. Even with disagreement, scores rarely differ more than one rank between observers. The core of the habitat range is from the border of California and Mexico in San Diego County through Orange County and Los Angeles County through Malibu. From 2002 to 2010, typically the median run strength in this core area was W2, with a small percentage of the runs at W4 or W5 level (Figure 1). Large spawning runs (W4 and W5) have been seen in every year, on occasion. On a year with a low median, the number of large runs is very low as well. Although large runs still occurred in 2018, in 6 of the past 8 years, 75% of the runs have been W2 or lower in the core habitat for this endemic species. Figure 1. View largeDownload slide When the Grunion Greeters started, median (heavy bars) run size was a moderate but effective W2 in the core species habitat of southern California. Since 2010, the median of runs reported has been no higher than W1, meaning that at least 50% of the runs observed do not hold significant spawning activity. In two years (2014 and 2016) the median was W0, meaning that >50% of the time runs were predicted, few or no spawning fish were present. From 2011 to 2018, the median across the traditional habitat range typically was W1 and twice was W0. N = 3462. Figure 1. View largeDownload slide When the Grunion Greeters started, median (heavy bars) run size was a moderate but effective W2 in the core species habitat of southern California. Since 2010, the median of runs reported has been no higher than W1, meaning that at least 50% of the runs observed do not hold significant spawning activity. In two years (2014 and 2016) the median was W0, meaning that >50% of the time runs were predicted, few or no spawning fish were present. From 2011 to 2018, the median across the traditional habitat range typically was W1 and twice was W0. N = 3462. Examining by county, runs in Los Angeles County, Orange County, and San Diego County have decreased in Walker Score over the time of the study (Figure 2). The five years 2004–2008 compared with the five years 2014–2018 show a significant decrease in the Walker Score of runs in the core habitat over time. This decline is consistent whether testing the three core counties together (Figure 1), looking within individual counties in southern California (Figure 2), or comparing across time within individual sentinel beaches (Figure 3). For the three core counties, significant differences are seen in frequencies of large and small runs between decades (N = 1952, X2 = 18.42, df = 5, p < 0.01). By county, these differences are also significant. For San Diego County, N = 742, X2 = 11.81, df = 5, p < 0.037; for Orange County, N = 500, X2 = 78.12, df = 5, p < 0.0001; and for Los Angeles County N = 465, X2 = 18.5, df = 5, p < 0.01). Figure 2. View largeDownload slide Reports from Grunion Greeters indicate that median (heavy bars) run size based on the Walker Scale have significantly decreased over time for each of the three southern counties. (a) San Diego, (b) Orange, and (c) Los Angeles. Figure 2. View largeDownload slide Reports from Grunion Greeters indicate that median (heavy bars) run size based on the Walker Scale have significantly decreased over time for each of the three southern counties. (a) San Diego, (b) Orange, and (c) Los Angeles. Figure 3. View largeDownload slide Proportions of runs that are small (W0 or W1), medium (W2 or W3), and large (W4 or W5) in five sentinel beaches in the core habitat range of southern California. Median runs dropped over the past decade and the likelihood of large runs decreased significantly in all cases. Figure 3. View largeDownload slide Proportions of runs that are small (W0 or W1), medium (W2 or W3), and large (W4 or W5) in five sentinel beaches in the core habitat range of southern California. Median runs dropped over the past decade and the likelihood of large runs decreased significantly in all cases. Runs are highly variable in space and time. Although on a given night one beach may hold a large run, other beaches on the same night or run series may show little activity (Figure 4). The proportion of runs that are small (W0 or W1) has significantly increased over the past 15 years (Spearman Rank Correlation Coefficient rs = 0.57, df = 13, p = 0.025). For the three counties of San Diego, Orange, and Los Angeles, small runs were 48.9% of reports from five years between 2004 and 2008, and increased to 65.4% of reports in the 5 years from 2014 to 2018. The proportion of runs at the W5 level has remained low and fairly consistent over the years, 1.58 ± 0.76% of reports in a given year. Figure 4. View largeDownload slide For one April night, beaches from San Diego, Orange, Los Angeles, Ventura, and Santa Barbara counties show the variability in run strength. The median run score is W2 for these 12 beaches. Figure 4. View largeDownload slide For one April night, beaches from San Diego, Orange, Los Angeles, Ventura, and Santa Barbara counties show the variability in run strength. The median run score is W2 for these 12 beaches. Runs north of the core habitat seem to be increasing according to our reports, although not yet significantly (Figure 5). The areas of northward range extension around San Francisco Bay underwent local extirpation in 2008 (Martin et al., 2013) but have been re-colonized in 2014. Runs in locations in and around San Francisco Bay start later, in May rather than March, and continue into August, with the largest runs usually in July and August. Figure 5. View largeDownload slide Runs appear to be increasing north of the core habitat range, but these differences are not significant. (a) Ventura and Santa Barbara Counties are north of the core habitat but within the traditional spawning range of L. tenuis. (b) L. tenuis colonized San Francisco Bay and points north in 2002, and then was locally extirpated by 2008. They returned in 2014 and runs are increasing in strength. Heavy line is median. Figure 5. View largeDownload slide Runs appear to be increasing north of the core habitat range, but these differences are not significant. (a) Ventura and Santa Barbara Counties are north of the core habitat but within the traditional spawning range of L. tenuis. (b) L. tenuis colonized San Francisco Bay and points north in 2002, and then was locally extirpated by 2008. They returned in 2014 and runs are increasing in strength. Heavy line is median. Grunion Greeters reported poaching (catching out of season, without a license, or with the use of any gear) in ∼20% of reports during closed season, and hunting or poaching for 93% of reports during open season. California fishers are not required to display a license while fishing. Informal questioning indicated that many adults hunting grunion during runs did not purchase a fishing license. Game Wardens were rarely observed during runs, <5 instances out of 5133 reports. Active hunting was often accompanied by loud, raucous crowds and high disturbance and prevention of spawning (Table 2). Table 2. Grunion Greeter reports indicate high levels of disturbance of spawning by people hunting. “Unruly THOUSANDS, some in water, all making noise. Looked like some sort of post-apocalyptic marine Mad Max.’’ “The few grunion that actually came up onto the beach were automatically grabbed by poachers. There were probably 20–30 people taking the fish last night.” “Hundreds of people on beach, many using buckets and strainers to collect fish; informed them of regulations.” (report from a marine biologist with California Department of Fish and Wildlife). “A large group of people gathered at least 10 plastic grocery bags full of grunion and women were walking behind them laughing and kicking the grunion. Many people were taking several hundred grunion home in trash bags.” “Over a hundred people in a frenzy to get the few fish that came in with each wave. Lots of screaming kids, dogs, and flashlights.” “Three families harvested hundreds.” “One goofy guy was running wildly up and down the beach with a flashlight and grabbing at any fish that started to spawn.” “Hunting–Splashing into water, capturing in water or at surf’s edge, noisy, yelling, screaming.” “Lots of youngsters excited and splashing in the shallows chasing grunion. Probably they harvested 200 or 300. There were maybe 50+ in groups of 4–10 running to and fro.” “There was a very rowdy group of ∼10 people, catching and collecting the grunion during the entire run, yelling and chasing after the fish into the water, up to even waist deep!” “Bad behavior: Kicking fish, throwing, stepping, or jumping on them.” “TONS of people. At the first big sighting of fish the people rushed the water & the grunion fled.” “There was a pack of ∼12–14 non-English speaking people stomping on and kicking fish on the beach. One run of grunion had started and when these people behaved in this way that run went back into the water and did not return to that location.” “Poachers continuously ignored our information very frustrating. Picking them up filling buckets and stepping on them and ripping them in half.” “Fish tried to come ashore but a crazy mob of people lined beach with buckets & lights.” “Unruly THOUSANDS, some in water, all making noise. Looked like some sort of post-apocalyptic marine Mad Max.’’ “The few grunion that actually came up onto the beach were automatically grabbed by poachers. There were probably 20–30 people taking the fish last night.” “Hundreds of people on beach, many using buckets and strainers to collect fish; informed them of regulations.” (report from a marine biologist with California Department of Fish and Wildlife). “A large group of people gathered at least 10 plastic grocery bags full of grunion and women were walking behind them laughing and kicking the grunion. Many people were taking several hundred grunion home in trash bags.” “Over a hundred people in a frenzy to get the few fish that came in with each wave. Lots of screaming kids, dogs, and flashlights.” “Three families harvested hundreds.” “One goofy guy was running wildly up and down the beach with a flashlight and grabbing at any fish that started to spawn.” “Hunting–Splashing into water, capturing in water or at surf’s edge, noisy, yelling, screaming.” “Lots of youngsters excited and splashing in the shallows chasing grunion. Probably they harvested 200 or 300. There were maybe 50+ in groups of 4–10 running to and fro.” “There was a very rowdy group of ∼10 people, catching and collecting the grunion during the entire run, yelling and chasing after the fish into the water, up to even waist deep!” “Bad behavior: Kicking fish, throwing, stepping, or jumping on them.” “TONS of people. At the first big sighting of fish the people rushed the water & the grunion fled.” “There was a pack of ∼12–14 non-English speaking people stomping on and kicking fish on the beach. One run of grunion had started and when these people behaved in this way that run went back into the water and did not return to that location.” “Poachers continuously ignored our information very frustrating. Picking them up filling buckets and stepping on them and ripping them in half.” “Fish tried to come ashore but a crazy mob of people lined beach with buckets & lights.” View Large Table 2. Grunion Greeter reports indicate high levels of disturbance of spawning by people hunting. “Unruly THOUSANDS, some in water, all making noise. Looked like some sort of post-apocalyptic marine Mad Max.’’ “The few grunion that actually came up onto the beach were automatically grabbed by poachers. There were probably 20–30 people taking the fish last night.” “Hundreds of people on beach, many using buckets and strainers to collect fish; informed them of regulations.” (report from a marine biologist with California Department of Fish and Wildlife). “A large group of people gathered at least 10 plastic grocery bags full of grunion and women were walking behind them laughing and kicking the grunion. Many people were taking several hundred grunion home in trash bags.” “Over a hundred people in a frenzy to get the few fish that came in with each wave. Lots of screaming kids, dogs, and flashlights.” “Three families harvested hundreds.” “One goofy guy was running wildly up and down the beach with a flashlight and grabbing at any fish that started to spawn.” “Hunting–Splashing into water, capturing in water or at surf’s edge, noisy, yelling, screaming.” “Lots of youngsters excited and splashing in the shallows chasing grunion. Probably they harvested 200 or 300. There were maybe 50+ in groups of 4–10 running to and fro.” “There was a very rowdy group of ∼10 people, catching and collecting the grunion during the entire run, yelling and chasing after the fish into the water, up to even waist deep!” “Bad behavior: Kicking fish, throwing, stepping, or jumping on them.” “TONS of people. At the first big sighting of fish the people rushed the water & the grunion fled.” “There was a pack of ∼12–14 non-English speaking people stomping on and kicking fish on the beach. One run of grunion had started and when these people behaved in this way that run went back into the water and did not return to that location.” “Poachers continuously ignored our information very frustrating. Picking them up filling buckets and stepping on them and ripping them in half.” “Fish tried to come ashore but a crazy mob of people lined beach with buckets & lights.” “Unruly THOUSANDS, some in water, all making noise. Looked like some sort of post-apocalyptic marine Mad Max.’’ “The few grunion that actually came up onto the beach were automatically grabbed by poachers. There were probably 20–30 people taking the fish last night.” “Hundreds of people on beach, many using buckets and strainers to collect fish; informed them of regulations.” (report from a marine biologist with California Department of Fish and Wildlife). “A large group of people gathered at least 10 plastic grocery bags full of grunion and women were walking behind them laughing and kicking the grunion. Many people were taking several hundred grunion home in trash bags.” “Over a hundred people in a frenzy to get the few fish that came in with each wave. Lots of screaming kids, dogs, and flashlights.” “Three families harvested hundreds.” “One goofy guy was running wildly up and down the beach with a flashlight and grabbing at any fish that started to spawn.” “Hunting–Splashing into water, capturing in water or at surf’s edge, noisy, yelling, screaming.” “Lots of youngsters excited and splashing in the shallows chasing grunion. Probably they harvested 200 or 300. There were maybe 50+ in groups of 4–10 running to and fro.” “There was a very rowdy group of ∼10 people, catching and collecting the grunion during the entire run, yelling and chasing after the fish into the water, up to even waist deep!” “Bad behavior: Kicking fish, throwing, stepping, or jumping on them.” “TONS of people. At the first big sighting of fish the people rushed the water & the grunion fled.” “There was a pack of ∼12–14 non-English speaking people stomping on and kicking fish on the beach. One run of grunion had started and when these people behaved in this way that run went back into the water and did not return to that location.” “Poachers continuously ignored our information very frustrating. Picking them up filling buckets and stepping on them and ripping them in half.” “Fish tried to come ashore but a crazy mob of people lined beach with buckets & lights.” View Large Clutches of eggs are buried 10–20 cm deep in beach sand in a band no >1–3 m wide parallel to shore on the upper beach in the mid to high intertidal zone. Considering a narrow strip on average ∼3 m wide along 483 km of sandy beaches in southern California results in a total spawning habitat area of 1.45 km2 for L. tenuis in its core primary habitat at the current time. Discussion California Grunion spawning runs can be assessed with the help of citizen scientists; in fact this may be the only way to obtain these extensive, hyperlocal data. The Walker Scale is currently used by professional resource biologists to monitor grunion runs for agencies such as US Army Corps of Engineers, California Department of Fish and Wildlife, California Coastal Commission, National Marine Fisheries Service, and California State Parks, as well as for public educational programs at Cabrillo Aquarium and Birch Aquarium at Scripps, among others (Martin et al., 2011). The Walker Scale is an effective, accurate, non-invasive although labour-intensive method for assessment of this species and other beach-spawning fishes. While the data from professional biologists monitoring grunion runs for coastal projects are certainly reliable, the number, locations, and frequency of these short-term projects are small relative to the substantial, long-term efforts of volunteer Grunion Greeters. Even though large runs can still be observed, the median Walker Score for California Grunion spawning on shore has declined significantly across much of the core habitat range in the past ten years (Figure 1). This pattern is consistent for this endemic fish across the three coastal counties constituting its core habitat (Figure 2) and within individual beaches known historically for large spawning runs of grunion (Figure 3). The occasional presence of large spawning aggregations may create the illusion of abundance even when a population is depleted (Erisman et al., 2011). These occasional large runs may tempt resource managers to believe that these kinds of runs are both more common and more widespread geographically than is the actual situation (Figure 4, Sadovy and Domeier, 2005). On the basis of reports from Grunion Greeters and resource biologists, California Grunion appear to be both shifting their habitat range northward (Figure 5) and decreasing in numbers in the more southern habitats (Figures 1 and 2). Warming trends in ocean water and the atmosphere may be affecting this species (Martin, 2015), along with ocean acidification (Tasoff and Johnson, 2019). There is an environmental component to sex determination of L. tenuis, so that warmer temperatures during early life result in greater proportions of males (Brown et al., 2014). Of more immediate concern, their critical spawning habitat is also declining (Dugan et al., 2008; Vitousek et al., 2017; King et al., 2018), potentially concentrating the spawning population into fewer locations on shore. The spawning zone of L. tenuis, the upper beach between the mid and high intertidal zone (Martin et al., 2006), is also the beach area that is most vulnerable to loss by coastal squeeze (Dugan and Hubbard, 2010; Schooler et al., 2017). The core spawning habitat total area of 1.45 km2 for L. tenuis is smaller than Dodger Stadium or the Los Angeles International Airport. The minimum size is 25 km2 for one Marine Protected Area (MPA) in California (Botsford et al., 2014), in a network of over 100 MPAs. This critical habitat for L. tenuis is likely to decrease, and is already <0.001% of the area of the California MPA network. Even though the species has managed to shift its habitat and colonize some northern bays, the northern ecotype grows to a smaller adult size, spawns less frequently, and produces significantly fewer, smaller eggs per clutch (Johnson et al., 2009; Martin et al., 2013). For these reasons the northern populations are more vulnerable to ecosystem perturbations and local extirpation than the populations in the traditional habitat. In addition, the more northern populations spawn on a different annual schedule than the southern populations of this species, and therefore the peak run times of the northern populations are not protected by the current closed season of April and May. These northern fish are neither different genetically (Johnson et al., 2009; Byrne et al., 2013) nor are they different in physiological response to temperature (Brown et al., 2012) from the southern grunion, so this habitat shift appears to be restricted to areas of bays that are warmer than the waters of the open ocean. Fished species that form spawning aggregations face an increased extinction risk (Sadovy and Erisman, 2012). Modern conservation practices almost universally protect the reproductive period and spawning aggregations of species (Hutchings, 2001). The regulations for fishing on California Grunion do the opposite by specifically targeting the spawning aggregations, striking this species at its most vulnerable and critical time, disrupting its ability to produce the next generations. Fishing on large aggregations can mask population declines or collapse (Erisman et al., 2011). Regulations put in place to protect the endemic California Grunion during spawning runs are rarely and unevenly enforced. Poaching during closed season is common on some urban beaches, and reported during ∼20% of closed season observations. Collection of spawning fish by people with or without fishing licenses is nearly universal during open season, identified in the vast majority of open season reports, disrupting runs, and preventing reproduction while removing ripe adults from the population (Table 2). Many grunion hunters do not fish for any other species, and do not possess fishing licenses. Children, not required to have a license, are very effective hunters (see Supplementary Material). Thus the potential number of people hunting California Grunion is far greater than the 2.5 million sport fishing licenses that were sold in California in 2016 (https://www.wildlife.ca.gov/Licensing/Statistics#SportFishingLicenses). Data from entrainment surveys are the only other long term dataset available for L. tenuis. The entrainment data conforms with CalCOFi nearshore trawl data pattern (Miller and McGowan, 2013). For California Grunion, usually less than one, or fewer than two individuals are seen per million cubic meter flow (E. Miller, pers. comm.). Compared with other local silverside fishes, for Topsmelt Atherinops affinis 14.6, and Jacksmelt Atherinopsis californiensis 39.4 are present per million cubic meters flow at a peak. Both A. affinis and A. californiensis are fished commercially and recreationally, with hundreds of thousands landed each year (Vejar, 2013). These fishery-independent surveys indicate at a minimum that L. tenuis abundance is substantially lower than its sister silverside species of similar size. Trawl surveys of San Diego Bay (Williams et al., 2016) and San Francisco Bay (Johnson et al., 2009) show large population fluctuations from year to year. In 2016 Williams et al. suggested a stock estimate for L. tenuis in San Diego Bay of 785,183 fish, but 92% were juveniles in surveys taken during the spawning season. This suggests substantially fewer, only 62,815 adult grunion in San Diego Bay in 2016. The human population of San Diego’s metropolitan area is 3.1 million, http://worldpopulationreview.com/us-cities/san-diego-population/ not including the city’s 35 million tourist visitors per year (https://www.sandiego.org/about.aspx). Because of the tendency of this species to aggregate, we hypothesize that even if fewer fish are present in the total population, large runs will still occur on occasion. Our observations suggest that it is likely that a minimum number of fish must be present for a spawning run to occur. Runs with fewer than a hundred individuals usually do not include spawning events or egg deposition. Therefore the presence of only small numbers of fish during a run suggests unsuccessful reproduction. As runs decline, fewer observations can be made. If the population declines, fewer locations will hold runs, and those runs will occur less frequently. The consistent pattern of decline in median run size is of great concern for this endemic indigenous species. We suggest it is possible that the numbers of adult fish could drop too low for successful spawning even when some members of the species are present and ripe. The sister species, Leuresthes sardina the Gulf Grunion, is endemic to the northern Gulf of California (Bernardi et al., 2003). This species shares the beach-spawning habits of L. tenuis (Thomson and Muench, 1976). Leuresthes sardina appears on the IUCN Red List as “Near Threatened” because of potential habitat loss and human interference (Findley et al., 2010). The California Grunion L. tenuis may face even greater threats because of larger human populations and more coastal development in California compared with Mexico. In summary, large spawning runs still occur for L. tenuis, but smaller runs have been much more common in the present decade than in the previous one in its core habitat range. There may be fewer California Grunion, or the fish may not able to spawn as frequently as in the past. Either way, reproductive output appears to be lower. For those populations that have moved north, the shift in habitat comes at the cost of smaller size and reduced clutch size, as well a shift in spawning season that is shorter and holds less frequent spawning. We strongly encourage increased protection of the spectacular spawning runs for this charismatic indigenous endemic marine fish. Its status as a managed species and an indicator species for climate change warrant greater concern. At minimum, a return to closed season from April to June, as originally designated in 1927, would help protect the southern population from fishing pressure. We recommend that the L. tenuis population on the central coast, in Monterey Bay and around San Francisco Bay, should be completely closed to take, as the populations there appear to be too small to withstand any fishing pressure. Outreach with the Grunion Greeters may help shift public perception of this species and their interaction with its runs. Greeters report with dismay that those hunting L. tenuis during its spawning runs exploit the vulnerability of these fish when out of water (Table 2). Unlike typical fishers who respectfully interact with the resource and take no more than they will use, grunion hunters often say they are following some sort of (perhaps misguided) cultural tradition. They scream and yell while running to wildly chase the fish that are trying to spawn. They sometimes step on the fish in their haste, breaking their backs; then toss them into buckets to expire. Instead, we hope that more and more people will come to quietly observe the run spectacle on its own terms, without disturbing the fish, as watchable wildlife. All should be able to simply enjoy the amazing sight of California’s original surfers dancing on the beach. Acknowledgements We are thankful for funding from US Fish & Wildlife Service, “Connecting People with Nature,” California Coastal Commission Whale Tail Program WT-13-22, National Science Foundation DBI 1062721, National Science Foundation, REU-1560352, USC Sea Grant College – Urban Oceans Program NOAA – NA14OAR4170089/Subaward 6094463, National Marine Fisheries Service, Southwest Region, Habitat Conservation Division Contract 8-819, National Geographic Society CRE 8105-07, and Pepperdine University. We are grateful to thousands of Grunion Greeters for their long walks on moonlit beaches. RD Martin provided helpful comments on the manuscript and C Davis, T Furlong, and M Perrault assisted with analyses. References Allen L. G. , Findlay A. M. , Phalen C. M. 2002 . Structure and standing stock of the fish assemblages of San Diego Bay, California from 1994 to 1999 . Bulletin Southern California Academy of Sciences , 101 : 49 – 85 . Aryafar H. , Carrillo A. , Berquist R. , Frank L. R. , Forsgren K. 2019 . Description of a male genital papilla in the California grunion, a beach-spawning marine silverside fish. Bulletin of the Southern California Academy of Sciences, in press. Barnhart P. S. 1918 . The spawning of the little-smelt Leuresthes tenuis (Ayres ). California Fish & Game , 4 : 181 – 182 . Bernardi G. , Findley L. , Rocha-Olivares A. 2003 . Vicariance and dispersal across Baja California in disjunct marine fish populations . Evolution , 57 : 1599 – 1609 . Google Scholar Crossref Search ADS PubMed Botsford L. W. , White J. W. , Carr M. H. , Caselle J. E. 2014 . Marine protected area networks in California, USA . Advances in Marine Biology , 69 : 205 – 251 . Google Scholar Crossref Search ADS PubMed Brown E. E. , Baumann H. , Conover D. O. 2012 . Absence of countergradient and cogradient variation in an oceanic silverside, the California Grunion Leuresthes tenuis . Marine Ecology Progress Series , 461 : 175 – 186 . Google Scholar Crossref Search ADS Brown E. E. , Baumann H. , Conover D. O. 2014 . Temperature and photoperiod effects on sex determination in Leurethes tenuis (fish), Supplement to: temperature and photoperiod effects on sex determination in a fish . Journal of Experimental Marine Biology and Ecology , 461 : 39 – 43 . Google Scholar Crossref Search ADS Byrne R. , Avise J. 2009 . Multiple paternity and extra-group fertilizations in a natural population of California grunion (Leuresthes tenuis), a beach-spawning marine fish . Marine Biology , 156 : 1681 – 1690 . Google Scholar Crossref Search ADS Byrne R. J. , Bernardi G. , Avise J. 2013 . Spatiotemporal genetic structure in a protected marine fish, the California Grunion (Leuresthes tenuis), and relatedness in the genus Leuresthes . Journal of Heredity , 104 : 521 – 531 . Google Scholar Crossref Search ADS PubMed Clark F. N. 1925 . The life history of Leuresthes tenuis, an Atherine fish with tide controlled spawning habits. Contribution No. 51, State Fisheries Laboratory. Fish Bulletin Number 10, California State Fish and Game Commission. Clark F. N. 1926 . Conservation of the grunion . California Fish & Game , 12 : 163 – 166 . Clark F. N. 1938 . Grunion in southern California . California Fish & Game , 24 : 49 – 54 . Defeo O. , McLachlan A. , Schoeman D. S. , Schlacher T. A. , Dugan J. , Jones A. , Lastra M. , et al. 2009 . Threats to sandy beach ecosystems: a review . Estuarine, Coastal and Shelf Science , 81 : 1 – 12 . Google Scholar Crossref Search ADS Dugan J. E. , Hubbard D. M. , Rodil I. F. , Revell D. L. , Schroeter S. 2008 . Ecological effects of coastal armoring on sandy beaches . Marine Ecology , 29 : 160 – 170 . Google Scholar Crossref Search ADS Dugan J. E. , Hubbard D. M. 2010 . Loss of coastal strand habitat in southern California: the role of beach grooming . Estuaries and Coasts , 33 : 67 – 77 . Google Scholar Crossref Search ADS Erisman B. E. , Allen L. G. , Claisse J. T. , Pondella D. J. II , Miller E. F. , Murray J. H. 2011 . The illusion of plenty: hyperstability masks collapses in two recreational fisheries that target fish spawning aggregations . Canadian Journal of Fisheries and Aquatic Science , 68 : 1705 – 1716 . Google Scholar Crossref Search ADS Findley L. , Espinosa H. , Collette B. , Rojas P. 2010 . Leuresthes sardina. The IUCN Red List of Threatened Species 2010: e.T183267A8083578. doi:10.2305/IUCN.UK.20103.RLTS.T183267A8083578.en. Gregory P. A. 2001 . Grunion. In California’s Living Marine Resources: A Status Report, pp. 246–247 . Ed. by Leet W. S. , Dewees C. M. , Klingbeill R. , Larson E. J. . California Department of Fish and Game , Sacramento, CA . 592 pp. Higgins B. A. , Horn M. H. 2014 . Suction among pickers: jaw mechanics, dietary breadth and feeding behavior in beach-spawning Leuresthes spp. compared with their relatives . Journal of Fish Biology , 84 : 1689 – 1707 . Google Scholar Crossref Search ADS PubMed Hutchings J. A. 2001 . Conservation biology of marine fishes: perceptions and caveats regarding assignment of extinction risk . Canadian Journal of Fisheries and Aquatic Science , 58 : 108 – 121 . Johnson P. B. , Martin K. L. , Vandergon T. L. , Honeycutt R. L. , Burton R. S. , Fry A. 2009 . Microsatellite and mitochondrial genetic comparisons between northern and southern populations of California Grunion Leuresthes tenuis . Copeia , 2009 : 467 – 476 . Google Scholar Crossref Search ADS King P. G. , Nelsen C. , Dugan J. E. , Hubbard D. M. , Martin K. L. 2018 . Valuing beach ecosystems in an age of retreat . Shore & Beach , 86 : 1 – 15 . Marine Protected Area Monitoring Action Plan. California Department of Fish and Wildlife and California Ocean Protection Council, California, USA. October 2018 . https://www.wildlife.ca.gov/Conservation/Marine/MPAs/Management/Monitoring/Action-Plan Martin K. L. M. 2015 . Beach-Spawning Fishes: Reproduction in an Endangered Ecosystem . Taylor & Francis Group, CRC Press , Oxford, UK . 219 pp. Martin K. L. M. , Hieb K. A. , Roberts D. A. 2013 . A southern California icon surfs north: local ecotype of California Grunion Leuresthes tenuis (Atherinopsidae) revealed by multiple approaches during temporary habitat expansion into San Francisco Bay . Copeia , 2013 : 729 – 730 . Google Scholar Crossref Search ADS Martin K. L. M. , Moravek C. L. , Martin A. D. , Martin R. D. 2011 . Community based monitoring improves management of essential fish habitat for beach-spawning California Grunion . Sandy Beaches and Coastal Zone Management: Proceedings of the Fifth International Symposium on Sandy Beaches, Rabat, Morocco. Travaux de l’ Institut Scientifique , 2011 : 65 – 72 . Martin K. , Speer-Blank T. , Pommerening R. , Flannery J. , Carpenter K. 2006 . Does beach grooming harm grunion eggs? Shore & Beach , 74 : 17 – 22 . Martin K. , Staines A. , Studer M. , Stivers C. , Moravek C. , Johnson P. , Flannery J. 2007 . Grunion Greeters in California: beach-spawning fish, coastal stewardship, beach management and ecotourism. In Proceedings of the 5th International Coastal & Marine Tourism Congress: Balancing Marine Tourism, Development and Sustainability , pp. 73 – 86 . Ed. by Lück M. , Gräupl J. , Auyong J. , Miller M. L. , Orams M. B. . New Zealand Tourism Research Institute , Auckland, New Zealand . Miller E. F. , McGowan J. A. 2013 . Faunal shift in southern California's coastal fishes: a new assemblage and trophic structure takes hold . Estuarine, Coastal and Shelf Science , 127 : 29 – 36 . Google Scholar Crossref Search ADS Robbins E. 2006 . Essential Fish Habitat in Santa Monica Bay, San Pedro Bay, and San Diego Bay: A Reference Guide for Managers. MS thesis, Duke University, 129 pp. Roberts D. , Lea R. N. , Martin K. L. M. 2007 . First record of the occurrence of the California Grunion, Leuresthes tenuis, in Tomales Bay, California; a northern extension of the species . California Fish & Game , 93 : 107 – 110 . 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 Sadovy Y. , Erisman B. E. , 2012 . The social and economic importance of aggregating species and the biological implications of fishing on spawning aggregations. In Reef Fish Spawning Aggregations: Biology, Research and Management Edition , pp. 225 – 284 . Ed. by Sadovy de Mitcheson Y. , Colin P. . Springer , New York . doi:10.1007/978-94-007-1980-4_8 Sandrozinski A. 2013 . California Grunion. Status of the Fisheries Report, an Update Through 2011 . California Department of Fish & Wildlife , Sacramento, CA . Schoeman D. S. , Schlacher T. A. , Defeo O. 2014 . Climate‐change impacts on sandy‐beach biota: crossing a line in the sand . Global Change Biology , 20 : 2383 – 2392 . Google Scholar Crossref Search ADS PubMed Schooler N. K. , Dugan J. E. , Hubbard D. M. , Straughan D. 2017 . Local scale processes drive long-term change in biodiversity of sandy beach ecosystems . Ecology and Evolution , 7 : 4822 – 4834 . Google Scholar Crossref Search ADS PubMed Spratt J. D. 1986 . The amazing grunion. Marine Resource Leaflet No. 3, California Department of Fish and Game, Sacramento, California. Tasoff A. J. , Johnson D. W. 2019 . Can larvae of a marine fish adapt to ocean acidification? Evaluating the evolutionary potential of California Grunion (Leuresthes tenuis ). Evolutionary Applications , 12 : 560 – 571 . Google Scholar Crossref Search ADS PubMed Thomson D. A. , Muench K. A. 1976 . Influence of tides and waves on the spawning behavior of the Gulf of California grunion, Leuresthes sardina (Jenkins and Evermann) . Bulletin of the Southern California Academy of Science , 75 : 198 – 203 . Thompson W. F. 1919 . The spawning of the grunion (Leuresthes tenuis ). California Fish & Game , 5 : 1 – 27 . Vejar A. 2013 . Silversides. Status of the Fisheries Report, an Update through 2011 . California Department of Fish & Wildlife , Sacramento, CA . Vitousek S. , Barnard P. L. , Limber P. , Erikson L. , Cole B. 2017 . A model integrating longshore and cross-shore processes for predicting long-term shoreline response to climate change . JGR Earth Surface , 122 : 782 – 806 . https://doi.org/10.1002/2016JF004065 Walker B. W. 1949 . The Periodicity of Spawning by the Grunion, Leuresthes tenuis, an Atherine Fish . Doctoral dissertation, University of California , Los Angeles . Walker B. W. 1952 . A guide to the grunion . California Fish & Game , 38 : 409 – 420 . Williams J. P. , Pondella D. J. II , Williams C. M. , Robart M. J. 2016 . Fisheries Inventory and Utilization Study to Determine Impacts from El Niño in San Diego Bay, San Diego, California for Surveys conducted in April and July 2016. Unified Port of San Diego, San Diego, CA. © International Council for the Exploration of the Sea 2019. 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)

Lonthair,, Joshua;Hwang,, Pung-Pung;Esbaugh, Andrew, J

doi: 10.1093/icesjms/fsaa023pmid: N/A

Abstract Ocean acidification (OA) and other climate change-induced environmental alterations are resulting in unprecedented rates of environmental degradation. This environmental change is generally thought to be too fast for adaptation using evolutionary process dependent on natural selection, and thus, resilience may be related to the presence of existing tolerant genotypes and species. Estuaries undergo natural partial pressure carbon dioxide (pCO2) fluctuations, with levels regularly exceeding predicted end of the century values. In this study, we use the estuarine orange-spotted grouper (Epinephelus coioides) to explore the intrinsic resilience to elevated pCO2. Our sensitivity endpoints included: survival, heart rate, growth, and yolk consumption. Furthermore, we attempted to determine whether their acid–base regulatory machinery was plastic in response to elevated pCO2 by analysing the gene expression of key transporters and ionocyte density. Survival was not significantly altered by exposure to elevated pCO2. Interestingly, the heart rate was significantly elevated at both 1500 and 3100 μatm exposure. However, other metrics of energetic consumption, such as yolk consumption and growth, were not significantly altered. Furthermore, we found no changes in gene expression in vha, nhe3, and nbc, as well as ionocyte density at elevated pCO2. Overall, these results support the hypothesis that estuarine species are resilient to the impacts of OA. Introduction Since the beginning of the industrial revolution, anthropogenic carbon dioxide (CO2) emissions have had measurable impacts on the oceanic carbonate chemistry—colloquially known as ocean acidification (OA). Models project that partial pressure CO2 (pCO2) could increase to 1000 μatm by the of the century, which would represent a 250% increase in pCO2 from present oceanic levels (Caldeira and Wickett, 2003; Solomon et al., 2007; Doney et al., 2009; McNeil and Sasse, 2016). This increase in pCO2 would result in a 0.3–0.4 units reduction in pH (Orr et al., 2005). Studies on the impacts of elevated pCO2 exposure on marine teleosts have found a plethora of negative outcomes across a variety of endpoints (Heuer and Grosell, 2014; Esbaugh, 2018), including impaired sensory systems (Munday et al., 2009b; Nilsson et al., 2012; Williams et al., 2019), alterations in aerobic scope (Munday et al., 2009a; Couturier et al., 2013), ionoregulatory physiology (Esbaugh et al., 2012; Heuer et al., 2012; Strobel et al., 2012), increased tissue damage (Frommel et al., 2012; Chambers et al., 2014; Frommel et al., 2014), and diminished growth and survival (Baumann et al., 2012). Recent studies have concluded that the negative effects of OA may be species specific, with a large number of articles measuring a variety of endpoints reporting no effects on fish (Munday et al., 2016; Lonthair et al., 2017; Baumann, 2019). The cumulative effects of OA are thought to stem from a compensated respiratory acidosis that results from the altered pCO2 gradients between the water and blood. In fact, previous studies on have demonstrated elevated plasma HCO3− and pCO2 coincident with normal pH in response to exposure to OA relevant pCO2 levels in marine fish species (Esbaugh et al., 2012; Heuer et al., 2012; Strobel et al., 2012; Esbaugh et al., 2016). The metabolic compensation pathways in teleosts generally involve apical excretion of H+ and basolateral re-uptake of HCO3− at the gills (Hwang et al., 2011). More specifically, the process occurs in a single branchial cell type called the ionocyte, whereby apical H+ transport occurs through Na+/H+ exchangers (NHE2 and NHE3) and/or the V-type H+ ATPase (VHA) (Marshall and Grosell, 2006). The basolateral re-uptake of HCO3− is thought to occur mainly through the electrogenic Na+/HCO3− cotransporter (NBC), which operates in the efflux direction due to the favourable electrochemical gradients (Hirata et al., 2003). It is important to note that other transporters play a role in acid-secretion mechanisms in other saltwater-acclimated and estuarine species (Takei et al., 2014; Liu et al., 2016). One major concern of exposure to elevated pCO2 is increased metabolic cost and potential metabolic reallocation that results from the increased transport of acid–base equivalents in larval teleosts. Previous work often uses additional metabolic costs as an explanation for physiological responses to OA (Stumpp et al., 2011, 2012). Furthermore, recent work on sea urchin larvae exposed to elevated pCO2 did not exhibit alterations in size, metabolic rate, biochemical content, and gene expression but did exhibit a metabolic reallocation, specifically in protein synthesis and ion transport (Pan et al., 2015). This alteration in energy allocation is critical for understanding the significance of sublethal stress, because individuals with maintenance costs less sensitive to environmental stressors are more likely to survive (Pan et al., 2015). Recent studies on OA have moved beyond defining the detrimental effects of elevated pCO2 on marine life to exploring the potential for resilience in marine systems. This is especially important for economic resources such as fish, and OA is predicted to have severe effects on fish populations (see reviews Hofmann and Todgham, 2010; Munday et al., 2010; Kelly and Hofmann, 2013; Pfister et al., 2014). The resilience of marine fish species to the long-term environmental degradation caused by OA is dependent on a number of factors. Evolutionary processes such as the rise in novel beneficial mutations that can accumulate in a population through natural selection are unlikely to facilitate resilience in species with generation times of months to years, although recent work has shown that rapid evolution may be possible (Torda et al., 2017; Ryu et al., 2018; Catullo et al., 2019). Thus, resilience to OA and climate change may depend on the presence of existing tolerant genotypes in a population, and the ability of individuals to alter their physiology to suit new environmental conditions; a process known as phenotypic plasticity (Bell, 2013; Gonzalez et al., 2013; Pespeni et al., 2013). Previous studies have demonstrated resilient responses to OA among a variety of marine teleosts from across a variety of ecosystems, including species from the Antarctic, estuaries, and coastal upwelling (Davis et al., 2016; Munday et al., 2016; Lonthair et al., 2017; Baumann, 2019). These studies have found that various endpoints are not altered by elevated exposure to OA, including survival, larval morphometrics, starvation rate, heart rate, enzymatic activity, and behaviour (Davis et al., 2016; Munday et al., 2016; Allmon and Esbaugh, 2017; Lonthair et al., 2017; Baumann et al., 2018). On this background, the current study sought to examine whether a sub-tropical species which migrates into an estuarine environment exhibits resilience to OA. We first experimentally assessed the sensitivity of a fast-developing economically important teleost species, the orange-spotted grouper (Epinephelus coioides). Orange-spotted grouper, also known as the estuary cod, are native to the Indo-Pacific and are critical species for aquaculture in the Asia-Pacific region (Zhou et al., 2011). We chose orange-spotted grouper because of their dependency on the tropical estuarine environment (Sheaves, 1993). We then tested whether this species exhibits plasticity of the acid–base regulatory machinery by measuring the gene expression of key exchangers in the mechanism. We hypothesized that orange-spotted grouper, owing to their estuarine-dependent life history, will be tolerant to the impacts of elevated pCO2 across a range of physiological, sublethal, and lethal endpoints. Methods Lethal and sublethal impacts All embryos were produced via common strip-spawning methods by an unknown number of captive orange-spotted grouper broodstock at the Academia Sinica, Institute of Cellular and Organismic Biology (ICOB) Marine Research Station (Jiaoxi, Taiwan), and transported to the Academia Sinica, ICOB (Taipei City, Taiwan), where all tests were initiated within 12 hpf, with hatching occurring within 24 hpf. For both development and pCO2 exposure experiments, analyses were completed on a minimum of two different spawning events with different parental pairings, although broodstock size is unknown. Seawater was filter sterilized using a Millipore ExpressPlus 0.22-µm filter, and salinity was corrected with deionized water to eliminate the potential bacterial growth during testing. pCO2 levels were achieved via methods outlined in Chapter 2, page 44, and Chapter 4, page 83, of Riebesell et al. (2010). Previous work in our laboratory on larval species has shown that bubbling CO2 can cause high mortality, so the method of adding HCl acid and HCO3 was chosen for these experiments, even though adding acid and bicarbonate does have its limitations we felt that the pros outweighed the cons (Lonthair et al., 2017). Both development and pCO2 exposure tests were completed in 24–1-l vacuum-sealed containers. Furthermore, animal care and experimentation were completed in accordance with IACUC-approved protocols through the University of Texas at Austin. Water quality analysis [temperature, salinity, pHNBS, and titratable alkalinity (TA)] was completed on the sterilized seawater to determine the necessary amount of hydrochloric acid and sodium bicarbonate needed to reach the desired pCO2 level. Sterilized seawater is the same as the control water used in the associated experiments, with no additives. The three pCO2 exposures included a control treatment (650 μatm), a medium pCO2 level (1500 μatm), and a predicted high coastal-upwelling pCO2 level (3100 μatm) (Cai et al., 2011; Lonthair et al., 2017). A final water sample was collected and analysed at the conclusion of the test to calculate the pCO2 using the CO2SYS software package developed by Lewis and Wallace (1998) (Table 1). Calculation preferences that were used in the software package include: CO2 constant—K1, K2 from Mehrbach et al. (1973)refit by Dickson and Millero (1987); KHSO4—Dickson (1990); pH scale—NBS scale (mol/kg H2O); total Boron—Uppstrom (1974); and air-sea flux—Wanninkhof (2014). Temperature and salinity were measured using a standard thermometer and refractometer. pH was measured with a combination of pH electrode, calibrated immediately before use (NBS scale), attached to an Orion Star A121 pH meter (Thermo Scientific). Importantly, pHNBS measurements are known to be uncertain up to 0.05 pHNBS units for seawater measurement, which could lead to uncertainty in estimated pCO2 (Riebesell et al., 2010). However, work has shown that, with careful use, these issues are minimized and that pHNBS is an appropriate method in biological CO2 manipulation experiments measuring differences over 100 μatm (Watson et al., 2017). TA was calculated via pH and total CO2, which was measured using a Corning 965 CO2 analyser. Water quality measures were monitored from a subset of replicates at the end of all embryonic incubation experiments, and pCO2 exposures were maintained throughout experiments (Table 1). Table 1. Mean (±1 SEM) temperature, salinity, pH, total alkalinity, and pCO2 from a subset of experiments with orange-spotted grouper (Epinephelus coioides) embryos and early life stages Experiment Treatment Temp. (°C) Salinity pHNBS TA (μmol/kg) pCO2 (μatm) Survival and morphological experiments Control (n = 7) 26.8  ±  0.1 33.2 ± 0.2 8.04 ± 0.01 2498 ± 16 656 ± 18 Medium pCO2 (n = 7) 27.0 ± 0.1 33.5 ± 0.3 7.72 ± 0.02 2519 ± 85 1545 ± 84 High pCO2 (n = 7) 27.0 ± 0.1 33.2 ± 0.3 7.43 ± 0.01 2498 ± 23 3143 ± 101 qPCR—24 h exposure Control (n = 8) 26.8 ± 0.2 33.7 ± 0.3 8.05 ± 0.01 2536 ± 21 692 ± 26 Medium pCO2 (n = 8) 26.9 ± 0.1 34 ± 0.4 7.71 ± 0.03 2501 ± 28 1778 ± 129 High pCO2 (n = 8) 27.0 ± 0.2 34 ± 0.3 7.39 ± 0.01 2520 ± 23 3509 ± 101 qPCR—48 h exposure Control (n = 8) 26.9 ± 0.2 33.7 ± 0.2 8.04 ± 0.02 2463 ± 13 613 ± 31 Medium pCO2 (n = 8) 27.0 ± 0 33 ± 0.2 7.74 ± 0.04 2482 ± 16 1239 ± 120 High pCO2 (n = 8) 27.1 ± 0.1 32.4 ± 0.3 7.46 ± 0.02 2426 ± 22 2982 ± 105 Experiment Treatment Temp. (°C) Salinity pHNBS TA (μmol/kg) pCO2 (μatm) Survival and morphological experiments Control (n = 7) 26.8  ±  0.1 33.2 ± 0.2 8.04 ± 0.01 2498 ± 16 656 ± 18 Medium pCO2 (n = 7) 27.0 ± 0.1 33.5 ± 0.3 7.72 ± 0.02 2519 ± 85 1545 ± 84 High pCO2 (n = 7) 27.0 ± 0.1 33.2 ± 0.3 7.43 ± 0.01 2498 ± 23 3143 ± 101 qPCR—24 h exposure Control (n = 8) 26.8 ± 0.2 33.7 ± 0.3 8.05 ± 0.01 2536 ± 21 692 ± 26 Medium pCO2 (n = 8) 26.9 ± 0.1 34 ± 0.4 7.71 ± 0.03 2501 ± 28 1778 ± 129 High pCO2 (n = 8) 27.0 ± 0.2 34 ± 0.3 7.39 ± 0.01 2520 ± 23 3509 ± 101 qPCR—48 h exposure Control (n = 8) 26.9 ± 0.2 33.7 ± 0.2 8.04 ± 0.02 2463 ± 13 613 ± 31 Medium pCO2 (n = 8) 27.0 ± 0 33 ± 0.2 7.74 ± 0.04 2482 ± 16 1239 ± 120 High pCO2 (n = 8) 27.1 ± 0.1 32.4 ± 0.3 7.46 ± 0.02 2426 ± 22 2982 ± 105 Open in new tab Table 1. Mean (±1 SEM) temperature, salinity, pH, total alkalinity, and pCO2 from a subset of experiments with orange-spotted grouper (Epinephelus coioides) embryos and early life stages Experiment Treatment Temp. (°C) Salinity pHNBS TA (μmol/kg) pCO2 (μatm) Survival and morphological experiments Control (n = 7) 26.8  ±  0.1 33.2 ± 0.2 8.04 ± 0.01 2498 ± 16 656 ± 18 Medium pCO2 (n = 7) 27.0 ± 0.1 33.5 ± 0.3 7.72 ± 0.02 2519 ± 85 1545 ± 84 High pCO2 (n = 7) 27.0 ± 0.1 33.2 ± 0.3 7.43 ± 0.01 2498 ± 23 3143 ± 101 qPCR—24 h exposure Control (n = 8) 26.8 ± 0.2 33.7 ± 0.3 8.05 ± 0.01 2536 ± 21 692 ± 26 Medium pCO2 (n = 8) 26.9 ± 0.1 34 ± 0.4 7.71 ± 0.03 2501 ± 28 1778 ± 129 High pCO2 (n = 8) 27.0 ± 0.2 34 ± 0.3 7.39 ± 0.01 2520 ± 23 3509 ± 101 qPCR—48 h exposure Control (n = 8) 26.9 ± 0.2 33.7 ± 0.2 8.04 ± 0.02 2463 ± 13 613 ± 31 Medium pCO2 (n = 8) 27.0 ± 0 33 ± 0.2 7.74 ± 0.04 2482 ± 16 1239 ± 120 High pCO2 (n = 8) 27.1 ± 0.1 32.4 ± 0.3 7.46 ± 0.02 2426 ± 22 2982 ± 105 Experiment Treatment Temp. (°C) Salinity pHNBS TA (μmol/kg) pCO2 (μatm) Survival and morphological experiments Control (n = 7) 26.8  ±  0.1 33.2 ± 0.2 8.04 ± 0.01 2498 ± 16 656 ± 18 Medium pCO2 (n = 7) 27.0 ± 0.1 33.5 ± 0.3 7.72 ± 0.02 2519 ± 85 1545 ± 84 High pCO2 (n = 7) 27.0 ± 0.1 33.2 ± 0.3 7.43 ± 0.01 2498 ± 23 3143 ± 101 qPCR—24 h exposure Control (n = 8) 26.8 ± 0.2 33.7 ± 0.3 8.05 ± 0.01 2536 ± 21 692 ± 26 Medium pCO2 (n = 8) 26.9 ± 0.1 34 ± 0.4 7.71 ± 0.03 2501 ± 28 1778 ± 129 High pCO2 (n = 8) 27.0 ± 0.2 34 ± 0.3 7.39 ± 0.01 2520 ± 23 3509 ± 101 qPCR—48 h exposure Control (n = 8) 26.9 ± 0.2 33.7 ± 0.2 8.04 ± 0.02 2463 ± 13 613 ± 31 Medium pCO2 (n = 8) 27.0 ± 0 33 ± 0.2 7.74 ± 0.04 2482 ± 16 1239 ± 120 High pCO2 (n = 8) 27.1 ± 0.1 32.4 ± 0.3 7.46 ± 0.02 2426 ± 22 2982 ± 105 Open in new tab For survival assays, 20 embryos were incubated in a 1-l vacuum-sealed container, with four replicates at each pCO2 treatment per spawn (minimum of two spawns). Survival was assessed at 60 hpf. This time point was chosen because the closed containers lack available food sources, and thus, survival measurements after the transition to exogenous feeding, which occurs at ∼72 hpf, would be the result of starvation. At the end of 48 h of exposure, unhatched and dead larval fish were removed and surviving larvae were anaesthetized using a buffered MS-222 solution (250 mg l−1) and counted. In some cases, we observed a hatch rate of zero in a control replicate; this necessitated us to remove the corresponding CO2 treatments from data analysis. A hatch rate of 0 in the control replicate would indicate that embryos were of poor quality and if included may create a bias and impact the results of the pCO2 treatments. A second series of morphometric analyses were performed on the animals from the survival assay to assess total length, yolk consumption, and heart rate at 48 hpf. Each treatment consisted of four experimental replicates and three larvae were sampled per replicate. Heart rate was analysed as described by Incardona et al. (2014). With only one beaker at a time being measured, so that all animals were anaesthetized for the same period of time. For all images, two to three larvae were mounted on 3% methylcellulose in sea water, which allowed for rapid processing of images and videos. Replicates were processed in random order, with controls being analysed throughout the imaging process. Videos and images were randomly numbered by a third-party laboratory member to remove any potential bias during analysis. Heart rate was manually determined from video, while images were analysed for total length and relative yolk sac area using the ImageJ free software programme. Figure 1 illustrates measurement methods of total length and yolk sac area. Figure 1. Open in new tabDownload slide Inverted microscope image of 60 hpf orange-spotted grouper (Epinephelus coioides). Overlaying illustration details methods used in ImageJ to determine total length (mm) and yolk sac area (mm2) at varying pCO2 conditions. Figure 1. Open in new tabDownload slide Inverted microscope image of 60 hpf orange-spotted grouper (Epinephelus coioides). Overlaying illustration details methods used in ImageJ to determine total length (mm) and yolk sac area (mm2) at varying pCO2 conditions. Acid–base regulatory plasticity Real-time polymerase chain reaction (PCR) primers were developed for ef1α, nbc, vha (B subunit), and nhe3. Full-length sequences for ef1α, nbc, vha, and nhe3 were identified from an in-house gill transcriptome. The identified sequences were then verified against the NCBI database using a standard Blast search. Primer pairs were identified using Primer3Plus software package. All primers and GenBank accession numbers for related sequences are found in Table 2. Table 2. List of primers used for real-time PCR Gene Accession # Orientation Sequence ef1-a #KU885470.1 L CTTCAACATCAAGAACGTGTCC R CATTAATCTGACCAGGGTGGTT nhe3 #MN511303 L TATCATGGTGTTTGGAGAGTCG R ATTAATTTTGGGTCCTCCCAGT nbc #MN511305 L TGAACGACATTTCTGACAAACC R CCGAGCAAGATGAATAAAAACC vha #MN511304 L CTAAGAAGACGGCCTGTGAGTT R CTGGATCATCTCCTCTGGGTAG Gene Accession # Orientation Sequence ef1-a #KU885470.1 L CTTCAACATCAAGAACGTGTCC R CATTAATCTGACCAGGGTGGTT nhe3 #MN511303 L TATCATGGTGTTTGGAGAGTCG R ATTAATTTTGGGTCCTCCCAGT nbc #MN511305 L TGAACGACATTTCTGACAAACC R CCGAGCAAGATGAATAAAAACC vha #MN511304 L CTAAGAAGACGGCCTGTGAGTT R CTGGATCATCTCCTCTGGGTAG All sequences are 5′–3′, and reverse primers are reverse compliments of the genetic sequence. Open in new tab Table 2. List of primers used for real-time PCR Gene Accession # Orientation Sequence ef1-a #KU885470.1 L CTTCAACATCAAGAACGTGTCC R CATTAATCTGACCAGGGTGGTT nhe3 #MN511303 L TATCATGGTGTTTGGAGAGTCG R ATTAATTTTGGGTCCTCCCAGT nbc #MN511305 L TGAACGACATTTCTGACAAACC R CCGAGCAAGATGAATAAAAACC vha #MN511304 L CTAAGAAGACGGCCTGTGAGTT R CTGGATCATCTCCTCTGGGTAG Gene Accession # Orientation Sequence ef1-a #KU885470.1 L CTTCAACATCAAGAACGTGTCC R CATTAATCTGACCAGGGTGGTT nhe3 #MN511303 L TATCATGGTGTTTGGAGAGTCG R ATTAATTTTGGGTCCTCCCAGT nbc #MN511305 L TGAACGACATTTCTGACAAACC R CCGAGCAAGATGAATAAAAACC vha #MN511304 L CTAAGAAGACGGCCTGTGAGTT R CTGGATCATCTCCTCTGGGTAG All sequences are 5′–3′, and reverse primers are reverse compliments of the genetic sequence. Open in new tab Embryo exposures were performed as described above. A total of 20 embryos were incubated in a 1-l vacuum-sealed container, with four replicates at each pCO2 treatment per spawn (minimum of two spawns). All surviving larvae were collected at 12, 36, and 60 hpf under both control conditions and after exposure to elevated pCO2 levels. Samples were flash frozen and stored in −80°C, until further processing was required. Total RNA isolation was performed using QIAzol (Qiagen) according to manufacturer’s protocols and quantified using an ND-1000 spectrophotometer (Thermo Scientific). Total RNA was treated for potential DNA contamination by incubating with DNase 1 (Roche), according to manufacturer’s protocols. complementary DNA (cDNA) synthesis was performed on 1 μg of total RNA using SuperScript IV reverse transcriptase (Invitrogen Live Technologies), according to manufacturer’s protocols. For all cDNA synthesis runs, no reverse transcriptase controls were performed to test for genomic DNA contamination. Samples were diluted tenfold using nuclease free water and stored at −20°C until quantitative polymerase chain reaction (qPCR) analysis. qPCR analysis was performed using the Maxima SYBR Green kit (Thermo Scientific). Reactions were prepared according to the manufacturer’s protocols with the exception that a 12.5-μl total reaction volume was used. All reactions were processed using an MX3000P qPCR machine (Stratagene) with accompanying software. A serial dilution was used for standard curves to determine the reaction efficiency of each primer pair. PCR efficiencies ranged from 70% to 97.4% with an R2 ≥ 0.97. For all genes, negative and no reverse transcriptase control reactions were performed. The CT values for each sample were used to assess the relative abundance of each gene in relation to the control gene ef1α using the delta–delta CT method. Immunofluorescence methods Samples were collected following the survival experiments described above and fixed overnight in Z-fix at 4°C. Samples were then washed one time with 100% methanol and then transferred to 100% methanol and stored at −20°C; this procedure constituted chill permeabilization of the sample. Samples were then washed with phosphate buffered saline (PBS) 1% triton X (PBST) four times for 5 min followed by 1 h in blocking buffer (PBST with 5% foetal calf serum) at room temperature. Blocking buffer was removed, and samples were then incubated with primary antibodies for Na+/K+ ATPase (NKA) (in 1:100 dilution in blocking buffer) at 4°C overnight on a rocker. The primary polyclonal rabbit antibody for NKA (sc-28800) was obtained from the Santa Cruz Biotechnology, and its effectiveness was verified using a western blot (Allmon and Esbaugh, 2017). Following primary incubation, samples were washed in blocking buffer three times for 5 min then incubated with secondary antibodies—goat anti-rabbit Alexa Flour 488 (1:500)—in the dark for 6 h at 4°C on a rocker. Samples were then washed with blocking buffer four times for 5 min and mounted using Vectashield hard-mount with DAPI and stored in the dark at 4°C until imaged. Imaging was completed using a Nikon C2+ confocal microscope system with a Nikon Eclipse Ti-E inverted microscope and utilizing NIS-Element imaging software for image acquisition, processing, and analysis. Images were randomly numbered to remove any potential bias during analysis. Ionocyte density was determined by creating a 0.0625-mm2 box over the yolk sac area and manually counting the number of ionocytes using the ImageJ free software programme. Figure 2 illustrates the immunofluorescence ionocyte density count methods. Figure 2. Open in new tabDownload slide Confocal microscopy image of ionocyte density using an antibody for NKA in 60 hpf orange-spotted grouper (Epinephelus coioides). The 0.0625 mm2 illustrates methods used to complete counts of number of ionocytes at varying pCO2 conditions. Figure 2. Open in new tabDownload slide Confocal microscopy image of ionocyte density using an antibody for NKA in 60 hpf orange-spotted grouper (Epinephelus coioides). The 0.0625 mm2 illustrates methods used to complete counts of number of ionocytes at varying pCO2 conditions. Statistical methods Survival and total length data across pCO2 exposures passed the Shapiro–Wilk normality test and was assessed using a one-way analysis of variance (ANOVA). Heart rate across pCO2 exposures was assessed via a Kruskal–Wallis one-way ANOVA on ranks due to failure of the Shapiro–Wilk normality test. Gene expression data across development passed the Shapiro–Wilk normality test after natural log transformation and was assessed using a one-way ANOVA and Holm–Sidak post hoc test against control values. Gene expression data across pCO2 exposures passed the Shapiro–Wilk normality test after natural log transformation and was assessed using a one-way ANOVA. Ionocyte density across pCO2 exposures passed the Shapiro–Wilk normality test after natural log transformation and assessed using a one-way ANOVA. All statistical measurements were completed using SigmaPlot (Systat Software, San Jose, CA, USA). Results Sensitivity experiments Lethal and sublethal impacts Survival was not impacted by 48 h exposure to increased pCO2 levels (Figure 3). Heart rate was significantly elevated by 48 h exposure to elevated pCO2 levels at both 1500 and 3100 μatm (p ≤ 0.05; Kruskal–Wallis ANOVA on ranks) (Figure 4a). The control heart rate in mean ± standard error of mean (SEM) was 134 ± 14 beats per minute (bpm) at 600 μatm, the 1500 μatm heart rate was 174 ± 10 bpm with, while the 3100 μatm heart rate was 190 ± 7 bpm, an increase of 40 bpm (30% increase) and 56 bpm (42% increase), respectively. No effects were observed in response to 48 h exposure to elevated pCO2 levels in both total length (Figure 4b) and yolk size (Figure 4c). Figure 3. Open in new tabDownload slide Mean (±SEM) survival of orange-spotted grouper (Epinephelus coioides) after 48 h exposure to control, 1500 μatm, and 3100 μatm pCO2. There were no significant differences between groups (ANOVA; N = 7 per treatment). Figure 3. Open in new tabDownload slide Mean (±SEM) survival of orange-spotted grouper (Epinephelus coioides) after 48 h exposure to control, 1500 μatm, and 3100 μatm pCO2. There were no significant differences between groups (ANOVA; N = 7 per treatment). Figure 4. Open in new tabDownload slide Mean (±SEM) (a) heart rate, (b) total length, and (c) yolk area of orange-spotted grouper (Epinephelus coioides) after 48 h exposure to control, 1500 μatm, and 3100 μatm pCO2. Asterisk indicated statistically significant difference from control (ANOVA; p < 0.05; N = 19–21 larvae per treatment). Figure 4. Open in new tabDownload slide Mean (±SEM) (a) heart rate, (b) total length, and (c) yolk area of orange-spotted grouper (Epinephelus coioides) after 48 h exposure to control, 1500 μatm, and 3100 μatm pCO2. Asterisk indicated statistically significant difference from control (ANOVA; p < 0.05; N = 19–21 larvae per treatment). Acid–base regulatory pathway plasticity Gene expression for nbc and vha exhibited significant up-regulation as a result of development at 60 h post-fertilization when compared with 12 h post-fertilization (p ≤ 0.05; ANOVA) (Figure 5a and c). In contrast, nhe3 exhibited no alterations in relative messenger RNA (mRNA gene expression as development progressed (Figure 5b). Interestingly, at two time points, 24 and 48 h, elevated pCO2 exposure at both 1500 and 3100 μatm had no effect on a variety of the transporters that are thought to play a critical role in acid–base regulation in marine teleosts, including nbc, nhe3, and vha (Figure 6). Furthermore, ionocyte density was not altered as a result of exposure to increased levels of pCO2 (Figure 7). Figure 5. Open in new tabDownload slide Whole animal gene expression of H+ excretion pathways: (a) nbc, (b) nhe3, and (c) vha, during development in orange-spotted grouper (Epinephelus coioides). All values are relative to housekeeping gene ef1a. Values set relative to control values denoted by dashed lines at 1.0. Significant differences from initial time point (12 h) denoted by an asterisk (ANOVA; p < 0.05; N = 8 per time point). All values are mean ± SEM survival. Figure 5. Open in new tabDownload slide Whole animal gene expression of H+ excretion pathways: (a) nbc, (b) nhe3, and (c) vha, during development in orange-spotted grouper (Epinephelus coioides). All values are relative to housekeeping gene ef1a. Values set relative to control values denoted by dashed lines at 1.0. Significant differences from initial time point (12 h) denoted by an asterisk (ANOVA; p < 0.05; N = 8 per time point). All values are mean ± SEM survival. Figure 6. Open in new tabDownload slide Whole animal gene expression of H+ excretion pathways: (a) nbc, (b) nhe3, and (c) vha, during 24 and 48 h exposure to control, 1500 μatm, and 3100 μatm in orange-spotted grouper (Epinephelus coioides). All values are relative to housekeeping gene ef1a. Values set relative to control values denoted by dashed lines at 1.0. There were no significant differences between groups (ANOVA; N = 8 per treatment). All values are mean ± SEM survival. Figure 6. Open in new tabDownload slide Whole animal gene expression of H+ excretion pathways: (a) nbc, (b) nhe3, and (c) vha, during 24 and 48 h exposure to control, 1500 μatm, and 3100 μatm in orange-spotted grouper (Epinephelus coioides). All values are relative to housekeeping gene ef1a. Values set relative to control values denoted by dashed lines at 1.0. There were no significant differences between groups (ANOVA; N = 8 per treatment). All values are mean ± SEM survival. Figure 7. Open in new tabDownload slide Natural log of the mean ionocyte density per 0.0625 mm2 area (±1 SEM) after 48 h exposure to control, 1500 μatm, and 3100 μatm pCO2 in orange-spotted grouper (Epinephelus coioides). There were no significant differences between pCO2 (ANOVA; N = 8–12 per treatment). Figure 7. Open in new tabDownload slide Natural log of the mean ionocyte density per 0.0625 mm2 area (±1 SEM) after 48 h exposure to control, 1500 μatm, and 3100 μatm pCO2 in orange-spotted grouper (Epinephelus coioides). There were no significant differences between pCO2 (ANOVA; N = 8–12 per treatment). Discussion Determining species resilience is a crucial aspect of understanding the impacts that OA and other climate change-induced environmental changes will have on future marine ecosystems, specifically within the discussion of evolutionary rescue. Evolutionary rescue is described as when genetic adaptation allows for a population to recover from deteriorating population effects, which were initiated by environmental change and would have otherwise caused a local extinction (Gonzalez et al., 2013). An important facet of evolutionary rescue is that a subset of individuals are resilient to the changing environment and have the appropriate phenotypic solutions. Thus, identifying species and ecosystems with tolerant traits that can defend against the physiological stresses of environmental degradation is critical. Here, we provide evidence that a fast growing and economically important estuarine-dependent species, the orange-spotted grouper, exhibits no clear detrimental effects of OA when exposed during the most sensitive early life stages. These findings are consistent with the hypothesis that fish endemic to coastal and upwelling regions that commonly experience elevated CO2 may have a level of intrinsic resilience (Baumann, 2019) and corroborates a case study on highly CO2-sensitive offshore fish species (Murray et al., 2019). It should be noted that the influence of parental CO2 exposure on the CO2 tolerance observed here in orange-spotted grouper is unknown as brood stock rearing conditions were not monitored. Prior work has shown that high CO2 conditions may contribute to CO2 tolerance in embryos in some species (Snyder et al., 2018). Early work on the impacts of OA on early life stage fish has emphasized survival as a critical endpoint, because larval survival and recruitment represents a crucial population bottleneck. Baumann et al. (2012) is a foundational study in the field, which found a 73% reduction in the survival of Menidia beryllina at elevated pCO2. More recent studies have found that other species exhibit reduced survival but none with such extreme sensitivity as M. beryllina, which likely indicates that M. beryllina is an outlier (Miller et al., 2012; Chambers et al., 2014; Lonthair et al., 2017). Conversely, other studies have shown that a number of species exhibit tolerance to high pCO2 (Esbaugh, 2018; Baumann, 2019) Furthermore, while a significant portion of the populations is unable to tolerate OA, there are tolerant individuals present in all species tested (Esbaugh, 2018). While larval survival is easy to interpret and has clear population level outcomes, it is also important to consider sublethal effects of OA that may indicate a poor prognosis for fish in later life. Following on our hypothesis that OA may place an additional energetic burden on endogenous feeding life stages, we sought to use two morphological traits that may indicate changing energy burdens: size at age and yolk sac area. Interestingly, there was no effect of elevated CO2 on either endpoint, which argues against our hypothesis and supports the premise that this species may have a level of intrinsic resilience. Previous work has highlighted that the impacts of OA on embryonic and larval growth can be variable depending on the model species. Some studies have found species exhibiting detrimental decreases in growth (Miller et al., 2012; Frommel et al., 2016), no effects on growth (Bignami et al., 2013; Frommel et al., 2013), and even increases in growth and size at age (Bignami et al., 2014; Chambers et al., 2014). This indicates that energy utilization may vary considerably between species when exposed to elevated pCO2, with some species allocating a larger portion of energy to maintaining growth. Plasticity in the acid–base regulatory machinery is a critical metric to understand resilience of OA in a teleost species. Our study indicates that orange-spotted grouper maintains acid–base regulatory transporters in a high enough abundance to correct for OA without a transcriptional change. We saw no alterations in gene expression in any of the acid–base regulatory transporters that we measured, including: nhe3, vha, and nbc. Other studies have found similar results via quick stabilization of net whole body titratable acid flux following pCO2 exposure (Edwards et al., 2005; Allmon and Esbaugh, 2017) and via the lack of plasticity in acid–base transport gene expression, enzyme activity, and protein abundance (Esbaugh et al., 2012; Michael et al., 2016; Allmon and Esbaugh, 2017). Furthermore, when measuring ionocyte density as a result of exposure to elevated pCO2, we found no significant changes. This further supports the argument that the orange-spotted grouper maintain high enough levels of acid–base regulatory machinery at even the earliest life stages. In contrast to the previously discussed data, the evidence provided from heart rate is consistent with the hypothesis that OA will result in elevated energetic costs of survival in orange-spotted grouper. Our data demonstrate that heart rate increases by ∼40% when exposed to elevated CO2 (3100 μatm). This trend is consistent with prior work (Lonthair et al. 2017); however, the magnitude of the response in grouper is much greater than previously described for other species. The significance of this finding is rooted in two physiological concepts. The first is that heart rate is an effective proxy for metabolic rate, whereby higher heart rates indicate greater energy utilization (Green, 2011). The second is that larval fishes generally have limited metabolic scope (Killen et al., 2007), which is the difference between the baseline costs of living and the maximum capacity of the system. This would suggest that the energetic costs of increased heart rate will remove available energy from other functions, such as future growth or activity. The significance of such metabolic reallocation is highlighted in Pan et al. (2015), which showed that sea urchin larvae expend increased energy on protein synthesis and ion transport under OA conditions, despite no evidence of OA induced effects on size or gene expression (Pan et al., 2015). Our data may suggest that grouper are undergoing similar metabolic reallocations to maintain the growth rate in the early life stage despite higher metabolic rates. While this would presumably aid the fish in reducing early life predation risk, the long-term cost of such metabolic reallocation is unknown. It is also important to consider the physiologically advantages of elevated heart rate to early life stage orange-spotted grouper. Adult teleosts control convective fluid movement across and through their respiratory epithelium in response to respiratory stress, such as reduced oxygen or elevated CO2 [see review in Gilmour and Perry (2007)]. In fact, OA relevant pCO2 exposures resulted in significant elevation in ventilatory parameters, which significantly reduces the magnitude of the metabolic compensation response in juvenile life stages (Ern and Esbaugh, 2016). Yet as such a benefit would seem unlikely in the current study given that early life fishes use cutaneous gas exchange with little role for convective fluid movement (Rombough, 2002; Rombough, 2007; Fu et al., 2010). It instead seems likely that the increased heart rate is a by-product of the developing sensory system related to cardiorespiratory control (Vulesevic and Perry, 2006; Miller et al., 2014). In conclusion, our study has shown that the estuarine-dependent orange-spotted grouper show no clear detrimental effects of OA exposure in the most sensitive life stage. This is consistent with the hypothesis that species that have evolved in habitats with natural fluctuations in pCO2 may have intrinsic resilience to the impact of OA. While these results are encouraging for the long-term prospects of orange-spotted grouper, it is important to recognize that our conclusions are limited to the early life stages as we were unable to complete extended grow-out studies due to the severe drop in survival that occurs at first feeding. Furthermore, we cannot inform on the potential behavioural effects of OA that may occur in later life stages, nor the potential implication of elevated temperature and reduced oxygen as additive stressors, both of which can exacerbate concerns regarding the baseline energetic cost of living in the future oceans. Funding This work was funded by National Science Foundation grants (EF 1315290 to AJE and OISE 1614168 to JL). Additional support for JL was provided by the Coastal Conservation Association (CCA) Texas, University of Texas at Austin: Summer Recruitment Fellowship, and the University of Texas at Austin Marine Science Institute—Lund Endowment. Orange-spotted grouper embryos were generously provided by the Institute of Cellular and Organismic Biology (ICOB) Marine Research Station (Jiaoxi, Taiwan). Elizabeth Allmon, Lee I-Chun, Kuan Bao-Long, and Amy Chow assisted in embryo collection, imaging, and sample collection with embryonic and early life stage work. References Allmon E. B. , Esbaugh A. J. 2017 . Carbon dioxide induced plasticity of branchial acid–base pathways in an estuarine teleost . Scientific Reports , 7 : 10 . Google Scholar Crossref Search ADS PubMed WorldCat Baumann H. 2019 . Experimental assessments of marine species sensitivities to ocean acidification and co-stressors: how far have we come? Canadian Journal of Zoology , 97 : 399 – 408 . Google Scholar Crossref Search ADS WorldCat Baumann H. , Parks E. M. , Murray C. S. 2018 . Starvation rates in larval and juvenile Atlantic silversides (Menidia menidia) are unaffected by high CO2 conditions . Marine Biology , 165 : 9 . Google Scholar Crossref Search ADS WorldCat Baumann H. , Talmage S. C. , Gobler C. J. 2012 . Reduced early life growth and survival in a fish in direct response to increased carbon dioxide . Nature Climate Change , 2 : 38 – 41 . Google Scholar Crossref Search ADS WorldCat Bell G. 2013 . Evolutionary rescue and the limits of adaptation . Philosophical Transactions of the Royal Society B: Biological Sciences , 368 : 6 . Google Scholar Crossref Search ADS WorldCat Bignami S. , Sponaugle S. , Cowen R. K. 2013 . Response to ocean acidification in larvae of a large tropical marine fish, Rachycentron canadum . Global Change Biology , 19 : 996 – 1006 . Google Scholar Crossref Search ADS PubMed WorldCat Bignami S. , Sponaugle S. , Cowen R. K. 2014 . Effects of ocean acidification on the larvae of a high-value pelagic fisheries species, Mahi-mahi Coryphaena hippurus . Aquatic Biology , 21 : 249 – 260 . Google Scholar Crossref Search ADS WorldCat Cai W.-J. , Hu X. , Huang W.-J. , Murrell M. C. , Lehrter J. C. , Lohrenz S. E. , Chou W.-C. et al. . 2011 . Acidification of subsurface coastal waters enhanced by eutrophication . Nature Geoscience , 4 : 766 – 770 . Google Scholar Crossref Search ADS WorldCat Caldeira K. , Wickett M. E. 2003 . Anthropogenic carbon and ocean pH . Nature , 425 : 365 . Google Scholar Crossref Search ADS PubMed WorldCat Catullo R. A. , Llewelyn J. , Phillips B. , Moritz C. C. 2019 . The potential for rapid evolution under anthropogenic climate change . Current Biology , 29 : R996 – R1007 . Google Scholar Crossref Search ADS PubMed WorldCat Chambers R. C. , Candelmo A. C. , Habeck E. A. , Poach M. E. , Wieczorek D. , Cooper K. R. , Greenfield C. E. et al. . 2014 . Effects of elevated CO2 in the early life stages of summer flounder, Paralichthys dentatus, and potential consequences of ocean acidification . Biogeosciences , 11 : 1613 – 1626 . Google Scholar Crossref Search ADS WorldCat Couturier C. S. , Stecyk J. A. W. , Rummer J. L. , Munday P. L. , Nilsson G. E. 2013 . Species-specific effects of near-future CO2 on the respiratory performance of two tropical prey fish and their predator . Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology , 166 : 482 – 489 . Google Scholar Crossref Search ADS WorldCat Davis B. E. , Miller N. A. , Flynn E. E. , Todgham A. E. 2016 . Juvenile Antarctic rockcod (Trematomus bernacchii) are physiologically robust to CO2-acidified seawater . Journal of Experimental Biology , 219 : 1203 – 1213 . Google Scholar Crossref Search ADS PubMed WorldCat Dickson A. G. 1990 . Standard potential of the reaction:, and and the standard acidity constant of the ion HSO4− in synthetic sea water from 273.15 to 318.15 K . The Journal of Chemical Thermodynamics , 22 : 113 – 127 . Google Scholar Crossref Search ADS WorldCat Dickson A. G. , Millero F. J. 1987 . A comparison of the equilibrium constants for the dissociation of carbonic acid in seawater media . Deep Sea Research Part A. Oceanographic Research Papers , 34 : 1733 – 1743 . Google Scholar Crossref Search ADS WorldCat Doney S. C. , Fabry V. J. , Feely R. A. , Kleypas J. A. 2009 . Ocean acidification: the other CO2 problem . Annual Review of Marine Science , 1 : 169 – 192 . Google Scholar Crossref Search ADS PubMed WorldCat Edwards S. L. , Wall B. P. , Morrison-Shetlar A. , Sligh S. , Weakley J. C. , Claiborne J. B. 2005 . The effect of environmental hypercapnia and salinity on the expression of NHE-like isoforms in the gills of a euryhaline fish (Fundulus heteroclitus) . Journal of Experimental Zoology Part A: Ecological and Integrative Physiology , 303A : 464 – 475 . WorldCat Ern R. , Esbaugh A. J. 2016 . Hyperventilation and blood acid–base balance in hypercapnia exposed red drum (Sciaenops ocellatus) . Journal of Comparative Physiology B: Biochemical, Systemic, and Environmental Physiology , 186 : 447 – 460 . Google Scholar Crossref Search ADS WorldCat Esbaugh A. J. 2018 . Physiological implications of ocean acidification for marine fish: emerging patterns and new insights . Journal of Comparative Physiology B: Biochemical Systems and Environmental Physiology , 188 : 1 – 13 . Google Scholar Crossref Search ADS WorldCat Esbaugh A. J. , Ern R. , Nordi W. M. , Johnson A. S. 2016 . Respiratory plasticity is insufficient to alleviate blood acid–base disturbances after acclimation to ocean acidification in the estuarine red drum, Sciaenops ocellatus . Journal of Comparative Physiology B: Biochemical Systemic and Environmental Physiology , 186 : 97 – 109 . Google Scholar Crossref Search ADS WorldCat Esbaugh A. J. , Heuer R. , Grosell M. 2012 . Impacts of ocean acidification on respiratory gas exchange and acid–base balance in a marine teleost, Opsanus beta . Journal of Comparative Physiology B , 182 : 921 – 934 . Google Scholar Crossref Search ADS WorldCat Frommel A. Y. , Maneja R. , Lowe D. , Malzahn A. M. , Geffen A. J. , Folkvord A. , Piatkowski U. et al. . 2012 . Severe tissue damage in Atlantic cod larvae under increasing ocean acidification . Nature Climate Change , 2 : 42 – 46 . Google Scholar Crossref Search ADS WorldCat Frommel A. Y. , Maneja R. , Lowe D. , Pascoe C. K. , Geffen A. J. , Folkvord A. , Piatkowski U. et al. . 2014 . Organ damage in Atlantic herring larvae as a result of ocean acidification . Ecological Applications , 24 : 1131 – 1143 . Google Scholar Crossref Search ADS PubMed WorldCat Frommel A. Y. , Margulies D. , Wexler J. B. , Stein M. S. , Scholey V. P. , Williamson J. E. , Bromhead D. et al. . 2016 . Ocean acidification has lethal and sub-lethal effects on larval development of yellowfin tuna, Thunnus albacares . Journal of Experimental Marine Biology and Ecology , 482 : 18 – 24 . Google Scholar Crossref Search ADS WorldCat Frommel A. Y. , Schubert A. , Piatkowski U. , Clemmesen C. 2013 . Egg and early larval stages of Baltic cod, Gadus morhua, are robust to high levels of ocean acidification . Marine Biology , 160 : 1825 – 1834 . Google Scholar Crossref Search ADS WorldCat Fu C. , Wilson J. M. , Rombough P. J. , Brauner C. J. 2010 . Ions first: Na+ uptake shifts from the skin to the gills before O-2 uptake in developing rainbow trout, Oncorhynchus mykiss . Proceedings of the Royal Society B: Biological Sciences , 277 : 1553 – 1560 . Google Scholar Crossref Search ADS WorldCat Gilmour K. M. , Perry S. F. 2007 . Branchial chemoreceptor regulation of cardiorespiratory function. In Sensory Systems Neuroscience , pp. 97 – 151 . Ed. by Hara T. J. , Zielinski B. S. . Elsevier Academic Press Inc ., San Diego . Google Preview WorldCat COPAC Gonzalez A. , Ronce O. , Ferriere R. , Hochberg M. E. 2013 . Evolutionary rescue: an emerging focus at the intersection between ecology and evolution . Philosophical Transactions of the Royal Society B: Biological Sciences , 368 : 8 . WorldCat Green J. A. 2011 . The heart rate method for estimating metabolic rate: review and recommendations . Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology , 158 : 287 – 304 . Google Scholar Crossref Search ADS WorldCat Heuer R. M. , Esbaugh A. J. , Grosell M. 2012 . Ocean acidification leads to counterproductive intestinal base loss in the Gulf toadfish (Opsanus beta) . Physiological and Biochemical Zoology , 85 : 450 – 459 . Google Scholar Crossref Search ADS PubMed WorldCat Heuer R. M. , Grosell M. 2014 . Physiological impacts of elevated carbon dioxide and ocean acidification on fish . American Journal of Physiology-Regulatory, Integrative and Comparative Physiology , 307 : R1061 – R1084 . Google Scholar Crossref Search ADS PubMed WorldCat Hirata T. , Kaneko T. , Ono T. , Nakazato T. , Furukawa N. , Hasegawa S. , Wakabayashi S. et al. . 2003 . Mechanism of acid adaptation of a fish living in a pH 3.5 lake . American Journal of Physiology-Regulatory, Integrative and Comparative Physiology , 284 : R1199 – R1212 . Google Scholar Crossref Search ADS PubMed WorldCat Hofmann G. E. , Todgham A. E. 2010 . Living in the now: physiological mechanisms to tolerate a rapidly changing environment. In Annual Review of Physiology , 72 , pp. 127 – 145 . Annual Reviews , Palo Alto . Google Preview WorldCat COPAC Hwang P. P. , Lee T. H. , Lin L. Y. 2011 . Ion regulation in fish gills: recent progress in the cellular and molecular mechanisms . American Journal of Physiology-Regulatory, Integrative and Comparative Physiology , 301 : R28 – R47 . Google Scholar Crossref Search ADS PubMed WorldCat Incardona J. P. , Gardner L. D. , Linbo T. L. , Brown T. L. , Esbaugh A. J. , Mager E. M. , Stieglitz J. D. et al. . 2014 . Deepwater Horizon crude oil impacts the developing hearts of large predatory pelagic fish . Proceedings of the National Academy of Sciences of the United States of America , 111 : E1510 – E1518 . Google Scholar Crossref Search ADS PubMed WorldCat Kelly M. W. , Hofmann G. E. 2013 . Adaptation and the physiology of ocean acidification . Functional Ecology , 27 : 980 – 990 . Google Scholar Crossref Search ADS WorldCat Killen S. S. , Costa I. , Brown J. A. , Gamperl A. K. 2007 . Little left in the tank: metabolic scaling in marine teleosts and its implications for aerobic scope . Proceedings of the Royal Society B: Biological Sciences , 274 : 431 – 438 . Google Scholar Crossref Search ADS WorldCat Lewis E. , Wallace D. 1998 . in Laboratory C. D. I. A. C.-O. R. N. (Ed). Program Developed for CO2 System Calculations. Liu S. T. , Horng J. L. , Chen P. Y. , Hwang P. P. , Lin L. Y. 2016 . Salt secretion is linked to acid–base regulation of ionocytes in seawater-acclimated medaka: new insights into the salt-secreting mechanism . Scientific Reports , 6 : 13 . Google Scholar Crossref Search ADS PubMed WorldCat Lonthair J. , Ern R. , Esbaugh A. 2017 . The early life stages of an estuarine fish, the red drum (Sciaenops ocellatus), are tolerant to high pCO2 . ICES Journal of Marine Science , 74 : 1042 – 1050 . Google Scholar Crossref Search ADS WorldCat Marshall W. , Grosell M. 2006 . Ion transport, osmoregulation, and acid–base balance. In The Physiology of Fishes , 3rd edn, pp. 177 – 230 . Ed. by Evans D. H. , Claiborne J. B. . CRC Press , Boca Raton, FL. Google Preview WorldCat COPAC McNeil B. I. , Sasse T. P. 2016 . Future ocean hypercapnia driven by anthropogenic amplification of the natural CO2 cycle . Nature , 529 : 383 – 386 . Google Scholar Crossref Search ADS PubMed WorldCat Mehrbach C. , Culberson C. H. , Hawley J. E. , Pytkowicx R. M. 1973 . Measurement of the apparent dissociation constants of carbonic acid in seawater at atmospheric pressure . Limnology and Oceanography , 18 : 897 – 907 . 10.4319/lo.1973.18.6.0897 Google Scholar Crossref Search ADS WorldCat Crossref Michael K. , Kreiss C. M. , Hu M. Y. , Koschnick N. , Bickmeyer U. , Dupont S. , Portner H. O. et al. . 2016 . Adjustments of molecular key components of branchial ion and pH regulation in Atlantic cod (Gadus morhua) in response to ocean acidification and warming . Comparative Biochemistry and Physiology Part B: Biochemistry & Molecular Biology , 193 : 33 – 46 . Google Scholar Crossref Search ADS WorldCat Miller G. M. , Watson S. A. , Donelson J. M. , McCormick M. I. , Munday P. L. 2012 . Parental environment mediates impacts of increased carbon dioxide on a coral reef fish . Nature Climate Change , 2 : 858 – 861 . Google Scholar Crossref Search ADS WorldCat Miller S. , Pollack J. , Bradshaw J. , Kumai Y. , Perry S. F. 2014 . Cardiac responses to hypercapnia in larval zebrafish (Danio rerio): the links between CO2 chemoreception, catecholamines and carbonic anhydrase . Journal of Experimental Biology , 217 : 3569 – 3578 . Google Scholar Crossref Search ADS PubMed WorldCat Munday P. L. , Crawley N. E. , Nilsson G. E. 2009a . Interacting effects of elevated temperature and ocean acidification on the aerobic performance of coral reef fishes . Marine Ecology Progress Series , 388 : 235 – 242 . Google Scholar Crossref Search ADS WorldCat Munday P. L. , Dixson D. L. , Donelson J. M. , Jones G. P. , Pratchett M. S. , Devitsina G. V. , Doving K. B. 2009b . Ocean acidification impairs olfactory discrimination and homing ability of a marine fish . Proceedings of the National Academy of Sciences of the United States of America , 106 : 1848 – 1852 . Google Scholar Crossref Search ADS WorldCat Munday P. L. , Dixson D. L. , McCormick M. I. , Meekan M. , Ferrari M. C. O. , Chivers D. P. 2010 . Replenishment of fish populations is threatened by ocean acidification . Proceedings of the National Academy of Sciences of the United States of America , 107 : 12930 – 12934 . Google Scholar Crossref Search ADS PubMed WorldCat Munday P. L. , Watson S. A. , Parsons D. M. , King A. , Barr N. G. , McLeod I. M. , Allan B. J. M. et al. . 2016 . Effects of elevated CO2 on early life history development of the yellowtail kingfish, Seriola lalandi, a large pelagic fish . ICES Journal of Marine Science , 73 : 641 – 649 . Google Scholar Crossref Search ADS WorldCat Murray C. S. , Wiley D. , Baumann H. 2019 . High sensitivity of a keystone forage fish to elevated CO2 and temperature . Conservation Physiology , 7 : coz084. WorldCat Nilsson G. E. , Dixson D. L. , Domenici P. , McCormick M. I. , Sorensen C. , Watson S. A. , Munday P. L. 2012 . Near-future carbon dioxide levels alter fish behaviour by interfering with neurotransmitter function . Nature Climate Change , 2 : 201 – 204 . Google Scholar Crossref Search ADS WorldCat Orr J. C. , Fabry V. J. , Aumont O. , Bopp L. , Doney S. C. , Feely R. A. , Gnanadesikan A. et al. . 2005 . Anthropogenic ocean acidification over the twenty-first century and its impact on calcifying organisms . Nature , 437 : 681 – 686 . Google Scholar Crossref Search ADS PubMed WorldCat Pan T. C. F. , Applebaum S. L. , Manahan D. T. 2015 . Experimental ocean acidification alters the allocation of metabolic energy . Proceedings of the National Academy of Sciences of the United States of America , 112 : 4696 – 4701 . Google Scholar Crossref Search ADS PubMed WorldCat Pespeni M. H. , Sanford E. , Gaylord B. , Hill T. M. , Hosfelt J. D. , Jaris H. K. , LaVigne M. et al. . 2013 . Evolutionary change during experimental ocean acidification . Proceedings of the National Academy of Sciences of the United States of America , 110 : 6937 – 6942 . Google Scholar Crossref Search ADS PubMed WorldCat Pfister C. A. , Esbaugh A. J. , Frieder C. A. , Baumann H. , Bockmon E. E. , White M. M. , Carter B. R. et al. . 2014 . Detecting the unexpected: a research framework for ocean acidification . Environmental Science & Technology , 48 : 9982 – 9994 . Google Scholar Crossref Search ADS PubMed WorldCat Riebesell U. , Fabry V. J. , Hansson L. , Gattuso J.-P. 2010 . Guide to Best Practices for Ocean Acidification Research and Data Reporting . European Commission , Luxembourg. Google Preview WorldCat COPAC Rombough P. 2002 . Gills are needed for ionoregulation before they are needed for O-2 uptake in developing zebrafish, Danio rerio . Journal of Experimental Biology , 205 : 1787 – 1794 . Google Scholar PubMed WorldCat Rombough P. 2007 . The functional ontogeny of the teleost gill: which comes first, gas or ion exchange? Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology , 148 : 732 – 742 . Google Scholar Crossref Search ADS WorldCat Ryu T. , Veilleux H. D. , Donelson J. M. , Munday P. L. , Ravasi T. 2018 . The epigenetic landscape of transgenerational acclimation to ocean warming . Nature Climate Change , 8 : 504 – 509 . Google Scholar Crossref Search ADS WorldCat Sheaves M. J. 1993 . Patterns of movement of some fishes within an estuary in tropical Australia . Australian Journal of Marine and Freshwater Research , 44 : 867 – 880 . Google Scholar Crossref Search ADS WorldCat Snyder J. T. , Murray C. S. , Baumann H. 2018 . Potential for maternal effects on offspring CO2 sensitivities in the Atlantic silverside (Menidia menidia) . Journal of Experimental Marine Biology and Ecology , 499 : 1 – 8 . Google Scholar Crossref Search ADS WorldCat Solomon S. , Qin D. , Manning M. , Chen Z. , Marquis M. 2007 . Climate Change 2007: The Physical Science Basis: Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change . Cambridge University Press , New York . Google Preview WorldCat COPAC Strobel A. , Bennecke S. , Leo E. , Mintenbeck K. , Portner H. O. , Mark F. C. 2012 . Metabolic shifts in the Antarctic fish Notothenia rossii in response to rising temperature and PCO2 . Frontiers in Zoology , 9 : 28 . Google Scholar Crossref Search ADS PubMed WorldCat Stumpp M. , Hu M. Y. , Melzner F. , Gutowska M. A. , Dorey N. , Himmerkus N. , Holtmann W. C. et al. . 2012 . Acidified seawater impacts sea urchin larvae pH regulatory systems relevant for calcification . Proceedings of the National Academy of Sciences of the United States of America , 109 : 18192 – 18197 . Google Scholar Crossref Search ADS PubMed WorldCat Stumpp M. , Wren J. , Melzner F. , Thorndyke M. C. , Dupont S. T. 2011 . CO2 induced seawater acidification impacts sea urchin larval development I: elevated metabolic rates decrease scope for growth and induce developmental delay . Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology , 160 : 331 – 340 . Google Scholar Crossref Search ADS WorldCat Takei Y. , Hiroi J. , Takahashi H. , Sakamoto T. 2014 . Diverse mechanisms for body fluid regulation in teleost fishes . American Journal of Physiology-Regulatory, Integrative and Comparative Physiology , 307 : R778 – R792 . Google Scholar Crossref Search ADS PubMed WorldCat Torda G. , Donelson J. M. , Aranda M. , Barshis D. J. , Bay L. , Berumen M. L. , Bourne D. G. et al. . 2017 . Rapid adaptive responses to climate change in corals . Nature Climate Change , 7 : 627 – 636 . Google Scholar Crossref Search ADS WorldCat Uppstrom, L. R. 1974. The boron/chemistry ratio of the deap-sea water from the Pacific Ocean. Deep-Sea Research I, 21 : 161 – 162 . Vulesevic B. , Perry S. F. 2006 . Developmental plasticity of ventilatory control in zebrafish, Danio rerio . Respiratory Physiology & Neurobiology , 154 : 396 – 405 . Google Scholar Crossref Search ADS PubMed WorldCat Wanninkhof, R. 2014. Relationship between wind speed and gas exchange over the ocean revisited. Limnology Oceanography Methods, 12 : 351 – 362 . Watson S. A. , Fabricius K. E. , Munday P. L. 2017 . Quantifying pCO(2) in biological ocean acidification experiments: a comparison of four methods . PLoS One , 12 : e0185469 . Google Scholar Crossref Search ADS PubMed WorldCat Williams C. R. , Dittman A. H. , McElhany P. , Busch D. S. , Maher M. T. , Bammler T. K. , MacDonald J. W. et al. . 2019 . Elevated CO2 impairs olfactory-mediated neural and behavioral responses and gene expression in ocean-phase coho salmon (Oncorhynchus kisutch) . Global Change Biology , 25 : 963 – 977 . Google Scholar Crossref Search ADS PubMed WorldCat Zhou J. G. , Wei J. G. , Xu D. , Cui H. C. , Yan Y. , Ou-Yang Z. L. , Huang X. H. et al. . 2011 . Molecular cloning and characterization of two novel hepcidins from orange-spotted grouper, Epinephelus coioides . Fish & Shellfish Immunology , 30 : 559 – 568 . Google Scholar Crossref Search ADS PubMed WorldCat © International Council for the Exploration of the Sea 2020. 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)

Todd, Victoria L, G;Williamson, Laura, D;Cox, Sophie, E;Todd, Ian, B;Macreadie, Peter, I

doi: 10.1093/icesjms/fsz077pmid: N/A

Abstract Offshore Oil & Gas (O&G) infrastructure creates artificial reef complexes that support marine communities in oceans. No studies have characterized the first wave of colonization, which can reveal information about habitat attraction and ecological connectivity. Here we used opportunistically-collected industrial Remotely Operated Vehicles (ROVs) to investigate fish and invertebrate colonization on a new North Sea O&G platform and trenching of an associated pipeline. We observed rapid colonization of fish communities, with increases in species richness (S), abundance (N), and diversity (H′) over the first four days (the entire study period). By contrast, there was minimal change in motile invertebrate communities over the survey period. After trenching, invertebrate S, N and H′ decreased significantly, whilst fish S, N and H′ increased. This study is the first to report on the pioneer wave of fish and invertebrate colonization on O&G infrastructure, thereby providing rare insight into formation of new reef communities in the sea. These short and opportunistic data are valuable in terms of showing what can be discovered from analysis of ‘pre-installation’ ROV footage of O&G structures, of which there are terabytes of data held by O&G companies waiting to be analyzed by environmental scientists. Introduction Sub-sea anthropogenic infrastructure often provide structurally complex hard substrata in contrast to the relatively featureless and sedimentary sea floor (Larcom et al., 2014). In turn, this can accommodate diverse sessile invertebrate communities comprising anemones, hydroids, bryozoans, sponges, mussels, barnacles, soft, and even hard corals (Guerin, 2009; Bergmark and Jørgensen, 2014; Larcom et al., 2014; Todd et al., 2018). Motile invertebrates are also associated with sub-sea infrastructure, using abundant refuge and food availability (Guerin, 2009; Langhamer and Wilhelmsson, 2009; Krone et al., 2013; Lengkeek et al., 2013). Commercially important fish have also been observed living in association with sub-sea infrastructure (Jørgensen et al., 2002; Løkkeborg et al., 2002; Soldal et al., 2002; Guerin, 2009; Friedlander et al., 2014; Fujii et al., 2014), many of which are juveniles that preferentially select structurally complex habitats (Sayer et al., 2005). Marine mammals have also been reported to aggregate around, rest on, and preferentially forage around structures and pipelines (Todd et al., 2009; Russell et al., 2014; Todd et al., 2016; Orr et al., 2017; Delefosse et al., 2018). Cessation of fishing pressure due to fishing exclusion zones, combined with plentiful food and refuge, has led some to suggest that certain sub-sea infrastructure may act as nursery grounds for fish and invertebrates (e.g. Sayer et al., 2005; Love et al., 2006; McLean et al., 2017; Todd et al., 2018). Fish-stock growth is dependent on juveniles reaching reproductive maturity (Sayer et al., 2005), and certain infrastructure may have potential to augment local fish populations (Cripps and Aabel, 2002; Jørgensen et al., 2002; Fujii et al., 2014; Fujii, 2015); however, it is disputed currently as to whether fish stocks are enhanced or merely concentrated by such habitat (Bohnsack, 1989; Fujii et al., 2014), also known as the “production vs. attraction” debate. In 2015, the Dogger Bank (DB) accommodated 79 offshore installations (OSPAR, 2015), in addition to wellheads, manifolds, and pipelines; however, due to diminishing hydrocarbon reserves and aging infrastructure, the number of structures requiring decommissioning is set to increase within the decade (Jørgensen, 2012). Current North Sea decommissioning legislation, principally the Oslo and Paris Convention (OSPAR) Decision 98/3, dictates that all so-called obsolete O&G infrastructure must be removed completely upon decommissioning. This will incur significant cost and likely cause major environmental disruption; consequently, many among the scientific community argue that there is scope in decommissioning obsolete infrastructure to form artificial reefs (e.g. Cripps and Aabel, 2002; Todd et al., 2009; Jørgensen, 2012; Macreadie et al., 2012; Bergmark and Jørgensen, 2014; Fowler et al., 2014; Todd et al., 2018). This is a paradigm referred to collectively as “rigs to reefs” (RTR). Despite a recent renaissance in North Sea RTR research (Fowler et al., 2018), there is still a profound need for improved scientific understanding of marine life associated with O&G infrastructure to accurately assess RTR applicability to the North Sea (Jørgensen, 2012; Macreadie et al., 2012; Fowler et al., 2014). The majority of literature to date has investigated marine life on mature O&G infrastructure, i.e. which has been in the oceans for decades, whereas no studies (to our knowledge) have characterized the primary wave of colonization when O&G infrastructure is first installed. Characterizing this first wave of colonization can provide key insight into ecological connectivity and can help with understanding what makes O&G infrastructure so effective as reef habitat. Moreover, information on the first wave of colonization is important for RTR discussions because it can help inform whether different types of decommissioning strategies (e.g. “topping” vs. “toppling”) might affect ongoing colonization and recruitment of marine life on O&G structures. The present study endeavours to help address the paucity of information on the first wave of colonization by analysing effects of sub-surface jacket installation and pipeline construction on benthic motile invertebrate and fish communities shortly after construction. Objectives This study made scientific use of routine industrially collected remotely operated vehicle (ROV) visual inspection data during trenching (the process of burying pipeline below the seafloor in a trench) of cylindrical concrete pipeline, CCP (henceforth “pipeline”) from the newly constructed “A-18” (55°06.406′N 3°50.045′E) satellite platform to the pre-existing “A-12” platform (55°23.654′N 3°48.484′E) located 32.6 km at a bearing of 357° from A-18, with an aim to quantify effects of platform on benthic-motile invertebrate and fish-community similarity, species richness (S), abundance (N), and diversity (H′) to contribute to wider evaluation of the utility of a North Sea RTR scheme. The diversity index is a mathematical measure of species diversity in a community (Magurran, 2004) that provides more information about community composition than simply species richness (i.e. the number of species present), taking the relative abundances of different species into account. Our analyses aimed to test the null hypotheses that there would be no significant variation in benthic-motile invertebrate and fish-community similarity, S, N, or H′, related to installing the A-18 satellite platform or trenching of the A-18 to A-12 pipeline. We acknowledge that our ability to make robust comparisons between the new (A-18) and mature (A-12) platforms is somewhat limited—having multiple new and mature platforms would make for a much more robust design; however, this study is opportunistic in nature, capitalizing on an opportunity to extract rare and brief industry data to improve ecological understanding. Material and methods Site description Surveys were performed in the Dutch sector of the DB, which is a vast expanse of submerged glacial moraine forming an offshore sandbank, situated between the thermally stratified north and isothermal southern North Sea (Pingree and Griffiths, 1978). This transitional location induces frequent frontal circulation systems that promote nutrient upwelling and facilitate year-round plankton production (Pedersen and Hansen, 1993). Such consistent and plentiful primary productivity nourishes the DB’s silt-sand substratum, supporting a diverse benthic (seafloor) community (Wieking and Kröncke, 2005), which in turn, supports rich invertebrate (Callaway et al., 2002), fish (Sell and Kröncke, 2013), and marine mammal assemblages (Matthiopoulos et al., 2004). A new satellite gas production platform (A18) was installed and a CCP laid between the A18 and the existing A12 platform 32.6 km away (Figure 1). Figure 1. View largeDownload slide (a) Chart of North East Atlantic, highlighting locations of A-18 and A-12 platforms in relation to bathymetry. (b) Inset shows approximate chart of DB (outline varies with season, and there is thus no accepted delineation), highlighting the red line of the A-18 to A-12 CCP. Figure 1. View largeDownload slide (a) Chart of North East Atlantic, highlighting locations of A-18 and A-12 platforms in relation to bathymetry. (b) Inset shows approximate chart of DB (outline varies with season, and there is thus no accepted delineation), highlighting the red line of the A-18 to A-12 CCP. Videographic sampling and timings ROV general visual inspection (GVI) footage of the A-18 satellite platform and pipeline was provided by Petrogas E&P Netherlands B.V, collected by Allseas Group SA and Bluestream Offshore BV using Ultra-High Definition (UHD)-42 and Seaeye Cougar XT ROVs. Pre-platform placement ROV footage was collected during “as-laid” and “as-trenched” pipeline surveys between 8 September 2015 and 27 September 2015. A18 platform legs were piled, and the platform placed into position over a construction period spanning 4–13 October 2015; consequently, ROV “pre-installation” and “post-installation” debris inspections of the A-18 location occurred between 10 October 2015 and 14 October 2015. These short timelines are typical of industry data collection studies, which are not designed to quantify biotic interactions with offshore infrastructure. As such, these data are considered rare and opportunistic. Motile invertebrate and fish sampling To quantify motile invertebrate and fish communities living in association with sub-surface infrastructure, ten separate 5-min sections of footage were extracted, using a random stratified technique, as per Todd et al. (2018), from both “as-laid” and “as-trenched” inspections of the pipeline and “pre-installation” and “post-installation” inspections at A-18. Five-minute timed-count surveys were then conducted, in which observed organisms were identified to the lowest possible taxonomic group, as per Hughes et al. (2010), Söffker et al. (2011), and Todd et al. (2018), and enumerated. Sessile organisms had not had time to establish on sub-surface infrastructure, and the newly installed platform was considered to be “clean”. Data analysis Statistical tests were performed using PRIMER-E v.7.0 (PRIMER-E Ltd., Ivybridge, UK) and Minitab v.17.0 (Minitab Ltd., Coventry, UK). In order to deduce motile invertebrate and fish community diversity indices (S, N, and H′), α-diversity analysis, as described by Magurran (2004), was applied to timed-count data using PRIMER-7. To assess significant differences between motile-invertebrate and fish-community similarity and diversity indices, between both “pre-installation” and “post-installation” inspections at A-18, in addition to “as-laid” and “as-trenched” inspections of the pipeline, parametric, and non-parametric tests with appropriate ad hoc tests were employed. Diversity indices which conformed to normal distribution (Kolmogorov–Smirnov tests, p > 0.05) and expressed equal variance (Levene’s tests, p > 0.05) underwent parametric t-tests assuming equal variances (Dytham, 2011; Fowler et al., 2013). Indices that conformed to normal distribution (Kolmogorov–Smirnov tests, p > 0.05) yet expressed unequal variances (Levene’s tests, p < 0.05) underwent parametric t-tests assuming unequal variances. Diversity indices which did not conform to normal distribution (Kolmogorov–Smirnov tests, p < 0.05) after appropriate normalization transformations underwent non-parametric Mann–Whitney U tests (Dytham, 2011; Fowler et al., 2013). To represent difference in motile invertebrate and fish-community similarity graphically between both “pre-installation” and “post-installation” inspections at A-18 in addition to “as-laid” and “as-trenched” inspections of the pipeline, four multidimensional scaling (MDS) ordination analysis tests were applied to corresponding Bray–Curtis similarity matrices (PRIMER-E, 2001). To determine if community similarity differed significantly, three analysis of similarity (ANOSIM) tests were applied to corresponding Bray–Curtis similarity matrices (PRIMER-E, 2001). Where community similarity was significantly different, Similarity of Percentages (SIMPER) analysis tests were also performed to elucidate which organisms expressed the greatest contribution to defining respective communities (PRIMER-E, 2001). Results The following is a simplified summary of comprehensive statistical analysis of community ecological dynamics. Full results can be found in the Supplementary Material. To serve as a reminder again here, S = species richness (total number of species in the community), N = abundance (or numbers), and H′ = diversity (index). A-18 satellite platform Motile invertebrates Motile invertebrate communities were dominated by the hermit crab (Pagurus bernhardus) followed to a lesser degree by starfish (Asteroidea), the gastropod whelk (Buccinum undatum), and the blue jellyfish (Cyanea lamarckii), all of which decreased in abundance once the A-18 jacket had been installed (Figure 2). Figure 2. View largeDownload slide SIMPER analysis highlighting average abundance (N) of most influential motile invertebrate species and percentage contribution to the overall 52.1% dissimilarity between pre- and post-installation communities. Figure 2. View largeDownload slide SIMPER analysis highlighting average abundance (N) of most influential motile invertebrate species and percentage contribution to the overall 52.1% dissimilarity between pre- and post-installation communities. Fish Installation of the A-18 sub-surface jacket had a much more profound effect on the benthic fish community: S, N, and H′ all expressed statistically significant increases once the jacket had been installed (Student t-test, Student t-test and Mann–Whitney U test, p < 0.05, respectively). As such, fish community similarity expressed a significant difference between pre- and post-installation samples (ANOSIM, p < 0.05). Common dab (Limanda limanda) and common dragonet (Callionymus lyra) dominated both communities (Figure 3); however, whiting (Merlangius merlangus) were present only during post-jacket installation. All species expressed greater abundance once the A-18 jacket had been installed. Figure 3. View largeDownload slide SIMPER analysis highlighting average abundance (N) of most influential fish species and percentage contribution to the overall 57.4% dissimilarity between pre- and post-installation communities. Figure 3. View largeDownload slide SIMPER analysis highlighting average abundance (N) of most influential fish species and percentage contribution to the overall 57.4% dissimilarity between pre- and post-installation communities. A-18 to A-12 pipeline Motile invertebrates Trenching of the pipeline had a clear effect on the local motile invertebrate population. Invertebrate S, N, and H′ all expressed statistically significant decreases after trenching (Student t-test, Mann–Whitney U test and Mann–Whitney U test, p < 0.05, respectively). As such, invertebrate community similarity expressed a significant difference between “as-laid” and “as-trenched” samples (ANOSIM, p < 0.05). P. bernhardus, swimming crab (Liocarcinus holsatus), edible crab (Cancer pagurus), and B. undatum all decreased in abundance after trenching, whereas, Asteroidea increased in abundance (Figure 4). Figure 4. View largeDownload slide SIMPER analysis highlighting average abundance (N) of most influential invertebrate species and percentage contribution to the overall 94.2% dissimilarity between “as-laid” and “as-trenched” communities. Figure 4. View largeDownload slide SIMPER analysis highlighting average abundance (N) of most influential invertebrate species and percentage contribution to the overall 94.2% dissimilarity between “as-laid” and “as-trenched” communities. Fish Trenching of the pipeline also had a clear effect on the local benthic fish population. Fish community S, N, and H′ all expressed statistically significant increases after trenching (Mann–Whitney U tests, p < 0.05). As such, fish community similarity expressed a significant difference between “as-laid” and “as-trenched” samples (ANOSIM, p < 0.05). L. limanda, C. lyra, and the grey gurnard (Eutrigla gurnardus) were all more abundant after trenching, whereas, M. merlangus were more abundant when the pipeline was exposed (Figure 5). Figure 5. View largeDownload slide SIMPER analysis highlighting average abundance (N) of most influential fish species and percentage contribution to the overall 82.5% dissimilarity between “as-laid” and “as-trenched” communities. Figure 5. View largeDownload slide SIMPER analysis highlighting average abundance (N) of most influential fish species and percentage contribution to the overall 82.5% dissimilarity between “as-laid” and “as-trenched” communities. Discussion A-18 satellite platform Motile invertebrates There was no significant variation in motile invertebrate community similarity immediately after installation of the A-18 satellite platform; therefore, in this instance, the null hypothesis was accepted. This is likely influenced by the short time period of the study. Langhamer and Wilhelmsson (2009), Krone et al. (2013), and Lengkeek et al. (2013) all observed increased abundance of hard substrata dwelling invertebrates as a result of refuge provided by the addition of solid sub-surface infrastructure. The limited change in invertebrate community composition observed here is likely because the jacket had only been in situ for 4 days when GVIs were conducted; therefore, sessile invertebrates, which form the base of hard substrata communities (Whomersley and Picken, 2003), would not have had time to establish, reducing habitat value for motile invertebrates. Langhamer and Wilhelmsson (2009) only observed a significant increase in motile invertebrate abundance 3 months after installation of artificial substrata. Fish Installation of the A-18 sub-surface jacket had a much clearer immediate effect on local fish communities. S, N, and H′ all expressed significant increases leading to significant variation between pre- and post-communities. These findings are similar to previous studies (e.g. Soldal et al., 2002; Love et al., 2006; Langhamer and Wilhelmsson, 2009; Martínez, 2011; Fujii, 2015), all of which observed increased fish abundance associated with sub-surface infrastructure. For example, Løkkeborg et al. (2002) conducted gillnet surveys at two North Sea platforms and documented a clear increase in fish density within 100 m of structures. Moreover, both Picken and McIntyre (1989) and Aabel et al. (1997) reviewed interactions between fish and sub-surface infrastructure and observed consistent patterns of association. Associations between fish and sub-surface infrastructure have been attributed to increased structural complexity (Guerin, 2009), refuge (Lengkeek et al., 2013), food availability (Fabi et al., 2004), and more recently reproduction (Todd et al., 2018). Sayer et al. (2005) demonstrated that fish, particularly juvenile gadoids, select preferentially the most complex habitat, concordant with the 88% increase observed in M. merlangus after A-18’s installation. Abundance of juvenile gadoids supports suggestions that sub-sea infrastructure may act as nursery areas (Sayer et al., 2005; Love et al., 2006); however, without site fidelity studies, such as those by Ridgway et al. (1991), Anthony et al. (2012), and Lowe et al. (2009), a conclusive outcome remains tenuous. Nevertheless, recent research conducted by Todd et al. (2018) on a platform distanced 26 km from A-18, suggests that certain obligate hard-substrata species actually use sub-surface infrastructure directly as vectors for reproduction. It should also be noted that obligate soft-sediment dwelling species are attracted to sub-surface infrastructure (Aabel et al., 1997; Guerin, 2009; Martínez, 2011) consistent with the 39% rise in L. limanda and 27% rise in C. lyra abundance observed post installation of A-18. Despite literature highlighting associations between fish and sub-sea infrastructure, Bohnsack (1989) and Fujii et al. (2014) discuss the paucity of information surrounding ecological dependency; specifically, if fish populations are attracted or are enhanced by sub-sea structures (the attraction vs. production debate). In this instance, the limited time A-18 had been in situ, and consequential lack of sessile life and invertebrate prey, would suggest the jacket was only attracting fish rather than actively supporting the local fish population. Nevertheless, it is highly probable that with time, sessile communities will establish (Whomersley and Picken, 2003; Meyer et al., 2018), attracting motile invertebrates (Langhamer and Wilhelmsson, 2009), which in turn support fish populations (Cowan and Rose, 2016), and possibly even marine mammals (Todd et al., 2009, 2016). When considered in the context of the RTR paradigm, if these findings, and supporting suggestions by De Groot (1982) and Osmundsen and Tveterås (2003) that sub-sea infrastructure may be acting as a network of de facto marine-protected areas (MPAs) and augmenting North Sea fish populations (Sayer et al., 2005; Fujii et al., 2014), the gradual loss of structures due to decommissioning legislation may remove refuge for many already depleted fish stocks; however, it should be noted that the dataset was small and only represented a short temporal scale. Moreover, image quality was poor on account of the ROV camera specification and light configuration, thus, smaller inconspicuous organisms may have been overlooked contributing to inaccurate representation of the community. In order to assess accurately the fish population, a greater volume of high-resolution footage over a longer temporal period is required. A-18 to A-12 pipeline Motile invertebrates Trenching of the A-18 to A-12 pipeline resulted in a significant decrease in motile invertebrate S, N, and H′, contributing to significant community variation between “as-laid” and “as-trenched” samples. The null hypothesis was therefore rejected. Despite representing a significant proportion of sub-sea infrastructure, pipelines have received little attention within RTR literature, a point stressed by Culwell and McCarthy (1998) and Lacey and Hayes (2019). As such, there are no peer-reviewed scientific studies documenting effects of pipeline trenching on benthic communities. Literature that discusses organisms associated with exposed pipelines (Allen et al., 1976; Lacey and Hayes, 2019) observed significant aggregations of motile invertebrates, comparable to the findings of this study. Both Allen et al. (1976) and Love and York (2005) suggested that invertebrates are attracted by abundant refuge and food compared to that of surrounding seabed. Reduction in decapod (crab) and mollusc (snail) presence is likely attributed to pipeline burial during trenching, resulting in loss of refugia. Furthermore, perturbation of the seabed is likely to have either evoked an avoidance response or buried invertebrates living in close association with the pipeline. Nevertheless, De Groot (1996) concluded that pipelines have minor impact on benthic communities, as disturbance of the seabed is restricted to the time of laying and burial. In practice, lack of scientific information regarding effects of pipeline construction and, in this case, absence of “pre-lay” footage to provide a control community, leave any conclusions tenuous until further research has been conducted. Whilst trenching reduced most benthic invertebrate presence, the abundance of Asterioids—specifically sand sea stars (Astropecten irregularis)—increased. A. irregularis are crepuscular, and feed only during twilight hours. Between periods of feeding activity, they remain buried in the substrate (Christensen, 1971). By chance, the majority of “as-trenched” footage was collected during twilight hours, whereas, “as-laid” footage was collected during periods of complete darkness or daylight. As such, it is likely that A. irregularis abundance was underrepresented during “as-laid” surveys. This issue was exacerbated by the limited quantity of useable GVI footage. It is also possible that usually buried A. irregularis were unearthed by the “digging Donald” during trenching and lay exposed on the seabed along with razor shells (Ensis sp.). A. irregularis may have also been attracted to the area due to the abundance of bivalve prey unearthed during the post-trenching process. Considering the paucity of information surrounding exposed pipelines, no definitive conclusions can be made regarding their importance as invertebrate habitat; however, the present study demonstrates that invertebrates are attracted to these structures. As pipelines contribute a significant proportion of sub-surface infrastructure, they may play a significant role in structuring benthic communities (Lacey and Hayes, 2019). Nevertheless, additional analysis over longer temporal scales using focused high-quality footage is required to determine accurately the significance of such habitat. Fish Trenching of the A-18 to A-12 pipeline had a significant effect on fish communities. S, N, and H′ all expressed significant increases post trenching resulting in significant variation between “as-laid” and “as-trenched” samples. The null hypothesis was therefore rejected. As with invertebrates, there are no peer-reviewed scientific studies documenting effects of pipeline trenching on fish communities. Likewise, literature discussing fish associated with exposed pipelines is also limited; nevertheless, significant aggregations of fish, around exposed pipelines have been reported (Allen et al., 1976; McLean et al., 2017). Greatest concentrations were observed in areas of high complexity. Allen et al. (1976) found this was primarily around nodes, terminals, and protective rocks, whereas Love and York (2005) noted highest concentrations in undercut areas, in some cases supporting seven times greater abundance than surrounding seabed. High proportions of fish associated with pipelines have been reported to be juveniles and larvae, supporting the theory that pipelines may act as nursery areas (Love and York, 2005; McLean et al., 2017). Post trenching, despite loss of habitat provided by the pipeline, there was a notable increase in benthic fish, predominantly L. limanda, C. lyra, and E. gurnardus, all of which are obligate bottom feeders that prey on small interstitial and epibenthic invertebrates (e.g. De Gee and Kikkert, 1993; Hinz et al., 2005; Griffin et al., 2012). Trenching is likely to have unearthed significant concentrations of interstitial invertebrate prey, increasing local food availability temporarily, and possibly contributing to increased abundance of benthic fish. Nevertheless, this phenomenon would have been ephemeral and would not sustain high concentrations of benthic fish for prolonged durations. Conversely, M. merlangus abundance decreased post trenching. M. merlangus is a member of the gadoid (cod) family and thus expresses strong affiliation with complex habitat (Sayer et al., 2005). Burial of the pipeline resulted in potential loss of refugia and reduced habitat, potentially explaining reduced abundance of this species. As with motile invertebrates, lack of information surrounding exposed pipelines and their interactions with fish make any conclusions regarding their importance as habitat premature; however, this study demonstrates that fish, particularly gadoids, are attracted to these structures. Given that pipelines contribute a significant proportion of offshore anthropogenic, sub-surface infrastructure, they may also play a significant role in structuring fish communities. This particular pipeline is now predominantly buried, and offers little habitat value, with the exception of exposed lay-down and start-up heads. Future research should build upon the findings of Love et al. (2006) by investigating significance of structures as feeding and spawning grounds for mature invertebrates and fish in addition to acting as nursery areas for juveniles. Considering the volume of sub-sea infrastructure currently due for decommissioning (OSPAR, 2015), combined with strong and consistent association with marine organisms (Picken and McIntyre, 1989; Aabel et al., 1997), there is an urgent need to elucidate roles these structures play within marine ecosystems. Future recommendations There are several disadvantages of using industry-obtained ROV footage for studying ecological interactions of O&G structures, as opposed to scientifically-collected ROV footage. In this section, we consider these disadvantages and frame them in terms of recommendations for future improvements to the way industry ROV footage could be collected to improve its scientific value. We recognize, however, that our recommendations add an effort burden to industry, so for suggestions to be implemented realistically by industry, there would need to be industry-academic partnerships (e.g. SEA SERPENT, Macreadie et al. 2018). Future studies should always include “pre-installation” footage and survey over longer temporal scales; this study only covered a 4-day period, due to limited availability of industry-collected ROV footage. It is also important to understand the benthic-community growth on both the structure (at different depth levels) as well as under and around the platform (including cuttings piles), which can constitute reefs within themselves. Again, because we were using industry footage that was collected for operational purposes—as opposed to ecological monitoring—we were limited in the data coverage of the structure. ROV surveys often use other technologies (pipe tracker, multibeam, blueview, obstacle avoidance sonar, etc.) over visual footage, resulting in image quality being of lower priority. As such, lighting configuration and camera specification are often sub-optimal. The key information to pass on to ROV pilots for this study is the importance of footage quality (camera specifications, lighting configuration, image stability, etc.) that is likely to yield greatest improvements in accuracy (Macreadie et al., 2018). In order to quantify fish throughout the water column, ROV pilots should also collect as much close visual inspection (CVI) and GVI footage as practicably possible. Many surveys concentrate simply on certain elements; however, CVI and GVI often cover the entirety of the sub-surface structure providing a good overview of the water column from the surface to the sea bed. It would also be beneficial to have data from different times of day and year for suitable comparisons to be made. In short, to improve the quality of data for the continuation of this study, ROV pilots should collect more high-definition (HD) footage, to enable the user to identify to species level, and also to differentiate between recreational and commercial fish species. This footage should be at timed intervals at various depths, to account for species differences due to depth and time of day. Footage should be collected so that a full range of vision (close up and at a distance) is achieved, to assess species composition. Conclusions Installation of the A-18 jacket had a strong and immediate aggregating effect on fish, but effects on invertebrates were inconclusive; Trenching of the A-18 to A-12 pipeline reduced abundance of motile invertebrates associated with the area, whilst temporally increasing local fish abundance; Post trenching, it is unlikely the pipeline will offer any meaningful habitat because it will predominantly be buried; Whilst the A-18 jacket was observed attracting only fish, drawing inference from literature, in time the structure is likely to accrue sessile life and in turn begin to support a diverse marine community; and, Collection of further high-quality video footage during CVI and GVI surveys would enhance the capacity of using industry-collected ROV data to answer scientifically-relevant questions. Acknowledgements Thanks to Petrogas E&P Netherlands B.V. for awarding the work and continued support of OSC’s ongoing RTR research. Specifically, we thank Alan Shand (Petrogas), who’s enthusiasm, passion, and appreciation for the project made this work so enjoyable. Finally, thanks to Jane Warley and Ed Lavallin (OSC) for participation in both analysis and editing earlier drafts during the original project. References Aabel J. , Cripps S. , Jensen A. , Picken G. 1997 . Creating Artificial Reefs from Decommissioned Platforms in the North Sea: Review of Knowledge and Proposed Programme of Research . Dames & Moore Group . Report of Dames and Moore, RF Rogaland Research, Southhampton Oceanography Centre and Cordah to the Exploration and Production Forum. Allen M. J. , Pecorelli H. , Word J. 1976 . Marine organisms around outfall pipes in Santa Monica Bay . Journal of Water Pollution Control Federation , 48 : 1881 – 1893 . Anthony K. M. , Love M. S. , Lowe C. G. 2012 . Translocation, homing behavior and habitat use of groundfishes associated with oil platforms in the east Santa Barbara channel, California . Bulletin, Southern California Academy of Sciences , 111 : 101 – 118 . Google Scholar Crossref Search ADS Bergmark P. , Jørgensen D. 2014 . Lophelia pertusa conservation in the North Sea using obsolete offshore structures as artificial reefs . Marine Ecology Progress Series , 516 : 275 – 280 . Google Scholar Crossref Search ADS Bohnsack J. A. 1989 . Are high densities of fishes at artificial reefs the result of habitat limitation or behavioral preference? Bulletin of Marine Science , 44 : 631 – 645 . Callaway R. , Alsvåg J. , De Boois I. , Cotter J. , Ford A. , Hinz H. , Jennings S. et al. 2002 . Diversity and community structure of epibenthic invertebrates and fish in the North Sea . ICES Journal of Marine Science , 59 : 1199 – 1214 . Google Scholar Crossref Search ADS Christensen A. M. 1971 . Feeding Biology of the Sea-Star: Astropecten Irregularis Pennant . Københavns Universitet . Cowan J. H. , Rose K. A. 2016 . Oil and gas platforms in the Gulf of Mexico: their relationship to fish and fisheries. In Fisheries and Aquaculture in the Modern World . Ed. by Makkola H. . IntechOpen , Rijeka, Croatia . pp. 95 – 122 . Cripps S. J. , Aabel J. P. 2002 . Environmental and socio-economic impact assessment of Ekoreef, a multiple platform rigs-to-reefs development . ICES Journal of Marine Science , 59 : S300 – S308 . Google Scholar Crossref Search ADS Culwell A. S. , McCarthy J. C. 1998 . Pipeline and power cable decommissioning. In Proceedings: Public Workshop, Decommissioning and Removal of Oil and Gas Facilities Offshore California: Recent Experiences and Future Deepwater Challenges, California , pp. 66 – 73 . Ed. by Manago F. , Williamson B. . De Gee A. , Kikkert A. 1993 . Analysis of the grey gurnard (Eutrigla gurnardus) samples collected during the 1991 International Stomach Sampling Project . ICES Journal of Marine Science , 1 : 14 – 26 . De Groot S. 1982 . The impact of laying and maintenance of offshore pipelines on the marine environment and the North Sea fisheries . Ocean Management , 8 : 1 – 27 . Google Scholar Crossref Search ADS De Groot S. 1996 . Quantitative assessment of the development of the offshore oil and gas industry in the North Sea . ICES Journal of Marine Science , 53 : 1045 – 1050 . Google Scholar Crossref Search ADS Delefosse M. , Rahbek M. L. , Roesen L. , Clausen K. T. 2018 . Marine mammal sightings around oil and gas installations in the central North Sea . Journal of the Marine Biological Association of the United Kingdom , 98 : 993 – 1001 . Google Scholar Crossref Search ADS Dytham C. 2011 . Choosing and Using Statistics: A Biologist’s Guide . John Wiley & Sons , Sussex, UK . Fabi G. , Grati F. , Puletti M. , Scarcella G. 2004 . Effects on fish community induced by installation of two gas platforms in the Adriatic Sea . Marine Ecology Progress Series , 273 : 187 – 197 . Google Scholar Crossref Search ADS Fowler A. M. , Jørgensen A.M. , Svendsen J. C. , Macreadie P. I. , Jones D. O. B. , Boon A. R. , Booth D. J. et al. 2018 . Environmental benefits of leaving offshore infrastructure in the ocean . Frontiers in Ecology and the Environment , 16 : 571 – 578 . Google Scholar Crossref Search ADS Fowler A. M. , Macreadie P. I. , Jones D. O. B. , Booth D. J. 2014 . A multi-criteria decision approach to decommissioning of offshore oil and gas infrastructure . Ocean & Coastal Management , 87 : 20 – 29 . Google Scholar Crossref Search ADS Fowler J. , Cohen L. , Jarvis P. 2013 . Practical Statistics for Field Biology . John Wiley & Sons , Sussex, UK . Friedlander A. M. , Ballesteros E. , Fay M. , Sala E. 2014 . Marine communities on oil platforms in Gabon, West Africa: high biodiversity oases in a low biodiversity environment . PLoS One , 9 : e103709. Google Scholar Crossref Search ADS PubMed Fujii T. 2015 . Temporal variation in environmental conditions and the structure of fish assemblages around an offshore oil platform in the North Sea . Marine Environmental Research , 108 : 69 – 82 . Google Scholar Crossref Search ADS PubMed Fujii T. , Walls A. , Horsfield M. 2014 . Is there a net benefit from offshore structures? In SPE International Conference on Health, Safety, and Environment. 17–19 March 2014, Long Beach, California, USA. 9 pp. Griffin R. , Pearce B. , Handy R. 2012 . Dietary preference and feeding selectivity of common dragonet Callionymus lyra in UK . Journal of Fish Biology , 81 : 1019 – 1031 . Google Scholar Crossref Search ADS PubMed Guerin A. J. 2009 . Marine Communities of North Sea Offshore Platforms, and the Use of Stable Isotopes to Explore Artificial Reef Food Webs. Faculty of Engineering, Science & Mathematics, School of Ocean & Earth Sciences . Doctoral dissertation, University of Southampton . 245 pp. Hinz H. , Kröncke I. , Ehrich S. 2005 . The feeding strategy of dab Limanda limanda in the southern North Sea: linking stomach contents to prey availability in the environment . Journal of Fish Biology , 67 : 125 – 145 . Google Scholar Crossref Search ADS Hughes S. J. M. , Jones D. O. B. , Hauton C. , Gates A. R. , Hawkins L. E. 2010 . An assessment of drilling disturbance on Echinus acutus var. norvegicus based on in-situ observations and experiments using a remotely operated vehicle (ROV) . Journal of Experimental Marine Biology and Ecology , 395 : 37 – 47 . Google Scholar Crossref Search ADS Jørgensen D. 2012 . OSPAR’s exclusion of rigs-to-reefs in the North Sea . Ocean & Coastal Management , 58 : 57 – 61 . Google Scholar Crossref Search ADS Jørgensen T. , Løkkeborg S. , Soldal A. V. 2002 . Residence of fish in the vicinity of a decommissioned oil platform in the North Sea . ICES Journal of Marine Science , 59 : S288 – S293 . Google Scholar Crossref Search ADS Krone R. , Gutow L. , Brey T. , Dannheim J. , Schröder A. 2013 . Mobile demersal megafauna at artificial structures in the German Bight – likely effects of offshore wind farm development . Estuarine, Coastal and Shelf Science , 125 : 1 – 9 . Google Scholar Crossref Search ADS Lacey N. C. , Hayes P. 2019 . Epifauna associated with subsea pipelines in the North Sea . ICES Journal of Marine Science , doi: 10.1093/icesjms/fsy196. Langhamer O. , Wilhelmsson D. 2009 . Colonisation of fish and crabs of wave energy foundations and the effects of manufactured holes – a field experiment . Marine Environmental Research , 68 : 151 – 157 . Google Scholar Crossref Search ADS PubMed Larcom E. A. , McKean D. L. , Brooks J. M. , Fisher C. R. 2014 . Growth rates, densities, and distribution of Lophelia pertusa on artificial structures in the Gulf of Mexico . Deep Sea Research Part I: Oceanographic Research Papers , 85 : 101 – 109 . Google Scholar Crossref Search ADS Lengkeek W. , Coolen J. , Gittenberger A. , Schrieken N. 2013 . Ecological relevance of ship wrecks in the North Sea . Nederlandse Faunistische Mededelingen , 41 : 49 – 59 . Løkkeborg S. , Humborstad O. B. , Jørgensen T. , Soldal A. V. 2002 . Spatio-temporal variations in gillnet catch rates in the vicinity of North Sea oil platforms . ICES Journal of Marine Science , 59 : S294 – S299 . Google Scholar Crossref Search ADS Love M. S. , Schroeder D. M. , Lenarz W. , MacCall A. , Bull A. S. , Thorsteinson L. 2006 . Potential use of offshore marine structures in rebuilding an overfished rockfish species, bocaccio (Sebastes paucispinis ). Fishery Bulletin , 104 : 383 – 390 . Love M. S. , York A. 2005 . A comparison of the fish assemblages associated with an oil/gas pipeline and adjacent seafloor in the Santa Barbara Channel, Southern California Bight . Bulletin of Marine Science , 77 : 101 – 118 . Lowe C. G. , Anthony K. M. , Jarvis E. T. , Bellquist L. F. , Love M. S. 2009 . Site fidelity and movement patterns of groundfish associated with offshore petroleum platforms in the Santa Barbara Channel . Marine and Coastal Fisheries , 1 : 71 – 89 . Google Scholar Crossref Search ADS Macreadie P. I. , Fowler A. M. , Booth D. J. 2012 . Rigs-to-reefs policy: can science trump public sentiment? Frontiers in Ecology and the Environment , 10 : 179 – 180 . Google Scholar Crossref Search ADS Macreadie P. I. , McLean D. L. , Thomson P. G. , Partridge J. C. , Jones D. O. B. , Gates A. R. , Benfield M. C. et al. 2018 . Eyes in the sea: unlocking the mysteries of the ocean using industrial, remotely operated vehicles (ROVs) . Science of the Total Environment , 634 : 1077 – 1091 . Google Scholar Crossref Search ADS PubMed Magurran A. E. 2004 . Measuring biological diversity . African Journal of Aquatic Sciences , 29 : 285 – 286 . Google Scholar Crossref Search ADS Martínez I. 2011 . Demersal Fish Assemblages around Sea Bed Features: Buzzard Oil & Gas Field Platform in the North Sea and Jones Bank, Celtic Sea . Doctoral dissertation, University of Aberdeen . 190 pp. Matthiopoulos J. , McConnell B. , Duck C. , Fedak M. 2004 . Using satellite telemetry and aerial counts to estimate space use by grey seals around the British Isles . Journal of Applied Ecology , 41 : 476 – 491 . Google Scholar Crossref Search ADS McLean D. L. , Partridge J. C. , Bond T. , Birt M. J. , Bornt K. R. , Langlois T. J. 2017 . Using industry ROV videos to assess fish associations with subsea pipelines . Continental Shelf Research , 141 : 76 – 97 . Google Scholar Crossref Search ADS Meyer K. S. , Li Y. , Young C. M. 2018 . Oceanographic and biological influences on recruitment of benthic invertebrates to hard substrata on the Oregon shelf . Estuarine, Coastal and Shelf Science , 208 : 1 – 8 . Google Scholar Crossref Search ADS Orr A. J. , Harris J. D. , Hirschberger K. A. , DeLong R. L. 2017 . Qualitative and quantitative assessment of use of offshore oil and gas platforms by the California sea lion (Zalophus californianus). U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-362. 72 pp. Osmundsen P. , Tveterås R. 2003 . Decommissioning of petroleum installations—major policy issues . Energy Policy , 31 : 1579 – 1588 . Google Scholar Crossref Search ADS OSPAR. 2015 . Biennial update of the inventory of oil and gas offshore installations in the OSPAR maritime area, 2015, OSPAR archives, London. Pedersen F. B. , Hansen L. 1993 . Structure of plankton communities in the Dogger Bank area (North Sea) during a stratified situation . Marine Ecology Progress Series , 95 : 115 – 131 . Google Scholar Crossref Search ADS Picken G. B. , McIntyre A. D. 1989 . Rigs to reefs in the North-Sea . Bulletin of Marine Science , 44 : 782 – 788 . Pingree R. , Griffiths D. 1978 . Tidal fronts on the shelf seas around the British Isles . Journal of Geophysical Research: Oceans (1978–2012) , 83 : 4615 – 4622 . Google Scholar Crossref Search ADS PRIMER-E. 2001 . PRIMER 6 for Windows. PRIMER-E Ltd, Plymouth. Ridgway M. S. , MacLean J. A. , MacLeod J. C. 1991 . Nest-site fidelity in a centrarchid fish, the smallmouth bass (Micropterus dolomieui) . Canadian Journal of Zoology , 69 : 3103 – 3105 . Google Scholar Crossref Search ADS Russell D. J. F. , Brasseur S. , Thompson D. , Hastie G. D. , Janik V. M. , Aarts G. , McClintock B. T. et al. 2014 . Marine mammals trace anthropogenic structures at sea . Current Biology , 24 : R638 – R639 . Google Scholar Crossref Search ADS PubMed Sayer M. , Magill S. , Pitcher T. , Morissette L. , Ainsworth C. 2005 . Simulation‐based investigations of fishery changes as affected by the scale and design of artificial habitats . Journal of Fish Biology , 67 : 218 – 243 . Google Scholar Crossref Search ADS Sell A. F. , Kröncke I. 2013 . Correlations between benthic habitats and demersal fish assemblages—A case study on the Dogger Bank (North Sea) . Journal of Sea Research , 80 : 12 – 24 . Google Scholar Crossref Search ADS Söffker M. , Sloman K. A. , Hall-Spencer J. M. 2011 . In situ observations of fish associated with coral reefs off Ireland . Deep Sea Research Part I: Oceanographic Research Papers , 58 : 818 – 825 . Google Scholar Crossref Search ADS Soldal A. V. , Svellingen I. , Jørgensen T. , Løkkeborg S. 2002 . Rigs-to-reefs in the North Sea: hydroacoustic quantification of fish in the vicinity of a “semi-cold” platform . ICES Journal of Marine Science , 59 : S281 – S287 . Google Scholar Crossref Search ADS Todd V. L. G. , Lavallin E. W. , Macreadie P. I. 2018 . Quantitative analysis of fish and invertebrate assemblage dynamics in association with a North Sea oil and gas installation complex . Marine Environmental Research , 142 : 69 – 79 . Google Scholar Crossref Search ADS PubMed Todd V. L. G. , Pearse W. D. , Tregenza N. C. , Lepper P. A. , Todd I. B. 2009 . Diel echolocation activity of harbour porpoises (Phocoena phocoena) around North Sea offshore gas installations . ICES Journal of Marine Science , 66 : 734 – 745 . Google Scholar Crossref Search ADS Todd V. L. G. , Warley J. C. , Todd I. B. 2016 . Meals on wheels? A decade of megafaunal visual and real-time Passive Acoustic Monitoring detections from on-tow and stationary offshore oil and gas rigs and platforms in the North and Irish Seas . PLoS ONE , 11 : 25 . Whomersley P. , Picken G. 2003 . Long-term dynamics of fouling communities found on offshore installations in the North Sea . Journal of the Marine Biological Association of the UK , 83 : 897 – 901 . Google Scholar Crossref Search ADS Wieking G. , Kröncke I. 2005 . Is benthic trophic structure affected by food quality? The Dogger Bank example . Marine Biology , 146 : 387 – 400 . Google Scholar Crossref Search ADS © International Council for the Exploration of the Sea 2019. 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)

Tewfik,, Alexander;Babcock, Elizabeth, A;Phillips,, Myles

doi: 10.1093/icesjms/fsaa008pmid: N/A

Klunder,, Lise;Lavaleye, Marc S, S;Filippidi,, Amalia;, van Bleijswijk, Judith D L;Reichart,, Gert-Jan;, van der Veer, Henk W;Duineveld, Gerard C, A;Mienis,, Furu

doi: 10.1093/icesjms/fsz252pmid: N/A

Advance access publication 12 September 2019. ICES Journal of Marine Science (2019), doi:10.1093/icesjms/fsy114. The following affiliation for Lise Klunder was not included in the earlier version of this article. This has now been added: Marine Evolution and Conservation, Groningen Institute of Evolutionary Life Sciences, University of Groningen, Nijenborgh 7, 9747 AG Groningen, The Netherlands © International Council for the Exploration of the Sea 2020. 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)

Sühring,, R;Cousins,, A;Gregory,, L;Moran,, C;Papachlimitzou,, A;Phillips,, C;Rowles,, R;Supple,, S;Wilczynska,, M;Birchenough, S N, R

doi: 10.1093/icesjms/fsz172pmid: N/A

Abstract The North Sea is one of the most studied and exploited ecosystems worldwide. The multiple uses from industrial, transport, as well as recreational activities have required researchers, regulators, and legislators to understand and, where possible, to minimize any expected negative environmental impacts. As with any international sea, assessing the current pressures and management actions resulting from these activities is centred on several national and international legislative instruments. This variety of co-existing legislations makes development processes and regulatory assessments cumbersome and time consuming. Hence there is a need to integrate environmental risk assessment and management across sectors, ensuring smart, cost-effective data generation, as well as supporting and standardizing environmental practices. This paper provides an overview of the changing regulatory frameworks regarding offshore chemicals used in the oil and gas industry, and the process of chemical risk assessment conducted under the Offshore Chemical Notification Scheme (ONCS) in the UK. Our view of methodological, research, and regulatory needs and challenges that should be addressed to ensure an adequate and sustainable assessment of offshore chemical use in the North Sea is discussed. Furthermore, we discuss the issues faced regarding chemicals used in the UK oil and gas sector with respect to declining hydrocarbon production. The development of risk and hazard assessments for offshore chemicals in the North Sea During the 1960–1970s, improved drilling technologies and knowledge of the local geology facilitated the exploitation of oil and gas in the North Sea, leading to a self-sufficient period in oil and gas up until the early 2000s for Britain (UK) and the Netherlands (NL) (Craig et al., 2018). These developments in the North Sea have required the use of chemicals on a large scale, combined with the need to consider the risks to the marine environment posed by their discharge. At the time the first wells were drilled, little thought was given to the potential environmental fate and hazard of chemicals used in drilling and the production of hydrocarbons. Drilling chemicals e.g. often included oil-based drilling muds that were discharged to sea. Although such discharges would be maintained for only a relatively short period required to drill each well, the subsequent production of oil and gas from successful wells would normally entail the discharge of chemicals on an ongoing basis. This is due to the presence of water in each reservoir, which on production installations must be separated from the oil and gas and is normally discharged. This “produced water” (PW) inevitably contains many of the chemicals used by the installation. Increasing scrutiny by e.g. non-governmental organizations (NGOs), such as Greenpeace and Friends of the Earth, led to increased pressure on the government to regulate and monitor such chemical discharges. Eventually, this led to both national and international regulations on the types of chemicals discharged and the requirement for justifications for the use of chemicals that are considered environmentally problematic (Craig et al., 2018). In the UK, a voluntary offshore notification scheme (OCNS) was set up in 1979 under which oil companies report quarterly on their chemical use and discharge (Cefas-OCNS, 2018a). This system is even in operation today, but has been amended to include the international reporting and assessment systems (see further details below). On a European level, monitoring and reduction of potential marine pollution from oil and gas production was initially included in regulation regarding pollution from land-based sources (Paris Convention, 1974). Since 1992, discharges from the offshore industry have been regulated separately through the OSPAR Convention (1992). One important measure regarding the reduction of offshore chemical discharge was to introduce the so called substitution warnings that require chemical products with particularly hazardous properties to be replaced by less hazardous ones (Oslo, 1972; Paris Convention, 1974; UN, 1974). In June 2000, the OSPAR Commission adopted Decision 2000/2 on a Harmonised Mandatory Control System (HMCS) for the Use and Reduction of the Discharge of Offshore Chemicals (OSPAR Decision, 2000/2). The aim of this framework was to reduce the use and discharge of harmful chemicals in the North–East Atlantic. Contracting parties to OSPAR, including the UK, subsequently incorporated the HMCS into national legislation. For the UK, the HMCS (and OCNS) is enforced through the Offshore Chemicals Regulations (OCR, 2002). Chemicals are assessed and registered using a multidisciplinary approach involving chemical hazard and risk assessors, at the Centre for Environment, Fisheries and Aquaculture Science (Cefas), on behalf of the regulator, the Department for Business, Energy & Industrial Strategy (BEIS). In 2007, the UK extended the service of OCR to the Netherlands on behalf of the State Supervision of Mines (SSM) Staatstoezicht op de Mijnen. Since then, Cefas has been contracted to assess and register offshore chemicals to be used in the UK or the NL waters, as well as to provide scientific and regulatory advice to the regulator, chemical suppliers, and offshore operators. Also, in 2007, the REACH Regulation (Registration, Evaluation, Authorisation and restriction of Chemicals) came into force. REACH is the system for controlling all chemicals before production/use/import in the European Union (EU) (European Commission, 2013). REACH replaced several national and EU chemical Directives and Regulations (see supporting information Supplementary Table S1) with a new single system, harmonizing chemical regulation and risk assessments across the EU. OSPAR recommended in 2006 (OSPAR, 2006) that the two systems (REACH and HMCS) could run in parallel until the time when they could successfully be merged. However, that merger remains an elusive goal. Current regulation and assessment of offshore chemicals The regulation of offshore chemical use and discharge in the UK follows a two-step process. Any chemical product registered for potential use and discharge is first assessed regarding its potential marine environmental hazard i.e. the potential severity of an impact based on the ecotoxicological properties of the substances within the product (see Section “Hazard assessment” for details) combined with information on the mode of use of the product. Secondly, for planned operations requiring chemical use, the marine environmental risk (hazard and likelihood of environmental exposure) of every individual product is assessed for the specific location and operation (see Section “Risk assessment” for details). The outcome of this risk assessment is used to inform the regulatory permitting (see Section “Permitting”). For both stages of the process, the standard means of assessment is the Chemical Hazard and Risk Management model (CHARM) (CIN, 2004). In the UK (and the Netherlands), offshore chemicals are registered for 3-year periods after which chemical suppliers must apply for re-registration. This rolling registration and periodic re-evaluation are combined with product differentiation using substitution warnings and colour bands to help encourage the use of more environmentally friendly chemical products (La Vedrine et al., 2015). Hazard assessment Hazard assessments are conducted for each substance in a chemical product submitted for registration, with the product assessment driven by the substance that is considered most hazardous. All registered products are collected on the Ranked List of Registered Products (ranked list) (Cefas-OCNS website, 2018a). The ranked list provides operators of offshore platforms in the UK/NL waters with information regarding the hazard assessment ranking (i.e. predicted environmental impact based on generic platform data) of all chemical products available for use in the UK and NL. Products that are expected to have the least environmental impact are assigned with Gold banding, Purple banding is reserved for the most hazardous products. Additionally, every registered product receives a “template,” containing information regarding the most hazardous substance in the product. This template enables operators to have the minimum information necessary to conduct their mandatory risk assessments without needing to know the full composition of the chemical product. Much of the focus of the assessment is on how chemical hazards are characterized and measured. The OCNS hazard assessment focuses primarily on whether a substance is likely to be persistent, bioaccumulative and/or toxic in the marine environment, based on the OSPAR Pre-Screening Scheme (2017). This assessment strategy captures three of the most problematic characteristics of chemicals and provides a model for the comparison of dissimilar substances. The approach can be criticized for ignoring risks that fall outside of this triad (e.g. endocrine disruption). Expert judgement and separate initiatives are relied upon to compensate for such shortcomings in the hazard assessment. Substances that are considered hazardous for the marine environment due to persistence and/or bioaccumulation potential and toxicity are assigned a so called “substitution warning.” While substitutable chemicals are allowed to be used offshore, the operator must apply for a permit and write a justification for the use and/or discharge of the chemical. Currently, the number of different substances recognized as distinct and registered for offshore use in the North Sea is just below 2000. The total annual chemical use (for the entire OSPAR region) is over 700 000 tonnes and the total annual discharge over 200 000 tonnes (OSPAR, 2018). The vast majority (>75%) of these substances are considered to pose little or no risk (PLONOR), but there are a number of applications where hazardous substances are still being used. Identified hazards vary substantially based on individual substances and application types (Figure 1). Half of the most commonly used product types (biocide, cement, corrosion inhibitor, etc.) have median hazardous properties around the regulatory threshold for high toxicity for organic substances (LC50 ≤ 10 mg/l) and around 20% have median hazardous properties around the regulatory threshold for high toxicity for inorganic substances (LC50 ≤ 1 mg/l) (OSPAR Decision, 2000/2) (Figure 1). Figure 1. Open in new tabDownload slide Lethal concentration 50% (LC50)/effect concentration 50% (EC50) in mg/l of the ten most commonly used offshore chemical product functions (as currently registered by Cefas). The black horizontal line represents the median value for the individual product function with the box representing the variability within the quartiles with error bars and outliers (black dots). The red line marks the regulatory threshold for a “highly toxic” classification (for organic substances) of 10 mg/l. Figure 1. Open in new tabDownload slide Lethal concentration 50% (LC50)/effect concentration 50% (EC50) in mg/l of the ten most commonly used offshore chemical product functions (as currently registered by Cefas). The black horizontal line represents the median value for the individual product function with the box representing the variability within the quartiles with error bars and outliers (black dots). The red line marks the regulatory threshold for a “highly toxic” classification (for organic substances) of 10 mg/l. To this day, most ecotoxicological testing still relies on animals, especially if the data are produced to inform regulatory assessment. Such test practices evoke ethical concerns and regulators, industry and science alike strive to develop methods to replace or, at least, reduce animal testing wherever possible. Strategies to reduce the need for new animal tests include the use of data from various species, testing of mixtures as mixtures, rather than compound by compound, read-across data from test results for similar substances, as well as the use of computational simulations (Quantitative Structure–Activity Relationships, or QSARs). This variety of acceptable data sources (as well as the uncertainty inherent to any laboratory experiment) can lead to a large range of submitted data for any given endpoint used to assess a substance. For quality assurance, the regulator can use independent testing to confirm submitted test results (for the given method). The results of such verification testing can help to identify which data/data sources are suitable and most reliable for the assessment of chemical hazard and marine environmental impact, if they are compared with data submitted by other registrants. Risk assessment For risk assessment, the information regarding the hazard of a chemical product is used to conduct a site- and operation-specific assessment of the potential severity of a marine environmental impact and the likelihood for such impact to occur. Different quantitative chemical risk assessment approaches are employed as part of the OCNS assessment, however all are based on the internationally accepted predicted environmental concentration (PEC): predicted no effect concentration (PNEC) ratio in which the PEC of a chemical in the water column is compared with the PNEC (Bascietto et al., 1990). The PNEC can be regarded as the chemical concentration limit, below which no environmental impact is expected. If the PEC>PNEC, then the potential for an environmental risk exists. In each approach, the PNECs are derived in a similar way, through the application of assessment factors that convert experimentally derived acute toxicity results, expressed as effect concentration (EC) or lethal concentration (LC) of 50% of the test organisms, to PNEC values. The main difference in the approaches relates to the calculation of the PEC. Permitting Following the registration of a product, any operator who wishes to use the product offshore must apply for a specific permit. The permit application process involves the operator producing a chemical risk assessment of the proposed use and discharge. For proposed operations in English waters, the OCNS risk assessment team at Cefas critically evaluates the chemical permit applications to establish if the proposed use represents a significant risk to the marine environment. Activities in Scottish waters are assessed by Marine Scotland. Comments are then relayed to the regulator who can decide if the permit is approved. For permitting, the information obtained from product-specific hazard assessments are combined with site-specific use and discharge information. Dilution factors which are selected based on the characteristics of the discharge, are then applied to the discharge concentrations to give PEC values at a distances of 500 m from the discharge point. The assessment factors used in CHARM to convert the experimentally derived toxicity results to PNEC values are chosen based on the type of discharge, continuous or batch and the amount of available toxicity data. The resultant PEC:PNEC ratio is called the CHARM RQ (risk quotient). An RQ is calculated for each relevant chemical, with values greater than unity indicative of a potentially significant risk to the environment. Any such instance must be justified by the operator, along with the use of any chemical that carries a substitution warning. The permit application should include a technical reason why the chemical is being used, what characteristics it has that result in the substitution warning, if there are ways to mitigate the environmental hazard and what alternatives have been considered (OCR, 2002, as amended 2011). Risk-based approach In addition to the submission of permit applications, the UK requires that the risks associated with the ongoing discharges of PW are assessed in a risk-based approach (RBA). Operators of installations that discharge PW are therefore required to conduct RBA assessments, in line with an OSPAR initiative (OSPAR Recommendation, 2012/5). RBA goes beyond the requirements of the HMCS and the simple CHARM model in requiring operators to take a holistic approach to the risk assessment of offshore discharges, which comprise not only offshore chemicals but also the naturally occurring substances that are present in PW. The sensitivity of the local environment must also be taken into consideration. These requirements have led to the UK adopting a four-tiered approach (DECC, 2014b), with the third and fourth tiers stipulating the use of dispersion modelling, using sophisticated tools such as the DREAM (Dose-related Risk and Effects Assessment Model) modelling software developed by The Foundation for Scientific and Industrial Research at the Norwegian Institute of Technology (SINTEF) in collaboration with a number of operators (2013). DREAM is a four-dimensional model which takes into consideration oceanographic data to predict the behaviour of each chemical upon discharge. The results are expressed in terms of volumes of PW in which the PEC>PNEC, and in terms of the percentage contribution that each chemical contributes to the risk. This allows key contributors to the risk to be identified and informs the PW management strategy. Listed chemicals For the OCNS scheme, several restriction lists apply. The most comprehensive of these is provided by REACH’s Annex XVII list of restricted substances. OSPAR specific lists are the OSPAR List of Chemicals for Priority Action and List of Substances of Possible Concern which have a specific focus on potential marine environmental impacts. The assessment of certain chemicals is truncated because they are considered benign. OSPAR provides a “green list” of chemicals referred to as the PLONOR list (OSPAR Agreement, 2013-06). These are permitted for use and discharge without the requirement for a formal risk assessment because they are considered to present a low risk to the marine environment. A measure of harmonization with REACH is provided by the consideration of Annex IV entries as PLONOR items. The same applies to certain categories of substances that are exempted from registration under Annex V of REACH. Challenges for the current scheme Chemical identification For simple, well-defined chemical substances identification is a facile task but many substances are challenging to describe. Less easily classified substances are often referred to as “substances of unknown or variable composition, complex reaction products and biological materials” or UVCBs for short. In terms of offshore chemicals, these are often distillates, polymers or complex reaction products. For some, typically polymers and distillates, the chemical structure may be well-defined, but the size of the molecule may vary. In other cases, stochastic reaction processes and heterogeneous starting materials can generate a range of products that are not described by a single chemical structure. To assist identification the OCNS scheme requires as much information as possible from suppliers. Unambiguous chemical descriptions and molecular weight are required as well as CAS and EC numbers if available. Since this information is often not readily available, characterization and therefore assessment of UVCBs remains a challenge. Substitution warning chemicals—challenges As described in Sections “The development of risk and hazard assessments for offshore chemicals in the North Sea” and “Permitting,” substitution warnings are applied to products that contain one or more substances with hazardous environmental properties. It is important to note that chemicals with a substitution warning should be replaced where possible but are not banned from use. While the number of substitution chemicals and their use has decreased since the inception of the UK National Plan, they are still being used and discharged in the North Sea (La Vedrine, 2015; Cefas-OCNS website, 2018b; OSPAR, 2018, described in the Supplementary Material). A challenge for substitution of chemicals is that the characteristics that give a chemical a substitution warning are often the properties most sought after for a specific function. For example, corrosion inhibitors are often surface-active substances with a low molecular weight—and therefore suspected bioaccumulative—as well as highly toxic, two criteria for substitution, yet products with these properties provide the best protection against microbial sources of corrosion because they adhere to surfaces protecting them longer. Similarly, chemicals that are highly persistent are often used in cement mixtures to increase the life or resilience of plugs and casings allowing an installation to be used for decades. The search for alternatives leads to the issue of potential “regrettable substitution” (Zimmerman and Anastas, 2015) i.e. that the environmental risk and impact of a non-substitutable chemical is increased compared with one carrying a substitution warning. Regrettable substitutions may include the use of a larger volume of chemical, or one which is highly toxic, and which causes a larger environmental risk in terms of the PEC:PNEC ratio (described in Section “Risk assessment”), than that of a substitutable chemical. Assessment challenges The current hazard and risk assessment strategies are based on standardized tests (Organisation for Economic Co-operation and Development; OECD) and established assessment criteria (PEC and PNEC). While these methods and assessment strategies are well researched and provide comparable results, they are not well suited to address issues outside the “classic” persistence, bioaccumulation potential, toxicity (PBT) realm. Issues that are not being addressed in the current assessment include endocrine disruptive properties, changes in hazard profiles due to sizes (nanoparticles), environmental hazards based on non-toxic properties (e.g. in case of plastics) and mixture effects. A step forward—the future of offshore chemical assessment in the North Sea Advances in scientific knowledge regarding the hazard and risk of chemicals, a movement towards decommissioning of installations and the continuous development of novel offshore chemicals with unknown environmental risks, challenge the established assessment methods. In this section, we present our view of methodological, research and regulatory needs and questions that should be addressed in order to ensure an adequate and sustainable assessment of offshore chemical use in the North Sea. Methodological/research needs The current processes in place have made it possible to establish a direct, credible and effective regulatory/advisory process for the management of offshore chemicals. However, with the novel challenges and recent events these current processes of legislation and advice may change at a UK, OSPAR, EU or even wider level. Challenges can come from many sources, such as the reduction in hydrocarbons due to well ageing and loss of viability, requiring the use of new chemistry to enhance oil recovery using polymers and/or fracturing. These novel techniques and materials can include chemical substances with new/unknown hazardous properties—not all of which can be captured by the current hazard and risk assessment process. Assessment and testing methods need to be reviewed and challenged to account for the potential environmental risk from novel chemical products (such as e.g. nanomaterials) or new identified environmental risks (e.g. endocrine disruption and chronic toxicity). For the UK and the Netherlands, Cefas has been tasked with this adaptation process by conducting research projects on behalf of BEIS and SSM. Regulatory research priorities include advice regarding plastics in offshore chemicals for potential regulatory measures to reduce marine plastic pollution, as well as on the potential leachate of endocrine disruptive materials (and how to test it). Longer range initiatives involving horizon scanning for potential issues is an ongoing and iterative process that Cefas and other agencies are constantly engaged in. Emerging issues are raised at international conferences and within academic and industry fora. The information gathered is assessed to provide informed decisions through projects including literature reviews, product reviews and experiments, to shape regulations to provide effective protection for the environment whilst providing time frames to allow industry to innovate and change their practices. However, innovation is not only needed with regards to novel contaminants but also in ensuring that current assessment and OECD testing methods represent the environmental risk based on the latest scientific information and for new processes such as decommissioning. Questions that need to be addressed include: Do the currently used LC/EC50 endpoints represent the environmental hazard of chemicals or are further criteria (e.g. endocrine disruptive potential) needed? ○ While currently acute toxicity is assessed on a substance basis for hazard assessment and later risk assessment other issues such as endocrine disruption is not. Novel methods for the identification and assessment of endocrine disruptive substances are urgently needed. Some of the needed novel methods include the identification of sources for endocrine disrupters. Could leachate from applications e.g. partition to the wastewater from wells and enter the marine environment? ○ Methods are needed to identify chemicals that are leached during the breakdown of plastics at extremely low levels and determined by specialist spectroscopic techniques. Emerging contaminants often need to be identified at extremely low levels to be able to potentially identify plastic sources and areas or particular plastic pollutants, and potentially trace the pollutant through time and the environment. Monitoring is critical in measuring the health of the marine environment. How can complex mixtures be characterized and assessed? What level of detail is needed to achieve a representative assessment? ○ Regulators are concerned with how chemicals made from complex mixtures can be characterized and grouped. This is both a pragmatic effort to ensure that data are not held separately for individual chemicals but also that the differences in toxicity results can be better understood. International collaborations to identify methods and provide guidelines for characterization of mixtures and how best to risk assess them are currently on going but need continuous engagement from regulators, academia and industry alike to be effective. If complex mixtures are not adequately characterized there is potential for unknown hazards and risks to go unnoticed. How can assessments be conducted with minimal or without need for animal testing? ○ QSARs can provide a surrogate for test data providing the source data are good quality and the methods employed are transparent. There are several freely available pieces of software that use transparent data sources and present it in such a way that it is compliant with both REACH and OSPAR requirements. However, stringent quality control of the derived data and clear information regarding the applicability domain and limitations of the individual methods are paramount for ensuring the regulatory acceptance of animal-alternative methods such as QSARs. What are the lessons learned from the current legislation of offshore chemicals and which improvements can be made—especially with regards to decommissioning? ○ Over the last few years, decommissioning of installations in the North Sea has been increasing. Since many of the wells were developed prior to the current chemical legislation, risk assessments must be made on the historic well contents from a data poor perspective. Chemicals in use prior to the OSPAR regulations may or may not be recorded and there may be little information regarding the composition. In these cases, the regulator may have to rely on the name of the chemical to identify what type of chemical could be present. Due to the lack of information, risk assessments can often only be conducted qualitatively based on similar types of chemicals and likelihood of substitution warnings and general risk. The INSITE initiatives phase I was created to address some of these questions to fully understand the effects of man-made structures and how these installations could be evaluated in the context of decommissioning practices (https://www.insitenorthsea.org/). This targeted research has provided the evidence to support regulators and industry alike, but the development of internationally accepted regulatory guidelines remains a challenging and time-consuming task. Assessment strategies for new activities Decommissioning of offshore oil and gas facilities in the North Sea is likely to gain momentum over the next 30 years (Royal Academy of Engineering, 2013). In the UK, decommissioning activities are predicted to take place on 214 fields between 2017 and 2025 and cost approximately £17 billion. Around 5500 km of pipeline and 1624 wells are forecast to be abandoned during that period while 98 platforms will need to be removed (Oil and Gas UK, 2017). Overall, over 3000 pipelines and approximately 5000 wells will require decommissioning (Oil and Gas UK, 2016) which must be carried out according to the requirements specified in the Petroleum Act 1998 (https://www.gov.uk/guidance/oil-and-gas-decommissioning-of-offshore-installations-and-pipelines). In addition, to comply with the Offshore Chemical Regulations 2002 (as amended 2011), these operations will be captured in the well intervention and decommissioning permit applications submitted to BEIS by the offshore oil and gas industry. There are several challenges relating to carrying out risk assessments of chemicals discharged during decommissioning operations. One of the biggest challenges is the evaluation of the risk presented by legacy chemicals that were first used prior to the introduction of the OCNS scheme and are therefore often not registered. Legacy chemicals are e.g. retained in a well from the point at which it was drilled, but must be handled (and often discharged) in order for the well to be decommissioned. While such chemicals do not appear on the permit application, they still need to be assessed. The current approach is, to use a contemporary surrogate chemical on which the risk assessment can be carried out. There are limitations to this approach because operators will often buy a bespoke chemical “package” from a specific chemical supplier who is not necessarily the producer for all the chemical components. This might mean that the chemical supplier has no information regarding the exact chemical composition of a legacy product or what could be a suitable modern equivalent for the risk assessment. Issues arise when surrogate chemicals are chosen inappropriately e.g. if the choice includes PLONOR listed substances instead of chemicals that are more hazardous. To mitigate this problem, the formulation of a legacy chemical is compared with that of a contemporary surrogate and it is identified where inappropriate surrogate chemicals have been used. This information is used to give advice regarding a chemicals substitution status and to provide a fuller picture of the risks associated with legacy chemicals. Regulatory needs REACH and OSPAR—how can the two legislative frameworks be harmonized? As detailed in Section “The development of risk and hazard assessments for offshore chemicals in the North Sea,” the use and discharge of chemicals in the North Sea are currently regulated by the Harmonised Mandatory Control Scheme developed by OSPAR (OSPAR Decision, 2000/2), as well as the EU REACH Regulation (European Commission, 2013). Since registration and assessment processes are time consuming for both the industry and the regulator, efforts are being made to harmonize HMCS with REACH as much as possible. Some ideas even go as far as to abandon the HMCS and rely fully on the REACH registration for the assessment of offshore chemicals. At a first glance, this idea seems attractive, as it has the potential to remove laborious and, apparently, duplicate assessments. However, the HMCS risk assessment focuses specifically on the risk for the marine environment, whereas REACH looks at the risk for human and environmental health in general. Furthermore, chemical substances with a production/import volume below a certain tonnage limit, as well as polymers, are exempt from REACH registration, while they must be registered and assessed under the HMCS. There is therefore concern that an assessment solely under REACH could reduce the current level of marine environmental protection. This concern was reinforced by the results of a recent study (Anderson et al., 2018). Here we compared the assessment results from offshore chemicals registered under the HMCS scheme with their assessments under REACH and found that the vast majority of assessments either could not be compared, because of the exemption criteria under REACH (Figure 2) or did not result in the same conclusions (Figure 3). Figure 2. Open in new tabDownload slide Substances meeting the criteria for concern under HMCS-registration status under REACH. Figure 2. Open in new tabDownload slide Substances meeting the criteria for concern under HMCS-registration status under REACH. Figure 3. Open in new tabDownload slide Comparison of assessment results under REACH and HMCS. Figure 3. Open in new tabDownload slide Comparison of assessment results under REACH and HMCS. Particularly the exemption of polymers under REACH is an obstacle for a harmonization with HMCS assessments, because many polymers are considered as substitutable substances under HMCS due to their high persistence. The difficulties surrounding both polymers and substitution criteria are discussed in the following sections, along with a further issue, which concerns the assessment factors used to calculate the RQ (HMCS) or risk characterization ratio (REACH). Assessment factors As described in Section “Permitting,” the key parameter that is used to determine chemical risk is the RQ, which becomes a concern when the PEC of the chemical exceeds the predicted no-effect concentration (PNEC). The latter figure is derived from estimated or measured ecotoxicity values, suitably adjusted by an assessment factor that provides a measure of environmental protection based on the level of uncertainty that exists in extrapolating the (usually lab-based) ecotoxicity data to a real-world scenario. However, different assumptions have been applied in the development of the CHARM model chosen by OSPAR to those made by the European Union for REACH (ECHA, 2008), and this has led to a difference of (typically) two orders of magnitude in the assessment factors computed in each case. As a result, chemical risk assessments conducted under REACH lead to risk characterization ratios that are higher by the same proportion than those generated in accordance with CHARM. This presents a dilemma for the offshore industry, since the conditions that have been used to permit the use of large numbers of chemicals under the OSPAR framework for many years will be regarded as inconsistent with safe use under REACH. The issue is the subject of current debate between industry groups and regulators, and OSPAR’s Offshore Industry Committee agreed in March 2019 to convene an intersessional correspondence group (ICG-REACH) to address this issue, the outcome of which is expected to be announced in 2020. Polymers/plastics: substances of low concern? Under REACH, polymers are exempted from registration and evaluation since they are considered to represent a low concern based on their high molecular weight (ECHA, 2012). However, whilst their molecular weight may restrict their ready uptake into organisms via transport across cell membranes, the increasing prevalence in the marine environment of plastics and microplastics can present additional risks (GESAMP, 2015,, 2016; UNEP, 2016). These range from the potential for physical harm to marine mammals from large plastic fragments (UNEP Resolution 1/6, 2014), to the ingestion of microplastics (House of Commons, 2016) by a wider range of marine animals. The persistence of plastics in the marine environment (Lithner et al., 2011) exacerbates the problem, whilst there are additional concerns created by the ability of plastics to adsorb other (potentially toxic) marine pollutants, thus facilitating their ingestion (Engler, 2012; DECC, 2014a). It must be stressed that the main source of such materials in the oceans is litter, with the discharge of offshore chemicals representing very minor quantities by comparison (OSPAR, 2017). However, as international concern over marine plastics mounts, the offshore industry faces increasing pressure to minimize its own contribution to a global environmental problem. Under the HMCS framework, plastics must be registered and are expected to be classified as substitutable based on their persistence, thereby facilitating regulatory control, but such action under REACH is hindered by the lack of registration data. This limitation does not prevent the regulatory control of plastics under REACH, since all substances potentially fall under the scope of the Restriction process. However, the progress of REACH in this area is currently limited, with a recent proposal under development for the restriction of microplastics (ECHA, 2019). Substitution criteria It should be noted that harmonization of REACH and OSPAR HMCS assessment of plastics would not be achieved by simply removing the exemption currently applied to polymers by ECHA. The reason for this is that although the OSPAR pre-screening scheme is based on the same elements of biodegradability, bioaccumulation potential and ecotoxicity that are used to highlight hazardous substances under the REACH Regulation, the scheme identifies persistence alone as sufficient qualification for a substitution warning. In contrast, REACH requires the existence of additional ecotoxicological issues to trigger the equivalent classification (PBT or vPvB). In addition, different thresholds for concern are employed under the two frameworks, with the overall outcome of far fewer substances being identified as candidates for substitution under REACH than under the OSPAR HMCS (Anderson et al., 2018, Figure 2). As a result, any moves towards the abandonment of OSPAR’s definition of hazardous substances in favour of the equivalent REACH criteria would risk the re-introduction of numerous substances already phased out under the OSPAR framework, along with jeopardizing the investment made by the chemical industry in the development of greener chemicals to replace them. To avert this dilemma, and impart a measure of harmonization, in theory additional restrictions could be imposed under REACH, to as noted in the context of plastics. However, this may not be a practical option given the number of substances involved, hence no immediate solution to this problem is in sight. Can sophisticated location-specific modelling approaches (e.g. DREAM) be accommodated within REACH (and OSPAR) frameworks? From the foregoing discussion, it is evident that REACH still has some way to go before the HMCS can be considered to be redundant. A further complication in a post-HMCS scenario would be the status of one of the most significant developments introduced by OSPAR’s Offshore Industry Committee in recent years, the RBA (see Section “Risk-based approach”). Under REACH, legitimate use of chemical substances relies upon the production of safety data sheets that include annexed Exposure Scenarios that cover the intended uses of the substance. These are required to demonstrate that if the prescribed conditions are followed, those uses will not present a risk to the environment (or human health). However, these Exposure Scenarios are essentially generic in nature (the use of CHARM is suggested; ECHA, 2016a, b), and make no allowance for site-specific parameters that could influence the risk. To address these, individual operators would need to produce their own Exposure Scenarios to demonstrate safe use. The use of such parameters is however also accommodated within the higher tier models (e.g. DREAM) that are a feature of UK RBA assessments. This factor may offer a future for the RBA initiative within REACH: however, in assessing the overall chemical risk of both synthetic and naturally occurring substances together, RBA has already taken a step beyond the requirements of both REACH and the HMCS, and for time being, it is more likely to remain as a stand-alone OSPAR initiative. Brexit On 23 June 2016, the British public voted to leave the European Union in a historic referendum. At the time of writing, the terms of the UK’s future relationship with the EU have yet to be finalized, although it is known that the UK will be excluded from the REACH Regulation, which has led to the UK announcing that it will impose the measures of REACH on a national basis. Conversely, the UK’s participation in OSPAR (as a non-EU body) will be unaffected. The ongoing debate with regards to Brexit still provides certain uncertainty and once the full situation has reached its closure, the new challenges associated with Brexit will have to be distil with regards to the uses of the UKEEZ. Conclusions The regulatory assessment of environmental risk from offshore chemicals in the UK (and OSPAR region) is based on methods and strategies that have been developed and used over decades. The established criteria and methods are a robust foundation for environmental risk assessment and protection. However, newly gained understanding on hazardous properties of plastics, nanomaterials as well as the continuous development of chemicals with novel properties (nanomaterials and emerging contaminants) call for a re-evaluation of the established criteria and a development of new ways of addressing challenges ranging from the assessment of endocrine disruptive properties, decommissioning (potential release of legacy chemicals), to the harmonization of international regulatory frameworks. In this article, we outlined some of the challenges we have encountered as a group of regulators and scientists within the UK regulatory framework. Furthermore, we described our strategies and views of how to address these challenges. Much work still needs to be done to address the old and emerging risks posed by offshore chemicals to the marine environment. To address these, we need new methods and assessment criteria, but, most importantly, more cross-disciplinary communication and collaboration of academia, regulators and industry to drive innovation and understanding of marine environmental protection. Acknowledgements This compilation was funded by The Centre for Environment, Fisheries and Aquaculture Science (Cefas) Corporate Funds (40101B-Learning and Development code) to support colleagues of the Offshore Chemical Notification Scheme (ONCS) team to produce this opinion document. References Anderson G. , Rowles R. , Supple S. , Phillips C. , Sühring R. 2018 . Comparison of substitution status of chemical substances under REACH and OSPAR legislation . Environmental Impact Assessment Review , 72 : 43 – 49 . Google Scholar Crossref Search ADS WorldCat Bascietto J. , Hinckley D. , Plafkin J. , Slimak M. 1990 . Ecotoxicity and ecological risk assessment. Regulatory applications at EPA. Part 1 . Environmental Science & Technology , 24 : 10 – 15 . Google Scholar Crossref Search ADS WorldCat Cefas-OCNS website. 2018a . https://www.cefas.co.uk/cefas-data-hub/offshore-chemical-notification-scheme/ (last accessed August 2018). Cefas-OCNS website. 2018b . https://www.cefas.co.uk/media/52537/uk-national-substitution-plan-5-dec-2010.pdf (last accessed August 2018). CIN (CHARM Implementation Network). 2004 . CHARM—Chemical Hazard and Risk Management User Guide Version 1.4. https://eosca.eu/wp-content/uploads/2018/08/CHARM-User-Guide-Version-1-5.pdf (last accessed August 2018). Craig J. , Gerali F. , MacAulay F. , Sorkhabi R. 2018 . History of the European Oil and Gas Industry. Geological Society, London, Special Publications, 465. DECC. 2014a . Guidance Notes: The Offshore Petroleum Activities (Oil Pollution Prevention and Control) Regulations 2005 (as amended). Department for Energy and Climate Change, UK. 47 pp. DECC. 2014b . Department of Energy and Climate Change. The United Kingdom Risk Based Approach (RBA) Implementation Programme. Department for Energy and Climate Change, UK. ECHA. 2008 . Guidance on Information Requirements and Chemical Safety Assessment Chapter R.10: Characterisation of Dose [Concentration]-Response for Environment, Section R.10.3.2.3. https://echa.europa.eu/documents/10162/13632/information_requirements_r10_en.pdf/bb902be7-a503-4ab7-9036-d866b8ddce69 (last accessed August 2018). ECHA. 2012 . Guidance for Monomers and Polymers, Version 2.0, 8 pp. https://echa.europa.eu/documents/10162/23036412/polymers_en.pdf/9a74545f-05be-4e10-8555-4d7cf051bbed (last accessed 23 August 18). ECHA. 2016a . Guidance on Information Requirements and Chemical Safety Assessment. Chapter R.16: Environmental Exposure Assessment, Version 3.0, 156. https://echa.europa.eu/documents/10162/13632/information_requirements_r16_en.pdf/b9f0f406-ff5f-4315-908e-e5f83115d6af (last accessed 24 August 2018). ECHA. 2016b . Report on the Operation of REACH and CLP 2016, 99. https://echa.europa.eu/documents/10162/13634/operation_reach_clp_2016_en.pdf (last accessed August 2018). ECHA. 2019 . Annex XV Restriction Report: Proposal for a Restriction on Intentionally Added Microplastics. https://echa.europa.eu/documents/10162/12414bc7-6bb2-17e7-c9ec-652a20fa43fc (last accessed May 2019). Engler R. E. 2012 . The complex interaction between marine debris and toxic chemicals in the ocean . Environmental Science and Technology , 46 : 12302 – 12315 . http://pubs.acs.org/doi/abs/10.1021/es3027105 Google Scholar Crossref Search ADS PubMed WorldCat European Commission. 2013 . General Report on REACH: Report from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions. COM/2013/49 final. European Commission, Brussels. 15 pp. GESAMP. 2015 . Report 90: Sources, Fate and Effects of Microplastics in the Marine Environment: A Global Assessment, 50, 46, 19, 65, 64. http://ec.europa.eu/environment/marine/good-environmental-status/descriptor-10/pdf/GESAMP_microplastics%20full%20study.pdf (last accessed August 2018). GESAMP. 2016 . Report 93: Sources, Fate and Effects of Microplastics in the Marine Environment: Part Two of a Global Assessment, 14. http://www.gesamp.org/site/assets/files/1275/sources-fate-and-effects-of-microplastics-in-the-marine-environment-part-2-of-a-global-assessment-en.pdf (last accessed August 2018). House of Commons. 2016. Report: Environmental Impact of Microplastics. Fourth Report of Session 2016–17. https://publications.parliament.uk/pa/cm201617/cmselect/cmenvaud/179/179.pdf (Last accessed August 2018). La Vedrine M. A. G. , Sheahan D. A. , Gioia R. , Rowles R. S. , Kroeger S. , Phillips C. , Kirby M. F. 2015 . Substitution of hazardous offshore chemicals in UK waters: an evaluation of their use and discharge from 2000 to 2012 . Journal of Cleaner Production , 87 : 675 – 682 . Google Scholar Crossref Search ADS WorldCat Lithner D. , Larsson A. , Dave G. 2011 . Environmental and health hazard ranking and assessment of plastic polymers based on chemical composition . Science of the Total Environment , 409 : 3309 – 3324 . Google Scholar Crossref Search ADS PubMed WorldCat Offshore Chemicals (Amended) Regulations. 2011 . http://www.legislation.gov.uk/uksi/2011/982/pdfs/uksi_20110982_en.pdf (last accessed August 2018). Oil and Gas UK. 2016 . Environment Report 2016. 60 pp. https://oilandgasuk.co.uk/wp-content/uploads/2016/11/Environment-Report-2016-Oil-Gas-UK.pdf (last accessed 16 August 2018). Oil and Gas UK. 2017 . Decommissioning Insight 2017. https://oilandgasuk.co.uk/wp-content/uploads/2017/11/Decommissioning-Report-2017-27-Nov-final.pdf (last accessed August 2018). Oslo. 1972 . Convention for the Prevention of Marine pollution by Dumping from Ships and Aircrafts. United Nations Treaty Series, Oslo. OSPAR Agreement 2013 -06. OSPAR List of Substances Used and Discharged Offshore which are Considered to Pose Little or No Risk to the Environment (PLONOR). https://www.cefas.co.uk/media/1384/13-06e_plonor.pdf (last accessed August 2018). OSPAR Convention. 1992 . Convention for the Protection of the Marine Environment of the North-East Atlantic. https://www.ospar.org/site/assets/files/1290/ospar_convention_e_updated_text_in_2007_no_revs.pdf (last accessed 21 August 2018). OSPAR Decision 2000 /2 on a Harmonised Mandatory Control System for the Use and Reduction of the Discharge of Offshore Chemicals (as amended by OSPAR Decision 2005/1) http://www.ospar.org/documents? d=32742 (last accessed May 2016). OSPAR Recommendation 2012 /5 for a Risk-Based Approach to the Management of Produced Water from Offshore Installations. https://www.ospar.org/documents? d=32922 (last accessed October 2018). OSPAR Recommendation 2017/1 on a Harmonised Pre-screening Scheme for Offshore Chemicals. https://www.ospar.org/documents? d=37431 (last accessed 24 October 2018). OSPAR Summary Report of Meeting. 2006 . Offshore Industry Committee, Copenhagen, Denmark, 20–24 February 2006, item 2.26. https://www.ospar.org/meetings/archive?q=summary&a=7458&y=1985&s=4 (last accessed 24 October 2018). OSPAR Summary Report of Meeting. 2017 . Offshore Industry Committee, Oslo, Norway, 15–17 March 2017, item 3.18. https://www.ospar.org/meetings/archive/offshore-industry-committee-675 (last accessed 23 August 2018). OSPAR. 2018 . Assessment of Discharges, Spills and Emissions from Offshore Oil and Gas Operations on the United Kingdom Continental Shelf, 2012–2016, OSPAR, Publication Number: 718. Paris Convention. 1974 . Prevention of Marine Pollution from Land Based Sources (as amended by the Protocol of 26 March 1986). REACH Ready website. http://www.reachready.co.uk/documents/REACHBasicsSummaryv3Jun13.pdf (last accessed August 2018). Royal Academy of Engineering. 2013 . Decommissioning in the North Sea. http://www.raeng.org.uk/publications/reports/decommissioning-in-the-northsea (last accessed August 2018). SINTEF. 2013 . MEMW 6.5 User Guide: SINTEF Marine Environmental Technology. MEMW 6.5 User’s manual. www.sintef.no. United Nations Treaty Series. 1974 . Convention for the Prevention of Marine Pollution by Dumping from Ships and Aircraft. https://treaties.un.org/doc/Publication/UNTS/Volume%20932/volume-932-I-13269-English.pdf (last accessed August 2018). UNEP. 2016 . Marine Plastic Debris and Microplastic: Global Lessons and Research to Inspire Action and a Guide to Policy Change, 48. https://wedocs.unep.org/rest/bitstreams/11700/retrieve (last accessed August 2018). UNEP Resolution 1/6. 2014. On ‘Marine Plastic Debris and Microplastics’. https://wedocs.unep.org/bitstream/handle/20.500.11822/17562/Marine%20Plastic%20Debris%20and%20Microplastic%20Technical%20Report%20Advance%20Copy%20Ann.pdf? sequence=2&isAllowed=y (last accessed August 2018). Zimmerman J. B. , Anastas P. T. 2015 . Toward substitution with no regrets . Science , 347 : 1198 – 1199 . Google Scholar Crossref Search ADS PubMed WorldCat © Crown copyright 2019. This article contains public sector information licensed under the Open Government Licence v3.0 (http://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/)

Droghini,, A;Fischbach, A, S;Watson, J, T;Reimer, J, P

doi: 10.1093/icesjms/fsaa014pmid: N/A

Abstract Minimal vessel traffic and cold water temperatures are believed to limit non-indigenous species (NIS) in high-latitude ecosystems. We evaluated whether suitable conditions exist in the Bering Sea for the survival and reproduction of NIS. We compiled temperature and salinity thresholds of NIS and compared these to ocean conditions projected during two study periods: recent (2003–2012) and mid-century (2030–2039). We also explored patterns of vessel traffic and connectivity for US Bering Sea ports. We found that the southeastern Bering Sea had suitable conditions for the year-round survival of 80% of NIS assessed (n = 42). This highly suitable area is home to the port of Dutch Harbor, which received the most vessel arrivals and ballast water discharge in the US Bering Sea. Conditions north of 58°N that include sub-zero winter water temperatures were unsuitable for most NIS. While mid-century models predicted a northward expansion of suitable conditions, conditions for reproduction remained marginal. Only 40% of NIS assessed (n = 25) had 6 or more weeks where conditions were suitable for reproduction. Our findings illustrate the potential vulnerability of a commercially important subarctic ecosystem and highlight the need to consider life stages beyond adult survival when evaluating limits to NIS establishment. Introduction Non-indigenous species (NIS) become invasive by passing through three stages: introduction, establishment, and spread (Blackburn et al., 2011). Aquatic NIS are often transported by vessels via ballast water or by fouling wetted surfaces such as hulls and anchors (Molnar et al., 2008). Thus, ports that receive high levels of vessel traffic are typically more vulnerable to biological introductions, though additional factors such as the amount of ballast water discharged and voyage duration also influence introduction risk (Verling et al., 2005; Seebens et al., 2013). After arriving at a location, NIS must survive and reproduce to establish a self-sustaining population (Blackburn et al., 2011). The final stage of the invasion process, in which NIS become “invasive”, happens when it disperses to and establishes a population at a new site (Blackburn et al., 2011). Most NIS never reach this endpoint, yet those that do can have large impacts on native biodiversity, commercial and recreational fisheries, and subsistence resources (Molnar et al., 2008). Because of the high potential costs of NIS, it is important to understand the factors that contribute to the success or failure of invasion and which areas are most at risk. Water temperature and salinity are among the primary factors that dictate the survival and reproduction of aquatic NIS (Iacarella et al., 2015; Dijkstra et al., 2017; Paiva et al., 2018). These variables directly affect an organism’s physiology, and conditions outside specific thresholds disrupt development, behaviour, metabolism, and ultimately, survival (Pörtner, 2001; Kassahn et al., 2009; Monaco and Helmuth, 2011). Since physiological thresholds for reproduction and development are typically narrower than adult survival (Pörtner, 2001; Woodin et al., 2013), predicting species’ distributions can be strengthened by considering physiological requirements across life-history stages (Kearney and Porter, 2009; Kingsolver et al. 2011; Woodin et al., 2013; Levy et al. 2015). Compared with most marine ecosystems, high-latitude ecosystems receive relatively little vessel traffic and experience cold water temperatures nearly year-round. While these factors may explain the low rates of NIS that have been reported (Ruiz and Hewitt, 2009; de Rivera et al., 2011; Chan et al., 2019), NIS are being transported by vessels from temperate to Arctic systems and are surviving the voyage (Chan et al., 2014; Ware et al., 2014; Chan et al., 2016). Modelling studies suggest that some NIS could survive year-round in high-latitude marine systems (de Rivera et al., 2011; Ware et al., 2016; Goldsmit et al., 2018), though few studies have considered the critical role of reproduction (but see de Rivera et al., 2007; Ware et al., 2014). These studies also predict that conditions will become increasingly suitable for NIS establishment (de Rivera et al., 2011; Ware et al., 2016; Goldsmit et al., 2018). High-latitude systems are already experiencing rapid oceanographic and socioeconomic changes that may facilitate NIS introduction and establishment. Warming temperatures have led to drastic declines in sea ice extent and thickness, and the length of the ice-covered season (Onarheim et al., 2018). These warmer conditions are expected to favour NIS whose survival or reproduction is currently limited by cold water temperatures (de Rivera et al., 2011; Ware et al., 2016; Goldsmit et al., 2018) and have been linked to the recent establishment of NIS in several high-latitude systems (Ruiz and Hewitt, 2009; Chan et al., 2019). Ocean warming is also expected to favour NIS introductions through increased commercial vessel traffic, industrial development, and tourism (Ruiz and Hewitt, 2009; Miller and Ruiz, 2014). Here, we explore the first two stages of the invasion process as they relate to the Bering Sea, a high-latitude and commercially important marine ecosystem that serves as a gateway to the Arctic. Our primary interest was to evaluate whether ocean temperature conditions in the Bering Sea were suitable for the survival and reproduction of NIS and whether forecasted oceanographic changes may facilitate NIS establishment. We compared species’ temperature and salinity thresholds with conditions of the Bering Sea under two study periods, recent (2003–2012) and mid-century (2030–2039). As a secondary interest, we examined patterns of vessel traffic and ballast water discharge for US ports in the Bering Sea. We hypothesized that recent ocean temperatures limit the survival and reproduction of most NIS. Given projections for increasing sea temperatures and receding sea ice by mid-century, we hypothesized that thermal barriers to NIS survival and reproduction would be reduced in the upcoming decades. Material and methods Study area The Bering Sea is a transition between subarctic and arctic marine systems that lies between the temperate North Pacific Ocean and the Arctic waters of the Chukchi Sea (Figure 1). It spans 15° of latitude from 51°N to 66°N and is characterized by strong latitudinal gradients in water temperature and sea ice formation. The southern Bering Sea is typically ice-free year-round (Stabeno et al., 2012). Meanwhile, the northern Bering Sea has historically been defined by the presence of seasonal sea ice, which reaches its maximum in February or March (Stabeno et al., 2007, 2012). The extent of sea ice in the spring, and the timing of its retreat, affects the timing of the phytoplankton bloom and the extent of a “cold pool”, an area of cold (<2°C) bottom water that is created by the melting of sea ice (Stabeno et al., 2007). Figure 1. Open in new tabDownload slide The Bering Sea extends from the Aleutian Islands north to the Bering Strait and from the Russian Far East west to the US state of Alaska. It is situated between the North Pacific Ocean and the Chukchi Sea and acts as a transition between subarctic and arctic systems. The southeastern Bering Sea’s continental shelf has some of the highest levels of marine productivity in the world and supports some of the largest commercial fisheries in the US sea ice extent from the National Snow and Ice Data Center: https://nsidc.org/data/seaice_index/archives. Figure 1. Open in new tabDownload slide The Bering Sea extends from the Aleutian Islands north to the Bering Strait and from the Russian Far East west to the US state of Alaska. It is situated between the North Pacific Ocean and the Chukchi Sea and acts as a transition between subarctic and arctic systems. The southeastern Bering Sea’s continental shelf has some of the highest levels of marine productivity in the world and supports some of the largest commercial fisheries in the US sea ice extent from the National Snow and Ice Data Center: https://nsidc.org/data/seaice_index/archives. Because most NIS inhabit nearshore or otherwise shallow waters (Ruiz et al., 2015), we restricted our analyses to the continental shelf (waters shallower than 200 m) and to the top 40 m of the water column, where the warmest temperatures occur (see Figure 5 in Hermann et al., 2016). Most of the continental shelf is shallow, with >60% of the shelf <60-m deep. The wide, southeastern continental shelf is of high global and regional value: it has some of the highest levels of marine productivity in the world and supports a USD 1 billion commercial fishing industry (NMFS, 2017). Only four animal NIS have been reported to date (Ruiz et al., 2006; Fofonoff et al., 2018): American shad (Alosa sapidissima), Atlantic salmon (Salmo salar), Japanese skeleton shrimp (Caprella mutica), and soft-shell clam (Mya arenaria). Of these, only the Japanese skeleton shrimp and the soft-shell clam have established populations (Ashton et al., 2008). Defining taxa-specific physiological thresholds We compiled a list of animal NIS from records in the National Exotic Marine and Estuarine Species Information System (Fofonoff et al., 2018) and the Nonindigenous Aquatic Species Database (Fuller and Benson, 2013). Because we could not feasibly consider all NIS, we assumed that NIS that were geographically close to the Bering Sea were more likely to survive the voyage and to become established because of their ability to tolerate similar environmental conditions. We, therefore, considered the 25 NIS that have been reported within two marine ecoregions of the Bering Sea, i.e. from the Bering Sea south to northern British Columbia and the Sea of Okhotsk; Spalding et al. (2007), as well as a subset of 24 NIS that occur in the third nearest ecoregions. This subset included all NIS reported from the Sea of Japan/East Sea in the northwest Pacific, and all NIS found along the northeast Pacific coast south to the mouth of Columbia River (on Washington–Oregon border), including major shipping hubs (Victoria, BC; Puget Sound, WA; and Astoria, OR). Beginning with this list of 49 taxa, we searched peer-reviewed publications, reports, and electronic databases to identify taxa-specific temperature (T) and salinity (S) thresholds for survival and reproduction. We explored temperature and salinity thresholds because they represent two primary physiological barriers that are vital aspects of environmental suitability for NIS establishment (Kassahn et al., 2009; Monaco and Helmuth, 2011; Iacarella et al., 2015; Paiva et al., 2018). Because brackish habitats are extremely limited in the Bering Sea, we only included taxa that could survive in salinities ≥30 parts per thousand (ppt). Anadromous NIS (n = 4) were only included in the survival component of our analyses. We prioritized T–S thresholds obtained from experimental studies but used thresholds inferred from geographic distributions if no other data were available. We defined survival thresholds as the absolute minimum and maximum values reported for that taxon, irrespective of life stage and geography. In contrast, reproductive thresholds represent the narrowest T–S range required for either sexual reproduction or ontogenetic development and growth. If multiple reproductive thresholds were available (e.g. different thresholds for spawning and metamorphosis), we chose the thresholds associated with the least tolerant life stage. Our models required information on minimum and maximum thresholds for both temperature and salinity. This information was unavailable for many taxa; however, we retained taxa with incomplete data in the following cases. For taxa that have been observed in temperatures that exceeded the Bering Sea maximum (∼17°C), but for which no maximum temperature threshold was reported, we assigned an arbitrary maximum temperature value that was greater than the maximum temperature of the Bering Sea. This step ensured that the taxon would be included in our analysis and that its upper-temperature threshold would not be the component that limited its predicted suitability. Similarly, for taxa that had no available salinity thresholds but that had been reported from waterbodies with salinities comparable to those of the Bering Sea, we assigned salinity values equal to average seawater (31–35 ppt), which conferred salinity survival to 99% of the Bering Sea shelf. Modelling ocean temperature and salinity conditions We derived T–S conditions of the Bering Sea from Regional Ocean Modeling Systems (ROMS) that were generated by downscaling one of three general circulation models: (i) CGCM3-t47, (ii) ECHO-G, and (iii) MIROC3.2 (Hermann et al., 2016). These models were chosen for downscaling because of their ability to satisfactorily represent mean conditions in the Bering Sea (Hermann et al., 2016). The ROMS predict ocean conditions using algorithms specific to the Bering Sea for physical processes, such as wind, solar radiation, sea ice dynamics, and freshwater input. They provide weekly T–S values for the entire Bering Sea and the northeastern Pacific Ocean as a square gridded surface with a resolution of 6 nautical miles (nmi), or ∼10 km. T–S values were available for several depth intervals (roughly every 10–300 m; Hermann et al., 2016) and every week of the year from 2003 to 2040. Rather than to evaluate T–S suitability at every available depth layer, we collapsed the depth dimension at each pixel into a single value by taking the maximum T–S values reported at that location. To evaluate changes in T–S suitability over time, we considered two 10-year periods: recent (2003–2012) and mid-century (2030–2039). Considering a decadal study period allowed us to incorporate model uncertainty inherent from annual fluctuations in Bering Sea conditions. Because each model uses different algorithms to predict ocean and atmospheric conditions, the T–S values they predict are different as well. For example, when applied to the Bering Sea, the ECHO-G model generally predicts the coldest ocean temperatures, while the CGCM3-t47 model predicts the warmest (Hermann et al., 2016). Nevertheless, all models predict an increase in sea surface temperature and a decrease in sea ice by 2039 (Hermann et al., 2016). Averaging across models is a commonly used technique to account for forecast uncertainties and minimize model-specific biases while retaining consensus patterns across the models. We present model-specific results in “Results” section but focus on these consensus patterns in the figures and “Discussion” section. Interested readers may consult Hermann et al. (2016) for an in-depth treatment of each model’s performance relative to past Bering Sea conditions. Analysing temperature and salinity suitability for survival and reproduction We modelled T–S suitability for: (i) year-round survival, (ii) weekly survival, and (iii) reproduction. Models were built separately for each taxon-ROMS-study period combination. Suitability across all taxa was determined by summing the number of taxa with suitable conditions in that pixel. Figures were generated for each study period by averaging these cumulative taxa results across the three ROMS. We also examined changes in suitable conditions by subtracting mid-century suitability results from recent ones. Positive change values indicate that more taxa are predicted to have suitable conditions by mid-century. Analyses were conducted in R Statistical Software version 3.3.2 (R Core Team, 2018) with support from the following packages: circlize (Gu et al., 2014), doSNOW (Microsoft Co. and Weston, 2017), ncdf4 (Pierce, 2017), plyr (Wickham, 2011), raster (Hijmans, 2017), rgdal (Bivand et al., 2018), rgeos (Bivand and Rundel, 2017), sp (Bivand et al., 2013), and tidyverse (Wickham et al., 2019). Our R code is available: https://doi.org/10.5281/zenodo.3546376 Suitability of temperature and salinity for year-round survival For each taxon, we defined a pixel as “suitable” if the pixel’s T–S values remained within the taxon’s T–S survival thresholds for all weeks of a given year. Within each 10-year study period, we classified the pixel as “suitable year-round” if it was suitable for at least 7 of the 10 years. We chose this threshold because we wanted to focus on taxa that could establish self-sustaining populations, rather than taxa that could only survive in anomalous years. Seasonal and latitudinal patterns of survival suitability We modelled survival on a weekly basis to investigate the spatial and temporal patterns of NIS survival. For each week, a taxon was considered to have suitable survival conditions if at least one pixel had T–S values within the taxon’s T–S survival range. As with our yearly survival analysis, we classified each “pixel-week” as suitable if the pixel was suitable for that week for at least 7 of the 10 years in a study period. We then summarized these results by latitude. We generated 16 1° latitudinal bands (from 51°N to 66°N) in ESRI ArcMap 10.5.1, by transforming Natural Earth’s 1° graticules lines (https://www.naturalearthdata.com/downloads/10m-physical-vectors/10m-graticules/) into polygons and manually correcting for errors that occurred when converting lines to polygons across the antimeridian. If a taxon had at least one pixel in the 1° latitudinal band that was suitable, we classified the entire latitude as suitable for that week. We summarized results across taxa by summing survival values for each week and each 1° latitude. Suitability of temperature and salinity for reproduction For each taxon, we defined a pixel as “suitable” if its T–S values were within the taxon’s T–S reproductive thresholds. For each year within our 10-year study periods, we calculated the number of consecutive weeks of suitable reproductive conditions, such that pixel values could range from 0 to 52. We used the average number of consecutive weeks within each study period to describe our results. For our figure, we displayed spatial patterns of suitable reproductive conditions by considering whether taxa had at least 1 week of suitable reproductive conditions. Doing so allowed us to express our findings as the average number of NIS per pixel, consistent with our other figures. Describing potential introduction vectors: commercial vessel traffic We analysed vessel traffic and ballast water discharge in the Bering Sea using two datasets: the National Ballast Information Clearinghouse (NBIC) and the National Marine Fisheries Service (NMFS) Vessel Monitoring Systems (VMS). These data allowed us to quantify the magnitude and spatial patterns of traffic arriving at US ports in the Bering Sea for fishing vessels and other large commercial vessels. First-order port connections were summarized based on the number of trips that originated in a particular origin region and ended (or discharged ballast water) in a particular destination port. NBIC data are publicly available (https://invasions.si.edu/nbic/search.html) and report vessel landings and their ballast water activities. Ballast water exchanges are reported when entering any port in the United States (33 CFR §§ 151). Because regulations have changed in the last decade, we only considered the three most recent, complete years (1 January 2014–31 December 2016). We queried Ship Arrival Records and Ballast Tank Records from the NBIC data portal for any vessel arriving in Alaska. Ports in Alaska were binned into one of the following regions: Arctic, Bering Sea/Aleutian Islands, Gulf of Alaska, or Southeast Alaska. Source ports with fewer than five reported trips were binned in a group labelled “other”. Records without a port name were removed (n = 13). Discharge reports that did not include valid source locations were also omitted. Many fishing vessels do not exchange ballast water or do not appear in the NBIC. However, these vessels may still transport NIS through the fouling of wetted surfaces such as hulls and anchors. We used VMS data (confidential under the US Code 16 §§ 1881a) from Alaska to examine patterns of fishing vessel traffic. NMFS regulations require VMS on all fishing vessels that target walleye pollock, Pacific cod, Atka mackerel, and crab in the Bering Sea and Gulf of Alaska. VMS-based locations are transmitted at 30-min intervals. The VMS data were matched with polygon boundaries around each port and were segmented into individual trips based on arrivals and departures into and out of ports, respectively (Watson and Haynie, 2016). We analysed a total of 4133 trips by 566 vessels from 2014 to 2016 for consistency with the NBIC data. Port connections with fewer than three vessels were omitted according to confidentiality rules. Some US fishing vessels appeared in both the NBIC and the VMS databases but are not easily queried because “fishing” is not a vessel type category in the NBIC database. To identify these records, we used their co-occurrence to create a “fishing” vessel type, thereby removing these vessels from the “other” category in the NBIC data. These vessels were identified by linking NMFS fishing permit numbers and US Coast Guard numbers to the International Maritime Organization vessel identifiers in the NBIC data via an NMFS vessel database (st.nmfs.noaa.gov/coast-guard-vessel-search/index). Results For the recent (2003–2012) study period, the ROMS predicted mean minimum (winter) temperatures ranging from −2.6°C in the northern Bering Sea to +4.2°C in the south. The predicted mean maximum (summer) temperatures ranged from +3.8 to +16.3. The warmest temperatures were predicted to occur in the southern Bering Sea and in shallow, coastal waters throughout the study area. Mid-century (2030–2039) models predicted slight increases in water temperatures, with mean minimum water temperatures ranging from −2.5 to +4.8°C, and maximum temperatures ranging from +5.4 to +18.6°C (see Hermann et al., 2016; note that reference acknowledges a “cold-bias” in model results). Predicted salinity for the Bering Sea had a mean value of 31.8 ± 1.1 (± SD) and ranged from 23.3 to 48.3 ppt, with seasonally fresher water in the south resulting from melting sea ice (Hermann et al., 2016). In the summer, shallow, coastal areas at the mouth of major rivers were predicted to have salinities <30 ppt. This area extended roughly from Bristol Bay north to Norton Sound (Figure 1). Salinity values were not expected to change appreciably by 2030–2039 (mean: 31.8 ± 1.16, range: 23.5–45.9). From our original list of 49 taxa, we obtained T–S survival thresholds for 42 NIS and reproductive thresholds for 25 marine NIS. The most common taxonomic groups were Crustacea (n = 15), Mollusca (n = 11), and Tunicata (n = 8). Minimum survival temperatures ranged from −2 to +10°C, with a median of 0°C. The median minimum temperature required for reproduction was 12°C (range: 4–18°C). Five taxa were capable of reproducing <10°C. All taxa have been documented to spread via anthropogenic vectors (e.g. ballast water, fouling, intentional introductions; Reimer et al., 2017). Physiological thresholds, predicted suitability models, and other data products associated with this project are available through The Knowledge Network For Biocomplexity Repository (doi: 10.5063/F1RB72ZR). Suitability of temperature and salinity for year-round survival Western Bristol Bay and the Aleutian Islands were predicted to support the highest number of NIS for both recent and mid-century time periods (Figure 2). Areas that were inhospitable for all taxa included northern Norton Sound (≥63.7°N) and the northern Gulf of Anadyr (≥65.0°N; Figure 2). For the recent period, all models predicted suitable conditions for a median of 10 NIS per pixel. The minimum number of taxa with suitable conditions predicted by all models was zero. The maximum number of taxa with suitable conditions varied by model and ranged from 33 to 35 NIS (Table 1). Under mid-century conditions, the CGCM3-t47 and the MIROC3.2 models predicted that the Bering Sea would become more suitable for NIS survival, both in terms of the maximum and the median number of taxa per pixel (Table 1). The ECHO-G model did not predict any change compared with recent predictions. Figure 2. Open in new tabDownload slide Number of non-indigenous species, averaged across three regional ocean models, which were predicted to have temperature and salinity conditions suitable for year-round survival. The rightmost panel shows the change in the average number of taxa from recent to mid-century conditions. Figure 2. Open in new tabDownload slide Number of non-indigenous species, averaged across three regional ocean models, which were predicted to have temperature and salinity conditions suitable for year-round survival. The rightmost panel shows the change in the average number of taxa from recent to mid-century conditions. Table 1. Median and maximum number of NIS that were predicted to have suitable temperature and salinity conditions for year-round survival in the Bering Sea, according to three regional ocean models. Recent predictions (2003–2012) Mid-century predictions (2030–2039) Models Median Maximum Change in median number Change in maximum number CGCM3-t47 10 35 +2 +1 ECHO-G 10 33 0 0 MIROC3.2 10 34 +1 +1 Recent predictions (2003–2012) Mid-century predictions (2030–2039) Models Median Maximum Change in median number Change in maximum number CGCM3-t47 10 35 +2 +1 ECHO-G 10 33 0 0 MIROC3.2 10 34 +1 +1 We considered 42 NIS and two 10-year study periods. Open in new tab Table 1. Median and maximum number of NIS that were predicted to have suitable temperature and salinity conditions for year-round survival in the Bering Sea, according to three regional ocean models. Recent predictions (2003–2012) Mid-century predictions (2030–2039) Models Median Maximum Change in median number Change in maximum number CGCM3-t47 10 35 +2 +1 ECHO-G 10 33 0 0 MIROC3.2 10 34 +1 +1 Recent predictions (2003–2012) Mid-century predictions (2030–2039) Models Median Maximum Change in median number Change in maximum number CGCM3-t47 10 35 +2 +1 ECHO-G 10 33 0 0 MIROC3.2 10 34 +1 +1 We considered 42 NIS and two 10-year study periods. Open in new tab By 2039, ∼50% (5148/10 224 pixels) of the Bering Sea continental shelf was predicted to become suitable for the year-round survival of additional NIS, whereas <6% (570.3 pixels) of the study area was predicted to be suitable for fewer NIS when compared with recent conditions. These values represent ∼635 600 km2 (185 328 nmi2) becoming more suitable by 2039 and ∼70 400 km2 (20 523 nmi2) becoming less suitable over the same time period. Regions between 57°N and 59°N were predicted to experience the greatest increases in NIS suitability (Figure 2). Seasonal and latitudinal patterns of suitability for survival In the first third of the year (weeks 1–16), temperature and salinity conditions were suitable only for NIS whose survival thresholds allowed year-round survival (Figure 3). In the second trimester (weeks 17–34), the number of NIS predicted to have suitable conditions reached a maximum, with conditions becoming suitable for the survival of all NIS we assessed (n = 42). Finally, in the last third of the year, the number of NIS with suitable survival conditions gradually declined (Figure 3). This seasonal pattern was similar under both recent and mid-century study periods. Mid-century conditions predicted that the number of consecutive weeks that could support all 42 NIS would increase from 1 to 3 weeks, depending on the model. Figure 3. Open in new tabDownload slide Number of non-indigenous species, averaged across three regional ocean models, with suitable temperature and salinity conditions for survival in the Bering Sea. Results are shown for every week of the year and for all latitudes in the Bering Sea. The bottom panel shows the qualitative change in the average number of taxa from recent to mid-century conditions. Latitudes at the edge of the study area are minimally represented in the Bering Sea and may not be representative of broader patterns. Figure 3. Open in new tabDownload slide Number of non-indigenous species, averaged across three regional ocean models, with suitable temperature and salinity conditions for survival in the Bering Sea. Results are shown for every week of the year and for all latitudes in the Bering Sea. The bottom panel shows the qualitative change in the average number of taxa from recent to mid-century conditions. Latitudes at the edge of the study area are minimally represented in the Bering Sea and may not be representative of broader patterns. In general, conditions at southern latitudes were conducive to the survival of the greatest number of NIS (Figure 3). Pixel conditions between 52°N and 53°N supported the most taxa year-round. Conditions between 58°N and 59°N supported slightly fewer taxa year-round but were suitable for all taxa for the greatest number of weeks. Mid-century models predicted that the number of weeks with maximum suitability would increase most for latitudes north of 60°N (Figure 3). Suitability of temperature and salinity for reproduction The number of consecutive weeks with suitable reproductive conditions declined as minimum temperature thresholds increased. Taxa with a minimum reproductive temperature of 7°C or less (n = 4) were predicted to have suitable conditions for at least 24 weeks of the year. NIS that required temperatures between 9 and 11°C (n = 7) were predicted to have >8 weeks of suitable conditions (range: 8.5–14.9), whereas taxa requiring 12°C for reproduction (n = 3) were predicted to have <6 weeks (range: 4.9–5.5). Lastly, taxa requiring temperatures ≥14°C (n = 11) were predicted to have <2 weeks of suitable conditions (range: 0–1.2). By mid-century, models predicted a slight increase in the number of consecutive weeks with suitable conditions. The biggest change was predicted for marine taxa whose minimum reproductive thresholds were between 7 and 12°C: by 2039, these taxa would have at least two additional weeks of suitable conditions. For taxa with minimum requirements of 15°C or above, mid-century models predicted that conditions would remain highly limited, averaging <1 week per year. Suitable conditions were predicted to be largely limited to the southern Bering Sea, i.e. from Bristol Bay south to the Aleutian Islands and west to the Kamchatka Peninsula (Figure 4). Our models predicted that temperature and salinity conditions north of 58°N were unsuitable for most NIS, except for Norton Sound (Figure 4). Mid-century models predicted that most areas would become suitable for a larger number of NIS, with Norton Sound, Bristol Bay, and waters off the Kamchatka Peninsula becoming suitable for the largest number of NIS (Figure 4). Figure 4. Open in new tabDownload slide Number of non-indigenous species, averaged across three regional ocean models, which were predicted to have suitable reproductive conditions for at least 1 week in a calendar year. The rightmost panel shows the change in the average number of taxa from recent to mid-century conditions. Figure 4. Open in new tabDownload slide Number of non-indigenous species, averaged across three regional ocean models, which were predicted to have suitable reproductive conditions for at least 1 week in a calendar year. The rightmost panel shows the change in the average number of taxa from recent to mid-century conditions. Patterns of commercial vessel traffic Dutch Harbor received the most traffic for both NBIC- and VMS-reported boats (Figure 5). Nome received the second highest amount of traffic for NBIC-reported vessels, while Akutan received the second highest amount of traffic for VMS-reported vessels. Arrivals originating outside of Alaska accounted for 83% of NBIC records (Figure 5a). California (n = 175), Washington (n = 142), and South Korea (n = 127) accounted for more vessel traffic than the more proximate Gulf of Alaska ports (n = 120). However, from VMS data, which predominantly include fishing vessels that do not report to the NBIC, an overwhelming majority of trips originated from Gulf of Alaska ports (n = 657; Figure 5b). Figure 5. Open in new tabDownload slide Total number of vessel trips to US Bering Sea ports from 2014 to 2016. The width of the coloured arcs or node segments represents the proportion of vessel traffic contributed to or received by each US Bering Sea port. Numbers in parentheses represent the number of vessel trips to or from that location. (a) Data from the National Ballast Information Clearinghouse. (b) Commercial fishing vessels whose locations are tracked through the NMFS Vessel Monitoring Systems. Figure 5. Open in new tabDownload slide Total number of vessel trips to US Bering Sea ports from 2014 to 2016. The width of the coloured arcs or node segments represents the proportion of vessel traffic contributed to or received by each US Bering Sea port. Numbers in parentheses represent the number of vessel trips to or from that location. (a) Data from the National Ballast Information Clearinghouse. (b) Commercial fishing vessels whose locations are tracked through the NMFS Vessel Monitoring Systems. Ballast water discharge reports for the US Bering Sea were distributed across nine vessel types, with Tankers accounting for >90% of the total ballast water volume discharged. Eighty-eight percent of ballast water discharged occurred in Dutch Harbor, equivalent to 57 736 metric tonnes (t). The port of Nome received the second largest discharge volume (7004 t). Despite more trips originating in the eastern Pacific Ocean, most ballast water released in the Bering Sea originated from Asia. South Korea and China each accounted for an order of magnitude more ballast water (18 728 and 17 453 t, respectively) than the next greatest source, Japan (7183 t). Approximately 20% of ballast water reports (representing 11% of the discharged volume) identified the source of their ballast water using coordinates instead of port names. Among these non-port sources, 76% of coordinates occurred in the open ocean, i.e. water depths >200 m. Twenty-five percent originated from locations in the northeast Pacific Ocean (defined here as latitudes >23.5°N, longitudes between 179.9°W and 110°W), and 15% originated from locations in the northwest Pacific Ocean (defined here as latitudes >23.5°N, longitudes between 100°E and 180°E). Discussion Suitability of temperature and salinity for NIS under recent oceanographic conditions Our results indicate that recent temperatures and salinities in the Bering Sea are suitable for the survival and reproduction of several NIS. Of the 42 taxa assessed, all were predicted to have suitable survival conditions for 6 weeks from early July to mid-August. Thirty-four taxa were predicted to have suitable conditions for year-round survival in at least one pixel. Moreover, 15 of 25 taxa had suitable reproductive conditions for at least 1 week in a calendar year. The southern Bering Sea was predicted to have suitable survival conditions for the greatest number of NIS. Far fewer were predicted to have suitable conditions north of 58°N (Figures 2 and 3). The 58°N “threshold” predicted by our models is coincident with the recent limit of seasonal sea ice extent in the Bering Sea (Stabeno et al., 2012). Above this boundary, year-round survival was limited to taxa that could tolerate sub-zero water temperatures. Suitable reproductive conditions were similarly focused in the southern Bering Sea, though several taxa were predicted to have suitable reproductive conditions in Norton Sound (∼64°N; Figure 4). Norton Sound freezes annually, but its shallow waters create a high-latitude hotspot in the summer that could present an opportunity for cold-tolerant NIS to establish populations in the northern Bering Sea. Suitability of temperature and salinity for NIS under future oceanographic conditions Mid-century models (2030–2039) predicted that conditions would support only a few additional NIS that cannot survive year-round under recent predictions (Table 1). However, models predicted a large expansion of suitable conditions for NIS that can already tolerate recent conditions. Suitable conditions are expected to expand northward, in agreement with other high-latitude modelling studies (de Rivera et al., 2011; Ware et al., 2016; Goldsmit et al., 2018) and surveys of native species in the Bering Sea (Mueter and Litzow, 2008; Kotwicki and Lauth, 2013; Barbeaux and Hollowed, 2018). With respect to reproduction, our models suggested slight-to-moderate changes in the number of consecutive weeks with suitable reproductive conditions. They also predicted increased suitability for large parts of the continental shelf, including shallow, coastal waters around the Kamchatka Peninsula and Norton Sound (Figure 4). Although our models vary with respect to the magnitude and seasonality of current and future predictions, they predict similar temporal and spatial patterns. However, the onset of warming conditions may be conservative, as recent observations have shown that temperatures in the Bering Sea are warming faster than predicted. The past few years have experienced several of the lowest sea ice extents on record and substantial decreases in the length of the ice-covered season (Duffy-Anderson et al., 2019). For example, sea ice during the winter of 2017–2018 was nearly absent from the northern Bering Sea (Duffy-Anderson et al., 2019). Observations have also documented extreme summer sea surface temperatures in Norton Sound (Eisner, 2019), where several NIS are predicted to have suitable reproductive conditions (Figure 4). The importance of life history to predicting species’ ranges Life-history considerations are currently underrepresented in the range shift literature. Yet, including life stages and processes beyond adult survival can provide critical information on limits to the establishment (Kearney and Porter, 2009; Kingsolver et al., 2011; Levy et al., 2015). These considerations may become especially salient at high latitudes where exposure to low temperatures can affect processes such as spawning or time required for development (Baba et al., 1999; de Rivera et al., 2007; Sunde et al., 2019). In our study, 80% of NIS had suitable conditions for year-round survival and 60% had at least 1 week of suitable reproductive conditions. However, only 40% of the NIS we assessed were predicted to have more than 6 consecutive weeks of suitable reproductive conditions and 20% were predicted to have 12 or more consecutive weeks. Interestingly, the two NIS that have established populations in the Bering Sea were among the five NIS that had reported minimum reproductive thresholds of 10°C and were the only two predicted to have suitable reproductive conditions nearly year-round. Collectively, these results suggest that some NIS may fail to establish at high latitudes because suitable conditions do not last long enough for them to reproduce and develop. However, it is important to note that, for nearshore waters that are heavily influenced by solar radiation and river discharge, the ROMS may underestimate summer water temperatures by as much as 3–5°C (MODIS imaging sensor, https://neo.sci.gsfc.nasa.gov/; T. Jorgenson, pers. comm.). These limitations likely lead to conservative estimates of suitable reproductive conditions but are of minor concern for predicting suitable year-round survival conditions, as outcomes for the latter are driven by minimum temperature thresholds. Further studies to identify life-history requirements such as physiological thresholds and time to development may improve our ability to predict biological invasions. More generally, these data could improve our ability to model species’ distributions for a variety of conservation and management purposes (Kearney and Porter, 2009; Woodin et al., 2013). Interactions between vessel traffic and suitable conditions Our analysis of vessel traffic and ballast water discharge identifies the port of Dutch Harbor as the most visited port in the US Bering Sea. Dutch Harbor is one of the highest volume fishing ports in the United States and has long been recognized as a potentially important point of entry for NIS based on patterns of vessel traffic (McGee et al., 2006; Verna et al., 2016). The vulnerability of Dutch Harbor to NIS introductions is particularly salient because of its location in the southeastern Bering Sea, which has highly suitable conditions for NIS survival and reproduction (Figures 2 and 4). Traffic to Dutch Harbor was up to two orders of magnitude more than traffic to other ports, and trips ending in Dutch Harbor originated from more than ten countries. Compared with ballast water data from large, commercial ships, fishing vessel traffic was highly regional, connecting the port of Dutch Harbor to Bering Sea ports in the Pribilof Islands, Bristol Bay, and Akutan. While many of these vessels may not discharge ballast water, they can still transport fouling organisms, which may be a larger contributor to propagules than ballast water (Chan et al., 2015). Studies have suggested that fewer organisms survive the voyage to northern ports such as Dutch Harbor because of the long transit times from temperate to high-latitude ports (Chan et al. 2014; Verna et al., 2016). Indeed, a study in the Canadian Arctic found an inverse relationship between the duration of the voyage and NIS richness and abundance (Chan et al., 2014). However, it is also true that at least some organisms are surviving the voyage (Chan et al., 2014; Ware et al., 2014; Chan et al., 2016). Although many vessels are transiting to the Bering Sea (Figure 5), measures of propagule pressure, related to the number, type, and quality of organisms being introduced, are unknown. In the absence of data on propagule pressure, future analyses of vessel traffic in the context of invasive species transport may benefit from using Automatic Identification System data, which are available for a greater portion of vessels travelling through Alaska waters. For example, a recent study on Arctic maritime activity projected a “most plausible scenario” of an ∼30% increase in vessel counts in the next decade (CMTS, 2019). Similar analyses of vessel traffic and ballast water discharge for ports in the Russian Bering Sea would also be vital for more fully projecting risk in this context. Considering factors beyond temperature and salinities: opportunities for future research In this article, we sought to explore temperature and salinity thresholds because they are two primary physiological barriers for NIS establishment. However, other physiological and ecological requirements may limit NIS establishment. For one, the ability of temperate NIS to adapt their behaviour and phenology to the short primary production season and the extreme light regime may continue to limit high-latitude invasions even as the climate becomes more favourable (Poloczanska et al., 2016; Sundby et al., 2016; Langbehn and Varpe, 2017). Indeed, a review of Arctic taxa suggests that the ability to reproduce and maintain high levels of activity throughout the year is a common adaptation; for these taxa, critical life-history processes are not reliant on warm water temperatures, light, or peaks in primary production (Berge et al., 2015). Second, requirements such as substrate type and water depth may be important considerations, particularly for sessile organisms (Ruiz et al., 2009). Although we did not explicitly account for substrate type, many communities in the Bering Sea have docks, piers, and jetties that could provide suitable habitat for NIS that require hard substrates. However, the ROMS have a horizontal resolution of ∼10 km, which limits the representation of intertidal zones and waters within a few kilometres from shore. Lastly, although we considered critical temperature and salinity thresholds, an organism’s performance may decline even before lethal thresholds are met (Kassahn et al., 2009; Monaco and Helmuth, 2011; Woodin et al., 2013). Experimental and field studies are necessary for obtaining richer datasets that move beyond critical thresholds and allow us to develop an understanding of species’ thermal response curves. Our examination of suitable conditions does not propose to be an exhaustive list of all possible environmental parameters relevant to each species or life-history stage, but it does propose a foundation by which NIS invasion risk may be examined in the context of a changing climate. While it is widely believed that NIS establishment to high-latitude systems is limited by cold water temperatures, our results indicate that (i) water temperatures do not limit the survival of NIS in the southern Bering Sea, but minimum water temperatures in areas with seasonal sea ice may limit survival of some NIS; (ii) the window of time where temperatures are suitable for reproduction may be too narrow for temperate NIS; (iii) suitable conditions are projected to expand by mid-century due to warming ocean temperatures; and (iv) mechanisms for NIS introduction by vessels overlap with suitable conditions. For most NIS that we assessed, temperature and salinity conditions prevent their successful establishment in the northern Bering Sea. However, suitable conditions exist in the southeastern Bering Sea, which experiences high vessel traffic, notably to and from the port of Dutch Harbor. As ocean temperatures and vessel traffic continue to increase, we expect the risk of NIS introduction and establishment to increase throughout the Bering Sea region, potentially adding to the list of stressors that are already being experienced by native species and the human communities that depend on them. Acknowledgements Tracey Gotthardt and Aaron Poe were involved in spearheading the project. Casey Greenstein, Lindsey Flagstad, Bonnie Bernard, Jaime Weltfelt, Curtis Whisman, Jen Karnak, Rob Bochenek, William Koeppen, Marcus Geist, and Al Hermann provided valuable data assistance and interpretative insight. Michael Carey, Matthew Carlson, Elizabeth Logerwell, Janine Powell, and three anonymous reviewers provided valuable feedback on this article. Funding Funding was provided by the North Pacific Research Board (project # 1523) and the Aleutian and Bering Sea Islands Landscape Conservation Cooperative. Disclaimer The findings and conclusions in the paper are those of the author(s) and do not necessarily represent the views of the National Marine Fisheries Service. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government. References Ashton G. V. , Riedlecker E. I. , Ruiz G. M. 2008 . First non-native crustacean established in coastal waters of Alaska . Aquatic Biology , 3 : 133 – 137 . Google Scholar Crossref Search ADS WorldCat Baba K. , Tada M. , Kawajiri T. , Kuwahara Y. 1999 . Effects of temperature and salinity on spawning of the brackish water bivalve Corbicula japonica in Lake Abashiri, Hokkaido, Japan . Marine Ecology Progress Series , 180 : 213 – 221 . Google Scholar Crossref Search ADS WorldCat Barbeaux S. J. , Hollowed A. B. 2018 . Ontogeny matters: climate variability and effects on fish distribution in the eastern Bering Sea . Fisheries Oceanography , 27 : 1 – 15 . Google Scholar Crossref Search ADS WorldCat Berge J. , Renaud P. E. , Darnis G. , Cottier F. , Last K. , Gabrielsen T. M. , Johnsen G. et al. 2015 . In the dark: a review of ecosystem processes during the Arctic polar night . Progress in Oceanography , 139 : 258 – 271 . Google Scholar Crossref Search ADS WorldCat Bivand R. , Keitt T. , Rowlingson R. 2018 . rgdal: Bindings for the “Geospatial” Data Abstraction Library (R Package Version 1.2-20) [Computer software]. https://CRAN.R-project.org/package=rgdal (last accessed 13 February 2020). Bivand R. , Rundel C. 2017 . rgeos: Interface to Geometry Engine—Open Source (‘GEOS’) (R Package Version 0.3-26) [Computer software]. https://CRAN.R-project.org/package=rgeos (last accessed 13 February 2020). Bivand R. S. , Pebesma E. , Gomez-Rubio V. 2013 . Applied Spatial Data Analysis with R , 2nd edn. Springer , New York . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Blackburn T. M. , Pyšek P. , Bacher S. , Carlton J. T. , Duncan R. P. , Jarošík V. , Wilson J. R. U. et al. 2011 . A proposed unified framework for biological invasions . Trends in Ecology and Evolution , 26 : 333 – 339 . Google Scholar Crossref Search ADS PubMed WorldCat Chan F. T. , Briski E. , Bailey S. A. , MacIsaac H. J. 2014 . Richness–abundance relationships for zooplankton in ballast water: temperate versus Arctic comparisons . ICES Journal of Marine Science , 71 : 1876 – 1884 . Google Scholar Crossref Search ADS WorldCat Chan F. T. , MacIsaac H. J. , Bailey S. A. 2016 . Survival of ship biofouling assemblages during and after voyages to the Canadian Arctic . Marine Biology , 163 : 250 . Google Scholar Crossref Search ADS PubMed WorldCat Chan F. T. , MacIsaac H. J. , Bailey S. A. , Krkošek M. 2015 . Relative importance of vessel hull fouling and ballast water as transport vectors of nonindigenous species to the Canadian Arctic . Canadian Journal of Fisheries and Aquatic Sciences , 72 : 1230 – 1242 . Google Scholar Crossref Search ADS WorldCat Chan F. T. , Stanislawczyk K. , Sneekes A. C. , Dvoretsky A. , Gollasch S. , Minchin D. , David M. et al. 2019 . Climate change opens new frontiers for marine species in the Arctic: current trends and future invasion risks . Global Change Biology , 25 : 25 – 38 . Google Scholar Crossref Search ADS PubMed WorldCat U.S. Committee on the Marine Transportation System (CMTS). 2019 . A Ten-year Projection of Maritime Activity in the U.S. Arctic Region, 2020–2030. Washington, DC. https://www.cmts.gov/downloads/CMTS_2019_Arctic_Vessel_Projection_Report.pdf (last accessed 13 February 2020). de Rivera C. E. , Hitchcock N. G. , Teck S. J. , Steves B. P. , Hines A. H. , Ruiz G. M. 2007 . Larval development rate predicts range expansion of an introduced crab . Marine Biology , 150 : 1275 – 1288 . Google Scholar Crossref Search ADS WorldCat de Rivera C. E. , Steves B. P. , Fofonoff P. W. , Hines A. H. , Ruiz G. M. 2011 . Potential for high-latitude marine invasions along western North America . Diversity and Distributions , 17 : 1198 – 1209 . Google Scholar Crossref Search ADS WorldCat Dijkstra J. A. , Westerman E. L. , Harris L. G. 2017 . Elevated seasonal temperatures eliminate thermal barriers of reproduction of a dominant invasive species: a community state change for northern communities? Diversity and Distributions , 23 : 1182 – 1192 . Google Scholar Crossref Search ADS WorldCat Duffy-Anderson J. T. , Stabeno P. , Andrews A. G. , Cieciel K. , Deary A. , Farley E. , Fugate C. et al. 2019 . Responses of the northern Bering Sea and southeastern Bering Sea pelagic ecosystems following record‐breaking low winter sea ice . Geophysical Research Letters , 46 : 9833 – 9842 . Google Scholar Crossref Search ADS WorldCat Eisner L. B. 2019 . The Bering Sea: current status and recent events . PICES Press , 27 : 33 – 39 . WorldCat Fofonoff P. W. , Ruiz G. M. , Steves B. , Simkanin C. , Carlton J. T. 2018 . National Exotic Marine and Estuarine Species Information System (NEMESIS). http://invasions.si.edu/nemesis/ (last accessed 13 February 2020). Fuller P. F. , Benson A. J. 2013 . Nonindigenous Aquatic Species Database (NAS). https://nas.er.usgs.gov/ (last accessed 13 February 2020). Goldsmit J. , Archambault P. , Chust G. , Villarino E. , Liu G. , Lukovich J. V. , Barber D. G. et al. 2018 . Projecting present and future habitat suitability of ship-mediated aquatic invasive species in the Canadian Arctic . Biological Invasions , 20 : 501 – 517 . Google Scholar Crossref Search ADS WorldCat Gu Z. , Gu L. , Eils R. , Schlesner M. , Brors B. 2014 . Circlize implements and enhances circular visualization in R . Bioinformatics , 30 : 2811 – 2812 . Google Scholar Crossref Search ADS PubMed WorldCat Hermann A. J. , Gibson G. A. , Bond N. A. , Curchitser E. N. , Hedstrom K. , Cheng W. , Wang M. et al. 2016 . Projected future biophysical states of the Bering Sea . Deep-Sea Research Part II: Topical Studies in Oceanography , 134 : 30 – 47 . Google Scholar Crossref Search ADS WorldCat Hijmans R. J. 2017 . raster: Geographic Data Analysis and Modeling (R Package Version 2.6-7) [Computer Software]. https://CRAN.R-project.org/package=raster (last accessed 13 February 2020). Iacarella J. C. , Dick J. T. A. , Alexander M. E. , Ricciardi A. 2015 . Ecological impacts of invasive alien species along temperature gradients: testing the role of environmental matching . Ecological Applications , 25 : 706 – 716 . Google Scholar Crossref Search ADS PubMed WorldCat Kassahn K. S. , Crozier R. H. , Pörtner H. O. , Caley M. J. 2009 . Animal performance and stress: responses and tolerance limits at different levels of biological organisation . Biological Reviews , 84 : 277 – 292 . Google Scholar Crossref Search ADS PubMed WorldCat Kearney M. , Porter W. 2009 . Mechanistic niche modelling: combining physiological and spatial data to predict species’ ranges . Ecology Letters , 12 : 334 – 350 . Google Scholar Crossref Search ADS PubMed WorldCat Kingsolver J. G. , Woods H. A. , Buckley L. B. , Potter K. A. , MacLean H. J. , Higgins J. K. 2011 . Complex life cycles and the responses of insects to climate change . Integrative and Comparative Biology , 51 : 719 – 732 . Google Scholar Crossref Search ADS PubMed WorldCat Kotwicki S. , Lauth R. R. 2013 . Detecting temporal trends and environmentally-driven changes in the spatial distribution of bottom fishes and crabs on the eastern Bering Sea shelf . Deep-Sea Research Part II: Topical Studies in Oceanography , 94 : 231 – 243 . Google Scholar Crossref Search ADS WorldCat Langbehn T. J. , Varpe Ø. 2017 . Sea-ice loss boosts visual search: fish foraging and changing pelagic interactions in polar oceans . Global Change Biology , 23 : 5318 – 5330 . Google Scholar Crossref Search ADS PubMed WorldCat Levy O. , Buckley L. B. , Keitt T. H. , Smith C. D. , Boateng K. O. , Kumar D. S. , Angilletta M. J. Jr 2015 . Resolving the life cycle alters expected impacts of climate change . Proceedings of the Royal Society B , 282 : 20150837 . Google Scholar Crossref Search ADS PubMed WorldCat McGee S. , Piorkowski R. , Ruiz G. 2006 . Analysis of recent vessel arrivals and ballast water discharge in Alaska: toward assessing ship-mediated invasion risk . Marine Pollution Bulletin , 52 : 1634 – 1645 . Google Scholar Crossref Search ADS PubMed WorldCat Microsoft Co Weston S. 2017 . doSNOW: Foreach Parallel Adaptor for the “snow” Package (R Package Version 1.0.16) [Computer Software]. https://CRAN.R-project.org/package=doSNOW (last accessed 13 February 2020). Miller A. W. , Ruiz G. M. 2014 . Arctic shipping and marine invaders . Nature Climate Change , 4 : 413 – 416 . Google Scholar Crossref Search ADS WorldCat Molnar J. L. , Gamboa R. L. , Revenga C. , Spalding M. D. 2008 . Assessing the global threat of invasive species to marine biodiversity . Frontiers in Ecology and the Environment , 6 : 485 – 492 . Google Scholar Crossref Search ADS WorldCat Monaco C. J. , Helmuth B. 2011 . Tipping points, thresholds and the keystone role of physiology in marine climate change research . Advances in Marine Biology , 60 : 123 – 160 . Google Scholar Crossref Search ADS PubMed WorldCat Mueter F. J. , Litzow M. A. 2008 . Sea ice retreat alters the biogeography of the Bering Sea continental shelf . Ecological Applications , 18 : 309 – 320 . Google Scholar Crossref Search ADS PubMed WorldCat National Marine Fisheries Service (NMFS). 2017 . Fisheries of the United States, 2016. NOAA Current Fishery Statistics No. 2016. https://www.fisheries.noaa.gov/webdam/download/62276710 (last accessed 13 February 2020). Onarheim I. H. , Eldevik T. , Smedsrud L. H. , Stroeve J. C. 2018 . Seasonal and regional manifestation of Arctic sea ice loss . Journal of Climate , 31 : 4917 – 4932 . Google Scholar Crossref Search ADS WorldCat Paiva F. , Barco A. , Chen Y. , Mirzajani A. , Chan F. T. , Lauringson V. , Baltazar-Soares M. et al. 2018 . Is salinity an obstacle for biological invasions? Global Change Biology , 12 : 3218 – 3221 . WorldCat Pierce D. 2017 . Interface to Unidata netCDF (Version 4 or Earlier) Format Data Files. (R package Version 1.16) [Computer software]. https://cran.r-project.org/package=ncdf4 (last accessed 13 February 2020). Poloczanska E. S. , Burrows M. T. , Brown C. J. , García Molinos J. , Halpern B. S. , Hoegh-Guldberg O. , Kappel C. V. et al. 2016 . Responses of marine organisms to climate change across oceans . Frontiers in Marine Science , 3 : 62 . Google Scholar Crossref Search ADS WorldCat Pörtner H. O. 2001 . Climate change and temperature-dependent biogeography: oxygen limitation of thermal tolerance in animals . Naturwissenschaften , 88 : 137 – 146 . Google Scholar Crossref Search ADS PubMed WorldCat R Core Team. 2018 . R: A Language and Environment for Statistical Computing (Version 3.3.2) [Computer Software]. https://www.r-project.org/ (last accessed 13 February 2020). Reimer J. P. , Droghini A. , Fischbach A. , Watson J. T. , Bernard B. , Poe A. 2017 . Assessing the risk of non-native marine species in the Bering Sea. Final report, Alaska Center for Conservation Science, University of Alaska Anchorage, AK. http://accs.uaa.alaska.edu/invasive-species/bering-sea-marine-invasives/ (last accessed 13 February 2020). Ruiz G. M. , Fofonoff P. W. , Carlton J. T. 2015 . Invasion history and vector dynamics in coastal marine ecosystems: a North American perspective . Aquatic Ecosystem Health & Management , 18 : 299 – 311 . Google Scholar Crossref Search ADS WorldCat Ruiz G. M. , Freestone A. L. , Fofonoff P. W. , Simkanin C. 2009 . Habitat distribution and heterogeneity in marine invasion dynamics: the importance of hard substrate and artificial structure. In Marine Hard Bottom Communities Studies , pp. 321 – 332. Ed by Wahl M. . doi: 10.1007/b76710. Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Ruiz G. M. , Hewitt C. L. 2009 . Latitudinal patterns of biological invasions in marine ecosystems: a polar perspective. In Smithsonian at the Poles: Contributions to International Polar Year Science , pp. 347 – 358 . Ed by Krupnik I. , Lang M. A. , Miller S. E. Smithsonian Institution, Washington, D.C., USA. doi: 10.5479/si.097884601X.0. Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Ruiz G. M. , Huber T. , Larson K. , McCann L. , Steves B. , Fofonoff P. W. , Hines A. H. 2006 . Biological invasions in Alaska’s coastal marine ecosystems: establishing a baseline. Report prepared for the Prince William Sound Regional Citizens’ Advisory Council & U.S. Fish & Wildlife Service. Smithsonian Environmental Research Center, Edgewater, MD. Springer, Berlin, Heidelberg, DEU. Seebens H. , Gastner M. T. , Blasius B. 2013 . The risk of marine bioinvasion caused by global shipping . Ecology Letters , 16 : 782 – 790 . Google Scholar Crossref Search ADS PubMed WorldCat Spalding M. D. , Fox H. E. , Allen G. R. , Davidson N. , Ferdaña Z. A. , Finlayson M. , Halpern B. S. et al. 2007 . Marine ecoregions of the world: a bioregionalization of coastal and shelf areas . BioScience , 57 : 573 – 583 . Google Scholar Crossref Search ADS WorldCat Stabeno P. J. , Bond N. A. , Salo S. A. 2007 . On the recent warming of the southeastern Bering Sea shelf . Deep-Sea Research Part II: Topical Studies in Oceanography , 54 : 2599 – 2618 . Google Scholar Crossref Search ADS WorldCat Stabeno P. J. , Kachel N. B. , Moore S. E. , Napp J. M. , Sigler M. , Yamaguchi A. , Zerbini A. N. 2012 . Comparison of warm and cold years on the southeastern Bering Sea shelf and some implications for the ecosystem . Deep-Sea Research Part II: Topical Studies in Oceanography , 65 : 31 – 45 . Google Scholar Crossref Search ADS WorldCat Sundby S. , Drinkwater K. F. , Kjesbu O. S. 2016 . The North Atlantic spring-bloom system—Where the changing climate meets the winter dark . Frontiers in Marine Science , 3 : 28 . Google Scholar Crossref Search ADS WorldCat Sunde J. , Larsson P. , Forsman A. 2019 . Adaptations of early development to local spawning temperature in anadromous populations of pike (Esox lucius) . BMC Evolutionary Biology , 19 : 1 – 13 . Google Scholar Crossref Search ADS PubMed WorldCat Verling E. , Ruiz G. M. , Smith L. D. , Galil B. , Miller A. W. , Murphy K. R. 2005 . Supply-side invasion ecology: characterizing propagule pressure in coastal ecosystems . Proceedings of the Royal Society B: Biological Sciences , 272 : 1249 – 1257 . Google Scholar Crossref Search ADS WorldCat Verna D. , Harris B. , Holzer K. , Minton M. 2016 . Ballast-borne marine invasive species: exploring the risk to coastal Alaska, USA . Management of Biological Invasions , 7 : 199 – 211 . Google Scholar Crossref Search ADS WorldCat Ware C. , Berge J. , Jelmert A. , Olsen S. M. , Pellissier L. , Wisz M. , Kriticos D. et al. 2016 . Biological introduction risks from shipping in a warming Arctic . Journal of Applied Ecology , 53 : 340 – 349 . Google Scholar Crossref Search ADS WorldCat Ware C. , Berge J. , Sundet J. H. , Kirkpatrick J. B. , Coutts A. D. M. , Jelmert A. , Olsen S. M. et al. 2014 . Climate change, non-indigenous species and shipping: assessing the risk of species introduction to a high-Arctic archipelago . Diversity and Distributions , 20 : 10 – 19 . Google Scholar Crossref Search ADS WorldCat Watson J. T. , Haynie A. C. 2016 . Using vessel monitoring system data to identify and characterize trips made by fishing vessels in the United States North Pacific . PLoS One , 11 : e0165173 – 20 . Google Scholar Crossref Search ADS PubMed WorldCat Wickham H. 2011 . The split-apply-combine strategy for data analysis . Journal of Statistical Software , 40 : 1 – 29 . WorldCat Wickham H. , Averick M. , Bryan J. , Chang W. , McGowan L. , François R. , Grolemund G. et al. 2019 . Welcome to the Tidyverse . Journal of Open Source Software , 4 : 1686 . Google Scholar Crossref Search ADS WorldCat Woodin S. A. , Hilbish T. J. , Helmuth B. , Jones S. J. , Wethey D. S. 2013 . Climate change, species distribution models, and physiological performance metrics: predicting when biogeographic models are likely to fail . Ecology and Evolution , 3 : 3334 – 3346 . Google Scholar PubMed WorldCat © International Council for the Exploration of the Sea 2020. 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)