Reducing uncertainty in fisheries stock status James Larcombe, Rocio Noriega and Ilona Stobutzki (editors)
Research by the Australian Bureau of Agricultural and Resource Economics and Sciences Publication series September 2015
© Commonwealth of Australia 2015
Ownership of intellectual property rights Unless otherwise noted, copyright (and any other intellectual property rights, if any) in this publication is owned by the Commonwealth of Australia (referred to as the Commonwealth). Creative Commons licence All material in this publication is licensed under a Creative Commons Attribution 3.0 Australia Licence, save for content supplied by third parties, logos and the Commonwealth Coat of Arms.
Creative Commons Attribution 3.0 Australia Licence is a standard form licence agreement that allows you to copy, distribute, transmit and adapt this publication provided you attribute the work. A summary of the licence terms is available from creativecommons.org/licenses/by/3.0/au/deed.en. The full licence terms are available from creativecommons.org/licenses/by/3.0/au/legalcode. Cataloguing data ABARES 2015, Reducing uncertainty in fisheries stock status, ABARES research report, Canberra, August. CC BY 3.0.
ISBN 978-1-74323-254-5 ABARES project 25679
Internet Reducing uncertainty in fisheries stock status is available at agriculture.gov.au/abares/publications. Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) Postal address GPO Box 858 Canberra ACT 2601 Switchboard +61 2 6272 2010 Facsimile +61 2 6272 2001 Email [email protected] Web agriculture.gov.au/abares Inquiries about the licence and any use of this document should be sent to [email protected]. The Australian Government acting through the Department of Agriculture, represented by the Australian Bureau of Agricultural and Resource Economics and Sciences, has exercised due care and skill in preparing and compiling the information and data in this publication. Notwithstanding, the Department of Agriculture, ABARES, its employees and advisers disclaim all liability, including for negligence and for any loss, damage, injury, expense or cost incurred by any person as a result of accessing, using or relying upon information or data in this publication to the maximum extent permitted by law.
Acknowledgements The Reducing Uncertainty in Stock Status (RUSS) research team at ABARES comprised: Belinda Barnes, Veronica Boero, Mark Chambers, James Larcombe, Adam Leatherbarrow, Rocia Noriega, Kevin McLoughlin, Heather Patterson, Lindsay Penrose, Ilona Stobutzki, Peter Ward and James Woodhams. The research team at CSIRO comprised: Natalie Dowling, Malcolm Haddon, Neil Klaer, Eva Plagányi, Tim Skewes and Sally Wayte. The project was funded by the (then) Department of Agriculture, Fisheries and Forestry.
Reducing uncertainty in stock status ABARES
Contents
1 Overview 1 Stocks and biological status 1 Reducing uncertainty in stock status 3 Reducing uncertainty in stock status assessments and papers 3 Discussion and conclusions 8 Management strategy evaluation 9 References 12 2 Coral Sea Fishery: Aquarium Sector assessments 14 Summary 14 Introduction 15 Data and information 16 Methods 18 Results 20 Discussion 25 Appendix A: Key commercial families collected 26 References 26 3 Assessing Coral Sea Fishery sea cucumber stocks using spatial methods 28 Summary 28 Introduction 29 Background 30 Methods 34 Results 40 Discussion 51 Appendix B 54 References 56 4 Coral Sea Fishery Line and Trap Sector: preliminary stock assessments 58 Summary 58 Introduction 59 Catch and effort statistics 60 Assessment approaches 70 Part 1: Habitat assessment 71 Part 2: The deep scalefish assemblage 73 Part 3: The reef scalefish assemblage 79 Part 4: The shark assemblage 88 Appendix C 91 References 97
iii Reducing uncertainty in stock status ABARES
5 Status determination for trochus and tropical rock lobster stocks in the Coral Sea Fishery Hand Collection Sector 100 Summary 100 Introduction 100 Tropical rock lobster 102 Trochus 106 References 106 6 Status determination for the deepwater prawn stock in the North West Slope Trawl Fishery 108 Summary 108 Introduction 108 Background information 110 Analysis 111 Discussion 125 Appendix D 126 References 127 7 North West Slope Trawl Fishery Scampi assessment 130 Summary 130 Introduction 130 Catch and effort 133 Methods 135 Results 138 Discussion and status 142 Appendix E 144 References 146 8 Elephant fish catch rate standardisation and Tier 4 assessment, 2009 148 Summary 148 Introduction 148 Data preparation 150 Catch per unit effort standardisation 151 Tier 4 assessment 154 Appendix F 157 References 161 9 Sawshark catch rate standardisation and Tier 4 assessment, 2009 162 Summary 162 Introduction 162 Data preparation 164 Catch per unit effort standardisation 165 Tier 4 assessment 168 Conclusion 170
iv Reducing uncertainty in stock status ABARES
Appendix G 171 References 175 10 Depletion analyses of Gould’s squid in the Bass Strait 176 Summary 176 Introduction 176 Data and methods 178 Results 183 Discussion 192 Conclusion 193 Appendix H 194 References 197 11 Stock status determination: weight-of-evidence decision-making framework 199 Summary 199 Introduction 199 Methods 200 Weighing evidence 213 Reporting of stock status 216 Appendix I 216 References 219
Tables
Table 2.1 Area (km²) of suitable habitat classes for aquarium fish in the Coral Sea Fishery 16 Table 2.2 The scoring system to assess the vulnerability of marine aquarium species in the Coral Sea Fishery Aquarium Sector 19 Table 2.3 Scoring system to assess the susceptibility of the marine aquarium species captured in the Coral Sea Fishery Aquarium Sector applied to species with medium or high vulnerability 20 Table 2.4 Estimates of maximum footprint for the sector 20 Table 2.5 Summary of observed minimum and maximum densities, estimated suitable habitat area, standing stock and annual extraction rates based on 2008–09 catch data 22 Table 2.6 Vulnerability risk scores for species within the Aquarium Sector of the Coral Sea Fishery 23 Table 2.7 Susceptibility risk scores for species within the Aquarium Sector of the Coral Sea Fishery 24 Table 3.1 Catch statistics by reef for hand collection operations in the CSF 32
v Reducing uncertainty in stock status ABARES
Table 3.2 Natural mortality estimates used by Skewes et al. (2004) for the Torres Strait and those used in these analyses 39 Table 3.3 Parameters (per hectare) used to estimate population size 41 Table 3.4 Habitat areas (in hectares) of the five habitat classes of interest 43 Table 3.5 Average animal weight for the four key commercial sea cucumber species in the Coral Sea Fishery 43 Table 3.6 Black teatfish population (numbers), biomass (kg) and maximum sustainable yield (kg) estimates by reef 45 Table 3.7 White teatfish population (numbers) biomass (kg) and maximum sustainable yield (kg) estimates by reef 46 Table 3.8 Prickly redfish population (numbers) biomass (kg) and maximum sustainable yield (kg) estimates by reef 47 Table 3.9 Surf redfish population (numbers), biomass (kg) and maximum sustainable yield (kg) estimates by reef 48 Table 3.10 Proportion of biomass remaining, 20th percentile starting biomass, Schaefer model and Pella–Tomlinson model 49 Table 3.11 Proportion of biomass remaining, median starting biomass, Schaefer model 50 Table 3.12 Proportion of biomass remaining, median starting biomass, Pella– Tomlinson model 51 Table 4.1 Coral Sea Fishery line sector annual financial year catch (kg) by species/species group 62 Table 4.2 Coral Sea Fishery trap sector annual financial year catch (kg) by species/species group 68 Table 4.3 Yield scenarios for the Coral Sea Fishery deep scalefish assemblage 77 Table 4.4 Yield calculations for the Coral Sea Fishery reef scalefish assemblage based on low and medium biomass scenarios (exploitation constant x=0.3, x=0.5 and x=0.7) 85 Table 7.1 Standing stock of scampi in North West Slope Trawl Fishery by Davis and Ward (1984) 132 Table 7.2 Biological parameters and model settings for North West Slope Trawl Fishery combined scampi used in stock reduction analysis 132 Table 7.3 A Stock Production Model Incorporating Covariates production model runs for North West Slope Trawl Fishery scampi 138 Table 7.4 Summary results of production model runs for North West Slope Trawl Fishery scampi 140 Table 8.1 Elephant fish 2009 Tier 4 methodology results (all areas) for two candidate reference periods 155 Table 9.1 Sawshark 2009 Tier 4 methodology results (base case, all areas) 170 Table 10.1 Results of the depletion analysis that used a natural mortality rate of 0.05 190
vi Reducing uncertainty in stock status ABARES
Table 11.1 The criteria for status based on the biomass and fishing mortality, in line with HSP 201 Table 11.2 Catch trends and potential implications for status 206 Table 11.3 Size structure and potential implications for status 207 Table 11.4 Age structure and potential implications for status 208 Table 11.5 Effort trends and potential implications for status 209 Table 11.6 Spatial trends in the fishery area and potential implications for status 209 Table 11.7 Spatial closures and potential implications for status 210 Table 11.8 Catch per unit effort trends and potential implications for status 211 Table 11.9 Surveys and potential implications for status 211 Table 11.10 Risk assessments and potential implications for status 212
Table B1 Area of all habitat types in the Coral Sea Fishery 54
Table C1 Proportions of species catch in the Coral Sea Fishery 91 Table C2 Coral Sea Fishery bathymetric contour lengths (nm) and depth strata areas (ha) 94 Table C3 Previous studies on species deep scalefish assemblage biomass within linear habitat 95
Table D1 Summary of parameter estimates from posterior densities of delay difference models with β = 1, β = 0.5 and β = 0.25, assuming a natural mortality of M = 0.6 year-1 126 Table D2 Summary of parameter estimates for sample size 12 000 from posterior densities of delay difference models with β = 1, β = 0.5 and β = 0.25, assuming a natural mortality of M = 0.4 year-1 127 Table D3 Summary of parameter estimates for sample size 12 000 from posterior densities of delay difference models with β = 1, β = 0.5 and β = 0.25, assuming a natural mortality of M = 0.8 year-1 127
Table E1 North West Slope Trawl Fishery, total catch by species (modelled), catch per unit effort indexes by species and total effort 144 Table E2 North West Slope Traw Fishery Zone Statistics: Total scampi catch (logbook), total effort (raw within the zone) and catch per unit effort indexes 145
Table F1 Elephant fish catch per unit effort and catch indexes 157 Table F2 Elephant fish gillnet and longline catch (t) by Southern and Eastern Scalefish and Shark Fishery shark sub-regions 158 Table F3 Elephant fish gillnet and longline catch (t) by Southern and Eastern Scalefish and Shark Fisheryshark gear type 159
vii Reducing uncertainty in stock status ABARES
Table F4 Elephant fish catches (t) and standardised catch per unit effort used in Tier 4 estimation 160
Figures
Figure 1.1 Biomass status (number of stocks), 2004 to 2012 2 Figure 1.2 Fishing mortality status (number of stocks), 2004 to 2012 3 Figure 2.1 Proportional catch for key commercial families between 200203 and 200809 18 Figure 3.1 The number of daily operations in the CSF Hand Collection Sector between 1997–98 and 2008–09 32 Figure 3.2 Total catch of key species by maximum depth field recorded in logbooks 33 Figure 3.3 Dive hours for the two hand collection methods and total catch 34 Figure 4.1 Coral Sea Fishery line sector annual financial year catch by fishing method, 1997–98 to 2008–09 61 Figure 4.2 Coral Sea Fishery line sector annual effort by fishing method, as (a) numbers of operations (records within the Australian Fisheries Management Authority log database) and (b) number of hooks set 64 Figure 4.3 Percentage of hooks set by depth class for each line method 65 Figure 4.4 Coral Sea Fishery trap sector catch and effort by depth of set 67 Figure 4.5 Annual catch of the deep scalefish assemblage in the combined Coral Sea Fishery line and trap sectors by financial year 74 Figure 4.6 Biomass estimates of the deep scalefish assemblage on the 200 metre isobath at island and seamounts/oceanic reef locations 75 Figure 4.7 Annual (financial year) catch of the reef scalefish assemblage in the Coral Sea Fishery Line and Trap Sector 79 Figure 4.8 A comparison of the average density of coral trout species per reef from the northern Great Barrier Reef, Coringa-Herald and Lihou Reef national nature reserves 80 Figure 4.9 Density of all serranids and P. laevis across three regions 81 Figure 4.10 Mean annual catch density of all demersal species from Effects of Line Fishing Experimental reefs zoned General Use B or Marine National Park B in each region and year since 1990 83 Figure 4.11 Annual catch of the shark assemblage in the Coral Sea Fishery 88 Figure 4.12 Annual catch of shark by taxa in the Coral Sea Fishery 89 Figure 5.1 Total annual catch and effort for tropical lobster in the Coral Sea Hand Collection Fishery 102 Figure 5.2 Time series of unstandardised catch per unit effort for tropical rock lobster in the Coral Sea 103 Figure 5.3 Flowchart for tropical lobster overfished (biomass) status determination. 105
viii Reducing uncertainty in stock status ABARES
Figure 5.4 Flowchart for tropical lobster overfishing (fishing mortality) status determination 105 Figure 6.1 Number of boats fishing in the North West Slope Trawl Fishery by financial year 110 Figure 6.2 Total combined reported prawn and combined scampi catch in tonnes and total effort in thousands of trawl hours by year in the North West Slope Trawl Fishery 110 Figure 6.3 Importance plot for random forests model predicting the probability of catching prawns on a given shot in the North West Slope Trawl Fishery, 1986 to 1993 112 Figure 6.4 Estimated smooths of variables in generalised additive model for conditional catch per unit effort 113 Figure 6.5 Standardised prawn catch per unit effort (shots reporting prawn catch) with a five quarter moving average line 114 Figure 6.6 Quarterly catch (tonnes, retained) and index of abundance, North West Slope Trawl Fishery deepwater prawns, 1986 to 1993 115 Figure 6.7 Graphical representation of suggested power law relationship between catch per unit effort and abundance 117 Figure 6.8 Prior distributions (dotted lines) and posterior distributions (solid lines) for estimation of each of specific quarterly recruitment rate, r,
carrying capacity, B0, observation error variance, τ2, and process error variance σ2 for the base case (M = 0.6, β = 1) 120 Figure 6.9 Time series plot of quantiles of posterior distribution of deepwater prawn biomass as a proportion of unfished biomass as estimated by quarterly delay difference model (M = 0.6, β = 1) 121 Figure 6.10 Predicted time series of deepwater prawn standing stock as a proportion of virgin biomass, M = 0.6 year-1, β = 0.5 122 Figure 6.11 Predicted time series of deepwater prawn standing stock as a proportion of virgin biomass, M = 0.6 year-1, β = 0.25. 122 Figure 6.12 Predicted time series of deepwater prawn standing stock as a proportion of virgin biomass, M = 0.8 year-1 123 Figure 6.13 Predicted time series of deepwater prawn standing stock as a proportion of virgin biomass, M = 0.4 year-1 124 Figure 6.14 Recovery trajectory of deepwater prawn standing stock as a proportion of virgin biomass 125 Figure 7.1 Total Metanephrops spp. catch by zone and total North West Slope Trawl Fishery effort 135 Figure 7.2 Combined scampi catch with accompanying base case catch per unit effort index (solid line) and catch per unit effort indexes that assume 1 per cent and 2 per cent annual efficiency increase 139 Figure 7.3 Base case (run 1) production model, fit to catch per unit effort index (1985 to 2011) 141
Figure 7.4 Base case (run 1) production model, biomass relative to BMSY and fishing mortality relative to FMSY 141
ix Reducing uncertainty in stock status ABARES
Figure 8.1 Elephant fish catch index, and raw and standardised catch per unit effort using the basic subset data (1980 to 2008) 153 Figure 8.2 Elephant fish catch index, and raw and standardised catch per unit effort series for the Bass Strait only (1980 to 2008) 154 Figure 8.3 Elephant fish catches (1976–2008) and reference period catch target levels 156 Figure 9.1 Probability of catching sawshark (base case, 1980–2008) 166 Figure 9.2 Sawshark base case standardised catch per unit effort index, catch index and raw catch per unit effort index using the basic subset data (1980– 2008) 167 Figure 9.3 Comparison of sawshark standardised catch per unit effort series (1980–2008) 168 Figure 9.4 Sawshark catches and reference period target catch levels 170 Figure 10.1 Length frequency data collected by Australian Fisheries Management Authority observers on foreign jig fishing records for 1980 (in the Bass Strait) and 1984 (in the Bass Strait and Central Zones) 180 Figure 10.2 Regression lines for each year’s weight (g) data 182 Figure 10.3 Regressions of weekly catch per unit effort against cumulative catch for each year (natural mortality rate fixed at 0.05 per week) 184
Figure 10.4 The effect of data truncation on estimates of initial stock size (N20) 185 Figure 10.5 a) Estimated numbers during the fishing season for each season (natural mortality rate fixed at 0.05 per week) 186
Figure 10.6 a) Initial stock size (N20) plotted by year 187 Figure 10.7 a) Exploitation rate plotted by year 188 Figure 10.8 Stock-recruitment relationship 190 Figure 10.9 a) Estimates of initial stock size 191 Figure 11.1 Diagram of the weight-of-evidence decision-making framework for status determination 203
Figure H1 Catch (kg) data for the SSJF and Commonwealth trawl fisheries in the Central Zone combined, 1995–2006 195 Figure H2 Nominal Southern Squid Jig Fishery catch per unit effort in the Central Zone data, 1995–2006 196
Maps
Map 2.1 Management area of the Coral Sea Fishery Aquarium Sector 15 Map 3.1 Extent of the Coral Sea Fishery 29 Map 4.1 Management area of the Line and Trap, and Trawl and Trap sectors of the Coral Sea Fishery 59
x Reducing uncertainty in stock status ABARES
Map 4.2 Map of 50 metre interval bathymetric strata adjacent to the Queensland east coast (Cairns to Bundaberg) 72 Map 4.3 Map of reef outer boundaries extending some 10–20 nautical miles from the reef 73 Map 5.1 Map of the Coral Sea Fishery 101 Map 6.1 North West Slope Trawl Fishery and recent relative fishing intensity 109 Map 7.1 Area of the North West Slope Trawl Fishery and recent relative fishing intensity 131 Map 7.2 Scampi zones used for spatially disaggregated analyses 134 Map 8.1 Relative fishing intensity in the a) Shark Gillnet and b) Shark Hook Sectors of the Southern and Eastern Scalefish and Shark Fishery, 2008 149 Map 8.2 Shark areas within the Southern and Eastern Scalefish and Shark Fishery 151 Map 9.1 Relative fishing intensity in the a) Shark Gillnet and b) Shark Hook sectors of the Southern and Eastern Scalefish and Shark Fishery, 2008 163 Map 9.2 Shark areas within the Southern and Eastern Scalefish and Shark Fishery 165 Map 10.1 Map of south-eastern Australia showing the distribution of squid catches in 2010 and the Bass Strait Central Zone, which was the subject of the current assessment 177
xi 1 Overview
Fisheries managers need to understand the biological status of the fish stocks and the economic status of the fishery. This is central to adaptive management, where assessment and reporting on status provides feedback on the effectiveness of current management approaches. Where necessary, managers can modify their approaches to ensure fisheries are achieving their objectives.
The Australian Fisheries Management Act 1991 (FMA) outlines the objectives for Commonwealth fisheries, which include ensuring ‘that the exploitation of fisheries resources and the carrying on of any related activities are conducted in a manner consistent with the principles of ecologically sustainable development’. In line with this, the Commonwealth Fisheries Harvest Strategy Policy 2007 (HSP) (DAFF 2007) directs that Commonwealth fisheries should be managed to pursue ‘the sustainable and profitable utilisation of Australia’s Commonwealth fisheries in perpetuity through the implementation of harvest strategies that maintain key commercial stocks at ecologically sustainable levels and within this context, maximise the economic returns to the Australian community’.
The status determinations are reported in the annual ABARES Fishery status reports. These document scientific and economic information for each Commonwealth fishery and continue the series of reports produced by the former BRS since 1992, when the Fisheries Management Act 1991 came into effect, and the Fishery economic status reports produced by the former ABARE in 2007 and 2008. The Fishery status reports provide governments, industry and the community with an independent overview of trends in the biological status of fish stocks and the economic status of fisheries for which the Australian Government has management responsibility. The reports also support the Australian Government’s directions within the HSP and provide stakeholders with information to help them meet the intent and purpose of the HSP.
Over time, the number of stocks considered in the Fishery status reports has generally increased, from 31 in 1992 to 101 in 2009 and 93 in 2013. As the number of stocks increased, the number classified as uncertain also increased, peaking at 52 stocks in 2007.
Lack of certainty in the biological status of a stock can hamper management and reduce confidence that the fishery is achieving its objectives. Uncertainty is often linked to low-value fisheries or stocks where there are limited resources for data collection and research. These fisheries often lack a formal stock assessment, which is usually the basis of status determination. Reducing uncertainty in the status of Commonwealth-managed fish stocks is critical to providing security to the fishing industry and confidence to the broader Australian community that fish stocks are being sustainably managed. Stocks and biological status
The 2007 Fishery status reports, the most recent report when the Reducing Uncertainty in Stock Status (RUSS) project commenced, covered 96 stocks, species or groups of species (all referred to as ‘stocks’). These stocks included those which were primary targets of fishing, had significant catches or were subject to a total allowable catch. Clearly separate sectors within a fishery, which target distinct species or suites of species, are differentiated. Some stocks comprise more than one species; for example, sawsharks (Pristiophorus spp.) in the Southern Eastern Scalefish and Shark Fishery (SESSF) comprise two species, and the Coral Sea Fishery Aquarium Hand Collection Sector takes hundreds of species but is treated as a single ‘stock’ in status report summaries. Reducing uncertainty in stock status ABARES
Biological status is assessed on two criteria based on the reference points in the HSP (DAFF 2007):
biomass of the stock (not overfished/overfished)—whether the biomass is adequate to sustain the stock in the long term and whether it is above the biomass limit reference point (BLIM) (DAFF 2007)
level of fishing mortality (not subject to overfishing/overfishing)—whether the stock is subject to a level of fishing mortality that would move the stock to an overfished state, or prevent recovery from an overfished state, and the amount of fishing does not exceed the limit reference point for fishing mortality (FLIM) (DAFF 2007).
Where there is inadequate information to determine status, the species is classified as uncertain with respect to either the biomass or fishing mortality status.
Of the 96 stocks assessed in the 2007 Fishery status reports, 45 stocks were classified as uncertain with respect to overfishing (fishing mortality) status, 52 stocks were classified as uncertain with respect to overfished (biomass) status and 39 stocks were classified as uncertain with respect to both overfishing and overfished status (Figure 1.1 and Figure 1.2).
Figure 1.1 Biomass status (number of stocks), 2004 to 2012 70 Not overfished 60 Uncertain B Overfished
50
40
30
Number ofstocks Number 20
10
0 2004 2005 2006 2007 2008 2009 2010 2011 2012
Source: Woodhams et al. 2013
2 Reducing uncertainty in stock status ABARES
Figure 1.2 Fishing mortality status (number of stocks), 2004 to 2012
90 Not subject to overfishing 80 Uncertain F
70 Subject to overfishing
60 50 40
30 Number ofstocks Number 20 10 0 2004 2005 2006 2007 2008 2009 2010 2011 2012
Source: Woodhams et al. 2013 Reducing uncertainty in stock status
The Reducing Uncertainty in Stock Status (RUSS) project was part of an expanded research programme to classify Commonwealth-managed species that were classified as uncertain. This report presents a series of stock assessments that were undertaken for species and species groups, particularly in smaller and lower-value Commonwealth fisheries during the RUSS project. Chapter 11 describes a weight-of-evidence framework for determining stock status from a structured review of scientific evidence and interpretation of indicators. A separate stream of work under the RUSS project involved management strategy evaluation (MSE) of several fishery's harvest strategies. This has been reported elsewhere (Dowling 2011; Haddon 2011; Klaer & Wayte 2011; Plagányi et al. 2011a, b). Reducing uncertainty in stock status assessments and papers
Nine assessments are presented in this report covering a substantial breadth of approaches and methods depending on the nature of the stocks and the data that were available. Some assessments cover multiple species or combined suites of species. The objective in all cases was to resolve the biomass and fishing mortality status of the stock, but this was not always possible. Several preliminary investigations of stocks were undertaken that did not proceed to fuller assessments or papers. These are also noted. In addition to the assessments, a weight-of- evidence framework was developed to facilitate status determination in a transparent manner. This framework uses a broader range of information sources. Coral Sea Fishery Aquarium Sector The Coral Sea Fishery (CSF) Aquarium Sector uses hand collection to take more than 600 coral reef aquarium finfish species for live trade. Given this diversity and the nature of the fishery- dependent data collected, traditional species-specific stock assessment methods were not considered appropriate. Therefore, three different assessment approaches were undertaken to build a weight-of-evidence for status determination:
3 Reducing uncertainty in stock status ABARES
A maximum footprint analysis estimated the total amount of reef area that operators could cover within a fishing season. The Aquarium Sector had a potential maximum footprint of around 7 per cent of the estimated suitable habitat area in a given fishing year, with a minimum of 93 per cent not exploited in any given season.
Annual extraction rates for key commercial families as a percentage of plausible total biomass were estimated. Extraction rates for the 2008–09 fishing season were estimated at less than 0.02 per cent of individuals from the total CSF population for all key commercial fish families. Under the present fishery effort management constraints, the maximum potential annual extraction rate is 0.04 per cent for any key commercial fish family.
A risk analysis was undertaken on 623 fish species based on vulnerability and susceptibility criteria. Eleven species were considered to be at medium risk of overexploitation after the vulnerability assessment but when assessed for susceptibility were all found to be at low risk of overexploitation.
Results from these analyses indicate that the Aquarium Sector is unlikely to have exerted a fishing pressure considered to be ‘overfishing’ and, as a result, the stock is unlikely to be overfished. Sea Cucumber Sector of the Coral Sea Fishery The Sea Cucumber Sector of the CSF uses hand collection to take a range of holothurian species. A habitat-based approach was employed to assess four species of sea cucumber: black teatfish (Holothuria whitmaei), prickly redfish (Thelenota ananas), white teatfish (Holothuria fuscogilva) and surf redfish (Actinopyga mauritiana). This assessment used coral reef habitat mapping along with historical surveys of sea cucumbers undertaken in two Nature Reserves. Together, these data were used to derive plausible scenarios of potential biomass of sea cucumber species. These were, in turn, used to estimate maximum sustainable yields (MSY) and to develop basic surplus production models.
Results indicated black teatfish and prickly redfish are not overfished and are not subject to overfishing at current catch levels. The deeper habitat of white teatfish was not well-sampled in the available surveys so its stock status remains uncertain. Surf redfish were present in low numbers in the surveys but were underestimated by the assessment and the stock is considered to be not subject to overfishing and not overfished.
The scarcity of data significantly impeded the ability of this study to produce reliable results for some species. Dedicated survey operations would substantially improve the robustness of outputs, but some results from this study may aid a revision of the harvest strategy. Line and trap sectors of the Coral Sea Fishery The combined CSF line and trap sectors take a suite of scalefish, sharks and crustaceans. Preliminary assessments were undertaken for three separate assemblages of species that comprise the combined line and trap sectors: a deep assemblage, a reef assemblage and a shark assemblage. The assessments used a habitat-based approach where biomass scenarios were developed from the published literature and yields were then estimated while considering the composition of the assemblage. Historic catch levels were then compared with the estimates of yield. The approach was specifically intended to be conservative (assuming lower biomass levels and exploitation rates) because of the lack of research information specific to the Coral Sea. This approach is distinct from a more data demanding approach that would give some confidence about what long-term maximum sustainable yields could be. A number of scenarios were developed where biomass and the constant used in the yield estimation equation were varied.
4 Reducing uncertainty in stock status ABARES
On the basis of these analyses, the deep and reef scalefish assemblages can be considered not subject to overfishing and not overfished up to and including the 2008–09 season. The same logic could not be applied to the shark assemblage because biomass scenarios could not be developed, largely because there was no estimate of sustainable harvest with which to compare historical catch of sharks in the fishery. As a result, the shark assemblage remains uncertain with regard to fishing mortality and biomass. The assessment results may be used to inform future harvest strategies, including development of trigger catch levels. Lobster and trochus in the Coral Sea Fishery Tropical rock lobster (Panulirus ornatus and P. versicolour) and trochus (Trochus niloticus and/or Tectus pyramis) have been taken sporadically and in low volumes from the CSF. The two species were subject to a brief analysis to establish likely status.
The status of the tropical rock lobster stock in the Coral Sea was considered under two alternative scenarios. The first scenario assumed that the Coral Sea population is part of a larger population with spawning grounds in the Torres Strait. The second scenario assumed that the Coral Sea population is a self-recruiting, self-sustaining population. Results from both scenarios suggest that the stock is not overfished and is not subject to overfishing.
The history of commercial exploitation of trochus in the Coral Sea Hand Collection Fishery is limited to a single trip to a single reef in 2001. As a result of negligible exploitation, trochus in the Coral Sea is classified as not overfished and not subject to overfishing. Deepwater prawns in the North West Slope Trawl Fishery The deepwater prawn stock in the North West Slope Trawl Fishery (NWSTF) comprises four species of penaeid and two species of carid prawn. For management purposes, deepwater prawns are considered a single stock, which was actively targeted between 1986 and 1993 but not in more recent years.
Logbook catch rates for deepwater prawns during the target years were standardised to estimate a relative abundance index. A delay difference model implemented within a Bayesian framework was developed for predicting the current condition of the stock. Life history information for Aristaeomorpha foliacea (the most commercially important species) was used to inform prior distributions, together with logbook records of prawn landings up to 2008. The sensitivity of the model to uncertainty in natural mortality was tested and the possible effect of hyperstability (maintenance of high catch rates when abundance declines) in the index of abundance was also explored.
The scenarios explored suggest that the stock is unlikely to be below 20 per cent unfished biomass and therefore not overfished. The catch and effort data were not sufficiently informative for the carrying capacity or absolute biomass to be estimated with reasonable precision. Recent low levels of catch are unlikely to constitute overfishing. Scampi in the North West Slope Trawl Fishery Scampi are the current target catch of the North West Slope Trawl Fishery (NWSTF), comprising predominantly Metanephrops australiensis, M. velutinus and M. boschmai. This study aimed to estimate the NWSTF scampi unfished biomass and assess the status of the stock.
Logbook catch rates for scampi were standardised to estimate a relative abundance index, then surplus production models were fitted to estimate quantities of management interest. The models estimated various parameters describing historic abundance and exploitation rates. The
5 Reducing uncertainty in stock status ABARES base case assessment modelled the combined population of scampi in the entire area where the NWSTF operates. Alternative scenarios were considered that model discrete areas within the fishery separately.
Trends in the ratios F/FMSY and B/BMSY for the base case indicate heavy initial exploitation but combined stock biomass has remained above BMSY and F has remained below FMSY (except in 1988) for the entire history of the fishery. These results from the production model indicate that the stock is not overfished and not subject to overfishing. Evidence of faunal changes in heavily trawled areas of the North West Slope (Wallner & Phillips 1995) means that the MSY estimates should not be interpreted as an opportunity to significantly increase catch. The high-cost nature of the fishery means that profitable catch rates require biomass levels well above BMSY. Elephant fish CPUE standardisation and Tier 4 assessment, 2009 Elephant fish (Callorhinchus milii) are commonly caught as by-product in the SESSF. This assessment comprised a Tier 4 analysis that was used to derive recommended biological catches in accordance with the SESSF harvest strategy. This requires an index of abundance derived from catch rates, selection of reference periods and application of a harvest control rule.
A delta model (also called a two-stage model) was used to derive an index of abundance from logbook catch and effort data. The abundance index encompassed 1980 to 2008 and showed considerable variability, but indicated a 20–40 per cent decline in abundance over that period.
Two reference periods were examined for applying the Tier 4 harvest control rule: 1980 to 1992 and 1998 to 2004. Results from 1980 to 1992 are not reliable because of the likely underestimation of trawl catch in this period and the variability in the catches and the abundance index. The 1998 to 2004 period resulted in a catch target of 109.7 tonnes and a recommended biological catch of 122.8 tonnes. The current abundance index is close to the target from this reference period, which indicates the stock is not overfished. There are some uncertainties because of lack of data on catch and discards from the trawl sector in the earlier years, unknown levels of discarding and the size of the recreational catch over time (Wilson et al. 2009). Results should be treated with caution because of the uncertainties regarding historic catch. Sawshark catch rate standardisation and Tier 4 assessment, 2009 Sawsharks (Pristiophorus cirratus, P. nudipinnis) are commonly caught as by-product in the SESSF. This assessment comprised a Tier 4 analysis that was used to derive recommended biological catches in accordance with the SESSF harvest strategy. This requires an index of abundance derived from catch rates, selection of reference periods and application of a harvest control rule. Common sawshark and southern sawshark are combined in these analyses because these two species were not differentiated in historical catch records. Fishery-dependent logbook records contain a high number of zero sawshark catches. A delta model was used to derive an index of abundance from these data.
The abundance index, which encompasses the period 1980 to 2008, implies a relatively consistent 40–50 per cent decline in sawshark abundance. Two reference periods were examined to apply the Tier 4 harvest control rule (1986 to 2001 and 2002 to 2008); however, there was no clear time period with a stable abundance index and stablecatch which would be desirable. The assessment is quite uncertain due to the lack of species-specific catch data and variable catch, and a lack of information on historical catch and discard levels. The steadily declining abundance index is a concern but no conclusion was reached on the status of sawshark.
6 Reducing uncertainty in stock status ABARES
Depletion analyses of Gould’s squid in Bass Strait Gould’s squid (Nototodarus gouldi) is taken by the Southern Squid Jig Fishery and by the Commonwealth Trawl Sector of the SESSF. This assessment used a modified depletion analysis that includes natural mortality and growth to estimate the abundance of Gould’s squid in the Central Zone of Bass Strait from 1995 to 2006. The purpose was to identify trends over seasons and to determine whether an impact of fishing could be detected across these seasons. Standardised catch per unit effort from the jig fishery logbook data was used as an index of abundance and total catches for the jig and trawl sectors were included in the analysis. The effects on the assessment of using a range of natural mortality values from similar squid stocks were explored.
The results suggest pre-season stock numbers of the order 107 animals and a proportional escapement of less than 40 per cent in about half the years considered, indicating considerable fishing impact in those years. Trends suggesting a slight overall stock decline between 1995 and 2006 were also detected. These conclusions strongly depend on key assumptions about natural mortality rates, recruitment, migration and growth of Gould’s squid. The analyses highlight that, if overfishing occurred within a fishing season, current arrangements for data collection and analysis would not allow management actions to be implemented in time to prevent squid biomass falling to unacceptable levels. Regular monitoring of the size composition of catches would greatly improve the reliability of the assessment and would allow the fishery’s harvest strategy to be effectively implemented. Preliminary analyses of selected stocks A preliminary investigation of deepwater bugs (Ibacus spp.) in the Western Deepwater Trawl Fishery was undertaken. Catch and effort data for deepwater bugs at four discrete concentration sites along the 200 metre isobath were analysed using depletion methods aimed at estimating catchability and abundance independently at each site during the period 2002 to 2004. Initial results indicated abundance declines at two sites, no trend at one site and an increase at one site. However, the approach was not considered to be robust so was not pursued further.
A preliminary review of striped marlin (Tetrapturus audax) status in the Eastern Tuna and Billfish Fishery was undertaken. This involved an attempt to reinterpret the south-west Pacific stock assessment (Langley et al. 2006) following previously uncertain biological parameters for striped marlin being resolved(Keller Kopf & Davie 2009)—particularly growth and natural mortality. However, this was not pursued because the complexity of the integrated assessment model meant that the model would need to be re-fit to the new data to give confidence in the status outcomes. A new assessment was subsequently undertaken in 2012 (Davies et al. 2012).
A brief study was undertaken to assess the status of the tropical rock lobster (Panulirus ornatus and P. versicolour) in the Coral Sea. Status of the tropical rock lobster was considered under two alternative stock structure scenarios (isolated Coral Sea and common with Torres Strait). Under either scenario the Coral Sea population was determined to be not overfished and not subject to overfishing (Woodhams et al. 2012). Weight-of-evidence framework Stock status determination and reporting is undertaken in most Australian states and territories, in other nations and by regional fisheries management organisations. This report describes a weight-of-evidence decision-making framework to help determine the status of some 100 fish stocks managed by the Commonwealth. The development of this framework recognised that not all stocks have formal stock assessments, but information may be available that can provide
7 Reducing uncertainty in stock status ABARES indicators of stock status. The framework’s objective is to provide for a structured scientific review and interpretation of indicators of biomass and fishing mortality and to arrive at a status determination through the cumulative weight of the evidence available. The framework aims to be expansive and inclusive in the types of evidence that could be considered and provide a systematic approach to evaluating that evidence. It is intended to provide a transparent and repeatable process especially for data or information poor stocks.
The framework acknowledges the role of expert judgement in determining status but emphasises documenting the key evidence and the rationale for the decision. The decision- making process is undertaken separately for biomass (overfished status) and fishing mortality (overfishing status). Discussion and conclusions
Following the commencement of the RUSS project (1 July 2008), the number of stocks classified as uncertain with respect to fishing mortality and biomass decreased considerably (Figure 1.1 and Figure 1.2).
Between 2008 and 2012 the number of stocks with an uncertain biomass status fell from 41 to 21, with most of these reclassified as not overfished (Figure 1.1). This was partly attributable to work under the RUSS project (assessments, MSE and adoption of the weight-of-evidence framework) and partly to new and updated assessments through the Australian Fisheries Management Authority’s (AFMA’s) research programme. Notable stocks where RUSS research directly contributed to resolving biomass status are: sea cucumbers in the CSF (4 stocks) and Torres Strait (2), CSF Rock Lobster and Aquarium sectors (1), NWSTF scampi (1) and prawns (1) and SESSF western gemfish (1).
In terms of fishing mortality status, between 2008 and 2012 the number of stocks that were uncertain fell from 33 to 12, with most of these stocks reclassified as not subject to overfishing (Figure 1.2). Again this was partly attributable to work under the RUSS project (assessments, MSE and adoption of the weight-of-evidence framework) and partly to new and updated assessment information through AFMA’s research programme. Notable stocks where RUSS research made a direct contribution to resolving fishing mortality status are: sea cucumbers in the CSF (4) and Torres Strait (3), CSF Aquarium/Line and Trap/Trawl and Trap sectors (3), NWSTF scampi (1) and prawns (1), SESSF western gemfish (1) and Torres Strait trochus (1). Fishing pressure relationship to information and status
The Commonwealth HSP (DAFF 2007) notes the concept of investment in information in the development of risk management policies. This concerns the appropriate amount of information decision-makers need to make informed judgments on whether risks to the stock from fishing are acceptable. While more information is desirable, there is a risk-cost-reward trade-off that must be made about the quantity and quality of information required to ensure limits are not breached and the likely cost of risk protection. The HSP further notes that where stock status is very uncertain harvesting levels (and targets) may be similarly precautionary.
In general, the determination of fishing mortality status was a more tractable question than biomass status. For many stocks it was easier to demonstrate that recent levels of fishing (and exploitation rates) had been low or negligible with respect to plausible levels of biomass for a particular stock. The extreme case of this was when catches had been zero or close to zero (for example, CSF rock lobster). In the case of the smaller fisheries managed by the Commonwealth, it was apparent that exploitation levels had been relatively light or moderate in recent years
8 Reducing uncertainty in stock status ABARES
(including all sectors of the CSF, NWSTF, Small Pelagics Fishery and the Southern Squid Jig Fishery).
If a deliberately conservative approach to assessment assumptions is taken in data poor circumstances, this can instil some confidence if a conclusion of not overfished or not subject to overfishing is reached (for example, the Coral Sea Fishery Line and Trap Sector). However, the same confidence does not exist for assessment outcomes suggesting overfishing or overfished status. This deliberately conservative approach is distinct from a more data-intensive approach that would give some confidence about what long-term maximum sustainable yields from the stock could be. In the studies undertaken by the RUSS project it was more difficult to resolve status for stocks where it was apparent that they may be subject to relatively high levels of fishing pressure (mortality) given their biological characteristics or where there was substantial uncertainty around this. In these cases the evidentiary burden is higher (noting the comments made earlier in this section on the HSP) and the assessment methods applied require a greater degree of rigour to determine status.
Data poor Most of the stocks examined were from smaller-scale fisheries or of low volume/value and would probably be termed ‘data poor’ by the standards of Commonwealth fisheries. However, a surprising amount of useful information was available, particularly when a sufficiently broad palette of assessment approaches was considered. In most, but not all, of these cases it was possible to advance our understanding of stock status to some extent through targeted studies. The risk-cost-reward trade-off is again relevant because the low or moderate exploitation rates in some fisheries place lower demands on data and assessments. Basket stocks
Basket stocks are those that contain multiple species which, for management reasons, are assessed together. In the RUSS studies these ranged from stocks containing two species (sawsharks in the SESSF) to stocks containing hundreds of species (Coral Sea Fishery Aquarium Sector and Line and Trap Sector). For sawsharks in the SESSF, the presence of two species in the catch with no means of discriminating them in the available logbook data was a major impediment to resolving status (along with uncertainty around earlier catches). The status of the shark assemblage in the Coral Sea Fishery was also unresolved, in part because of the great variety of shark species (and life histories) present in the catch. Sharks presented a significant challenge for assessment because of their biology and a lack of information to inform assessments (compared with scalefish).
A risk assessment approach, supported by two other approaches, was applied to the Coral Sea Aquarium Sector which was successful in demonstrating low risk across the large basket of species taken. In the Coral Sea Line and Trap Sector, suites of species were combined and assessment was undertaken on the assemblage as a whole, while accounting for the life history characteristics of the taxa within the assemblages (and hence their robustness to fishing). Management strategy evaluation
A separate stream of work under the RUSS project involved management strategy evaluation (MSE) of the harvest strategies in place in an array of different, predominately smaller and low value fisheries and their likely capacity for achieving the requirements of the Commonwealth HSP (including avoiding limits and achieving targets). The work for each of these fisheries has been reported elsewhere and Haddon (2012) provides a complete summary. Some of the findings are summarised here.
9 Reducing uncertainty in stock status ABARES
The five fisheries considered were the Bass Strait Central Zone Scallop Fishery (Haddon 2011), the Southern and Eastern Scalefish and Shark Fishery (Klaer & Wayte 2011), the North West Slope Trawl Fishery (Dowling 2011), the Torres Strait Bêche-de-mer Fishery (Plagányi et al. 2011a) and the Sea Cucumber Sector of the Coral Sea Fishery (Plagányi et al. 2011b).
For the fisheries considered, each of the MSEs demonstrated that, after making certain assumptions and noting some limitations, the harvest strategies had the capacity to achieve some aspects of the Commonwealth Harvest Strategy Policy. The harvest strategies tended to perform better on the sustainability objective of the policy that requires the avoidance of excessive biomass depletion (to below the limit reference point). However, the policy objective to maximise economic returns (through maximum economic yield) was more difficult to measure and so performance of the harvest strategies on this objective was not well demonstrated across the fisheries.
For harvest strategies in small and low value fisheries to be effective, commercial catches of key species need to be adequately monitored and resources need to be available to appropriately collect data and produce analysis in a timely fashion. Any positive findings of the management strategy evaluations depend on this action. North West Slope Trawl Fishery MSE Using scampi as a case study, Dowling (2011) evaluated the harvest strategies of the North West Slope Trawl and the Western Deepwater Trawl fisheries. The North West Slope Trawl Fishery harvest strategy consists of trigger catch levels linked to a scaled assessment and potential management responses. This approach recognises the trade-off between risk, cost and catch because the investment in information and science increases as the catch (and exploitation rate) increases.
Most of the scenarios examined successfully maintained or recovered the spawner biomass above the limit reference point, well over 90 per cent of the time as required by the policy, except when the population had been heavily fished historically.
Trigger-based harvest strategies such as this can be effective provided the triggers are correctly specified and are related to the size of the stock and its productivity. If the fishery is small relative to the size of the resource at the lower catch triggers, the fishing mortality is assumed to also be small. Demonstrating this assumption is important (for example, see the Coral Sea Fishery Line and Trap Sector assessment in this volume). Catch history is not necessarily a good predictor of this.
The research found that a trigger-based harvest strategy (with its stepped approach to assessments and management action) can create high volatility in recommended annual catches. This is an undesirable feature of a harvest strategy and options for minimising this are examined. Bass Strait Central Zone Scallop Fishery MSE Haddon (2011) assessed the performance of the harvest strategy for the Bass Strait Central Zone Scallop Fishery. The harvest strategy in place in 2011 required that viable beds be identified through surveys and a proportion of those beds be closed to fishing (reserving a minimum of 500 tonnes) before fishing would be permitted to occur on the balance.
10 Reducing uncertainty in stock status ABARES
The MSE used an agent-based operating model containing 25 beds (populations) that were essentially treated as separate entities. In any year a bed may contain scallops based on probabilities for a recruitment event, magnitude of the recruitment, growth rates and survival.
The harvest strategy was able to maintain the exploitable biomass (across all beds) above the 500 tonne reserve level with high probability (94 per cent). If this is regarded as a proxy for the limit reference point then the harvest strategy appears to be successful at achieving the sustainability objective of the policy that requires the avoidance of excessive biomass depletion.
The variable nature of the simulated scallop populations and the decision rules contained in the harvest strategy meant there would often be years when no fishery could occur. The probability of there being a fishery every year varied from 0.25 to 0.66. However, it was also found that if the scallop beds in the Tasmania and Victoria fisheries were to be managed together with the Commonwealth fishery the overall performance would be greatly enhanced.
Haddon (2011) contends that, because of the variable nature of Bass Strait scallop populations, the usual equilibrium concepts are inappropriate. This may undermine the development of depletion-based reference points that reference an unfished biomass (B0) and the notion that if a target fishing mortality is applied then the stock will move to a stable biomass level. Southern and Eastern Scalefish and Shark Fishery MSE Klaer and Waite (2011) examined stock status uncertainty and the performance of the tiered harvest strategy framework in the Southern and Eastern Scalefish and Shark Fishery (SESSF).
Following testing and evaluation, all current SESSF harvest strategies were found to achieve the HSP objective of maintaining stocks above the biomass limit reference point more than 90 per cent of the time.
In addition to the current four tiers contained in the SESSF harvest strategy framework, some additional harvest strategies (assessment method plus associated harvest control rule) were developed and tested. An average fish length harvest strategy was developed that included specific target and limit reference points designed to achieve policy objectives. This was found to perform acceptably well for typical SESSF demersal trawl species that had relatively high productivity and a good knowledge of the length-at-age relationship. Another harvest strategy that used a production model was developed and tested. This was found to be effective provided that data (catch and catch rates) have sufficient contrast so as to be informative on status (a recognised limitation of production models). Torres Strait Bêche-de-mer Fishery MSE Plagányi et al. (2011a) assessed the performance of the harvest strategy used for the Torres Strait Bêche-de-mer Fishery. The harvest strategy consists of size limits and species-specific, total allowable catches derived from abundance survey results (through a fixed exploitation rate).
The MSE testing was undertaken separately for eight species of sea cucumber. A spatial and age- structured operating model was developed that included sea cucumber habitat across 27 reef units of the Torres Strait and simulated the population trajectories of all species in all units.
The work drew a number of conclusions for each of the species about the current total allowable catch and the size limit and their consistency with the HSP. It suggested that sandfish, despite the present zero allowable catch, may not recover in the short term even in the absence of fishing and that a larger size limit may be more appropriate for this species. It suggested that
11 Reducing uncertainty in stock status ABARES black teatfish and surf redfish (currently also at zero allowable catch) could sustain small experimental quotas without unduly increasing risk but there was an ongoing risk of localised depletion in some locations. The current total allowable catches (TACs) for white teatfish and prickly redfish were found to perform well in controlling the risk of overfishing.
It was suggested that an improved harvest strategy could be more spatially explicit and allow for the closure of only locally depleted zones rather than the entire fishery for a species. The MSE framework has clear value for helping develop new management approaches for Torres Strait sea cucumber. Coral Sea Fishery, sea cucumber MSE Plagányi et al. (2011b) assessed the performance of the rotational harvest strategy used for the Coral Sea Fishery Sea Cucumber Sector. The harvest strategy consists of a three-year rotation of harvest across the Coral Sea reefs along with fixed total allowable catches and size limits. The MSE testing was undertaken separately for eight species of sea cucumber using an operating model with similar structure to that used for the Torres Strait work (described in the Torres Strait Sea Cucumber Fishery MSE section). As it is largely a data poor fishery, the operating model was constructed to readily represent a large number of uncertainties, notably alternative hypotheses on recruitment.
The testing found that the spatial rotational harvest strategy resulted in a substantial decrease in the risk of local (reef) and overall depletion for sea cucumbers.
The species predicted to be most at risk under the current harvest strategy was surf redfish because the TAC was too large for the primary single source reef and catch needed to be spread among the other reefs and zones. The prickly redfish TAC appeared to be appropriate but could be problematic if catches are not sufficiently distributed spatially. The black teatfish and white teatfish TACs were found to be low enough to pose minimal risk to the resource.
Given the remoteness and large geographical area of the Coral Sea, the fleet dynamics and location choice settings of the fishery simulation model were found to be important. Fishery location models based on profitability resulted in much heavier depletion of reefs close to the major ports because of large travel distances to some reefs in the Coral Sea.
The risks to sea cucumber stocks are substantially reduced because of the relatively large area of reef protected by the Lihou Reef and Coringa-Herald national nature reserves. The biomass in these closed areas was explicitly included in the operating model. References
DAFF 2007, Commonwealth Fisheries Harvest Strategy Policy and Guidelines, Department of Agriculture, Fisheries and Forestry, Commonwealth of Australia, Canberra.
Davies, N, Hoyle, S & Hampton, J 2012, ‘Stock assessment of striped marlin (Kajikia audax) in the southwest Pacific Ocean’, working paper WCPFC-SC8-2012/SA-WP-05, WCPFC Scientific Committee eighth regular session, Busan, Republic of Korea, 7–15 August.
Dowling, N 2011, Management strategy evaluation testing of the management strategies used with North West Slope Trawl Fisheries, CSIRO Marine and Atmospheric Research Report.
12 Reducing uncertainty in stock status ABARES
Haddon, M (ed.) 2012, Reducing uncertainty in stock status: harvest strategy testing, evaluation, and development—general discussion and summary, CSIRO Marine and Atmospheric Research Report.
Haddon, M 2011, Management strategy evaluation testing of the management strategies used with South-Eastern Scallop Fisheries, CSIRO Marine and Atmospheric Research Report.
Klaer, N & Wayte, S 2011, Demersal MSE for trawl fish in the Southern and Eastern Scalefish and Shark Fishery and other like-species, CSIRO Marine and Atmospheric Research Report.
Keller Kopf, R & Davie, P 2009, Population biology and habitat preferences of striped marlin, Kajikia audax in the southwest Pacific Ocean, report for NSW Fisheries and the Australian Fisheries Management Authority, Canberra.
Langley, A, Moloney, B, Bromley, D, Yokawa, K & Wise, B 2006, ‘Stock assessment of striped marlin (Tetrapturus audax) in the south west Pacific Ocean’, working paper WCPFC-SC2- 2006/SA WP-6, Scientific Committee Second Regular Session, Philippines, 7–18 August.
Plagányi, E, Skewes, T, Dowling, N & Haddon, M 2011a, Evaluating management strategies for data‐poor sea cucumber species in the Coral Sea Fishery, CSIRO Marine and Atmospheric Research Report.
Plagányi, E, Skewes, T, Dowling, N, Haddon, M, Woodhams, J, Larcombe, J & Chambers, M 2011b, Evaluating management strategies for data‐poor sea cucumber species in the Coral Sea Fishery, CSIRO Marine and Atmospheric Research Report.
Wallner, BG & Phillips BF 1995, Development of a trawl fishery for deepwater metanephropid lobsters off the northwest continental slope of Australia: designing a management strategy compatible with species life history, ICES Marine Science Symposium, vol. 199, pp. 379–90.
Wilson, D, Curtotti, R, Begg, G & Phillips, K (eds) 2009, Fishery status reports 2008: status of fish stocks and fisheries managed by the Australian Government, Bureau of Rural Sciences & Australian Bureau of Agricultural and Resource Economics, Canberra.
Woodhams, J, Vieira, S & Stobutzki, I (eds) 2013, Fishery status reports 2012, Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra.
Woodhams, J, Vieira, S & Stobutzki, I (eds) 2012, Fishery status reports 2011, Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra.
13 Reducing uncertainty in stock status ABARES 2 Coral Sea Fishery: Aquarium Sector assessments
Adam Leatherbarrow and James Woodhams Summary
The Aquarium Sector stock of the Coral Sea Fishery (CSF) has been classified as uncertain for both biomass and fishing mortality in the ABARES Fishery status reports since 2006. The sector is complex, comprising more than 600 aquarium finfish species, and has limited data available to fishery managers and scientists. The sector is managed primarily through input controls and spatial closures.
Given the diversity of species taken and the nature of the fishery dependent and independent data collected, standard, species-specific stock assessment methods were not appropriate for this sector. Alternative quantitative and qualitative methods were explored that, when considered in a weight-of-evidence context, can give some confidence about stock status. Three approaches were selected to examine the potential impact of the fishery on target stocks:
A maximum footprint analysis, which aimed to estimate the total area operators could cover within a fishing season. This analysis combined estimates of coral reef area in the fishery with operation information to estimate the potential footprint that can be covered by the sector, under the constraints of the current management arrangements.
Estimation of annual extraction rates for key commercial families. This method estimated plausible standing stock numbers by multiplying the total area of suitable habitat in the fishery by a range of published density estimates for each of the key commercial fish families. These standing stock estimates were then compared with 2008–09 catches (chosen as they were the highest recorded at the time of analysis) to estimate extraction rates.
Species-specific risk analysis, based on vulnerability and susceptibility. This method calculated the overall risk of overexploitation by the collection fishery, at the species level. The overall risk to a species was assumed to be a combination of its vulnerability to exploitation and its susceptibility to overexploitation.
Under the current harvest strategy for the Aquarium Sector, the potential maximum footprint achievable by the fishery was around 7.4 per cent of the estimated suitable habitat area in a given fishing year. As a result, a large proportion (more than 92 per cent) of the suitable habitat is not exploited in any given season. Although this is a low area of potential annual coverage, the potential of localised depletion at the reef or sub-reef level should be considered and analysed further.
The estimated extraction rate for the 2008–09 fishing season for all key commercial families was less than 0.02 per cent of potential population size. If effort is scaled up to represent the potential maximum effort in the fishery, the maximum potential annual extraction rate possible in the sector would be approximately 0.04 per cent of potential population size. Because this analysis is conducted at the family level, the potential for depletion of individual species within these family groups should be considered.
The risk assessment considered 623 species. Eleven of the 623 species were considered to be at medium risk of overexploitation after the vulnerability assessment. These 11 species were then
14 Reducing uncertainty in stock status ABARES assessed under the susceptibility criteria and all were found to be at low risk or very low risk of overexploitation.
Results from these analyses indicate that it is unlikely the sector is currently, or has previously, exerted a level of fishing pressure considered to be ‘overfishing’. As a result, the stock is unlikely to be overfished. Therefore, the CSF aquarium stock could be classified as not subject to overfishing and not overfished at current fishing levels. Introduction
The Aquarium Sector of the CSF is a multispecies, multigear fishery extending from Cape York to Sandy Cape in Queensland (Map 2.1).
Map 2.1 Management area of the Coral Sea Fishery Aquarium Sector
Data source: 2009 Fishery status reports
The Aquarium Sector is primarily managed through input controls. These include limited entry, gear restrictions including maximum net size/diameter, mesh sizes, handle/shaft lengths and net lengths, and a prohibition on chemical or explosive use. Spatial closures also apply in the form of Coringa-Herald and Lihou Reef national nature reserves (Map 2.1). The current harvest strategy (AFMA 2008) also includes a 200 days trigger upon which a further evaluation of the sustainability is required. Currently there are two operators within the fishery and each are permitted to use two tender (collection) vessels, operating from a mother vessel.
The ABARES 2009 Fishery status reports (the most recent report at the beginning of this study) classified the stock as uncertain with respect to both overfished and overfishing status (Wilson et al. 2010). The sector had not been previously assessed because of the complexities around the large number of species collected, the relatively low value of the sector and the limited data available to fishery managers and scientists.
15 Reducing uncertainty in stock status ABARES
Approach to status determination The aim of this study is to examine the potential impact of the sector on target stocks. Given the diversity of species taken and the nature of the fishery dependent and independent data collected, species-specific stock assessment methods were not considered appropriate and alternative quantitative and qualitative methods were explored. The three approaches selected were: a maximum footprint analysis that aimed to estimate the total amount of area operators could cover within a fishing season; estimation of annual extraction rates for key commercial families; and a species-specific risk analysis, based on vulnerability and susceptibility. The assessment is structured into sections on data and information, methods, results and discussion. Data and information Habitat area This report uses data from the Millennium Coral Reef Mapping Project. The project was undertaken by the Institute for Marine Remote Sensing, University of South Florida (IMaRS/USF) and Institut de Recherche pour le Développement (IRD/UR 128, Centre de Nouméa). IMaRS/USF performed a supervised classification of Landsat Thematic Mapper satellite imagery to create geomorphologic maps of reefs within the CSF.
Habitat area was estimated by importing the IMaRS/USF geomorphological classification into a geographic information system. Of the 15 geomorphological habitat classes in the dataset, five were thought to best reflect suitable habitat for the aquarium species taken in the CSF. These habitat classes represent the depth ranges subject to collection effort by the sector. Investigation of logbook data—specifically, hours per day, dives per day and the number of divers per day— suggests (assuming decompression tables are adhered to) dive effort is concentrated within the first 12 metres of the water column. ABARES are aware of deeper, experimental collection activity within the sector for particular species but this was not considered ‘typical’ activity.
The five classes used in the analyses were reef flat, pass, fore reef, inner slope and sub-tidal reef flat (Table 2.1). Reef classes excluded were either deep or sandy bottom habitats because these were not considered to be a key component of the fishery and survey density estimates for most species in those classes are lower.
Some reefs of interest to the management of CSF stocks are not covered by the USF dataset. These include Shark, Cato, Malay, McDermott and Mellish reefs. The calculated suitable habitat area is considered a conservative estimate because some reefs and reef classes were excluded from this analysis so the actual habitat area (and standing stock estimate) is probably larger. This conservative bias should be noted when using these analyses to inform management decisions.
Table 2.1 Area (km²) of suitable habitat classes for aquarium fish in the Coral Sea Fishery
Reefs Fore reef Inner slope Pass Reef flat Sub-tidal Total area reef flat Open reefs Dart 1.90 – – 3.96 – 5.86 Flinders Reefs 25.41 51.16 70.34 45.43 21.35 213.69 Flora Reef 3.94 7.16 – 7.99 1.59 20.67 Herald Surprise 2.72 – – 4.17 0.00 6.90 Holmes Reefs 16.20 48.94 – 54.08 6.98 126.20 Moore Reefs – – – 5.31 2.24 7.55
16 Reducing uncertainty in stock status ABARES
Willis Islets 31.01 – – 7.03 62.89 100.93 Abington Reef 1.22 – – 2.57 – 3.79 Bougainville 2.33 – – 5.58 – 7.91 Diamond Islets 3.04 – – 5.80 41.58 50.42 Dieanne Bank 8.01 – 16.10 0.43 144.95 169.49 Frederick Reef 10.87 – 1.26 6.28 2.93 21.35 Kenn Reef 28.79 46.23 11.47 16.05 26.88 129.42 Osprey Reef 11.10 22.45 0.91 40.77 0.47 75.69 Tregrosse Reefs 1.58 – – 1.02 25.62 28.23 Wreck Reefs 8.56 – 3.83 15.28 0.05 27.71 Magdelaine Cays 2.24 – – 5.33 – 7.57 Closed reefs Lihou Reefs 3.86 – – 14.14 – 18.00 Coringa Islets 22.01 – – 2.19 108.19 132.39 Herald Cays 90.14 106.73 70.64 65.48 43.73 376.72 Total area 274.95 282.68 174.54 308.88 489.44 1 530.48 Area within national nature 116.01 106.73 70.64 81.81 151.92 527.11 reserves (%) (42.19) (37.76) (40.47) (26.49) (31.04) (34.44) Source: ABARES
Density estimates and extraction rates Estimates of density for the key commercial families collected within this sector were obtained from published research using underwater visual census (UVC) and chemical-based survey techniques (Matoto et al. 1996; Ackerman & Bellwood 2000; Depczynski & Bellwood 2005; Ceccarelli et al. 2008). The estimates represent the known minimum and maximum density for each key commercial family. Underwater visual census has been widely used as a sampling technique to monitor the abundance of coral reef fish. Estimates of natural densities from UVC are likely to be biased downwards (underestimates) because of fish behaviour (Samoilys & Carlos 2000). Chemical-based survey techniques are thought to provide more realistic estimates for small cryptic species (Blenniidae, Gobiidae and Pomacentridae) but this technique is likely to be less reliable for larger species such as serranids and labrids (Depczynski & Bellwood 2005), which may move away from the area surveyed. Fishery dependent data Catch data has historically been recorded at genus or family level, although operators maintain private databases that include data on the catch (by number) of individuals at the species level. Data may also include fish size, effort expended and detailed spatial information.
Logbook data for the period August 2002 to October 2009 were used in these analyses but these data are suspected to be incomplete. Data from the 2008–09 fishing season were used to estimate annual extraction rates because this season had the highest annual seasonal catch from the available data.
Logbook data were tabulated and sorted into family groups to identify the key commercial families and the proportions these make of the total catch (Figure 2.1). Analyses were conducted at the family level because of the resolution of logbook data. Six key commercial families constitute around 80 per cent of the total catch. Further detail on the biological characteristics of these key commercial families is at Appendix A.
17 Reducing uncertainty in stock status ABARES
Figure 2.1 Proportional catch for key commercial families between 200203 and 200809
100% Other 90% 80% Pomacanthidae
70% Acanthuridae 60% Gobiidae & Bleniidae 50% Pomacentridae 40% 30% Serranidae
Collection (numbers)Collection 20% Labridae 10% 0% 2002–03 2003–04 2004–05 2005–06 2006–07 2007–08 2008–09
Season (financial year)
Source: Data from AFMA
Risk assessment The Australian Fisheries Management Authority (AFMA) provided ABARES with a list of aquarium species taken in the CSF. The list was informed by the two existing Coral Sea aquarium permit holders (AFMA, pers. comm., 2010). The Queensland Department of Agriculture, Fisheries and Forestry provided ABARES with a database that included risk scoring information for aquarium species in the Queensland Marine Aquarium Finfish Fishery (MAFF). Methods Footprint analysis Estimates of the area of suitable habitat in the CSF (Table 2.1) were combined with operations information to estimate the potential footprint that could be covered by the sector, under the constraints of the current management arrangements. The aim of this work was to understand the potential area that could be covered by the fishery, relative to the extent of suitable habitat for the species taken.
This method will give an upper bound estimate of the potential extent of the fishery. The actual extent of the fishery in a given year is likely to be less than that estimated because operators rarely use all available fishing days, are unlikely to cover the maximum potential area of a single operation and may revisit sites.
Calculation of the potential footprint of the fishery is undertaken by multiplying the maximum area that could be covered in a single fishing operation, the maximum number of sites that could be fished per day and the maximum number of fishing days per season (200 days as prescribed by the current harvest strategy).
The maximum area of a single collection site was defined as a circle where the radius is equal to the length of the divers’ hookah hose (assuming the tender vessel is anchored). Hookah hose length is reported as 150 metres (Donnelly 2009) resulting in a maximum single site area of 70 686 square metres (0.0707 square kilometres). The ‘maximum area of a single operation’ is
18 Reducing uncertainty in stock status ABARES the maximum amount of area that a diver could physically cover as restricted by the length of hookah hose. A diver would not realistically search this amount of area because of the difficulty in collection technique and constraints on the amount of time a diver can safely spend underwater without risking decompression sickness.
The total number of sites per day is calculated based on permit conditions and logbook data. Permit holders can use two tenders. Logbook data show that divers generally conduct two dive operations per day. With two permit holders in the fishery, the maximum number of sites per day is likely to be eight (AFMA 2009). Annual extraction rates Plausible standing stock numbers are estimated for each of the key commercial families by multiplying the total area of suitable habitat in the fishery by a range of published density estimates (Matoto et al. 1996; Ackerman & Bellwood 2000; Depczynski & Bellwood 2005; Ceccarelli et al. 2008). These standing stock estimates are then compared with 2008–09 catches (the highest recorded catch from the available data) to estimate extraction rates. As logbooks only attribute 94 days to the catch in 2008–09, the catch for this season was scaled up to represent 200 days of fishing effort to reflect a ‘worst-case’ scenario under the current management constraints. Scenarios of 300 and 400 days of fishing effort are also provided for reference. Risk assessment The risk of overexploitation to a species taken in this fishery is defined as a combination of its vulnerability to exploitation and its susceptibility to overexploitation. This technique has been used to assess the risk to species collected within the adjacent, and somewhat similar, Queensland-managed MAFF. Vulnerability and susceptibility scores generated by Roelofs and Silcock (2008) for the MAFF were applied to the same species captured in the CSF. For those species where vulnerability and susceptibility scores were not available, the methodology employed by Roelofs and Silcock (2008) was used to calculate a risk score. Vulnerability assessment The vulnerability assessment uses a defined set of ecological and economic criteria to score a species’ attractiveness and overall vulnerability to fishers. Ponder and Grayson (1998) used a similar approach to assess the risk posed to Australian marine gastropods from collection activities. The criteria include species distribution, accessibility, market value and ecological niche. The cumulative score across criteria (VAR) provides a measure of vulnerability (Table 2.2).
Table 2.2 The scoring system to assess the vulnerability of marine aquarium species in the Coral Sea Fishery Aquarium Sector
Vulnerability Cumulative Description risk score (VAR) Very low 0–8 Average ranking across each criteria is 2 or below. These species are not vulnerable to collection activity in the CSF (Aquarium Sector). Low 8.5–12 Average ranking is between 2 and 3. These species are at low risk from CSF collection activity. Medium 12.5–16 Average ranking is between 3 and 4. These species have characteristics that make them moderately vulnerable to overfishing by the fishery. High >16 Average ranking is >4. These species have characteristics that make them highly vulnerable to over-collection by the fishery.
19 Reducing uncertainty in stock status ABARES
Susceptibility assessment The method employed by Roelofs and Silcock (2008) requires that any species scored as medium risk or higher in the vulnerability assessment be carried through into the susceptibility assessment. This assessment considers a species’ ability to recover from exploitation, based on its life history traits, including age at maturity, fecundity, longevity and biological development. Table 2.3 details the scoring system used for assessing susceptibility of a species. For further detail on the methodologies applied, see Roelofs and Silcock (2008).
Table 2.3 Scoring system to assess the susceptibility of the marine aquarium species captured in the Coral Sea Fishery Aquarium Sector applied to species with medium or high vulnerability
Susceptibility Cumulative Description risk score(SAR) Very low 0–8 Average ranking across each criteria is 2 or below. These species are not susceptible to collection activity in the CSF (Aquarium Sector). Low 8.5–12 Average ranking is between 2 and 3. These species are at low risk from CSF collection activity. Medium 12.5–16 Average ranking is between 3 and 4. These species have characteristics that make them moderately susceptible to over-collection by the fishery. High >16 Average ranking is >4. These species have characteristics that make them highly vulnerable to over-collection by the fishery. Results Footprint analysis The maximum area that can be covered by collection effort, in a fishing season, under the current management arrangements was estimated to be 113.12 square kilometres. This figure represents approximately 7.4 per cent of the suitable reef area within the CSF (Table 2.4). At 400 days of fishing effort, an estimated maximum of 14.8 per cent of suitable reef area could be covered.
Table 2.4 Estimates of maximum footprint for the sector
Available fishing days 200 300 400 Area of single site (km²) 0.0707 0.0707 0.0707 Sites per day 8 8 8 Maximum annual fishery coverage (km²) 113.12 169.68 226.24 Total suitable reef area (km²) 1 530.48 1 530.48 1 530.48 Percentage of reef area covered 7.39 11.09 14.78 Note: Fishery management arrangements currently allow a maximum of 200 days in the fishery. Scenarios of 300 and 400 days of fishing effort are provided for reference.
Annual extraction rates Standing stock and extraction rate estimates for key commercial families in the CSF are presented in Table 2.5. The table provides information for deriving standing stocks estimates— both a maximim and minimum—based on the cited literature and habitat areas. Table 2.5 also provides information on the extraction rate (per cent of individuals removed) based on the standing stock estimates and the catch from the fishery catch in the 2008–09 fishing season. Potential extraction rates from scaling up of 2008–09 catch (94 days) to 200 days (maximum provided for under the harvest strategy) are also provided in Table 2.5.
20 Reducing uncertainty in stock status ABARES
Under all scenarios of standing stock (minimum and maximum) and catch (2008–09 or the rescaled 200 days catches), the extraction rates for all families were estimated to be very low. All estimates were less that 0.04 per cent annually and mostly far less than this. These rates could be considered negligible but they apply at the fish family level and rates may be higher for particular species within a family.
21 Reducing uncertainty in stock status ABARES
Table 2.5 Summary of observed minimum and maximum densities, estimated suitable habitat area, standing stock and annual extraction rates based on 2008–09 catch data
Family Observed density Estimated Standing stock estimate (absolute n) 2008–09 aquarium 2008-09 extraction rate- Maximum extraction (m²) suitable habitat catch (n) 94 days (%) rate-200 days (%) e (km²) minimum maximum minimum maximum minimum maximum Serranidae 0.04a 0.30b 1 530.48 56 627 823 459 144 510 11 538 0.003 0.02 0.04 Labridae 0.06c 0.30b 1 530.48 91 828 902 459 144 510 8 520 0.002 0.009 0.02 Pomacentridae 0.88c 10.00b 1 530.48 1 346 823 896 15 304 817 000 7 665 0.00005 0.0005 0.001 Acanthuridae 0.11c 0.32a 1 530.48 168 352 987 487 917 566 3 915 0.00002 0.0008 0.002 Blenniidae & Gobiidae 0.04a 12.00d 1 530.48 61 219 268 18 365 780 400 2 178 0.00001 0.004 0.008 Pomacanthidae 0.04a na 1 530.48 61 525 364 na 1 988 na 0.003 0.007 a Matoto et al. 1996; b Ackerman & Bellwood 2000; c Ceccarelli et al. 2008; d Depczynski & Bellwood 2005. e This column contains a scaled-up extraction rate based on 200 days of fishing effort.
22
22
Reducing uncertainty in stock status ABARES
Risk assessment Vulnerability assessment In total, 623 species were assessed against the vulnerability criteria. Of these 449 were covered by the work undertaken by Roelofs and Silcock (2008). The remaining 174 species were assessed by ABARES. Of the 623 species, 11 species were assessed at medium vulnerability risk (a VAR score between 12.5 and 16). All other species were assessed as low or very low vulnerability risk.
Table 2.6 Vulnerability risk scores for species within the Aquarium Sector of the Coral Sea Fishery
Family Species Common name Distribution Ecological Market Accessibility Total (VAR) Risk niche value Labridae Cirrhilabrus bathyphilus Deepwater wrasse 5 2 3 3.5 13.5 Medium Pomacentridae Amphiprion melanopus Blackback anemonefish 2 5 2 4.5 13.5 Medium Blenniidae Escenius aequalis Great Barrier Reef blenny 4 2 2 5 13 Medium Palaemonidae Periclimenes brevicarpalis Pacific clown anemone shrimp 1 5 2 5 13 Medium Pomacentridae Amphiprion perideraion Skunk anemonefish 2 5 2 4 13 Medium Pomacentridae Amphiprion akindynos Barrier Reef anemonefish 2 5 2 4 13 Medium
23 Pomacentridae Amphiprion chrysopterus Orange-fin anemonefish 2 5 2 4 13 Medium
Chaetodontidae Chaetodon rainfordi Rainford’s butterflyfish 2 5 2 4 13 Medium Pomacentridae Premnas biaculeatus Maroon anemonefish 2 5 2 4 13 Medium Blenniidae Ecsenius tigris Tiger blenny 5 2 2 3.5 12.5 Medium Siganidae Siganus corallinus Coral rabbitfish 2 4 2 4.5 12.5 Medium Note: All vulnerability criteria were given the same weighting. Scores were scaled in increasing order of risk (5 = higher level of risk).
23
Reducing uncertainty in stock status ABARES
Susceptibility assessment The 11 species ranked as medium vulnerability risk were assessed for their susceptibility to overfishing, based on life history traits. All species were assessed as either ‘low’ or ‘very low’ risk in terms of susceptibility (Table 2.7).
Table 2.7 Susceptibility risk scores for species within the Aquarium Sector of the Coral Sea Fishery
Family Species Common name Total Maturity Fecundity Longevity Development Total Risk (VAR) (SAR) Labridae Cirrhilabrus bathyphilus Deepwater wrasse 13.5 1 3 2 3 9 Low Pomacentridae Amphiprion melanopus Blackback anemonefish 13.5 2 4 3 3 12 Low Blenniidae Escenius aequalis Great barrier reef blenny 13 3 3 3 2 11 Low Palaemonidae Periclimene brevicarpalis Pacific clown anemone shrimp 13 1 1 1 5 8 Very low Pomacentridae Amphiprion perideraion Skunk anemonefish 13 2 2 3 3 10 Low Pomacentridae Amphiprion akindynos Barrier Reef anemonefish 13 2 4 3 3 12 Low Pomacentridae Amphiprion chrysopterus Orange-fin anemonefish 13 2 4 3 3 12 Low Chaetodontidae Chaetodon rainfordi Rainford’s butterflyfish 13 2 2 3 1 8 Very low 24 Pomacentridae Premnas biaculeatus Maroon anemonefish 13 2 4 3 3 12 Low
Blenniidae Ecsenius tigris Tiger blenny 12.5 2 2 3 1 8 Very low Siganidae Siganus corallinus Coral rabbitfish 12.5 3 1 3 1 8 Very low Note: All susceptibility criteria were given the same weighting. Scores were scaled in increasing order of risk (5 = higher level of risk).
24
Reducing uncertainty in stock status ABARES
Discussion
Together, these analyses indicate that the CSF aquarium stock is not being exploited at a level considered to be excessive. Further, no species are considered as being at greater than low risk from the fishery in its current form.
The total maximum footprint achievable under current management arrangements is approximately 7.4 per cent (Table 2.4). As a result, a large proportion (more than 92 per cent) of the suitable habitat is not exploited in any given season. Although the area of potential coverage is low, the potential for localised depletion at the reef or sub-reef level should be considered and analysed further.
The highest extraction level observed for any family group (up to and including 2008–09) was 0.02 per cent for 94 days of fishing in the 2008–09 fishing season. If the 2008–09 fishing effort is scaled up to represent the maximum effort available under the current harvest strategy (200 days), the maximum potential annual extraction rate in the sector would be approximately 0.04 per cent. Because this analysis is conducted at the family level, the potential of depletion of individual species should still be considered.
The risk assessment considered 623 species using the methodology employed by Roelofs and Silcock (2008). Eleven species were considered to be medium risk post the vulnerability assessment. However all 11 scored ‘low’ or ‘very low’ risk under the susceptibility analysis (Table 2.6 and Table 2.7).
These analyses do not consider data before 1999 but anecdotal evidence suggests exploitation before this time was less than the period for which data was available.
Approximately 35 per cent of the estimated suitable habitat area for the aquarium stock is protected within existing national nature reserves (Table 2.1). This means that a substantial area of the fishery is now, and is likely to continue to be, unexploited.
The evidence presented provides a strong weight-of-evidence approach for this stock being classified as not subject to overfishing and not overfished. This would hold true as long as the fishery continues to operate as it has for the data period. If changes are made to the species harvested or operational conditions, the assumptions within these analyses may need to be considered further.
One of the difficulties with conducting this work was the logbook design. The logbook used in the CSF is the same as that used for the Queensland-managed MAFF and, as such, is designed for those species collected in the MAFF. The logbook should require reporting of the catch of each of the seven key commercial families described in Figure 2.1. Important individual species should also be recorded separately. These should include at least species showing greater than negligible risk after a susceptibility assessment and any listed species. Space should also be designated for newly discovered species. Fine-scale effort information, such as that obtained from diver data-loggers, would provide fishery managers with additional data that may facilitate more in-depth analysis.
For the risk assessment the scoring criterion for the vulnerability and susceptibility assessment/s are deterministic in nature. Because of the variable and uncertain nature of the biological and economic parameters relied on within the methodologies, this work would need to be updated to maintain relevance should there be substantial changes to the fishery, fishing methods or biological information.
25 Reducing uncertainty in stock status ABARES
Appendix A: Key commercial families collected Blenniidae and Gobiidae In Australian waters there are around 100 genera and more than 450 species of blenniidae and gobiidae. Both families are known to be guarders of demersal eggs (Kuiter 1996). In the CSF Aquarium Sector more than 52 species of blenniidae and gobiidae families are known to be collected (AFMA, pers. comm., 2010). Labridae The family labridae (wrasses) contains more than 60 genera and more than 400 species. They are carnivores, feeding on a wide range of small invertebrates. Labridae typically have a pelagic egg distribution and are harem spawners (Kuiter 1996). In the CSF Aquarium Sector more than 91 species of wrasse are known to be collected (AFMA, pers. comm., 2010). Serranidae Serranidae are a large family of fishes containing well over 400 species in more than 50 genera and all are carnivorous. Many species are protogynous hermaphrodites, meaning that they start out as female and change sex to male later in life. They are known to produce large quantities of pelagic eggs (Kuiter 1996). In the CSF Aquarium Sector more than 44 species of serranid are known to be collected (AFMA, pers. comm., 2010). Pomacanthidae The family pomacanthidae consist of five genera; approximately 30 species are found in Australian waters. They are a highly sought after aquarium species. Pomacanthidae are known to be protogynous hermaphrodites and pelagic spawners; most are secretive on reefs (Kuiter 1996). In the CSF Aquarium Sector more than 22 pomacanthid species are known to be collected (AFMA, pers. comm., 2010). Acanthuridae Acanthuridae consist of six genera and approximately 70 species; approximately 35 species are found within Australian waters. The planktivores of the family are usually found in large schools and the benthic feeders are found in either pairs or schools (Kuiter 1996). In the CSF Aquarium Sector more than 25 species of acanthurid are known to be collected (AFMA, pers. comm., 2010). Pomacentridae Pomacentridae are a large fish family, primarily tropical with around 15 genera and 132 species found within Australian waters. These fishes are guarders of demersal eggs (Kuiter 1996). In the CSF Aquarium Sector more than 48 species of pomacentridae are known to be collected (AFMA, pers. comm., 2010). References
Ackerman, J & Bellwood, D 2000, ‘Reef fish assemblages: a re-evaluation using enclosed rotenone stations’, Marine Ecology Progress Series, vol. 206, pp. 227–37. AFMA 2009, Management arrangements Coral Sea, Australian Fisheries Management Authority, Canberra. —— 2008, Harvest Strategy – Coral Sea Fishery, Hand Collection Sector: Aquarium, Australian Fisheries Management Authority, Canberra.
26 Reducing uncertainty in stock status ABARES
Ceccarelli, D, Choat, JH, Ayling, AM, Richards, Z, van Herwerden, L, Ayling, A, Ewels, G, Hobbs, JP & Cuff, B 2008, Coringa-Herald National Nature Reserve Marine Survey 2007, report to the Department of the Environment, Water, Heritage and the Arts by C&R Consulting and James Cook University. Depczynski, M & Bellwood, D 2005, ‘Wave energy and spatial variability in community structure of small cryptic coral reef fishes’, Marine Ecology Progress Series, vol. 303, pp. 283–293. Donnelly, R 2009, Pro-vision Reef: Stewardship Action Plan. A statement of operational standards and climate change contingency planning, Pro-vision Reef Inc. Cairns, Australia.
Kuiter, R 1996, Guide to sea fishes of Australia, New Holland, Sydney. Matoto, S, Ledua, E, Mou-Tham, G, Kulbicki, M & Dalzell, P 1996, The aquarium-fish fishery in Tongatupu, Tonga. Status and recommendations for management, South Pacific Commission, BP D5, Noumea, New Caledonia. Ponder, W & Grayson, J 1998, The Australian marine molluscs considered to be potentially vulnerable to the shell trade, report prepared for Environment Australia, Canberra. Roelofs, A & Silcock, R 2008, A sustainability assessment of marine fish species collected in the Queensland marine aquarium trade, Department of Primary Industries and Fisheries, Brisbane. Samoilys, M & Carlos, G 2000, Determining methods of underwater visual census for estimating the abundance of coral reef fishes, Northern Fisheries Research Centre, Department of Primary Industries, Cairns, Queensland. Wilson, DT, Curtotti R, & Begg GA (eds) 2010, Fishery status reports 2009: status of fish stocks and fisheries managed by the Australian Government, Australian Bureau of Agricultural and Resource Economics – Bureau of Rural Sciences, Canberra.
27 Reducing uncertainty in stock status ABARES 3 Assessing Coral Sea Fishery sea cucumber stocks using spatial methods
James Woodhams, Mark Chambers and Lindsay Penrose Summary
The biomass status of sea cucumber stocks in the Coral Sea Fishery (CSF) have mostly been classified as uncertain in the ABARES Fishery status reports since 2006. Sea cucumber stocks in the Coral Sea Fishery are subject to total allowable catches (TACs), catch triggers and move-on provisions (AFMA 2008). However, the scientific basis for these TACs and catch triggers is unclear, so compliance with the TACs was not sufficient to establish stock status.
A habitat-based approach was the primary assessment method employed in this study. The approach focused on four species of sea cucumber and relied on coral reef habitat mapping, along with historical surveys of sea cucumbers undertaken in the Coringa-Herald and Lihou Reef national nature reserves. Together, these data were used to derive plausible potential biomass scenarios of sea cucumber species.
A geographic information system was developed comprising geo-located survey data and geomorphological habitat classes to map the reefs within the CSF. Together, these data were used to derive estimates of population size and biomass for commercially important sea cucumber species in the CSF. Estimates of biomass then formed the basis surplus production models and estimation of maximum sustainable yields (MSY).
The reliability of the outputs of these analyses was highly influenced by the availability of data. Data availability was poor across all species, but particularly poor for white teatfish (Holothuria fuscogilva) and surf redfish (Actinopyga mauritiana). Although estimates of density, population size and sustainable yields have been derived, the limitations imposed by the availability of data need to be taken into account. The analyses predict higher biomasses and sustainable yields at reefs that have historically supported larger catches.
Other potential assessment methods were investigated including catch per unit effort, size (weight) trends over time and stock depletion analysis. However, these methods were found to be inappropriate because of data quality/quantity issues or violation of assumptions that underlie the method.
The analyses presented here were undertaken at the reef level but status determination is undertaken at the fishery level to reflect the unit of management. The reconciliation of status was based on the surplus production models, which provide an estimate of biomass in 2010 as a proportion of biomass at the start of the assessment period (1997). Using this approach, black teatfish (Holothuria whitmaei) and prickly redfish (Thelenota ananus) would be classified as not overfished and not subject to overfishing.
Recent catches for surf redfish have been less than the median fishery level estimate of MSY. As such, the stock is classified as not subject to overfishing. Given that surf redfish catch was less than the median estimate of MSY for 11 of the 14 seasons since 1997–98 (excluding Cato Reef), and surplus production model results indicate biomass levels of 70 per cent (Schaefer model)
28 Reducing uncertainty in stock status ABARES and 91 per cent (Pella–Tomlinson model) at the end of the modelling period, it is unlikely that the stock is overfished. As such, the stock is classified as not overfished.
Plausible initial biomass estimates could not be established for white teatfish so it was not possible to place the historical catches into context. Results from the surplus production model are not reliable for this species. As such, the CSF white teatfish stock remains uncertain with respect to biomass and fishing mortality. Introduction
The CSF is a multispecies, multigear fishery extending from Sandy Cape in southern Queensland to Cape York in Northern Queensland (Map 3.1). The Australian Fisheries Management Authority (AFMA) produces a statement of management arrangements (similar to a management plan) for the fishery, specifying the controls on fishing for each season (currently aligned with the financial year). AFMA has also produced harvest strategies for the fishery, which detail the catch and effort controls, triggers and decision rules.
The Sea Cucumber Sector of the CSF is a targeted, hand collection fishery, taking a suite of sea cucumber species. Primary target species include black teatfish (Holothuria whitmaei), white teatfish (H. fuscogilva), surf redfish (Actinopyga mauritiana) and prickly redfish (Thelenota ananus). Another dozen species have either been taken or could be taken in the fishery should a market arise.
There are two licenses in the sector. These permit the taking of sea cucumbers by hand, through either free-dive or hookah (surface supplied air) fishing methods. Fishing is typically undertaken from smaller tenders associated with a mothership.
Map 3.1 Extent of the Coral Sea Fishery
Source: ABARES
29 Reducing uncertainty in stock status ABARES
The ABARES 2009 Fishery status reports (the most recent report at the start of this study) and earlier reports classified all sea cucumber stocks as uncertain if overfished and uncertain if subject to overfishing (Woodhams et al. 2010). Approach to status determination The purpose of this study was to derive plausible potential biomass scenarios for four sea cucumber species in the CSF, from which yield may be estimated and compared with historical catches. A habitat-based approach was adopted but other approaches such as catch per unit effort, size trends over time and stock depletion analysis were initially investigated. The chapter is organised into five sections: background, methods, results and discussion. Background
Total allowable catches (TACs) came into effect in the 2002–03 season. Currently, TACs are set for black teatfish (1 tonne), white teatfish (4 tonnes), sandfish (Holothuria spp.; 1 tonne), surf redfish (10 tonnes) and prickly redfish (20 tonnes). Greenfish (Stichopus chloronotus) and lollyfish (H. atra) are covered by a 10 tonne combined TAC. All other species are subject to a 10 tonne per species TAC (AFMA 2008). There is also an overarching total annual TAC of 150 tonnes for all species in the fishery. Permit conditions also contain move-on provisions, whereby after a total of 5 tonnes of sea cucumber (mixed species) has been taken from any one reef annually, no further collection may occur within 15 nautical miles of that location. The weight of catch in the sector is recorded as a processed wet weight, which means the animal has been gutted and partially processed prior to the catch being attributed to the TAC.
A reef rotation policy has been in place since July 2005. This policy stipulates a three-year rotational harvesting strategy. The plan identifies 21 reefs and stipulates the number of days fishing on each reef with each reef only open one year in three. This rotational harvest policy is included in permit conditions.
Harvest strategies were introduced in the CSF on 1 July 2008. The sea cucumber harvest strategy employs TACs, spatial closures, move-on provisions and size limits. The process used to set the current TACs is not well documented and the basis of TACs for individual species is unclear. Analysis by Hunter et al. (2002) may have informed the current TACs for black teatfish and white teatfish, but no documentation supporting the decision-making process has been identified. As such, compliance with the harvest strategy could not be relied on in the status determination process as evidence of sustainable extraction rates. Previous assessments Hunter et al. (2002) analysed sea cucumber catch per unit effort (CPUE) in the CSF using data from 2000 and 2001. Fishing records were spatially attributed to reefs and analysis undertaken comprised:
daily and yearly total catch for each species, at each reef
average dive time for both free-dive and hookah dive methods for vessels targeting sea cucumber
CPUE (in kilograms per hour) for each species at each reef, daily and monthly.
Catch rates were standardised using linear models with trip and year as explanatory variables. Separate standardisations were undertaken for free-dive and hookah across five reefs and six sea cucumber species.
30 Reducing uncertainty in stock status ABARES
In undertaking this work, Hunter et al. (2002) identified difficulties in separating catch according to the type of method applied (hookah or free-dive). There were also issues with the applicability of CPUE as an indicator of abundance for a highly targeted, spatially discrete, hand collection fishery. These issues remain relevant.
Hunter et al. (2002) propose that free diving is more targeted toward taking black teatfish, surf redfish and lollyfish and hookah is more targeted toward taking white teatfish and amberfish (Thelenota anax). Prickly redfish appeared to be taken using both methods.
Hunter et al. (2002) reported a decline in the catch of high value species (white teatfish, black teatfish, prickly redfish, sandfish, surf redfish) and declines in the CPUE of a some high value species at a number of reefs. The authors recommended:
ceasing fishing effort on black teatfish
considering further restrictions on the TAC and catch of white teatfish and prickly redfish in particular
modifying logbook design to incorporate recommended changes (to allow the separation of catches according to fishing method).
It is understood that the black teatfish TAC was subsequently reduced to 1 tonne and that logbook design was amended in 2010.
Management strategy evaluation (MSE) simulations were recently undertaken as part of the RUSS project (stream 2) to identify if the present harvest strategy achieves the intention of the Commonwealth Fisheries Harvest Strategy Policy (Haddon 2012). MSE is identified within the Harvest Strategy Policy as an appropriate method for assessing management options for stock target and limit reference points. Catch and effort These analyses contain data back to February 1998. Although there was fishing activity in the area before 1998, ABARES has been unable to access these data. The extent of hand collection fishing operations before 1998 is unknown.
Dive operations in the fishery may take sea cucumber, lobster and trochus in the one trip. As such, some summary statistics are provided for lobster and trochus but the status of these stocks are not considered in this report.
A total of 528 dive operations were recorded between 1997–98 and 2008–09 (Figure 3.1). Fishers are required to complete daily logbook entries, so each record reflects one fishing day, by one primary vessel. There may be a number of tenders and multiple divers associated with each vessel attributed to each daily record.
The overall trend is for declining catch and effort in recent years, from the peak in the 2000–01 season, although the pattern of exploitation varies for individual species (Figure 3.1). The reasons for the decline are not clear but may relate to management changes (such as the introduction of TACs) and the considerable costs associated with operating in this remote fishery. It is also possible that the catches being taken in the peak of the fishery could not be sustained by the available biomass.
31 Reducing uncertainty in stock status ABARES
Figure 3.1 The number of daily operations in the CSF Hand Collection Sector between 1997–98 and 2008–09
180 160
140 120 100 80 60
Number ofrecords Number 40 20 0 1997– 1998– 1999– 2000– 2001– 2002– 2003– 2004– 2005– 2006– 2007– 2008– 98 99 00 01 02 03 04 05 06 07 08 09 Season
Source: Data from AFMA
Approximately 33 per cent of all operations and 29 per cent of total catch has occurred on a single reef (
Table 3.1). Approximately 67 per cent of the lobster catch (by weight) was taken from two reefs. Another reef accounted for approximately 10 per cent of the number of hand collection records and around 21 per cent of the total dive catch (by weight).
Table 3.1 Catch statistics by reef for hand collection operations in the CSF
Reef Records (no.) Proportion of Total catch Proportion of Proportion of Proportion of total records (kg) total dive catch key commercial lobster / sea cucumber trochus within within total total catch catch Reef 1 175 33.1% 43 748.4 29.0% 70.6% 0.5% Reef 2 65 12.3% 20 084.7 13.3% 50.2% 12.0% Reef 3 53 10.0% 32 106.4 21.3% 83.2% 0.1% Reef 4 50 9.5% 6 371.4 4.2% 70.3% 27.4% Reef 5 43 8.1% 14 611.8 9.7% 93.4% 3.0% Reef 6 38 7.2% 9 045.8 6.0% 81.2% 0.2% Reef 7 29 5.5% 5 577.0 3.7% 76.4% 9.7% Reef 8 22 4.2% 5 220.5 3.5% 95.1% 0.0% Reef 9 11 2.1% 948.4 0.6% 56.4% 40.5% Reef 10 9 1.7% 5 012.0 3.3% 98.5% 0.0% Reef 11 8 1.5% 2 137.0 1.4% 94.4% 0.0% Reef 12 5 0.9% 978.0 0.6% 95.9% 0.0% Reef 13 4 0.8% 1 031.0 0.7% 94.2% 0.0% Reef 14 4 0.8% 181.6 0.1% 99.7% 0.0% Reef 15 4 0.8% 1 256.5 0.8% 99.6% 0.0% Reef 16 3 0.6% 1 223.5 0.8% 99.6% 0.0%
32 Reducing uncertainty in stock status ABARES
Reef 17 2 0.4% 16.0 0.0% 100.0% 0.0% Reef 18 1 0.2% 88.0 0.1% 90.9% 0.0% Reef 19 1 0.2% 85.0 0.1% 75.3% 0.0% Reef 20 1 0.2% 910.0 0.6% 100.0% 0.0% Reef 21 0 – 0.0 – – – Reef 22 0 – 0.0 – – – Reef 23 0 – 0.0 – – – Totals 528 – 150 633 – – – Note: As a result of AFMA’s confidentiality rules, the catch of a single reef cannot be disclosed publically. The reefs that form part of this analysis are as follows (in alphabetical order): Abington Reefs, Bougainville Reef, Cato Reef, Coringa Islets, Dart Reef, Diamond Islets, Dieanne Bank, Flinders Reefs, Flora Reef, Frederick Reef, Herald Cays, Herald Surprise, Holmes Reefs, Kenn Reef, Lihou Reefs, Magdelaine Cays, Mellish Reef, Moore Reefs, Osprey Reef, Tregrosse Reefs, unknown reef, Willis Islets, Wreck Reefs. Source: Data from AFMA
Two high value species, white and black teatfish, made up the largest individual components of hand collection catch over the history of the fishery, totalling approximately 58 tonnes and 23 tonnes, respectively. Over the same period, surf redfish and prickly redfish, two medium– high value species, contributed 16 tonnes and 18 tonnes, respectively. These four species represent approximately 80 per cent of the total sea cucumber catch (by weight) since 1997–98. Two other medium–low value species, amberfish and lollyfish, make up most of the balance, contributing another 17 per cent (combined) of the total sea cucumber catch (by weight).
A maximum depth is recorded in logbooks for each operation (Figure 3.2). Although the proportion, or amount of catch, taken at this maximum depth is not known, these data may indicate the preferred depth for taking certain species.
More than 65 per cent of the white teatfish catch was taken with a maximum depth recorded of 30–40 metres and approximately 88 per cent of white teatfish catch had a maximum depth of 20–40 metres. Greater than 50 per cent of the black teatfish catch was taken with an associated maximum depth of 0–10 metres. Harvesting surf redfish may be preferred at depths less than 10 metres because 78 per cent of the harvest of this species recording a maximum depth of 0– 10 metres. As surf redfish and prickly redfish are considered secondary target species, it is unclear if these data indicate a true preference of harvest depth for these species.
Figure 3.2 Total catch of key species by maximum depth field recorded in logbooks
Exploitation by depth (maximum depth field) 45 40 White teatfish 35
30 Black teatfish 25
Catch (t)Catch 20 Surf redfish 15 10 Prickly redfish 5 0 0–9 10–19 20–29 30–39 40–49 Depth (m)
33 Reducing uncertainty in stock status ABARES
Source: Data from AFMA
Total hookah dive hours have exceeded free-dive hours in all but one season (2001–02; Figure 3.3). The split between the two methods was not recorded before the 1999–2000 season. Peak effort occurred in the same season as peak catch (2000–01) (Figure 3.3). Effort applied in the 2003–04 season was largely attributed to the take of lobster.
Figure 3.3 Dive hours for the two hand collection methods and total catch
1 400 60
1 200 50
1 000
40
800 30
600 Catch (t) Catch 20
Dive hours Dive 400
200 10
0 0 1997– 1998– 1999– 2000– 2001– 2002– 2003– 2004– 2005– 2006– 2007– 2008– 98 99 2000 01 02 03 04 05 06 07 08 09 Season
Free-dive effort (hours) Hookah effort (hours) Catch (t)
Source: Data from AFMA Methods Investigation of methods Initially, a variety of traditional stock assessment approaches were considered for assessing CSF sea cucumber stocks. Methods considered were standardised catch rates, depletion/removals models and analysis of changes in sea cucumber size (weight) through time. All these approaches suffered from data quality or quantity issues or violations of the underlying principles inherent in the method and, as such, were found to be unsuitable. Perry et al. (1999) suggest that depletion experiments and CPUE measurements, at an appropriately fine spatial scale, might provide useful information for stock assessment purposes, particularly where fished areas can be compared with unfished areas. In addition, depletion and removal methods have been attempted using fishery dependent data elsewhere (for example, DeLury 1947; Carle & Strubb 1978). These analyses were not fruitful, largely as a result of the inability to construct an index of abundance (CPUE series).
CPUE analysis was not considered to provide a reasonable index of abundance because:
our current understanding of sea cucumbers is that they are not evenly distributed across the reef structure. Instead, they are found in patches associated with suitable habitat. For some species, there may also be a seasonal effect where animals aggregate for reproduction. Because of this aggregation, and because sea cucumbers are relatively sedentary, catch data give very little information about the presence of cucumbers in areas
34 Reducing uncertainty in stock status ABARES
that were not fished. The resolution at which catch data is currently reported in the CSF (approximately 1 hectare) also makes it difficult to isolate and investigate fishing at the patch level.
the potential for hyper-stability in catch and effort data is difficult to account for. It is possible that fishers could move throughout the fishery, sustaining catch rates while patches of the stock are serially depleted. This concern could not be allayed through investigating position information in the logbooks because of the coarse resolution of these data.
separating the catch data according to fishing method is not possible. Logbooks up to and including the 2008–09 season reflect the methods used to harvest sea cucumbers in the fishery (free-dive and hookah). However, catch taken on trips where both methods are used is not separated according to method. Therefore, the relative harvesting power of these methods is difficult to investigate and/or account for in a standardisation process.
the scarcity of data, particularly in recent years, raises issues of low sample size and poor statistical power.
The lack of a reliable CPUE-based index of abundance precludes a variety of simple modelling methods such as depletion analysis and biomass dynamics models (requiring both catch and an index of abundance) (Cowx 1983; Pollock 1991).
Using average size or weight as an indicator of stock status was difficult for many of the same reasons as CPUE. Such an analysis is also likely to be affected by the targeted sampling nature of the fishery and minimum and maximum size limits. In response to being handled, sea cucumbers may eviscerate internal organs, contract in size, dispel water or even split in half. This makes it difficult to apply length and weight-based metrics to assess sea cucumber stocks. Further, the numbers of individuals caught or the unit weight (as recorded in the logbooks) appears to have been estimated or approximated on a number of occasions. Habitat-based approach: the method applied The spatial, habitat-based approach undertaken relies on estimates of habitat area and sea cucumber density. These can be used to derive estimates of population size. Habitat area was derived from geomorphological classification performed in the Millennium Coral Reef Mapping Project. Density estimates for these habitat classes were derived from survey data of the Lihou Reef and Coringa-Herald national nature reserves in the Coral Sea. Survey data from the Torres Strait were also considered in the case of white teatfish because data for this species were scarce in surveys of Lihou Reef and Coringa-Herald national nature reserves. Catch data were used to estimate average animal size to facilitate biomass estimates. Habitat mapping The Millennium Coral Reef Mapping Project was undertaken by the Institute for Marine Remote Sensing, University of South Florida (IMaRS/USF) and Institut de Recherche pour le Développement (IRD/UR 128, Centre de Nouméa). IMaRS is funded by the Oceanography programme of the National Aeronautics and Space Administration (NASA) to provide a worldwide inventory of coral reefs using high-resolution satellite imagery (Landsat 7 images acquired between 1999 and 2002). As at 17 May 2006, 1 724 images had been collected (1 661 with permission to distribute). By using consistent datasets of high-resolution (30 metres) multispectral data, the USF characterised, mapped and estimated the extent of shallow coral reef ecosystems in the main coral reef provinces around the world. These data were subjected to a
35 Reducing uncertainty in stock status ABARES supervised classification (by IMaRS/USF) to generate geomorphological classes. These data are available at the Institute for Marine Remote Sensing’s website.
ABARES imported the USF geomorphological dataset into a GIS and calculated areas for each of the geomorphological classes (polygons). The area of these polygons to generate the total area of each geomorphological class, at each reef.
Some reefs of interest to the management of CSF sea cucumber stocks were not covered by the IMaRS/USF dataset and, for this reason, could not be progressed through these analyses. These reefs include Cato and Mellish. Therefore, some of the relevant geomorphological classes in the CSF may be underestimated, in turn underestimating the corresponding estimate of population size (and biomass). Habitat selection Geomorphological class was used as a surrogate for habitat. The relationship between the geomorphological classes within the IMaRS/USF dataset and sea cucumber habitat preference was determined using survey data (see Survey samples section later in this chapter) from the CSF, the USF satellite imagery, consultation with experts and license holders and expert judgement.
The survey data and IMaRS/USF geomorphological data included a number of habitat descriptors. This information was matched with an understanding of the preferred habitat characteristics of the key commercial species. These matchings were then discussed with experts and licence holders.
Of the 15 geomorphological classes covered by the IMaRS/USF dataset, five were chosen as appropriate sea cucumber habitat for one or more species. These classes were reef flat (shallow reef flat), pass (deeper, high flow areas between sub-reef units), fore-reef (the front or ocean facing edge of the reef), inner slope (sloping reef on the inside of the reef crest toward the lagoon) and sub-tidal reef flat (reef flat of variable depth). Other habitat types may also be important to certain species but obtaining appropriate densities for other habitat types presents a challenge. Appendix B provides reef by reef area estimates for all habitat types.
Deep terrace (deeper reef structure associated with a bank or rise rather than an atoll) and deep lagoon (deeper reef inside atoll type structures) were considered for inclusion as an appropriate habitat type for white teatfish; however, their very large areal extent and low (or no) number of survey transects for these habitat types precluded their use. Following consultation with stakeholders, the pass habitat type was not included as habitat for black teatfish.
Habitat suitability (and therefore density) is likely to be governed by site-specific conditions, including current stress, turbulence, temperature, competition, food availability and predation. These are difficult to account for using the remotely sensed data. Large scale and episodic environmental forces are also likely, such as high (relatively) water temperature events or storm activity that can affect habitat on particular reefs. Cyclone damage to coral habitats was noted during the 2003 Coringa-Herald Survey (Oxley et al. 2003) and extensive coral bleaching events were noted at Lihou Reef National Nature Reserve (Oxley et al. 2004). Survey samples The Department of the Environment commissioned surveys of the Lihou Reef and Coringa- Herald national nature reserves. These surveys used several visual census sampling techniques to estimate the presence/abundance of coral, fin fishes and benthic invertebrates, including sea
36 Reducing uncertainty in stock status ABARES cucumbers and trochus. The surveys considered within this report and their primary sampling methods are:
Coringa-Herald, March–April 2003 (Oxley et al. 2003) - snorkel swims of reef flat—500 metres by 5 metres transects (1–5 metres depth) - manta tows of back reef—~325 metres by 2 metres transects Lihou Reef, March 2004 (Oxley et al. 2004; data for Elizabeth and Middleton reefs were included in this dataset but these reefs are outside the CSF and these data was not used) - snorkel swims of reef flat—500 metres by 10 metres transects (2–6 metres) - SCUBA search of reef back, front and flanks—50 metres by 5 metres transects (down to 9 metres) Coringa-Herald, May and October 2007 (Ceccarelli et al. 2008) - snorkel swims of reef flat—500 metres by 10 metres transects - SCUBA search of reef back, front and flanks—500 metres by 5 metres transects (down to 20 metres) Lihou Reef, December 2008 (Ceccarelli et al. 2009) - snorkel swims of reef flat—~500 metres by 10 metres transects - SCUBA search of reef back, front and flanks—500 metres by 5 metres transects. Electronic datasets for each of the surveys were obtained from the Australian Institute of Marine Science (AIMS) for the 2003 and 2004 surveys and from C&R Consulting for the 2007 and 2008 surveys. Records were processed and associated with linear features within a geographic information system (GIS) using the following approaches:
Coringa-Herald 2003 and Lihou Reef 2004—start and finish coordinates were available from AIMS for some transects and the balance where created by digitising the maps published in Oxley et al. (2003, 2004)
Coringa-Herald 2007—start and finish coordinates were available from C&R Consulting for most transects and these where checked and amended by reference to the maps published in Ceccarelli et al. (2008)
Lihou Reef 2008—C&R Consulting provided GPS track data for all transects.
In cases where two samples were associated with one spatial feature (for example, side by side snorkel swims) the samples were regarded as one and the survey results averaged into one record with twice the sampling area. The sample by sample tabular data generally consisted of a record code, sampling method, some location and/or habitat information and densities (per hectare) of species surveyed.
Survey samples from Coringa-Herald and Lihou Reef national nature reserves were associated with geomorphological (habitat) classes, based on the overlap of the survey transect with the habitat class. There were a small number of locations where the survey transects spanned two or more geomorphological classes. In these cases, the survey record was allocated to the habitat type where most of the transect fell.
White teatfish densities for the fore-reef and inner-slope habitats were sourced from surveys of the Torres Strait (Skewes et al. 2004).
37 Reducing uncertainty in stock status ABARES
Population estimation Estimates of population size of four sea cucumber species of interest were obtained by multiplying sea cucumber density in each of five habitats (reef flat, pass, fore-reef, inner slope and sub-tidal reef flat) with the area of these habitats at each of 23 reefs within the fishery. Population densities (and associated error bounds) for each species in each habitat were estimated from survey data. Population density was treated as a stochastic variable so the distribution of possible population sizes could be generated by simulation within each reef. Density Estimates were made of the overall mean population density for each species in each habitat, the uncertainty around this mean (standard error and confidence intervals) and the variance of this density across reefs.
Of the four sea cucumber species and five habitat types, there was sufficient data to analyse six species–habitat combinations in detail. For each species–habitat combination, negative binomial distributions were assumed and were fitted to the survey density data by allowing species density to vary across the 10 to 15 sub-reefs within the Coringa-Herald and Lihou Reef national nature reserves (sub-reef refers to the sub-reef units within the surveyed Coringa-Herald and Lihou Reef). These analyses provided estimates for an overall mean population density for a habitat, an across sub-reef density variance and confidence intervals for these estimates.
For the remaining 13 species–habitat combinations, simple raw density means were calculated, without simultaneously estimating between reef variances. Using the six species–habitat combinations for which there was sufficient data, general relationships between raw means and the confidence intervals for these means, and the across sub-reef density variances, were estimated using regression. These relationships are crude but provide some basis for estimating parameters that cannot be directly estimated in instances of limited data. Between reef density variance was predicted as 9.09 raw mean density and overall density mean standard error was predicted by square root of the raw mean density/n, where n is the number of quadrats in a survey. Negative binomial distributions were then sampled for each simulation of the 19 species–habitat combinations of interest using the predicted population parameters. Population size Simulation was used to estimate the range of possible population sizes for each species. This involved 5 000 random draws of likely negative binomial parameters used to generate likely sea cucumber densities at each reef. Through this approach, mean and median population sizes were estimated, along with 20th and 80th population percentiles. Assumptions Several assumptions apply to the population estimates and their associated errors:
between reef density variance was the same as between sub-reef density variance within the reef complex’s surveyed
distribution of sea cucumber densities across the Coral Sea could be described by the negative binomial distribution, which is a mixture of Poisson distributions whose means are described by a gamma distribution
distributions of species with smaller sample sizes were assessed to be similar to the distributions of species with larger sample sizes. That is, all sea cucumber species had the same relationship between overall mean population density, the standard error of the
38 Reducing uncertainty in stock status ABARES
mean and across reef gamma density and these relationships were assessed to hold across habitats. Mean animal weight A mean animal weight is required to convert numerical estimates of abundance (numbers) into biomass (tonnes). Mean animal weight was estimated from fishery dependent data (logbook data). These data will be influenced by the sampling methods of the fishery and any processing of the animal before the logbook is completed. If, as suspected, weights are reported after some processing, then the real live, wet, unprocessed average animal weight is likely to be larger than that reported. Fishers may also preferentially select for larger individuals for ease of processing and/or higher unit price. This would bias the average animal weight up. Minimum size limits may also increase the average animal size.
Mean weights for each species were calculated for each reef and year (sum of kilograms taken/number of individuals taken). Two-thousand bootstrapped replicates of the aggregated means obtained in this way were simulated. The means of the bootstrapped replicates for each statistic were then used as weight value for each species. Biomass estimation Having estimated population size for each species at each reef, biomass was estimated by simple multiplication of the numerical population size (such as the median or 20th percentile) with average animal weight. Yield estimation Sustainable yield was estimated using a formula adapted from Gulland (1983) by Perry et al. (1999) and used by Skewes et al. (2004) in assessing Torres Strait sea cucumber stocks:
MSY=0.2MB
where M = natural mortality and B = biomass.
There are no direct estimates of natural mortality for sea cucumbers in the CSF. Estimates of natural mortality from Skewes et al. (2004) were halved in this assessment to build in an additional level of conservatism (Table 3.2). This measure was considered appropriate given the assumptions of the formula used, uncertainties in the model structure and particulars of sea cucumber biology.
Table 3.2 Natural mortality estimates used by Skewes et al. (2004) for the Torres Strait and those used in these analyses
Species This study Torres Strait Black teatfish 0.3 0.6 White teatfish 0.3 0.6 Prickly redfish 0.3 0.6 Surf redfish 0.4 0.8 Source: Skewes et al. 2004
Surplus production models Two surplus production models of population dynamics were developed to investigate the response of stocks to historical fishing levels. Although these stocks may have been exploited before 1997–98, ABARES was unable to obtain data to inform this assumption. As such, any catch before 1998 is not incorporated into these production models and populations as at 1997–
39 Reducing uncertainty in stock status ABARES
98 are assumed to be unfished. Both surplus production models used an estimate of unfished sea cucumber biomass at the start of the simulation period. Biomass was modelled at both the reef level and the total fishery level. The first model followed a Schaefer form for surplus production.
The Schaefer model was: B = (Bt – C) + [((4*MSY)/B0) * (Bt*(1-Bt/B0))]
where:
B = biomass in the current year
B0 = the calculated biomass
Bt = biomass from the previous season
C = catch from the previous season
MSY = 0.2*M*B. MSY in this case is pre-determined and not estimated within the model.
This model achieves maximum surplus production at 50 per cent of the pre-fished level, with the rate of production declining both below and above this point.
The second surplus production model used a Pella–Tomlinson production curve (Haddon 2001). Maximum production was set to occur at 75 per cent of the pre-fished biomass. This model was intended to reflect what is understood of the Allee effect (Allee 1939), where a decrease in the breeding population leads to reduced production and/or survival of eggs or offspring (effectively depensation). This concept may explain the slow recovery of overfished stocks such as sandfish (Holothuria scabra) in the Torres Strait (Flood & George 2013).
The Pella–Tomlinson model was: B = (Bt – C) + (r/p) Bt (1-(Bt/K)p)
where:
B = biomass in the current year
Bt = biomass from the previous year
C = catch from the previous season
r = the rate of production (0.88 for surf redfish and 0.67 for the other species)
p = the asymmetrical term that fixes surplus production at a specified level, in this case B75 (7.25 for surf redfish and 7.36 for other species)
K = the carrying capacity or in this case, the biomass estimate. Results Density Mean population density across reefs is shown in Table 3.3 for six species–habitat combinations with sufficient data. For the remaining 13 species–habitat combinations, a raw mean density was calculated. Raw mean density is derived from the raw counts of sea cucumbers as measured in sampling transects, and mean population density is the mean population density across reefs that is calculated in conjunction with population variances. These raw means have large errors associated with them, as a result of the small size of the samples (n, Table 3.3) from which they were calculated.
40 Reducing uncertainty in stock status ABARES
Overall, black teatfish presented the highest overall mean population density. The inner slope habitat followed by reef flat and fore reef showed the highest density for black teatfish. Prickly redfish showed the highest mean population density across reefs in the fore reef habitat followed by inner slope. Surf redfish presented the lowest mean population density (Table 3.3).
Table 3.3 Parameters (per hectare) used to estimate population size
Species Habitat n. Raw Overall mean Lower Upper α β Reef mean population confidence confidence density density density (Ha-1) interval (95%) interval (95%) variance (Ha-1) (Ha-1) (Ha-1) Black Fore reef a 14 3.93 3.394 1.38 9.5 0.3132 10.58 34.61 teatfish Inner slope a 10 8 7.339 3.61 16.18 0.743 9.626 67.65 Reef flat a 15 4.33 4.141 2.25 7.98 0.7833 5.155 20.62 Sub tidal reef flat 3 2.33 – – – – – – Prickly Fore reef a 14 3.86 2.96 0.88 12.05 0.159 18.07 50.91 redfish Inner slope a 10 2.2 1.811 0.55 6.31 0.306 5.561 9.25 Pass 2 16.5 – – – – – – Reef flat 7 0 – – – – – – Sub tidal reef flat 3 2.33 – – – – – – Surf Fore reef 14 0.35 – – – – – – redfish Inner slope 10 0.6 – – – – – – Pass 2 0 – – – – – – Reef flat a 15 1.13 1.056 0.42 2.66 0.4151 2.427 2.44 Sub tidal reef flat 3 0.33 – – – – – – White Pass 2 1 – – – – – – teatfish Reef flat 15 0.07 – – – – – – Sub tidal reef flat 3 0 – – – – – – Fore reef 159 4.29 – – – – – – Inner slope 159 4.29 – – – – – – a Parameters estimated from data, and otherwise predicted from raw mean density and standard error of the overall population density. Notes: Raw mean density is the overall population density estimated from counts of sea cucumbers in transects. n is the number of transects used to estimate population densities. Overall mean population density is the mean population density across reefs. Alpha (α) and beta (β) are negative binomial parameters describing across reef population densities. Reef density variance is the between reef population density for a species–habitat.
Habitat area The habitat area by reef for the five habitats of interest are provided in
41 Reducing uncertainty in stock status ABARES
Table 3.4. A significant proportion of these habitats (28–43 per cent) occur within the Coringa- Herald and Lihou Reef national nature reserves. These habitat types are thought to best reflect the habitat of the species of interest, although estimates of habitat area for all habitat classes are provided at Appendix B. This work does not consider any other marine reserves such as those established by the Australian Government in 2012.
42 Reducing uncertainty in stock status ABARES
Table 3.4 Habitat areas (in hectares) of the five habitat classes of interest
Reef name Fore reef Inner slope Pass Reef flat Sub-tidal reef flat Open reefs Abington Reefs 122 – – 257 – Bougainville Reef 233 – – 558 – Dart Reef 190 – – 396 – Diamond Islets 304 – – 580 4 158 Dieanne Bank 801 – 1 610 43 14 495 Flinders Reefs 2 541 5 116 7 034 4 543 2 135 Flora Reef 394 716 – 799 159 Frederick Reefs 1 087 – 126 628 293 Herald Surprise 272 – – 417 – Holmes Reefs 1 620 4 894 – 5 408 698 Kenn Reef 2 879 4 623 1 147 1 605 2 688 Moore Reefs – – – 531 224 Osprey Reef 1 110 2 245 91 4 077 47 Tregrosse Reefs 158 – – 102 2 562 Willis Islets 3 101 – – 703 6 289 Wreck Reefs 856 – 383 1 528 5 Reefs in reserves Coringa Islets 2 201 – – 219 10 819 Lihou Reefs 9 014 10 673 7 064 6 548 4 373 Herald Cays 386 – – 1 414 – Magdelaine Cays 224 – – 533 – Total open 15 669 17 595 10 390 22 174 33 752 Total reserves 11 825 10 673 7 064 8 714 15 191 Percentage in reserves 43 38 40 28 31 Total 27 495 28 268 17 454 30 888 48 944 Average animal weight estimates The average weights of the four key species of sea cucumber in the CSF are presented in Table 3.5. Table 3.5 also presents the average animal weight reported by Skewes et al. (2004) in the Torres Strait for comparison. CSF average animal weight is notably smaller than those of the Torres Strait, most likely as a result of weight being recorded after some processing.
Overall, the average animal weights from the CSF are considered appropriate for the purposes of this assessment because comparisons of reported catch with TACs are likely to be in roughly the same units (level of processing).
Table 3.5 Average animal weight for the four key commercial sea cucumber species in the Coral Sea Fishery
Species Weight (kg; s.e.)—this study Weight (kg; s.e.)—Torres Strait Black teatfish 0.99 (s.e. 0.20) 1.70 (s.e. 0.19) White teatfish 1.28 (s.e. 0.27) 2.34 (s.e. 0.28) Surf redfish 0.42 (s.e. 0.14) 0.82 (s.e. 0.12) Prickly redfish 1.56 (s.e. 0.46) 2.81 (s.e. 0.73) s.e. Standard error.
43 Reducing uncertainty in stock status ABARES
Population, biomass and maximum sustainable yield results Population, biomass and MSY estimates for the four key sea cucumber species are presented in Table 3.6 to Table 3.9. Mean, median and 20th and 80th percentile estimates of population size for each species at each reef and across the CSF are provided.
Black teatfish was the most abundant of all species. The highest population was found at Lihou Reefs (median 200 067), with a biomass of 196 066 kilograms and a MSY of 11 764 kilograms. The species was also abundant at Flinders Reefs (115 441), with a biomass of 113 132 kilograms and a MSY of 6 788 kilograms (Table 3.6).
Prickly redfish was the second most abundant species. The highest population was found at Flinders Reefs (126 659), with a biomass of 195 055 kilograms and a MSY of 11 703 kilograms. The species was also abundant at Lihou Reefs (139 494) with a biomass of 215 521 kilograms and a MSY of 12 931 kilograms (Table 3.8).
Surf redfish and white teatfish show considerably lower population numbers. The highest population for surf redfish was found at Lihou Reefs (21 720), with a biomass of 8 905 kilograms and a MSY of 712 kilograms, followed by Coringa Islets (1687), with a biomass of 692 kilograms and a MSY of 55 kilograms (Table 3.9). White teatfish was most abundant at Flinders Reefs (4 686), with a biomass of 5 998 kilograms and a MSY of 360 kilograms. The species was also abundant at Kenn Reef (3 037), with a biomass of 3 887 kilograms and a MSY of 233 kilograms.
These results are likely to be underestimates of the true population size at some reefs and for some species (white teatfish and surf redfish) because of the paucity of data. A particularly large degree of uncertainty is associated with those population samples drawn using predicted population parameters, which generated a wider range of population draws than would otherwise be the case. The 20th percentile population estimate for a number of reefs was zero, as a result of the distribution of simulated population sizes for the species less common in the survey data. Estimates for black teatfish and prickly redfish are substantially larger and less uncertain than those of white teatfish and surf redfish. However, some caution should be applied to the estimates of MSY for these species. It would not be sensible to go from the current TAC for black teatfish (1 tonne) to the median MSY estimate of nearly 20 tonnes (open fishery). The reef level estimates of MSY and how possible localised depletion would be prevented should be considered.
44 Reducing uncertainty in stock status ABARES
Table 3.6 Black teatfish population (numbers), biomass (kg) and maximum sustainable yield (kg) estimates by reef
Reef Number Biomass (kg) MSY (kg) mean median 20th 80th mean median 20th 80th mean median 20th 80th percentile percentile percentile percentile percentile percentile Open reefs Abington Reefs 1 201 787 233 1 926 1 177 771 228 1 887 71 46 14 113 Bougainville Reef 2 442 1 616 494 4 003 2 393 1 584 484 3 923 144 95 29 235 Dart Reef 1 863 1 240 363 3 003 1 826 1 215 356 2 943 110 73 21 177 Diamond Islets 12 513 5 652 1 719 17 843 12 263 5 539 1 685 17 486 736 332 101 1 049 Dieanne Bank 53 290 28 796 11 193 73 644 52 224 28 220 10 969 72 171 3 133 1 693 658 4 330 Flinders Reefs 133 103 115 441 63 290 192 646 130 441 113 132 62 024 188 793 7 826 6 788 3 721 11 328 Flora Reef 10 263 8 428 4 051 15 332 10 058 8 259 3 970 15 025 603 496 238 902 Frederick Reefs 8 183 5 671 2 483 12 225 8 019 5 558 2 433 11 981 481 333 146 719 Herald Surprise 2 247 1 473 441 3 603 2 202 1 444 432 3 531 132 87 26 212 Holmes Reefs 63 961 50 865 23 859 96 008 62 682 49 848 23 382 94 088 3 761 2 991 1 403 5 645 Kenn Reef 71 698 60 080 31 879 104 650 70 264 58 878 31 241 102 557 4 216 3 533 1 874 6 153
45
Moore Reefs 1 857 1 085 263 3 011 1 820 1 063 258 2 951 109 64 15 177 Osprey Reef 35 414 28 379 13 308 53 521 34 706 27 811 13 042 52 451 2 082 1 669 783 3 147 Tregrosse Reefs 6 672 2 413 622 9 410 6 539 2 365 610 9 222 392 142 37 553 Willis Islets 29 705 17 569 5 039 47 565 29 111 17 218 4 938 46 614 1 747 1 033 296 2 797 Wreck Reefs 10 846 8 659 3 953 16 275 10 629 8 486 3 874 15 950 638 509 232 957 Reefs in reserves Coringa Islets 37 344 17 175 12 825 48 800 36 597 16 832 12 569 47 824 2 196 1 010 754 2 869 Herald Cays 5 224 3 386 1 015 8 478 5 120 3 318 995 8 308 307 199 60 499 Lihou Reefs 218 366 200 067 165 710 261 505 213 999 196 066 162 396 256 275 12 840 11 764 9 744 15 376 Magdelaine Cays 2 334 1 555 464 3 734 2 287 1 524 455 3 659 137 91 27 220 Total 708 526 560 337 343 204 977 182 694 355 549 130 336 340 957 638 41 661 32 948 20 180 57 458 Total open 445 258 338 154 163 190 654 665 436 353 331 391 159 926 641 572 26 181 19 883 9 596 38 494 Total reserves 263 268 222 183 180 014 322 517 258 003 217 739 176 414 316 067 15 480 13 064 10 585 18 964 Percentage in reserves 37% 40% 52% 33% 37% 40% 52% 33% 37% 40% 52% 33%
Reducing uncertainty in stock status ABARES
Table 3.7 White teatfish population (numbers) biomass (kg) and maximum sustainable yield (kg) estimates by reef
Reef Number Biomass (kg) MSY (kg) mean median 20th percentile 80th percentile mean median 20th percentile 80th percentile mean median 20th percentile 80th percentile Open reefs Abington Reefs 93 0 0 38 119 0 0 49 7 0 0 3 Bougainville Reef 186 0 0 86 238 0 0 110 14 0 0 7 Dart Reef 146 0 0 71 187 0 0 91 11 0 0 5 Diamond Islets 230 0 0 93 294 0 0 119 18 0 0 7 Dieanne Bank 2 104 134 0 2 604 2 693 172 0 3 333 162 10 0 200 Flinders Reefs 16 945 4 686 304 27 040 21 690 5 998 389 34 611 1 301 360 23 2 077 Flora Reef 1 514 260 6 2 099 1 938 333 8 2 687 116 20 0 161 Frederick Reefs 830 36 0 723 1 062 46 0 925 64 3 0 56 Herald Surprise 216 0 0 83 276 0 0 106 17 0 0 6 Holmes Reefs 9 559 1 518 33 13 001 12 236 1 943 42 16 641 734 117 3 998 Kenn Reef 10 970 3 037 198 16 579 14 042 3 887 253 21 221 842 233 15 1 273
46 Moore Reefs 31 0 0 0 40 0 0 0 2 0 0 0
Osprey Reef 4 420 962 55 6 111 5 658 1 231 70 7 822 339 74 4 469 Tregrosse Reefs 112 0 0 43 143 0 0 55 9 0 0 3 Willis Islets 2 022 2 0 838 2 588 3 0 1 073 155 0 0 64 Wreck Reefs 1 054 68 0 1 281 1 349 87 0 1 640 81 5 0 98 Reefs in reserves Coringa Islets 0 0 0 0 0 0 0 0 0 0 0 0 Herald Cays 376 0 0 153 481 0 0 196 29 0 0 12 Lihou Reefs 6 885 24 0 5 308 8 813 31 0 6 794 529 2 0 408 Magdelaine Cays 172 0 0 78 220 0 0 100 13 0 0 6 Total 57 865 10 727 596 76 229 74 067 13 731 763 97 573 4 444 824 46 5 854 Total open 50 432 10 703 596 70 690 64 553 13 700 763 90 483 3 873 822 46 5 429 Total reserves 7 433 24 0 5 539 9 514 31 0 7 090 571 2 0 425 Percentage in reserves 13% 0% 0% 7% 13% 0% 0% 7% 13% 0% 0% 7%
Reducing uncertainty in stock status ABARES
Table 3.8 Prickly redfish population (numbers) biomass (kg) and maximum sustainable yield (kg) estimates by reef
Reef Number Biomass (kg) MSY (kg) mean median 20th 80th mean median 20th 80th mean median 20th 80th percentile percentile percentile percentile percentile percentile Open reefs Abington Reefs 936 376 39 1 535 1 441 579 60 2 364 86 35 4 142 Bougainville Reef 1 745 685 65 2 873 2 687 1 055 100 4 424 161 63 6 265 Dart Reef 1 469 569 59 2 413 2 262 876 91 3 716 136 53 5 223 Diamond Islets 11 784 4 665 884 17 030 18 147 7 184 1 361 26 226 1 089 431 82 1 574 Dieanne Bank 66 151 45 726 22 215 92 026 101 873 70 418 34 211 141 720 6 112 4 225 2 053 8 503 Flinders Reefs 146 709 126 659 67 757 214 353 225 932 195 055 104 346 330 104 13 556 11 703 6 261 19 806 Flora Reef 4 367 2 655 804 7 070 6 725 4 089 1 238 10 888 404 245 74 653 Frederick Reefs 11 357 6 522 2 849 16 783 17 490 10 044 4 387 25 846 1 049 603 263 1 551 Herald Surprise 1 999 765 75 3 254 3 078 1 178 116 5 011 185 71 7 301 Holmes Reefs 20 500 12 184 3 567 33 160 31 570 18 763 5 493 51 066 1 894 1 126 330 3 064 Kenn Reef 53 588 42 923 22 138 78 321 82 526 66 101 34 093 120 614 4 952 3 966 2 046 7 237
47
Moore Reefs 559 91 1 811 861 140 2 1 249 52 8 0 75 Osprey Reef 13 114 7 877 2 883 20 193 20 196 12 131 4 440 31 097 1 212 728 266 1 866 Tregrosse Reefs 7 521 2 692 592 10 769 11 582 4 146 912 16 584 695 249 55 995 Willis Islets 38 944 22 336 4 743 62 603 59 974 34 397 7 304 96 409 3 598 2 064 438 5 785 Wreck Reefs 13 056 9 705 4 812 19 346 20 106 14 946 7 410 29 793 1 206 897 445 1 788 Reefs in reserves Coringa Islets 40 689 19 230 14 494 51 934 62 661 29 614 22 321 79 978 3 760 1 777 1 339 4 799 Herald Cays 2 821 1 113 117 4 564 4 344 1 714 180 7 029 261 103 11 422 Lihou Reefs 159 429 139 949 82 942 224 739 245 521 215 521 127 731 346 098 14 731 12 931 7 664 20 766 Magdelaine Cays 1 764 676 67 2 826 2 717 1 041 103 4 352 163 62 6 261 Total 598 502 447 398 231 103 866 603 921 693 688 993 355 899 1 334 569 55 302 41 340 21 354 80 074 Total open 393 799 286 430 133 483 582 540 606 450 441 102 205 564 897 112 36 387 26 466 12 334 53 827 Total reserves 204 703 160 968 97 620 284 063 315 243 247 891 150 335 437 457 18 915 14 873 9 020 26 247 Percentage in reserves 34% 36% 42% 33% 34% 36% 42% 33% 34% 36% 42% 33%
Reducing uncertainty in stock status ABARES
Table 3.9 Surf redfish population (numbers), biomass (kg) and maximum sustainable yield (kg) estimates by reef
Reef Number Biomass (kg) MSY (kg) mean median 20th 80th mean median 20th 80th mean median 20th 80th percentile percentile percentile percentile percentile percentile Open reefs Abington Reefs 135 6 0 153 55 2 0 63 4 0 0 5 Bougainville Reef 249 14 0 282 102 6 0 116 8 0 0 9 Dart Reef 198 9 0 212 81 4 0 87 6 0 0 7 Diamond Islets 1 639 57 0 857 672 23 0 351 54 2 0 28 Dieanne Bank 5 408 15 0 892 2 217 6 0 366 177 0 0 29 Flinders Reefs 6 113 935 22 7 952 2 506 383 9 3 260 201 31 1 261 Flora Reef 848 136 3 1 111 348 56 1 456 28 4 0 36 Frederick Reefs 675 43 0 641 277 18 0 263 22 1 0 21 Herald Surprise 219 11 0 240 90 5 0 98 7 0 0 8 Holmes Reefs 5 355 843 26 6 888 2 196 346 11 2 824 176 28 1 226 Kenn Reef 5 339 670 14 6 013 2 189 275 6 2 465 175 22 0 197
48
Moore Reefs 241 10 0 264 99 4 0 108 8 0 0 9 Osprey Reef 2 857 454 11 4 040 1 171 186 5 1 656 94 15 0 133 Tregrosse Reefs 803 15 0 243 329 6 0 100 26 0 0 8 Willis Islets 3 554 111 1 2 016 1 457 46 0 827 117 4 0 66 Wreck Reefs 797 41 0 879 327 17 0 360 26 1 0 29 Reefs in reserves Coringa Islets 5 114 1 687 1 102 2 730 2 097 692 452 1 119 168 55 36 90 Herald Cays 551 25 0 622 226 10 0 255 18 1 0 20 Lihou Reefs 23 347 21 720 18 544 26 143 9 572 8 905 7 603 10 719 766 712 608 857 Magdelaine Cays 230 11 0 252 94 5 0 103 8 0 0 8 Total 63 672 26 813 19 723 62 430 26 106 10 993 8 086 25 596 2 088 879 647 2 048 Total open 34 430 3 370 77 32 683 14 116 1 382 32 13 400 1 129 111 3 1 072 Total reserves 29 242 23 443 19 646 29 747 11 989 9 612 8 055 12 196 959 769 644 976 Percentage in reserves 46% 87% 100% 48% 46% 87% 100% 48% 46% 87% 100% 48%
Reducing uncertainty in stock status ABARES
Surplus production model results Table 3.10 provides an estimate of black teatfish and prickly redfish biomass in 2010 as a proportion of biomass at the start of the modelling period (1997–98 season), estimated by the Schaefer and the Pella–Tomlinson surplus production models using the 20th percentile biomass estimate. Outputs of the two surplus production models estimated biomass in 2010 to be greater than 99 per cent of that at the start of the modelling period for black teatfish and prickly redfish stocks.
Estimates for white teatfish and surf redfish were implausible (reduced to zero) and are therefore not presented here (Table 3.10).
Table 3.10 Proportion of biomass remaining, 20th percentile starting biomass, Schaefer model and Pella–Tomlinson model
Reef Schaefer model Pella–Tomlinson model Black teatfish Prickly redfish Black teatfish Prickly redfish Open reefs Abington Reefs 100% 100% 100% 100% Bougainville Reef 98% 94% 100% 100% Dart Reef 100% 93% 100% 97% Diamond Islets 94% 66% 99% 68% Dieanne Bank 100% 100% 100% 100% Flinders Reefs 99% 100% 100% 100% Flora Reef 98% 97% 100% 100% Frederick Reefs 86% 90% 89% 93% Herald Surprise – – – – Holmes Reefs 99% 89% 100% 98% Kenn Reef 98% 99% 99% 100% Moore Reefs 88% – 93% – Osprey Reef 100% 48% 100% 23% Tregrosse Reefs 86% 39% 91% 35% Willis Islets 100% 100% 100% 100% Wreck Reefs – 97% – 100% Reefs in reserves Coringa Islets na na 100% 100% Herald Cays na na 100% 100% Lihou Reefs 100% 100% 100% 100% Magdelaine Cays na na 100% 100% Total biomass 99% 99% 100% 100% Note: A dash (–) appears where the estimate of biomass at a given reef was reduced to zero. Table 3.11 provides an estimate of the proportion of stock biomass remaining on reefs in the CSF in 2010, estimated by the Schaefer surplus production model using the median biomass estimate.
Table 3.12 provides an estimate of the proportion stock biomass remaining on reefs in the CSF in 2010, estimated by the Pella–Tomlinson surplus production model using the median biomass estimate.
49 Reducing uncertainty in stock status ABARES
Table 3.11 Proportion of biomass remaining, median starting biomass, Schaefer model
Reef Black teatfish White teatfish Surf redfish Prickly redfish Open reefs Abington Reefs 100% – – 100% Bougainville Reef 100% – – 100% Dart Reef 100% – – 99% Diamond Islets 98% – 100% 95% Dieanne Bank 100% 95% 100% 100% Flinders Reefs 99% – – 100% Flora Reef 99% 99% – 99% Frederick Reefs 94% – – 96% Herald Surprise 94% – – 97% Holmes Reefs 99% – 98% 98% Kenn Reef 99% 82% – 99% Moore Reefs 97% – – – Osprey Reef 100% – 92% 93% Tregrosse Reefs 97% – – 90% Willis Islets 100% – 94% 100% Wreck Reefs 85% 100% – 99% Reefs in reserves Coringa Islets na na na na Herald Cays na na na na Lihou Reefs 100% 100% 100% 100% Magdelaine Cays na na na na Total biomass 99% – 70% 100% Note: A dash (–) appears where the estimate of biomass at a given reef was reduced to zero.
50 Reducing uncertainty in stock status ABARES
Table 3.12 Proportion of biomass remaining, median starting biomass, Pella–Tomlinson model
Reef Black teatfish White teatfish Surf redfish Prickly redfish Open reefs Abington Reefs 100% – – 100% Bougainville Reef 100% – – 100% Dart Reef 100% – – 100% Diamond Islets 100% – 100% 99% Dieanne Bank 100% 100% 100% 100% Flinders Reefs 100% – – 100% Flora Reef 100% 100% – 100% Frederick Reefs 95% – – 97% Herald Surprise 100% – – 100% Holmes Reefs 100% – 100% 100% Kenn Reef 100% 89% – 100% Moore Reefs 99% – – – Osprey Reef 100% – 100% 99% Tregrosse Reefs 99% – – 94% Willis Islets 100% – 100% 100% Wreck Reefs 85% 100% – 100% Reefs in reserves Coringa Islets 100% – 100% 100% Herald Cays 100% – 100% 100% Lihou Reefs 100% 100% 100% 100% Magdelaine Cays 100% – 100% 100% Total biomass 100% – 91% 100% Note: A dash (–) appears where the estimate of biomass at a given reef was reduced to zero.
Using median biomass estimates, total biomass for black teatfish, prickly redfish and surf redfish was greater than 70 per cent of the initial (1997–98 season) biomass for both models at the end of the modelling period (2010). For white teatfish, the modelling resulted in a biomass reducing to zero in 2010—a result that is not plausible and is likely to result from an underestimate of the initial biomass. Discussion
The overall objective of this assessment was to reconcile the status of sea cucumber stocks in the CSF. A number of approaches were examined and a habitat-based approach with production modelling was considered to have the best chance of producing outputs useful for determining status.
The production modelling relies on an initial biomass that is reflective of the true biomass. ABARES confidence in the estimates is greater for black teatfish and prickly redfish but is low for white teatfish and surf redfish. Given that the accumulated catch for white teatfish and surf redfish exceeds the estimates of biomass at some reefs, the biomass estimates for these stocks cannot be considered reliable. This shortcoming stems from the poor representation of these species in the survey data, in part because the surveys were conducted on only a few reefs and focused on shallower habitats, possibly less suited to these two species.
51 Reducing uncertainty in stock status ABARES
Black teatfish and prickly redfish The estimated biomass in 2010 was greater than 99 per cent of that at the start of the simulation period (1997–98) for both stocks using both the median (Table 3.11 and Table 3.12) and 20th percentile biomass estimate (Table 3.10). As a result, both stocks are classified as not overfished and not subject to overfishing. Surf redfish Surf redfish population sizes and MSY are considered to be underestimated by this assessment. As a result, biomass from the estimates was not sufficiently high to support historical catches on some reefs. The surf redfish median MSY estimate for the fishery (all reefs) was 879 kilograms. Catch for surf redfish in the 2008–09 season was zero. Total fishery catch of this species between 2008–09 and 2010–11 was less than the median [under] estimate of MSY. On this basis, surf redfish are classified as not subject to overfishing.
Cato Reef was not included in this analysis because of missing habitat information. Cato Reef has supported approximately 64 per cent of the total fishery catch of surf redfish between 1997–98 and 2008–09 and some analysis of the effect of historical catch at this reef should be undertaken when data are available. Although catch from the balance of the reefs assessed has exceeded the mean estimate of MSY in three of the 14 seasons between 1997–98 and 2010–11 (with the largest catch being approximately double the estimate of MSY and the other two years substantially less), catch in the remaining 11 seasons is substantially less than the median estimate of MSY. The surplus production modelling resulted in 70 per cent depletion from initial biomass at the end of the modelling period for the Schaefer model (Table 3.11) and 91 per cent for the Pella–Tomlinson model Table 3.12). Given these results, surf redfish (excluding Cato Reef) could be classified as not overfished. White teatfish A comparison of catch (at the reef level) with the estimated biomass (at the reef level) showed that these assessment results underestimate biomass for white teatfish. The catch of white teatfish in 2008–09 was 1 799 kilograms, which is above the median MSY estimate (824 kilograms) but below the mean MSY estimate (4 444 kilograms). The results of the surplus production modelling have been provided for white teatfish, but are not considered reliable.
While we do not have a conclusive evidence base for status determination, catches since 2002– 03 have been below the mean MSY estimate. However, the catch of white teatfish is more than double the catch of any other species within the CSF, with approximately 70 per cent of the catch being taken in the three seasons from 1999–2000 to 2001–02. The uncertainties associated with this species within these analyses do not allow us to evaluate what effect these historical catches may have had on the stock. On this basis, the stock remains uncertain with respect to biomass and fishing mortality. General comments The status determination process described in this chapter was undertaken at the fishery level. These analyses show that populations of sea cucumbers and historical catch are not evenly distributed across the fishery. There is a reef rotation component to the fishery’s harvest strategy (AFMA 2008), but there may be scope for revision based on the information from this report. Revision could take account of the potential habitat areas (Table 3.4), estimates of population size and sustainable yield and the regularity with which reefs are open to fishing. The size of the fishery and the distances between viable and profitable populations of sea cucumbers should also be taken into account in any revision of the harvest strategy.
52 Reducing uncertainty in stock status ABARES
The scarcity of data significantly impeded the ability of this study to produce reliable results for some species. The robustness of outputs from this project could be substantially improved by dedicated survey operations across the fishery—possibly coupled with commercial operations of the fishery in order to contain costs.
Some caution should be applied to the estimates of MSY for the species considered in this report. The authors would not recommend strict application of median MSY estimates as TACs for future application. For example, it would not be sensible to go from the current TAC for black teatfish (1 tonne) to the median MSY estimate of nearly 20 tonnes (estimate for open fishery), particularly because total fishery catch for this species up to 2008–09 was only 23 tonnes. The reef level estimates of MSY, how many reefs would be open in a given year and how potential localised depletion could be prevented should all be considered.
53 Reducing uncertainty in stock status ABARES
Appendix B Table B1 Area of all habitat types in the Coral Sea Fishery
Reef Reef flat Forereef Inner Pass Subtidal Enclosed Deep Deep Lagoon Land on Drowned Pass Shallow Shallow Shallow Total slope reef flat lagoon lagoon terrace pinnacle reef bank reef flat lagoon lagoonal terrace terrace Abington 257 122 – – – – – – – – – – – 10 – 389 Reefs Bougainville 558 233 – – – – – – 1 – – – 273 – 290 1 356 Reef Coringa Islets 219 2 201 – – 10 819 – – 228 280 – 35 6 879 – – 352 – 248 785 Dart Reef 396 190 – – – – – – – – – – 415 – – 1 000 Diamond 580 304 – – 4 158 – – 314 409 – 52 – – – 334 – 319 838 Islets Dieanne Bank 43 801 – 1 610 14 495 – 88 382 – – 7 – – – – – 105 338 Flinders Reefs 4 543 2 541 5 116 7 034 2 135 – 68 491 – 108 3 – – – – – 89 971 Flora Reef 799 394 716 – 159 – 271 – – – – – – – – 2 338
54 Frederick 628 1 087 – 126 293 – – 6 287 – – – – – 306 – 8 727
Reefs Herald Cays 1 414 386 – – – 141 – 4 958 – 73 – – – – – 6 972 Herald 417 272 – – – – – – 10 – – – 359 – – 1 059 Surprise Holmes Reefs 5 408 1 620 4 894 – 698 343 5 881 1 275 – – – – – – – 20 120 Kenn Reef 1 605 2 879 4 623 1 147 2 688 – 14 039 – 7 3 – – – – – 26 990 Lihou Reefs 6 548 9 014 10 673 7 064 4 373 – 201 195 – 21 120 – – – – – 239 008 Magdelaine 533 224 – – – – – – – 51 – – – 283 – 1 092 Cays Moore Reefs 531 – – – 224 – – – – – – – – 175 – 930 Osprey Reef 4 077 1 110 2 245 91 47 – 11 369 – 4 – – 3 – – – 18 945 Tregrosse 102 158 – – 2 562 – – – – 8 – – – – – 2 831 Reefs Willis Islets 703 3 101 – – 6 289 – – 62 010 – 46 – – – 1 007 – 73 155
54 Reducing uncertainty in stock status ABARES
Wreck Reefs 1 528 856 – 383 5 – – 15 883 – 17 – – – – – 18 671 unknown – – – – – – – – – – 1 578 – – – – 1 578 unknown – – – – – – – – – – 2 205 – – – – 2 205 Total 30 888 27 495 28 268 17 454 48 944 484 389 627 633 103 151 415 10 662 3 1 047 2 467 290 1 191 298 Marion Reef 4 058 3 154 10 137 5 363 735 – 64 070 – 64 3 – – – – – –
55
55 Reducing uncertainty in stock status ABARES
References
AFMA 2008, Harvest strategy—Coral Sea Fishery, hand collection sector: sea cucumber, Australian Fisheries Management Authority, Canberra. Allee, WC 1939, Animal Aggregations: A Study in General Sociology, University of Chicago Press, Chicago, Illinois.
Carle, F & Strub, M 1978, ‘A new method for estimating population size from removal data’, Biometrics, vol. 34, pp. 621–830.
Ceccarelli, D, Choat, J, Ayling, AM, Richards, Z, van Herwerden, L, Ayling, AL, Ewels, G, Hobbs, J & Cuff, B 2008, Coringa-Herald National Nature Reserve Marine Survey—2007, report to the Department of the Environment, Water, Heritage and the Arts by C&R Consulting and James Cook University.
Ceccarelli, D, Ayling, AM, Choat, JH, Ayling, AL, Williamson, DH & Cuff, B 2009, Lihou Reef National Nature Reserve Marine Survey—2008, report to the Department of the Environment, Water, Heritage and the Arts by C&R Consulting, Townsville.
Cowx, I 1983, ‘Review of the methods for estimating fish population size from survey removal data’, Fish Management, vol. 14, issue 2, pp. 67–82.
DeLury, D 1947, ‘On the estimation of biological populations’, Biometrics, vol. 3, pp. 145–67.
Flood, M & George, D 2013, ‘Torres Strait Bêche de mer and Trochus fisheries’, in Woodhams, J, Vieira S, & Stobutzki, I (eds) 2013, Fishery status reports 2012: status of fish stocks and fisheries managed by the Australian Government, Australian Bureau of Agricultural and Resource Economics – Bureau of Rural Sciences, Canberra. Gulland, J 1983, Fish Stock Assessment; a manual of basic methods, FAO/Wiley Inter-Science, New York.
Haddon, M 2001, Modelling and Quantitative Methods in Fisheries, Chapman and Hall/CRC. Florida, USA.
Haddon, M (ed.) 2012, Reducing uncertainty in stock status: harvest strategy testing, evaluation, and development—general discussion and summary, CSIRO Marine and Atmospheric Research Report.
Hunter, C, Skewes, T, Burridge, C & Dennis, D 2002, Research for management of the Coral Sea Collector Fishery (Sea cucumber), CSIRO Division of Marine Research, Cleveland, Australia.
Oxley, W, Ayling, A, Cheal, A & Thompson, A 2003, Marine Surveys undertaken in the Coringa- Herald National Nature Reserve, March–April 2003, Australian Institute of Marine Science. Townsville, Australia.
Oxley, W, Emslie, M, Muir, P & Thompson, A 2004, Marine Surveys undertaken in the Lihou Reef National Nature Reserve, March 2004, Australian Institute of Marine Science.
Perry, R, Walters, C & Boutillier, J 1999, ‘A framework for providing scientific advice for the management of new and developing invertebrate fisheries’, Reviews in Fish Biology and Fisheries, vol 9, pp. 125–59.
56 Reducing uncertainty in stock status ABARES
Pollock, K 1991, ‘Modelling Capture, Recapture, and Removal Statistics for Estimation of Demographic Parameters for Fish and Wildlife Populations: Past, Present, and Future’, Journal of the American Statistical Association, vol 86, issue 413, pp. 225–38.
Skewes, T, Dennis, D, Koutsoukos, A, Haywood, M, Wassenberg, T & Austin, M 2004, Stock Survey and Sustainable Harvest of Strategies for Torres Strait Bêche-de-mer, CSIRO, Cleveland, Australia.
Woodhams, J, Chambers, M & Pham, T 2010, ‘Coral Sea Fishery’, in Wilson, DT, Curtotti R, Begg GA & Phillips, K (eds) 2010, Fishery status reports 2009: status of fish stocks and fisheries managed by the Australian Government, Australian Bureau of Agricultural and Resource Economics – Bureau of Rural Sciences, Canberra.
57 Reducing uncertainty in stock status ABARES
4 Coral Sea Fishery Line and Trap Sector: preliminary stock assessments
James Larcombe and Justin Roach Summary
The Line and Trap, and Trawl and Trap sectors of the Coral Sea Fishery (CSF) have had an uncertain biological status in the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) Fishery status reports since first being assessed in 2006.
This preliminary assessment considers historical catch levels against conservative yield estimates to potentially reconcile stock status. The assessment is specifically intended to be conservative, in the sense of assuming lower biomass levels and exploitation rates, because of the lack of research information specific to the Coral Sea. This is distinct from a more data- intensive approach that would give some confidence about possible long-term maximum sustainable yields from the stocks. Three separate assemblages of species taken in the combined line and trap sectors were assessed: a deep scalefish assemblage, a reef scalefish assemblage and a shark assemblage. Only a partial assessment was possible for the shark assemblage because of a lack of information. Deep scalefish assemblage Under the most conservative scenario (low biomass and lowest exploitation constant), the yield estimate for the deep scalefish assemblage was 82 tonnes for the CSF as a whole. Exploitation levels have been higher on some individual reefs in the Coral Sea, with catches exceeding the yield estimates under some scenarios. However, the reliability of the yield estimates at the scale of reefs is questionable. Taking the Coral Sea Fishery as a whole, the deep scalefish assemblage would be considered not subject to overfishing and would not be overfished. Reef scalefish assemblage Under the most conservative scenario (low biomass and lowest exploitation constant), the yield estimate for the reef scalefish assemblage was 96 tonnes for the CSF as a whole. Exploitation levels, both peak and mean, have been higher on some individual reefs in the Coral Sea, with catches exceeding the yield estimates under some scenarios. However, the reliability of the yield estimates at the scale of reefs is questionable, with likely significant between-reef differences in biomass per unit of habitat. Taking the CSF as a whole, the reef scalefish assemblage would be considered not subject to overfishing and would not be overfished. Shark assemblage A partial assessment was undertaken for the shark assemblage, with no yield estimation because of a lack of published information on shark densities and the prevalence of some wide ranging species in the catch (such as tiger shark). Shark assemblage exploitation per unit of habitat was calculated assuming the 200 metre isobath as a habitat metric. The mean annual catch of all sharks from the CSF was 21.2 tonnes, which equates to 14.6 kilograms per nautical mile per year. The peak catch of 111 tonnes in 2005–06 equated to a peak of 77 kilograms per nautical mile. No status is proposed for sharks in the CSF from this assessment.
58 Reducing uncertainty in stock status ABARES
Approximately 20 per cent of deep scalefish assemblage habitat and 25 per cent of reef scalefish assemblage habitat in the CSF is located within the Coringa-Herald and Lihou marine reserves and not accessible to fishing. Estimates in this study apply to all habitats in the Coral Sea, including marine reserves.
This report makes no specific recommendations for setting of future catches of any of the assemblages or the species taken within them. However, the results may be used to inform the future harvest strategies, including development of trigger catch levels. Introduction
The Line and Trap, and Trawl and Trap sectors of the CSF extend from Cape York to Sandy Cape, Queensland (Map 4.1). The sectors take a large suite of finfish, sharks and crustaceans.
Map 4.1 Management area of the Line and Trap, and Trawl and Trap sectors of the Coral Sea Fishery
Source: ABARES
Fishing methods allow the use of traps, demersal longlines, demersal and midwater trawl gear, trotlines, droplines, set lines and handlines (Woodhams et al. 2010). If prior approval is obtained from the Australian Fisheries Management Authority (AFMA), automatic baiting equipment can also be used.
In 2009–10 there was only one active vessel in the Line and Trap Sector and there were no active vessels in the Trawl and Trap Sector. There have been no trawl operations since 2008. There are no designated ‘target’ species for the sectors. The Line and Trap, and Trawl and Trap sectors are managed through input controls, including limited entry, spatial closures and size limits. The CSF harvest strategy contains triggers to detect changes in the fishery, designed to lead to investigation of the reasons for the change (Woodhams et al. 2010). Triggers include changes to catch levels and catch composition. Actions following a trigger include analysis of
59 Reducing uncertainty in stock status ABARES
logbook catch and effort data, industry consultation and revised risk analysis. Following this, a management response may be invoked.
Before this study no quantitative stock assessment had been carried out for the sectors. The ABARES Fishery status reports 2010 (the most recent report at the start of this study) and earlier reports classified the sectors as uncertain if overfished. As no trawl or trap operations took place during the 2009–10 season, that sector was classified as not subject to overfishing. The Line and Trap Sector was classified as uncertain if subject to overfishing (Woodhams et al. 2011). Approach to status determination The purpose of this study is to examine the biological status of key commercial stocks taken for the Line and Trap, and Trawl and Trap sectors. An approach was adopted where suites of species are combined and the assemblage as a whole (multispecies assessment) is assessed, while accounting for the life history characteristics of the taxa within the assemblages (and hence their robustness to fishing). The other key aspect of the assessment approach is to place the fishery in the context of measured quantities of habitat, allowing for estimated yields of each assemblage to be scaled to the size of habitat available. This allows the level of catches in the fishery (and estimated yields) to be scaled to the size of the habitat that supports those catches. In addition, comparisons can then be made to similar fisheries where there is more information.
Analysis and results are presented in four sections that reflect the components of the work.
1) habitat assessment
2) deep scalefish assemblage
3) reef scalefish assemblage
4) shark assemblage. Catch and effort statistics
While trawl operations were considered in the early stages of these analyses (some data summaries are provided), they did not progress to the assessment stage because of the limited spatial and temporal data available. Further consideration of the species taken while trawling (predominantly deepwater seamount associated finfish like alfonsino and crustaceans such as deepwater prawns, bugs and scampi) may reconcile the status of these stocks, but they will not be discussed further within these assessments. Commonwealth records for the present combined CSF line and trap sectors commence in 1997–98 (Figure 4.1). It is unclear what fisheries may have existed before this; any such fisheries were probably under Queensland jurisdiction. Line sector Total catches using line fishing methods peaked in 2005–06 at approximately 160 tonnes and have been below 60 tonnes since then (Figure 4.1). Most of the catch has been taken by multi- hook methods such as dropline and demersal longline. Operations are located on or adjacent to reef structures and have fished most reefs since Commonwealth records began, but with a particular focus on some reefs. The large catches of 2003 (Figure 4.1) were taken mainly from southern reefs, moving northward in 2004 and 2005.
The catch is taxonomically diverse (Table 4.1), in part because of the range of depths fished and gears employed but also because of the diversity of predatory fish fauna commonly found in
60 Reducing uncertainty in stock status ABARES
proximity to Indo-Pacific coral reefs (Bellwood & Hughes 2001). The deep slope rosy snapper (Pristipomoides filamentosus; 23.9 per cent of catch weight) and ruby snapper (Etelis carbunculus; 9.5 per cent of catch weight) are important components of the catch, along with a range of other deep slope scalefish taxa (such as Pristipomoides spp., Etelis spp., Aphareus rutilans and Hyperoglyphe antarctica). Sharks have also historically been a prominent part of the line catch weight, particularly blacktip sharks (Carcharhinus spp.; 9.5 per cent), tiger shark (Galeocerdo cuvier; 8.7 per cent), whitetip reef sharks (Triaenodon obesus; 4.2 per cent) and scalloped hammerhead (Sphyrna lewini; 3.5 per cent). A smaller proportion of the total catch is shallower coral reef associated taxa such as Plectropomus spp., Lethrinus spp. and Lutjanus spp. (Table 4.1).
Figure 4.1 Coral Sea Fishery line sector annual financial year catch by fishing method, 1997–98 to 2008–09
180 troll
160 trotline handline 140 dropline (manual) dropline (hydraulic) 120
dropline 100 longline (bottom) longline (auto) Tonnes 80
60
40
20
0 1997- 1998- 1999- 2000- 2001- 2002- 2003- 2004- 2005- 2006- 2007- 2008- 98 99 00 01 02 03 04 05 06 07 08 09 Financial year
Source: Data from AFMA
61 Reducing uncertainty in stock status ABARES
Table 4.1 Coral Sea Fishery line sector annual financial year catch (kg) by species/species group
Common name Species name Annual catch (kg) 2000– 2001– 2002– 2003– 2004– 2005– 2006– 2007– 2008– Total % of 01 02 03 04 05 06 07 08 09 total Rosy jobfish/king Pristipomoides filamentosus 30 075 14 780 35 052 64 171 15 094 8 462 3 221 3 116 1 338 175 309 23.86 snapper Blacktip sharksa Carcharhinus spp. 257 90 108 4 069 6 480 35 536 – 21 719 1 366 69 625 9.48 Northwest ruby fish Etelis carbunculus 4 630 12 213 22 849 12 479 8 083 5 186 647 3 112 279 69 477 9.46 Tiger sharka Galeocerdo cuvier 1 652 – 1 030 – 20 325 24 460 9 866 6 409 523 64 265 8.75 Bar rockcod Epinepbelus ergastularius and E. 1 367 3 568 6 555 9 122 7 994 3 123 1 355 1 721 211 35 016 4.77 septemfasciatus Flame snapper Etelis coruscans – – 188 7 778 12 850 7 992 914 3 485 815 34 021 4.63
5 Whitetip reef sharka Triaenodon obesus 431 85 148 760 5 163 22 707 1 638 – 32 30 964 4.21 Shark othera Sharks—other 5 412 9 930 2 299 69 15 61 22 1 085 – 18 893 2.57 Scalloped hammerheada Sphyrna lewini 550 – – – 1 855 18 615 1 804 2 636 114 25 574 3.48 Jobfish Aphareus rutilans 363 641 5 848 6 331 3 214 1 708 293 439 2 093 20 930 2.85
62 Red emperor Lutjanus sebae 1 769 411 312 1 836 1 531 2 733 775 1 451 183 11 001 1.50 Goldband snappers Pristipomoides multidens and P. typus 561 651 5 126 8 015 422 41 55 – 18 14 889 2.03 Blue eye trevalla Hyperoglyphe antarctica 14 6 000 5 535 48 236 6 5 83 52 11 979 1.63 Coral trout Plectropomus and Variola spp. 3 595 217 2 100 2 733 72 7 – – 6 8 729 1.19 Mixed fish Mixed fish 101 1 647 502 1 560 788 844 372 70 314 6 197 0.84 Grey reef sharka Carcharhinus amblyrhynchos 5 574 – 2 372 – – – 2 120 – – 10 066 1.37 Whaler sharka Carcharhinidae family 900 – – – – – 498 8 437 500 10 335 1.41 Large mouth nannygai Lutjanus malabaricus 676 – 506 161 – 7 4 – – 1 354 0.18
Japanese sea bream Gymnocranius euanus – – – 54 795 4 963 1 896 864 704 9 276 1.26 Green jobfish Aprion virescens 644 844 938 2 775 509 1 421 434 325 585 8 474 1.15 Mozambique bream Wattsia mossambica 530 255 1 023 2 505 2 180 592 163 192 143 7 583 1.03 Amberjack Seriola dumerili 73 223 718 2 970 674 810 152 683 560 6 863 0.93 Rock cods Aethaloperca, Anyperodon and Epinephelus 223 166 205 3 555 230 523 215 405 970 6 492 0.88 spp.
Reducing uncertainty in stock status ABARES
Sea bream snapper Gymnocranius spp 799 864 802 15 21 – – – 959 3 460 0.47 Sandbar sharka Carcharhinus plumbeus – – – – – 4 135 8 1 159 – 5 302 0.72 Long tail/ruby snapper Etelis spp. – 471 1 922 1 048 1 265 – – 199 – 4 905 0.67 a Elasmobranches. Note: Taxa are sorted by total catch since 2000.
5
63
Reducing uncertainty in stock status ABARES
Effort in the line sector, expressed as numbers of operations and hooks set, peaked around 2002 and 2003 (Figure 4.2) but is not high compared with other Commonwealth fisheries. Hydraulic dropline sets dominate the operations counts from 2000 to 2005. However, the number of hooks set is dominated by bottom longline and auto-longline methods.
Significant difficulties were encountered with the recorded effort figures for line methods in the AFMA logbook databases. There appears to be record-to-record inconsistency and confusion between the different effort figures (number of lines, hooks per line, line length) for a single log operation. This made further analysis of hook and line effort difficult; there may be errors in the hook counts for some methods because of this.
Logbook records contain an average depth field, which provides an estimate of the depth at which the gear was set. These data are incomplete with many missing values. Gear may be set across a range of depths so some caution is needed when using such data. The percentage of hooks set across depth classes for each gear is given in Figure 4.3. The predominant dropline and demersal longline gears are set at a wide range of depth from near surface to more than 600 metres. Dropline tends to be focused in the 100–300 metres depth range (shallower for manual and deeper for hydraulic) and auto-longline hooks are set mainly at 100–400 metres depth.
Figure 4.2 Coral Sea Fishery line sector annual effort by fishing method, as (a) numbers of operations (records within the Australian Fisheries Management Authority log database) and (b) number of hooks set (a)
450
400 troll
trotline 350 handline
300 dropline (manual) dropline (hydraulic)
250 dropline
longline (bottom) 200 longline (auto) 150
Number of operations of Number 100
50
0 1997- 1998- 1999- 2000- 2001- 2002- 2003- 2004- 2005- 2006- 2007- 2008- 98 99 00 01 02 03 04 05 06 07 08 09 Financial year
Source: Data from AFMA
64 Reducing uncertainty in stock status ABARES
(b)
300 000 troll trotline 250 000 handline dropline (manual) 200 000 dropline (hydraulic) dropline longline (bottom) 150 000 longline (auto)
100 000 Number of of hooks Number
50 000
0 1997- 1998- 1999- 2000- 2001- 2002- 2003- 2004- 2005- 2006- 2007- 2008- 98 99 00 01 02 03 04 05 06 07 08 09 Financial year
Source: Data from AFMA
Figure 4.3 Percentage of hooks set by depth class for each line method
100 Troll
90 Trotline
80 Handline
Dropline manual 70
Dropline hydraulic
60 Dropline 50 % hooks % Bottom longline
40 Auto longline
30
20
10
0 0-100 100-200 200-300 300-400 400-500 500-600 600+ Depth class (m)
Note: Gears may extend across a range of depths in a single set so figures are approximate. Source: Data from AFMA
65 Reducing uncertainty in stock status ABARES
Trap sector Records of substantial trap fishing are for 2004–05 to 2007–08. There was no trap fishing in the 2008–09 season. Catch and effort between 2004–05 and 2007–08 ranged from 48–94 tonnes and 6 500–1 100 trap lifts per season (Figure 4.4). Traps are set across a range of depths (Figure 4.4) and this is reflected in the composition of the catch, which contains significant quantities of shallower, reefal species such as red emperor (Lutjanus sebae; 21.8 per cent of the catch) and trumpet emperor (Lethrinus miniatus; 18.1 per cent) as well as deeper species such as crimson jobfish (Pristipomoides filamentosus; 17 per cent) (Allen 1985; Carpenter & Allen 1989; Froese & Pauly 2010) (Table 4.2).
Figure 4.4 Coral Sea Fishery trap sector catch and effort by financial year, 2000–01 to 2008–09
100 12 000 catch (kg) 90 trap lifts 10 000 80
70
8 000
60
50 6 000
40
Catch (tonnes) Catch 4 000
30 lifts) (trap effort
20 2 000 10
0 0 2000-01 2001-02 2002-03 2003-04 2004-05 2005-06 2006-07 2007-08 2008-09 Financial year
Source: Data from AFMA
66 Reducing uncertainty in stock status ABARES
Figure 4.4 Coral Sea Fishery trap sector catch and effort by depth of set
100 16 000 catch (kg) 90 no. trap lifts 14 000 80 12 000
70
10 000 60
50 8 000
40
6 000 effort (trap lifts) (trap effort
Catch (tonnes) Catch 30 4 000 20 2 000 10
0 0 0-50 50-100 100-150 150-200 200-250 250-300
Depth class (m)
Note: Traps may be strung across a range of depths in a single set so figures are approximate. Source: Data from AFMA
67 Reducing uncertainty in stock status ABARES
Table 4.2 Coral Sea Fishery trap sector annual financial year catch (kg) by species/species group
Common name Species name 2004–05 2005–06 2006–07 2007–08 Total % of total Cum % Red emperor Lutjanus sebae 3 692 38 047 7 469 9 613 58 821 21.8 21.8 Trumpet emperor Lethrinus miniatus 22 348 11 406 8 256 6 628 48 638 18.1 39.9 Crimson jobfish Pristipomoides filamentosus 18 125 14 278 5 752 7 627 45 781 17.0 56.9 Large-eye bream Gymnocranius euanus 5 333 13 670 9 486 12 611 41 100 15.3 72.1 Grass emperor Lethrinus laticaudis 2 325 2 525 1 922 2 029 8 801 3.3 75.4 Snapper spp. Lutjanus spp. 9 20 2 330 5 354 7 713 2.9 78.3 Speckled blue grouper Epinephelus cyanopodus 1 392 5 213 290 492 7 387 2.7 81.0 Spotcheek emperor Lethrinus rubrioperculatus 1 230 2 738 819 2 463 7 249 2.7 83.7 Grouper spp. Aethaloperca and Anyperodon spp. – 1 442 1 560 2 041 5 043 1.9 85.6 Miscellaneous fish Miscellaneous fish 2 970 865 – – 3 835 1.4 87.0 Highfin grouper Epinephelus maculates 1 589 522 409 843 3 363 1.2 88.2 Sharks—other Sharks—other 54 3 134 24 30 3 242 1.2 89.5 Mackerel spp. Family Carangidae 291 1 227 409 991 2 918 1.1 90.5
68 Leopard coral grouper Plectropomus leopardus – 1 236 585 1 041 2 862 1.1 91.6
Two-spot red snapper Lutjanus bohar – 25 2 606 0 2 631 1.0 92.6 Painted sweetlip Diagramma pictum – 1 003 462 943 2 408 0.9 93.5 Mixed reef fish Mixed reef fish 1 034 565 350 6 1 955 0.7 94.2 Whitetip reef shark Triaenodon obesus 0 0 635 949 1 584 0.6 94.8 Comet grouper Epinephelus morrhua 692 241 276 260 1 469 0.5 95.3 White trevally Pseudocaranx dentex 179 382 812 – 1 373 0.5 95.8 Longface emperor Lethrinus olivaceus 462 333 245 65 1 105 0.4 96.2 Spangled emperor Lethrinus nebulosus 2 491 347 184 1 024 0.4 96.6 Large-eye bream Wattsia mossambica 69 33 342 491 935 0.3 97.0 Blacktip grouper Epinephelus fasciatus – – 216 651 867 0.3 97.3 Greater amberjack Seriola dumerili 150 217 95 398 860 0.3 97.6 Goldbanded jobfish Pristipomoides multidens and P. typus 250 352 – 231 833 0.3 97.9 Gobies Plectropomus and Variola spp. 422 330 – – 752 0.3 98.2
Reducing uncertainty in stock status ABARES
Green jobfish Aprion virescens 68 252 103 252 675 0.3 98.5 Bluestripe snapper Lutjanus kasmira 150 114 51 200 515 0.2 98.6 Triggerfish and filefish Balistidae and Monacanthidae 35 0 120 196 351 0.1 98.8 Maori grouper Epinephelus undulatostriatus 47 102 – 154 303 0.1 98.9 Golden-eye jobfish Pristipomoides flavipinnis 95 155 – – 250 0.1 99.0 Note: Taxa are sorted by total catch since 2004.
69
Reducing uncertainty in stock status ABARES
Assessment approaches
This assessment is focused on the key commercial species taken by line and trap methods in the CSF. Assessing each of these species (around 30 taxa comprising 90 per cent of the catch) separately with traditional stock assessment approaches was outside the scope of resources available, both time and data/information requirements. Instead, suites of species are combined and the assessment is undertaken on the assemblage as a whole (multispecies assessment), accounting for the biological and life history characteristics of the taxa within the assemblage (and therefore their robustness to fishing).
Estimated quantities of habitat within the CSF area are used to allow catches to be considered in relation to the size of the habitat that supports those catches. Comparisons can then be made to similar fisheries where more information is available. Potential yields are also considered in the assemblage context.
The approach described is a preliminary or first stage assessment. A decision to proceed to additional modelling (where possible) would depend on consideration of the per-unit-area exploitation rates found in the Coral Sea. If rates are light in comparison to other similar fisheries, further assessment may not be warranted. As a corollary, it should also be possible to recommend trigger levels of catch per-unit-area, above which further assessment or management action is recommended. Some background and rationale behind the methods used is provided. Multispecies assessments Scalefish on tropical deep slope and coral reef habitats have frequently been assessed at the level of multispecies assemblages. To some extent this has resulted from the practicality of dealing with a large number of species, but there has been some theory developed that supports the approach. For reviews of multispecies assessment, particularly in a tropical context see Polovina (1992) and Stephenson et al. (2001).
For this study, species were grouped into a deep scalefish assemblage (120–350 metres), a reef scalefish assemblage (coral reef associated 0–150 metres) and a shark assemblage (all shark species). Scalefish were classified into the deep and reef scalefish assemblages using depth and habitat information from Fishbase (Froese & Pauly 2010). Relatively small catches of several invertebrate taxa were judged to be mislabelled trawl catch and excluded (Table C1). Small catches of several ‘very deep’ taxa (including alfonsino and gemfish) were judged to represent a different assemblage to the balance of the deep species and were excluded (they may also represent mislabelled trawl catch).
Mees and Rosseau (1996) investigated the effects of fishing on multispecies bottomfish fisheries in the Pacific and Indian Oceans. Results suggest, at least for the cases studied, that single species and aggregate single species models can substitute for more complex ecosystem models when applied to multispecies fisheries (Ralston et al. 2004).
The total biomass Schaefer model (Pope 1979) is a production model that has been used to estimate total finfish production maximum sustainable yield (MSY) in a number of multispecies assemblages. These models tend to fit the catch and catch per unit effort (CPUE) data better than if models were fitted to individual species. Ralston and Polovina (1982) examined the performance of the total biomass Schaefer model applied to the deep slope fishery of Hawaii, with various levels of species aggregation (grouping of like species). They found that estimation of management statistics (MSY, yield per unit area) were substantially improved by aggregating
70 Reducing uncertainty in stock status ABARES
rather than analysing the (13) individual species. Estimation was also slightly improved by aggregating associated species from similar depth habitats (30–140 metres, 80–240 metres and 200–350 metres), rather than grouping all species together.
Estimates of yield derived from production models (for example, Ralston & Polovina 1982), fishing surveys (Polovina & Ralston 1986) or depletion methods (Grandcourt 2003) can be associated with the quantity of habitat to estimate yield per unit area. By doing so, these estimates have far better utility for comparing amongst studies and, importantly for the CSF, provide a benchmark for likely yields in new or unexploited fisheries (Polovina 1992). Habitat metrics used for reef fish have typically been linear, such as length of 200 metres isobath or length of reef edge and, occasionally, actual areas expressed in hectares of habitat. Part 1: Habitat assessment
Two methods of describing fish habitat were developed for the CSF—bathymetric contour (isobath) lengths and reef edge lengths—applicable to the different species assemblages.
Isobaths at intervals of 25 metres were derived from the Geoscience Australia 250 metres scale bathymetric grid (Map 4.2). Area estimates were also calculated for 25 metres depth interval bathymetric strata (Appendix C).
The coral reefs of the Coral Sea have been mapped by the Millenium Coral Reef Mapping Project (Andréfouët et al. 2005). Gross reef perimeters were calculated from Millenium Coral Reef Mapping Project data. The reef perimeter was defined as the outer edge of the entire mapped reef, including all reef zones (Map 4.2).
71 Reducing uncertainty in stock status ABARES
Map 4.2 Map of 50 metre interval bathymetric strata adjacent to the Queensland east coast (Cairns to Bundaberg)
Note: Contouring is derived from Geoscience Australia 250 metres scale bathymetric grid. Sources: AFMA; Bureau of Rural Sciences; Geoscience Australia
Thirty individual reef and seamount units were generated (Map 4.3) that encompassed 83 per cent of line operations and 95 per cent of trap operations. A large number of line operations appear to be in very deep waters that would not be accessible to demersal line fishing gear. These appear to be either errors in reporting of position or of gear/fishery designation.
72 Reducing uncertainty in stock status ABARES
Map 4.3 Map of reef outer boundaries extending some 10–20 nautical miles from the reef
Sources: AFMA; Bureau of Rural Sciences; Geoscience Australia Part 2: The deep scalefish assemblage
A variety of studies have examined predatory scalefish biomass/density and yield in the deep slope demersal tropical habitats centred on the 200 metre isobath (Polovina & Shomura 1990; Fry et al. 2006). The deep slope assemblage of the present study contains a similar set of taxa to those studies, particularly Pristipomoides spp., Etelis spp. and Epinephelus spp. Catch of the deep scalefish assemblage increased to a peak of 115 tonnes in 2003 and has since declined to less than 20 tonnes in recent years (Figure 4.5).
73 Reducing uncertainty in stock status ABARES
Figure 4.5 Annual catch of the deep scalefish assemblage in the combined Coral Sea Fishery line and trap sectors by financial year
140
120
100
80
60 Tonnes
40
20
0 1997- 1998- 1999- 2000- 2001- 2002- 2003- 2004- 2005- 2006- 2007- 2008- 98 99 00 01 02 03 04 05 06 07 08 09 Financial year
Source: Data from AFMA
Biomass scenarios Studies have estimated predatory scalefish biomass or density in deeper, demersal tropical habitats similar to the type exploited by the combined line and trap sectors of the CSF (Appendix C). None of these estimates are made for stocks in the CSF. Biomass estimates for the deep scalefish assemblage are presented in Figure 4.6. Following the research of Polovina and Shomura (1990), estimates from around islands have been separated from those of seamounts and isolated oceanic reefs. There is considerable variability in these estimates. Biomasses are lower around islands with a median of 700 kilograms per nautical mile, with most estimates below 1 000 kilograms per nautical mile. This may reflect natural conditions and some historical exploitation. Around seamounts and isolated reefs the estimates are substantially higher, with a median of 2 700 kilograms per nautical mile, with most estimates over 2 000 kilograms per nautical mile. Polovina and Shomura (1990) suggested that, in new and exploratory Pacific fisheries with no surveys, standing stocks of 700 kilograms per nautical mile for reefs and 2 700 kilograms per nautical mile for seamounts could be assumed. In this assessment the fished locations in the CSF would be classed as seamounts (isolated oceanic reefs) (Figure 4.6).
74 Reducing uncertainty in stock status ABARES
Figure 4.6 Biomass estimates of the deep scalefish assemblage on the 200 metre isobath at island and seamounts/oceanic reef locations
10 000
8 000
6 000
4 000 kg/nautical mile kg/nautical 2 000
0 Islands Seamounts
Note: Horizontal bar shows the median. Source: Data from AFMA
Two unfished biomass scenarios were applied to the CSF deep scalefish assemblage with its 1 446 nautical miles of 200 metre isobath (Figure 4.6). The medium scenario was the base case and assumed biomass at the rate of 2 tonnes per nautical mile, resulting in a standing stock estimate of 2 892 tonnes for the deep scalefish assemblage. The low scenario assumed biomass at the rate of 0.7 tonnes per nautical mile of 200 metre isobath, which resulted in a standing stock estimate of 1 012 tonnes. These biomass scenarios were attributed to the deep scalefish assemblage species in proportion to the catch (for example, Pristipomoides filamentosus represents 49.6 per cent of the catch, so under the low scenario P. filamentosus has an unfished biomass of 502 tonnes of the total 1 012 tonnes of unfished biomass of the deep assembly species group). This assumes that catches of all species are in proportion to their relative biomass in the Coral Sea. Potential yields for the deep scalefish assemblage Yield, as a percentage of unfished biomass, has been estimated at between 5 and 30 per cent for the deep scalefish assemblage in the Pacific (Clark 1993). Much of the variation in this range is a result of differing estimation procedures (mainly Gulland 1971; Pauly & Mines 1982) and from refinements over time to species life history traits, specifically better estimation of age/growth and implied natural mortality levels (for example, Fry et al. 2006). The method for estimating yield from multispecies unfished biomass in these analyses is adapted from Ralston and Polovina (1982) and used the Gulland (1971) yield estimator expressed as:
MSY = xMB0 where x is a sustainable exploitation constant, M is natural mortality and B0 is unfished biomass. Three scenarios for the sustainable exploitation constant were examined—0.3, 0.5 and 0.7— with 0.3 being a highly conservative value and 0.5 and 0.7 considered more likely to accurately estimate MSY for the scalefish species involved.
Results of the yield estimation for species within the deep scalefish assemblage are summarised in Table 4.3. Unfished biomass scenarios are provided for each of the deep scalefish assemblage species for the low and medium scenarios. Natural mortalities for each species are based on the Hoenig (1983) method except where indicated.
75 Reducing uncertainty in stock status ABARES
When the most conservative exploitation constant of 0.3 was used the fishery level (all species) yields for the deep scalefish assemblage low biomass and medium biomass scenarios were 82 tonnes and 235 tonnes, respectively (8.1 per cent of unfished biomass in each case) (Table 4.3). When an exploitation constant of 0.5 was used (13.5 per cent of unfished biomass) an overall MSY of 137 tonnes and 391 tonnes was calculated for the low and medium biomass scenarios, respectively. When an exploitation constant of 0.7 was used, an overall MSY of 192 tonnes and 548 tonnes was calculated for the low and medium biomass scenarios, respectively.
76 Reducing uncertainty in stock status ABARES
Table 4.3 Yield scenarios for the Coral Sea Fishery deep scalefish assemblage
Common name Scientific name Biomass Mortality Catch MSY Exploitation Exploitation Exploitation constant = 0.3 constant = 0.5 constant = 0.7 Low Medium (M) Mean Low (t) Medium Low (t) Medium Low Medium scenario scenario annual (t) (t) (t) (t) (0.7 t / (2 t / nm) (t) nm) Rosy snapper Pristipomoides filamentosus 502.4 1 435.3 0.38a 18.8 57.3 163.6 95.5 272.7 133.6 381.8 Ruby snapper Etelis carbunculus 155.3 443.7 0.09a 5.8 4.2 12.0 7.0 20.0 9.8 28.0 Bar rockcod Epinephelus ergastularius and E. septemfasciatus 78.6 224.6 0.14b 2.9 3.3 9.4 5.5 15.7 7.7 22.0 Flame snapper Etelis coruscans 76.3 217.9 0.21a 2.9 4.8 13.7 8.0 22.9 11.2 32.0 Rusty jobfish Aphareus rutilans 49.3 140.9 0.26a 1.8 3.8 11.0 6.4 18.3 9.0 25.6 Goldband snapper Pristipomoides multidens and P. typus 37.5 107.0 0.25a 1.4 2.8 8.0 4.7 13.4 6.6 18.7 Blue-eye trevalla Hyperoglyphe antarctica 26.9 76.8 0.18a 1.0 1.5 4.1 2.4 6.9 3.4 9.7 Mozambique seabream Wattsia mossambica 19.0 54.2 0.38a 0.7 2.2 6.2 3.6 10.3 5.0 14.4 d
77 Ruby snapper Etelis spp. 10.9 31.2 0.23 0.4 0.8 2.2 1.3 3.7 1.8 5.1 c Hapuku Polyprion oxygeneios 10.5 30.1 0.09 0.4 0.3 0.8 0.5 1.4 0.7 1.9 Comet grouper Epinephelus morrhua 8.6 24.6 0.30a 0.3 0.8 2.2 1.3 3.7 1.8 5.2 Tang’s snapper Lipocheilus carnolabrum 3.1 8.9 0.30b 0.1 0.3 0.8 0.5 1.3 0.7 1.9 Saddleback snapper Paracaesio kusakarii 1.3 3.7 0.22a 0.0 0.1 0.2 0.1 0.4 0.2 0.6 Goldeneye snapper Pristipomoides flavipinnis 1.1 3.1 0.38a 0.0 0.1 0.4 0.2 0.6 0.3 0.8 All deep taxa – 1 012.3 2 892.3 0.24 37.9 82.2 234.7 136.9 391.2 191.7 547.7 a Fry et al. 2006. b Fishbase. c Wakefield et al. 2010. d Mean of the two other Etelis spp. MSY Maximum sustainable yield. nm Nautical mile.
Reducing uncertainty in stock status ABARES
Deep scalefish assemblage catch and status Analyses were undertaken for each of the identified reefs and seamounts in Map 4.3, but detailed results are presented only for the Coral Sea as a whole to comply with AFMA data confidentiality requirements.
Mean annual catch for the CSF deep scalefish assemblage for the period 1997–98 to 2008–09 was 37.9 tonnes, equivalent to 26.2 kilograms per nautical mile of 200 metre isobath. Over the same period, the peak of the three-year moving average annual catch was 91.9 tonnes per year (Figure 4.5), equivalent to 63.5 kilograms per nautical mile.
Under the most conservative scenario of low biomass and an exploitation constant of 0.3, the mean annual catch of 37.9 tonnes from the deep scalefish assemblage is well below the MSY of 82 tonnes (Table 4.3). However the peak of the three-year moving average (91.9 tonnes) exceeded the MSY for the low scenario by approximately 10 tonnes during the period of higher catches from 2002 to 2005. Under the remaining five scenarios of biomass and exploitation constant, the estimated yields ranged from 137–548 tonnes (Table 4.3). Mean annual and peak catches seen over the past decade have been well below these levels.
Under an exploitation constant of 0.3, reefs and seamounts where mean annual catch exceeded MSY under the low scenario include Frederick, Kenn, Cato, Fraser seamount, Unknown–west of Diamond and Wreck reefs. No reefs or seamounts had mean annual catch levels that exceeded MSY under the medium scenario. Reefs and seamounts where the peak of the three-year moving average exceeded MSY under the low scenario included Diamond, Frederick, Kenn, Cato, Fraser seamount, Unknown-west of Diamond, Seamount south-east of Kenn and Wreck reefs. With the exception of Diamond, these reefs also had peaks of three-year moving average that exceeded MSY under the medium scenario.
Under an exploitation constant of 0.5, fewer reefs and seamounts had average annual catch (1997–98 to 2008–09) that exceeded MSY estimates. Average annual catches at Cato, Unknown– west of Diamond and Wreck reefs exceeded MSY under the low scenario. These reefs, as well as the Frederick, Kenn, Fraser seamount and the Seamount south-east of Kenn also exceeded the peak of the three-year moving average under the low scenario. Under the medium scenario the peak of the three-year moving average exceeded MSY at Cato and Unknown–west of Diamond reefs.
Under an exploitation constant of 0.7, only Cato Reef had an average annual catch that exceeded MSY under the low scenario. Frederick, Cato, Fraser seamount, Unknown–west of Diamond, Seamount south-east of Kenn and Wreck reefs all had a peak of the three-year moving average that exceeded MSY under the low scenario. Only Cato Reef had a peak of the three-year moving average catch that exceeded MSY under the medium scenario. Concluding remarks This assessment considers catch levels against conservative yield estimates to reconcile stock status for the deep scalefish assemblage. Exploitation levels appear to have been higher on some reefs than others, with catches exceeding the MSY under some scenarios for some reefs. However, the reliability of the MSY estimates at the scale of reefs is questionable, with likely significant between-reef differences in biomass per unit of habitat. Considering the Coral Sea Fishery as a whole, the deep scalefish assemblage would be considered not subject to overfishing and would not be overfished. This conclusion is based on a comparison of the annual average catch for the period 1997–98 to 2008–09 with the most conservative scenario
78 Reducing uncertainty in stock status ABARES
examined. In this instance, the mean annual catch for the deep scalefish assemblage was less than half the estimated yield.
This report makes no specific recommendations on future catches of the deep scalefish assemblage or the individual species taken within it. However, the results presented in Table 4.3 may be used to inform future refinements of harvest strategies within the fishery.
Approximately 20 per cent of deep scalefish assemblage habitat in the CSF is located within the Coringa-Herald and Lihou marine reserves, and not accessible to fishing. All calculations in this study apply to all deep scalefish assemblage habitats in the Coral Sea, including marine reserves. If marine reserves are regarded as outside of the stock then appropriate adjustments may be made to MSY. Part 3: The reef scalefish assemblage
The reef scalefish assemblage is predominantly fishes of the families Lutjanidae, Lethrinidae and Serranidae. Some 57 per cent of the entire catch is made up of the three species: red emperor (Lutjanus sebae), redthroat emperor (Lethrinus miniatus) and paddletail seabream (Gymnocranius euanus). Catch of the reef scalefish assemblage increased to a peak of 93 tonnes in 2005 (two years later than the deep scalefish assemblage peak catch) and has since declined to under 10 tonnes in 2008–09 (Figure 4.7).
Figure 4.7 Annual (financial year) catch of the reef scalefish assemblage in the Coral Sea Fishery Line and Trap Sector
100 90 80 70 60 50 40
30 Tonnes 20 10 0 1997- 1998- 1999- 2000- 2001- 2002- 2003- 2004- 2005- 2006- 2007- 2008- 98 99 00 01 02 03 04 05 06 07 08 09 Financial year
Source: Data from AFMA
Biomass scenarios Visual census A large number of studies use the underwater visual census methods (Halford & Thompson 1996) to estimate abundance of shallow coral reef fishes. Visual census methods applied widely on the Great Barrier Reef (GBR) and in several surveys in the marine reserves of the Coral Sea typically involve a 50 metre transect with a width of 5 metres at depth of 6–9 metres running
79 Reducing uncertainty in stock status ABARES
parallel with the reef edge. This approach has two significant problems for providing per-unit- area estimates of biomass for the reef scalefish assemblage of this study. First, the visual census samples are taken from the very upper bound of the 0–150 metres depth range that defines the reef scalefish assemblage—most of the line and trap fishing is deeper than the top 10 metres. Second, the method provides a relatively poor sample of larger, highly mobile predatory scalefish that are the primary target of the fishery. The samples are likely biased downwards, particularly for species such as Lethrinus miniatus and Gymnocranius euanus, and to some extent for Lutjanus sebae (Cappo & Brown 1996). Visual census is an acceptable method for shallower species such as coral trout (Plectropomus spp.).
On the other hand, visual census estimates may be useful for comparing the abundance of large scalefish predators in Coral Sea with the more intensively studied GBR. This should highlight any major differences that would suggest caution if GBR surveys and studies were to be applied to the Coral Sea. In 2003–04, Oxley et al. (2004) compared the surveyed abundance of coral trout at the Lihou Reef and Coringa-Herald national nature reserves in the Coral Sea to the Cairns and Townsville sections of the GBR (Figure 4.8). They found that relative coral trout abundances were high at Lihou (Plectropomus spp. ~20 per hectare) and moderate at Coringa-Herald (~4 per hectare), compared with the GBR (Cairns ~3 per hectare and Townsville ~10 per hectare). However, Oxley et al. (2004) found that in general there was a low abundance and diversity of other commercially targeted finfish groups (Lutjanidae, Lethrinidae and Serranidae) in the habitats visually sampled.
Figure 4.8 A comparison of the average density of coral trout species per reef from the northern Great Barrier Reef, Coringa-Herald and Lihou Reef national nature reserves
Source: Oxley et al. 2004
In 2007, Ceccarelli et al. (2008) undertook surveys at Coringa-Herald reef and made comparisons to the GBR and the reefs of the north Tasman Sea (the southern-most reefs of the Australian east coast). Again, the visual census samples found Plectropomus laevus in very high abundances compared with the GBR and north Tasman (Figure 4.9). The balance of serranids appears to be in lower abundances at Coringa-Herald reef compared with the GBR (Figure 4.9).
80 Reducing uncertainty in stock status ABARES
Figure 4.9 Density of all serranids and P. laevis across three regions
s.e. Standard error; CHNNR Coringa-Herald National Nature Reserve Source: Ceccarelli et al. 2008
From this we can tentatively conclude that there are reasonably high abundances of coral trout in Coral Sea shallow reef habitats, compared with the GBR. The abundance of non-coral trout Serranidae and the families Lutjanidae and Lethrinidae appear to be lower in the Coral Sea shallow reef habitats. The CSF reef scalefish assemblage species that dominate catch (Lutjanus sebae, Lethrinus miniatus and Gymnocranius euanus) are clearly present in some abundance in the Coral Sea because they comprise more than half the catch but they are not sampled at all in the Coral Sea by the shallow visual census methods. Catch density In the course of the GBR Effects of Line Fishing experiment, reported in part by Mapstone et al. (2004), extensive experimental line fishing was undertaken at reefs with different protection zonings and across multiple years. Catches for all line-caught species combined were expressed as mean annual catch density, with units of kilograms of catch per kilometre of reef edge (Figure 4.10). This experimental fishing should provide a reasonable basis for developing density scenarios because the species selectivity of the gear will be more similar to that of the CSF (though the CSF does fish through a wider depth range). For our purposes the GBR catch densities are regarded as minimum biomass estimates for the reefs. This approach is considered to be conservative, given that reefs in the GBR study yield catches year after year (Figure 4.10) and the main target species have a ‘sustainably fished’ status (DEEDI 2009) with exploitation rates considered to be moderate. Catches in the CSF are low in comparison to these.
Two unfished biomass scenarios were applied to the CSF reef scalefish assemblage with its 1 421 nautical miles of reef perimeter (the habitat metric, Table 4.4). The low scenario assumed biomass at the rate of 0.8 tonnes per nautical mile of reef perimeter (equivalent to 432 kilograms per kilometre in Figure 4.10) and resulted in a standing stock estimate of 1 137 tonnes for the reef scalefish assemblage. The medium scenario assumed biomass at the rate of 1 500 kilograms per nautical mile (equivalent to 810 kilograms per kilometre in Figure 4.10) and resulted in a standing stock estimate of 2 134 tonnes. These reef scalefish assemblage
81 Reducing uncertainty in stock status ABARES
biomass scenarios are slightly less than the deep scalefish assemblage scenarios relating to the 200 metre isobath. The shallower habitats of the reef scalefish assemblage would be expected to have higher productivity, which further suggests conservative assumptions for the reef scalefish assemblage biomass scenarios. Three exploitation constant scenarios (0.3, 0.5 and 0.7) are also presented for the reef scalefish assemblage.
82 Reducing uncertainty in stock status ABARES
Figure 4.10 Mean annual catch density of all demersal species from Effects of Line Fishing Experimental reefs zoned General Use B or Marine National Park B in each region and year since 1990
Note: GU = General Use B, MNP = Marine National Park B. Catch density relates to kilograms fish per kilometre reef perimeter. Data were averaged over reefs within zones having the same status (closed or open) or subject to the same experimental treatment (pulsed) in each year. The effectiveness of attempts to increase fishing are inferred from comparing reefs ‘pulsed’ in 1997 or 1999 with similar but un-manipulated reefs in the same or previous years. Source: Mapstone et al. 2004
83 Reducing uncertainty in stock status ABARES
Potential yields for reef scalefish assemblage Unfished biomass scenarios are provided for each of the reef scalefish assemblage species for the low and medium scenarios (Table 4.4). When the most conservative exploitation constant of 0.3 was used, the fishery level yields for the reef scalefish assemblage were 96 tonnes for the low scenario and 180 tonnes for the medium scenario (8.45 per cent of unfished biomass in each case; Table 4.4). When an exploitation constant of 0.5 was used (14 per cent of unfished biomass) an overall MSY of 160 tonnes and 300 tonnes was calculated for the low and medium scenarios, respectively. When an exploitation constant of 0.7 was used, an overall MSY of 224 tonnes and 420 tonnes was calculated for the low and medium scenarios, respectively.
84 Reducing uncertainty in stock status ABARES
Table 4.4 Yield calculations for the Coral Sea Fishery reef scalefish assemblage based on low and medium biomass scenarios (exploitation constant x=0.3, x=0.5 and x=0.7)
Common name Scientific name Biomass Mortality Catch MSY Exploitation Exploitation Exploitation constant 0.3 constant 0.5 constant 0.7 Low scenario Medium scenario (M) Mean Low Medium Low Medium Low Medium (0.8 t/nm) (1.5 t/nm) annual (t) (t) (t) (t) (t) (t) (t) Red emperor Lutjanus sebae 274.8 515.2 0.14a 6.50 11.6 21.7 19.3 36.2 27.0 50.6 Redthroat emperor Lethrinus miniatus 188.9 354.1 0.43b 4.47 24.4 45.7 40.6 76.1 56.9 106.6 Paddletail seabream Gymnocranius euanus 179.1 335.8 0.34c 4.23 18.3 34.3 30.4 57.1 42.6 79.9 Coral trout Plectropomus and Variola spp. 47.0 88.1 0.25c 1.11 3.5 6.5 5.8 10.9 8.1 15.2 Rockcod Aethaloperca, Anyperodon and 40.9 76.6 0.30d 0.97 3.6 6.8 6.0 11.3 8.5 15.8 Epinephelus spp. Saddletail snapper Lutjanus malabaricus 35.2 66.1 0.30c 0.83 3.1 5.9 5.2 9.8 7.3 13.7 Green jobfish Aprion virescens 34.6 64.9 0.22c 0.82 2.3 4.3 3.9 7.2 5.4 10.1 Grass emperor Lethrinus laticaudis 32.8 61.4 0.37c 0.77 3.6 6.8 6.0 11.3 8.4 15.8
85 Comet grouper Epinephelus morrhua 31.0 58.1 0.44d 0.73 4.1 7.7 6.8 12.8 9.5 17.9 Tropical snapper Lutjanus spp. 28.5 53.4 0.22e 0.67 1.9 3.6 3.2 6.0 4.4 8.3 Amberjack Seriola dumerili 27.3 51.1 0.37c 0.64 3.0 5.6 5.0 9.4 7.0 13.2 Spotcheek emperor Lethrinus rubrioperculatus 25.8 48.5 0.34c 0.61 2.6 4.9 4.4 8.2 6.2 11.5 Sea bream, snapper Gymnocranius spp. 22.8 42.8 0.37f 0.54 2.5 4.7 4.2 7.9 5.9 11.0 Trevallys Carangidae – undifferentiated 20.9 39.1 0.22g 0.49 1.4 2.6 2.3 4.4 3.3 6.1 Common coral trout Plectropomus leopardus 17.9 33.5 0.25c 0.42 1.3 2.5 2.2 4.1 3.1 5.8 Purple rockcod Epinephelus cyanopodus 17.4 32.6 0.15c 0.41 0.8 1.5 1.3 2.5 1.9 3.5 Robinson’s seabream Gymnocranius grandoculis 13.3 24.9 0.37c 0.31 1.5 2.7 2.4 4.6 3.4 6.4 Highfin grouper Epinephelus maculates 12.6 23.6 0.40c 0.30 1.5 2.8 2.5 4.7 3.5 6.6 Samson fish Seriola hippos 11.1 20.8 0.16h 0.26 0.5 1.0 0.9 1.7 1.2 2.3 Fusiliers, seaperches and Caesionidae and Lutjanidae - 10.7 20.1 0.22e 0.25 0.7 1.3 1.2 2.2 1.7 3.1 tropical snappers undifferentiated Red bass Lutjanus bohar 9.3 17.4 0.09i 0.22 0.3 0.5 0.4 0.8 0.6 1.1
Reducing uncertainty in stock status ABARES
Painted sweetlip Diagramma labiosum 8.6 16.0 0.20c 0.20 0.5 1.0 0.9 1.6 1.2 2.3 Bluespotted emperor Lethrinus spp. j 7.5 14.0 0.28k 0.18 0.6 1.2 1.1 2.0 1.5 2.8 Pearl perch Glaucosoma spp. 6.3 11.8 0.19c 0.15 0.4 0.7 0.6 1.1 0.8 1.5 Snapper Pagrus auratus 6.2 11.6 0.11c 0.15 0.2 0.4 0.3 0.7 0.5 0.9 Silver trevally Pseudocaranx dentex 5.9 11.0 0.18c 0.14 0.3 0.6 0.5 1.0 0.7 1.4 Longnose emperor Lethrinus olivaceus 4.6 8.7 0.30c 0.11 0.4 0.8 0.7 1.3 1.0 1.8 Tripletail maori wrasse Cheilinus trilobatus 4.0 7.4 0.28c 0.09 0.3 0.6 0.6 1.0 0.8 1.5 Blacktip rockcod Epinephelus fasciatus 3.4 6.4 0.25c 0.08 0.3 0.5 0.4 0.8 0.6 1.1 Mangrove jack Lutjanus argentimaculatus 2.9 5.4 0.14d 0.07 0.1 0.2 0.2 0.4 0.3 0.5 Highfin amberjack Seriola rivoliana 2.5 4.8 0.20c 0.06 0.2 0.3 0.3 0.5 0.4 0.7 Yellowedge coronation trout Variola louti 2.1 3.9 0.24c 0.05 0.1 0.3 0.2 0.5 0.3 0.6 Bluestriped snapper Lutjanus kasmira 1.9 3.5 0.33c 0.04 0.2 0.3 0.3 0.6 0.4 0.8 All reef taxa na 1 137.4 2 132.6 na 26.89 96.2 180.3 160.3 300.5 224.4 420.7 Note: Natural mortality sources: a Newman & Dunk 2002; b Townsville region average—Williams et al. 2007; c Fishbase maximum age—Hoenig 1983; d Fry et al. 2006; e same as L. sebae; f same as G. grandocolis; g assumed 20 year maximum age—Hoenig 1983; h Rowland 2009; i Marriott et al. 2007; j Carpenter, pers comm; k average of other Lethrinus spp.
86
Reducing uncertainty in stock status ABARES
Reef scalefish assemblage catch and status Analysis was undertaken for each of the identified reefs in Map 4.3, but detailed results are presented only for the Coral Sea as a whole to comply with AFMA confidentiality requirements.
The mean annual catch for the CSF reef scalefish assemblage for the period 1997–98 to 2008–09 was 26.9 tonnes, equivalent to 18.7 kilograms per nautical mile of reef perimeter. Over the same period, the peak of the three-year moving average annual catch was 65 tonnes per year equivalent to 45 kilograms per nautical mile of reef perimeter (Table 4.4).
Under the most conservative scenario of low biomass and an exploitation constant of 0.3, the mean annual catch of 26.9 tonnes and the peak of the three-year moving average (65 tonnes) are below the estimated MSY of 98 tonnes. Under the remaining five scenarios of biomass and exploitation constants, the estimated yields ranged from 160–421 tonnes. Mean annual and peak catches seen over the past decade have been well below these levels.
Catches on individual reefs or seamounts were variable, ranging from essentially no fishing on several reefs to more than 9 tonnes per year average on other reefs (including Diamond Reef). The Coringa, Cato and Wreck reefs were also important in terms of catch, averaging more than 87 2 tonnes per year.
Under an exploitation constant of 0.3, reefs and seamounts where mean annual catch (1997–98 to 2008–09) exceeded MSY under the low scenario include Cato, Fraser seamount and Unknown-west of Diamond reefs. The mean annual catch at Cato and Fraser seamount also exceeded MSY under the medium scenario. Reefs and seamounts where the peak of the three- year moving average annual catch exceeded MSY under the low scenario included Diamond, Kenn, Mellish and Wreck reefs. Diamond reef and Unknown-west of Diamond reef had peaks of three-year moving average catch that exceeded MSY under the medium scenario.
Under an exploitation constant of 0.5, fewer reefs and seamounts had average annual catch that exceeded MSY estimates. Average annual catches at Cato and Fraser seamount exceeded MSY under both low and medium scenarios. These reefs, as well as Unknown-west of Diamond and Diamond, also exceeded the peak of the three-year moving average catch that exceeded MSY under both the low and medium scenarios.
Using the exploitation constant of 0.7, the average annual catch at Cato and Fraser reefs exceeded MSY under both low and medium scenarios. For Diamond and Unknown-west of Diamond reefs, the peak three-year moving average catch levels under the low scenario continue to exceed MSY. Concluding remarks It appears that exploitation levels for the reef scalefish assemblage have been higher on some reefs, with catches exceeding the MSY under some scenarios. However, the reliability of the MSY estimates at the scale of reefs is questionable, with likely significant between-reef differences in biomass per unit of habitat. Considering the Coral Sea Fishery as a whole, the reef scalefish assemblage would be considered not subject to overfishing and would not be overfished. This conclusion is based on a comparison of the annual average catch and the high of the three-year moving average for the period 1997–98 to 2008–09 with the most conservative scenario examined. In this instance, the mean annual catch for the reef scalefish assemblage was less than one-third of the low biomass scenario and less than two-thirds of the high of the three-year moving average at the fishery level.
87 Reducing uncertainty in stock status ABARES
This report makes no specific recommendations for future catches of the reef scalefish assemblage or the species taken within it. However, the results presented in Table 4.3 can inform the future refinement of harvest strategies within the fishery.
Approximately 25 per cent of reef scalefish assemblage habitat in the CSF is located in the Coringa-Herald and Lihou marine reserves and is not accessible to fishing. All calculations in this study apply to all habitats in the Coral Sea, including marine reserves. Part 4: The shark assemblage
The shark assemblage is not well identified, with much of the catch ascribed to groups at the family level or higher. Catches peaked in 2005–06 (Figure 4.11) associated with targeting of sharks with longline methods, a practice which has since ceased. Catches are predominantly blacktip sharks (Carcharhinus spp. with black tipped fins, such as C. amblyrhynchoides, C. melanopterus, C. limbatus and C.tilstoni), tiger shark (Galeocerdo cuvier), whitetip reef shark (Triaenodon obesus), miscellaneous sharks, scalloped hammerhead (Sphyrna lewini) and grey reef shark (Carcharhinus amblyrhynchos) (Figure 4.12). Together these categories make up approximately 90 per cent of the catch by weight for the CSF between 1997–98 and 2008–09.
88
Figure 4.11 Annual catch of the shark assemblage in the Coral Sea Fishery
120
100
80
60 Tonnes
40
20
0 1997- 1998- 1999- 2000- 2001- 2002- 2003- 2004- 2005- 2006- 2007- 2008- 98 99 00 01 02 03 04 05 06 07 08 09 Financial year
Note: Catch is expressed in tonnes (left axis) and as kilograms per nautical mile of 200 metre isobath within the entire Coral Sea Fishery (right axis) Source: Data from AFMA
88 Reducing uncertainty in stock status ABARES
Figure 4.12 Annual catch of shark by taxa in the Coral Sea Fishery
40 Blacktip shark Tiger shark 35 Whitetip reef shark Sharks (other) 30 Scalloped hammerhead Grey reef shark
25 Whaler and weasel sharks (Carcharhinidae, Hemigaleidae)
Sandbar shark 20 Bronze whaler
Tonnes School shark 15 Greeneye dogfish
10
5
0
89
Financial year
Note: Catch is expressed in tonnes (left axis) and as kilograms per nautical mile of 200 metre isobath within the entire Coral Sea Fishery (right axis) Source: Data from AFMA
Shark assemblage exploitation rates and status Biomass scenarios were not developed for the shark assemblage because there was very limited information on shark densities. Potential yields could not be estimated; however, shark assemblage exploitation per unit of habitat was calculated. The 200 metre isobath was used as the habitat metric but the shark assemblage consists of shallow reef species (such as whitetip reef shark) as well as deeper slope (black tip sharks) and mobile oceanic species (hammerhead sharks and tiger shark). Therefore, this may not be an appropriate habitat metric for all sharks.
The Coral Sea Fishery contains some 1 446 nautical miles of 200 metre isobath habitat. For the period 1997–98 to 2008–09 the mean annual catch of all sharks was 21.2 tonnes, which equates to 14.6 kilograms per nautical mile per year. The peak catches of 111 tonnes in 2005–06 equated to a peak of 77 kilograms per nautical mile (Figure 4.11). Catch densities per unit of habitat for shark species are provided in Figure 4.12.
Shark fishing has tended to focus on a few reefs and over a short period of time. The northern reefs of Diamond and Cato and other features in that vicinity have supported most of the catch, with catch densities of 30–40 kilograms per nautical mile (mean annual 1997–98 to 2007–08) and as high as 125 kilograms per nautical mile on one reef during the years 2004 to 2006.
Management measures such as size limits, trip catch limits, reduced total allowable catches and spatial closures have been introduced in some regions. There is currently no quantitative means of evaluating the effectiveness of these management initiatives. In multispecies shark fisheries, where gear selectivity precludes species-specific targeting, management measures should be directed towards the most vulnerable (high risk) species within catches. To some degree, this has been done in the CSF with the catch limits on commonly caught grey reef shark and white- tip reef shark.
89 Reducing uncertainty in stock status ABARES
No status is proposed for sharks in the CSF resulting from this assessment. Conclusions and status The Coral Sea Fishery Line and Trap Sector has been classified as uncertain for fishing mortality and biomass in the ABARES Fishery status reports since 2006. Before this assessment there had been no estimates of sustainable harvest levels with which to compare catch levels. Although not attempting to undertake classical, species level stock assessments, this work has addressed the absence of assemblage level estimates of yield to facilitate status determination.
The deep and reef scalefish assemblages can be considered not subject to overfishing and not overfished up to and including the 2008–09 season. We have been unable to apply the same logic to the shark assemblage, largely because there is no level of sustainable harvest with which to compare historical catch of sharks in the fishery. As a result the shark assemblage remains uncertain with regard to fishing mortality and biomass.
The status of the trawl sector has not been addressed within these analyses. The variety of species and the scarcity of data meant the approach taken for the deep and the reef scalefish assemblages was not feasible.
90
The assessment approach is particularly useful for new or relatively lightly exploited demersal fisheries where a measure of habitat can be derived. For this assessment it was necessary to rely on biomass density estimates from studies outside of the fishery and hence levels of precaution were introduced in the biomass estimation and the MSY calculation. If it can be established that the fishery (or assemblage on this case) is fished within MSY limits in this conservative case, no further assessment may be needed. A more precisely defined sustainable yield for higher levels of exploitation would require a more rigorous assessment process. For the multispecies/habitat method used here, this would require surveyed density estimates from the fishery area that would also allow a fishery independent understanding of species composition (effectively density by species). Alternatively, species-specific approaches would need to be adopted, but noting the data limitations.
90 Reducing uncertainty in stock status ABARES
Appendix C
Table C1 Proportions of species catch in the Coral Sea Fishery
CAAB Standard name Scientific name Group Total catch (kg) Cumulative % 37346032 Rosy snapper Pristipomoides filamentosus deep 225 503 21.61 37346004 Red emperor Lutjanus sebae reef 77 903 29.08 37346014 Ruby snapper Etelis carbunculus deep 69 711 35.76 37018901 Blacktip shark Carcharhinus, Loxodon and Rhizoprionodon spp. shark 69 695 42.44 37018022 Tiger shark Galeocerdo cuvier shark 64 265 48.60 37351009 Redthroat emperor Lethrinus miniatus reef 53 491 53.73 37351022 Paddletail seabream Gymnocranius euanus reef 50 817 58.60 37311910 Bar rockcod Epinephelus ergastularius and Epinephelus septemfasciatus deep 35 281 61.98 37346038 Flame snapper Etelis coruscans deep 34 238 65.26 37018038 Whitetip reef shark Triaenodon obesus shark 32 685 68.40 37990003 Sharks (other) Sharks—other shark 27 813 71.06
91 37019001 Scalloped hammerhead Sphyrna lewini shark 25 584 73.52
37346001 Rusty jobfish Aphareus rutilans deep 22 133 75.64
91 37346901 Goldband snapper Pristipomoides multidens and P. typus deep 16 813 77.25
37445001 Blue-eye trevalla Hyperoglyphe antarctica deep 12 074 78.41 37311905 Coral trout Plectropomus and Variola spp. reef 11 812 79.54 37311901 Rockcod Aethaloperca, Anyperodon and Epinephelus spp. reef 11 598 80.65 37999999 Mixed fish Mixed fish unclear 10 455 81.65 37018000 Whaler and weasel sharks: Carcharhinidae, Carcharhinidae, Hemigaleidae—undifferentiated shark 10 385 82.65 Hemigaleidae—undifferentiated 37018030 Grey reef shark Carcharhinus amblyrhynchos shark 10 072 83.61 37346027 Green jobfish Aprion virescens reef 9 516 84.53 37351006 Grass emperor Lethrinus laticaudis reef 9 299 85.42 37351027 Mozambique seabream Wattsia mossambica deep 8 518 86.23 37346007 Saddletail snapper Lutjanus malabaricus reef 8 432 87.04 37226790 Atlantic cod Gadus morhua reef 8 233 87.83
Reducing uncertainty in stock status ABARES
37346905 Tropical snapper Lutjanus spp. reef 8 087 88.61 37337025 Amberjack Seriola dumerili reef 7 733 89.35 37351012 Spotcheek emperor Lethrinus rubrioperculatus reef 7 334 90.05 37337000 Trevallies and jacks Carangidae—undifferentiated reef 5 908 90.62 37018007 Sandbar shark Carcharhinus plumbeus shark 5 302 91.12 37311078 Common coral trout Plectropomus leopardus reef 5 071 91.61 37311145 Purple rockcod Epinephelus cyanopodus reef 4 940 92.08 37346914 Ruby snapper Etelis spp. deep 4 905 92.55 37311006 Hapuku Polyprion oxygeneios deep 4 735 93.01 28711048 Redspot king prawn Melicertus longistylus invertebrate 4 535 93.44 37351901 Sea bream, snapper. Gymnocranius spp. reef 4 403 93.86 37018001 Bronze whaler Carcharhinus brachyurus shark 4 369 94.28 37311151 Comet grouper Epinephelus morrhua deep 3 860 94.65 28821000 Shovel-nosed/slipper lobsters Scyllaridae—undifferentiated invertebrate 3 840 95.02 37351005 Robinson’s seabream Gymnocranius grandoculis reef 3 762 95.38 23270007 Commercial scallop Pecten fumatus invertebrate 3 615 95.73
92
37311011 Highfin grouper Epinephelus maculates reef 3 571 96.07 37337007 Samson fish Seriola hippos reef 3 080 96.37 37346000 Fusiliers, seaperches and tropical snappers Caesionidae, Lutjanidae—undifferentiated reef 3 037 96.66 37346029 Red bass Lutjanus bohar reef 2 631 96.91 37350003 Painted sweetlip Diagramma labiosum reef 2 429 97.14 37351001 Bluespotted emperor Lethrinus spp. [Carpenter, pers comm] reef 2 048 97.34 37320901 Pearl perch Glaucosoma spp. reef 1 781 97.51 37353001 Snapper Pagrus auratus reef 1 668 97.67 37337062 Silver trevally Pseudocaranx dentex reef 1 667 97.83 37346031 Tang’s snapper Lipocheilus carnolabrum deep 1 396 97.96 37351004 Longnose emperor Lethrinus olivaceus reef 1 309 98.09 37384044 Tripletail maori wrasse Cheilinus trilobatus reef 1 125 98.20 37441024 Wahoo Acanthocybium solandri pelagic 973 98.29 37311014 Blacktip rockcod Epinephelus fasciatus reef 971 98.38
Reducing uncertainty in stock status ABARES
37258002 Alfonsino Beryx splendens very deep 926 98.47 37020007 Greeneye dogfish Squalus mitsukurii shark 875 98.55 37439002 Gemfish Rexea solandri very deep 830 98.63 37346015 Mangrove jack Lutjanus argentimaculatus reef 820 98.71 37337052 Highfin amberjack Seriola rivoliana reef 719 98.78 37311166 Yellowedge coronation trout Variola louti reef 586 98.84 37346060 Saddleback snapper Paracaesio kusakarii deep 581 98.89 37017008 School shark Galeorhinus galeus shark 537 98.94 37346044 Bluestriped snapper Lutjanus kasmira reef 534 99.00 CAAB Codes for Australian Aquatic Biota.
93
Reducing uncertainty in stock status ABARES
Table C2 Coral Sea Fishery bathymetric contour lengths (nm) and depth strata areas (ha)
Depth contour (m) Length Depth class Area (nm) (m) (ha) 0 107 na na 25 592 0–25 43 707 –50 1 594 25–50 228 585 –75 1 137 50–75 819 540 –100 1 147 75–100 62 999 –125 1 158 100–125 46 912 –150 1 172 125–150 38 171 –175 1 185 150–175 35 593 –200 1 205 175–200 35 670 –225 1 219 200–225 41 416 –250 1 245 225–250 49 401 –275 1 280 250–275 69 472 –300 1 307 275–300 94 168 –325 1 305 300–325 150 213 –350 1 286 325–350 118 178 –375 1 311 350–375 121 413 –400 1 330 375–400 130 243 –425 1 339 400–425 134 580 –450 1 369 425–450 161 276 –475 1 391 450–475 168 011 –500 1 419 475–500 168 313 –525 1 495 500–525 195 649 –550 1 602 525–550 243 575 –575 1 596 550–575 289 639 –600 1 593 575–600 286 859 –625 1 547 600–625 317 893 –650 1 534 625–650 272 270 –675 1 530 650–675 236 796 –700 1 529 675–700 233 634 –725 1 556 700–725 240 647 –750 1 586 725–750 235 234 –775 1 593 750–775 245 469 –800 1 676 775–800 287 132 –825 1 678 800–825 348 278 –850 1 678 825–850 321 908 –875 1 759 850–875 369 909 –900 1 842 875–900 450 869 –925 1 989 900–925 489 356 –950 1 981 925–950 506 986 –975 2 361 950–975 580 413 –1 000 2 634 975–1 000 947 200 Note: Contouring is derived from Geoscience Australia 250 metre scale bathymetric grid (note that the 200 metre isobath length was subsequently refined for some reefs using RAN nautical charts).
94 Reducing uncertainty in stock status ABARES
Table C3 Previous studies on species deep scalefish assemblage biomass within linear habitat
Location/description Main taxa Estimates Marianas, Pacific Ocean. Pristipomoides spp. (4) kg/nm of 200 m contour (125–275 m) Hook and line surveys of “deep slope” habitats between 125 and Etilis spp. (2) All species - Northern banks and islands = 717 kg/nm 275 m. Caranx sp. (1) All species - Southern banks and islands = 441 kg/nm Biomass estimates derived from previously estimated gear Other (numerous) All species - Western seamounts = 8 989 kg/nm catchability for the 200 m depth contour. Combined Commercial = 748 kg/nm Various yields estimated using age at first capture, YPR, F and F0.1. Caranx lugubris = 62.1 (Polovina & Ralston 1986) Pristipomoides filamentosus = 33.1 P. auricilla = 73.2 P. flavipinnis = 41.3 P. zonatus = 273.8 Etilis coruscans = 76.3 E. carbunculus = 60.8 Others = 118.1 Polovina and Shomura (1990) summarised results from a range of Pristipomoides filamentosus kg/nm of 200 m contour depletion experiments and associated estimates of unfished P. flavipinnis Four Tongan seamounts = 2 700, 2 700, 2 400, 8 500 kg/nm
95 biomass densities; Tonga Etelis coruscans
E. carbunculus Polovina and Shomura (1990), as above; Fiji E. coruscans Five Fijiian seamounts = 2 500, 3 800, 1 500, 1 200, 7 000 kg/nm E. carbunculus Four Fijiian islands = 700, 700, 600, 500 kg/nm Polovina and Shomura (1990), as above; Vanuatu E. carbunculus Two Vanuatu islands = 2 300, 700 kg/nm E. coruscans E. radiosus Lutjanus malabaricus Polovina and Shomura (1990), as above; Papua New Guinea P. multidens One PNG seamount (Kavieng) = 3 331 kg/nm P. filamentosus One PNG island (Schoutten ) = 191 kg/nm E. carbunculus Wattsia mossambica L. malabaricus Paracaesio stonei
Reducing uncertainty in stock status ABARES
Location/description Main taxa Estimates Saya de Malha Bank, Southwest Indian Ocean. Pristipomoides filamentosus Density in 55–130m = 2 364 kg/km2 (95 per cent CI = 2 044–2 976) Depletion (Leslie) model from a 13-day intensive fishing period. Density on 100 m isobath = 2 233 kg/nm (95 per cent CI = 1 929–2 811) MSY estimated at 24 per cent of Bo. (Grandcourt 2003) Lihir Island group, Papua New Guinea. Mainly commercial species such as kg per nm of 200 m contour Dropline surveys between 125–350 m depth. Etelis, Lutjanus and Pristipomoides. Total = 395 kg/nm Biomass estimated by using catchability coefficient from Polovina Caranx lugubris = 15 kg/nm and Ralston (1986). Caranx tille = 0.5 kg/nm Yield calculated using three different methods. Wattsia mossambica = 13 kg/nm (Fry et al. 2006) Aphareus rutilans = 12 kg/nm Etelis carbunculus = 105 kg/nm Etelis coruscans = 26 kg/nm Lutjanus timorensis = 6 kg/nm Paracaesio kusakarii = 26 kg/nm Paracaesio stonei = 19 kg/nm Pristipomoides filamentosus = 13 kg/nm Pristipomoides flavipinnis = 7 kg/nm
96 Pristipomoides multidens = 16 kg/nm
Pristipomoides zonatus = 9 kg/nm Cephalopholis sexmaculata = 0.2 kg/nm Epinephelus morrhua = 12 kg/nm Variola albimarginata = 0.4 kg/nm ‘Others’ = 115 kg/nm Main Hawaiin Islands, Pacific Ocean. ‘Deep 7’ kg/nm of 200 m contour Stock assessment of Hawaiian bottomfish (including Deep 7 Pristipomoides filamentosus Main Hawaiian islands = 1 115 kg/nm species) using data to 2007. Etelis coruscans Historical biomass estimated from standardised catch and CPUE Etelis carbunculus using a Bayesian production model. Epinephelus quernus B0 for the main Hawaiian island is estimated at some Pristipomoides zonatus 3 500 000 pounds (1 587 t) of which the Deep 7 comprise 2 510 000 Pristipomoides seiboldii pounds (1 143 t). There is some 1 025 nm of 200 m isobath habitat. Aphareus rutilans (Brodziak et al. 2009)
Reducing uncertainty in stock status ABARES
References
Allen, GR 1985, ‘Pristipomoides multidens in Jordan Goldbanded Jobfish’,
Andréfouët, S, Muller-Karger, FE, Robinson, JA, Kranenburg, CJ, Torres-Pulliza, D, Spraggins, SA & Murch, B 2005, ‘Global assessment of modern coral reef extent and diversity for regional science and management applications: a view from space’ in Suzuki, Y, Nakamori, T, Hidaka, M, Kayanne, H, Casareto, BE, Nadaoka, K, Yamano, H, Tsuchiya, M, & Yamazato, K, (eds.) 10th International Coral Reef Symposium, Japanese Coral Reef Society, Okinawa, Japan. CDROM. pp. 1732–45.
Bellwood, DR & Hughes, TP 2001, Regional scale assembly rules and biodiversity of coral reefs, Centre for coral reef biodiversity, Townsville, Queensland.
Brodziak, J, Moffitt, R & DiNardo, G 2009, ‘Hawaiian bottomfish assessment update for 2008’, Pacific Islands Fish. Sci. Cent., Natl. Mar. Fish. Serv., NOAA, Honolulu, HI 96822-2396. Pacific Islands Fish. Sci. Cent. Admin. Rep. H-09- 02.
Carpenter, KE & Allen, GR 1989, FAO species catalogue, Vol. 9. Emperor fishes and large-eyed breams of the world, Food and Agricultural Organization of the United Nations, Rome.
Cappo, M & Brown, IW 1996, Evaluation of sampling methods for reef fish populations of commercial and recreational interest, CRC Reef Research Centre technical report no. 6 Townsville.
Ceccarelli, D, Choat, JH, Ayling, AM, Richards, Z, van Herwerden, L, Ayling, A, Ewels, G, Hobbs, JP & Cuff, B 2008, Coringa-Herald National Nature Reserve Marine Survey 2007, report to the Department of the Environment, Water, Heritage and the Arts by C&R Consulting and James Cook University.
Clark, WG 1993, The effect of recruitment variability on the choice of a target level of spawning biomass per recruit, International Pacific Halibut Commission, Seattle, Washington.
DEEDI 2009, Plan for the Assessment of Queensland East Coast Shark Resources 2009–14, Department of Employment, Economic Development and Innovation, Brisbane.
Froese, R & Pauly, D 2010, FishBase, electronic publication, www.fishbase.org, version (07/2010).
Fry, GC, Brewer, DT & Venables, WN 2006 ‘Vulnerability of deepwater demersal fishes to commercial fishing; evidence from a study around a tropical volcanic seamount in Papua New Guinea’, Fisheries Research, vol 81, pp. 126–41.
Grandcourt, EM 2003, ‘The effect of intensive line fishing on the virgin biomass of a tropical deepwater snapper, the crimson jobfish (Pristipomoides filamentosus)’, Fishery Bulletin, vol. 101, pp. 305–11.
Gulland, JA 1971, ‘Science and fishery management’, ICES Journal of Marine Science, vol. 33, no. 3, pp. 471–77.
Halford, A & Thompson, AA 1996, Visual Census Surveys of Reef Fish, Standard Operational Procedure 3, AIMS Townsville.
97 Reducing uncertainty in stock status ABARES
Hoenig, JM 1983, ‘Empirical use of longevity data to estimate mortality rates’, US National Marine Fisheries Service Bulletin, vol. 81, pp. 898–903.
Mapstone, BD, Davies, CR, Little, LR, Punt, AE, Smith, ADM, Pantus, F, Lou, DC, Williams, AJ, Jones, A, Ayling, AM, Russ, GR & McDonald, AD 2004, The effects of line fishing on the Great Barrier Reef and evaluations of alternative potential management strategies, CRC Reef Research Centre, technical report no. 52, CRC Reef Research Centre, Townsville, Australia.
Marriott, RJ, Mapstone, BD & Begg, GA 2007, ‘Age-specific demographic parameters, and their implications for management of the red bass, Lutjanus bohar (Forsskal 1775): A large, long-lived reef fish’, Fisheries Research, vol. 83, pp.204–15.
Mees, CC & Rousseau, J 1996, Management of multi-species tropical fisheries, final report to ODA. FMSP Project R5484. MRAG, 193 pp.
Newman, SJ & Dunk, IJ 2002, ‘Growth, Age Validation, Mortality, and other Population Characteristics of the Red Emperor Snapper, Lutjanus sebae (Cuvier, 1828), off the Kimberley Coast of North-Western Australia Estuarine’, Coastal and Shelf Science, vol. 55, issue 1, July 2002, pp. 67–80.
Oxley, WG, Emslie, M, Muir, P & Thompson, A 2004, Marine surveys undertaken in the Lihou Reef Naitonal Nature Reserve, Australian Institute of Marine Science, produced for the Department of Environment and Heritage.
Pauly, D & Mines, AN 1982, Small scale fisheries of San Miguel Bay Phillipines: biology and stock assessment, ICLARM technical report.
Polovina, JJ & Ralston, S 1986, ‘An approach to yield assessment for unexploited resources with application to the deep slope fishes of the Marianas’, Fishery Bulletin, vol. 84, no. 4.
Polovina, JJ & Shomura RS 1990, ‘United States Agency for International Development and National Marine Fisheries Service Workshop on Tropical Fish Stock Assessment’, 5–26 July 1989, Honolulu, Hawaii. U.S. Dep. Commer., NOAA Tech. Memo. NOAA-TM-NMFS-SWFSC-148.
Polovina, JJ 1992, Modelling fish stocks: applicability, problems and requirements for multispecies and multigear fisheries in the tropics, National Marine Fisheries Service, Hawaii.
Pope, J 1979, Stock assessment in multispecies fisheries, with special reference to the trawl fishery in the Gulf of Thailand, South China Sea Fisheries Development Program, FAO/UNDP South China Sea Fisheries Development and Coordination Programme, SCS/DEV/79/19.
Ralston, S & Polovina, LJ 1982, ‘A multispecies analysis of the commercial deep-sea handline fishery in Hawaii’, Marine Fishery Bulletin, vol. 80, no. 3.
Ralston, S, Cox, S, Labelle, M & Mees, C 2004, Bottomfish Stock Assessment Workshop: Final Panel Report, January 13–16 2004, Western Pacific Fishery Management Council, Hawaii.
Rowland, AJ 2009, ‘The biology of Samson fish Seriola hippos with emphasis on the sportfishery in Western Australia’, Doctor of Philosophy Thesis, Murdoch University, Perth.
Stephenson, PC, Edmonds, JS, Moran, MJ & Caputi, N 2001, ‘Analysis of stable isotope ratios to investigate stock structure of red emperor and Rankin cod in northern Western Australia’, Journal of Fish Biology, vol. 58, pp. 126–144.
98 Reducing uncertainty in stock status ABARES
Wakefield, CB, Newman, SJ & Molony, BW 2010, ‘Age-based demography and reproduction of hapuku, Polyprion oxygeneios, from the south coast of Western Australia: implications for management’, ICES Journal of Marine Science, vol. 67, pp. 1164–74.
Williams, AJ, Mapstone, BD & Davies, CR 2007, ‘Spatial patterns in cohort specific mortality of red throat emperor, Lethrinus miniatus, on the Great Barrier Reef’, Fisheries Research, vol. 84, pp. 328–37.
Woodhams, J, Chambers, M & Pham, T 2010 ‘Coral Sea Fishery’, in DT Wilson, R Curtotti, & GA Begg (eds) Fishery status reports 2009: status of fish stocks and fisheries managed by the Austrlaian Government, Australian Bureau of Agricultural and Resource Economics – Bureau of Rural Sciences, Canberra.
99 Reducing uncertainty in stock status ABARES 5 Status determination for trochus and tropical rock lobster stocks in the Coral Sea Fishery Hand Collection Sector
Mark Chambers Summary
This brief study assesses the status of the tropical rock lobster and trochus stocks in the Coral Sea. The ABARES 2008 Fishery status reports classified the stocks as uncertain if overfished but not subject to overfishing.
This study considered the status of the tropical rock lobster under two alternative scenarios. The first scenario assumed that the Coral Sea population are part of a larger population with spawning grounds in the Torres Strait. The second scenario assumed that the Coral Sea population are a self-recruiting, self-sustaining population.
If the first scenario is true, historical catch in the Coral Sea was found to be negligible compared with catch in the Torres Strait Tropical Rock Lobster Fishery (TSTRLF). In this case, the current (2008) not overfished and not overfishing status of the TSTRLF implies that the combined Torres Strait–Coral Sea population is also not overfished and is not subject to overfishing.
If the Coral Sea tropical rock lobster population is assumed to be self-recruiting, historical effort has not been sufficient to cause substantial fishing mortality. Even under very precautionary criteria, a Coral Sea population of less than 16 000 individuals would be sufficient to sustain the highest recorded annual catch given productivity estimates for tropical rock lobster. Estimates of suitable Coral Sea habitat combined with observed catch rates would suggest that a self- recruiting population would be much larger than 16 000. Results show that under the self- recruiting scenario, the Coral Sea Rock Lobster Fishery is not overfished and not subject to overfishing.
The history of commercial exploitation of trochus in the Coral Sea Hand Collection Fishery is limited to a single trip to a single reef in 2001. This study concludes that, because of negligible exploitation, trochus in the Coral Sea is not overfished and not subject to overfishing. Introduction
The Lobster and Trochus Sector is part of the Coral Sea Fishery (CSF), which extends from Cape York to Sandy Cape in Queensland (Map 5.1). The Lobster and Trochus Sector of the CSF allows lobster and trochus to be collected by hand, with or without underwater breathing apparatus.
Historical catch records suggest that at least two species of tropical rock lobster have been taken. Ornate rock lobster (Panulirus ornatus) has accounted for most of the total catch and a smaller quantity of painted rock lobster (P. versicolour) has also been recorded. There is some doubt over which species of trochus is found in the Coral Sea. Two possibilities are Trochus niloticus and Techtus pyramus (Wilson et al. 2010).
100 Reducing uncertainty in stock status ABARES
Map 5.1 Map of the Coral Sea Fishery
X
Current management methods in the CSF involve catch triggers, spatial closures, move-on provisions and size limits.
Before this assessment, the ABARES 2008 Fishery status reports (Wilson et al 2009) classified the CSF lobster and trochus stock as uncertain if overfished but not subject to overfishing. The uncertain status determination was based on the lack of reliable mechanism for comparing the current biomass with an estimate of the biomass in an unfished state. The lobster stock was classified as not subject to overfishing as a result of no catch in 2007–08. Similarly, there has been no reported catch of trochus since 2001 (Wilson et al 2009). Approach to status determination The aim of this brief study is to assess the status of the tropical rock lobster and trochus stocks in the Coral Sea. Ideally an assessment of the lobster stock would be based on regular scientific surveys of lobster abundance. However, the gross value of Coral Sea tropical rock lobster production does not warrant the expense associated with scientific surveys so no such data are available. Additionally, no biological data are available specific to the Coral Sea stock of tropical rock lobster. In the absence of these data, a minimum population size that would be required to sustain the maximum annual catch of tropical rock lobster in the Coral Sea is estimated and compared with what might be expected to be a plausible minimum population for a self- sustaining tropical rock lobster population in the Coral Sea. Similarly, limited data are available for trochus. In the absence of these data, stock status are estimated based on known natural mortality and logbooks records from 2001.
The study is divided into two parts. In the first part, the status of the tropical rock lobster is considered under two alternative scenarios. The first scenario assumes the tropical lobster stock
101 Reducing uncertainty in stock status ABARES in the CSF is part of a larger population with spawning grounds in the Torres Strait. The second scenario assumes the Coral Sea stock is a self-recruiting and self-sustaining population. The second part deals with the historical catch and recovery potential of trochus in the Coral Sea. Tropical rock lobster
Catch and effort records from the Hand Collection Sector in the Coral Sea suggest that at least two species have been taken. Ornate rock lobster (Panulirus ornatus) has accounted for most of the total catch, with a smaller quantity of painted rock lobster (P. versicolour) recorded.
A number of authors (for example, Dennis et al. 2001; MacFarlane & Moore 1986) have suggested that most ornate rock lobster in the Coral Sea arrive from larval dispersal rather than migration. This has led to a commonly held view that ornate rock lobster in the Coral Sea can be regarded as a sink population within a larger stock that includes the Torres Strait. Whether ornate rock lobster in the Coral Sea are a sink population within the larger stock is a key question for management to consider, although it does not affect current status.
Total catch over the history of the fishery has been less than 10 tonnes and the highest annual catch was recorded in 2003 (2.5 tonnes; Figure 5.1).
Figure 5.1 Total annual catch and effort for tropical lobster in the Coral Sea Hand Collection Fishery
Total Catch Total Effort
2500 500
2000 400
1500 300
Landings (kg) Landings Effort (dive hours) (dive Effort
1000 200
500 100
0 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 Source: Data from AFMA
Using geomorphological data provided by the Millennium Coral Reef Mapping Project (Andréfouët et al. 2005) shallow reef area in the Coral Sea is estimated to be greater than 1 million hectares (this estimate excludes Mellish Reef and Cato Island, which are not covered by the Millennium Reef Project).
Observed catch rates have varied from around 2 to 12 kilograms per dive hour (Figure 5.2). These catch rates are consistent with values for the Torres Strait reported in Chittleborough (1974), suggesting that densities in areas that have been fished in the Coral Sea are comparable with the areas surveyed in the Torres Strait. Ye and Dennis (2009) derived a standardised Tropical Rock Lobster Fishery CPUE timeseries for the Torres Strait Fishery taking fishing
102 Reducing uncertainty in stock status ABARES power into account. They found this standardised CPUE timeseries to be roughly proportional to survey estimated densities.
Figure 5.2 Time series of unstandardised catch per unit effort for tropical rock lobster in the Coral Sea
14 Lobster CPUE
12
10
8
6 kg per dive hour dive per kg
4
2
0
2000 2001 2002 2003 2004 2005 2006 2007 2008
Source: Data from AFMA
Calculation of population size to sustain maximum catch Numerical values need to be assumed for some characteristics of the lobster population. ABARES assume values for natural mortality and target fishing mortality from the nearby Torres Strait rock lobster fishery, which harvests the same species. Plagányi et al. (2010) recommended a target fishing mortality (FMSY) of 0.15 year-1. Natural mortality was estimated to be 0.68 year-1, and between 0.53 year-1 and 0.82 year-1 with 90 per cent confidence (Plagányi et al. 2010).
Given this information, a basic population model can be used to estimate the population size at the time of recruitment that would support the maximum observed catch for the target fishing mortality.
For a lobster population of size P0 just after an assumed annual recruitment pulse, the population size at time t throughout the year (t measured in years), where M is natural mortality and F is fishing mortality, is given by:
M F t Pt P0 e
The average population size across the year is then:
1 M F t P0 (M F ) P P0 e dt e 1 t0 M F
The total catch is equal to the average population size throughout the year multiplied by the fishing mortality.
103 Reducing uncertainty in stock status ABARES
FP0 (M F ) FP e 1 M F
If we assume that the reduction in population size because of fishing mortality is equal to the number of individuals taken (reasonable for a hand collection fishery), we can write:
FP0 (M F ) FP0 (M F ) Catch e 1 1 e M F M F
The minimum initial population can be expressed in terms of annual catch, fishing mortality and natural mortality:
* CM F P 0 F 1 e(M F )
According to logbook records from the Coral Sea Fishery Hand Collection Sector the maximum number of lobsters taken in a calendar year was 1 285 in 2003. If we estimate fishing mortality at FMSY = 0.15 year-1we can calculate a lobster population that would support the historic maximum catch, P*MSY 1285 0.68 0.15 P* MSY 0.151 e0.680.15
A few points should be mentioned about the calculated minimum population size. The calculation relies upon parameter estimates of M and FMSY from a population in Torres Strait that is assumed to be distinct from the Coral Sea population. The assumed fishing mortality is applicable at the biomass associated with maximum sustainable yield rather than the unfished biomass. The low level of tropical lobster catch in the Coral Sea before 2003 (Figure 5.2) suggests that the Coral Sea population was likely to have been in an approximately unfished state in 2003. Plagányi et al. (2010) estimate that maximum sustainable yield for the Torres Strait tropical rock lobster stock occurs at a spawning stock biomass of roughly 80 per cent of unfished levels. Our simple calculation makes no distinction between the exploitable stock biomass (in terms of size) and total biomass and applies fishing mortality to the entire population for the entire year, so the figure is probably best considered as the population vulnerable to the fishery.
Based on the simple calculation, the virgin biomass required to support the historic maximum catch is around 12 700 individuals or 27 tonnes, which is the number of individuals multiplied by the mean weight of captured lobsters (2.15 kilograms) in the Coral Sea. If it was assumed that the figure of 12 700 individuals refers to the population size at maximum sustainable yield, the required unfished population vulnerable to the fishery would be 15 900 individuals or approximately 34 tonnes. Alternatively, given that the average catch in the Coral Sea has been considerably less than 1 285 individuals (2.7 tonnes) in most years, a one-off catch of this size might be treated as contributing to a fish down toward BMSY.
Two scenarios were considered to assess the status of tropical lobster in the Coral Sea. The first scenario assumed that the Coral Sea population are part of a larger population with spawning grounds in the Torres Strait. The second scenario assumed that the Coral Sea population are a self-recruiting, self-sustaining population. A flowchart diagram of the Coral Sea lobster status
104 Reducing uncertainty in stock status ABARES determination logic is given in Figure 5.3 for overfished status and Figure 5.4 for overfishing status.
Figure 5.3 Flowchart for tropical lobster overfished (biomass) status determination.
Source: ABARES
Figure 5.4 Flowchart for tropical lobster overfishing (fishing mortality) status determination
Source: ABARES
Under the assumption that the Coral Sea population is part of a larger population with spawning grounds in the Torres Strait, historical catch in the Coral Sea was found to be negligible compared with catch in the Torres Strait Rock Lobster Fishery. In this case, the current (2008) not overfished and not overfishing status of the Torres Strait Rock Lobster Fishery would also apply to the Coral Sea sub-population.
105 Reducing uncertainty in stock status ABARES
If the Coral Sea tropical rock lobster population is assumed to be self-recruiting, historical fishing has not been sufficient to cause substantial fishing mortality. Even under very precautionary criteria, a Coral Sea population of less than 16 000 individuals would be sufficient to sustain the highest recorded annual catch given productivity estimates for tropical rock lobster. Estimates of suitable Coral Sea habitat combined with observed catch rates would suggest that a self-recruiting population would be much larger than 16 000 individuals. The conclusion under the self-recruiting scenario is that the Coral Sea Rock Lobster Fishery is not overfished and not subject to overfishing. Trochus
Logbooks record that trochus catch has been limited to 160 kilograms from a single reef in 2001.
For a stock to be considered overfished their biomass should fall below 20 per cent (B20) of the unfished abundance (biomass). For trochus to have been overfished in 2001, a Coral Sea trochus unfished biomass of less than 200 kilograms would be required. Trochus are a somewhat cryptic species (T Skewes, pers. comm., 2010). It is not plausible that a single trip in 2001 to one reef would have yielded more than 80 per cent of the trochus biomass in the Coral Sea.
For trochus to be classified as overfished in 2009, more than 80 per cent of unfished Coral Sea trochus biomass would have had to be taken in the single trip in 2001 and the standing stock biomass would need to be yet to recover to 40 kilograms (the equivalent of B20 if B0 was 200 tonnes). This is also not plausible. As such, the trochus stock in the Coral Sea is not overfished. There has been no recorded trochus catch from the Coral Sea since 2001, so trochus is not subject to overfishing. References
Andréfouët, S, Muller-Karger, FE, Robinson, JA, Kranenburg, CJ, Torres-Pulliza, D, Spraggins, SA & Murch, B 2005, ‘Global assessment of modern coral reef extent and diversity for regional science and management applications: a view from space’, in Y Suzuki, T Nakamori, M Hidaka, H Kayanne, BE Casareto, K Nadaoka, H Yamano, M Tsuchiya, & K Yamazato (ed.), 10th International Coral Reef Symposium, Japanese Coral Reef Society, Okinawa, Japan. CDROM, pp. 1732–45.
Chittleborough, RG 1974, The tropical rock lobster Panulirus ornatus (Fabr.) as a resource in Torres Strait, CSIRO (Division of Fisheries and Oceanography), No. 58.
Dennis, DM, Pitcher, CR & Skewes, TD 2001, ‘Distribution and transport pathways of Panulirus ornatus (Fabricius, 1776) and Panulirus spp. larvae in the Coral Sea, Australia’, Marine & Freshwater Research, vol. 52, pp. 1175–85.
MacFarlane, JW & Moore, R 1986, ‘Reproduction of the Ornate Rock Lobster, Panulirus ornatus (Fabricius), in Papua New Guinea’, Australian Journal of Marine and Freshwater Research, vol. 37, pp. 55–65.
Plagányi, E, Dennis, D & Kienzle, M 2010, Updated 2010 Assessment of the Tropical Rock Lobster (Panulirus ornatus) Fishery in the Torres Straits, report incorporating comments received by TRL Resource Assessment Group, October 2010, CSIRO, Cleveland, Queensland.
Wilson, DT, Curtotti, R, Begg, GA & Phillips, K (eds) 2009, Fishery status reports 2008: status of fish stocks and fisheries managed by the Australian Government, Bureau of Rural Sciences and Australian Bureau of Agriculture and Resource Economics, Canberra.
106 Reducing uncertainty in stock status ABARES
Wilson, DT, Curtotti, R & Begg, GA (eds) 2010, Fishery status reports 2009: status of fish stocks and fisheries managed by the Australian Government, Australian Bureau of Agriculture and Resource Economics – Bureau of Rural Sciences, Canberra.
Ye, Y & Dennis, D 2009, ‘How reliable are the abundance indices derived from commercial catch- effort standardization?’, Canadian Journal of Fisheries and Aquatic Science, vol. 66, pp. 1169–78.
107 Reducing uncertainty in stock status ABARES 6 Status determination for the deepwater prawn stock in the North West Slope Trawl Fishery
Mark Chambers Summary
The deepwater prawn stock in the North West Slope Trawl Fishery (NWSTF) comprises chiefly four species of penaeid and two species of carid prawn. For management purposes, deepwater prawns in the NWSTF are considered a single stock. In 2008, the ABARES 2008 Fishery status reports (the most recent report at the start of this study) classified the stock as uncertain for both overfished and overfishing status (Wilson et al. 2009). In recent years, there appears to have been little or no effort targeted at prawns in the NWSTF. Recent catch and effort data are believed to be incompatible with the period when prawns were targeted. This rules out methods of stock assessment that rely on catch per unit effort (CPUE) being used as an index of relative abundance throughout the entire time series in the conventional manner.
To provide an assessment of this stock, ABARES analysed logbook data from the years 1986 to 1993, when the prawn stock was actively targeted. A delay difference model implemented within a Bayesian framework was developed, which predicted the current condition of the stock. Life history information for Aristaeomorpha foliacea (the most commercially important species) was used to inform prior distributions together with logbook records of prawn landings up to 2008. The sensitivity of the model to uncertainty in natural mortality is examined. The possible effect of hyperstability in the index of abundance is also explored by assuming a power law relationship between abundance and CPUE (see for example Harley et al. 2000). The scenarios explored suggest that the current stock biomass is unlikely to be below 20 per cent of its average unfished level and is therefore not overfished. The catch and effort data were not sufficiently informative for the carrying capacity or absolute biomass to be estimated with reasonable precision. Recent low levels of catch are unlikely to constitute overfishing. Introduction
The NSWTF operates off north-western Australia (Map 6.1). The fishery currently targets scampi and deepwater prawns are a by-product of the fishery. The deepwater prawn stock in the NWSTF comprises chiefly four species of penaeid and two species of carid prawn. For management purposes, the fishery’s deepwater prawns are considered a single stock.
The NWSTF was established in 1985 after exploratory trawls suggested the existence of a potentially commercial scampi stock in the area (Evans 1992). After initially targeting scampi, fishers found they were able to locate aggregations of giant red prawns (Aristaeomorpha foliacea) using echo sounding equipment (B Wallner, pers. comm., 2010). The prawn fleet reached its peak size in 1987 (Evans 1992), after which catch rates fell sharply. Since 1994, effort in the NWSTF has been predominantly from a small number of vessels based in Western Australia.
108 Reducing uncertainty in stock status ABARES
Map 6.1 North West Slope Trawl Fishery and recent relative fishing intensity
Source: ABARES
The fishery is managed through a limited number of permits (seven) and gear restrictions (codend mesh size ≤ 50 mm).
The ABARES 2008 Fishery status reports (the most recent report at the start of this study) classified the stock as uncertain for both overfished and overfishing status (Wilson et al. 2009). The sector had not been previously assessed. Approach to status determination This study aims to estimate the NWSTF deepwater prawn unfished biomass and assess the status of the stock. Logbook data were analysed from 1986 to 1993, when the prawn stock was actively targeted. A delay difference model implemented within a Bayesian framework was developed to predict the current condition of the stock. Life history information for Aristaeomorpha foliacea (the most commercially important species) was used to inform prior distributions together with logbook records of prawn landings up to 2008. The sensitivity of the model to uncertainty in natural mortality was analysed. The possible effect of hyperstability in the index of abundance was also explored by assuming a power law relationship between abundance and CPUE (see, for example, Harley et al. 2000). The assessment is structured into sections on background information, analysis and discussion.
109 Reducing uncertainty in stock status ABARES
Background information Catch and effort The NWSTF was established in 1985 (Evans 1992) and by 1987 the prawn fleet had reached its peak size of 21 vessels (Evans 1992) (Figure 6.1). The total prawn catch reached an historical high of 792 tonnes in the same calendar year (Figure 6.2). After 1987 catch rates fell sharply and total prawn catch in the NWSTF had declined to less than 67 tonnes by 1993. The number of fishers operating in the fishery greatly reduced and those that remained reverted to targeting the higher priced scampi species. Since 1994, effort in the NWSTF has been predominantly from a small number of vessels based in Western Australia.
Figure 6.1 Number of boats fishing in the North West Slope Trawl Fishery by financial year
25
20
15
10 Vesselcount
5
0
86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11
– – – – – – – – – – – – – – – – – – – – – – – – – –
1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 1985
Source: Data from AFMA
Figure 6.2 Total combined reported prawn and combined scampi catch in tonnes and total effort in thousands of trawl hours by year in the North West Slope Trawl Fishery
Total prawn landings Total scampi landings Total effort
1000 20
800 15
600
10 Landings (tonnes) Landings
400 Hours trawled (thousands) trawled Hours
5 200
0 0 1988 1992 1996 2000 2004 2008
Source: Data from AFMA
110 Reducing uncertainty in stock status ABARES
Trawl gear, targeting and catching deepwater prawns in NWSTF Wadley (1992) describes Aristaeomorpha foliacea as a highly aggregating species. The stomach contents of deepwater prawns along the north west slope and consideration of variation in catch rates during the day compared with at night suggest that A. foliacea and the two carid species, Heterocarpus woodmasoni and H. sibogae, migrate into midwater at night to feed. The other prawn species, Haliporoides sibogae, Aristeus virilis and Plesiopenaeus edwardsianus, are thought to be predominantly bottom feeders.
Aggregations of A. foliacea were able to be detected on echo sounders and targeted. Fishers were able to more effectively target prawn aggregations over time. Reduced marketability of the prawns has meant that the species have been subject to little or no targeted fishing for a number of years.
Evans (1992) suggests that when targeting prawns, 60 millimetre mesh is used on the wings compared with 90 millimetre mesh used on the wings when targeting scampi. In the NWSTF, a maximum codend mesh gauge of 50 millimetres is specified to discourage the targeting of finfish. Typically 45 millimetre codend mesh is used when targeting either prawns or scampi. Wadley (1992) analysed length frequency data of A. foliacea caught with a prawn configuration having 57 millimetre wing mesh size and 45 millimetre codend mesh. She found at various times both male and female cohorts with modal carapace lengths less than 30 millimetres. These were assumed to belong to the 1+ year cohorts in each case.
Fishers in the NWSTF can identify species targeted in their logbook records. Records of target species commenced in late 2007 and its importance will increase with time. Among existing records providing information on the target species, Australian scampi (Metanephrops australiensis) was the predominant target. No records specified any prawn species as the target. Current fishing mortality of deepwater prawns is as a result of bycatch from effort targeted at scampi.
A number of observer reports from trips taken with NWSTF fishers were sourced from the Australian Fisheries Management Authority (AFMA). As well as detailed records of catch quantities by species, these reports provide summaries of vessel gear type, target species, bycatch and discarding, logbook recording compliance and other compliance. The observer reports suggest that bycatch in the fishery accounts for a high proportion of total catch. A wide variety of taxa are recorded, with prawn tending to be the group caught in greatest quantities by weight. Because of discarding, total prawn catches in the NWSTF since 1993 are likely to have been substantially higher than quantities reported in fishers’ logbooks. Analysis Catch rate standardisation The distinct change in targeting behaviour of fishers in the NWSTF, as well as gear changes, makes constructing a single index of abundance for the deepwater prawn stock problematic. The catchability of prawns when fishers are targeting A. foliacea is likely to be much higher than when prawns are caught as bycatch by fishers targeting scampi. Some aspects of the targeting behaviour can be accounted for by standardising, but the effect on catchability of detecting an aggregation of prawns in time and space using echo sounding equipment cannot. To address this problem we exclude shots that we assess as having low probability of catching prawns from the analysis. In addition, since it appears that minimal targeting of prawns has occurred since 1993, and considering the possibility of unreported bycatch, we limit the period of analysis to the years 1986 to 1993.
111 Reducing uncertainty in stock status ABARES
Stephens and MacCall (2004) address the problem of species targeting in multispecies fisheries using logistic regression to subset logbook records where specific information identifying the target species was not available. A similar procedure is applied here except that the random forests data mining algorithm (Breiman 2001; Liaw & Wiener 2002) in the statistical software R is used to estimate the probability of each shot catching prawns instead of a generalised linear model. Lennert-Cody et al. (2008) used R’s random forests algorithm to investigate the determinants of catching bigeye tuna in purse-seine shots in the eastern Pacific Ocean. We estimate the probabilities of catching prawns at an average biomass across the period of analysis by excluding temporal variables (such as year) from the model. Variables latitude, longitude, depth, month, start time and effort were used in the random forests procedure to estimate the probability of catching prawns on a given shot. The importance of the variables, as measured by the associated mean decrease in the model Gini coefficient, is shown in the importance plot (Figure 6.3). Month was the most important single variable and the relative importance of latitude and longitude are likely to be understated because they are highly correlated with location of NWSTF effort concentrated along a strip parallel to the Kimberley coast line. Other variables of less importance were the mean depth in the c-square (CSQR_MEAN), the maximum depth in the c-square (CSQR_MIN), the maximum depth in the c-square (CSQR_MAX), the longitude and latitude of the c-square (CSQR_LONG and CSQR_LAT) and CSQR-RANGE, which is the difference between the maximum depth and the minimum depths in the c-square (Figure 6.3). The categorical variable c-square specifies the location of each trawl as a discrete six- minute latitude by six-minute longitude (Rees 2003). The number of variables should not affect results because the algorithm of the random forest model selects only the variables that are most useful at any point in time.
Figure 6.3 Importance plot for random forests model predicting the probability of catching prawns on a given shot in the North West Slope Trawl Fishery, 1986 to 1993 Targeting Prawns
MONTH
EFFORT
START_TIME
DEPTH
LONG_START
LAT_START
CSQR_RANGE
MAX_DEPTH
CSQR_MEAN
CSQR_LONG
CSQR_MIN
CSQR_MAX
MIN_DEPTH
CSQR_LAT
0 5 10 15 20 25 30 35 MeanDecreaseGini
Source: Data from AFMA
Changes in the abundance of prawns over the period of analysis is expected to explain some of the variation in the observed probability of catching prawns. The intention of the random forests model is to estimate the degree of targeting by fishers, which is interpreted as the probability of catching prawns averaged across the actual levels of abundance during the period of analysis. Ideally characteristics of fishing gear would be included in the random forests model as they are
112 Reducing uncertainty in stock status ABARES indicative of the targeting decisions of fishers and would also explain differences in the probability of catching prawns controlling for all other factors.
A delta lognormal model (Fletcher 2008) was used to estimate a relative abundance index from the subset logbook data. Expected prawn CPUE is given by the product of a binomial generalised linear submodel and a lognormal generalised additive submodel properly back transformed.
The binomial submodel included as predictors the random forests calculated proxy for targeting described previously in this chapter as well as the factor ‘quarter’. Here quarter denotes the main temporal variable for the index of abundance, not a seasonal variable that might be used to detect changes in prawn catchability at different times of year. The probability of catching prawns on shot i, given an estimated level of targeting and the quarter between 1986 and 1993 in which the shot occurred, is given by:
Among the shots deemed to be targeting prawns, those successful in catching prawns were fitted to a lognormal generalised additive model (GAM) (Figure 6.4). The lognormal GAM submodel has CPUE as its target variable and includes smooths for start time (time of day when trawling started), trawl depth and depth range of 6 minute c-square (an indication of slope of seabed) and a month factor.
Figure 6.4 Estimated smooths of variables in generalised additive model for conditional catch per unit effort
Note: Solid lines are the variables means, dashed lines are the 95 per cent confidence intervals and the short lines on the X axes are the rag plots.
113 Reducing uncertainty in stock status ABARES
Source: ABARES
Once the effect on conditional CPUE of the various factors is estimated in the GAM submodel, a standardised conditional CPUE index can be estimated. Figure 6.5 shows a constant decline in standardised conditional CPUE from 1987 to 1993. The questionable sharp increase in prawn biomass between 1986 and 1987 suggests the data for this period is unreliable. A hypothetical dataset is created with all combinations of the predictors for each quarter in the period of analysis. The GAM submodel is used to predict conditional CPUE for each row in the hypothetical dataset and standardised conditional CPUE is estimated by averaging the fitted values for each quarter over all combinations of predictors. The conditional CPUE model can be represented as:
log Y | Y 0 f X f X f X f X γX ij ij 1 1ij 2 2ij 3 3ij 4 4ij 5ij ij f , f , f , f X X X X where 1 2 3 4 are smooth functions, 1ij , 2ij , 3ij , 4ij are assumed to be known, given the depth, quarter, start time and c-square range of shot i in quarter j. γ is a row vector of X coefficients specifying the effects of each month factor, 5ij is a column vector that picks out the appropriate element from γ, reflecting the month of trawl shot i in quarter j. The ij are 2 residuals lognormally distributed about zero, where the ij have variance j in quarter j. An approximately unbiased estimate of standardised conditional CPUE in quarter j is given by: ˆ exp Y 1 σˆ 2 j j 2 j Y where j is the mean of the predicted values across all levels of the hypothetical dataset corresponding to quarter j.
Figure 6.5 Standardised prawn catch per unit effort (shots reporting prawn catch) with a
five quarter moving average line
60
50
40
30
Conditional CPUE 20
10 0
1986 1987 1988 1989 1990 1991 1992 1993
Source: Data from AFMA
The overall index of abundance is calculated as the product of the standardised prawn strike rate and the standardised conditional CPUE (Figure 6.6):
114 Reducing uncertainty in stock status ABARES
ˆ ˆ ˆ j j j Figure 6.6 Quarterly catch (tonnes, retained) and index of abundance, North West Slope Trawl Fishery deepwater prawns, 1986 to 1993
Source: Data from AFMA
Delay difference model Delay difference models that incorporate a time step of less than one year have been applied to assess prawn stocks in recent years. For example Dichmont et al. (2003) used a weekly delay difference model for the abundance of tiger prawns (Penaeus esculentus and Penaeus semisulcatus) in Australia’s Northern Prawn Fishery, and O’Neill and Turnbull (2006) used a monthly model for the abundance of brown tiger prawns (P. esculentus) in the Torres Strait. Taking into account the slower growth of deepwater prawns, a delay difference model with a quarterly (three-month) time step is applied using the index of abundance derived from standardised CPUE.
Meyer and Millar (1999) used a state space implementation of the delay difference model in a Bayesian framework to allow flexible estimation of both observation error and process error variance. For the delay difference model, process error results from variation in recruitment, growth and mortality.
B B B Expected biomass in year t 1, t1 , given t and t1 can be expressed as:
Equation 1
M 2M Bt Ct E Bt1 | Bt , Bt1 1 e Bt Ct e Bt1 Ct1 Bt B C eM t t R R t 2, , N B t t1 t
115 Reducing uncertainty in stock status ABARES
k1 where k , k1 is the pre-recruit weight and k is the weight at recruitment. The constant, , is Ford’s growth coefficient given by exp K such that 0 1. M is natural mortality. Ct and Rt are total catch and recruitment in year t , respectively, and K is Brody’s growth coefficient from the von Bertalanffy growth relationship. Markov Chain Monte Carlo implementation The Markov Chain Monte Carlo (MCMC) algorithm implemented in the statistical software WinBUGS (Spiegelhalter et al. 2003) is a powerful and flexible method of parameter estimation, and also allows quantities related to risk (such as the probability of a stock falling below a particular proportion of unfished biomass) to be estimated in a straightforward manner.
Meyer and Millar (1999) report extremely slow convergence of Equation 1 partly as a result of large correlation between parameters. They address this problem by sampling from transformed state equations obtained by dividing through by the carrying capacity, B0. This gives state transition equations:
Equation 2
P1 1 u1
P k C P eM 1 eM P k C r 1 eM 1 1 u 2 1 1 P 2 1
M 2M Pt k Ct Pt1 1 e Pt k Ct e Pt 1 k Ct 1 Pt P k C r 1 eM t t u for t 2, , N P t 1 t and observation equations:
Equation 3
I QP v for t 1, , N t t t
Bt 1 R Pt k r B B B Q qB0 where 0 , 0 , 0 and .
The quantities, Pt, are the unknown biomasses in years t as a proportion of the biomass in 1986, which is assumed to be at an unfished level. Recruitment, R, is assumed to be constant each year, but is not assumed to be equal to recruitment at equilibrium. Catchability, q, is also assumed to be constant each year. The process errors, ut, are assumed to be independent and identically 2 v normally distributed, such that ut ~ N 0, . The observation errors, t , are assumed to be independent and normally distributed with approximately constant coefficient of variation, such v ~ N 0,w 2 w that t t . The weights, t , are proportional to the squared fitted values obtained using a robust smoothing of time series of means by running medians (using the R function smooth) equivalent to the procedure described by Meyer and Millar (1999, p. 42).
116 Reducing uncertainty in stock status ABARES
Hyperstability Equation 3 assumes that abundance is proportional to standardised CPUE. This is questionable in a fishery where the target species aggregate into fishable shoals and are able to be detected using echo sounding equipment. The ability for trawls to be actively targeted to aggregations of prawns suggests that fishery dependent catch-effort data might be prone to hyperstability, which is graphically represented in Figure 6.7. True effort may not be well represented by trawl hours if the time that fishers spend searching for aggregations using their echo sounders before commencing trawling varies considerably. If reductions in standing stock caused by fishing resulted in fewer aggregations, but the aggregations that remained were of a similar size and density to those fished previously, then catch rates may be maintained to some degree. The additional time that fishers spend searching for aggregations as they become more scarce is not recorded in the logbook data. Zhou et al. (2007) found evidence of increased catchability of banana prawns with decreased biomass. Harley et al. (2000) suggest that, of a number of possible nonlinear relationships between abundance and catch per unit effort, the simplest can be represented by:
I QP v for t 1, , N t t t .
In this representation, 0 1 indicates hyperstability and 1indicates a hyperdepletion fishery. The possibility of 1can theoretically be considered by estimating this parameter within the MCMC process or by testing the effect of various plausible values through sensitivity analysis.
Figure 6.7 Graphical representation of suggested power law relationship between catch per unit effort and abundance
hyperstability
= 0.25
cpue = 0.5
= 1 hyperdepletion
= 2
= 4
abundance
Die and Ellis (1999) present a counterargument to the suggestion of hyperstability—for species where only a proportion of the total stock is aggregated at any given time, the rate of decline for the total stock is lower than for the aggregated proportion of the population. According to this argument, the extent of hyperstability will depend on the proportion of the stock on average in
117 Reducing uncertainty in stock status ABARES detectable aggregations. Aside from A. foliacea, the other prawn species were less able to be detected using acoustic gear so the catchability for these species is likely to have been considerably lower. Specification of prior distributions for model parameters Punt and Hilborn (1997) advocate using informative prior distributions for parameters that can be informed by existing data and previous analyses.
The findings of Baelde (1994) were used as prior information for parameters related to growth and mortality. Meyer and Millar (1999) used the approximation 0 assuming that the pre- recruit weight of albacore tuna was zero. This approximation seems reasonable for deepwater prawns given growth curves for Haliporoides sibogae in Baelde (1994, p. 621), although the use of a quarterly time step may diminish its reasonableness. Nevertheless, the approximation 0 was made allowing simplification of the state transition equations giving:
Equation 4
P1 1 u1
M M P2 e 1 e P1 k C1 r u2 P k C P 1 eM P k C e2M t t P k C r u for t 2, , N t1 t t P t 1 t 1 t 1 t The rapid decrease in standardised catch rates in response to catch levels between 1987 and 1993 suggests that the catch in these years represented a large proportion of the standing stock at that point in time. Since exploitation began at roughly this time, the standing stock at the beginning of the fishery might be expected to be a reasonable estimate of the carrying capacity, although prawn stocks can be prone to fluctuations in population size because of environmental or other factors. Therefore, it seems unlikely that the carrying capacity is much more than 4 000 tonnes. A normal prior is defined for the stock carrying capacity with mean 1 tonne, but truncated on the left at its mean so that 80 per cent of the probability mass is between 1 and 20 000 tonnes.
To construct a prior for specific recruitment, r, Meyer and Millar (1999, p. 42) relate the prior distribution for this quantity to natural mortality, M, based on the assumption of linear growth and a stock in equilibrium:
Equation 5 1 r M a exp M a 1 a 1
Baelde (1994, p. 617) suggests a natural mortality of around 0.6 year −1 for Haliporoides sibogae. If we interpolate natural mortality as 0.15 quarter −1, Equation 5 yields specific recruitment of around 0.02 quarter−1. We define a gamma (1.1, 5) prior for quarterly specific recruitment, r, which is largely flat over likely values of r and has roughly 90 per cent of probability mass for r less than 0.5.
118 Reducing uncertainty in stock status ABARES
2 The observation error variance, , can be estimated from the residual variance of the standardised CPUE model that was used to produce the index of abundance series (Maunder & Punt 2004). Observation error is estimated to be around 0.68 on the scale of the index of abundance. Here we specify a log-normal distribution with mean −0.38 and variance 0.5 on the log scale. This distribution has roughly 80 per cent of its probability mass between 0.5 and 0.95.
2 A non-informative prior is defined for the process error variance, . Specifically, the standard deviation of process error is assigned a uniform prior over the interval 0.0001 and 10.
Meyer and Millar (1999) assign fixed values for the mortality and growth parameter of albacore tuna. Baelde (1994) suggests values for natural mortality of 0.6 year-1 and Brody’s growth coefficient, K, for female Haliporoides sibogae of 0.37 year-1. We assign a fixed value for natural mortality of 0.15 quarter−1 as a base case. Papaconstantinou and Kapiris (2003) estimate the Von Bertalanffy growth coefficient, K, for an A. foliacea stock in the Greek Ionian Sea to be 0.460 year-1 or equivalently 0.115 quarter-1. Fords’s growth coefficient is given by exp K (see, for example, Schnute 1985). We assign a fixed value of 0.89 . It was found that, when natural mortality was estimated, the correlation between natural mortality and recruitment was particularly strong. Convergence diagnostics for MCMC parameter estimation Reliable parameter estimation requires that MCMC chains be tested for convergence. Initially three chains were used to enable the use of the Gelman-Rubin diagnostic and to better check for good mixing as well as sensitivity to initial parameter values. Autocorrelation of MCMC parameter estimates was persistent at high lags. Autocorrelated chains tend to yield posterior distributions with credible intervals that are too narrow. To address this problem parameter chains were thinned so that only every 30th sample from the posteriors was kept. We ran 430 000 iterations of the MCMC chain, discarded the first 70 000 as a ‘burn-in’ period, then thinned the chain by sampling every 30th iteration giving a final chain length of 12 000.
The thinned chains exhibited no autocorrelation problems. The Raftery-Lewis diagnostic suggested that the length of the chain was sufficient and that the period of burn-in was sufficient. Geweke diagnostics suggested no evidence of difference between the beginning and final sections of chains. The Heidelberger-Welch diagnostic suggested that the mean of the posterior distributions had been estimated with satisfactory precision. For one scenario the chain estimating the final biomass failed the Heidelberger-Welch diagnostic. The MCMC was rerun and convergence confirmed. Plots of cumulative quantiles were stable.
A range of alternative priors for each stochastic parameter were trialled. The posterior for the carrying capacity was sensitive to the specified priors. The sensitivity of other parameters was minimal.
The MCMC algorithm strongly favours a large, relatively unproductive stock (Figure 6.8). This conflicts with common perceptions about the nature of prawn stocks. Deepwater prawns, although less productive than prawns from shallow waters, are still thought to be highly productive. This unexpected result might be because of particular features of the relationship between the time series of catch and the index of relative abundance as shown in Figure 6.6. The index of abundance suggests that, following the quarter of largest catch, biomass increased slightly. This characteristic is likely to contribute to parameter likelihood functions that are uninformative or poorly located in the parameter space.
119 Reducing uncertainty in stock status ABARES
Figure 6.8 Prior distributions (dotted lines) and posterior distributions (solid lines) for estimation of each of specific quarterly recruitment rate, r, carrying capacity, B0, observation error variance, τ2, and process error variance σ2 for the base case (M = 0.6, β = 1)
Note: Where prior distributions have been truncated, the graphs have been normalised to have unit area. Source: ABARES
Incorporation of hyperstability reduced the tendency of the algorithm to estimate a dubious peak in biomass in 1987 (Figure 6.9). The period of high catches in the NWSTF lasted only two years. Results from models assuming hyperstability (Figure 6.10) and severe hyperstability (Figure 6.11) suggest the stock was depleted to a greater extent by the end of 1993 and would be expected to have experienced only a slight recovery.
The apparent failure of the stock to return to unfished biomass after a significant period of negligible fishing mortality suggested by all models can probably be explained by limitations in the delay difference model used in this analysis. The model assumes that recruitment is independent of biomass and estimates a single value that is best supported by the catch and effort data given this assumption. The predicted recruitment parameter is likely to be a compromise between high recruitment expected with stock levels at or near carrying capacity
120 Reducing uncertainty in stock status ABARES and somewhat diminished recruitment that would have been expected when the spawning biomass was much lower. The assumption of biomass independent recruitment also casts doubt on the predicted speed of the recovery. The incorporation of a realistic stock recruitment relationship might be expected to produce a recovery that was initially slower, but ultimately returned the stock to a level nearer its carrying capacity.
Figure 6.9 Time series plot of quantiles of posterior distribution of deepwater prawn biomass as a proportion of unfished biomass as estimated by quarterly delay difference
model (M = 0.6, β = 1)
1.4
1.2
1.0
0
0.8
B/B
0.6
0.4 0.2
1990 1995 2000 2005
Note: Shown are the posterior median (solid line), upper and lower quartiles (dashed lines), 80 per cent credible interval (dotted lines). Estimates from 1994 are projections taking into account reported catch. Source: ABARES
121 Reducing uncertainty in stock status ABARES
Figure 6.10 Predicted time series of deepwater prawn standing stock as a proportion of
virgin biomass, M = 0.6 year-1, β = 0.5
1.4
1.2
1.0
0
0.8
B/B
0.6
0.4 0.2
1990 1995 2000 2005
Note: Shown are the posterior median (solid line), upper and lower quartiles (dashed lines) and 80 per cent credible interval (dotted lines). Source: ABARES
Figure 6.11 Predicted time series of deepwater prawn standing stock as a proportion of
virgin biomass, M = 0.6 year-1, β = 0.25.
1.4
1.2
1.0
0
0.8
B/B
0.6
0.4 0.2
1990 1995 2000 2005
Note: Shown are the posterior median (solid line), upper and lower quartiles (dashed lines), 80 per cent credible interval (dotted lines). Source: ABARES
122 Reducing uncertainty in stock status ABARES
Sensitivity to natural mortality The base case assumed natural mortality was 0.6 year-1. Baelde (1994) considers a natural mortality of M = 0.4 year-1 as a ‘low’ figure for deepwater Haliporoides sibogae and M = 0.8 year-1 as a ‘high’ figure. We explore the sensitivity of the model to these values of natural mortality.
The models are seen as quite insensitive to the value of natural mortality assumed (Figure 6.12 and Figure 6.13). This might be a result of correlations between natural mortality and other parameters. The suggested increase in biomass between 1986 and 1987 is most likely an artefact of increased targeting of prawns relative to scampi in 1987 that was not fully accounted for in the standardisation process.
Figure 6.12 Predicted time series of deepwater prawn standing stock as a proportion of
virgin biomass, M = 0.8 year-1
1.4
1.2
1.0
0
0.8
B/B
0.6
0.4 0.2
1990 1995 2000 2005 Note: Shown are the posterior median (solid line), upper and lower quartiles (dashed lines), 80 per cent credible interval (dotted lines). Source: ABARES
123 Reducing uncertainty in stock status ABARES
Figure 6.13 Predicted time series of deepwater prawn standing stock as a proportion of
virgin biomass, M = 0.4 year-1
1.4
1.2
1.0
0
0.8
B/B
0.6
0.4 0.2
1990 1995 2000 2005
Note: Shown are the posterior median (solid line), upper and lower quartiles (dashed lines), 80 per cent credible interval (dotted lines). Source: ABARES
Exploratory delay difference models: recovery trajectory Schnute (1985) notes that a characteristic of delay difference models is their biological realism when compared for instance with surplus production models. This biological realism enables estimates of biological parameters for stocks of interest to be incorporated into the models. Expected population trajectories can then be calculated subject to assumptions about the size of the stock, the relative biomass at a particular point in time and the accuracy of the biological parameter estimates.
In a separate modelling exercise we examine the likely recovery trajectory after targeted prawn fishing ceased. The maximum annual prawn catch of 790 tonnes is used as a conservative estimate of the size of the prawn stock. We assume that, at the end of 1993, the stock was depleted to 10 per cent of its unfished biomass and that since 1993 unrecorded bycatch has been the minimum of 50 tonnes and 90 per cent of total prawn biomass in that year. We model the stock according to Equation 1 with ω = 0. The stock recruitment relationship is assumed to follow a hockey stick form (Barrowman & Myers 2000) with lognormally distributed errors.
Uncertainty in the biological parameters is incorporated into the modelling procedure by sampling from distributions as a Monte Carlo simulation. Natural mortality is sampled from a normal distribution with mean 0.6 year-1 and standard deviation 0.2 year-1 truncated at 0.4 year- 1 and 0.8 year-1. The growth parameter is held fixed at K = 0.37. Maximum expected recruitment is set at a level to maintain the stock at 790 tonnes in the absence of fishing given the sampled M. A steepness is selected from a normal distribution with a mean of 0.5 and standard deviation 0.1, truncated at 0.2 and 1. A coefficient of variation of 0.60 is assumed for the lognormal errors in the stock recruitment relationship. The stock characteristics are set to be conservative rather than best estimates. For instance Hoenig’s equation (Hewitt & Hoenig 2005) suggests a natural mortality of M = 0.73 for a crustacean living six years and M = 0.83 for a crustacean living five years.
124 Reducing uncertainty in stock status ABARES
Even with what are believed to be conservative assumptions about the stock size, depletion, bycatch and prawn biology, the stock is predicted to recover with a high degree of confidence (Figure 6.14).
Figure 6.14 Recovery trajectory of deepwater prawn standing stock as a proportion of virgin biomass
Note: Shown are medians and 90 per cent and 10 per cent confidence intervals of 500 Monte Carlo draws from delay difference model, with five individual trajectories for illustration. Source: ABARES Discussion
The models used are not able to provide precise estimates for the unfished biomass of the deepwater prawn stock in the NWSTF from the catch and effort data. The models do seem to suggest that the stock is likely to be larger than 5 000 tonnes. Schnute (1985) points out that delay difference models can perform poorly in practice when heavily reliant on catch and effort data. For a number of reasons the catch and effort data used in this analysis are particularly problematic. The ratio of predicted biomass in 2008, B2008, to unfished biomass, B0, also exhibits considerable uncertainty as might be expected given the length of time over which biomass is projected. For most sensitivities tested, the full 80 per cent confidence interval suggests that the stock is not overfished. It is surprising that the median estimates of depletion (B/B0) do not return to close to an unfished state (1) despite an extended period of very low catch. The separate modelling of the recovery trajectory following cessation of targeted fishing, under a pessimistic set of parameters, did indicate a recovery of the stock to well above default target levels.
Some doubt surrounds the ability to obtain a meaningful index of abundance from catch rates obtained from the NWSTF logbook data given evidence of aggregation targeting. Although some effort has been made to consider the effects of possible hyperstability, it is possible that none of
125 Reducing uncertainty in stock status ABARES
the scenarios tested provides a reasonable approximation of the effect of targeted fishing on the deepwater prawn stock.
Among the nine scenarios investigated, the maximum estimated probability of biomass in 2008 being below 20 per cent of unfished biomass was 7.1 per cent, occurring when M = 0.4 and the hyperstability parameter β was 0.25 (Table D1). Even in this worst case scenario the stock would be considered not overfished with greater than 90 per cent confidence. The recent low levels of reported catch indicate that the stock is unlikely to be subject to overfishing. Further work Given questions about selectivity and targeting of deepwater prawns within current NWSTF effort and the lack of prawn catch recorded in logbooks in recent years, fishery independent data would need to be obtained to complete a comprehensive study. The aggregating nature of prawns would mean that sampling by random transects would result in estimates with high variance. An adaptive sampling methodology might be more suitable where the probability of encountering an aggregation with echo sounding equipment per swept area is multiplied by estimates of the size of the aggregations. Collection of a complete set of length frequency data tracking all cohorts monthly for 12 months would allow improved estimation of growth, recruitment and natural mortality characteristics of the stock.
However, the low economic value of the stock, the fact that it is not being targeted and that fishing mortality appears to be low suggests that such a study would be difficult to justify financially.
Existing bycatch data provided by AFMA observers should be compiled and looked at carefully. If the level of prawn bycatch was considered to be a problem, the effect of lifting the restriction on codend mesh gauge could be trialled, possibly in conjunction with other methods of bycatch mitigation. The effect on the selectivity of scampi would be a primary concern in any such trial. Appendix D Table D1 Summary of parameter estimates from posterior densities of delay difference models with β = 1, β = 0.5 and β = 0.25, assuming a natural mortality of M = 0.6 year-1 1 0.5 0.25 Parameter 25% mean 75% 25% mean 75% 25% mean 75%
P1993 0.621 0.723 0.818 0.462 0.626 0.745 0.311 0.443 0.525
P2008 0.618 0.822 1.04 0.522 0.759 0.986 0.326 0.524 0.683
B0 9 274 16 133 21 397 7 694 14 666 19 891 9 643 16 367 21 639 r 0.023 0.027 0.033 0.019 0.025 0.031 0.012 0.017 0.022 σ2 0.001433 0.0028 0.003653 0.000593 0.002 0.002835 6.60E-05 0.001 0.00138 τ2 0.126 0.167 0.197 0.168 0.217 0.253 0.182 0.23 0.266 Q 0.364 0.402 0.436 0.346 0.386 0.424 0.349 0.381 0.413
Pr(P2008) – 0.03 – – 0.043 – – 0.052 – <0.2 deviance –26.17 –21.89 –17.78 –17.25 –13.59 –10.19 –13.77 –11.74 –9.95
126 Reducing uncertainty in stock status ABARES
Table D2 Summary of parameter estimates for sample size 12 000 from posterior densities of delay difference models with β = 1, β = 0.5 and β = 0.25, assuming a natural mortality of M = 0.4 year-1 Parameter 1 0.5 0.25
25% mean 75% 25% mean 75% 25% mean 75%
P1993 0.624 0.731 0.818 0.458 0.599 0.726 0.38 0.481 0.485
P2008 0.519 0.776 1.048 0.468 0.708 0.93 0.329 0.515 0.505
B0 9 200 16 157 21 526 8 836 15 914 21 306 9 166 15 248 19 373 r 0.011 0.015 0.019 0.01 0.014 0.017 0.006 0.009 0.009 σ2 0.000947 0.002 0.00252 0.000648 0.0016 0.00217 1.70E-05 4.00E-04 0.000489 τ2 0.13 0.172 0.202 0.159 0.204 0.238 0.183 0.23 0.265 Q 0.362 0.4 0.437 0.354 0.39 0.425 0.344 0.375 0.405
Pr(P2008) < 0.2 – 0.058 – – 0.047 – – 0.071 – deviance –25.26 –21.01 –17.14 –18.75 –15.41 –12.41 –13.75 –11.62 –9.79 Table D3 Summary of parameter estimates for sample size 12 000 from posterior densities of delay difference models with β = 1, β = 0.5 and β = 0.25, assuming a natural mortality of M = 0.8 year-1
Parameter
25% mean 75% 25% mean 75% 25% mean 75%
P1993 0.619 0.727 0.827 0.422 0.692 0.938 0.411 0.701 1.007
P2008 0.591 0.81 1.039 0.444 0.778 1.107 0.515 0.817 1.149
B0 9 057 16 111 21 372 8 562 15 621 20 924 7 068 14 235 19 694 r 0.035 0.041 0.048 0.026 0.039 0.052 0.028 0.041 0.055 σ2 0.002279 0.0042 0.005409 0.00048 0.0026 0.003837 0.000592 0.0024 0.003439 τ2 0.125 0.168 0.2 0.174 0.224 0.263 0.185 0.235 0.272 Q 0.365 0.405 0.442 0.328 0.383 0.433 0.316 0.353 0.388
Pr(P2008) – 0.028 – – 0.037 – 0 0.039 0 <.2 deviance –26.38 –21.78 –17.2 –16.26 –12.47 –8.78 –13.32 –10.68 –8.34 References
Baelde, P 1994, ‘Growth, mortality and yield-per-recruit of deep-water royal red prawns (Haliporoides sibogae) off eastern Australia, using the length-based MULTIFAN method’, Marine Biology, vol. 118, pp. 617–25.
Barrowman, NJ & Myers, RA 2000, ‘Still more spawner-recruitment curves: the hockey stick and its generalizations’, Canadian Journal of Fisheries and Aquatic Sciences, vol. 57, pp. 665–6.
Breiman, L 2001, ‘Random forests’, Machine Learning, vol. 45, pp. 5–32.
Dichmont, CM, Punt, AE, Deng, A, Dell, Q & Venables, W 2003, ‘Application of a weekly delay- difference model to commercial catch and effort data for tiger prawns in Australia’s Northern Prawn Fishery’, Fisheries Research, vol. 65, pp. 335–50.
Die, DJ & Ellis, N 1999, ‘Aggregation dynamics in penaeid fisheries: banana prawns (Penaeus merguiensis), Marine & Freshwater Research, vol. 50. pp. 667–75.
127 Reducing uncertainty in stock status ABARES
Evans, D 1992, ‘The Western Deep Water Trawl and North West Slope Trawl Fisheries’, pp. 19– 27, in The fisheries biology of deepwater crustacea and finfish on the continental slope of Western Australia, Rainer, S. F. (ed), final report FRDC project 1998/74.
Fletcher, D 2008, ‘Confidence intervals for the mean of the delta-lognormal distribution’, Environmental and Ecological Statistics, vol. 15, pp. 175–189.
Harley, SJ, Myers, RA & Dunn, A 2000, ‘Is catch-per-unit-effort proportional to abundance?’, Canadian Journal of Fisheries and Aquatic Sciences, vol. 58, pp. 1760–72.
Hewitt, DA & Hoenig, JM 2005, ‘Comparison of two approaches for estimating natural mortality based on longevity’, Fisheries Bulletin, vol. 103, pp. 433–37.
Liaw, A & Wiener, M 2002, ‘Classification and Regression by randomForest’, R News, vol. 2, no. 3, pp. 18–22.
Lennert-Cody, CE, Roberts, JJ & Stephenson, RJ 2008, ‘Effects of gear characteristics on the presence of bigeye tuna (Thunnus obesus) in the catches of the purse-seine fishery in the eastern Pacific Ocean’, ICES Journal of Marine Science, vol. 65, no. 6, pp. 970–78.
Maunder, MN & Punt, AE 2004, ‘Standardizing catch and effort data: a review of recent approaches’, Fisheries Research, vol. 70, pp. 141–59.
Meyer, R & Millar, RB 1999, ‘Bayesian stock assessment using a state-space implementation of the delay difference model’, Canadian Journal of Fisheries and Aquatic Sciences, vol 56, no. 1, pp. 37–52.
Ntzoufras, I 2009, Bayesian Modeling Using WinBUGS, John Wiley & Sons, Hoboken, New Jersey.
O’Neill, MF & Turnbull, CT 2006, Stock Assessment of the Torres Strait Tiger Prawn Fishery (Peneaus esculentus), Department of Primary Industries and Fisheries, Queensland.
Papaconstantinou, C & Kapiris, K 2003, ‘The biology of the giant red shrimp (Aristaeomorpha foliacea) at an unexploited fishing ground in the Greek Ionian Sea’, Fisheries Research, vol. 62, pp. 37–51.
Punt, AE & Hilborn, R 1997, ‘Fisheries stock assessment and decision analysis: the Bayesian approach’, Reviews in Fish Biology and Fisheries, vol. 7, pp. 35–63.
Rees, T 2003, ‘“C-Squares” a New Spatial Indexing System and its Applicability to the Description of Oceanographic Datasets’, Oceanography, vol. 16, no. 1, pp. 11–19.
Schnute, J 1985, ‘A General Theory for Analysis of Catch and Effort Data’, Canadian Journal of Fisheries and Aquatic Sciences, vol 42, pp. 414–29.
Spiegelhalter, D, Thomas, A & Best, NG 2003, WinBUGS User Manual, Version 1.4, MCR Biostatistics Unit, Cambridge, United Kingdom.
Stephens, A & MacCall, A 2004, ‘A multispecies approach to subsetting logbook data for purposes of estimating CPUE’, Fisheries Research, vol. 70, pp. 299–310.
Wadley, VA 1992, ‘Distribution, growth and reproductive development of the Giant Red Prawn, Aristaeomorpha foliacea (Risso, 1827), in the North West Slope Trawl Fishery’, pp. 75–93. In The fisheries biology of deepwater crustacea and finfish on the continental slope of Western Australia, Rainer, S. F. (ed). Final Report FRDC Project 1998/74.
128 Reducing uncertainty in stock status ABARES
Wilson, D, Curtotti, R, Begg, G & Phillips, K (eds) 2009, Fishery status reports 2008: status of fish stocks and fisheries managed by the Australian Government, Bureau of Rural Sciences & Australian Bureau of Agricultural and Resource Economics, Canberra.
Zhou, S, Dichmont, C, Burridge, CY, Venables, WN, Toscas, PJ & Vance, D 2007, ‘Is catchability density-dependent for schooling prawns?’, Fisheries Research, vol. 85, pp. 23–36.
129 Reducing uncertainty in stock status ABARES 7 North West Slope Trawl Fishery Scampi assessment
Mark Chambers and James Larcombe Summary
Three species of scampi (Metanephrops spp.) have been harvested by the North West Slope Trawl Fishery (NWSTF) since 1983 using demersal trawl gear adapted for fishing at depths of around 400 metres. In the late 1980s, annual catch of the combined scampi species in the NWSTF averaged around 100 tonnes. At the peak of the fishery, around 20 vessels were operating in the NWSTF with fishers able to sell both their scampi and deepwater prawn catch. When markets for deepwater prawns disappeared the fishery became less profitable and the number of active vessels declined, resulting in a decline in the NWSTF scampi catch.
Before this assessment the scampi stock in the NWSTF had not been assessed. The ABARES 2008 Fishery status reports and earlier reports classified it as uncertain if overfished and uncertain if subject to overfishing (Wilson et al. 2009).
We use surplus production models to assess the scampi stock. The models provide estimates of various parameters describing historic abundance and exploitation rates. The base case assessment models the combined population of scampi in the area where the NWSTF operates. Alternative scenarios are considered that model discrete areas within the fishery separately. Based on these assessments it is unlikely the stock is overfished or that overfishing is currently occurring. Introduction
Scampi species are the target catch of the NSWTF. The predominant species caught are Metanephrops australiensis, M. velutinus and M. boschmai.
The NWSTF operates off north-western Australia (Map 7.1) and was established in 1983 after exploratory trawls suggested a potentially commercial scampi stock existed in the area (Evans 1992). After initially targeting scampi, fishers found they were able to locate aggregations of giant red prawns (Aristaeomorpha foliacea) using echo sounding equipment (B Wallner, pers. comm., 2010). By 1987 the prawn fleet had reached its peak size of 21 vessels (Evans 1992). After 1987 prawn and scampi catch rates fell sharply. The number of fishers operating in the fishery greatly reduced and those that remained reverted to targeting the higher priced scampi species.
Since 1994, effort in the NWSTF has been predominantly from a small number of vessels based in Western Australia. The fishery is managed by a limited number of permits (seven) and gear restrictions (codend mesh size ≤ 50 mm).
The ABARES 2008 Fishery status reports (the most recent report at the start of this study) and earlier reports classified the stock as uncertain to both overfished and overfishing status. The stock had not been previously assessed.
130 Reducing uncertainty in stock status ABARES
Map 7.1 Area of the North West Slope Trawl Fishery and recent relative fishing intensity
Source: ABARES
Approach to status determination This chapter is divided into four sections: catch and effort, catch rate and standardisation, surplus production models and discussion. The NWSTF scampi stock was assessed by standardising catch rates to develop an index of abundance, then applying surplus production models to estimate quantities of management interest. The models provide estimates of various quantities describing historic abundance and exploitation rates. The base case assessment models the combined population of scampi in the area where the NWSTF operates. Alternative scenarios are considered that model discrete areas within the fishery separately. Previous assessments Around 1983 there were research surveys by CSIRO and exploratory fishing trials by the Kailis & France Foods company to investigate the commercial potential of the scampi off the North–west Shelf. Davis and Ward (1984) reported the discovery of new NWSTF scampi grounds after a systematic trawl survey of the area, with additional concentrated surveys of grounds that had been located previously. Davis and Ward (1984) also estimated standing stock using the swept area density and the estimated total area of the ground. Their results are presented in Table 7.1.
131 Reducing uncertainty in stock status ABARES
Table 7.1 Standing stock of scampi in North West Slope Trawl Fishery by Davis and Ward (1984)
Ground Location Area Density Biomass “M. andamanicus ground” –17.75, 118.5 (~ central zone) 1 260 km2 7 kg/ha 880 t “M. australensis ground” –16.91, 119.9 (~ north central zone) 250 km2 12 kg/ha 300 t “M. boschmai ground” –14.82, 121.6 (~ north zone) 600 km2 11 kg/ha 660 t Total na na na 1 840 t Based on these results and assuming a natural mortality of 0.1 year-1 (from Nephrops norvegicus), Davis and Ward (1984) estimated a minimum annual sustainable catch of 184 tonnes from full exploitation of all identified grounds. Their estimated sustainable catch figure relied on the rule of thumb that a fishing mortality equal to natural mortality maximises sustainable yield. They suggested this was a conservative estimate but that work was required to refine the basic biological parameters assumed.
Wallner and Phillips (1995) found the following characteristics for Metanephrops in the NWSTF:
relatively long lived and slow growing, taking possibly six to eight years to attain commercial size
females brood eggs for 9–10 months of the year and probably spawn annually
produce only 100 to 900 larvae per brood; considerably less than other types of lobster
larvae settle to a benthic habitat soon after hatching, with little likelihood of wide dispersal to re-populate heavily fished areas.
Fowler and McLoughlin (1996) undertook a stock reduction analysis with catch and CPUE data up to and including 1992. They estimated a maximum sustainable yield of approximately 100 tonnes. The biological parameters used in their analysis are given in Table 7.2.
Table 7.2 Biological parameters and model settings for North West Slope Trawl Fishery combined scampi used in stock reduction analysis
Parameter Model setting
Natural_Mortality 0.46 year-1
L∞ 81 mm (CL) k 0.187 year-1 t0 0 a (growth) 0.0012 b (growth) 2.89 age at recruitment 3 years age at maturity 4 years steepness 0.75 recruitment cv 0 maximum age 10 years minimum initial biomass 500 tonnes maximum initial biomass 2 000 tonnes Note: Parameters apply to both sexes. Source: Fowler & McLoughlin (1996)
132 Reducing uncertainty in stock status ABARES
Lynch and Garvey (2005) undertook analysis of catch, effort, CPUE and size information sourced from the fishery logbooks. They adjusted raw catch rates to calculate indexes of abundance using three approaches. The first approach used only shots that targeted scampi by excluding shots with more than 50 per cent prawns and/or no scampi. The second approach was a representative area method, which examined catch rates in two grounds defined spatially and by depth—north of Rowley Shoals (M. australensis, less than 500 metres depth, north central zone) and south of Rowley Shoals (M. velutinus, less than 450 metres depth, central zone) (Map 7.1). The third approach was a proportional catch method after Garvey et al. (1992), where the fishery is spatially zoned and CPUE is weighted in each zone by that zones proportional area and effort level.
Several important findings of Lynch and Garvey (2005) were:
fishing grounds appear to be fully explored and it is unlikely that discovery of new grounds or exploratory effort is distorting current CPUE
no evidence that catches and CPUE were being sustained through serial exploitation of new grounds (hyperstability)
no evidence of growth overfishing (from size frequencies)
MSY of 75 tonnes likely to be a realistic sustainable yield (compared with 100 tonnes estimated by Fowler & McLoughlin, 1996).
Wayte et al (2007) undertook a risk assessment of the NWSTF using the productivity susceptibility method (PSA, level 2). Wayte et al (2007) comments in their summary:
Australian scampi, boschma’s scampi and velvet scampi are currently the main target species in the NWSTF. They are assessed at medium risk in the PSA. These species have been assessed in more detail in other analyses (Lynch & Garve, 2005). Although catch rates have declined, they are not considered to be over-exploited at current catch levels.
There is no information available for any of the target species on the overlap of their range with effort in the fishery. Fishing for scampi in the NWSTF has been confined to relatively small areas, and there is no evidence of serial depletion of scampi in the fishery. (Lynch & Garvey 2005) Zhou et al (2009) undertook a risk assessment of the NWSTF using the SAFE method (level 3); however, crustaceans were not considered in this work. Catch and effort Source and status of data Shot by shot logbook data have been collected since 1985. The format of the logbooks has changed twice in this time, becoming more detailed each time. Historic fishery data were stored in three separate databases. These data were combined into a continuous series using a common fields approach. This combined series of fisheries dependent data was the primary source data used in this assessment. Spatial distribution Scampi are taken from several spatially distinct grounds along a 1 000 kilometre stretch of the continental shelf break in north-west Australia. Scampi abundance is centred on the upper slope at depths of around 400 metres. The fishery was divided into four zones based on the history of fishing and spatial arrangement of shots reporting scampi (Map 7.2). These zones were considered as potential stocks in sensitivity analyses. Historically there has been some variation
133 Reducing uncertainty in stock status ABARES in the exploitation of grounds from year to year. For example, effort has shifted in and out of the northern and southern most grounds at various times. Map 7.2 Scampi zones used for spatially disaggregated analyses
Sources: ABARES; Geoscience Australia
Catch—total scampi Fishing in the NWSTF commenced in 1983 (Wallner & Phillips 1995) with catch and effort data available from fishers’ logbooks from 1985. Scampi are a primary target species of high value and there is little or no discarding or high grading of product (in contrast to NWSTF deepwater prawns). There does not appear to have been any significant fishing mortality on scampi before the fishery existed (aside from some surveys in the early 1980s). The only nearby fishery taking scampi is within the NPF and is located north of Darwin, some 700 kilometres north-east of the NWSTF. NWSTF logbook data alone give a complete picture of catches that are likely to affect abundance of scampi within the NWSTF.
Scampi catch and total effort increased rapidly from 1985 and remained high until 1992, when effort and catch declined greatly (Figure 7.1). In these early years annual catches averaged around 100 tonnes, and since 1993 annual catches have fluctuated around an average of 50 tonnes. Catch is derived from a different combination of zones each year, with no clear trend indicating sequential or serial exploitation.
134 Reducing uncertainty in stock status ABARES
Figure 7.1 Total Metanephrops spp. catch by zone and total North West Slope Trawl Fishery effort
180 25 000 north north central 160 central south central 140 hours Total fishery 20 000 120
15 000 100
80 hours tonnes 10 000 60
40 5 000 20
0 0
1999 2007 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 2000 2001 2002 2003 2004 2005 2006
Source: Data from AFMA Methods Catch rate standardisation The index of abundance for Metanephrops spp was based on total CPUE of combined scampi species modelled using an extension of the common delta lognormal approach.
An initial scampi targeting model was fitted in an attempt to capture the effects of historic changes in the fishery related to different levels of targeting deepwater prawns as opposed to scampi. The targeting model was used to estimate levels of scampi targeting as a factor. The targeting factor was then used as one of the predictors in the delta lognormal model.
Fisheries tend to become more efficient at catching target species over time. Improvements in the technology used for fishing might lead to increases in catch rates for a given abundance. The effects on CPUE indexes of an assumed 1 per cent and 2 per cent annual increase in fishing efficiency are explored. Estimating scampi targeting When modelling fishers’ catch and effort data, it is often necessary to account for different species targeting practices over time. In the NSWTF, for a given scampi abundance, much lower scampi catch would be expected for shots targeting prawns. Prawns were the predominant target in the NWSTF briefly in the late 1980s and into the early 1990s. To reasonably model the relative abundance of scampi across the history of the fishery from fishers’ logbook data, the effect of prawn targeting needs to be treated appropriately.
To address this problem, a preparatory scampi targeting model was fitted to fishers’ logbook data. The purpose of this model was to estimate the probability a NWSTF trawl shot in a particular location at a particular time of year would catch scampi given average scampi abundance. A binary generalised linear mixed model was used for this purpose. Probability of
135 Reducing uncertainty in stock status ABARES catching scampi was assumed to depend upon depth, month, vessel and zone. A normally distributed random c-square effect (Rees 2003) was incorporated to consider the effect of fine scale location on the probability of scampi catch. The categorical variable c-square specifies the location of each trawl as a discrete six-minute latitude by six-minute longitude:
The fitted model was used to predict the probability of scampi catch at average abundance for each shot. Each shot was then classified to one of four levels according to a scampi so that there were roughly equal numbers in each category. Delta lognormal models The delta lognormal (two part) model is commonly used to standardise fisheries CPUE data (see for example, Maunder & Punt 2004). Fletcher et al. (2005) advocate the use of the delta lognormal model in ‘a wide range of applications’ when modelling skewed data with many zeros.
Indexes based on the delta lognormal are derived from the outputs of two sub-models fitted to catch and effort data. The first part is a binary model that estimates the effects of various covariates on the probability, , of catching the target species. The second part is a lognormal model that estimates the expected log transformed catch of each shot in the event the shot were to catch the target species.
The binary model (part 1) component of the delta lognormal modelled the probability of non- zero scampi catch given the scampi targeting category, month, duration of trawl and year was:
logit log
The lognormal model (part 2) included fixed effects for scampi targeting, month, time of day and year as well as random vessel effects was:
log where is scampi catch per hour trawled for trawl shots with nonzero scampi catch. Fishing efficiency Fishing processes are expected to become more efficient over time as technological innovations are introduced and vessels and gear are upgraded. Such improvements facilitate higher catches for a given stock abundance so increase the effective catchability of the stock. The introduction of global positioning systems (GPS) in the early 1990s for instance is likely to have allowed operators to locate and exploit the most favourable trawl paths with greater efficiency. Conversely Garvey et al. (1992) reported that average total headline length decreased from 65 metres in 1985 to 54 metres in 1990, resulting in an estimated 17 per cent reduction in average swept area over this period. This suggests that, while some components of fishing power increased over the reference period, others have decreased to some extent in this time. The problem of changes in fishing efficiency was approached both explicitly and implicitly in modelling.
Ideally if good data on the characteristics of individual vessels were available, the effects of various innovations can be estimated and adjusted for as these innovations are fitted to individual vessels (for example, Bishop et al. 2008). Catch per unit effort models typically standardise for vessel as either a fixed or random effect. Recognising that existing vessels can upgrade by installing more powerful engines and gear improvements, vessel as a nested random
136 Reducing uncertainty in stock status ABARES effect of year in vessel allows the estimated relative efficiency of individual vessels to vary by year. In addition, the effect of explicitly indexing logbook recorded effort by a compounding 1 per cent and 2 per cent per year was explored as sensitivity analyses. Effort was indexed according to:
YEAR 1985 Effort Effort 1 Creep % ind raw 100 The catch per unit effort is then adjusted according to the indexed effort Catch CPUE ind Effort ind Surplus production models Fishery information used to manage low value or data poor fisheries is often limited to indicators such as CPUE indexes and total catch time series. This information is typically used to gauge the relative abundance of a stock at different points in time. However, CPUE alone does not provide estimates of absolute abundance, historic fishing mortality or maximum sustainable yield. The simplest population models that do provide estimates of these parameters are surplus production (or biomass dynamics) models. Only time series of catch and relative abundance are required for estimation of the model parameters. Despite their simplicity, surplus production models have been shown, in some circumstances, to be more accurate and more precise than more complex models (Polacheck et al 1993).
Surplus production models have been used to assess lobster fisheries in the past (for example, Jensen 1986; Morgan 1979) and simulation studies have suggested that they can be suitable for application in these fisheries (for example, Breen & Kendrick 1998; Yoshimoto & Clarke 1993).
Model fitting was undertaken using A Stock Production Model Incorporating Covariates (ASPIC) Version 5.34 (Prager 1994), part of the NOAA Fisheries Toolbox. ASPIC is a non-equilibrium implementation of the surplus production models of Schaefer, Fox and Pella & Tomlinson. ASPIC is able to incorporate multiple catch and CPUE (abundance) index series; however, in this analysis the data used in each run were simply catch and a single standardised catch rate.
A variety of model runs were undertaken (Table 7.3).The base case (run 1) used combined scampi catch from all species, assumed a single stock and fitted a Fox model with the initial biomass fixed at K (B0). Runs 1 to 5 were for combined scampi species assuming a single stock and explore sensitivities around effort creep and model parameters. Runs 6 to 9 were for combined scampi in each of the four zones (south central, central, north central and north (Map 7.2). Poor or no fit was obtained in some runs and this is noted under the comments in Table 7.3 along with any additional adjustments made to achieve a fit. Appendix E contains all time series inputs of catch and CPUE index.
137 Reducing uncertainty in stock status ABARES
Table 7.3 A Stock Production Model Incorporating Covariates production model runs for North West Slope Trawl Fishery scampi
Run Spatial structure Model CPUE series Initial biomass Comment 1 single Fox CPUE index B1/K = 1 fixed Base case 2 single Fox CPUE 1% effic B1/K = 1 fixed na 3 single Fox CPUE 2% effic B1/K = 1 fixed na 4 single Fox CPUE index B1/K estimated na 5 single Schaefer CPUE index B1/K = 1 fixed na 6 South central Fox CPUE index B1/K = 1 fixed na 7 Central Fox CPUE index B1/K = 1 fixed na 8 North central Fox CPUE index B1/K estimated Poor fit to base model so allowed B1/K to be estimated 9 North Fox CPUE index B1/K estimated Poor fit to base model so allowed B1/K to be estimated Results Catch rate standardisation Standardised CPUE for combined NWSTF scampi (Figure 7.2) exhibit a rapid decreasing trend in the first years of the fishery (1985 to 1987), followed by a stable period at around 0.4 during 1988 to 1993. CPUE then increased and fluctuated around 0.7 (1994 to 2001) before decreasing to 0.5 in recent years (2002 to 2007). The standardised CPUE that assumes a 1 per cent and 2 per cent annual fleet efficiency increase results in the same patterns but lower indexes over time. The average CPUE of the most recent four years is 0.52 (0.41 and 0.33 for 1 and 2 per cent assumed annual efficiency increases, respectively).
138 Reducing uncertainty in stock status ABARES
Figure 7.2 Combined scampi catch with accompanying base case catch per unit effort index (solid line) and catch per unit effort indexes that assume 1 per cent and 2 per cent annual efficiency increase
totalTotal Metanephrops Metanephrops (logs) spp. * (logs)* 180 indexIndex total total Metanephrops Metanephrops spp. 1.2 TotalTotal Metanephrops Metanephrops with spp. 1% with annual 1% annual 160 TotalTotal Metanephrops Metanephrops with spp. 2% with annual 2% annual 1.0 140
120 0.8
100 0.6
tonnes 80
60 0.4 CPUE index
40 0.2 20
0 0.0
1991 1985 1986 1987 1988 1989 1990 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Source: Data from AFMA
Appendix E contains complete annual time series of CPUE indexes catches for each species and, in the case of M. australiensis and M. velutinus, CPUE indexes by zone.
139 Reducing uncertainty in stock status ABARES
Surplus production models Results for each of the nine runs are summarised in Table 7.4.
Table 7.4 Summary results of production model runs for North West Slope Trawl Fishery scampi
Parameter Combined scampi, single stock Combined scampi by zone base 1% 2% B1/K Scha SC C NC N Run 1 2 3 4 5 6 7 8 9 Goodness-of-fit R-squared in CPUE 0.44 0.39 0.17 0.51 0.42 0.19 0.50 0.09 0.28 Total object funct (Loss) 1.14 1.57 2.00 1.08 1.42 2.50 1.35 0.57 3.33 MSE 0.06 0.08 0.10 0.06 0.07 0.14 0.07 0.03 0.19 RMSE 0.24 0.28 0.32 0.24 0.27 0.37 0.27 0.16 0.43 Contrast index (ideal = 1.0) 0.59 0.56 0.56 0.78 0.60 0.60 0.72 0.24 0.54 Nearness index (ideal = 1.0) 0.96 0.93 0.92 0.95 1.00 0.97 1.00 0.61 1.00 Model parameter estimates B1/K (Starting biomass) 1 1 1 1.19 1 1 1 1 0.37
K (B0) 403.4 673.6 788.7 372.7 322.4 116.2 117.4 370.4 231.1 phi (BMSY/K) 0.37 0.37 0.37 0.37 0.50 0.37 0.37 0.37 0.37 q (Catchability Coefficient) 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.00 0.00 Management & derived parameters MSY 101.4 81.7 74.4 102.1 99.1 16.9 19.9 103.4 30.6
BMSY 148.4 247.8 290.2 137.1 161.2 42.8 43.2 136.3 85.0
FMSY 0.68 0.33 0.26 0.75 0.62 0.40 0.46 0.76 0.36 Exponent in production 1 1 1 1 2 1 1 1 1
B2008/BMSY 2.16 1.91 1.77 2.16 1.70 2.12 1.95 2.49 2.45
F2007/FMSY 0.26 0.36 0.43 0.25 0.33 0.33 0.51 0.09 0.00 Ye. (Equilibrium yield 2008) 50.7 54.9 56.6 50.5 50.4 8.8 12.9 23.0 7.7 Note: See Table 7.3 for description of runs. Source: ABARES
Fits to the combined scampi, single stock base case and sensitivities (runs 1 to 5) were generally good. The time series data for these runs show good contrast with early high catches resulting in a rapid fall in CPUE, followed by a fall in catches and a recovery in CPUE. In all cases CPUE fits better in the earlier years, but these models consistently predict CPUE to be higher than observed in recent years (Figure 7.3). Base case results with associated 95 per cent confidence ranges from bootstrapping were: MSY = 104 tonnes (82–132 tonnes), B2008/BMSY = 2.15 (1.95– 2.29) and F2007/FMSY = 2.57 (1.83–3.49) (Figure 7.4). Runs which assumed 1 per cent and 2 per cent annual efficiency increase gave lower estimates of MSY (82 tonnes and 74 tonnes, respectively), lower relative biomass and higher fishing mortality relative to FMSY. The base case was not sensitive to allowing the initial stock size to be estimated (run 4). Fitting a Schaefer (logistic) shape model (run 5) resulted in an MSY of 99 tonnes.
140 Reducing uncertainty in stock status ABARES
Figure 7.3 Base case (run 1) production model, fit to catch per unit effort index (1985 to 2011)
Source: ABARES
Figure 7.4 Base case (run 1) production model, biomass relative to BMSY and fishing mortality relative to FMSY
F / FMSY B / BMSY
2.8
2.4
2.0
1.6
1.2
0.8
0.4
0.0
1988 1992 1996 2000 2004 2008
Source: ABARES
The success of model fitting for the zone specific runs (6 to 9) was variable. The contrast seen in the combined data (runs 1 to 5) was absent from some of runs 6 to 9, so model fits tended to be poorer, yielded implausible values or sometimes failed altogether.
141 Reducing uncertainty in stock status ABARES
Discussion and status Catch rates: status implications Conclusions about stock status using CPUE alone require the standardised CPUE to reasonably accurately reflect abundance and biomass at the start of the fishery (in 1985) was close to an unfished level (B0). The almost 50 per cent fall in combined scampi CPUE in the second year of the fishery implies a 50 per cent fall in biomass after the catch of some 100 tonnes of fish in
1985 (and hence a B0 of some 200 tonnes assuming no recruitment). This is not plausible given annual catches greater than 100 tonnes in the four subsequent years (1986 to 1989), which suggest a larger initial stock size. Similarly, the survey estimates of Davis and Ward (1984) indicated biomass of more than 1 000 tonnes. Other factors are most likely responsible for this rapid initial decline.
It has been observed (B Wallner, pers. comm., 2010) that once ground has been well trawled, scampi catch rates drop quickly, but if grounds are rested the catch rates tend to recover also quite quickly. This suggests that the reductions in catch rates, in some circumstances, might result from scampi staying in their burrows because of disturbance of the sea bed rather than depletion as a result of fishing mortality. That is, observed declines in catch rates might be partly a result of scampi spending less time out of their burrows following fishing activity. The catch rate abundance indexes developed here are probably still quite informative. The absolute number of scampi out of their burrows and available to be caught at least at the first trawl over a particular region is likely to be strongly related to scampi abundance in that region. Surplus production models: status implications For the purpose of status determination, discussion here has been limited to the combined scampi, single stock runs (1 to 5), which were most robust (Table 7.3). MSY estimates for these scenarios ranged from 74 tonnes to 102 tonnes, in the context of mean annual catches of
71 tonnes and a catch peak of 153 tonnes in 1988 (Table 7.4). Trends in the ratios F/FMSY and B/BMSY for the base case (Figure 7.4) indicate heavy initial exploitation but combined stock biomass has remained above BMSY and F has remained below FMSY (except in 1988) for the entire history of the fishery. Across runs 1 to 5 the biomass ratios for the most recent year (B2008/BMSY) ranged from 1.7 to 2.16 (Table 7.4)—this would indicate that the stock is not overfished. Across runs 1 to 5 the fishing mortality ratios for the most recent year (F2007/FMSY) ranged from 0.25 to 0.43 (Table 7.4)—this would indicate that the stock is not subject to overfishing.
The MSY estimates for the NWSTF should not be interpreted as an opportunity for a significant increase in catch for a number of reasons. Wallner and Phillips (1995) reported evidence of faunal changes in heavily trawled areas in the North West Slope and evidence of increased CPUE in areas permitted periods of rest. The burrowing behaviour of scampi might mean that prerecruit classes are susceptible to increased mortality because of disruption of the seabed resulting from trawling. Additionally, mortality of berried (with eggs) females and detachment of undeveloped eggs particularly in M. velutinus might result in reduced productivity of the resource if effort is substantially increased. The high cost nature of the fishery is likely to mean that profitable catch rates require that biomass levels be sustained well above BMSY.
It was observed in the section on Catch rates—status implications that catchability may be non constant—with catchability higher in early shots on a ground and falling away because more scampi stay in their burrows. In practice this would mean catch rates would decline as more effort was put into the grounds, partly because of this disturbance effect where biomass becomes unavailable for capture. This also might act as a kind of protection response to the
142 Reducing uncertainty in stock status ABARES biomass in the face of increasing effort levels. The effect of a change in catchability might explain why the amount taken from the Central Zone is much less than the estimated maximum sustainable yield.
143 Reducing uncertainty in stock status ABARES
Appendix E Table E1 North West Slope Trawl Fishery, total catch by species (modelled), catch per unit effort indexes by species and total effort
Catch (tonnes) CPUE index CPUE indexes with assumed fishing efficiency increase Effort Year Total Metanephrops (logs) a Index total Metanephrops Total Metanephrops with 1% annual Total Metanephrops with 2% annual Effort (total fishery hrs) 1985 93.9 1.000 1.000 1.000 5 812 1986 115.1 0.529 0.518 0.513 11 197 1987 118.9 0.408 0.395 0.387 19 359 1988 152.8 0.392 0.374 0.363 17 242 1989 109.2 0.333 0.318 0.306 13 975 1990 79.5 0.305 0.270 0.257 16 751 1991 60.9 0.327 0.289 0.272 11 130 1992 124.8 0.425 0.360 0.336 15 420 1993 29.7 0.377 0.275 0.254 6 656 1994 38.1 0.578 0.437 0.400 3 667
144 1995 37.3 0.653 0.471 0.427 3 187
1996 44.2 0.810 0.595 0.534 3 017 1997 67.9 0.546 0.411 0.365 5 651 1998 35.9 0.506 0.390 0.343 3 257 1999 55.7 0.715 0.554 0.482 4 709 2000 73.5 0.809 0.661 0.571 5 170 2001 102.7 0.611 0.486 0.415 8 648 2002 42.1 0.351 0.253 0.214 4 916 2003 55.8 0.421 0.284 0.238 5 484 2004 41.3 0.510 0.403 0.334 3 788 2005 72.2 0.543 0.400 0.328 7 024 2006 26.4 0.499 0.386 0.314 2 549 2007 56.8 0.352 0.259 0.209 5 899 a :Catch of total Metanephrops confined to the fishery boundaries (Lat <=–12 and >=–21.5, Long >=114 and <=125).
Reducing uncertainty in stock status ABARES
Table E2 North West Slope Traw Fishery Zone Statistics: Total scampi catch (logbook), total effort (raw within the zone) and catch per unit effort indexes
Year Catch (tonnes) Effort (raw hours) CPUE indexes (base) South Central North North Hours total South Central North North South Central North North central central fishery central central central central
1985 0.3 55.4 37.9 0.3 5 812 27 3 258 2 488 44 NA 1 0.767 0.299 1986 26.3 26.9 54.5 7.3 11 197 2 417 4 001 4 241 537 0.518 0.409 0.579 0.394 1987 37.3 39.9 14.0 27.7 19 359 3 438 11 907 1 525 2 489 0.498 0.283 0.592 0.435 1988 30.0 24.7 97.7 2.7 17 242 2 081 8 287 6 594 450 0.578 0.221 0.755 0.357 1989 14.5 23.2 44.2 27.3 13 975 1 480 5 784 4 613 2 097 0.348 0.21 0.69 0.545 1990 18.1 13.6 44.0 3.7 16 751 3 985 7 274 5 054 440 0.293 0.235 0.513 0.379 1991 24.3 9.9 14.4 12.3 11 130 4 262 4 053 1 554 1 261 0.312 0.216 0.688 0.424 1992 14.2 28.2 49.8 32.6 15 420 3 034 4 239 5 262 2 885 0.352 0.329 0.623 0.481
145 1993 3.3 9.3 15.5 1.6 6 656 1 835 2 967 1 682 170 0.205 0.303 0.768 0.587 1994 0.0 0.0 30.0 8.1 3 667 — 5 2 847 815 — — 0.693 0.572
1995 18.5 8.3 10.4 0.0 3 187 1 450 789 948 — 0.514 0.661 0.597 — 1996 4.6 9.7 24.0 5.9 3 017 291 742 1 627 356 1 0.562 1 1 1997 16.7 8.5 37.1 5.7 5 651 1 286 655 3 225 486 0.509 0.504 0.733 0.605 1998 4.6 14.4 12.7 4.2 3 257 452 1 405 1 136 264 0.348 0.434 0.648 0.649 1999 7.4 8.9 19.4 20.1 4 709 717 853 1 789 1 349 0.631 0.596 0.771 0.898 2000 9.9 31.3 27.5 4.8 5 170 671 2 131 2 104 270 0.88 0.727 0.977 0.706 2001 16.7 39.5 34.5 15.8 8 648 1 318 3 127 3 273 1 366 0.693 0.489 0.714 0.742 2002 5.0 6.4 12.7 17.9 4 916 958 690 1 542 1 726 0.312 0.226 0.723 0.677 2003 22.0 7.3 24.4 2.0 5 484 2 265 716 2 292 212 0.333 0.403 0.852 0.622 2004 1.7 3.4 10.4 25.7 3 788 250 418 1 045 2 075 0.349 0.443 0.607 0.596 2005 4.3 6.0 54.0 7.9 7 024 320 661 5 096 958 0.754 0.471 0.716 0.506 2006 4.7 6.8 9.8 5.1 2 549 421 719 1 036 372 0.415 0.492 0.563 0.772 2007 12.1 20.8 23.6 0.3 5 899 991 2 039 2 086 784 0.416 0.345 0.567 0.124
Reducing uncertainty in stock status ABARES
References
Bishop, J, Venables, WN, Dichmont, CM & Sterling, DJ 2008, ‘Standardizing catch rates: is logbook information by itself enough?’, ICES Journal of Marine Science, vol. 65, no. 2, pp. 255–66.
Breen, PA & Kendrick, TH 1998‚ ‘An evaluation of surplus production analysis for assessing the fishery for New Zealand red rock lobsters (Jasus edwardsii)’, in Proceedings of the North Pacific Symposium on Invertebrate Stock Assessment and Management, edited by G. S. Jamieson and A. Campbell, Canadian Special Publication Fisheries and Aquatic Sciences, 125. pp. 213–23.
Davis, TLO & Ward, TJ 1984, ‘CSIRO finds two new scampi grounds off the North West Shelf’ Australian Fisheries, vol. 43 no. 8, pp. 41–5.
Evans, D 1992, ‘The Western Deep Water Trawl and North West Slope Trawl Fisheries’, pp. 19– 27, in The fisheries biology of deepwater crustacea and finfish on the continental slope of Western Australia, Rainer, S. F. (ed), Final Report FRDC Project 1988/74.
Fletcher, D, Mackenzie, D & Villouta, E 2005, ‘Modelling skewed data with many zeros: A simple approach combining ordinary and logistic regression’, Environmental and Ecological Statistics, vol. 12, pp. 45–54.
Fowler, J & McLoughlin, K (eds) 1996, North West Slope Trawl Fishery and Western Deepwater Trawl Fishery 1994, Fisheries Assessment Report compiled by the Fisheries Resource Assessment Group. Australian Fisheries Management Authority, Canberra.
Garvey, JR, Wadley, VA & Phillips, BF 1992, ‘Estimation of the Relative Abundance of Commercial Crustaceans from the North West Slope Trawl Fishery Using Commercial Catch and Effort Data’, pp. 131–55, in The fisheries biology of deepwater crustacea and finfish on the continental slope of Western Australia, Rainer, S. F. (ed), final report FRDC project 1988/74.
Jensen, A L 1986, ‘Assessment of the Maine lobster fishery with surplus production models’, North American Journal of Fisheries Management, vol. 6, pp. 63–68.
Lynch, AW & Garvey, JR 2005, North West Slope Trawl Fishery Scampi Stock Assessment 2004, Data Group, Australian Fisheries Management Authority, Canberra.
Maunder, MN & Punt, AE 2004, ‘Standardizing catch and effort data: a review of recent approaches’, Fisheries Research, vol. 70, no. 2–3, pp. 141–59.
Morgan, GR 1979, ‘Assessment of the stocks of the western rock lobster Panulirus cygnus using surplus yield models’, Australian Journal of Marine and Freshwater Research, vol. 30, pp. 355–63.
Polacheck, T, Hilborn, R & Punt, AE 1993, ‘Fitting Surplus Production Models: Comparing Methods and Measuring Uncertainty’, Canadian Journal of Fisheries and Aquatic Research, vol. 50, pp. 2597–607.
Prager, MH 1994, ‘A suite of extensions to a non-equilibrium surplus production model’, Fisheries Bulletin 92 (1994), pp. 374–89.
Rees, T 2003, ‘“C-Squares” a New Spatial Indexing System and its Applicability to the Description of Oceanographic Datasets’, Oceanography, vol. 16, no. 1, pp. 11–19.
146 Reducing uncertainty in stock status ABARES
Wallner, BG & Phillips, BF 1995, Development of a trawl fishery for deepwater metanephroid lobsters off the northwest continental slope of Australia: designing a management strategy compatible with species life history, CSIRO Division of Fisheries, North Beach, Western Australia.
Ward, TJ & Davis, TLO 1987, ‘Diel periodicity of Metanephrops australiensis and other deep- water crustaceans of northwest Australia’, Fisheries Research, vol. 5, no. 1, pp. 91–97.
Wayte, S, Dowdney, J, Williams A, Fuller, M, Bulman, C, Sporcic, M & Hobday, A 2007, Ecological Risk Assessment for the Effects of Fishing: Report for the North West Slope Trawl Fishery, report for the Australian Fisheries Management Authority, Canberra.
Wilson, DT, Curtotti R, Begg GA & Phillips, K (eds) 2009, Fishery status reports 2008: status of fish stocks and fisheries managed by the Australian Government, Bureau of Rural Sciences & Australian Bureau of Agricultural and Resource Economics, Canberra. Yoshimoto, SS & Clarke, RP 1993, ‘Comparing dynamic versions of the Schaefer and Fox production models and their application to lobster fisheries’, Canadian Journal of Fisheries and Aquatic Sciences, vol. 50, no. 1, pp. 181–89.
Zhou, S, Fuller, M & Smith, A 2009, Rapid quantitative risk assessment for fish species in seven Commonwealth fisheries April 2009, Australian Fisheries Management Authority.
147 Reducing uncertainty in stock status ABARES 8 Elephant fish catch rate standardisation and Tier 4 assessment, 2009
Veronica Boero and Kevin McLoughlin Summary
Elephant fish (Callorhinchus milii) are commonly caught as by-product in the Southern and Eastern Scalefish and Shark Fishery (SESSF). Before this study, no formal assessments of elephant fish had been adopted by the Southern Shark Fishery Resource Assessment Group (SharkRAG). The ABARES 2008 Fishery Status Reports listed the stock as uncertain to both overfished and overfishing status (Wilson et al. 2009). The purpose of this study was to assess the status of the elephant fish stock by undertaking a standardisation of catch rates for use within a Tier 4 analysis as described by the SESSF harvest strategy (AFMA 2009).
Elephant fish are not targeted by the commercial sector and logbook records contain a high number of zero catches. A delta model (also called a two-stage model) was used to derive an index of abundance from logbook catch and effort data. Following discussions at the SharkRAG in 2008 and 2009, several approaches to data selection were examined. A model that used data from all regions was adopted as the base case and the factor of vessel was the most important, followed by area, month and gear. The base case index of abundance shows an overall declining trend over the period 1980 to 2008 (from an index of 1 to around 0.7), with substantial variability from year to year.
The Tier 4 harvest strategy requires a reference period to be identified that represents desirable conditions of catch rate, catches and status of the fishery. Two candidate reference periods, 1980 to 1992 and 1998 to 2004, are examined. The 1980 to 1992 period was considered to be unreliable because of the likely underestimation of trawl catch and discarding and the variability in the catches and the index of abundance. The 1998 to 2004 period resulted in a catch target of 109.7 tonnes and recommended biological catch (RBC) of 122.8 tonnes. The current catch rate index is close to the target for this reference period, which would indicate the stock is not overfished. The current catch levels and increasing recent catch rate index trend suggests that the stock is not subject to overfishing.
The present Tier 4 analysis has some uncertainties because of a lack of data on catch and discards from the SESSF Commonwealth Trawl Sector (CTS) in the earlier years, unknown levels of discarding in the Gillnet, Hook and Trap sectors (GHT), and the unknown size of the recreational catch over time (Wilson et al. 2009). Caution is also advised given the lack of information on catches before the late 1970s and the current assumption that elephant fish were in a relatively unfished state at that time. Introduction
Elephant fish (Callorhinchus milii) are taken as by-product in the Southern and Eastern Scalefish and Shark Fishery (SESSF). The species has been mainly caught in the Shark Gillnet and Shark Hook Sectors (SGSHS) of the SESSF, which cover waters off South Australia, Tasmania and Victoria and include both Commonwealth and state waters (Map 8.1). The SGSHS are primarily managed through output controls in the form of total allowable catches with individual
148 Reducing uncertainty in stock status ABARES transferable quotas, as well as some input controls such as gear restrictions and closed areas (Wilson et al. 2009).
Map 8.1 Relative fishing intensity in the a) Shark Gillnet and b) Shark Hook Sectors of the Southern and Eastern Scalefish and Shark Fishery, 2008 a)
b)
Source: ABARES
149 Reducing uncertainty in stock status ABARES
The ABARES 2008 Fishery status reports (the most recent report at the start of this study) classified the stock as uncertain to both overfished and overfishing status (Wilson et al. 2009). At the start of this study, no formal assessments of sawshark had been adopted by the Southern Shark Fishery Resource Assessment Group (SharkRAG). Approach to status determination This study examined approaches to elephant fish catch rate standardisation and undertook a Tier 4 analysis. Under the SESSF harvest strategy (AFMA 2009), Tier 4 analyses are an approach used to derive recommended biological catches (RBCs) for species with only limited available information. Data required for a Tier 4 analysis include a time series of total catches and an index of abundance, usually derived from standardisation of commercial catch rates. The Tier 4 methods were revised in 2008 (Little et al. 2008) and the use of this approach for elephant fish has been discussed at meetings of the SharkRAG.
The assessment detailed in this chapter is divided into three sections: data preparation, catch rate standardisation and Tier 4 assessment. Data preparation
Catch data was restricted to the period from 1980 to 2008 as requested during the May 2009 SharkRAG meeting. Data before 1980 was considered inadequate because of the different fishing behaviour at that time. Data records which clearly showed the intention was to avoid catching elephant fish were excluded. Data entry errors and clear outliers were also excluded.
The SESSF area has been divided into a number of shark areas or regions which extend from the coast line to the 200 metre water depth mark and are 1 degree wide (Map 8.2). Areas 99 to 108, 112 to 115, 126 to 140, 148,149 and 201 were excluded as they hold less than 5 per cent of the observations with positive elephant fish catch. Including or excluding data from these areas makes little difference to the elephant fish CPUE series.
Data from gillnets with a mesh size of 6”, 6.5” or 7” were analysed. Some shots were undertaken with two nets, resulting in two records for effort in the database. Observations with more than one record for effort were excluded as well as effort records taken with a net length of less than 1 000 metres. The latter were not considered to be representative of standard fishing practices.
Between 1980 and 2008 there were 421 distinct vessels in the fishery. For these vessels elephant fish was a low or nonexistent component of their catch. To minimise data ‘noise’, vessels that caught less than 10 per cent of elephant fish per shot were excluded from the analysis. This resulted in the exclusion of records from 147 vessels.
The subsetting described in this section resulted in a dataset of 138 018 fishing operations (observations).
150 Reducing uncertainty in stock status ABARES
Map 8.2 Shark areas within the Southern and Eastern Scalefish and Shark Fishery
Source: AFMA Catch per unit effort standardisation
The CPUE standardisation is based on gillnet logbook catch and effort data. Elephant fish are not targeted by the gillnet sector, so logbook records contain a high number of zero catches. A delta model (also called a two-stage model) is appropriate under these circumstances and was used to derive an index of abundance from logbook catch and effort data. This model calculates the probability of an observation being zero and then calculates the CPUE for the non-zero observations. The two parts are combined to produce the index.
First, a logistic model was fitted to the indicator variable presence/absence of elephant fish, using year, month, area, average depth and gear (mesh size) as explanatory variables. Average depth was divided in three levels: 0 to 19 metres, 20 to 80 metres and greater than 80 metres. Predictions using all combinations of model parameter levels were calculated and back transformed and the year average of the predictions was calculated. Those averages are the standardised probability of a positive elephant fish catch. The model fitted was:
p log iljdk year area month depth gear 1 p 11111 i l j d k iljdk iljdk where piljdk is the probability of catching elephant fish in year i, area l, month j, depth d and with gear k. The intercept μ11111 is the parameter for i = 1 corresponding to year 1980, area l = 1 corresponding to area 4, month j = 1 corresponding to January, depth d = 1 corresponding to depth between 0 and 19 metres and for k = 1 corresponding to gillnet 6”; yeari is departure from the intercept for being in year i > 1, areal is the departure from the intercept for being in area l > 1, monthj is the departure from the intercept for being on month j >1, depthd is the departure from the intercept for being at depth d > 1 and geark is the departure for gear k > 1.
In the second stage a linear mixed model was fitted to the non-zero log-transformed elephant fish catches including year, month, area, average depth and gear as fixed effects explanatory
151 Reducing uncertainty in stock status ABARES variables and vessel as a random effect. The standardisation used the same set of factors as the standardisation of the probability of a positive catch. The model fitted for observations with positive elephant fish catch was:
log(CPUE ) year area month depth gear iljdkv v1111 vi l vj vd vk v ijdkv where log(CPUEijdkv) is log-transformed CPUE for a shot done by vessel v in year i, in month j, at depth d and with gear k; βv1111 is the parameter for i = 1corresponding to year 1980, area l = 1 corresponding to area 4, month j = 1 corresponding to January, depth d = 1 corresponding to depth between 0 and 19 metres and for k = 1 corresponding to gillnet 6”; yeari is departure from the intercept for being in year i > 1, areal is the departure from the intercept for being in area l > 1, monthj is the departure from the intercept for being on month j >1, depthd is the departure from the intercept for being at depth d > 1 and geark is the departure for gear k > 1; γv is the between vessel variability (vessel random effect) assumed to be normally distributed with mean zero and variance ψ2 independently distributed across vessels; and the within vessel errors εijdkv are assumed to also be normally distributed with mean zero and variance σ2, independent from vessel to vessel and independent of γv.
The standardised probability of a positive catch and the standardised CPUE from the non-zero observations are multiplied to obtain the final standardised CPUE. The standardised CPUE and the raw CPUE are compared in Figure 8.1 and Table F1.
Results from the analyses of variance indicate the vessel factor accounted for the highest proportion of variability in the model, followed by area, month and gear. Depth was not significant.
The standardised elephant fish CPUE series (Figure 8.1) has several peaks and troughs over the period 1980 to 2008. Despite a decline of approximately 50 per cent of the 1980’s standardised CPUE in 2004, the overall decline in CPUE has been approximately 20 per cent over the 1980 to 2008 period. If the CPUE series is considered indicative of abundance then this decline suggests that abundance has fallen by approximately 20 per cent over the 1980 to 2008 period. However, it should be noted that the stock was subject to exploitation before 1980. Historic records indicate catches of around 70 tonnes per year from 1976 to 1979 and it is likely that there was also some level of trawl catch before this time. Despite the catches, SharkRAG have indicated that elephant fish have never been a target species that it is unlikely the species was significantly depleted by 1980. The year 1980 was used as a reference for the standardisation.
152 Reducing uncertainty in stock status ABARES
Figure 8.1 Elephant fish catch index, and raw and standardised catch per unit effort using the basic subset data (1980 to 2008)
Source: ABARES
Standardisation for Bass Strait At the request of SharkRAG, a standardisation was undertaken on catch and effort data from Bass Strait only (Figure 8.1, area codes 6 to 35) using the same specification as described previoulsy in this chapter. The index for Bass Strait is shown in Figure 43 and the trends are similar to those seen in the ‘all areas’ analysis (Figure 8.1). The ‘all regions’ index was selected as the base case because it is desirable to include as much data as possible and both series have similar trends in any case.
153 Reducing uncertainty in stock status ABARES
Figure 8.2 Elephant fish catch index, and raw and standardised catch per unit effort series for the Bass Strait only (1980 to 2008)
Source: ABARES Tier 4 assessment
A full description of the SESSF harvest strategy, including the functioning of Tier 4, is provided by AFMA (2009). Elephant fish is designated a Tier 4 species under the SESSF harvest strategy due to limited information on the species’ current biomass or current exploitation rate (AFMA 2009). As well as an index of abundance, Tier 4 requires a time series of catches that account for all sources of mortality.
SESSF catches were extracted from AFMA fishery logbook data prior to 2002 and landings data after 2002. Catches from the gillnet, longline and trawl sectors of the SESSF are presented in the Tables F2, F3 and F4.
Estimates of discarding mortality from the trawl sectors were based on the analysis of Walker and Gason (2007). The level of shark discarding from the Gillnet, Hook and Trap sectors (GHT) has not been well estimated. A base case level of 10 per cent is assumed, though Braccini et al. (2009) suggests levels as high 50 per cent based on survey information. Estimates of elephant fish caught in state managed fisheries were taken from Pribac (2007) up to 2006 and then assumed to be constant at that level. Since the mid 1990s recreational fishing was assumed to have taken a fixed 29 tonnes of elephant fish annually. Table F4 contains the complete catch and abundance index data used in the Tier 4 assessment for elephant fish.
The Tier 4 assessment requires use of data from a ‘reference’ period, a historical period identified as desirable in terms of CPUE, catches and status of the fishery. SharkRAG experienced difficulty in selecting a suitable reference period and two candidate periods were chosen: 1980 to 1992 and 1998 to 2004.
154 Reducing uncertainty in stock status ABARES
The target catch levels and the historic catches used in the analysis are shown in Figure 8.3 and Table F4. Total elephant fish catches between 1976 and 2008 range from 46 tonnes to 129 tonnes. During the 1980 to 1992 reference period (target catch 82.8 tonnes) catches fluctuated from 65 tonnes in 1982 to 121 tonnes in 1985. During the 1998 to 2004 reference period (target catch 109.7 tonnes) catches remained relatively stable ranging from 95 tonnes in 2001 to 133 tonnes in 2003. A major factor in the differences between the periods is the higher catches in later years from the trawl and recreational sectors. The level of trawl catch and discarding is likely to have been underestimated in the earlier period.
Applying the Tier 4 with a 1980 to 1992 reference period results in a catch target of 82.8 tonnes and a recommended biological catch (RBC) for 2010 of 52.8 tonnes (Table 8.1 Elephant fish 2009 Tier 4 methodology results (all areas) for two candidate reference periods). However, the likely underestimation of trawl catch and discarding and the variability in the catches and CPUE are a potential problems with the 1980 to 1992 reference period. The 1998 to 2004 period results in a catch target of 109.7 tonnes and an RBC of 122.8 tonnes (Table 8.1 Elephant fish 2009 Tier 4 methodology results (all areas) for two candidate reference periods). The level of discarding from the gillnet and longline sectors is also poorly known. These results are based on an assumed annual level of 10 per cent. If discards for the gillnet and longline sectors are assumed to be 20 per cent, the catch target for the 1998 to 2004 period becomes 114.3 tonnes and the RBC 127.9 tonnes (however, a higher level of discards would also be subtracted to provide a similar level of RBC available for allocation). Conclusion Results from this analysis suggest that the current CPUE is close to the target CPUE from the 1998 to 2004 reference period, which indicates that the species is not overfished (Table 8.1). The current catch and increasing CPUE trend suggests that the stock is not subject to overfishing. However, results should be treated with caution due to the different assumptions regarding historic catch. The present Tier 4 analysis has some uncertainties because of a lack of data on catch and discards from the trawl sector in the earlier years, unknown levels of discarding in the GHT sectors, and the size of the recreational catch over time (Wilson et al. 2009). Table 8.1 Elephant fish 2009 Tier 4 methodology results (all areas) for two candidate reference periods
2009 Tier 4 results Reference period 1980 to 1992 1998 to 2004
CPUE target (CPUEtarg 0.898 0.656
CPUE limit (CPUElim) 0.359 0.262 0.703 0.703 Recent CPUE average ( CPUE ) No. years in average (m) 4 4 Catch target (tonnes) (C*) 82.76 109.72 Scaling factor 0.638 1.12 2010 RBC (tonnes) 52.80 122.81 State catch (tonnes) 9.81 9.81 Recreational (tonnes) 29.0 29.0 Discards (tonnes) 7.28 7.28 RBC – discards, recreational, State 6.70 76.71 (tonnes) Note: CPUE = catch per unit effort (index of abundance); RBC = recommended biological catch.
155 Reducing uncertainty in stock status ABARES
Figure 8.3 Elephant fish catches (1976–2008) and reference period catch target levels
140
120 GHAT Total catch 100
C* 1980 to 1992
80 C* 1998 to 2004
Catch (t)Catch 60
40
20
0 1970 1980 1990 2000 2010
Note: C* 1980 to 1992 refers to target catch if a reference period of 1980 to 1992 is used and C* 1998 to 2004 refers to target catch if a reference period of 1998 to 2004 is used. Source: ABARES
156 Reducing uncertainty in stock status ABARES
Appendix F Table F1 Elephant fish catch per unit effort and catch indexes
Probability of Standardised CPUE Standardised CPUE Catch index Raw CPUE index Year catching elephant index referenced to index referenced to 1980 referenced to 1980 fish 1980 1980 0.3222 0.0013 1.0000 1.0000 1.0000 1981 0.3381 0.0015 1.1993 0.9776 0.9345 1982 0.3219 0.0012 0.9650 0.7109 0.6601 1983 0.2831 0.0011 0.8480 1.0356 0.7640 1984 0.2510 0.0010 0.8024 0.7894 0.5386 1985 0.2464 0.0011 0.8366 0.8385 0.5352 1986 0.1887 0.0009 0.7008 1.0007 0.4368 1987 0.1597 0.0008 0.6519 1.1082 0.4421 1988 0.1857 0.0010 0.7997 0.8041 0.3295 1989 0.2138 0.0011 0.8625 0.8627 0.3798 1990 0.1887 0.0011 0.8838 0.9818 0.3769 1991 0.2170 0.0014 1.1159 1.4224 0.5968 1992 0.2081 0.0013 1.0025 1.5345 0.5617 1993 0.1730 0.0007 0.5492 0.8447 0.2792 1994 0.1759 0.0009 0.7323 0.8344 0.2505 1995 0.2508 0.0012 0.9248 1.0457 0.4479 1996 0.2752 0.0012 0.9329 1.2362 0.6140 1997 0.3038 0.0008 0.6592 0.7314 0.3630 1998 0.3115 0.0007 0.5822 0.6219 0.3222 1999 0.3713 0.0009 0.7235 0.5911 0.3386 2000 0.4123 0.0009 0.7348 0.4694 0.3302 2001 0.4431 0.0010 0.7980 0.4426 0.3639 2002 0.3755 0.0009 0.6872 0.5081 0.3627 2003 0.3233 0.0007 0.5848 0.6050 0.3922 2004 0.3012 0.0006 0.4792 0.4706 0.2792 2005 0.3697 0.0008 0.5871 0.4374 0.3476 2006 0.3755 0.0009 0.6781 0.4564 0.3805 2007 0.4106 0.0010 0.7546 0.4676 0.4415 2008 0.4266 0.0010 0.7906 0.4381 0.4224
157 Reducing uncertainty in stock status ABARES
Table F2 Elephant fish gillnet and longline catch (t) by Southern and Eastern Scalefish and Shark Fishery shark sub-regions
Year WSA CSA SAV WBS EBS WTAS ETAS NSW Unknown/other Total 1970 0 0 0 0.884 5.166 0.476 0.007 0 0 6.533 1971 0 0 0 1.155 2.852 0 0.034 0 0 4.041 1972 0 0 0 2.333 8.686 0 0.041 0 0 11.06 1973 0 0 0 7.715 22.253 0.024 1.324 0 0 31.316 1974 0.029 0.17 0.931 4.651 50.927 0.135 8.551 0 0 65.394 1975 0 0 0 5.87 56.79 0 6.894 0 0 69.554 1976 0 0 0.03 10.903 28.981 0.452 1.822 0 0 42.188 1977 0 0 2.087 10.219 54.2 0 1.828 0 0 68.334 1978 0 0 0.029 20.587 36.249 0 8.71 0 0 65.575 1979 0 0 0.339 12.968 80.013 0 7.261 0 0 100.581 1980 0 0 0.215 20.337 54.688 0.217 6.826 0 0 82.283 1981 0 0.003 0.357 24.093 52.571 0.155 4.886 0 0 82.065 1982 0 0 0.254 24.494 28.959 0.615 4.341 0 0 58.663 1983 0 0 0.356 27.246 47.466 0.244 5.166 0 0 80.478 1984 0 0.003 0.073 16.601 45.888 3.627 12.003 0 0 78.195 1985 0.148 0.182 0.053 19.479 42.485 3.642 42.998 0 0 108.987 1986 0 0 0.052 11.962 33.639 2.365 17.35 0 0 65.368 1987 0 0.325 0.263 17.128 24.393 0.742 20.512 0 0 63.363 1988 0 0.442 0.623 16.882 21.23 1.046 26.877 0 0 67.1 1989 0 0.065 0.08 11.178 22.757 0.478 27.551 0 0 62.109 1990 0 0.3 0.058 13.618 13.599 1.194 27.023 0 0 55.792 1991 0.022 0.025 0.027 13.689 39.226 0.093 16.118 0 0 69.2 1992 0 0.116 0.371 14.543 26.426 5.894 23.721 0 0 71.071 1993 0 0.007 0.025 15.537 12.642 4.37 21.754 0 0 54.335 1994 0 0.057 0.031 10.551 12.739 1.859 34.265 0 0 59.502 1995 0 1.867 0.906 21.388 12.001 1.589 14.085 0 0 51.836 1996 0 1.267 0.718 24.274 21.291 1.794 27.767 0 0 77.111 1997 0 2.306 3.072 15.511 17.999 0.797 20.172 0 0 59.857 1998 0.012 2.264 0.409 15.443 20.58 1.761 12.363 0 0 52.832 1999 0.008 4.501 1.267 10.825 30.737 0.48 11.381 0 0 59.199 2000 0.148 3.133 0.509 8.417 25.942 0.655 15.084 0 0 53.888 2001 0.047 6.597 0.833 3.289 29.32 1.242 6.002 0 0 47.33 2002 0 2.086 0.519 6.654 24.328 0.084 6.899 0 0 40.57 2003 0.115 3.905 0.627 6.168 30.463 1.465 6.09 0 0 48.833 2004 0.152 1.689 0.83 4.588 20.913 0.661 6.732 0.02 0 35.585 2005 0.173 2.041 0.149 6.998 20.896 0.463 5.568 0.013 0 36.301 2006 0.856 1.498 0.085 3.227 21.521 1.275 4.824 0 0 33.286 2007 0.332 2.492 0.121 2.558 20.269 0.368 8.587 0.04 0 34.767 2008 0.158 2.155 0.163 2.384 20.126 0.207 4.174 0 0 29.367 Note: WSA: Western South Australia; WBS: Western Bass Strait; ETAS: Eastern Tasmania; CSA: Central South Australia; EBS: Eastern Bass Strait; SAV: South Australia – Victoria; WTAS: Western Tasmania.
158 Reducing uncertainty in stock status ABARES
Table F3 Elephant fish gillnet and longline catch (t) by Southern and Eastern Scalefish and Shark Fisheryshark gear type
Year Longline Unknown mesh 6.5" mesh 6" mesh 7" mesh 8" mesh Unknown/other Total 1970 2.942 3.108 0 0 0 0 0.483 6.533 1971 0.148 3.761 0 0 0 0 0.132 4.041 1972 1.143 6.593 0 0 0 0 3.324 11.06 1973 0.425 0.399 0 2.16 26.108 1.738 0.486 31.316 1974 0.932 4.368 0 20.045 10.834 0.17 29.045 65.394 1975 2.361 7.206 0 38.34 2.861 0 18.786 69.554 1976 1.163 4.623 0 28.692 2.928 0.03 4.752 42.188 1977 0 8.599 0 51.979 0.003 0 7.753 68.334 1978 0 16.712 0.13 46.966 0.005 0 1.762 65.575 1979 0 23.797 0.285 75.706 0.793 0 0 100.581 1980 0.002 24.897 3.998 52.32 1.066 0 0 82.283 1981 0 20.179 0.341 60.991 0.554 0 0 82.065 1982 0.064 15.461 0.1 43.014 0.024 0 0 58.663 1983 0.02 19.95 0.02 60.016 0.064 0 0.408 80.478 1984 0.236 39.972 0.486 37.498 0.003 0 0 78.195 1985 0.565 72.741 0.026 35.44 0.215 0 0 108.987 1986 2.589 33.395 0.059 28.795 0 0 0.53 65.368 1987 0.069 33.299 0 29.449 0.277 0.269 0 63.363 1988 7.735 30.75 0 26.89 1.71 0 0.015 67.1 1989 1.022 36.314 0 23.857 0.107 0.025 0.784 62.109 1990 2.295 31.905 0.181 20.856 0.31 0 0.245 55.792 1991 1.38 21.913 0 45.698 0.025 0 0.184 69.2 1992 6.899 25.674 0.07 37.512 0.893 0.001 0.022 71.071 1993 3.762 26.413 0 19.896 0.911 2.41 0.943 54.335 1994 0.504 38.013 0.021 19.55 0.156 0.46 0.798 59.502 1995 0.291 0 2.395 40.482 8.576 0 0.092 51.836 1996 0.437 0 1.046 58.894 14.525 0 2.209 77.111 1997 0.069 0.5 4.437 42.999 8.944 0 2.908 59.857 1998 1.347 0 2.303 46.809 0 0 2.373 52.832 1999 0.435 0.498 4.557 48.278 0.041 0 5.39 59.199 2000 0.033 0.12 4.204 46.473 0 0 3.058 53.888 2001 0.053 0.122 5.917 41.007 0 0 0.231 47.33 2002 0.122 0.089 2.069 38.28 0 0 0.01 40.57 2003 0.096 0 2.95 44.924 0 0 0.863 48.833 2004 0.525 0 2.265 29.575 0 0 3.22 35.585 2005 0 0 1.674 32.562 0 0 2.065 36.301 2006 0.003 0 3.345 28.859 0 0 1.079 33.286 2007 0.037 0 4.893 29.564 0 0 0.273 34.767 2008 0.007 0 2.457 26.891 0 0.01 0.002 29.367
159 Reducing uncertainty in stock status ABARES
Table F4 Elephant fish catches (t) and standardised catch per unit effort used in Tier 4 estimation
Year GHAT SET GAB State Recreational Discards Discards Total Std (SSF) (SET & GAB) catch CPUE 1976 42.188 – – – – 4.219 – 46.407 – 1977 68.334 – – – – 6.833 – 75.167 – 1978 65.575 – – – – 6.558 – 72.133 – 1979 100.581 – – – – 10.058 – 110.639 – 1980 82.283 – – – – 8.228 – 90.511 1 1981 82.065 – – – – 8.207 – 90.272 1.199 1982 58.663 – – – – 5.866 – 64.529 0.966 1983 80.478 – – – – 8.048 – 88.526 0.848 1984 78.195 – – – – 7.820 – 86.015 0.802 1985 108.987 0.911 – – – 10.899 – 120.797 0.837 1986 65.368 5.154 – – – 6.537 – 77.059 0.701 1987 63.363 1.846 – – – 6.336 – 71.545 0.652 1988 67.100 12.200 0.100 – – 6.710 – 86.110 0.800 1989 62.109 3.207 0.144 – – 6.211 – 71.671 0.862 1990 55.792 1.892 0.045 – – 5.579 – 63.308 0.884 1991 69.200 5.385 0.032 – – 6.920 – 81.537 1.116 1992 71.071 5.698 0.060 – – 7.107 – 83.936 1.002 1993 54.335 2.725 0.000 – – 5.434 – 62.494 0.549 1994 59.502 3.987 0.710 – – 5.950 – 70.149 0.732 1995 51.836 2.819 0.039 – – 5.184 – 59.878 0.925 1996 77.111 5.410 0.275 – 29.000 7.711 4.813 124.320 0.933 1997 59.857 5.598 0.095 – 29.000 5.986 3.587 104.123 0.659 1998 52.832 7.900 0.070 – 29.000 5.283 3.256 98.341 0.582 1999 59.199 7.460 0.965 15.000 29.000 5.920 3.528 121.071 0.724 2000 53.888 8.913 0.000 4.000 29.000 5.389 2.800 103.990 0.735 2001 47.330 8.444 0.106 3.000 29.000 4.733 2.800 95.413 0.798 2002 24.659 29.658 0.374 12.261 29.000 2.466 2.800 101.218 0.687 2003 48.794 28.960 2.384 16.129 29.000 4.879 2.800 132.946 0.585 2004 35.225 32.928 1.700 9.905 29.000 3.523 2.800 115.081 0.479 2005 40.523 40.204 1.855 10.474 29.000 4.052 2.800 128.908 0.587 2006 40.143 26.054 2.329 9.592 29.000 4.014 2.800 113.932 0.678 2007 42.713 19.090 1.415 9.592 29.000 4.271 2.800 108.881 0.755 2008 55.970 24.693 0.680 9.592 29.000 5.597 2.800 128.332 0.791 Note: Catches for GHAT, SET and GAB are taken from logbooks before 2002 so they are from landings data. Discards for the gillnet and longline sectors (SSF) presented here are an assumed 10 per cent of total catch (based on Braccini et al. 2009; T Walker, pers. comm., 2010). Other discards, for the SET and GAB, are based on Walker and Gason (2007) since 2000.
160 Reducing uncertainty in stock status ABARES
References
AFMA 2009, Harvest Strategy - for the Southern and Easter Scalefish and Shark Fishery, Australian Government Department of Agriculture, Fisheries and Forestry, Commonwealth of Australia, Canberra.
Braccini, JM, Walker, TI & Gason, AS 2009, GHATF shark survey of population abundance and population size composition for target, by-product and bycatch species, R2006/823, final report to the Australian Fisheries Management Authority, Canberra.
Little, R, Tuck, G, Haddon, M, Day, J, Klaer, N, Smith, T, Thomson, R & Wayte, S 2008, Developing CPUE targets for the Tier 4 harvest strategy of the SESSF, report to the Shelf Resource Assessment Group.
Pribac, F 2007, Tier 4 Harvest Control Rule applied to Elephant Fish and Saw Shark 2007, report to the Shark Resource Assessment Group.
Walker, TI & Gason, AS 2007, Shark and other chondrichthyan by-product and bycatch estimation in the Southern and Eastern Scalefish and Shark Fishery, project No. 2001/007, report to the Fisheries Research and Development Corporation, Canberra.
Wilson, DT, Curtotti, R & Begg, GA 2009, Fishery Status Reports 2009: status of fish stocks and fisheries managed by the Australian Government, Australian Bureau of Agricultural and Resource Economics – Bureau of Rural Sciences, Canberra.
161 Reducing uncertainty in stock status ABARES 9 Sawshark catch rate standardisation and Tier 4 assessment, 2009
Veronica Boero and Kevin McLoughlin Summary
Sawshark (Pristiophorus cirratus and P. nudipinnis) are commonly caught as by-product in the Southern and Eastern Scalefish and Shark Fishery (SESSF). At the start of this study, no formal assessments of sawshark had been adopted by the Southern Shark Fishery Resource Assessment Group (SharkRAG). The ABARES 2008 Fishery status reports listed the stock as uncertain to both overfished and overfishing status (Wilson et al. 2009). The purpose of this study was to assess the status of the sawshark stock by undertaking a standardisation of catch rates for use within a Tier 4 analysis as described by the SESSF harvest strategy (AFMA 2009).
Common and southern sawshark are not differentiated in historical catch data and the two species were treated as a combined stock. Sawshark are not targeted by the commercial sector and logbook records contain a high number of zero catches. A delta model (also called a two- stage model) was used to derive an index of abundance from logbook catch and effort data. Analysis was restricted to 1980–91 and 1997–2008 because data before 1980 and during 1992– 96 was considered inadequate due to changes in fishing and/or reporting behaviour for sawshark. A model that used data from all regions was adopted as the base case and the factor of area was the most important, followed by vessel, month, depth and gear. The base case index of abundance shows a continuous declining trend over the period 1980–2008 (from an index of 2 to around 1), noting the missing years during 1992–96. A number of alternate indexes were developed from different data subsets and these showed similar overall trends to the base case.
The Tier 4 harvest strategy requires a reference period to be identified that represents desirable conditions of catch rate, catches and status of the fishery. There was no clear time period with stable catch rates and catch so two candidate reference periods, 1986–2001, and 2002–08, were examined. The 2009 catch rate index of abundance was below the target but above the limit under both the candidate reference periods. The 1986–2001 period resulted in a catch target of 301.1 tonnes and a recommended biological catch (RBC) for 2010 of 202.5 tonnes whereas the 2002–08 period resulted in a catch target of 391.4 tonnes and an RBC for 2010 of 369.6 tonnes.
Lack of data from the Commonwealth Trawl Sector, unknown discarding rates by different sectors and particularly the lack of species-specific catch data make results from this assessment uncertain and the status for sawshark was not determined. Introduction
Common sawshark (Pristiophorus cirratus) and southern sawshark (P. nudipinnis) are a by-product of the Southern and Eastern Scalefish and Shark Fishery (SESSF). At the time of this assessment the species were mainly caught in the Shark Gillnet and Shark Hook sectors (SGSHS) which covers waters off South Australia, Tasmania and Victoria and include both Commonwealth and state waters (Map 9.1). The SGSHS are primarily managed through output controls in the form of total allowable catches with individual transferable quotas, as well as some input controls such as gear restrictions and closed areas.
162 Reducing uncertainty in stock status ABARES
Map 9.1 Relative fishing intensity in the a) Shark Gillnet and b) Shark Hook sectors of the Southern and Eastern Scalefish and Shark Fishery, 2008 a)
b)
Source: ABARES
The Shark Gillnet and Shark Hook sectors are primarily managed through input controls such as gear restrictions and closed areas. Output controls involve individual transferable quotas.
163 Reducing uncertainty in stock status ABARES
The Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) 2008 Fishery status reports (the most recent report at the start of this study) classified the stock as uncertain to both overfished and overfishing status (Wilson et al. 2009). At the start of this study, no formal assessments of sawshark had been adopted by the Southern Shark Fishery Resource Assessment Group (SharkRAG). Approach to Status Determination This study examined approaches to sawshark catch rate standardisation and undertook a Tier 4 analysis. Common sawshark and southern sawshark were combined due to a lack of historical differentiation of these two species in catch records. Under the SESSF harvest strategy (AFMA 2009), Tier 4 harvest control rules are the default approach used to derive recommended biological catches (RBCs) for species with only limited available information. Data required for a Tier 4 analysis include a time series of total catches and an index of abundance, usually derived from standardisation of commercial catch rates. The Tier 4 methods were revised in 2008 (Little et al. 2008) and the use of this approach for sawshark has been discussed at recent meetings of the Shark Resource Assessment Group (SharkRAG).
The assessment detailed in this chapter is divided into 3 sections: data preparation, catch rate standardisation, and Tier 4 assessment. Data preparation
Catch data was restricted to the periods 1980 to 1991 and 1997 to 2008, as recommended during the 2009 May SharkRAG meeting. Data before 1980 was considered inadequate due to the different fishing behaviour at that time. Data between 1992 and 1996 was considered inadequate due to changes in fishing and/or reporting behaviour before the introduction of quotas for sawshark.
The SESSF area has been divided into a number of shark areas or regions which extent from the coast line to the 200 metre water depth mark and are 1 degree wide (Map 9.2). Data from areas 108, 122, 129, 132, 136, 140 and 201 (Map 9.2) were excluded as they hold less than 5 per cent of the observations with positive sawshark catch. Including or excluding data from these areas made little difference to the sawshark CPUE series.
Data from gillnets with a mesh of 6”, 6.5” or 7” were analysed. Some shots were undertaken with two nets, resulting in two records for effort in the database. Observations with more than one record for effort were excluded as well as effort records taken with a net length of less than 1000 metres. The latter were not considered to be representative of standard fishing practices.
Between 1980 and 2008 there were 421 distinct vessels in the fishery. For these vessels sawshark was a low or nonexistent component of their catch. In order to minimise noise, vessels that caught less than 10 per cent of elephant fish per shot were excluded from the analysis. This resulted in the exclusion of records from 71 vessels. In May 2009, SharkRAG suggested the analysis of data from vessels that have fished sawshark for a minimum of three years however this was not pursued because this resulted in a loss of the majority of observations.
The subsetting described in this section resulted in a data set of 174 149 fishing operations (observations) that formed the basic data subset.
164 Reducing uncertainty in stock status ABARES
Map 9.2 Shark areas within the Southern and Eastern Scalefish and Shark Fishery
Source: AFMA Catch per unit effort standardisation
A number of standardised catch rate time series were developed including a base case index using comprehensive data (from the basic data subset) and a Bass Strait only index restricted to that region (zones 6 to 12, 18 to 23 and 30 to 35) (Map 9.2). Tests were also conducted using alternative models or inclusion of additional factors and these are described under the relevant headings below.
Sawsharks are not targeted by the gillnet sector, so logbook records contain a high number of zero catches. A Delta model (also called a two-stage model) is appropriate under these circumstances and was used to derive an index of abundance from logbook catch and effort data. This model calculates the probability of an observation being zero and then calculates the CPUE for the non-zero observations. The two parts are combined to produce the index. Base case (basic data subset) In the first stage, a logistic model was fitted to the indicator variable presence/absence of sawshark using year, month, area, average depth and gear (mesh size) as explanatory variables. Average depth was divided into three levels: 0 to 19 metres, 20 to 80 metres and greater than 80 metres. Predictions using all combinations of model parameter levels were calculated and back transformed and the year average of the predictions was calculated. Those averages are the standardised probability of a positive sawshark catch. The model fitted was:
p log iljdk year area month depth gear 1 p 11111 i l j d k iljdk iljdk were piljdk is the probability of catching sawshark in year i, area l, month j, depth d and with gear k. The intercept μ11111 is the parameter for i = 1corresponding to year 1980, area l = 1 corresponding to area 4, month j = 1 corresponding to January, depth d = 1 corresponding to depth between 0 and 19 metres and for k = 1 corresponding to gillnet 6”; yeari is departure from the intercept for being in year i > 1, areal is the departure from the intercept for being in
165 Reducing uncertainty in stock status ABARES
area l > 1, monthj is the departure from the intercept for being on month j >1, depthd is the departure from the intercept for being at depth d > 1 and geark is the departure for gear k > 1.
In the second stage a linear mixed model was fitted to the non-zero log-transformed sawshark catches including year, month, area, average depth and gear as fixed effects explanatory variables and vessel as a random effect. The standardisation used the same set of factors as the standardisation of the probability of a positive catch. The model fitted for observations with positive sawshark catch was:
log(CPUE ) year area month depth gear iljdkv v1111 vi l vj vd vk v ijdkv where log(CPUEijdkv) is log-transformed CPUE for a shot by vessel v in year i, in month j, at depth d and with gear k; βv1111 is the parameter for i = 1corresponding to year 1980, area l = 1 corresponding to area 4, month j = 1 corresponding to January, depth d = 1 corresponding to depth between 0 and 19 metres and for k = 1 corresponding to gillnet 6”; yeari is departure from the intercept for being in year i > 1, areal is the departure from the intercept for being in area l > 1, monthj is the departure from the intercept for being on month j >1, depthd is the departure from the intercept for being at depth d > 1 and geark is the departure for gear k > 1; γv is the between vessel variability (vessel random effect) assumed to be normally distributed with mean zero and variance ψ2 independently distributed across vessels; and the within vessel errors εijdkv are assumed to also be normally distributed with mean zero and variance σ2, independent from vessel to vessel and independent of γv.
Figure 9.1 plots the probability of a sawshark catch over time resulting from the base case standardisation. There is a slight increasing trend in the probability of a sawshark catch over time. The reasons for this are not clear but may be because of changes in discarding rates over time or increased targeting of gummy shark.
Figure 9.1 Probability of catching sawshark (base case, 1980–2008)
Probability
Source: ABARES
Results from the analyses of variance indicated the variable ‘area’ is responsible for the highest proportion of variability in the model, followed by ‘vessel’, ‘month’, ‘depth’ and ‘gear’.
166 Reducing uncertainty in stock status ABARES
The standardised probability of a positive catch (Figure 9.2) and the standardised CPUE from the non-zero observations are multiplied to obtain the final standardised CPUE. The year 2002 was used as a reference year (set to 1) for comparison of with other years and plotting.
The base case sawshark catch rate index shows a large decline in the early years, but it has been relatively stable at around 50 per cent of the 1980 levels since 2002 (Figure 9.2 and Table G1).
Figure 9.2 Sawshark base case standardised catch per unit effort index, catch index and raw catch per unit effort index using the basic subset data (1980–2008)
Source: ABARES
Bass Strait A separate standardisation was undertaken for the Bass Strait area (codes from 6 to 35 in Map 9.2) following the same methodology as for the base case. Results showed similar trends to the base case (Figure 9.3, Bass Strait original). Following presentation to the May 2009 SharkRAG meeting, it was suggested that the interaction between month and area factors may be very important. A model considering interactions between month and area could not be applied to the full dataset for all areas due to lack of computer memory. However, as the standardised CPUE series for Bass Strait was quite similar to the series for all the regions, an additional model (Bass Strait alternative) using interactions was applied to the Bass Strait data. A two-stage model was fitted, such that:
p log iljd year area month depth area *month 1111 i l j d l j iljd 1 piljd
log(CPUEiljdkv) v1111 yearvi areal monthvj depthvd areal *monthj v ijdkv Note the gear effect was removed and an interaction between area and month was added.
167 Reducing uncertainty in stock status ABARES
Figure 9.3 provides a comparison of the various models. There is very little difference between the three series, especially since 1997.
Figure 9.3 Comparison of sawshark standardised catch per unit effort series (1980–2008)
Source: ABARES
Additional catch per unit effort standardisations Following suggestions from the March 2009 SharkRAG meeting number of additional standardisations were undertaken to examine sensitivity to alternative models. Some operators indicating that large catches of sawshark occur when gummy shark catches are low and so a sensitivity analysis was undertaken where gummy shark was included as an explanatory variable. Standardisations were also undertaken for observations where gummy and school sharks catch was small. Two levels of catch were used, 100 and 500 kilograms for each species. School shark was also included in this sub-setting because similarly to gummy shark it has been a targeted species throughout most of the time series.
The catch rate indexes resulting from these additional models were found to be very similar to the series presented in the base case and for Bass Strait Figure 9.3, and so the index was not sensitive to these alternative data or model formulations. Tier 4 assessment
A full description of the SESSF harvest strategy, including the functioning of Tier 4, is provided by AFMA (2009).
The Tier 4 approach requires consideration of all sources of sawshark fishing mortality. Catches from the gillnet and longline fishery are presented in Table G2 and G3. Other sources of mortality include trawl fishing in the south east region and in the Great Australian Bight (Table G4). Catches from these sectors have been extracted from AFMA’s fishery logbook and landings data (landings data was used from 2002 as it is considered more accurate). Information on the level of mortality associated with discarded sawshark from the trawl fisheries was based on the
168 Reducing uncertainty in stock status ABARES analysis of Walker and Gason (2007). Discarding from the Southern Shark Fishery is not well understood and a level of 6% of recorded catch was assumed based on observations from surveys (Braccini et al. 2009). Sawshark are also caught in state-managed fisheries. Estimates of these catches were provided by Pribac (2007) up to 2006 and then assumed to be constant at that level. Appendix Table G4 Sawshark catches (t) and catch per unit effort used in Tier 4 estimation contains the data used in the Tier 4 assessment.
The Tier 4 assessment also requires use of data from a “reference” period, an historical period identified as desirable in terms of CPUE, catches and status of the fishery. SharkRAG experienced difficulty in selecting a suitable reference period and two candidate periods were chosen: 1986– 2001 and 2002–08. During the 1986 to 2001 period catches ranged from 225 tonnes to 487 tonnes and during the 2002 to 2008 reference period catches ranged from 317 tonnes to 452 tonnes (Table G4).
Overall there has been a decline in standardised CPUE of around 50 per cent from 1980 to 2008 (base case index, Figure 9.2). If the CPUE series is considered indicative of abundance then this suggests that abundance has been approximately halved over this period. Consideration of the state of the stock in relation to initial levels requires assumptions about the level of depletion that occurred before 1980. Catches from the shark database date from 1976 and indicate catches of around 240 tonnes per year from 1976 to 1979 (Figure 9.4), this implies that significant catches were being taken before the data used in CPUE standardisation. Trawl catches of sawshark are not available from this period, however it is likely that catches were being taken at this time. There is evidence of trawl catches dating back to the 1950s. Klaer (2006) reports catches of ‘sawfish’ of around 30 tonnes per year in the 1950s from the south east trawl sector. Despite the described historical catches SharkRAG believes that it is unlikely the species had been severely depleted by 1980 and sawshark has never been a target species.
Tier 4 analyses results using the 1986–2001 reference period result in a catch target (C*) of 301.1 tonnes and an RBC for 2010 of 202.5 tonnes, whereas the 2002–2008 reference period results in a catch target of 391.4 tonnes and an RBC of 369.6 tonnes (Figure 9.4). There are two major reasons for the difference between the reference periods (Table 9.1). Catch estimates are considerably higher in the latter period (Table G4) which result in a higher catch target estimate and a higher RBC. In addition, the average CPUE in the earlier reference period is higher than the latter, producing a scaling factor that also contributes to a lower RBC. The confidence in the accuracy of these two elements is important in making a decision on the appropriate RBC. Although there are uncertainties in the shark sector data, the trends are consistent across the scenarios examined. However, there is greater uncertainty in the estimates of discarding and catch estimates from other sectors (Table G4). For example, it is likely that the trawl catches are underestimated in the earlier years of the series. Following the introduction of quotas and the introduction of more accurate monitoring of landings, 114.6 tonnes of sawshark were reported as landed in 2002 compared with 41.6 tonnes from the 2001 logbook data. Higher values for these earlier years would produce a higher catch target. If it is assumed that catches were at this higher level since the mid-1980s then the 1986–2001 period catch target would be 386.7 tonnes and the 2010 RBC would be 190.3 tonnes. The 2002–2008 values would not be affected by changes to the earlier catch estimates.
There are currently no values assumed for discarding by the trawl sector for years before 1996 though it is likely there would have been some level of discarding. The level of discarding from the gillnet and longline sectors is also poorly known. If discards for the gillnet and longline sectors are assumed to be 20 per cent (rather than 6 per cent), then the catch target for the 1986–2001 period becomes 333.2 tonnes and the RBC 224.1 tonnes (however, a higher level of
169 Reducing uncertainty in stock status ABARES discards would also be subtracted to provide a similar level of RBC). Assuming 20 per cent level of discarding with the 2002–2008 reference period results in a catch target of 415.6 tonnes and RBC of 392.5 tonnes.
Table 9.1 Sawshark 2009 Tier 4 methodology results (base case, all areas)
Reference period 1986-2001 (1993-1997 2002-2008 omitted) CPUE target (CPUEtarg) 1.224 1.018 CPUE limit (CPUElim) 0.490 0.407 0.984 0.984 Recent CPUE average ( CPUE ) No. years in average (m) 4 4 Catch target (C*) 301.07 391.40 Scaling factor 0.673 0.944 2010 RBC 202.50 369.61 State catch 40.44 40.44 Discards 29.41 29.41 RBC – discards and State catch 132.66 299.76 Note: CPUE – Catch per unit effort. RBC – Recommended biological catch Source: ABARES
Figure 9.4 Sawshark catches and reference period target catch levels
500 450 400 350
300 GHAT 250 Total catch
Catch (t)Catch 200 C* 1986 to 2001 150 C* 2002 to 2008 100 50 0 1970 1980 1990 2000 2010
Note: C*_1986_2001 refers to target catch if a reference period of 1986-2001 is used and C*_2002_2008 refers to target catch if a reference period of 2002-2008 is used. Source: ABARES Conclusion
Results from this Tier 4 assessment are substantially uncertain due to the lack of species-specific catch data, variable catch as well as lack of information on historical catches and discard levels. The base case index of abundance shows a clear decline through time and this was supported by alternative formulations of the standardisation model. Although two alternative reference periods were explored, there is no clear time period with stable CPUE and catch which are preferred for Tier 4 reference periods. The differentiation of species in future catches would
170 Reducing uncertainty in stock status ABARES help to reduce uncertainty in future assessments. No conclusion was reached on the status of sawshark from this assessment. Appendix G Table G1 Shark Gillnet and Shark Hook sectors sawshark catch per unit effort and catch indexes Standardised CPUE Probability of Standardised CPUE Catch index Raw CPUE index Year index referenced to catching saw shark index referenced to 2002 referenced to 2002 2002 1980 0.3314 0.0016 2.0319 3.8093 2.1768 1981 0.2690 0.0013 1.5789 2.9286 1.7400 1982 0.2555 0.0012 1.4697 3.2579 1.9701 1983 0.2757 0.0012 1.5255 2.6719 1.6397 1984 0.2812 0.0012 1.5439 2.5223 1.4965 1985 0.2804 0.0012 1.5451 2.5591 1.5108 1986 0.2528 0.0010 1.2875 2.4794 1.5156 1987 0.2436 0.0010 1.2033 2.9018 1.7964 1988 0.2312 0.0011 1.3062 2.3340 1.8285 1989 0.2190 0.0010 1.1968 2.1055 1.7885 1990 0.2203 0.0009 1.1444 2.0687 1.6178 1991 0.2471 0.0010 1.2768 2.0693 1.6142 1997 0.2998 0.0010 1.2381 1.3205 1.3662 1998 0.3109 0.0010 1.2058 1.1602 1.3204 1999 0.3134 0.0010 1.1898 0.9833 1.1386 2000 0.3282 0.0010 1.2239 1.0783 1.1675 2001 0.3342 0.0010 1.1912 1.0247 1.0518 2002 0.3195 0.0008 1.0000 1.0000 1.0000 2003 0.3213 0.0009 1.1025 1.1305 1.0917 2004 0.3323 0.0009 1.0860 1.0925 1.0203 2005 0.3420 0.0008 0.9854 1.0794 0.9977 2006 0.3454 0.0008 0.9975 0.9787 1.0189 2007 0.3765 0.0007 0.9051 0.8006 0.7334 2008 0.3643 0.0008 1.0462 0.9403 0.8663
171 Reducing uncertainty in stock status ABARES
Table G2 Sawshark gillnet and longline catch (t) by shark sub-regions
Year WSA CSA SAV WBS EBS WTAS ETAS NSW Unknown Total catch 1970 0 0 1.728 7.247 15.839 0 1.503 0 0 26.317 1971 0 0 1.197 26.772 16.309 0 0.034 0 0 44.312 1972 0 0 4.591 35.638 25.717 0 3.976 0 0 69.922 1973 0 0.224 13.384 46.323 87.869 0.209 0.377 0 0 148.386 1974 0.028 0.197 3.701 45.441 162.459 2.393 13.059 0 0 227.278 1975 0 0.02 4.57 49.025 166.018 0.008 1.198 0 0 220.839 1976 0 0.431 35.518 85.531 122.478 0.452 4.24 0 0 248.65 1977 0 0.32 13.665 98.608 114.037 2.028 1.719 0 0 230.377 1978 0 0 11.417 133.508 116.783 0.171 7.321 0 0 269.2 1979 0 0 5.391 118.55 108.596 0 4.223 0 0 236.76 1980 0 0.12 4.023 113.508 108.781 0.37 1.167 0 0 227.969 1981 0.018 0 4.462 84.54 102.892 0.441 1.239 0 0 193.592 1982 0 0 8.292 130.106 101.91 0.96 2.779 0 0 244.047 1983 0 0 6.425 95.327 130.641 0.575 1.705 0 0 234.673 1984 0 0.194 4.206 92.811 124.495 3.621 5.138 0 0 230.465 1985 0.025 1.452 5.239 109.432 134.874 5.219 6.672 0 0 262.913 1986 1.603 2.538 4.595 114.5 143.532 8.793 4.968 0 0 280.529 1987 5.19 8.396 13.374 159.258 131.112 7.23 2.805 0 0 327.365 1988 1.167 7.483 13.21 108.335 107.515 4.399 6.599 0 0 248.708 1989 0.751 5.81 7.314 80.008 109.515 1.969 7.223 0 0 212.59 1990 2.669 3.753 4.472 70.764 90.263 5.2 3.002 0 0 180.123 1991 2.824 4.383 17.32 89.838 89.511 2.709 5.021 0 0 211.606 1992 1.179 3.356 6.951 85.717 90.742 14.755 6.542 0 0 209.242 1993 0.211 2.137 8.629 113.184 142.494 17.241 5.309 0 0 289.205 1994 0.996 2.538 7.48 129.354 176.588 5.348 5.102 0 0 327.406 1995 0.473 6.951 31.963 185.733 152.381 2.518 10.964 0 0 390.983 1996 0.623 3.431 16.519 136.674 136.033 5.591 11.956 0 0 310.827 1997 2.528 7.98 28.404 86.981 95.393 2.536 9.507 0 0 233.329 1998 2.866 8.6 16.997 81.192 126.555 2.019 5.918 0 0 244.147 1999 1.969 7.007 18.561 60.195 117.536 1.924 5.581 0 0 212.773 2000 0.826 5.802 12.247 39.917 134.874 1.424 5.92 0 0 201.01 2001 1.059 10.114 8.187 22.578 129.058 1.406 3.41 0 0 175.812 2002 0.722 4.247 15.288 44.423 95.083 1.511 5.517 0 0 166.791 2003 1.17 5.079 18.542 66.343 106.741 2.51 4.149 0 0 204.534 2004 0.492 5.05 15.069 74.666 92.126 0.808 5.061 0.041 0 193.313 2005 1.948 6.989 23.809 65.962 68.609 0.845 3.928 0.558 0 172.648 2006 0.917 5.89 33.696 34.518 73.469 6.505 3.628 0 0.024 158.647 2007 0.556 5.768 12.76 28.365 55.516 1.156 3.744 0.004 0 107.869 2008 0.497 6.349 7.235 12.943 65.207 0.187 2.583 0 0.02 95.021 Note: WSA: Western South Australia, WBS: Western Bass Strait, ETAS: Eastern Tasmania, CSA: Note:Central South Australia, Source: EBS: Eastern Bass Strait, SAV: South Australia – Victoria, WTAS: Western Tasmania
172 Reducing uncertainty in stock status ABARES
Table G3 Sawshark gillnet and longline catch (t) by Southern and Eastern Scalefish and Shark Fisheryshark gear type
Year Longline Unknown Mesh 6.5" mesh 6" mesh 7" mesh 8" mesh Unknown/other Total catch 1970 2.597 23.698 0 0 0 0 0.022 26.317 1971 6.558 37.72 0 0 0 0 0.034 44.312 1972 2.35 63.314 0 0 0 0 4.258 69.922 1973 1.753 17.14 0.029 28.995 97.02 3.098 0.351 148.386 1974 23.237 12.342 0 130.558 54.112 0.099 6.93 227.278 1975 1.98 20.197 0 176.178 20.868 0.004 1.612 220.839 1976 3.33 34.024 0 192.222 17.212 0.612 1.25 248.65 1977 3.745 23.134 0 194.402 1.934 0.358 6.804 230.377 1978 0.58 32.858 0.363 228.811 0.771 0.015 5.802 269.2 1979 0.235 27.075 2.519 198.518 8.119 0 0.294 236.76 1980 0.114 20.04 2.302 199.781 5.732 0 0 227.969 1981 0.104 39.775 0.984 146.912 5.817 0 0 193.592 1982 0.119 42.14 0.238 199.48 2.07 0 0 244.047 1983 0.348 50.154 0.22 182.923 0.14 0 0.888 234.673 1984 0.259 67.008 0.4 162.274 0.524 0 0 230.465 1985 2.658 89.538 2.302 166.407 1.996 0 0.012 262.913 1986 4.583 94.038 0.828 176.238 3.841 0.031 0.97 280.529 1987 3.839 99.034 0.79 209.04 11.556 1.211 1.895 327.365 1988 2.942 61.455 1.761 177.952 4.401 0.072 0.125 248.708 1989 4.66 67.549 1.231 132.396 4.088 0.105 2.561 212.59 1990 6.351 44.211 1.279 123.735 3.612 0 0.935 180.123 1991 6.313 52.778 1.942 147.796 2.476 0 0.301 211.606 1992 19.994 36.704 1.112 146.919 4.494 0.002 0.017 209.242 1993 19.396 46.993 0.237 211.349 5.683 4.032 1.515 289.205 1994 4.904 44.199 0.57 271.137 4.132 1.654 0.81 327.406 1995 1.472 0 15.541 359.725 13.894 0.045 0.306 390.983 1996 1.234 0 2.58 293.207 12.991 0.018 0.797 310.827 1997 0.613 3.849 33.976 187.891 5.084 0.097 1.819 233.329 1998 0.358 0.002 15.731 226.693 0.232 0 1.131 244.147 1999 0.399 0 10.179 199.836 0.599 0 1.76 212.773 2000 0.371 0.386 12.415 186.984 0.137 0 0.717 201.01 2001 4.307 0.136 9.059 162.149 0 0 0.161 175.812 2002 0.066 1.458 8.665 156.478 0 0 0.124 166.791 2003 0.174 0 6.515 197.613 0.07 0 0.162 204.534 2004 0.146 0 9.508 179.773 0 0 3.886 193.313 2005 0.209 0 11.886 159.53 0 0 1.023 172.648 2006 0.061 0 9.851 148.524 0 0 0.211 158.647 2007 0.062 0 7.396 99.319 0 0 1.092 107.869 2008 0.061 0 7.809 86.466 0 0.09 0.595 95.021
173 Reducing uncertainty in stock status ABARES
Table G4 Sawshark catches (t) and catch per unit effort used in Tier 4 estimation
Year GHAT SET GAB State Discards Discards (SET & GAB) Total catch Std CPUE (SSF) 1976 248.650 – – – 14.919 – 273.515 – 1977 230.377 – – – 13.823 – 253.415 – 1978 269.200 – – – 16.152 – 296.120 – 1979 236.760 – – – 14.206 – 260.436 – 1980 227.969 – – – 13.678 – 250.766 2.032 1981 193.592 – – – 11.616 – 212.951 1.579 1982 244.047 – – – 14.643 – 268.452 1.470 1983 234.673 – – – 14.080 – 258.140 1.526 1984 230.465 – – – 13.828 – 253.512 1.544 1985 262.913 4.110 – – 15.775 – 293.314 1.545 1986 280.529 19.478 – – 16.832 – 328.060 1.287 1987 327.365 16.431 0.015 – 19.642 – 376.548 1.203 1988 248.708 30.514 0.505 – 14.922 – 304.598 1.306 1989 212.590 18.273 3.983 – 12.755 – 256.105 1.197 1990 180.123 17.463 9.601 – 10.807 – 225.199 1.144 1991 211.606 20.737 14.442 – 12.696 – 267.946 1.277 1992 209.242 25.183 25.255 – 12.555 – 280.604 – 1993 289.205 30.484 20.506 – 17.352 – 369.116 – 1994 327.406 42.596 17.139 – 19.644 – 419.882 – 1995 390.983 32.427 24.365 – 23.459 – 486.873 – 1996 310.827 37.818 29.462 0 18.650 31.078 440.268 – 1997 233.329 35.992 27.609 0 14.000 24.773 345.036 1.238 1998 244.147 29.329 25.726 0 14.649 25.010 348.626 1.206 1999 212.773 35.153 23.123 16 12.766 22.156 330.483 1.190 2000 201.010 53.421 23.645 13 12.061 20.150 331.327 1.224 2001 175.812 41.698 33.684 10 10.549 20.150 298.925 1.191 2002 174.041 114.605 28.482 17.732 10.442 20.150 372.414 1.000 2003 205.600 103.005 55.000 33.021 12.336 20.150 437.336 1.103 2004 213.850 116.290 40.535 40.062 12.831 20.150 452.272 1.086 2005 192.016 110.088 43.298 49.793 11.521 20.150 434.547 0.985 2006 175.096 144.101 53.274 37.325 10.506 20.150 447.456 0.997 2007 117.559 96.871 32.888 37.325 7.054 20.150 316.549 0.905 2008 132.815 100.399 23.698 37.325 7.969 20.150 327.669 1.046 Note: Catches for GHAT, SET and GAB are taken from logbooks before 2002. Subsequently they Note:are from landings Source: data. Discards for the gillnet and longline sectors (SSF) presented here are an assumed 6% of total catch (based on Braccini et al. 2009) Other discards (SET and GAB) are based on Walker and Gason (2007) since 2000.
174 Reducing uncertainty in stock status ABARES
References
AFMA 2009, Harvest Strategy - for the Southern and Easter Scalefish and Shark Fishery, AFMA, Canberra.
Braccini, JM, Walker, TI & Gason, AS 2009, GHATF shark survey of population abundance and population size composition for target, by-product and bycatch species, R2006/823, final report to Australian Fisheries Management Authority, Fisheries Research Branch, Department of Primary Industries, Victoria, Australia.
Klaer, N 2006, Changes in the Structure of Demersal Fish Communities of the South East Australian Continental Shelf from 1915 to 1961, PhD thesis.
Little, R, Tuck, G, Haddon, M, Day, J, Klaer, N, Smith, T, Thomson, R, & Wayte, S 2008, Developing CPUE targets for the Tier 4 harvest strategy of the SESSF report to the Shelf Resource Assessment Group, August 2008.
Pribac, F 2007, Tier 4 Harvest Control Rule applied to Elephant Fish and Saw Shark 2007, report to the Shark Resource Assessment Group, October 2007.
Walker, TI & Gason, AS 2007, Shark and other chondrichthyan by-product and bycatch estimation in the Southern and Eastern Scalefish and Shark Fishery, project No. 2001/007, final report to the Fisheries Research and Development Corporation.
Wilson, D, Curtotti, R, Begg, G & Phillips, K 2009, Fishery status reports 2008: status of fish stocks and fisheries managed by the Australian Government, Bureau of Rural Sciences & Australian Bureau of Agricultural and Resource Economics, Canberra.
175 Reducing uncertainty in stock status ABARES 10 Depletion analyses of Gould’s squid in the Bass Strait
Belinda Barnes, Peter Ward and Veronica Boero Summary
A modified depletion analysis that included natural mortality and growth was used to estimate the abundance of Gould’s squid (Nototodarus gouldi) in the Central Zone of Bass Strait from 1995 to 2006. The purpose was to identify possible trends over seasons and to determine whether an impact of fishing could be detected across these seasons. Standardised catch per unit effort (CPUE) from the jig fishery logbook data was used as an index of abundance, while total catches for the jig and trawl sectors were included in the analysis. Affects on the assessment of a range of natural mortality values from similar squid stocks were explored. The results suggest pre-season stock numbers of around 10 million animals, and a proportional escapement of less than 40 per cent in about half the years considered, indicating considerable fishing impact in those years. Trends suggesting some overall stock declines during 1995 to 2006. These conclusions strongly depend on assumptions about natural mortality rates, recruitment, migration and growth of Gould’s squid. Assessment conclusions would be quite different, for example, if squid do not all recruit at the same time each year and if they do not grow at the rate derived from 1980 and 1984 size composition data. Nevertheless the analyses highlight that, if overfishing occurred within a fishing season, current arrangements for data collection and analysis would not allow management actions to be implemented in time to prevent squid biomass falling to unacceptable levels. Regular monitoring of the size composition of catches would greatly improve the reliability of the assessment and would allow the effective implementation of the fishery’s harvest strategy. Introduction
Gould’s squid is taken by the Southern Squid Jig Fishery (SSJF) and by the Commonwealth Trawl Sector of the Southern and Eastern Scalefish and Shark Fishery. Gould’s squid is the only target of the SSJF. The SSJF is located offshore from the coastal waters of New South Wales, Victoria, Tasmania and South Australia, and in a small area of oceanic water off southern Queensland (Map 14).The SSJF is a single method fishery managed by input controls such as gear based statutory fishing rights and annual total allowable effort (Woodhams et al. 2011). The SSJF harvest strategy, developed in 2007, does not specify target or limit reference points or associated indicators (AFMA 2007). Instead, responses to the harvest strategy’s triggers require that depletion analyses be conducted for major fishing regions and the entire fishery. A rapid response is required to collect biological data to determine the cohort(s) being fished through size or age frequency data. These data should feed into the depletion analyses to enable squid abundance to be estimated (AFMA 2007). If the assessment shows that depletion has not been significant for any region, then fishing can be allowed to proceed until a limit is reached. Alternatively, a total allowable catch or season closure can be determined for the whole fishery or fishery region (AFMA 2007; Ward et al. 2013). The harvest strategy’s intermediate triggers are 3000 tonnes for the jig catch or 4000 tonnes for the combined jig and trawl catch or when fishing effort reaches 30 standard jig vessels in any season (AFMA 2007).
176 Reducing uncertainty in stock status ABARES
Map 10.1 Map of south-eastern Australia showing the distribution of squid catches in 2010 and the Bass Strait Central Zone, which was the subject of the current assessment
Source: ABARES
Before 1978 the annual Australian catch of Gould’s squid was less than 100 tonnes, largely as a by-product of demersal trawling and danish-seining off south-eastern Australia. During the late 1970s Japanese commercial squid jig vessels carried out feasibility fishing in southern waters under a joint venture with Australian companies (Woodhams et al. 2011). Annual catches ranged up to 8 485 tonnes (1980), but interest waned because of variable catch rates and competition from other squid fisheries (Woodhams et al. 2011). Since 1986 the annual catch of Gould’s squid by jig and trawl has not exceeded 3 000 tonnes, with recent catches below 1 000 tonnes.
In the past the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) Fishery status reports have classified squid taken by the SSJF as uncertain if overfished and uncertain if subject to overfishing. The 2008 Fishery status reports (the most recent report at the start of this study) classification was changed to not overfished and not subject to overfishing based on historically low catches in recent years (caused by economic factors rather than availability of squid), maintenance of good jig catch rates when fishing had been undertaken, and the known ability of squid species to thrive when environmental conditions are favourable (Wilson et al. 2009). Approach to status determination The purpose of this study was to apply depletion analyses to historical squid catch and fishing effort data to determine whether an impact of fishing could be detected. It also provides an opportunity to highlight key data inputs and population parameters that influence assessment conclusions. This includes the arrangements for managing fishing activities through the SSJF harvest strategy (AFMA 2007). This paper uses a modified depletion analysis that includes
177 Reducing uncertainty in stock status ABARES natural mortality and growth to estimate the abundance of squid in the Central Zone of Bass Strait from 1995 to 2006. Standardised catch per unit effort (CPUE) from jig fishery logbook data was used as an index of abundance, while catches for the SSJF and trawl sector have been included in the analysis. Data and methods
The short life-span of squid (maximum life span of 12 months) (Woodhams et al. 2011), their highly variable inter-annual abundance and weak stock-recruitment relationships make stock assessment difficult (Young et al. 2004). Consequently, depletion analysis is often used to estimate the status of squid stocks (Young et al 2004; Chen et al. 2008). Such analysis has been applied to squid fisheries around the Falkland Islands (Beddington et al. 1990; Rosenberg et al. 1990, Basson et al. 1996, Agnew et al. 1998), in Scottish waters (Young et al. 2004), the Northwest Pacific Ocean (Chen et al. 2008) and the English Channel (Royer et al. 2002). In these fisheries, analysis and management have been carried out in real-time to avoid overfishing. In the absence of real-time data, the analysis can be carried out retrospectively to explore stock trends and fishing impacts, as is the case in this paper.
The Leslie-Delury depletion model (Table H1) in a modified form that included natural mortality was applied to examine squid abundance trends and to assess the impact of fishing in the Australian jig and trawl fishery in the Bass Strait Central Zone from 1995 to 2006. Little information on the species’ population dynamics is available, but depletion analyses require only catch and effort data and estimates of natural mortality and growth. While the analysis provides estimates of stock abundance throughout each season, the results should be interpreted with caution. This modelling approach is highly sensitive to assumptions about recruitment, migration, growth and natural mortality. Nevertheless, it does provide a means of relative assessment over the years which can be used to inform the fishery’s management. The model Triantafillos (2008) presented a preliminary depletion analysis of Gould’s squid in the SSJF. He used generalized linear models to develop abundance indexes and showed how those indexes might be used to predict initial biomass and exploitation rates in the Central Zone for one fishing season (2001). The present report extends that work with the inclusion of foreign catch, fishing effort and size data and uses many of the data processing rules described by Triantafillos (2008).
Equation 1
The depletion model (Table H1) applied in this analysis includes constant natural mortality and growth: where t is time in weeks, N1 is the initial number of squid at the start of the data series, Ci is catch in numbers at time i, wt is average weight (g) at time t, M is natural mortality (per week), CPUEt is standardised CPUE at time t, and is catchability. The right-hand side of the equation is referred to as ‘modified CPUE’ in this report. We refer to the term in square brackets as the ‘modified cumulative catch’. Given data for the two modified terms, regression allows an estimate of N1 and , thereby providing an estimate of stock numbers at the start of the data series.
178 Reducing uncertainty in stock status ABARES
An important assumption underpinning depletion analyses is that CPUE is proportional to abundance—an assumption of linearity. This assumption has been widely criticised (Rosenberg et al. 1990) although it is often used in fish stock modelling. It follows that depletion analyses should only include the portion of the time-series of CPUE where the linear assumption is reasonable (Rosenberg et al. 1990). Typically, the beginning of the series (where catch rate saturation is likely) and the end of the series are excluded (Rosenberg et al. 1990, Basson et al. 1996). Abrupt declines in CPUE at the end of the series may reflect changes in vulnerability as squid move away from the fishing grounds or mortality following spawning. In practice, the point at which CPUE peaks is usually chosen as the start of the series (Rosenberg et al. 1990, Basson et al. 1996). Thus, dealing with the assumption of linearity introduces an element of subjectivity to the analysis.
This analysis used the CPUE sub-series for each fishing season that exhibited depletion, with the beginning and end of each series truncated where it was considered to represent catch rate saturation or reduced availability or mortality associated with spawning. The impacts of including truncated points (towards the end of the series) on predicted stock numbers was also investigated. We note that depletion analysis is not appropriate for all datasets and may fail completely or provide a poor fit (Agnew et al. 1998). This was found to be the case for several years in the Bass Strait dataset.
The model used here describes the number of individuals, rather than biomass, as was the case in Rosenberg et al. (1990), Beddington et al. (1990), Basson et al. (1996) and Agnew et al. (1998), because as individuals grow, biomass may increase steadily in the presence of considerable fishing, and thus depletion (in terms of biomass) may not be evident until very late in the fishing season. Analyses of squid abundance using biomass may therefore overestimate stock size. In contrast, numbers in a closed population will decrease under fishing pressure and provide a more appropriate series. The data Weekly estimates of catch weight consisted of Commonwealth trawl sector and SSJF landings reported through the Australian Fisheries Management Authority (AFMA) catch disposal records (CDRs) (2001–08), domestic logbooks (1995–2000) and foreign logbooks (1977–87). Total catch estimates do not include state commercial or recreational catches.
We used observer-collected measurements of squid length and their distributions to convert catch estimates to number of individuals. Australian observers were deployed on foreign longliners during 1980–88. Summary spreadsheets of size (and sex) frequencies were available for 23 trips, consisting of 27 466 measurements. All but two trips were in the 1980 or 1984–85 fishing season. Note that details of the latitude and longitude of each night’s fishing operation were available, but those details were not available for size samples; the data consisted of size frequencies (Figure 10.1) combined for each trip.
AFMA also deployed observers on two domestic jig vessels in 2005 and 2007 over 26 days of fishing. We did not use those data because the small sample size was unlikely to be representative of the size structure of the squid population.
179 Reducing uncertainty in stock status ABARES
Figure 10.1 Length frequency data collected by Australian Fisheries Management Authority observers on foreign jig fishing records for 1980 (in the Bass Strait) and 1984 (in the Bass Strait and Central Zones)
180 Reducing uncertainty in stock status ABARES
Source: Data from AFMA
Catch data from the Australian jig and trawl fisheries were provided in weight (kilograms). We used the relationship developed by Willcox et al. (2001) for Gould’s squid caught in Tasmanian waters where lengths were converted to weight:
5 3.1006 w (1.4 10 )L where male and female squid are assumed to have the same relationship. We applied a methodology similar to Agnew et al. (1998) to convert catch weights to the number of individuals. For each series (1980 and 1984), for each week (t), an average individual weight
(wt) for squid was calculated using the Willcox et al. (2001) relationship. Since data were not available for all weeks, results were interpolated (Figure 10.2), although interpolation outside of the limits of the dataset were not attempted. The data were combined since the two series (1980 and 1984) were in reasonable agreement.
181 Reducing uncertainty in stock status ABARES
Figure 10.2 Regression lines for each year’s weight (g) data
Note: Dotted line is regression for 1980 data, dashed line is regression for 1984 data, and solid line is regression for all data combined. Dots are the data for the 1980 and 1984 combined regression. Source: ABARES
Unless otherwise indicated, the analyses presented here are limited to the Central Zone. For the squid population in this area we assume a single breeding season, a single recruitment event on 1 July prior to the fishing season (assumed to start 1 September) and a lifespan of one year. For short-lived species, there is usually only a single age group present in the fishery at any one time, and animals do not survive from one year to the next (Rosenberg et al. 1990). Observers did not report discards on foreign or domestic jig vessels, so that catches are equivalent to landings, and we assume a closed population—there was no emigration from, or immigration to, the population during the period of analysis.
Estimates of weekly natural mortality rates (M) were not available for the Central Zone squid stock. We used the value of M = 0.05 that was estimated for the Falkland Islands fishery for a squid species with similar life characteristics (Rosenberg et al. 1990). Beddington et al. (1990) and Basson et al. (1996) used M = 0.06 for the Falkland Islands fishery. Like the Falkland Islands assessments, we assume M to be constant throughout the season. Chen et al. (2008) considered a constant M, but examined a wide range of values from 0.03 to 0.10. Csirke (1987) included seasonal variation in M with M = 0.03 for most of the season and a higher value (M = 0.06) at the season end (Basson et al. 1996). For a variety of squid species, Pauly (1985) estimated an annual M of between 0.53 and 2, suggesting a weekly M of between 0.01 and 0.04 (Young et al. 2004). In view of the wide range of values for comparable stocks, in this assessment we explored the sensitivity of results to values of M ranging from 0.01 to 0.06. Measures of stock status and fishing impacts The Commonwealth Fisheries Harvest Strategy Policy does not specify biomass and fishing mortality reference points that can be directly applied to depletion analyses of annual stocks like squid (DAFF 2007). In this chapter we present measures of stock status and fishery impacts, based on those applied elsewhere.
For the Falkland Islands squid fishery, a 40% proportional escapement level was applied; that is
182 Reducing uncertainty in stock status ABARES
(number of spawners at season close with fishing) ≥ 0.4 Equation 2 (number of spawners at season close without fishing)
Although this reference point was not adequate in preventing declining stocks for the Falkland Islands fishery (Agnew et al. 1998), we present it for comparison purposes and as a measure of seasonal fishing pressure.
Another reference point is exploitation rate:
(total season catch) Equation 3 (initial population size)
This value measures catch as a proportion of total available stock for a fishing season. It is dependent on the initial population size in week 1 of the season. However, average weight estimates and regression analyses begin much later (approximately week 20). Thus, for this measure, we have used the predicted ‘initial’ population size at week 20 (N20) and catches from week 20 onwards. This is reasonable as the measure is used here for comparative purposes across seasons and, moreover, catches prior to this week are small relative to catches in the main part of the fishing season. Results
Results of regression analyses are presented in Figure 10.3. The model predicts stock size (N20) at the start of the series used for regression, after week 20. We could estimate numbers at week 1 from catch data, assumed growth and natural mortality rates, but such estimates rely on an interpolation of the weight–time relationship well outside the range of data.
The data for 1998 and 2002 are excluded from subsequent analysis because their CPUE series show no downward trend. This suggests that fishing mortality may have had little impact on the stock size during those years. Alternatively, recruitment or immigration during the season may have replenished the stock at a rate sufficient to mask depletion. This feature is not unusual in the analysis of short-lived squid stocks (Agnew et al. 1998).
For 2004, the data show no clear declining trend (Figure 10.3). Although the results for this year have been included in the analysis, it is likely that the model provides a poor estimate of stock size. Summary results therefore exclude this year. Catches and fishing effort in 1999 were orders of magnitude lower than in other years. The 1999 data are presented in Table H2, but this year has been excluded from the analysis.
183 Reducing uncertainty in stock status ABARES
Figure 10.3 Regressions of weekly catch per unit effort against cumulative catch for each year (natural mortality rate fixed at 0.05 per week)
Note: Modified catch per unit effort and modified cumulative catch are outlined in equation 1 Source: ABARES
All size data have been included in the analysis for the years of foreign fishing (1980 and 1984). For all other years, however, data were truncated at both ends of the series as described in the Data and Methods Section. For 2001, the last three points were truncated because they involved very small catches (200–300 kilograms) with high fishing effort, almost double the average effort for other years. For all other years, we consider the range of initial stock levels (N20) predicted when one, two or more of the final points of the series are truncated (Figure 10.4). The maximum number of points truncated depends on the series and relates to sudden increases or decreases that impact on the regression. The results of progressively truncating points (Figure 10.4) provide a rough measure of the uncertainty around estimated stock levels.
184 Reducing uncertainty in stock status ABARES
The data from 2004 are extremely noisy. A wide range of initial stock sizes (22×106 to 78×106) was estimated for 2004, indicating that the results for this year are unreliable. For 1996, there is some variability, with the initial value with truncation a lower bound.
Figure 10.4 The effect of data truncation on estimates of initial stock size (N20)
Note: Black symbols are the stock levels at week 20 predicted by the model. Red intervals cover the range of predicted initial stock levels as the number of points in the data series is progressively truncated from the end of the series. The upper bound on the interval for 2004 (7.8×107) is not shown. Natural mortality fixed at 0.05 per week. Source: ABARES
Estimated numbers and total biomass over the season are presented for each year in Figure 10.5a and b. The impact of fishing is evident, with predicted numbers at the end of the season ranging from 0.2 million (1997) to 1.4 million (1996). The number in 2004 is 3.6 million, but this series is uncertain, as discussed previously in the Results section. The impact of growth is evident in the plot for biomass; biomass increases before decreasing. For the Falkland Islands fishery a clear peak in biomass was evident in weeks 27 to 30 (March) (Beddington et al. 1990). While the peak is observed in March in some years in Bass Strait, this timing was not consistently observed for all years.
185 Reducing uncertainty in stock status ABARES
Figure 10.5 a) Estimated numbers during the fishing season for each season (natural mortality rate fixed at 0.05 per week)
b) biomass during the fishing season for each year
Note: Natural mortality rate fixed at 0.05 per week. Source: ABARES
186 Reducing uncertainty in stock status ABARES
As discussed earlier, depletion analysis cannot provide reliable, exact stock size values. Instead we examine relative quantities, which are compared across years. The purpose of this analysis is to determine the impact of fishing, so we examine variation in initial numbers, the exploitation rate and the proportional escapement of the stock (Figure 10.6 and Figure 10.7). Regression lines in figures 8 and 9 are used to illustrate a general trend and are not intended to ‘fit’ the data.
Figure 10.6a shows that, when 2004 is excluded, there is a slow and steady decline in initial stock numbers over the time period under consideration, and that these values are, in general, lower than those estimated for the 1980s. Notably, catch numbers have declined to a lesser degree and, when all catches are considered (that is, including 1998, 1999, 2002 and 2004), have remained steady (although lower than those in the 1980s).
The graph of exploitation rate (Figure 10.7a, equation 3) illustrates a relatively stable exploitation rate of between 20 per cent and 40 per cent over the years, in spite of highly variable initial stock sizes. For proportional escapement (Figure 10.7b, equation 2) it is notable that for 5 out of the 11 years (including 1984), proportional escapement fell to or below the 40 per cent level used as a conservative threshold in the Falklands Islands fishery to terminate fishing during the season. If the 2004 series is excluded, there is an upward trend in exploitation rate over the years and a downward trend in escapement, although both trends are slight.
In order to establish any direct relationship of one season’s fishing on the following year’s recruits, we considered total catch, escapement and exploitation relative to recruits the following year. However, we could find no evidence of a consistent relationship that might provide evidence that high catches or exploitation result in low subsequent recruitment, although the trends in Figure 10.6b and Figure 10.7b suggest this result. This supports the notion of high and unpredictable inter-annual variation between stock sizes, consistently reported in the literature (see, for example, Royer et al. 2002).
Figure 10.6 a) Initial stock size (N20) plotted by year
Note: The solid line is the regression of all data points. The dashed line is the regression excluding 2004. Estimates for 1980 and 1984 are shown on y-axes for reference. Source: ABARES
187 Reducing uncertainty in stock status ABARES b) Total catch plotted by year
Note: The solid line is the regression line of all data points. The dashed line is the regression line excluding 1998, 1999, 2002 and 2004. Estimates for 1980 and 1984 are shown on y-axes for reference Source: ABARES
Figure 10.7 a) Exploitation rate plotted by year
Note: The red dashed line is the exploitation rate for 1980 and the blue line is that for 1984. The solid line is the regression of all the data points. The dashed line is the regression excluding 2004. Source: ABARES
188 Reducing uncertainty in stock status ABARES b) Proportional escapement plotted by year
Note: The red dashed line is the exploitation rate for 1980 and the blue line is that for 1984. The dotted line at 0.4 represents the threshold applied in the Falkland Island Squid Fishery. The solid line is the regression of all the data points. The dashed line is the regression excluding 2004. Source: ABARES
Table 10.1 provides a summary of the values for initial stock size, final population size (a proxy for stock spawning biomass), exploitation rates and proportional escapement values for the years considered and the results presented.
Stock-recruitment curves are notoriously difficult to establish for annual stocks like squid (Young et al. 2004). For the present analysis, plotting the estimated number of recruits against the estimated number of spawners in the previous year (Figure 10.8) highlights an anomaly where low spawner numbers in 2003 are followed by very high recruit numbers in 2004 (point 3), and the resulting high number of spawners in 2004 is then followed by very low recruit numbers in 2005 (point 2). It should be noted, however, that points 2 and 3 include data from 2004, already discussed as being possibly problematic or unreliable. In general, these results serve to support the notion that annual squid stocks are highly variable between years, and that this analysis does not provide a well defined relationship between spawners and recruits.
189 Reducing uncertainty in stock status ABARES
Table 10.1 Results of the depletion analysis that used a natural mortality rate of 0.05
Year N20 N50 Exploitation Proportional Comments (million (million rate escapement individuals) individuals) 1995 6.2 0.7 0.31 0.46 na
1996 12.4 1.4 0.22 0.54 N20 sensitive to end values of data series 1997 2.0 0.2 0.34 0.38 na 1998 – – – – Analysis failed, no depletion series 1999 0.4 0.003 0.55 0.04 Very low stock levels and catches, not used in analysis 2000 9 .0 0.8 0.35 0.33 na 2001 7.5 1.1 0.16 0.70 Very short depletion series 2002 – – – – Analysis failed, no depletion series 2003 6.2 0.3 0.4 0.32 na 2004 21.3 3.6 0.13 0.76 Unreliable, uncertain results 2005 3.3 0.3 0.44 0.38 na 2006 5.8 0.6 0.28 0.50 na
Note: N20 is the predicted initial stock size at the season ‘start’ (week 20). N50 is the predicted final stock size at the season end (week 50). Exploitation rate and proportional escapement as defined in section Methodology.
Figure 10.8 Stock-recruitment relationship
Note: Point 1 represents 1997 recruits from 1996 spawners, point 2 represents 2005 recruits from 2004 spawners and point 3 represents 2004 recruits from 2003 spawners. Source: ABARES
190 Reducing uncertainty in stock status ABARES
For the preceding analyses natural mortality was fixed at a weekly rate of 0.05. However, this value has not been established for the Australian stock. As discussed in the Data section, from the literature a rate within the range of 0.01 to 0.06 is plausible for stocks with similar characteristics. Thus we have explored the effect of natural mortality (M) on the estimation of initial stock numbers and proportional escapement (Figure 10.9).
Figure 10.9b considers proportional escapement, and suggests that, approximately, escapement varies linearly with natural mortality. Within this range, M has a considerable impact on the number of years that fall below the threshold level, and thus assumptions concerning natural mortality have implications for management. However, the downward trend in proportional escapement over the time interval remains qualitatively unchanged, although it becomes more pronounced as M decreases.
Figure 10.9 a) Estimates of initial stock size
191 Reducing uncertainty in stock status ABARES b) Proportional escapement over a range of values of natural mortality rate (M)
Note: The dashed line is at 0.4 escapement and is the conservative management threshold for that is used in the Falkland Islands fishery. Source: ABARES Discussion
This application of depletion analyses suggests initial annual population sizes of the order 10 million individual squid in the SSJF each year. While depletion analysis cannot reliably predict exact numbers, its value is in the relative assessment across a number of seasons as presented here, and the exploitation and proportional escapement values that provide support for management.
There are two main results that emerge from this study: first, there appears to be a slow downward trend in recruit numbers and proportional escapement over the time period considered; and second, for approximately half the years examined, proportional escapement (a direct measure of fishing impact) falls below the threshold level set for the Falkland Islands fishery, which, although considered precautionary, was not adequate in preventing declining stocks in that fishery (Agnew et al. 1998). These results suggest that, while practices between 1995 and 2006 did not constitute serious overfishing, increased levels may well do so. And further, that an extension of the time series to include more recent years may verify these results.
For the results presented here, we highlight some caveats to the approach. An assumption about average weight over time was made from length-frequency data from 1980 and 1984, whereas our analysis is from 1995 to 2006. While the data from the two earlier seasons are in reasonable agreement, this extrapolation may not be appropriate. Young et al. (2004) report statistically significant inter-annual differences in squid growth in Scottish waters, but our analysis was unable to take into account between-year variation in the seasonal pattern of body-weight. While we used the available data, our results are sensitive to the growth function chosen.
192 Reducing uncertainty in stock status ABARES
Further, recruitment beyond the beginning of the season is not considered here. In some other stocks, two distinct recruitment pulses have been observed (Young et al. 2004), including in the Falkland Islands stock where work on two distinct cohorts has been carried out (Agnew et al. 1998). Observed increases in CPUE towards the latter part of the season are thought to be due to recruitment pulses when multiple cohorts are present (Agnew et al. 1998, p156; Hatfield 1996). This may have a considerable impact on the results presented here. In the Falkland Islands this second cohort was observed to occur in May, and in our analysis this coincides with weeks 36 to 40 where changes in stocks were observed in, for example, the 1998 dataset (see Table H2). We excluded 1998 from our analysis for this reason, but a two cohort model may have been appropriate for this year.
Natural mortality (M) estimates for the Australian squid stocks were not available and thus we considered a range of plausible values, which illustrated that our quantitative results are sensitive to this rate, although the relative trends over time are similar. Further, our analysis assumes a constant M during the depletion period, as was the case in the work of Beddington et al. (1990), Rosenberg et al. (1990), Agnew et al. (1998), Young et al. (2004), Royer et al. (2002) and Chen et al. (2008). Triantafillos (2008) did not include natural mortality. However, this constant assumption may be unrealistic. A high value for M towards the end of the season may provide more accurate results (Basson et al. 1996), although we did not model the population beyond the time period of the dataset. These results highlight the need for improved biological information for reliable model predictions.
We also assume a constant catchability for a season and, in our defence, serious departures from this assumption were not observed in the Falkland Islands squid analysis (Beddington et al. 1990). However, it is possible that individuals change behaviour as they grow from juveniles to adults (Royer et al. 2002). Further, we have assumed no migration or additional unreported catch, which are assumptions usually made but not easily met in typical squid stocks (Royer et al. 2002). Conclusion
The main purpose of this study was to illustrate the application of a modified version of the depletion analysis and to investigate whether an impact of fishing on the squid stock could be detected across seasons using the available data.
Probable declining stock numbers between 1995 and 2006 were discerned, and our analysis indicates that, for the Australian Southern Squid Jig Fishery, proportional escapement falls on or below the threshold applied in the Falkland Islands fishery for almost half of the years considered. Thus, our results strike a cautionary note. However, depletion analysis cannot provide exact stock size values and this modelling approach is highly sensitive to assumptions about recruitment levels, migration, growth and mortality rates.
It should be noted that if overfishing does occur, the current arrangements for data collection and analysis are unlikely to facilitate the implementation of management actions outlined in the SSJF harvest strategy. Regular size composition sampling of catches is not currently in place and if data collection starts after catch limits and effort triggers have been reached there will be a time-lag in those data being available for analysis. The modified depletion analysis presented in this study is a useful tool for stock status determination, however the current lack of size and growth data hinders its reliability. Regular monitoring of the size composition of catches is therefore essential for the effective implementation of the harvest strategy.
193 Reducing uncertainty in stock status ABARES
Appendix H
The standard Leslie-Delury method (DeLury 1947; Hilborn & Walters 1992) allows for a simple regression procedure to predict initial stock size. Underpinning this approach is the assumption that individuals die from fishing alone, that is, natural mortality is assumed negligible. Further, the population is assumed to be closed in the sense that there is no migration to or from the population over the time period considered, typically a fishing season.
The modelled number of individuals alive at time t is given by
t N N C t 1 i1 i where N1 is the initial stock size, and Ct the catch at time t. Assuming CPUE is approximately a linear function of N, we have
CPUEt qNt t qN q C 1 i1 i Using a series of data for CPUE and catches C, simple linear regression can be applied to find estimates for q and N1. We note that, equivalently, CPUE could be assumed a linear function of biomass, whence
CPUEt q' Bt q' N w t t where wt is the average weight of individuals at time t.
This depletion method performs best when a sizeable portion of the stock is removed by fishing, that is, there is a clear period of stock depletion.
In most cases, the assumption of negligible natural mortality is unrealistic, particularly for annual stocks (Rosenberg et al. 1990). Thus, assuming a closed population, we can model abundance as
M M M N 2 (N1 C1 )e N1e C1e M 2M 2M M N 3 (N 2 C2 )e N1e C1e C2e t1 N (N C )eM N e(t1)M C e(ti)M t t1 t1 1 i1 i where, N1 is the initial stock size. Assuming, that CPUE is a linear function of N (or biomass B), linear regression can be used to estimate N1 and q. For the analysis presented in the text the following formulation has been used:
CPUEt q' Bt
q' N t wt (t1)M t1 (i1)M wt e q' N1 q' Cie i1 so that
194 Reducing uncertainty in stock status ABARES
CPUEt t1 (i1)M q' N1 q' Cie w e(t1)M i1 t Figure H1 Catch (kg) data for the SSJF and Commonwealth trawl fisheries in the Central Zone combined, 1995–2006
Source: ABARES
195 Reducing uncertainty in stock status ABARES
Figure H2 Nominal Southern Squid Jig Fishery catch per unit effort in the Central Zone data, 1995–2006
Note: CPUE by season from 1995 to 2006 where CPUE is estimated from the jig fishery in the Central Zone. The year relates to the season start on 1 September. Source: ABARES
Catch per unit effort standardisation The CPUE, calculated as the number of squid caught per jig hour, was standardised using a linear mixed model. The data used in the model was the Jig Fishery Central Zone catch and effort logbook data from the 1995 to the 2006 fishing seasons. The data was truncated to only observations from week 20 to week 46, with each fishing season arbitrarily given a commencement date of 1 July of each year.
A square root transformation of CPUE was necessary to meet the normality assumption of the residuals. The final model, fitted to the entire dataset, was:
CPUE week season subzone moonphase vessel ij i i i i j ij
196 Reducing uncertainty in stock status ABARES where week, 12-month fishing season, subzone (central-east or central-west) and moonphase (dark or light) are factors that were fitted as fixed effects. Vessel was fitted as random effects 2 assumed independent and normally distributed with a mean of zero and variance of v . The within-vessel errors ij are assumed independent and normally distributed with a mean of zero 2 and variance of and independent of the vessel random effects.
This model was used to obtain average weekly CPUE predictions for each fishing season, which were the standardised CPUE series used in the depletion models.
In developing the model, we explored the inclusion of other factors that might affect squid CPUE, such as daily average wind velocity and cloud cover recorded at Portland. However, those factors did not improve the model’s performance, as judged by Akaike’s information criterion (AIC). References
AFMA 2007, Southern Squid Jig Fishery harvest strategy, Australian Fisheries Management Authority, Canberra.
Agnew, DJ, Baranowski, R, Beddington, JR, des Clers, S & Nolan, CP 1998, ‘Approaches to assessing stocks of Logligo gahi around the Falkland Islands’, Fisheries Research, vol. 35, pp. 155–69.
Basson, M, Beddington, JR, Crombie, JA, Holden, SJ, Purchase, LV & Tingley, GA 1996, ‘Assessment and management techniques for migratory annual squid stocks: the Illex argentinus fishery in the Southwest Atlantic as an example’, Fisheries Research, vol. 28, pp. 3–27.
Beddington, JR, Rosenberg, AA, Crombie, JA and Kirkwood, GP 1990, ‘Stock assessment and the provision of management advice for the Short Fin Squid Fishery in the Falkland Islands waters’, Fisheries Research, vol. 8, pp. 351–65.
Chen, X, Chen, Y, Tian, S, Liu, B & Qian, W 2008, ‘An assessment of the west winter-spring cohort of neon flying squid (Ommasrephes bartramii) in the Northwest Pacific Ocean’, Fisheries Research, vol. 92, pp. 221–30.
Csirke, J 1987, ‘The Patagonian fishery resources and the offshore fisheries in the South-West Atlantic’, Fisheries Tech. Pap, 286, FOA, Rome.
DAFF 2007, Commonwealth Fisheries Harvest Strategy: policy and guidelines, Australian Government Department of Agriculture, Fisheries and Forestry, Canberra.
DeLury, DB 1947, ‘On the estimation of biological populations’, Biometrics, vol.3, pp. 145–67.
Hatfield, EMC 1996, ‘Towards resolving multiple recruitment into loliginid fisheries: Loligo gahi in the Falkland Islands fishery’, ICES Journal of Marine Science, vol.53, pp. 65–75.
Hilborn, R & Walters, C 1992, Quantitative fisheries stock assessment: choice, dynamics and uncertainty, Chapman and Hall, London
Pauly, D 1985, ‘Population dynamics of short-lived species, with emphasis on squids’, Science Council Studies, vol.9, pp. 143–54.
197 Reducing uncertainty in stock status ABARES
Rosenberg, AA, Kirkwood, GP, Crombie, JA & Beddington, JR 1990, ‘The assessment of stocks of annual squid species’, Fisheries Research, vol.8, pp.335–50.
Royer, J, Peries, P & Robin, JP 2002, ‘Stock assessments of English Chanel Loliginid squid: updated depletion methods and new analytical methods’, ICES, Journal of Marine Science, vol.59, pp. 445–57.
Woodhams, J, Stobutzki, I, Vieira, S, Curtotti, R & Begg GA (eds) 2011, Fishery status reports 2010: status of fish stocks and fisheries managed by the Australian Government, Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra.
Triantafillos, L 2008, Use of depletion analysis in the Southern Squid Jig Fishery of Australia, report to the Australian Fisheries Management Authority
Ward, P, Marton, N, Moore, A, Patterson, H, Penney, A, Sahlqvist, P, Skirtun, M, Stephan, M, Vieira, S & Woodhams, J 2013, Technical reviews for the Commonwealth Fisheries Harvest Strategy Policy 2007: Implementation Issues, Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) report to client prepared for the Fisheries Research and Development Corporation, Canberra, May 2013.
Willcox, S, Lyle, J & Steer, M 2001, Tasmanian arrow squid fishery – Status Report 2001, Internal Report of the Tasmanian Aquaculture and Fisheries Institute, Tasmania.
Wilson, D, Curtotti, R, Begg, G & Phillips, K 2009, Fishery status reports 2008: status of fish stocks and fisheries managed by the Australian Government, Bureau of Rural Sciences & Australian Bureau of Agricultural and Resource Economics, Canberra.
Young, IAG, Pierce, GJ, Daly, HI, Santos, MB, Key, LN, Bailey, N, Robin, JP, Bishop, AJ, Stowasser, G, Nyegaard, M, Cho, SK, Rasero, M & Pereira, JMF 2004, ‘Application of depletion models to estimate stock size in squid Loligo forbesi in Scottish waters (UK)’, Fisheries research, vol.69, pp. 211–27.
198 Reducing uncertainty in stock status ABARES 11 Stock status determination: weight- of-evidence decision-making framework
Ilona Stobutzki, James Larcombe, James Woodhams and Heather Patterson Summary
Stock status determination and reporting is undertaken in most Australian states/territories. Similar determination processes are undertaken by other nations and organisations , however challenges arise in comparing the determinations made by the different jurisdictions due to the varying approaches taken. This paper describes a weight-of-evidence decision-making framework designed to aid in determining the status of some 100 fish stocks managed by the Commonwealth. The development of this framework recognised that while not all stocks have formal stock assessments, other sources of data and information may be available that can provide indicators of stock status. The framework aims to provide for a structured scientific review and interpretation of indicators of biomass (B) and fishing mortality (F) and to arrive at a status determination through the cumulative weight of the evidence available. The framework aims to be expansive and inclusive in terms of the types of evidence that could be considered and provide a systematic approach to evaluating that evidence. It is intended to provide a transparent and repeatable process especially for data/information-poor stocks.
Implementation of framework will not remove all uncertainty in stock determination due to insufficient information and data gaps that exist for some stocks. In these cases, the framework will assist in identifying information gaps. In all cases there is a role for expert judgement in the process of status determination. The framework acknowledges this but places a strong emphasis on documenting the key evidence and the rationale for the decision. The decision-making process is undertaken separately for biomass (overfished status) and fishing mortality (overfishing status). Introduction
Stock status determination and reporting is undertaken in most Australian states/territories (for example, Flood et al. 2012, Woodhams et al. 2013, Fletcher & Santoro 2013), at a national level (for example, NOAA 2012, European Environment Agency 2011, NZ Ministry for Primary Industries 2013), within regional fisheries management organisations (see ISSF 2013 for a summary of tuna status across multiple oceans), and internationally by the Food and Agriculture Organization of the United Nations (FAO 2012). There are also similarities in the processes that underpin conservation status determination, such as the listing of protected species under the Australian Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act) or internationally (for example, IUCN Red List (IUCN 2012)).
In Australian fisheries managed by the Commonwealth, stock status has been reported in the annual Fishery status reports since 1992. This paper describes a weight-of-evidence decision- making framework to aid in determining the status of some 100 Commonwealth-managed fish stocks. The development of this framework recognises that while not all stocks have formal stock assessments, data and information may be available that can provide indicators of stock status. The framework aims to provide for a structured scientific review and interpretation of indicators of biomass (B) and fishing mortality (F) and to arrive at a status determination
199 Reducing uncertainty in stock status ABARES through the cumulative weight of the evidence available, with the intention of providing a transparent and repeatable process especially for data/information-poor stocks.
The framework builds on 18 years of experience in the Fishery status reports. The framework aims to be expansive and inclusive in terms of the types of evidence that could be considered and provide a systematic approach to evaluating the evidence. However, given the diverse nature of fisheries, there is likely to be other evidence that could also be included for particular stocks, therefore the decision process is not limited to the evidence types listed. For most Commonwealth fish stocks, particularly in the smaller fisheries, only a subset of the potential information/data may be available and/or useful. However, the framework aims to ensure all possible evidence sources are considered.
The implementation of the framework will not resolve all uncertain status; some stocks will have insufficient information/data, conflicting indicators or substantial uncertainty in assessments. In this situation the framework will assist in identifying key information gaps will influence status determination.
There is a role for expert judgement in the process of status determination. The framework acknowledges this but places a strong emphasis on documenting the key evidence and the rationale for the decision. To aid in documenting the lines of evidence, proformas are provided in the appendixes. The decision-making process is undertaken separately for biomass (overfished status) and fishing mortality (overfishing status).
This paper provides a definition of status as used by the Commonwealth and outlines the weight-of-evidence framework which includes the compilation of evidence, the weighing of evidence and the determination of status. Methods
Published status determination processes were reviewed, including previous Commonwealth Fishery status reports, approaches used in other Australian jurisdictions, other countries and internationally. This background was used to develop the framework and which was trialled in the development of the ABARES 2009 Fishery status reports.
In 2010 the proforma and framework were revised following a review by external experts (Dr Malcolm Haddon CSIRO, Dr James Skandol and Dr Mark Bergman of the Australian Centre of Excellence for Risk Analysis). The revised framework was then applied in the drafting of the 2010 Fishery status reports. This report documents the framework and is presented in this report. Status definitions Unit of assessment The term ‘fish stock’ is generally defined as ‘a functionally discrete population that is largely distinct from other populations of the same species and can be regarded as a separate entity for management or assessment purposes’ (Flood et al. 2012). In assessing and managing fish stocks this is the fundamental biological unit. However, in practice the operational management unit is not necessarily a biological fish stock as defined by Flood et al. (2012). For a number of Commonwealth-managed stocks there is a clear divergence between population structure and the management unit applied via quotas and other controls (Stobutzki et al. 2011). In addition a biological fish stock comprises a single species, whereas in practice management units can comprise several species.
200 Reducing uncertainty in stock status ABARES
In the ABARES Fishery status reports (Woodhams et al. 2013) the unit of assessment is the management unit and 96 stocks examined comprise species or groups of species (all referred to as ‘stocks’). Shared stocks Some of the stocks assessed straddle management boundaries and may be taken by fisheries within other Australian jurisdictions, on the high seas or within the exclusive economic zones (EEZs) of other countries. In the latter situations, these stocks may be part of regional fisheries management organisations (RFMO) or bilateral fisheries agreements. Where an RFMO undertakes stock assessments these inform the Australian status assessment. However, the status terminology and reference points used by RFMOs may differ from that used within Australia. Biological status criteria The biological status of a stock depends on its current stock size (biomass) and the rate of removals from it (fishing mortality) and is based on reference points (Table 11.1 The criteria for status based on the biomass and fishing mortality, in line with HSP ). The Australian Commonwealth Harvest Strategy Policy (DAFF 2007) defines target and limit reference points for Commonwealth fisheries, in terms of biomass (BTARG and BLIM, respectively) and fishing mortality (FTARG and FLIM, respectively). Stocks are classified independently with respect to the biomass status and the level of fishing mortality.
Table 11.1 The criteria for status based on the biomass and fishing mortality, in line with HSP
Status Fishing mortality (F) F < FTARG FTARG < F < FLIM F > FLIM B >= Not overfished; Not overfished; Not overfished; overfishing BTARG overfishing is not occurring. overfishing may not be occurring. is occurring; possible fish- down. BTARG > Not overfished; Not overfished; Not overfished; B > BLIM overfishing is not occurring. overfishing may not be occurring if F will overfishing is occurring. allow rebuilding to target or close to target and a step down in F is planned. B < BLIM Overfished; Overfished; Overfished; overfishing is not occurring as overfishing may not be occurring if overfishing is occurring. long as F will enable projections/indicators show current F rebuilding to BLIM within the will enable rebuilding to target.
required timeframe. Biomass (B) Biomass Source: DAFF (2007)
The Australian Commonwealth Harvest Strategy Policy (HSP) specifies default limit of half the biomass required for MSY (0.5BMSY) or 20 per cent of the unfished biomass (0.2B0). For biomass status, stocks can be classified as:
Not overfished where the biomass is above BLIM and therefore above the level where the risk to the stock is unacceptable.
Overfished where the biomass is below BLIM. The HSP requires that fish stocks remain above a biomass level reference point at least 90 per cent of the time.
Uncertain where there is inadequate information to determine whether the biomass is above or below the limit reference point.
201 Reducing uncertainty in stock status ABARES
For fishing mortality status, stocks can be classified as:
Not subject to overfishing when the fishing mortality does not exceed the limit reference point (FLIM)—the stock is not subject to a level of fishing that would move the stock to an overfished state.
Subject to overfishing when the fishing mortality exceeds FLIM. The stock is subject to a level of fishing that would move the stock to an overfished state, or prevent it from rebuilding from an overfished state. The HSP indicates that an overfished stock should not be subject to any directed (targeted) fishing. Also:
- Fishing mortality in excess of FLIM will not be defined as overfishing if the stock is above the target level (BTARG) and a formal ‘fish-down’ or similar strategy has been developed.
- When the stock is less than BMSY but greater than BLIM, FLIM will decrease in proportion to the level of biomass relative to BMSY.
- Any fishing mortality will be defined as overfishing if the stock level is below BLIM, unless fishing mortality is below the level that will allow the stock to recover (to or above BLIM) within a period of 10 years plus one mean generation time, or three times the mean generation time, whichever is less.
Uncertain where there is inadequate information to determine whether the level of fishing mortality represents overfishing or not.
Weight-of-evidence decision-making framework There is always some level of uncertainty in the determination of stock status. Even for well studied stocks for which we have detailed, integrated assessments, a determination must be made on the reliability of the assessment and its suitability for determining stock status. The ‘quality’ of an assessment, as judged by the quality/representativeness of input data, goodness of fit and the application of expertise, is usually an implicit part of the assessment process—a weighing of the evidence is undertaken. The intention of developing this weight-of-evidence framework was to also enable a structured review of evidence available for status determination in cases where traditional stock assessment models are absent or equivocal.
Many uncertain stocks do not have a reliable stock assessment for deriving status. However, in many cases there are indicators and information to make a reasoned assessment of likely status. This framework provides a way of systematically reviewing evidence, weighing that evidence (applying logic) and then determining status where possible. The intention is that the process be transparent (documented), repeatable and allow uninvolved parties to clearly understand the basis and logic of a determination. Figure 11.1 provides a diagram of the framework which moves through the stages of evidence compilation, weighing of evidence and status determination. These are discussed in detail in this chapter.
202 Reducing uncertainty in stock status ABARES
Figure 11.1 Diagram of the weight-of-evidence decision-making framework for status determination
Source: ABARES
Documentation of lines of evidence There are a range of potential lines of evidence for the stock status and there are also biological characteristics of the stock/species and characteristics of the fishery/ies for the species that may influence the interpretation of the lines of evidence. The potential lines of evidence documented here are quite extensive and there may be others that are not listed. For for a particular stock, however, only some may be available. The relative weighting of the different lines of evidence is also discussed. Stock/species and fishery attributes Biological characteristics of the stock/species and characteristics of the fishery/ies for the species that may influence the interpretation of the lines of evidence are presented here. Biological characteristics: Productivity The productivity of a species is linked to its life span, age at maturity, fecundity, trophic level, taxonomy and habitat. In general species with a low productivity cannot sustain high levels of fishing mortality and will require longer timeframes to recover from an overfished status or high fishing mortality. Species with a high productivity can sustain higher levels of fishing mortality and are likely to recover more quickly from an overfished status or high fishing mortality. Biological characteristics: Natural mortality estimates Estimates of natural mortality (M) can be compared with estimates of fishing mortality. As a guide, fishing mortalities in excess of M are potentially overfishing. Estimates of M may be derived from empirical data using equations such as supplied by Hoenig (1983).
203 Reducing uncertainty in stock status ABARES
Biological characteristics: Distribution The spatial distribution of the stock in relation to the area fished influences the amount of the population impacted by fishing. Greater overlap may result in greater vulnerability to overfishing. Distribution of the stock outside the area fished may provide a refuge for part of the stock, however, the impact of other fisheries will need to be considered. Distribution in this context refers to three-dimensional space, as fish stocks may extend below the fishable area of certain fisheries, as well as geographically beyond accessible fishing grounds. Consideration must also be given to mobility in this context Biological characteristics: Aggregation behaviour Some species display obligate aggregating/schooling behaviour or at specific times for spawning, feeding or migration. Aggregating behaviour can increase vulnerability of a species to fishing (increased catchability) and therefore increase that species vulnerability to overfishing. The catch rate of aggregating species may be subject to hyperstability where declines in abundance are not reflected in a proportional change in catch rates until biomass is low. This would need to be considered when interpreting catch rates as an abundance index. Biological characteristics: Mobility Mobility of a species has been demonstrated to be important when considering the level of protection that may be afforded by closed or unfished areas of a stocks range (Barnes & Sidhu 2013). In general, highly mobile species are not well protected by closed areas or unfished areas of habitat because they are exposed to fishing in the course of their movement. There is then a continuum through to sedentary species for which closed areas are highly effective at removing a proportion of the adult stock from fishing pressure and providing a reserve of spawning potential that may contribute to the stock as a whole. This is a relevant issue when considering biomass status.
Sedentary or sessile species may also be more vulnerable to localised overfishing. Biological characteristics: Stock structure The spatial structure of the stock relative to spatial structure of the management needs to be considered in status determination. If there is significant stock structure at a finer scale than the management (eg a quota covers several biological stocks) the status assessment should be done separately for each stock, if information is available. These can then be combined for a status that corresponds to the management unit. Fishery characteristics: Separation of species The taxonomic separation of species in catch reporting needs to be considered. In for some species they are reported as part of a group or at a taxonomic level above species. Likewise, the management unit (including quota) may be the higher level (for example, deepwater sharks). Reporting grouped catches may mask depletion or overfishing of some species and may result in greater uncertainty.
If there is data available that enables separation of species then separate assessments of status may be undertaken. If it is not possible to separate species, the impact on the least productive species could be considered as an indicator and the results from Productivity-Susceptibility analyses (PSA) (Hobday et al. 2011) may assist in interpreting the indicators based on the least productive or highest risk species.
204 Reducing uncertainty in stock status ABARES
Fishery characteristics: Targeting Species that are targeted are potentially subject to higher fishing mortality than by-product species. However, the productivity and co-occurrence of by-product species needs to be taken into account. There may be greater potential for unreported discards for by-product species and therefore an underestimate of fishing mortality. The consistency of fishing practises over time with respect to targeting also needs to be considered as species may change between target and by-product over time. Targeting practices can have a profound effect on catch rates and their utility as indexes of abundance.
Species that are under rebuilding strategies may be managed under ‘bycatch TACs’; these species were previously by-product or target species, that have been depleted. In this case assessment of status needs to consider the rebuilding strategy, levels of catch and effort and whether these will facilitate rebuilding. For these species, the level of discarding and post- release survival may also be significant in determining fishing mortality. There may be increased potential for unreported discards and underestimation of the total catch for these species. Fishery characteristics: Sources of fishing mortality The status assessment should account for all sources of fishing mortality, including all fisheries, sectors, other jurisdictions and recreational fishing. If information is not available from all sources of fishing mortality there may be greater uncertainty in the status, depending on the likely scale of different sources. Empirical indicators Empirical indicators are relatively simple, non model-based, metrics that respond to stock status. Each indicator needs to be critically reviewed for a species in terms of its representativeness of the species and fishery, and its validity as an indicator of B or F for that species. The latter is related to the indicators responsiveness to changes in B or F, the impact of other drivers on the indicator (for example, management or markets) and any potential bias.
Empirical indicators can provide an indication of trends or changes in B or F. The utility of empirical indicators for status determination is improved when they are used in association with reference levels. Reference levels should be consistent with existing harvesting policy (for example, DAFF 2007).
Empirical indicators are usually key data inputs to integrated, quantitative stock assessment models and so understanding the trends and robustness of these indicators is a key aspect of assessing the reliability of integrated assessments and their outputs.
Empirical indicators for status determination include:
catch trends
size structure of the catch
age structure of the catch
effort trends
spatial distribution of the fishery
catch rates (standardised).
205 Reducing uncertainty in stock status ABARES
Catch trends The trends in catch by sector, gear or fishery over time may provide indications of the level of depletion and the level of fishing mortality in a stock. A recent history of catches that are within a recommended biological catch (RBC) limits from a robust harvest strategy is evidence of that overfishing is not likely. Explanations for large changes in catch should be sought.
Aspects that should be considered include:
Whether the catch is from a validated source or if there is an agreed catch history?
Does the catch series represent total mortality? Is there information on all sources of catches, are there any significant illegal, unregulated or unreported (IUU) catches, is discarding likely, if so is there an estimate available, is it likely to vary over time?
Does the time period covered by records represent the entire history of exploitation (is there an undocumented earlier history of catches)?
Caution is required because catch is highly subject to other drivers (apart from stock status) include management changes (for example, quotas or other forms of restrictions), market trends, structural adjustment, processing capacity and effort trends.
Table 11.2 Catch trends and potential implications for status
Catch trend Implication for biomass status (B) Implication for fishing mortality status (F)
Declining catch or previously declined or Potentially depleted Overfishing may be occurring low relative to historic highs
Stable catch (without changes in the area Potentially not overfishing, unless fished) and not substantially lower than there are indications the species is historic catches overfished
Increasing catch in a young fishery Potentially not overfishing, may be a ‘fish down’ Increasing catch in comparison to historic Biomass likely to be declining Fishing mortality likely to be catches increasing; overfishing may be occurring Current catch is > RBC or historic time Potentially to overfishing but this Potentially overfishing but it depends series depends on B level on the status of B.
Zero catch Overfishing is not occurring
Insignificant catch: If catches history is also insignificant Overfishing is not occurring The level regarded as ‘insignificant’ the stock is not overfished depends on the stock and history of exploitation Size structure of the catch Size information may come in several forms:
the trend in the length frequency distribution of the catch
the trend in the mean size of the catch
206 Reducing uncertainty in stock status ABARES
size at capture in comparison to size at maturity.
In interpreting size information from the catch, it needs to be considered whether sampling has been consistent and representative over time, whether gears and sectors may be size selective and whether this has changed over time (for example, with changes in net mesh size). Issues such as size-related schooling, changes in fishing grounds, discarding (in particular high- grading) or targeting may have influenced the trends in size data. Market preferences can also drive trends in the size structure of catches.
Table 11.3 Size structure and potential implications for status
Size structure Implication for biomass status (B) Implication for fishing mortality status (F)
Stable mean size over time Potential stable B, not overfished Potential stable F, not subject to overfishing Reduction in the mean size caught over Potential declining B Potential increasing F, overfishing time may be occurring Length classes caught in the fishery < If the fishery has been occurring for a Potential overfishing occurring index of maturity substantial time, this may indicate depletion
Length classes caught in the fishery > If fishery has been occurring for a Potentially not overfishing index of maturity substantial time, may indicate it is not be overfished
Length frequency shows declining Potentially overfished (depending on Potentially overfishing numbers or proportion of recruits the level of decline) entering the fishery, or much lower than historical levels
Length frequency shows declining Potentially overfished (depending on Potentially overfishing numbers or proportion of mature fish, or the level of decline) much lower than historical levels
Age structure of the catch Age structure information may come in several forms:
the trend in the age frequency distribution of the catch
the trend in the mean age of the catch
age at capture in comparison to age at maturity
size at age trend.
In interpreting age structure information, similar to size structure information, the representativeness of the catch sampling needs to be considered. Specific gears and sectors may be selective with respect to age and this may change over time (for example, with changes in net mesh size). Issues such as age-related schooling, changes in fishing grounds, discarding (in particular high-grading) or targeting may influence the trends. The longevity of the species should be considered in the interpretation, as longer lived species with numerous age classes in the fishery may be slower in showing indications of fishing impacts to recruitment.
207 Reducing uncertainty in stock status ABARES
Other potential drivers for trends in age structure of the catch include markets, if there is a preferred size, or management changes.
Table 11.4 Age structure and potential implications for status
Age structure Implication for biomass status (B) Implication for fishing mortality status (F) Stable mean age over time Potential stable B, not overfished Potential stable F, not subject to overfishing Reduction in mean age over time Potential declining B Potential increasing F, overfishing may be occurring
Age classes caught in the fishery < index If the fishery has been occurring for a Potential overfishing occurring of maturity substantial time, this may indicate depletion
Age classes caught in the fishery > index If fishery has been occurring for a Potentially not overfishing of maturity substantial time, may indicate it is not be overfished
Age frequency shows a declining Potentially overfished (depending on Potentially overfishing numbers or proportion of recruits the level of decline) entering the fishery, or much lower than historical levels
Age frequency shows declining numbers Potentially overfished (depending on Potentially overfishing or proportion of mature fish, or much the level of decline) lower than historical levels
Effort trends Changes in the level of fishing effort over time and also by trends in the timing and length of the fishing season can inform status determination. However, in general catch per unit effort will be more informative.
In interpreting effort trends consideration should to be given to whether the effort data is representative of the fishery, its reliability, the period covered relative to the history of the fishery, the measure of effort and whether targeting can be distinguished and whether this has changed over time. Interpretation of effort trends will vary depending on whether the species is a target or by-product. Discarding can mask effort that is catching the species and discarding practises may change over time. The potential for changes in fishing efficiency (power) over the time period needs to be carefully considered.
Other drivers for trends in effort include management restrictions, market demands, structural adjustment, operating costs, weather.
208 Reducing uncertainty in stock status ABARES
Table 11.5 Effort trends and potential implications for status
Effort trend Implication for biomass status (B) Implication for fishing mortality status (F) Increasing fishing effort, indicated by Potential declining B Potential increasing F, overfishing quantitative measures such as greater may be occurring if it has occurred number of fishing days or sets, length of over a substantial timeframe net set etc, or qualitative with larger vessels, longer fishing trips (longer seasons and fishing a larger area)
Historically low effort, few vessels or Potentially not overfished, unless Potentially low F and not overfishing little targeting, possibly capped by there are indicators the stock is (unless there are indicators stock is markets or management (restrictions on depleted and so effort has shifted to depleted/overfished) effort) other species
Declining effort na Potentially falling F but interpretation will depend on indicators of biomass
Stable effort na Potentially not overfishing
Effort in the most recent year is higher na Potentially increased F and than historic values overfishing, depending on the B status. Zero effort in the most recent year na Overfishing is not occurring
Insignificant effort in the most recent na Overfishing is not occurring. year Depending on whether a species is in recovery
Spatial distribution of the fishery Changes in the spatial distribution of catch, effort and CPUE over time, including changes in the depths fished can inform status determination. This may show stability in the grounds where the species is caught, or evidence of serial depletion, or maintaining catch or CPUE with expansion of fishing grounds. Other potential drivers of changes in the spatial distribution of the fishery include markets, fuel costs, targeting different species, exploration and discovery of new fishing grounds.
Table 11.6 Spatial trends in the fishery area and potential implications for status
Spatial trend Implication for biomass status (B) Implication for fishing mortality status (F) If it is not a developing fishery and fished Possible depletion or serial depletion Potentially overfishing as moving to area/depths expanding or moving to new maintain catch areas
Developing fishery and expanding or Not likely to be overfished Not likely to be overfishing moving to new areas/depths
209 Reducing uncertainty in stock status ABARES
Formal spatial fishing closures can also influence status determination. In considering the potential impact of closures on status aspects that need to be considered are the proportion of the species distribution that is in areas closed to fishing and the level of connectivity between closed and open areas (this includes mobility and dispersal). The robustness of the information on the distribution of the species and abundance in different areas should be considered, also the length of time for which the closures have been in place. The process for considering formal closures in setting TACs should be considered.
Informal closures, where fishers are potentially avoiding areas where species are known to be caught, due to restrictive quota or other issues can also be considered in status determination.
Table 11.7 Spatial closures and potential implications for status
Spatial closure Implication for biomass status (B) Implication for fishing mortality status (F) Significant proportion of species Some biomass may be protected Some biomass may not be subject to distribution is not fished (providing a (depending on connectivity) F (depending on connectivity) potential refuge) due to formal or informal closures
Important areas such as nursery grounds Some biomass may be protected Some biomass may not be subject to or spawning aggregation sites are closed (depending on connectivity) F (depending on connectivity)
Most of the distribution is fished, or they Little biomass protected Most of the population is subject to F are fished in key aggregating areas
Catch per unit effort Trends in the catch per unit effort (CPUE), rather than catch or effort trends by themselves, can provide an indicator of trends in biomass and are often a critical input to integrated stock assessment models. While nominal CPUE can be considered, standardised CPUE is more robust and preferable. Standardisation is a statistical modelling process (such as generalised linear modelling) that uses explanatory factors that affect catches (such as time of day, vessel, depth etc) to better reflect the underlying trend in biomass. Ideally the standardisation is able to account for changes in catches or catch rates associated with all fishing behaviours, including targeting. The robustness of standardised CPUE as an index of biomass needs to be considered, including the representativeness of the catch and effort data (such as impact of quota), potential changes in fishing efficiency/power over time, and are these reflected in the effort measure or CPUE index. Additional considerations for standardising CPUE are: what factors are taken into account; likelyhood or hyperstability or hyperdepletion, and; do the scale of changes in CPUE make sense given the biology of the species and the trends in catch and effort (CPUE indexes can be highly variable through time, implying changes to stock biomass that may not be plausible given the lifespan of the species).
Standardization should consider targeting, while taking spatial and temporal changes into account, however data availability is likely to provide a constraint through the time series. Factors such as management (quota) and market influences also need to be considered.
210 Reducing uncertainty in stock status ABARES
Table 11.8 Catch per unit effort trends and potential implications for status
CPUE trend Implication for biomass status (B) Implication for fishing mortality status (F) Declining CPUE or stable CPUE in recent Depletion of biomass Possibly overfishing depending on the years at lower levels than historically timeframe/ possibly different interpretation if species is no longer targeted
Increasing CPUE Increasing biomass Fishing mortality is allowing a biomass increase. Possibly not overfishing.
Increasing CPUE in a developing fishery Learning and expansion may be na occurring Stable CPUE Possibly stable biomass, compare to Possibly not overfishing, unless CPUE historic values for level of depletion is stable at a low level
Fishery independent surveys Fishery independent surveys (FIS) can provide estimates of absolute or relative biomass that can be more robust than fishery dependent CPUE. The FIS can be used to infer the impact of fishing, particularly if reference points are available. They may also be incorporated into quantitative stock assessment models.
Interpreting survey data and results requires that consideration is given to the survey objectives and design to ensure they are appropriate and representative for the stock under consideration.
If there is a historical time series, early estimates may be equivalent to B0 otherwise earlier fishing history must be taken into account.
Indirect methods include surveys of a species eggs that may be used to infer parental biomass (through the daily egg production method—for example, Ward & McLeay 1998) and approaches that examine the genetic composition of a sample of fish to infer relative or absolute abundance (such as the close kin method—for example, Bravington et al. 2012).
Table 11.9 Surveys and potential implications for status
Survey examples Implication for biomass status Implication for fishing mortality status (B) (F)
Time series of biomass estimates are Estimate of B compared to BTARG If declines are ongoing potentially available and can provide a trend. Early and BLIM (Table 1) overfishing, if increasing biomass or estimates can be a proxy for B0 or lightly stable probably not overfishing fished, so that BLIM and BTARG and Bcurr can be estimated
Only current biomass estimates available If current biomass estimates are na from FIS and a valid reference point be << reference point from early inferred from historic catches catches, B may be declined or overfished
211 Reducing uncertainty in stock status ABARES
Risk assessments The Australian Fisheries Management Authority (AFMA) has commissioned ecological risk assessments for all the major Commonwealth fisheries. The two main approaches have been the SAFE approach (Zhou & Griffiths 2008) and the Productivity–Sustainability Analysis (PSA; Hobday et al. 2011). These provide an assessment of two types of risk at a point in time and can contribute to status determination.
The SAFE approach (Zhou & Griffiths 2008) provides an estimate of cumulative (across gears/fisheries) fishing mortality for the assessment period, as well as reference points. It provides information on the status in terms of F at a point in time. In using the results of a SAFE assessment it should be considered whether the assumptions are appropriate for the stock, the data inputs and from where the assumed M has been derived.
In Productivity–Sustainability Analysis, ‘productivity’ is defined as the capacity of a species to recover from a depleted state and ‘susceptibility’ is the likelihood or exposure of the species to capture and mortality from the fishery (Stobutzki et al. 2001). PSA incorporates some of the species/stock attributes described in previous sections of this report. In using PSA results it should be considered whether the assumptions in the PSA process are appropriate for the stock and the data inputs (particularly how much is unknown). The productivity and susceptibility scores can be informative by themselves, but note that they overlap with the species/stock attributes.
Following the undertaking of a risk assessment there may have been management responses or other changes in the fishery that change the level of risk and this should also be considered.
Table 11.10 Risk assessments and potential implications for status
Risk assessment Implication for biomass status (B) Implication for fishing mortality status (F) PSA measure of the relative risk of If species at a lower risk are known to High risk are more vulnerable to overfishing in that fishery at a single be overfished it may be likely that this overfishing; High susceptibility score point in time, not an annual indicator but species is. more vulnerable to overfishing may provide an indication of F or vulnerability. SAFE estimate of Fcumulative, and na Estimate of Fcumm > Flim, Fcumm > reference points - max sustainable fishing Fcrash, may indicate overfishing mortality Fmsm, Flim, Fcrash, with some measure of uncertainty. Quantitative stock assessment models Quantitative stock assessment models are distinguished from empirical indicators in that they integrate data from more than one source into a mathematical framework that models aspects of the species biology. Complex quantitative models generally benefit from a synthesis of more information sources and may make fewer assumptions, however, they can be highly demanding of data and the skills of the analyst. Less complex models rely on fewer information sources, make more assumptions but are less demanding of data and so may be applicable across a wider range of circumstances. With informative input data, stock status can be reliably informed by both simple and complex models. This section does not attempt to describe the range of approaches available or the different assumptions or sensitivities of the approaches. There are a range of publications that provide this (Haddon 2011; Hilborn & Walters 1992).
In general, the quantitative stock assessment should be reviewed with respect to the data inputs, appropriateness of the model, appropriateness of assumptions, sensitivities examined and
212 Reducing uncertainty in stock status ABARES appropriate reference points. The consideration of trends in empirical indicators and other lines of evidence will help in understanding the robustness of the quantitative assessments including its inputs. In many Australian Commonwealth fisheries the resource assessment group plays a key role in the development and interpretation of quantitative assessments. The comments of these groups should be taken into consideration.
In the context of Australian Commonwealth fisheries, quantitative assessments should ideally provide estimates of:
BCURR relative to BMEY (1.2BMSY), BLIM, (B20 or 0.5BMSY),
FCURR relative to FMEY (1.2FMSY), FLIM
The modelling should also include an exploration of uncertainty through sensitivity runs where inputs parameters are adjusted within plausible ranges.
Where a robust quantitative stock assessment model is available, the interpretation of the outputs relative to status of B and F follow Table 11.1. Harvest strategies Since the development of the Commonwealth Harvest Strategy Policy (HSP), most Commonwealth fisheries have harvest strategies in place. In principle, status determination should be straightforward for a stock fished under a harvest strategy that has been demonstrated to be fully compliant with the HSP. In practice this is not always the case, particularly for smaller fisheries and harvest strategies that use simpler or untested approaches.
For harvest strategies to inform stock status, the following requirements should be noted:
evidence that the reference points or proxies are consistent with the HSP
HS testing to demonstrate that the HS control rules have a high probability of B>BLIM and F required data collected and included all forms of mortality have been taken into account, as appropriate the stock assessment that drives the harvest control rules is reliable (see previous sections) management decisions and actual practice in the fishery follows the control rules. Weighing evidence Key lines of evidence There should be clear documentation of the key lines of evidence used in the status determination, the interpretation of the evidence and justification for using these lines of evidence. This provides for transparency in the status determination, so that others can understand how the determination was reached. Weighting of lines of evidence If different weighting is applied to different lines of evidence this must be explicitly documented and justification provided. Again this provides for transparency and review by others. The weighting is qualitative where some lines of evidence are weighted higher in the cumulative 213 Reducing uncertainty in stock status ABARES weight-of-evidence and others medium or low. The weighting is likely to be fishery and stock specific. Inconsistent of lines of evidence Any inconsistent lines of evidence should be documented, along with justification for weighting these lower within the status determination. Key areas of uncertainty Areas of uncertainty in the status determination should be documented. The status of a species with respect to B or F status may be uncertain if: there is insufficient data, information or indicators to make a scientifically robust determination of status the available lines of evidence show inconsistent trends, or are substantially influenced by external drivers, and there is no basis to weight particular lines of evidence more highly than others the available quantitative assessments are very sensitive to parameter assumptions, lack key data inputs (for example, observer data, aging data, length frequency data), or have large confidence intervals in terms of the outputs the available quantitative assessments, may be uncertain if the approach is unlikely to be appropriate for the species involved, the assessment is dated and there are no valid indicators of recent trends. Conclusions on overfished (B) or overfishing (F) status The conclusions in terms of overfished (B, overfished/not overfished/uncertain) or overfishing (F, subject to overfishing/not subject to overfishing/uncertain) status should be clearly stated based on the documentation and weighting of evidence. Expert input and review The status determination involves expert judgement. Even when quantitative stock assessments are available, these have an element of expert judgement in their development, assumptions and weighting of inputs. For Commonwealth fisheries, resource assessment group often provides this input. The status determinations in the Australian Commonwealth Fishery status reports (Woodhams et al. 2013) are reviewed internally within the Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) at a stock status determination workshop. Through this workshop expert opinion is provided on the lines of evidence that have been considered, interpretation of the evidence, weighting and the conclusions. Any additional evidence is highlighted or suggestions for differences in interpretation or weighting. This may be an iterative process. The AFMA resource assessment group processes also provide for indirect expert input through the discussion of data, assessments, trends in indicators and advice on recommended biological catches. The status determinations are provided to resource assessment group chairs or appropriate scientists for expert scientific comment. This may provide comment on the lines of evidence, 214 Reducing uncertainty in stock status ABARES interpretation, weighting and conclusions. These expert comments are then considered by ABARES and again this may be an iterative process. Guidelines for weighing evidence While the weight-of-evidence framework aims to increase the transparency and systematic process for status determination, it remains an adaptable approach, particularly in data poor fisheries, without quantitative stock assessment models. Points in this section provide some general guidelines for status determination. A robust integrated stock assessment model, where available, with appropriate reference points (or proxies) is the basis for status determination (following Table 11.1 The criteria for status based on the biomass and fishing mortality, in line with HSP ). Any inconsistent indicators or areas of uncertainty should be noted. For quantitative assessments, the relative weighting from highest to lowest would be: 1) Integrated stock assessment models 2) A robust form of assessment, with appropriate reference points (or proxies), 3) A robust catch curve, is weighted most heavily in the F status determination 4) A robust CPUE analysis in B status determination. For empirical indicators, the relative weighting from highest to lowest: 5) Representative fishery independent surveys over time 6) Standardised CPUE, with appropriate reference points 7) Nominal CPUE 8) Size/age data over time 9) Spatial and temporal changes 10) Catch, Effort trends. In the absence of appropriate reference points, empirical indicators are likely to provide trends rather than absolute status. However, in the context of the attributes of the species and the fishery, a status determination can be made based on the cumulative weight-of-evidence. The weighting of indicators should reflect their representativeness of the species and fishery and their robustness in terms of the influence of other drivers. A clear interpretation of each indicator, relative weighting and the implications for status needs to be provided. In making a status determination in this way, there needs to be an explicit consideration of whether a change in status would be detected with the available types of indicators. If a species has been fished previously, but not in recent years, the probable recovery time (based on information such as natural mortality, growth and steepness parameters) in the absence of fishing provides evidence of whether the species is likely to have recovered. 215 Reducing uncertainty in stock status ABARES Reporting of stock status When reporting stock status, the keys lines of evidence and their weighting (including contradictory evidence) should be clearly stated and the rationale for the status determination provided in a transparent manner. Appendix I Table I1 Proforma for stock/species and fishery attributes Attribute type Attribute Description and implications for status determination Biological Productivity characteristics Natural mortality estimate Distribution Aggregating behaviour Mobility Stock structure Fishery Separation of species characteristics Targeting Sources of fishing mortality 216 Reducing uncertainty in stock status ABARES Table I2 Proforma for empirical lines of evidence Indicator type Indicator Representativeness of Other potential Implication Implication for Relative weighting the stock drivers that influence for status (B) status (F) in status interpretation determination Catch volume Catch trend by sector/gear or fishery (time series) Catch in the most recent year(s) Size structure of catch Trends in length frequency Trend in the size composition of the catch (by sector) over time. Mean size of catch over time Size at capture versus size at maturity (by sector) Age structure of catch Trends in age frequency age at capture vs age at maturity (by sector) Trend in the age composition of the catch (by sector) over time. Mean age of catch over time Mean length at age trend Effort trends Effort trend, by sector/gear or fishery Effort in the most recent year, compared to previous Spatial distribution of the Spatial distribution of catch/effort over fishery time (including depths) Proportion of the species’ distribution that is fished (accounting for spatial closures) Catch per unit effort Trends in nominal CPUE Trends in standardised CPUE - by sector and fishery. Reducing uncertainty in stock status ABARES Table I3 Proforma for lines of evidence for survey, risk assessment, stock assessment model and harvest strategy Indicator type Indicator Representativeness Other potential Implication Implication Relative of the stock drivers that for status for status weighting in influence (B) (F) status interpretation determination Fishery Independent Surveys Trends in estimates of biomass (relative or absolute) (includes egg (FIS) and larval surveys, acoustic surveys) Recent FIS estimates of biomass (relative or absolute) (includes egg and larval surveys, acoustic surveys) Risk Assessments Productivity–Sustainability Analysis (PSA) measure of the relative risk of overfishing. SAFE approach estimates F relative to reference points. Integrated stock assessment Assessment outputs: estimates of BCURR, FCURR, BTARG, FTARG, model BLIM, FLIM Delay-difference model Assessment outputs: estimates of BCURR/BTARG, FCURR/FTARG, BLIM, FLIM Non-equilibrium surplus Assessment outputs: estimates of BCURR/BTARG, FCURR/FTARG, BLIM, production model FLIM Catch Curve Analysis Assessment outputs: estimates of FCURR, FTARG, FLIM CPUE trend analysis Assessment outputs: estimates of BCURR/BTARG, BLIM Fishery-dependent depletion Assessment outputs: estimates of B at the start of the season analysis Proxies for reference points Harvest strategy proxies that are consistent with the HSP Harvest strategy testing Influence of the initial state of the stock on harvest strategy performance Range of uncertainties that have been tested HS Control rules Probability of harvest strategy maintaining B>BLIM based on testing Probability of harvest strategy maintaining F Reducing uncertainty in stock status ABARES Table I4 Proforma for documentation of status determination Biomass Key lines of evidence status Weighting of lines of evidence Inconsistent indicators (if any) Key areas of uncertainty Interpretation of evidence for biomass status Conclusion on overfished status: Overfished/not overfished/uncertain Fishing Key lines of evidence mortality status Weighting of lines of evidence Inconsistent indicators (if any) Key areas of uncertainty Interpretation of evidence for fishing mortality status Conclusion on overfishing status: Subject to overfishing/not subject to overfishing/uncertain References Barnes, B & Sidhu, H 2013, ‘The impact of marine closed areas on fishing yield under a variety of management strategies and stock depletion levels’, Ecological Modelling, vol. 269, pp 113–25. Bravington, M, Grewe, P & Davies, C 2012, Report of the close-kin project: estimating the absolute spawning stock size of SBT using genetics, Commission for the Conservation of Southern Bluefin Tuna working paper CCSBT-ESC/1208/19. DAFF 2007, Commonwealth Fisheries Harvest Strategy Policy and Guidelines, Australian Government Department of Agriculture Fisheries and Forestry, Canberra. European Environment Agency 2011, Status of marine fish stocks, FAO 2012, The State of World Fisheries and Aquaculture 2012, FAO Fisheries and Aquaculture Department, Food and Agriculture Organization of the United Nations, Rome. Fletcher, WJ & Santoro, K (eds.) 2013, Status Reports of the Fisheries and Aquatic Resources of Western Australia 2012/13: The State of the Fisheries, Department of Fisheries, Western Australia. Flood, M, Stobutzki, I, Andrews, J, Begg, G, Fletcher, W, Gardner, C, Kemp, J, Moore, A, O’Brien, A, Quinn, R, Roach, J, Rowling, K, Sainsbury, K, Saunders, T, Ward, T & Winning, M (eds) 2012, 219 Reducing uncertainty in stock status ABARES Status of key Australian fish stocks reports 2012, Fisheries Research and Development Corporation, Canberra. Haddon, M 2011, Modelling and Quantitative Methods in Fisheries, Second Edition, Chapman and Hall/CRC, New York. Hilborn, R & Walters, CJ 1992, Quantitative Fisheries Stock Assessment: Choice, Dynamics and Uncertainty, Chapman and Hall, New York. Hobday, AJ, Smith, ADM, Stobutzki, IC, Bulman, C, Daley, R, Dambacher, JM, Deng, RA, Dowdney, J, Fuller, M, Furlani, D, Griffiths, SP, Johnson, D, Kenyon, R, Knuckey, IA, Ling, SD, Pitcher, R, Sainsbury, K, Sporcic, M, Smith, T, Turnbull, C, Walker, TI, Wayte, SE, Webb, H, Williams, A, Wise, BS, & Zhou, S 2011, ‘Ecological risk assessment for the effects of fishing’, Fisheries Research, vol. 108, pp. 372–84. Hoenig, JM 1983, ‘Empirical use of longevity data to estimate mortality rates’, Fisheries Bulletin vol. 82, pp. 898–903. ISSF 2013, ISSF Tuna Stock Status Update, 2013(3): Status of the world fisheries for tuna. ISSF Technical Report 2013-04B, International Seafood Sustainability Foundation, Washington D.C. IUCN 2012, IUCN Red List Categories and Criteria: Version 3.1, Second edition. Gland, Switzerland and Cambridge, UK: International Union for Conservation of Nature. NOAA 2012, Status of Stocks 2012, Annual Report to Congress on the Status of U.S. Fisheries, US Department of Commerce, National Oceanic and Atmospheric Administration, National Marine Fisheries Service. NZ Ministry for Primary Industries 2013, The Status of New Zealand Fisheries 2013, Stobutzki, I, Miller, M & Brewer, D 2001, ‘Sustainability of fishery bycatch: a process for assessing highly diverse and numerous bycatch’, Environmental Conservation vol. 28, pp. 167– 81. Ward, TM & McLeay, LJ 1998, Use of the Daily Egg Production Method to estimate the spawning biomass of pilchards (Sardinops sagax) in shelf waters of central and western South Australia in 1998, Report to PIRSA Fisheries, SARDI Aquatic Sciences, Adelaide. Woodhams, J, Vieira, S & Stobutzki, I (eds) 2013, Fishery status reports 2012, Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. Zhou, S & Griffiths, SP 2008, ‘Sustainability Assessment for Fishing Effects (SAFE): a new quantitative ecological risk assessment method and its application to elasmobranch bycatch in an Australian trawl fishery’, Fisheries Research, vol. 91, pp. 56–68. 220