Quantification of the recreational harvest utilised data collected from boat-based recreational fishing surveys and tour operator returns (charter logbooks) across the multiple bioregions of Western Australia (Fig. 1), and the scaling of these data back to 1941, which is considered the start of shark fishing in the State 14,15. Recreational fishers in Western Australia are required to possess a Recreational Boat Fishing Licence which provides a cost-effective sampling frame for recreational fishing surveys. Biennial statewide recreational fishing surveys have been conducted in Western Australia in 2017/18 (September 2017 to August 2018), 2015/16 (September 2015 to August 2016), 2013/14 (May 2013 to April 2014) and 2011/12 (March 2011 to February 2012) 13. A random sample of Recreational Boat Fishing licence holders were selected for an initial Screening Survey to identify intending fishers for recruitment into a 12-month Phone-Diary Survey. Survey participants were given diary kits containing a welcome letter, species identification guide 32, fishing location guide 33 and diary card. After receiving the diary kits, a brief diary explanation was provided by trained interviewers, and participants were encouraged to use the diary card to record fishing effort (start and end times of a fishing event), catch (numbers by species, both retained and released), location, method and trip time (boat launched and retrieval times) of every fishing event during the 12-month period. Survey participants were then contacted regularly by interviewers during the 12-month period to collect their fishing information. Further details about the survey design, annual estimates and measure of uncertainty for elasmobranch species are available from 13,34–36.
Reporting of effort and catch data from charter-boat recreational fishing have been monitored with mandatory Tour Operator Returns (charter logbooks) in Western Australia since 2001. These data include fishing effort (number of clients on charter boat), catch (numbers by species, both retained and released) and location 37. For consistency with time series reconstructions (see below), only those years when the statewide surveys were conducted were considered in this study. As the Tour Operator data are from an assumed census, there are no measures of uncertainty associated with aggregated catch.
Data were adjusted to account for known sources of bias and to improve accuracy of estimated recreational harvest based on the following steps: 1) proportional allocation of catches from generic to specific taxa, 2) adjustment of estimated catch from private-boat recreational fishing to account for shore-based recreational fishing, 3) adjustment of estimated catch to account for PRM, and 4) conversion of estimated catch in numbers to weights. Following data adjustments, the recreational harvest time series were reconstructed. All analyses were done using the statistical package R 38. The computer code developed for data adjustments and analyses can be accessed at https://github.com/JuanMatiasBraccini/Git_Stock.assessments/blob/master/Catch.recons.Recreational.R.
Reapportioning of generic taxa.
Both the statewide surveys and Tour Operator Returns have some records for generic taxa that do not discriminate between species (e.g. ‘whaler shark’, ‘other shark’ or ‘hammerhead’). Therefore, for both data sets, the estimated harvest of sharks reported as ‘whaler shark’ was reapportioned using the overall proportion of whaler sharks reported at a species level by bioregion and year. Similarly, the estimated harvest of sharks reported as ‘other shark’ were reapportioned using the proportion of sharks reported to the species level. The estimated harvest of ‘hammerhead’ was reapportioned into scalloped (Sphyrna lewini) or smooth (S. zygaena) hammerhead according to the bioregion in which the harvest was reported based on the spatial distribution of these two species 39. Although great hammerhead (S. mokarran) also occurs in northern Western Australia, they are much less common than scalloped hammerhead so they were not considered in the reapportioning. Hence, the ‘hammerhead’ harvest reported in the Gascoyne and North Coast bioregions was allocated to scalloped hammerhead whereas the harvest reported in the West Coast and South Coast bioregions was allocated to smooth hammerhead. Finally, records of bull shark (C. leucas) were considered to be pigeye shark (C.amboinensis) due to their morphological similarities but much higher abundance of pigeye shark in the study area 40.
Calculation of shore-based recreational fishing.
Currently, there are no year- and region-specific estimates of shore-based recreational fishing harvest. Therefore, the estimated harvest from private-boat surveys was adjusted to account for shore-based recreational fishing using a private-boat to shore-based ratio (1:0.96) available for 2000-01 for elasmobranch species across Western Australia (DPIRD unpublished data). Thus, for each statewide survey year, shore-based recreational fishing harvest was calculated as 0.96 of the estimated catch of each elasmobranch species from private-boat recreational fishing.
Accounting for post-release mortality.
To account for PRM, species-specific PRM from line fishing was obtained from the literature. This information was only available for gummy (Mustelus antarcticus), school (Galeorhinus galeus), Port Jackson (Heterodontus portusjacksoni) and hammerhead sharks, and rays and skates 41–45. For species without available estimates, a conservative value of 30% was assumed (Table 2). The estimated harvest for each species was therefore the sum of the retained catch and the product of the released catch with the assumed PRM. This process was repeated for each year for estimated catches from both the statewide surveys and charter logbooks.
Conversion of numbers to weights.
Average weights for species retained by private-boat recreational fishers were determined from boat ramp recreational fishing surveys conducted in 2011/12, 2013/14, 2015/16 and 2017/18 46. For elasmobranch species, average weights were available for gummy, whiskery (Furgaleus macki) and whaler sharks, wobbegongs, and rays and skates. Hence, for requiem (with the exception of nervous and sliteye sharks), hammerhead and lamnid sharks, and species of the orders Orectolobiformes and Rhinopristiformes without available estimates, the average weight for whaler sharks was used whereas the average weight for gummy shark was used for the remaining species without available estimates (Table 2). For each species, an average weight across all survey years was applied to the estimated total harvest (the sum of private-boat, shore-based and charter-boat catches). Given differences in the reliability of estimated catches across statewide survey years 13, higher credence was given to estimates with lower relative standard error (RSE) by calculating an average weight, weighted by 1/RSE.
Reconstruction of time series.
For each species, the calculated average annual harvest (in weight) across survey years was then multiplied by a relative time series of residential population size and an estimated participation rate in recreational fishing (29.8%), which corresponds to the average participation rate between 2009-10 and 2018-19 47. The relative time series of residential population size was obtained by dividing the Estimated Residential Population of Western Australia between 1941 and 2018 (Australian Bureau of Statistics, https://www.abs.gov.au, accessed 25 November 2019) by 2011. The year 2011 was chosen because it was the first year of the statewide surveys. Sensitivity tests were done to test the assumptions made on average weights and PRM using ‘high’ and ‘low’ catch reconstruction scenarios (Table 1). Total shark and ray catches in WA (recreational and commercial) were plotted to provide a perspective of the reconstructed recreational harvest in relation to total extractions. Two time series of commercial shark and ray catches were presented: those reported in statutory catch and effort returns (i.e. landings only) and those reconstructed by 18 which also account for illegal and unreported catches and discards.
Patterns in species composition.
To determine patterns in species composition among bioregions and between private- and charter-boat fishing, multivariate analyses were done using the ‘vegan’ package 48. Non-metric multidimensional scaling (nMDS) on squared-root transformed data and Bray-Curtis dissimilarities was used to collapse the multivariate species composition into two dimensions. Similarity Percentage (SIMPER) Analysis was used to determine the contribution of each species to the dissimilarity between bioregions.