Assessing variability in standardised harvest rates from shore-based recreational fishing surveys
Date of Award
Master of Science (Environmental Management)
School of Science
Field of Research Code
Recreational fishing is important in Australia and world-wide, providing both social and economic benefits to communities and stakeholders. The popularity of recreational fishing requires both stewardship and monitoring for sustainable conservation of resources. In Western Australia, recreational fishing is highly valued with 30% of the population participating each year, and with more than 50% of recreational catches being harvested from the shore, the development of an index of harvest rates from shore-based recreational fishing would be a valuable addition to stock assessments for some species. However, the large spatio-temporal scope of this fishery poses logistical and financial challenges to adequate monitoring. Harvest rates can provide an ongoing index for a weight of evidence approach capable of detecting variability in the thresholds set in harvest strategies. However, nominal harvest rates can mislead sustainability and stock assessments, and subsequent management decisions. Standardising harvest rates has proven to improve estimates of harvest rates for both commercial and recreational fisheries, however, the use of this technique in shore-based recreational fisheries are few and have not yet been applied to fisheries data collected in Western Australia. The focus of this study is on assessing generalised linear models (GLM) that can improve the harvest rates from shore-based recreational fishing in the Perth metropolitan region by standardising harvest rates according to influential factors. This study applied GLMs to analyse catch and effort data from a roving creel survey of shore-based recreational fishers in the Perth metropolitan area to determine the impact of factors (survey year and month, targeting, fishing platform, fishers avidity, time of day and day type) on estimates of harvest rates for key nearshore species. Due to the numbers harvested, and the high frequency of fishers targeting them, the species of interest were Australian herring (Arripis georgianus), School whiting Sillago spp, Southern garfish (Hyporhamphus melanochir), Western Australian salmon (Arripis truttaceus) and Silver trevally (Pseudocaranx georgianus).
GLMs were assessed via five-fold cross-validation for five distribution families, including the zero-altered gamma (ZAG) and Tweedie models, to account for differences in catch distributions among species and the high prevalence of fishing events where no catch was reported. A single model, with specific distribution and significant factors, was selected for each species according to its performance in relation to correlation, bias and root mean squared error. Model fit and significant variables varied among species with Lognormal, ZAG and Tweedie models having the best performance overall. The distribution family for the preferred model varied among species with Tweedie most appropriate for Australian herring and School whiting, ZAG for Southern garfish and Lognormal for Western Australian salmon and Silver trevally. The Gaussian and Gamma distributions were not suitable for any species.
With regards to the variables that contributed to the preferred models, targeted was significant for all five species, the fishing platform was significant for Southern garfish and Western Australian salmon, and time of day was only significant for Australian herring. Other factors, such as avidity, day type and month, did not contribute significantly to any models. As most stock assessments and harvest strategy reviews require comparison of harvest rates over time, year remained a fixed factor in each model, and was significant for all species, except School whiting.
Harvest rates for each species were compared among four survey years according to the preferred standardised models. Harvest rates for Australian herring (Tweedie model) were significantly higher in 2010 (1.75 ± 0.21SE, fish per fishing party per day) compared with 2016 (0.28 ± 0.06SE) (p<0.001). Although harvest rates for School whiting ranged from 0.14 ± 0.04SE in 2010 to 0.31 ± 0.07SE in 2014, there were no significant differences among years (p=0.040). Harvest rates for Southern garfish significantly decreased from 2010 (0.45 ± 0.07SE) to 2016 (0.003 ± 0.002SE). Harvest rates for Western Australian salmon significantly increased from 0.01 ± 0.00SE in 2010 to 0.08 ± 0.02SE in 2016 (p<0.001). Harvest rates for Silver trevally was slightly higher in 2010 (0.034 ± 0.047SE) and stayed relatively stable in the subsequent survey years 2014 (0.006 ± 0.037SE) (p<0.001), 2015(0.007 ± 0.037SE) (p<0.001), and 2016 (0.003 ± 0.037SE).
The declines in harvest rates for Australian herring and Southern garfish were consistent with stock assessments, for which the stock status is indicated as inadequate. This has prompted management actions with a reduction in the daily bag limit of Australian herring from 30 to 12 fish per fisher in 2015, and a fishing closure for Southern garfish in the Perth metropolitan area in 2017. Ongoing monitoring of harvest rates for these species will provide an indicator of the impacts of these management changes, and assessment of recovery relative to previous higher levels of shore-based recreational fishing. This study has demonstrated the importance of standardising harvest rates for key nearshore species in the Perth metropolitan shore-based recreational fishery. The development of preferred models for each species, with specific distribution and explanatory variables, provides an approach that can be used to develop cost-effective survey designs, improve estimates of harvest rates, and inform future stock assessments and management of these valuable fisheries.
Access to this thesis is embargoed until 26 February 2019. At the expiration of the embargo period, access to the full text will be restricted to current ECU staff and students by author's request.
Email queries to email@example.com
Tate, A. (2017). Assessing variability in standardised harvest rates from shore-based recreational fishing surveys. Retrieved from http://ro.ecu.edu.au/theses/2057