Gear selectivity and sample size effects on growth curve selection in shark age and growth studies
Thorson, James T., and Simpfendorfer, Colin A. (2009) Gear selectivity and sample size effects on growth curve selection in shark age and growth studies. Fisheries Research, 98 (1-3). pp. 75-84.
|PDF (Published Version) - Repository staff only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader|
View at Publisher Website: http://dx.doi.org/10.1016/j.fishres.2009...
The effects of gear selectivity and sample size on growth model selection for shark length-at-age data was investigated using simulated data sets for dusky (Carcharhinus obscurus) and Atlantic sharpnose (Rhizoprionodon terraenovae) sharks. Simulated data sets were generated using four different individual growth functions (von Bertalanffy, Gompertz, logistic and Schnute) with five different sample sizes and seven different sampling gear selectivity functions. Five growth models were fit to each data set and the Akaike information criterion (AIC) and AIC weights used to evaluate which fit best to the data. The accuracy of each of the fitting models was also determined and compared with estimates derived from AIC multi-model inference (MMI). The results demonstrated that no one growth model outperformed all other models in all situations, with the two-parameter von Bertalanffy model being selected most often by AIC. There were clear differences between the fitting models that were ranked the highest based on AIC selection or AIC weights and those based on models accuracy of length and growth parameter estimates. In most situations, the two-parameter von Bertalanffy model provided the least accuracy in parameter recovery, while the Schnute model was the most accurate individual model in most situations. The AIC MMI approach improved estimation accuracy over most individual models, except at very low sample sizes. These simulation results suggest that sample sizes of 200 are required to consistently achieve good accuracy for growth parameters, and that sampling gear selectivity has clear effects on which growth models fit best to the data. The use of the AIC MMI approach is recommended, as it provides the most robust estimates of growth parameters.
|Item Type:||Article (Refereed Research - C1)|
|Keywords:||individual growth model; model selection; length-selective sampling; Akaike information criterion; multi-model estimation|
|FoR Codes:||07 AGRICULTURAL AND VETERINARY SCIENCES > 0704 Fisheries Sciences > 070402 Aquatic Ecosystem Studies and Stock Assessment @ 100%|
|SEO Codes:||83 ANIMAL PRODUCTION AND ANIMAL PRIMARY PRODUCTS > 8302 Fisheries - Wild Caught > 830204 Wild Caught Fin Fish (excl. Tuna) @ 100%|
|Deposited On:||11 Feb 2010 09:32|
|Last Modified:||20 May 2013 01:03|
Last 12 Months: 0
|Citation Counts with External Providers:||Web of Science: 80|
Repository Staff Only: item control page