Publication Abstract

Meta-assessment for elasmobranchs using Bayesian networks and dietary data

T.R. Hammond and J.R. Ellis

We estimated historic biomass trends for elasmobranchs of the Irish Sea using a food web derived from published dietary data. With a new technique we call 'meta-assessment', we incorporated into our estimates the results from virtual population analysis (VPA) for Irish Sea demersal stocks. We employed two Bayesian networks in making these estimates. To assess the accuracy of our networks, we used them to estimate historic trends in cod and plaice biomass using trends in fishing effort and in the biomass of sole, whiting and Nephrops norvegicus. We compared predicted annual trends to those derived from VPA. Our predictions about cod and plaice trends were accurate 72% and 59% of the time, respectively. We also compared elasmobranch biomass trends estimated from an annual trawl survey to corresponding network predictions. For the 4 elasmobranch species with the lowest survey index CV, we predicted survey trends correctly 61% of the time. Our assessment results suggest that of 11 elasmobranch species considered, the angel shark (Squatina squatina) decreased the most over the period from 1987 to 1997. Survey results also suggest this species has declined. When we applied our approach to the common skate (Raja batis) over the period 1965-1978 (during which time the skate nearly disappeared from the Irish Sea), a marked decline in biomass was predicted. We conclude that meta-assessment can provide timely and cost-effective identification of threatened stocks.

Reference:

T.R. Hammond and J.R. Ellis (2002) Meta-assessment for elasmobranchs using Bayesian networks and dietary data. Ecological Indicators, 1(3):197-211.