| Summary: | Herein we propose a nested boostrap scheme to explore di erent sources of variability in discard data. The proposed tool resamples on trips, hauls within trips and lengthclasess within hauls to include all sources of variability identified in the Spanish DCR sampling program scheme. Megrim (Lepidorhombus wiffiagonis) and Hake (Merluccius merluccius) discard data collected in 2008 from OTB-51 metier operating in ICES Division VI - VII, and Mediterranea DEF-Southern metier operating in the GSA 6S area were considered as case studies. Bootstrap Error and Coeficients of Variation associated to mean discards per trip were obtained and compared with asymptotic estimates. Furthermore, we use the bootstrap to quantify the contribution of the given sources of variability to global variability, and to assess simulated sampling scenarios differing each others in sampling effort. The last application of this method is for detecting outliers in 2008 sampled units (Trips). The good performance of the bootstrap method validates its use to obtain reliable error estimates in further regression and/or classiffication studies on discard data.
|
|---|