An observing system for the collection of fishery and oceanographic data

Fishery Observing System (FOS) was developed as a first and basic step towards fish stock abundance now-casting/forecasting within the framework of the EU research program Mediterranean Forecasting System: Toward an Environmental Prediction (MFSTEP). The study of the relationship between abundance and environmental parameters also represents a crucial point towards forecasting. Eight fishing vessels were progressively equipped with FOS instrumentation to collect fishery and oceanographic data. The vessels belonged to different harbours of the Central and Northern Adriatic Sea. For this pilot application, anchovy (Engraulis encrasicolus, L.) was chosen as the target species. Geo-referenced catch data, associated with in-situ temperature and depth, were the FOS products but other parameters were associated with catch data as well. MFSTEP numerical circulation models provide many of these data. In particular, salinity was extracted from re-analysis data of numerical circulation models. Satellite-derived sea surface temperature (SST) and chlorophyll were also used as independent variables. Catch and effort data were used to estimate an abundance index (CPUE-Catch per Unit of Effort). Considering that catch records were gathered by different fishing vessels with different technical characteristics and operating on different fish densities, a standardized value of CPUE was calculated. A spatial and temporal average CPUE map was obtained together with a monthly mean time series in order to characterise the variability of anchovy abundance during the period of observation ( October 2003-August 2005). In order to study the relationship between abundance and oceanographic parameters, Generalized Additive Models (GAM) were used. Preliminary results revealed a complex scenario: the southern sector of the domain is characterised by a stronger relationship than the central and northern sector where the interactions between the environment and the anchovy distribution are hidden by a higher percentage of variability within the system which is still unexplained. GAM analysis showed that increasing the number of explanatory variables also increased the portion of variance explained by the model. Data exchange and interdisciplinary efforts will therefore be crucial for the success of this research activity.