Analysing and forecasting spatio-temporal abundance of Illex argentinus in the Falkland Islands zone based on oceanographic features

Illex argentinus, commonly known as the Argentine shortfin squid, holds significant ecological and economic importance in the Patagonian Shelf ecosystem. Previous studies found the spatial and temporal distribution of I. argentinus linked to oceanographic features. This study used an extensive time series dataset, state-of the art Generalized-Additive-Mixed-Effect Models and novel approach of LSTM Neural Networks (NN) to analyse and forecast I. argentinus spatio-temporal abundance measured as Catch per Unit Effort. Furthermore, we tested for any lag between environmental variables and temporal abundance, specifically, 6 months previous to sampling. Higher temperatures and micronekton abundances had overall positive effects on CPUE at the time of sampling, whereas 6 months earlier, they had negative influences. Increased number of eddies within the wider FI zone and the Falkland Current had a positive impact on CPUE, other areas and the 6-month delay had negative influence on CPUE with increased eddy density.

The overall best performing model was a NN model, which used temperature, zooplankton abundance and associated fronts for a weekly CPUE forecast with 3 weeks projection span.

Apart from confirming findings from previous studies, that I. argentinus abundance is dependent on temperature and ocean fronts, we found more relevant predictor variables, e.g. micronekton abundance, sea surface height and number of eddies per area, which are influencing temporal and spatial abundance of I. argentinus. We showed that NNs are capable of forecasting temporal abundance, despite complexity of oceanographic processes on the Patagonian Shelf, coupled with complex migration patterns of I. argentinus.