Tuna schools / aggregations in surface longline data 1993 – 98

Exploratory data analysis of a highresolution (hook-by-hook), 6-year time series (1993– 98) of observed longline catch data for tunas was used to investigate fine-scale spatial patterns along individual sets that may be indicative of social behaviour (i.e., schooling) and/or the response of individual fish to favourable extrinsic conditions (i.e., aggregation). Methods of spatial data analysis (i.e., nearest neighbour analysis) that have previously been applied in various other sciences (e.g., forestry and astronomy) were used. Results indicate strong clustering of individual tunas at characteristic scales within the set. Mean Nearest Neighbour Distances (NNDs) were between 100 and 200 m, compared with NNDs of 200–700 m predicted by a heterogeneous Poisson process on the same spatial domain. The results suggest that these adult tunas were either schooling or aggregating at the time of capture; this may therefore be related either to social behaviour or to sub-mesoscale oceanographic features. An aggregation index was derived from the NNDs, giving a classification method that may be used for similar data and the development of empirical models attempting to relate patterns in fish catch distributions to environmental variables. The success of such models will ultimately depend on elucidating the ecological processes reflected in oceanographic features at biologically meaningful spatial scales.