Report - Aquatic Environment and Biodiversity (AEBR)
News & Resources, Publications, Fishing, Fisheries
New Zealand long-line fishery data were fitted with Boosted Regression Tree models to examine the extent to which biological and physical environmental forces might predict the distribution of highly migratory species.
Models fitted to presence/absence generally performed better than those predicting the number caught. Smaller datasets, or those with limited spatial distributions had poorer predictive ability.
Sea surface temperature, latitude and subarea were most influential in Binomial models