Evaluation of environmental conditions as predictors for mako shark CPUE using generalized linear mixed modeling and quantile regression

Environmental conditions were evaluated for their influence on catch per unit effort (CPUE) of shortfin mako (Isurus oxyrinchus). Catch rates of shortfin mako were calculated from the US pelagic longline observer program (1992-2016) using a generalized linear mixed model (GLMM) with a delta-lognormal approach. The GLMM analysis included consideration of the following environmental variables as predictor variables: sea surface height, sea surface temperature, and bathymetry. The addition of environmental predictor variables resulted in an index that spans 2003-2012. The final index was used to predict average catch per unit effort (CPUE) based on environmental conditions. The two portions of the delta-lognormal approach retained different suites of variables with sea surface temperature and bathymetry retained to predict proportion of positive sets while bathymetry was retained to predict the CPUE of positive catches. Quantile regression was also performed to evaluate whether environmental variables can predict spatial areas with high CPUE. As with the delta approach, environmental data were used to predict conditions that favor high CPUE. Maps generated from both the approaches will later be used for determining mako shark habitat for a spatial management strategy evaluation.