Model rankings of Poisson GLM for large pelagic species richness in the 95% confidence set of models and the relative importance of each predictor variable.

Models are ranked based on minimum Akaike Information Criterion corrected for small sample sizes (AICc). Predictors used at each model were: 1) time—time of deployment (morning 6—10 am/afternoon 1—3 pm); 2) camera depth—camera depth level (shallow 10m/deep 25m; 3) seamount depth—summit depth level (shallow 400m); 4) drift—drifting distance of each deployment during the effective recording time (soak time); 5) MPA distance—minimum distance of each seamount to the boundaries of Cocos or Galápagos Islands; 6) temp—mean temperature at the camera depth level and 7) chla—mean chorophyll-a concentration. Log of soak time (hours of effective recording) was used as an offset in the models. Degrees of freedom (df), maximum Log Likelihood (LL), degrees of freedom (df), difference of AICc of a given model to the model with best fit (ΔAICc) and relative model probability expressed as AICc weight (wAICc) are shown for each model. The sum of the wAICc was used to calculate the relative importance of each variable in the 95% confidence set of models following [79,122]. The function dredge from the MuMin Package in R version 3.6.3 was used to obtain this table. (XLSX)