Multiple regression models of δC and δN for fish populations in the eastern Gulf of Mexico

Multiple regression models were created to explain spatial and temporal variation in the δC and δN values of fish populations on the West Florida Shelf (eastern Gulf of Mexico, USA). Extensive trawl surveys from three time periods were used to acquire muscle samples from seven groundfish species. Isotopic variation (δCvar and δNvar) was calculated as the deviation from the isotopic mean of each fish species. Static spatial data and dynamic water quality parameters were used to create models predicting δCvar and δNvar in three fish species that were caught in the summers of 2009 and 2010. Additional data sets were then used to determine the accuracy of the models for predicting isotopic variation (1) in a different time period (fall 2010) and (2) among four entirely different fish species that were collected during summer 2009. The δNvar model was relatively stable and could be applied to different time periods and species with similar accuracy (mean absolute errors 0.31–0.33‰). The δCvar model had a lower predictive capability and mean absolute errors ranged from 0.42 to 0.48‰. δN trends are likely linked to gradients in nitrogen fixation and Mississippi River influence on the West Florida Shelf, while δC trends may be linked to changes in algal species, photosynthetic fractionation, and abundance of benthic vs. planktonic basal resources. These models of isotopic variability may be useful for future stable isotope investigations of trophic level, basal resource use, and animal migration on the West Florida Shelf. & 2014 Elsevier Ltd. All rights reserved.