EVALUATING THE POTENTIAL FOR USING SPECIES PRESENCE DATA COLLECTED BY COMMERCIAL FISHERMEN FOR SPECIES DISTRIBUTION MODELING IN THE GULF OF MAINE

Fishermen and scientists have noted that fish species distributions are changing along the Northeast Continental Shelf in response to local climate variability and large scale warming trends. NEFSC semiannual trawl survey data have been modeled to demonstrate these distributional and poleward shifts in species assemblages, but New England groundfishermen argue that management structures are slow to reflect these changes and that species distribution modeling of trawl survey data has failed to accurately characterize them. Fishermen have requested an outlet to submit information to scientists regarding where they are encountering shifting species while fishing. Data collected by fishermen themselves may be suitable for gaining a deeper understanding of moving fish populations. In response, the Gulf of Maine Research Institute and Island Institute have proposed to create a phone application that would allow fishermen to take photographs of species they encounter and log the location at the time of the photograph. The purpose of this study is to evaluate whether data submitted though such an application would be useful to scientists in modeling distributional shifts of commercially important species. I created a proxy for data submitted by commercial fishermen on red hake and used it as input data in maximum entropy models to determine whether such data would be comprehensive enough or spatially distributed in a manner that would make it suitable for species distribution modeling. All “citizen science” datasets of varying sample size and spatial distribution produced models with AUC values greater than 0.75, while the baseline models using presence data from the trawl survey had lower AUC values around 0.7. Our results indicate that concentrated data from fishing areas can lead to underestimation of species presence probability in regions where no fishing occurred. This happened because the fishing data used to generate random points from fishermen only covered part of the study region. Findings of this study suggest that presence data from fishermen are suitable for presence-only maximum entropy species distribution models, but actual Citizen science data collection and further modeling should include data from fishermen in the Mid-Atlantic to produce more robust and reliable models.