NMFS Trawl Survey data was used to fit species distribution models and the resulting modeled predictions for presence/absence and abundance.

We analyzed the influence of environmental characteristics on the spatial distribution of our eight focal species using data collected by the Northeast Fisheries Science Center (NEFSC) spring (March-May), and fall (September-November) bottom trawl surveys along the Northeast US Shelf (65-75°W longitude, and 35-45°N latitude) for the period 1968-2014. Sea surface temperature, sea bottom temperature, and depth were sampled concurrently with trawl samples. We used sediment grain size as a measure of substrate type using existing data layers from the Nature Conservancy. Surveys were trimmed to strata that were sampled in at least 43 of the 46 years. Data from the fall and spring surveys 1968-2014 were combined in order to fit species distribution models (SDMs) using a generalized additive model (GAM) fit separately for each species. Presence or absence of species x in haul location i in year y was modeled using a logistic model with a binomial error distribution and a logit link function. The probability of species occurrence in each haul was modeled as an additive function of the five environmental variables and regional species biomass: haul-specific observations of sea surface temperature, sea bottom temperature, depth, sediment grain size, and average region-wide biomass (kg/tow) of species x in year y in the season in which the haul was conducted. Penalized regression splines were fitted using the “gam” function in the mgcv package in R.

In addition to examining changes in overall range size, we also predicted historical and projected biomass using a delta-lognormal GAM that combines the predictions of the presence/absence model with that for biomass when present.

Range size and species overlap wer calculated using the R-file species_overlap_BCO.R available in the "Supplemental Documents" section on this page.