Simulating Habitat Preferences and Stock Assessment of Atlantic Fish Under the Pressures of Climate

The Northeast United States continental shelf spans from the Outer Banks of North Carolina to the Gulf of Maine. Fish stocks in this highly productive and economically important region are managed by the National Oceanic and Atmospheric Administration’s (NOAA) Northeast Fisheries Science Center (NEFSC) in Woods Hole, Massachusetts. Federal fishery biologists assess the health and abundance of each commercial fish stock using fishery-independent bottom trawl survey data that has been collected by NOAA throughout the region since 1963. I am currently contracted with NOAA’s NEFSC federal fishery scientists to consider the impact of climate change on future stock assessment.


Due to a combination of climate change and shifts in circulation, the Northeast United States continental shelf has experienced rapid warming in recent decades, resulting in a shift in spatial distributions of many species. Since stock assessment models rely on accurate descriptions of population dynamics and contemporary patterns of spatial abundance, there is concern that rapid undocumented changes in spatial distributions of species will bias future stock assessments. The implication of this is that the bottom trawl survey is actually sampling the population during a different life cycle stage than was originally assumed, which can lead to biased stock assessments. We are therefore interested in analyzing the impact of climate change on the accuracy of future stock assessment models as measured by NOAA’s ongoing bottom-trawl survey along the East coast.


I am modeling fish behavior and their response to environmental change by editing the R package MixFishSim to address our research question. Population dynamics in the model follow a modified two-stage Deriso-Schnute delay difference model, where the biomass in cell d in time step t+1 is given by


Recruitment dynamics in the model are defined using the Beverton-Holt relationship

The probability of movement from cell I to cell J is a combination of the distance between cells, the species-specific habitat preferences, and species-specific temperature tolerances, as defined by


An early task included relating spatial trawl data to the covariates depth and mean seafloor sediment to derive species-specific fish habitats in the model (see Image below).



Fish behavior in the model relies on integrating species-specific habitat preferences (above) with species-specific temperature tolerances to create species-specific habitat domains that develop as the temperature changes (see image below). Model parameters will be calibrated using NOAA’s extensive trawl survey data.




Using a temperature gradient in our simulation that increases over time we are able to simulate how the spatial locations of different species will respond under the pressures of climate change. We can then calculate the perceived abundance using typical survey based methods such as stratified mean and compare estimates to model based methods such as the Vector Autoregressive Spatio-Temporal (VAST) model. In both cases we can compare our estimates to found in the model (see final image below for preliminary result). Discrepancies will be accounted for in VAST by including environmental covariates in our models.