Predicted Annual Probability of Observing at least One Black-legged Kittiwake

NOTE: Two data download links are provided. The first includes the data described below as a geographic point layer and as a .csv file. The second link is a data package containing: the annual probability of observing one individual, the annual probability of encountering a large flock, and the monthly probability of observing one individual, for the full set of 24 species (in .csv format), and the associated report “Mapping the distribution, abundance and risk assessment of marine birds in the Northwest Atlantic.” To improve display of this data on Data Basin the point data was converted to a raster grid.

This map depicts the mean predicted probability of observing at least one individual Black-legged Kittiwake (Rissa tridactyla) (NOTE: the name of this dataset changed - formally Black Tern - the name is now correct and matches the data which has remained the same) throughout the Northeastern and Mid-Atlantic coast of the United States based on the "double-hurdle model" of the probability distribution of avian count data. The model was developed to better handle the excessive amount of observed zeros and the few but relatively extreme counts simultaneously within a single framework. Each location represents the centroid of a .041 x .041 decimal degree grid (approximately 4x4 km) and the value is a probability from 0 to 1. The model is spatially-explicit: the value at each cell (location) depends not only on covariate information available at that cell, but on information available in neighboring cells. This map was developed as part of a project to predict the distribution and abundance of marine birds in the Northwest Atlantic. Annual and monthly maps were developed for 24 species of marine birds, illustrating both predicted probability of observing at least one individual bird and predicted probability of observing at least one large count of individuals. The models were based on data from the Atlantic Seabird Compendium of continuous time strip surveys for the time period July 2002 to November 2010. Four biophysical covariates were used to predict seabird occurrence and abundance: sea surface temperature, ocean depth, chlorophyll-a concentration, and distance to shore.

The project was led by North Carolina State University; also involved the NOAA National Ocean Service, the Biodiversity Research Institute, and CSI – City University of New York; and was sponsored by the North Atlantic Landscape Conservation Cooperative. Detailed documentation on the development of all of the species datasets are available at: https://www.fws.gov/science/catalog. The documentation includes a list of all the species for which models have been or are being developed and discusses limitations and constraints for using the datasets.