MEPS 299:239-255 (2005)  -  doi:10.3354/meps299239

Winter distribution of blue crab Callinectes sapidus in Chesapeake Bay: application and cross-validation of a two-stage generalized additive model

Olaf P. Jensen1,3,*, Ralf Seppelt2,4, Thomas J. Miller1, Laurie J. Bauer1

1University of Maryland Center for Environmental Science Chesapeake Biological Laboratory, PO Box 38, 1 Williams St., Solomons, Maryland 20688, USA
2Technical University Braunschweig, Institute of Geoecology, Department for Environmental System Analysis, Langer Kamp 19c, 38106 Braunschweig, Germany
3Present address: University of Wisconsin Center for Limnology, 680 N. Park St., Madison, Wisconsin 53706, USA
4Present address: UFZ Centre for Environmental Research Leipzig, Department for Applied Landscape Ecology, Permoserstraße 15, 04318 Leipzig, Germany

ABSTRACT: We present a 2-stage generalized additive model (GAM) of the distribution of mature female blue crab Callinectes sapidus in Chesapeake Bay based on data from a fishery-independent winter dredge survey. The distribution and abundance of blue crabs was modeled as a flexible function of depth, salinity, water temperature, distance from the Bay mouth, distance from submerged aquatic vegetation (SAV), and bottom slope for each of the 13 yr of data available. Depth, salinity, temperature, and distance from the Bay mouth were found to be the most important environmental determinants of mature female blue crab distributions. The response curves for these variables displayed patterns that are consistent with laboratory and field studies of blue crab/habitat relationships. The generality of the habitat models was assessed using intra- and inter-annual cross-validation. Although the models generally performed well in cross-validation, some years showed unique habitat relationships that were not well predicted by models from other years. Such variability may be overlooked in habitat suitability models derived from data collected over short time periods.

KEY WORDS: Blue crab · Chesapeake Bay · GAM · Cross-validation · Habitat suitability model

Full text in pdf format