9781888569810-ch5

Shark Nursery Grounds of the Gulf of Mexico and the East Coast Waters of the United States

Spatial Delineation of Summer Nursery Areas for Juvenile Sandbar Sharks in Chesapeake Bay, Virginia

R. Dean Grubbs and John A. Musick

doi: https://doi.org/10.47886/9781888569810.ch5

Abstract.—The lower Chesapeake Bay is the largest summer nursery for sandbar sharks Carcharhinus plumbeus in the western Atlantic. The objective of this study was to define essential fish habitat for juvenile sandbar sharks in this estuary. The longline survey conducted by the Virginia Institute of Marine Science was expanded from 1990 to 1999 to include ancillary stations throughout the Virginia portion of Chesapeake Bay to delineate this nursery spatially. We analyzed catch per unit of effort data from 83 stations as a function of nine physical and environmental variables using tree-based regression models. The highest abundance of juvenile sandbar sharks was predicted where salinity was greater than 20.5 (practical salinity scale) and depth was greater than 5.5 m. The models also suggested that dissolved oxygen concentration may influence shark distribution. To increase applicability of the models to management practices, we introduced distance to the mouth of the estuary as a surrogate variable for salinity. The models estimated that the highest abundance of sharks was in areas less than 34.5 km from the mouth of the estuary and in depths greater than 5.5 m. The areas of the estuary that met the criteria of the models, based on the threshold values of the variables, were mapped spatially in a geographic information system. The resulting response surfaces were interpreted to represent essential nursery habitat for sandbar sharks in Chesapeake Bay. Both models performed very well using several dependent and independent measures to estimate their classification and predictive ability. We used logistic regression with presence/absence data to validate the tree models. The logistic regression models agreed very well with the tree-based regression models, selecting the same variable combinations to predict sandbar shark presence and absence.