Use of Recovery Probabilities Can Improve Sampling Efficiency for Throw Traps in Vegetated Habitats

Abstract

Throw traps are commonly used to assess small fish and macroinvertebrate (SFI) communities in aquatic habitats. Traditionally, numerous passes with a bar seine are used to remove all SFIs from within the throw trap, providing density estimates that are comparable across habitat types. However, similar information can be gathered with fewer seine passes than are traditionally used, thus allowing more sampling in a given field effort. The appropriate number of seine passes to meet study objectives can be determined if habitat-specific recovery probabilities are estimated. We investigated how SFI recovery probabilities within throw traps varied among five submerged aquatic vegetation habitat types and how these recovery probabilities can be used to determine the number of seine passes necessary to assess and detect differences in SFI relative abundance. Using a 1-m² throw trap and bar seine, we found that estimates of SFI recovery probability were high (75–80%) and similar across habitats. Because of the high recovery probabilities, the use of about three bar-seine passes produced accurate estimates of true SFI density within a throw-trap sample, and additional bar-seine passes did not substantially improve the power to detect decreases in SFI density. These results are applicable to studies that use throw traps to sample SFIs in submerged aquatic vegetation habitats, and they illustrate how improvements in sampling efficiency can be obtained by decreasing the number of passes used per sample.

Received January 25, 2010; accepted December 4, 2010

ACKNOWLEDGMENTS

Funding in support of this research was provided by the State Wildlife Initiative Grant Program of the Florida Fish and Wildlife Conservation Commission and by the University of Florida. We are also grateful for the field and laboratory assistance provided by a number of people, including Morgan Edwards, Brandon Baker, Ashley Williams, and Andrew Dutterer. Finally, we thank Lew Coggins for assistance with the multinomial recovery probability estimation procedure.