Sampling and Statistical Considerations for Hydroacoustic Surveys Used in Estimating Abundance of Forage Fishes in Reservoirs

Abstract

An understanding of the spatial distribution of forage fish resources is required to make informed fishery management decisions. We used mobile hydroacoustics to assess the distribution and abundance of forage fish in Badin Lake, a reservoir in central North Carolina. By sampling a series of cross-channel and longitudinal transects and analyzing the data using geostatistics, we characterized both large- and small-scale spatial patterns in forage fish density. Forage fish were observed in higher densities in upstream regions of the reservoir and were seen only in surface waters during July 2000 owing to the existence of a strong thermo–oxycline and in two layers (surface and near bottom) during mixed conditions in December 2001. We observed differences in the scale of patchiness (200–700 m) in forage fish distribution depending on the region of the reservoir where sampling took place, and we infer that these patterns are governed by prevailing limnological conditions. Modeling the spatial variation in the acoustic data using geostatistics resulted in similar average densities (July 2000: 0.56 ± 0.28 (mean ± SD) fish/m²; December 2001: 0.57 ± 0.49 fish/m²) and improvements in the precision of abundance estimates based on approximated variance (July 2000: 7.05 × 10⁶ ± 8.30 × 10⁵ fish; December 2001: 7.07 × 10⁶ ± 2.10 × 10⁶ fish) when compared with arithmetic averaging and extrapolation (July 2000: 0.60 ± 0.61 fish/m², 7.5 × 10⁶ ± 4.61 × 10⁶ fish; December 2001: 0.60 ± 0.69 fish/m², 8.00 × 10⁶ ± 6.34 × 10⁶ fish). We found that sampling along longitudinal transects is a more efficient way to characterize the spatial patterns of forage fish distribution and to estimate systemwide abundance and biomass than using data collected with both a longitudinal and cross-channel sampling design in this system.