Comparison of fish-community size spectra based on length frequencies and mean lengths: a note

Abstract

Estimates of fish-community size spectra are promising indicators of the impact of fishing on fish assemblages. Size spectra consist of logarithmic graphs of abundance plotted against fish body size. Size spectra may either be constructed from length frequency data or estimated from the mean sizes and abundances of the species in an assemblage — very often mean sizes are the only historical data available. Changes in the slopes of size spectra are interpreted to indicate changes in the relative abundances of small vs large fish, whereas changes in intercept (height) suggest changes in the overall abundance of the fish assemblage. A comparison of the size spectra of linefish catches of the Cape region, South Africa, for the period 1986–1998 reveals that statistics of the size spectra calculated from mean length data are significantly larger (heights) and shallower (slopes) than those calculated using length frequencies (paired t-tests, p < 0.001). Therefore, use of mean lengths in size spectra overestimates the overall abundance of the community. Mean length was also found to overestimate the relative abundance of larger fish in the assemblage, thereby underestimating the effects of fishing. In a time-series of size spectra, it is therefore necessary to use one method consistently for comparative purposes.