Target strength (TS) may be estimated as a function of fish length (L) by comparing
in situ acoustic measurements with the fish size distribution from trawl catches. The
acoustic and fishing data are directly comparable only if they are samples from the same
population, but the TS measurements depend on the detection of isolated targets which may
not be representative of the ensonified population. Errors may occur through the false
detection of multiple targets, and the size-dependence of schooling density means that
larger fish are more likely to be detected. We show that such effects can result in the
estimated TS being biased high. We describe a stochastic method for estimating the
constant b in the formula TS=20 Log (L)-b, by a least-squares fitting procedure. It is
assumed that the echo-amplitude has a Rayleigh distribution. The method has been applied
to data from surveys of cichlids in Lake Malawi. From 33 independent measurements of
Diplotaxodon spp. around 15 cm in length, b is 67.4 ± 0.2 dB.