The efficiency of a seine net

We present a method to predict the capture efficiency of a 25-m, 5-mm mesh seine net as a function of fish size and taxon from a diverse fish communit y. This allows true abundance and size distribution to be estimated from ob served catches. Predicted capture efficiency from an empirical model of fie ld calibrations from the Amazon River floodplain was a positively skewed, u nimodal function of fish length, whose magnitude depended on method of sein e operation and fish taxonomic group. Capture efficiency is the product of efficiency of encirclement as the net is laid (which decreases with increas ing fish size) and efficiency of retention as the net is hauled (which incr eases with increasing fish size). Retention was determined by modeling mark -recapture data. Dividing observed capture efficiency by this retention yie lded empirical encirclement efficiency, which was then compared with encirc lement efficiency determined from a simulation model of fishes' evasive beh avior. The simulation accounts for the fishes' swimming speed relative to t he speed of deployment of the seine, threshold distance (how close the dist urbance from laying the net must be to initiate evasion), appraisal time (h ow long a fish continues evasive behavior when it moves outside the thresho ld distance), and the directionality of evasive movements. Simulated result s of encirclement efficiency corresponded to empirically based predictions within plausible ranges of the simulation variables above, although for fis h of length exceeding about 50 cm there is a high coefficient of variation in captured biomass due to small numbers and low catchability. We conclude that the method can be used for a wide range of conditions to convert seine capture data to unbiased estimates of abundance and size distribution, but that empirical determinations will still be needed for different net speci fications and sampling conditions.