A correction factor for, and its application to, visual censuses of littoral fish

Abstract

Visual estimation techniques were used between October 1976 and May 1977 on the fish fauna of an enclosed mid-intertidal pool in the eastern Cape, South Africa, to test the usefulness and accuracy of this non-consumptive census method. In January, 1977, two pools were censused for fish, then poisoned using the ichthyocide ‘pronoxfish’ and all fish collected. Forty species in 21 families were obtained from the two stations. The accuracy of visual estimates varied with the species of fish. Secretive and cryptically coloured types, as well as species which Inhabit crevices, were underestimated, between 0 and 86% being counted. Other fish with protective colours but that occur more openly and do not react adversely to divers, as well as schooling species, were also underestimated, but up to 100% of the total were observed. Those which occur singly or in small groups in the water column were counted more accurately, with 57 to 100% of the actual number being seen. Correction factors were calculated from these data and applied to a census taken of one pool in May. Comparisons of corrected and actual numbers Indicate that these factors were relatively accurate for non-secretive species. The factors will, however, vary for each species, from observer to observer, depending on their personal experience, and from area to area.