Abstract

We compare the accuracy of an interference visual census technique (IVC), in which dismantling of the habitat is performed, to traditional underwater visual census (VC) and anaesthetic census. We compare the performance of these techniques applied to a temperate cryptobenthic fish assemblage using two strategies: sampling over the whole depth extent of the rocky bottom, and stratified sampling over the main microhabitats present at the study site. T h e number of species encountered was lower using the traditional VC. Fish density estimates were significantly higher using the interference technique compared to the traditional VC, in the transect strategy. These differences were larger for clingfishes and some gobies which occurred preferably under cobble and small rocks. No differences were found when comparing the IVC and anaesthetic census in the habitat strategy, for each microhabitat considered. We conclude that dismantling the habitat increases the performance of the visual census technique and is therefore a valuable approach when applied to temperate cryptobenthic fish assemblages.