RESUMO
Remote underwater videos are widely employed to assess the structure and composition of reef fish assemblages but the sampling effort employed on each survey differs considerably, indicating that both the number of assessments and video length could be optimized. We searched for this optimal sampling effort in remote video samples to conduct rapid assessments of community composition and discussed the relation between number of replicates and video length, and how it impacts the method's efficiency to characterize species assemblages. Remote video recordings from tropical reefs in northeastern Brazil were used to investigate how fish species richness and composition builds across time and number of assays. Videos as short as 5 min successfully recorded species richness, requiring about five repetitions to record most species that compose 80% of the total biomass. Recording species composition required even less time in these reefs, setting a minimum of 3 min with the same five videos. By comparing the detected richness per analysed time unit, we found several shorter videos recorded for more species than a few longer videos, indicating that increasing the sampling coverage in the reef area might be better than just extending the video length for rapid assessments.