Influence of instantaneous variation on estimates of coral reef fish populations and communities

The magnitude of different sources of variation in coral reef fish abundance data needs to be known if temporal changes in population and community data are to be correctly estimated. A particularly important missing component of the variability is the 'instantaneous' change in fish, largely caused by the interaction between fish movement and observer recognition. This variation occurs at a time scale less than that influenced by the focus of previous studies, including time of day, tides, migration, or birth and death processes. Without this measure of variance, estimates of temporal change are confounded. To determine the magnitude of this instantaneous variance, belt-transect visual counts of damselfish, surgeonfish, and parrotfish were conducted during a short interval at midday during neap tides over consecutive days in the calm season and compared to similar samples in 1992 ano 2003. Within-site, or our estimate of instantaneous variation, was the greatest source of variability for the whole assemblage and for the surgeonfish/parrotfish group but not for damselfish. Direct between-year comparisons produced estimates of population change over time that were twice as high as those derived by an indirect method where the instantaneous spatial component was subtracted from the total variation. Because the inherent spatial component of variability makes it difficult to detect site change over time, we recommend sampling designs that use random sampling and have greater statistical power to detect change. Furthermore, aggregate metrics, such as numbers of species or density at the family, community, or the functional group, will have greater potential to detect change for sample sizes typical of coral reef studies. Otherwise, when life history traits and species level change are important, high replication will be required.