A comparison between two methods for surveying reef fish in St. Eustatius
Coral reef ecosystems are considered the most productive and diverse of all marine systems. However, these are being dramatically degraded due to different anthropogenic impacts, such as coastal pollution, climate change, and overfishing. The latter is the main cause of the decline of reef fish species which play important roles on the well-functioning and health of coral reefs. The fish families Acanthuridae, Scaridae, Serranidae, and Lutjanidae have an important ecological and commercial role in the wider Caribbean reefs and are considered key families to determine the health status of coral reefs. In order to properly monitor and survey reef fish assemblage scientists and managers have to choose the appropriate methodology. Underwater Visual Census (UVC) has been popularly used to survey fish species richness and abundances as it is a non-destructive technique, it is accessible and repeatable. However, a shift towards video-based methodologies is occurring as developments on video technology offer clear advantages to improve fish surveys. An example is the stereo-Diver Operated Video (sDOV) which is based on two cameras set in stereo to provide three-dimensional images and thus accurately and precisely estimate fish lengths and in turn fish biomass. This methodological shift requires studies to contrast different techniques and to appropriately recognize changes in fish density and biomass patterns. This thesis aims to investigate whether UVC and sDOV estimates of fish densities, biomass, and length can be used interchangeably in surveys performed in the Dutch Caribbean island of St. Eustatius. Additionally, it will study whether different UVC observers estimate fish density, biomass and length differently and if two different habitat complexities support different fish densities, biomass and lengths. From February 2016 to April 2016 twelve sites with five transects (30m x 2m) per site were simultaneously surveyed by an UVC observer, counting and estimating the total lengths (TL) of key fish families, and a sDOV operator, recording the whole transect and who estimated the three metrics afterwards. Forty out of sixty transects were surveyed by Matt & Paula (MP), and twenty by Steve & Paula (SP) changing the UVC observer. Fish data obtained by both methods was tested for mean density, biomass and length differences as well as inter-calibrated by regression Model II. Additionally, data was analysed to detect for an UVC observer effect and a habitat complexity effect. Results showed that inter-calibration between UVC and sDOV was possible for overall fish, surgeonfish, and parrotfish density estimates despite showing mean fish density differences. However, the carnivorous families were subjected to high variability and thus decreasing the reliability of the inter-calibration model. Moreover, UVC observer was found to differently estimate overall fish densities and surgeonfish biomass and length which can be explained by the difference in training background from both UVC observers. Finally, habitat complexity was not found to affect fish density, biomass and length estimates but the habitat complexity categories surveyed did not differ much. These results demonstrated that method interchangeability should be carefully considered as not all fish families are similarly surveyed by both methods. It strongly suggests further studies to improve the data comparability and to obtain more results mainly for the carnivorous families and to obtain larger ranges of length estimates. It strongly suggests to continue investigating differences between UVC observers as it is well known to affect fish estimates.
Leo Nagelkerke (Leo.Nagelkerke@wur.nl)
Wageningen University, Aquaculture & Fisheries Group