Sampling effort required to recover the main gradients in marine benthic species composition

Marine benthic surveys and monitoring programs may have several goals. When the aim of such studies is to reveal and describe gradients in species composition, the gain in ecological insight achieved by the study depends on how well the spatial variability of species' occurrence and abundance within the study area is covered by the sampling design. The common practice in such programs is to collect four or five grab samples from each visited site to capture the site's species composition. The rationale for this practice is based on the aim of capturing a large proportion of species at a specific site (i.e. point). We argue that this practice has been established without taking Iterative approaches into account, such as decreasing the sampling effort at each site and instead increase the number of sites visited. The collection and processing of data from marine sediment ecosystems are time-consuming and, thus, expensive. Therefore, finding the optimal trade-off between the number of sites visited and the number of samples per site is important. We examined this trade-off by exploring how reduction of sampling effort per site affects the stability and recovery of gradients in species composition of marine benthic macrofauna. Our full data set consisted of four grab samples from each of 28 sites from the inner part of the Oslofjord, SE Norway. Procrustes analysis and calculation of Kendall's rank correlation coefficients showed that the main gradient in species composition extracted by detrended correspondence analysis of data sets with reduced per-site sampling effort closely resembled the main gradient extracted for the full data set. Our results therefore suggest that marine benthic surveys that aim to identify the main gradient structure of species composition and relate this structure to main environmental complex gradients are likely to benefit from reducing the sampling effort at each site and, instead, increasing the number of visited sites. We argue that coverage of both the variation in species composition and the environment in the study area will then be improved. Monitoring programs with other aims than to identify the main gradients in species composition may also benefit from the adoption of a many-sites, one-sample sampling strategy. This may be particularly true if additional sampled sites are based on a stratified sampling strategy that ensures that the spatial variability in the study area is accounted for.