Abstract
A distance-based multivariate control chart is a useful tool for ecological monitoring to detect changes in biological community resulting from natural or anthropogenic disturbances at permanent monitoring sites. It is based on a matrix of any distances or dissimilarities among observations obtained from species composition and abundance data, and bootstrapping techniques are used to set upper confidence bounds that trigger an alarm for further investigations. We extended the use of multivariate control charts to stratified random sampling and analyzed reef fish monitoring data collected annually on shallow (≤30 m) reefs across the Northwestern Hawaiian Islands (NWHI), part of the Papahānaumokuākea Marine National Monument. Fish assemblages in the NWHI were mostly stable, with exceptions in the south region (Nihoa, Mokumanamana and French Frigate Shoals) in 2012 and 2015 where changes in the assemblage structure exceeded the upper confidence bounds of multivariate control charts. However, these were due to changes in relative abundances of native species, and potentially related to the small numbers of survey sites and relatively low coral covers at the sites, particularly in 2015. The present study showed that multivariate control charts can be used to evaluate the status of biological communities in a very large protected area. Future monitoring of fish assemblages in the Papahānaumokuākea Marine National Monument should be accompanied by specific habitat or environmental variables that are related to potential threats to its shallow-water ecosystems. This should allow for more detailed investigations into potential causes and mechanisms of changes in fish assemblages when a multivariate control chart triggers an alarm.