Ecological Indicators (Oct 2021)
The reliability of low taxonomic and numerical resolutions for biodiversity monitoring is site specific and dependent on the statistical method
Abstract
Ecologists are challenged to detect and offer measures to mitigate the ever-growing impacts on ecosystems (e.g., eutrophication and global warming). An important part of the challenge is to get the necessary biodiversity data over large spatial extents and continuously over time. This challenge is even more evident when one considers the scarcity of research funds and specialized personnel to conduct biomonitoring programs. Thus, by necessity, most of them are based on some sort of shortcut, including the use of higher taxa, presence-absence data and a limited number of taxonomic groups. However, there is a scarcity of studies evaluating the reliability of these shortcuts in temporal analyses of community structure. Here, using zooplankton communities monitored over a period of 85 months in a tropical reservoir, we tested whether data with low taxonomic and numerical resolutions were able to predict beta diversity and ordination patterns generated with species abundance data. The results of two methods, commonly used to measure the relationships between multivariate data (Mantel and Procrustes tests), indicated a high correlation between datasets with low and high taxonomic resolutions. However, the Mantel test results indicated that resemblance matrices derived from presence-absence data were, in general, poorly correlated with those matrices derived from abundance data. Finally, based on the simple correlation between ordination axes derived from data with different taxonomic and numerical resolutions, we found that none of the shortcuts provided reliable results for the different sites analyzed. We suggest that further studies should raise the bar for the proposal of shortcuts and that high-resolution data are key to achieve biomonitoring goals.
