Abstract: Terrestrial insects are declining at estimated rates of 1-2% per year across long-term monitoring datasets, however, these estimated rates likely differ substantially from true rates of change due to site selection bias. Many long-term ecological datasets come from relatively undisturbed habitat, such as from nature preserves, which were selected to reduce confounding factors and because they are expected to be available to sample for decades. Insect populations at these sites are often not representative of those in the surrounding landscape, however, which are subject to land use change and degradation. To assess how site selection bias affects estimated rates of insect biodiversity change, we used combined results derived from a systematic review of multi-year studies on Hymenoptera (bees, wasps, and ants).
We identified over 200 multi-year datasets that can be used for the analysis, ranging from two consecutive years to over 30 (median time series length = 3). We compared rates of biodiversity change from long-term monitoring studies to those from control plots in experimental studies designed for other purposes (e.g. untreated controls from tests of pesticide effectiveness), which we expect to be less subject to such biases. Preliminary results from the meta-analysis suggest that rates of change estimated from data collected from control plots are less negative than rates estimated from observational studies. One possible explanation for the weaker trends in control plots is that they are situated in areas that start out with low abundance and diversity of insects and the data reflect patterns for species that can persist in degraded land. The results of this systematic review and meta-analysis will not only be useful for designing new insect biodiversity sampling schemes for monitoring efforts that capture a more representative sample of the landscape, but will also help to increase our understanding of insect biodiversity change in the Anthropocene.
