Robust estimates of biodiversity change require high-resolution time series

Accurate estimates of biodiversity change are more important than ever before. However, biodiversity data often suffer from limitations that could produce unreliable trends. One common limitation is low temporal resolution, with many broad-scale biodiversity studies quantifying trends using just two or three sampling years per site. Such low-resolution data could potentially inflate trend errors, producing misleading conclusions about biodiversity change. At the same time, low resolution data are often the only data available, and conclusions drawn from these data can still be broadly accurate, suggesting certain limitations may be acceptable without introducing substantial bias. In this presentation, we discuss how different temporal data limitations affect biodiversity trend estimates. We do so using European-scale data for 1,353 river invertebrate communities sampled almost annually for 10 to 29 years across 18 countries. Using each time series, we simulated lower sampling frequencies from annual to every 2–6 years, and shorter durations from 10 total years to 9–2 years, then compared how these changes affected trend directions (i.e., positive or negative) and the estimated magnitude of change. We further supplemented these analyses with a comparison of our data with those of a lower resolution dataset from the European Environment Agency (EEA) for the same rivers. Reducing sampling frequency had a minor effect on trend direction errors, with 87–74% sites still matching in direction. Conversely, magnitude errors increased more substantially, from matches of 89% when sampling occurred every 2 years down to only 12% when sampling every 6 years. We found similar changes in error rates with reduced duration, for shorter (≥10 years) versus longer (≥20 years) time series, and for our comparison of the two monitoring datasets. Our findings show that temporal data resolution can greatly impact the estimated magnitude of biodiversity change, whereas its direction is less sensitive. Consequently, obtaining accurate estimates of both magnitude and direction requires high-resolution time series, whereas lower-resolution data may only reliably capture direction. These results highlight the value and limitations of temporal biodiversity data, with implications for future monitoring and broad-scale investigations of biodiversity change.