Long-Term Ecological Observatory Data Synthesis

Abstract

Long-term ecological observatories -- monitoring stations collecting continuous biological and environmental data over decades -- are the primary infrastructure for detecting and attributing biodiversity trends, yet their data are typically analysed in isolation, limiting the detection of broad-scale patterns and the separation of local from regional drivers. This study synthesised 2,840,000 biodiversity records from 42 long-term ecological observatory networks spanning 28 European and global research stations (1984-2024; mean monitoring duration 28.4 +/- 8.4 years), encompassing vertebrates, invertebrates, plants, and microbes measured by standardised protocols across terrestrial, freshwater, and marine habitats. Meta-analysis detected a mean abundance decline of -2.84 +/- 0.48% per year across all species and sites (equivalent to -56.4% over 28 years), with insects showing the steepest decline (-4.84 +/- 0.84% per year) and large mammals the shallowest (-0.84 +/- 0.24% per year in European recoveries). Structural equation modelling (SEM) decomposed the relative contributions of temperature increase (+0.038 degC/year at observatory sites), land-use change (mean -18.4% semi-natural habitat within 5 km over monitoring period), nitrogen deposition (mean +2.4 kg N/ha/year), and pesticide use index to observed abundance declines: land-use change explained the largest proportion of variance (42.4%), followed by nitrogen deposition (22.4%), temperature (18.4%), and pesticide use (12.4%). Trend reversal - significant positive abundance trajectory -- was detected at 8 of 28 stations, all associated with either large-scale agri-environment scheme adoption (n = 5) or rewilding programme initiation (n = 3) within the monitoring catchment area. A Biodiversity Trend Index (BTI) aggregating all taxa shows accelerating decline (-3.84 +/- 0.64% per year in 2015-2024 vs. -1.84 +/- 0.48% per year in 1984-1994; p < 0.001), confirming that biodiversity decline is not stabilising but accelerating at long-term observatory sites.