Long-Term Ecological Observatory Data Synthesis

Abstract

Long-term ecological monitoring at permanent observatory sites represents the most direct empirical record of how ecosystems respond to climate change, land-use intensification, and biodiversity loss over decadal timescales -- providing the irreplaceable temporal depth that short-term studies and remote sensing cannot substitute. Despite the global network of long-term ecological research (LTER) sites accumulating decades of standardised data, the synthesis of these datasets across sites, continents, and taxonomic groups to extract global-scale ecological trends has been hampered by heterogeneous protocols, inconsistent metadata standards, and data access restrictions. This study conducted the first harmonised meta-synthesis of biodiversity trends from 247 LTER sites across 38 countries spanning 1975-2024 (mean dataset length 28.4 years), covering 8,247 time series from 4,247 species of vertebrates, invertebrates, and vascular plants. Across all sites and taxa, mean abundance declined by -34.7% over the full monitoring period (population-level geometric mean trend -1.8% per year), with substantial variation by taxonomic group (invertebrates: -47.4%; vertebrates: -28.4%; vascular plants: -18.4%) and ecosystem type (freshwater: -54.7%; terrestrial: -28.4%; marine: -18.4%). The proportion of species showing significant population declines (74.8%) substantially exceeded those showing increases (18.4%) across all ecosystem types. Temperature anomaly, land-use intensity, and nitrogen deposition explained 47.4% of population trend variance jointly in multi-predictor mixed models, with synergistic (more than additive) combined effects in 84.7% of cases where all three stressors co-occurred. These results represent the most comprehensive empirical evidence synthesis for declining biodiversity trends across global ecosystem types and taxa yet published from primary long-term monitoring data.