Anthropogenic climate and land-use change drive short- and long-term biodiversity shifts across taxa

The study addresses how anthropogenic climate change and land-use change jointly shape biodiversity across taxa and time. Although both drivers are known to influence biodiversity, the lack of historical land-use data has limited understanding of their interactions and the role of baseline environmental conditions. The authors aim to quantify changes in species richness, beta diversity (biotic homogenization), and community temperature index (CTI) for birds, butterflies, and plants in Great Britain at two timescales (long term: 1960s–2010s; short term: 1990s–2010s). They test: (1) whether richness, beta diversity, and CTI have increased or decreased; (2) how changes relate to climate and land-use change, their interactions, and baseline conditions including microclimatic heterogeneity; and (3) which environmental characteristics are associated with grid-cell contributions to national beta diversity (LCBD), to inform conservation prioritization. The work is motivated by evidence that habitat availability mediates climate-driven range shifts, that land-use change alters local microclimates, and that baseline conditions can set trajectories for community change; yet long-term, nation-scale analyses integrating historical land use, climate, and multi-taxon biodiversity have been scarce.

The paper synthesizes prior research showing: intensified agriculture and forestry across Europe since the early 20th century have generally reduced biodiversity and increased homogenization; the global climate has warmed by about 1°C with poleward range shifts across taxa; habitat availability is critical for climate-driven expansions and fragmented landscapes can impede colonization and persistence; land use alters microclimate, influencing species’ experienced warming; and baseline environmental conditions can govern both community change and current biodiversity distributions. Previous studies often relied on space-for-time substitutions or modern land cover due to a lack of historical land-use data, leaving climate–land-use interactions and baseline effects largely unexplored over decadal scales. The study builds on work indicating time-lagged biodiversity responses (extinction debts and colonization credits), microclimatic buffering effects, and the conservation value of semi-natural grasslands.

Scope and scale: Analyses were conducted at the 10 km British National Grid cell resolution across Great Britain for birds, butterflies, and plants, focusing on local-assemblage responses rather than species-specific trends. Two timescales were examined: long term (1960s to 2010s) and short term (1990s to 2010s), using only grid cells with data in both periods per comparison. Biodiversity data: 3,715,724 occurrence records across n = 2,670 grid cells. Species sets included 250 breeding birds, 55 butterflies, and 1,587 vascular plants, filtered to stable taxonomic concepts. Time periods used: birds (1968–72; 1988–91; 2008–11), plants (1930–60; 1987–99; 2010–19 subset), butterflies (1970–74; 1990–94; 2010–14). Effort correction: Presence-only data with uneven recorder effort were corrected using Frescalo to estimate effort per taxon and period. For birds, a Frescalo-derived benchmark species metric was available. Effort terms were incorporated into models (log scale for richness, natural scale for beta diversity and LCBD). For CTI, a species-richness threshold per grid cell was applied (butterflies ≥10 spp; plants ≥214 spp) to include well-recorded cells; birds required no threshold. Environmental data: Land-use categories (arable, semi-natural grasslands, improved grasslands, forest, urban; plus water) were compiled from historical maps (1930s–40s) and LCM 1990/2015 at 25 m, aggregated to 10 km proportions. Climate variables were average annual mean temperature and annual precipitation from Met Office HadUK-Grid, aggregated to 10 km for 1965–75, 1985–95, and 2005–15. Microclimatic heterogeneity was proxied by the grid-cell standard deviation of solar index (2000–2010), reflecting topography-driven thermal heterogeneity. Biodiversity metrics: Species richness (SR); beta diversity (BD) as community dissimilarity of each focal cell to its eight neighboring cells using Sørensen index via betapart; CTI as the mean species temperature index (STI) of species present in a cell, using STI datasets for each taxon. Modeling framework: Bayesian inference via INLA (R-INLA). Spatial structure was included using ICAR models (and Leroux for some count models) to account for spatial autocorrelation. (1) Temporal differences: For each taxon and timescale, GLMMs estimated changes between periods: SR (Poisson for birds; negative binomial for butterflies and plants; log link), BD (Beta GLMM; logit link), CTI (Gamma GLMM; log link). Predictors included period (categorical) and recorder effort (except CTI). Spatial random effects: ICAR for BD and CTI; SR used Leroux structure to manage overdispersion. (2) Drivers of change: For each taxon, metric, and timescale, spatial linear mixed models (Gaussian) regressed observed change in biodiversity on: changes in climate (temperature, precipitation), changes in land-use proportions, all two-way climate×land-use interactions, baseline land use and climate, baseline biodiversity (SR and BD where relevant), microclimatic heterogeneity, and recorder effort terms (initial and change where applicable). Predictors were standardized; ICAR spatial random effects were included (except in BD change models to avoid overfitting). Model performance assessed via marginal and conditional R²; sequential regression gauged the added contribution of variable groups. (3) LCBD modeling: For each taxon and period (1960s, 1990s, 2010s), LCBD (from adespatial) was modeled with Beta GLMMs (logit link) as a function of land use, climate, microclimatic heterogeneity, and recorder effort, including ICAR spatial random effects. Robustness: Tenfold cross-validation compared fixed-effect estimates from full and CV models. Sensitivity tests probed the short-term butterfly CTI decrease by substituting 1995–99 for 1990–94 (result remained a slight decrease with credible interval overlapping zero). Priors were weakly informative Gaussian for fixed effects and penalized complexity priors for spatial components.

- Across grid cells, species richness increased for birds, butterflies, and plants over both long (1960s–2010s) and short (1990s–2010s) terms. Example: birds showed an average ~2% increase in richness from the 1960s to 2010s, with an estimated 58–60% probability that a given grid cell increased in richness. - Biotic homogenization (declining beta diversity) generally increased over time, except birds exhibited increased beta diversity in the long term. - CTI increased over the long term for all taxa, indicating shifts toward warmer-adapted communities. Notably, butterflies showed a counterintuitive short-term CTI decrease despite warming climates. - Anthropogenic climate and land-use change individually associated with increased richness and homogenization, particularly over the long term. Increases in temperature, precipitation, and anthropogenic land uses (arable, improved grasslands, forest, urban) were linked to these trends, largely at the expense of semi-natural grasslands. - Climate–land-use interactions: Long-term interactions influenced richness and homogenization (directions varied by taxon). For CTI, short-term interactions were synergistic (anthropogenic land-use change amplified climate-driven CTI increases in birds and butterflies), while long-term for butterflies some interactions were antagonistic (arable/improved grassland cover lessened warming effects on CTI). - Baseline conditions were especially influential over the short term: higher initial cover of semi-natural grasslands was associated with smaller increases in richness and lower homogenization (i.e., more stability). Cooler and wetter baseline climates also moderated change in some taxa (e.g., wetter/cooler baselines linked to lower increases in richness and homogenization in birds and butterflies; cooler baselines linked to smaller CTI shifts for butterflies and plants). - Baseline biodiversity strongly constrained change: higher baseline richness predicted smaller richness increases; higher baseline beta diversity predicted larger declines in beta diversity. Including baseline biodiversity increased marginal R² by roughly 0.4, indicating substantial explanatory power. - Microclimatic heterogeneity was associated with larger increases in plant and butterfly richness (both timescales) and with reduced CTI increases for birds; however, higher microclimatic heterogeneity was linked to lower LCBD in the 1990s and 2010s across taxa. - LCBD: Higher proportions of semi-natural grassland consistently conferred higher local contributions to national beta diversity for all taxa and periods, with effect sizes increasing over time, implying that semi-natural grasslands became increasingly important. Greater forest and improved grassland cover reduced LCBD. Warmer areas had higher LCBD for birds and plants across all periods. - Data scope: 3,715,724 records; 2,670 grid cells; 250 bird species, 55 butterfly species, 1,587 plant species analyzed.

Findings demonstrate that climate warming and land conversion have reorganized communities, increasing local richness while generally decreasing beta diversity, yielding national-scale biotic homogenization and warmer-adapted communities. Interactions between land use and climate significantly modulated richness and homogenization, especially over longer timescales, underscoring that climate impacts are contingent on land-use context. The prominence of baseline environmental conditions in explaining short-term changes supports an inertia or lag hypothesis: recent community shifts partly reflect delayed responses to past environmental change, including agricultural intensification and habitat conversion. Semi-natural grasslands emerged as stabilizing habitats (lower rates of change) and increasingly vital contributors to national beta diversity, likely reflecting their distinctive biotas and the disproportionate loss of these habitats. Apparent increases in richness amid global biodiversity concerns may reflect prior landscape transformations (pre-1930s) and recent expansions of widespread and non-native species, alongside climatic range shifts; delayed extirpations may still unfold. Taxa differed in magnitude and direction of changes: butterflies showed stronger richness increases and homogenization (consistent with rapid generational turnover and sensitivity to habitat change), birds uniquely showed long-term beta diversity increases (potentially due to high dispersal capacity and ability to track isotherms), and CTI patterns indicated complex, sometimes lagged thermal responses, including short-term decreases in butterflies possibly linked to habitat degradation in warmer regions. Microclimatic heterogeneity moderated community reorganization in differing ways: providing refugia (dampening CTI increases in birds) but also potentially expanding niche opportunities that promote colonization (richer plant and butterfly assemblages). Overall, while fixed effects captured important patterns, spatial structure accounted for a large portion of variation, indicating unmeasured spatially structured processes.

Climate and land-use change jointly drive substantial biodiversity reorganization across British birds, butterflies, and plants over both multi-decadal and decadal scales. Protecting and restoring natural and semi-natural habitats—especially semi-natural grasslands—is critical, as these areas both stabilize local biodiversity dynamics and increasingly underpin national-scale beta diversity. Management and policy should go beyond simple richness metrics, considering species-specific requirements, community composition, and functional attributes. Given that climate and land conversion explained a limited fraction of observed change, future work should integrate additional drivers (e.g., land-use intensification, topography, species functional traits, and potential epigenetic factors) and consider finer spatial resolutions and species-level analyses to better guide conservation actions.

- Spatial resolution: Analyses at 10 km grid cells may miss finer-scale habitat and species responses to land-use change and microclimate. - Historical data biases: Presence-only records with uneven observer effort and potential inconsistencies in historical recording could influence inferred trends, despite Frescalo corrections and thresholding. - Temporal mismatches: Historical land-use data precede some biological datasets; while major land-use shifts occurred post-1960s, temporal alignment is imperfect. - Model explanatory power: Fixed effects (climate, land use, interactions, baselines) explained a relatively low proportion of variance, with spatial random effects capturing much variation, suggesting unmeasured drivers. - CTI sensitivity: The counterintuitive short-term CTI decrease for butterflies could be affected by recording biases; sensitivity analyses indicate the decrease persisted but with uncertainty overlapping zero when using 1995–1999 data. - Generalizability: Results are specific to Great Britain’s taxa and history of land-use change; responses may differ in other regions and taxonomic groups.