Rare and declining bird species benefit most from designating protected areas for conservation in the UK

Biodiversity loss is accelerating and threatens to exceed planetary boundaries. Expanding protected areas (PAs) is a key policy response (for example, the Aichi 17% target and the post-2020 proposal of 30% of land and sea conserved), yet evidence for PA effectiveness is mixed and causal links between designation and biodiversity outcomes are rarely tested. PAs differ in objectives and management, making effectiveness hard to measure. Birds in the UK are well monitored and a substantial proportion are of high conservation concern, with national (SSSI) and European (SPA/SAC) designations offering primary protection. However, whether positive associations are due to protection per se or underlying habitat correlates remains unclear. This study leverages three large-scale UK citizen science programmes to evaluate, across most of the UK avifauna, whether PA extent is associated with: (1) higher occurrence and abundance (state); (2) higher colonization and persistence (range dynamics); (3) more positive abundance trends; (4) stronger effects in bird-targeted SPAs than in SACs; (5) demographic mechanisms via higher productivity; (6) stronger benefits for rare, declining, or habitat specialist species; and (7) community-level effects indicative of climate change adaptation (greater specialization, colder-community affinities, diversity patterns). Analyses control for land cover, elevation, climate, and human population density and use both regression and statistical matching to address confounding.

Prior work shows PAs often target biodiversity-rich areas but do not always capture the most sensitive species or habitats effectively. Reported PA impacts are mixed: positive associations between PA extent and biodiversity trends (species diversity, population trends) are sometimes observed but not consistently. PAs may facilitate climate adaptation by providing refugia and aiding range shifts, yet species responses vary widely. The distinction between effects of protection versus underlying land cover/habitat selection is often unclear, and causal inference is limited. In Europe and the UK, overlapping designations (SSSI, SPA, SAC) complicate evaluation; SPAs are targeted to bird species, whereas SACs focus on habitats and other biodiversity. Monitoring frameworks (for example, Common Standards Monitoring) assess condition of designated features but not broader biodiversity benefits. This study addresses these gaps by testing associations and mechanisms at national scale while controlling for key confounders and employing matching.

Design: National-scale observational analysis linking PA coverage to multiple avian metrics while controlling for environmental covariates, using both regression and statistical matching. Data sources and response variables: - Occurrence, colonization, persistence (range dynamics): UK breeding bird atlases 1988–1991 and 2007–2011 across 2×2 km tetrads (42,561 and 46,390 tetrads; 29,851 surveyed in both). Species included if likely breeding (probable/confirmed breeders within containing 10 km square), native, and present in ≥20 tetrads; final n=180 species (168–172 usable for occurrence models; 169–171 for colonization; 129–130 for persistence, depending on PA type). - Abundance and abundance trend: BTO/JNCC/RSPB Breeding Bird Survey (BBS), 1994–2019, 1 km squares, two 1 km transects per square, two visits/season. 6,718 squares total; 133 species included (recorded in ≥100 squares/year on average). Abundance per square-year: max of two visits. Trend modeled via year terms. - Productivity and productivity trend: Constant Effort Scheme (CES) mist-netting, 1990–2019, 490 sites (97 in 1990, 114 in 2019). Productivity index: proportion juveniles among captures per site-year; 22 species included after data-quality filters. Protected areas and covariates: - PA layers: SSSI (national), SPA (bird-focused), SAC (habitat-focused). Proportion coverage per survey square (1 km for BBS/CES; 2 km for Atlas). Designation dates not modeled; all treated as designated across study period due to boundary/date complexities. - Environmental controls: Land Cover Map 2015 aggregate classes (% cover of 9 categories, excluding arable to avoid collinearity), elevation (ASTER GDEM v003), geographic location (easting, northing), human population density (Global Human Settlement Layer; linear and quadratic terms), and climate/topography via tensor smooth of elevation×easting×northing. Year as random effect for abundance and productivity; quadratic year for abundance to capture overall trends. Modeling framework: - Species-level models: Generalized additive mixed models (mgcv). Outcomes: • Occurrence/colonization/persistence: binomial with logit link. Persistence defined among squares occupied in the first atlas; colonization among squares unoccupied initially. • Abundance: negative binomial with log link (Poisson overdispersed). Includes linear proportion PA and its interaction with year for trend effects. • Productivity: binomial (events-trials), juveniles as “successes”; includes PA×year interaction for temporal change. Model diagnostics via mgcv gam.check and DHARMa; excluded overdispersed/zero-inflated failures and extreme outliers. - Matching analysis: For each PA type, matched squares with >10% PA coverage to squares with 0% coverage using Mahalanobis distance (MatchIt), without replacement/callipers, partial matching to retain robust SEs. Matched analyses re-fit with covariate adjustment; tests presence/absence of PA rather than continuous extent. Matching reduced imbalance for Atlas and BBS; CES imbalance remained; sample sizes reduced (28% of PA squares matched; 14–27% by designation; 36–60 BBS species removed). - Traits analysis: Phylogenetically weighted regression (MCMCglmm) using an Ericson tree averaged from 1,000 trees (birdtree.org). Response: per-species PA effect estimates (coefficients) from species-level models. Explanatory variables: conservation status (BoCC categories; EU Annex I; UK Schedule 1) with/without log population size; and ecological traits (log body mass, log population size, log population change, Species Specialisation Index, Species Temperature Index, primary habitat association). Inverse-variance weighting of species estimates; priors as specified; 50,000 iterations, 5,000 burn-in, thinning=25; diagnostics via trace/density plots. - Community analysis: For each BBS square-year, computed detectability-corrected community metrics: species richness, Hill’s N2 diversity, evenness (N2/richness), Community Specialisation Index (density-weighted mean SSI), Community Temperature Index (density-weighted mean STI). Fitted GAMs with land cover, climate/location smooths, PA extent, and PA×year interaction. Statistical summaries: Across species, counted significant positive vs negative PA associations (binomial tests) and compared mean effect sizes (t-tests). Compared SPA vs SAC effects via paired t-tests.

- State (occurrence and abundance): 48% of species showed significantly positive associations between occurrence and PA extent versus 21% negative (χ²=17.1, P<0.001); mean occurrence slope 0.49 ± 0.07. For abundance, 48% positive (χ²=8.6, P=0.003); mean effect 0.25 ± 0.05. Species with negative responses tended to be urban-associated. - Range dynamics: Colonization showed similar numbers of significant positives (25%) and negatives (26%) (χ²=0.04, P=0.83), but an overall positive mean effect (0.27 ± 0.08). Persistence increased with PA extent (0.23 ± 0.09), with more species positive (30%) than negative (20%). - Abundance trends: No overall significant effect of PA extent; similar counts of significant positive (20%) and negative (23%) species; mean effect ~0 (non-significant). - Designation type: Effects strongest for SPAs (bird-focused). With increasing SPA extent, more species were likely to occur (41% significantly positive vs 30% significantly negative) and had higher abundances (39% vs 25%). Mean relationships for occurrence, colonization, and abundance trend were significantly stronger in SPAs than in SACs. - Demographic mechanism: Across CES species, site-level productivity tended to be negatively correlated with overall PA extent, but SPA-specific analyses showed that species with higher productivity where SPA extent was greater also had higher abundances with greater SPA extent. Species whose productivity increased more over time in SPAs also tended to have more positive abundance trends with SPA extent (relationship remains, marginal if excluding an outlier). This supports higher productivity as a mechanism for positive SPA effects. - Beneficiaries: After accounting for body mass and phylogeny and weighting by estimate precision, positive relationships of PA extent with occurrence, colonization, persistence, and abundance were strongest for rarer species (smaller population size) and habitat specialists. Species declining nationally had more positive (or less negative) abundance trends in squares with greater PA extent, with stronger relationships for SPA than SAC. Legally protected or BoCC-listed species had higher occurrence and persistence with greater PA coverage, though abundance effects were less marked after adjusting for population size. - Habitat groups: Wetland and woodland species were more likely to occur and persist, and were more abundant, in areas with greater PA coverage; wetland species also had more positive abundance trends. Urban species were less likely to occur and had lower abundances in high-PA areas but showed more positive trends. - Communities: Squares with greater PA coverage had lower species richness and experienced reductions in diversity and evenness over time, but supported more specialist and more cold-dwelling communities (higher CSI, lower CTI and slower CTI increases). This indicates PAs may buffer climate-driven community warming and favor specialists. - Robustness: Matched analyses produced broadly similar conclusions, often with more positive responses, though effect sizes for trait analyses were generally smaller and sample sizes reduced.

Findings indicate that the UK PA network, particularly bird-focused SPAs, has been well targeted for birds: species, on average, occur more frequently and are more abundant where PA extent is greater, independent of broad-scale habitat, topography, climate, and human density. Beyond selection effects, PAs are associated with more favorable range dynamics (higher colonization and persistence), especially for species of greatest conservation concern—rare, declining, and habitat specialists—though average abundance trends are not significantly more positive inside PAs. Evidence suggests a plausible causal mechanism via demography: higher breeding productivity in SPAs links to higher abundance and more positive abundance trends, aligning with the known efficacy of targeted management actions that improve breeding success for species of concern. The absence of a general positive relationship between productivity and abundance trends across all PAs could reflect variable habitat quality (many sites are in unfavourable condition) and density-dependent constraints where PAs attract higher densities. Effects are strongest for SPAs, consistent with designation targeted at bird species and with previous continent-wide and single-species findings. Benefits are pronounced for wetland and woodland species—habitats targeted by conservation and often fragmented in the UK. Despite heterogeneity and unmodeled variation in site management intensity and habitat quality, positive associations persist, suggesting broad-scale benefits of the PA network. Community-level analyses show that PAs shape assemblages toward specialists and colder-affinity species and may slow climate-driven community warming, indicating a role for PAs in climate adaptation. Reduced richness/diversity in high-PA areas likely reflects the nature of protected habitats (for example, wetlands, uplands) that do not necessarily harbor high richness but are critical for specialists. Overall, results support the premise that expanding and appropriately targeting PAs can contribute meaningfully to biodiversity recovery, particularly for the species most in need, while underscoring the importance of management quality within PAs.

This study provides a comprehensive, multi-metric assessment of PA effectiveness for a national avifauna, showing that UK PAs—especially SPAs—are associated with higher occurrence, abundance, and improved range dynamics, with strongest benefits for rare, declining, and habitat specialist species. Independent productivity data support a demographic mechanism underlying positive effects in SPAs. At the community level, PAs favor specialist and colder-affinity assemblages and may buffer climate-driven warming signals. Policy implications include strong support for increasing PA coverage (for example, to 30% by 2030) alongside strategic targeting toward threatened species and habitats and ensuring effective management to translate presence into sustained population growth. Large-scale citizen science monitoring provides a powerful framework for ongoing evaluation of PA networks. Future research should: (1) incorporate site-level management intensity and habitat quality to better explain variation in outcomes; (2) refine causal inference with improved temporal designation/management data; (3) assess species-specific mechanisms (productivity, survival, movement) across taxa; and (4) evaluate connectivity and landscape context effects, including climate adaptation performance.

- Causality and designation timing: Lack of comprehensive designation dates and management start times led to treating all sites as designated throughout, limiting causal attribution and timing analyses. - Unmeasured management/habitat quality: Variation in site management intensity and habitat condition was not modeled; many sites are in unfavourable condition, which could mask stronger potential effects. - Residual confounding: Although models controlled for land cover, climate/topography, and human density, complete separation of protection effects from habitat selection is challenging. Matching reduced but did not eliminate imbalance; CES matching imbalances remained. - Data scope and exclusions: Some species and models were excluded due to small sample sizes, overdispersion, or zero-inflation; seabirds (except gulls/terns) and non-breeding records were excluded, potentially limiting generalizability. - Scale and detectability: Analyses at 1–2 km scales may miss finer-scale habitat/management effects; detectability corrections were applied at the community level but species-level detection heterogeneity may persist. - Abundance trends: No overall positive PA effect on abundance trends suggests potential limitations in management effectiveness or density-dependent constraints; interpretation is complex and context-dependent.