Luxury and legacy effects on urban biodiversity, vegetation cover and ecosystem services

Urban green spaces (UGS) and associated biodiversity underpin multiple ecosystem services (ES) crucial for urban residents’ health and quality of life. Beyond biophysical determinants, socio-economic factors and urban development shape heterogeneous urban landscapes and biodiversity patterns. Two prominent mechanisms are: (1) the luxury effect, whereby wealthier residents—proxied here by higher educational attainment—can influence greening and biodiversity distribution through greater resources and political agency; and (2) the legacy effect, whereby past land-use planning and management leave enduring imprints on contemporary vegetation and biodiversity, often resulting in higher plant diversity in older neighbourhoods due to longer establishment and successional times. These dynamics can drive environmental injustices by differentially distributing access to nature’s benefits. This study asks how luxury and legacy effects influence spatial patterns of biodiversity, vegetation cover, and regulating ES, and how interactions with habitat quality and population density modulate these effects. The mid-sized European city of Vitoria-Gasteiz—recognized for sustained greening policies including a green belt and a green infrastructure plan—offers a suitable case to test: i) relationships between high educational attainment (luxury) and neighbourhood development age (legacy) with biodiversity and vegetation cover; and ii) how these effects influence regulating ES provided by public trees. The authors expected positive luxury and legacy effects on biodiversity and vegetation cover and, hence, on ES, modulated by population density, vegetation cover, and habitat quality.

Prior research has linked socio-economics to urban biodiversity and vegetation patterns across taxa (plants, birds, and some mammals, lizards, arthropods), often documenting a luxury effect where wealthier areas harbor more biodiversity and green cover. Traditional proxies like income can oversimplify social complexity; educational attainment may better capture socio-economic variation relevant to biodiversity outcomes. Legacy effects from past land-use and UGS management shape contemporary urban forests via successional trajectories, species establishment, and long-lived trees. Population density interacts with socio-economics and urban form, sometimes negatively associated with vegetation and biodiversity in compact historical cores, but relationships can be mixed depending on context and management. Despite extensive literature on luxury and legacy effects for biodiversity and vegetation, evidence linking these effects directly to regulating ES remains limited, prompting this study’s focus.

Study site: Vitoria-Gasteiz (Basque Country; 248,087 inhabitants) with long-standing greening policies including a 731-ha, 35 km green belt and extensive public tree inventory. Data and indicators: Biodiversity proxied by species richness from two datasets: (1) public land tree inventory (city council), and (2) bird census (10,016 observations across 100 sampling points, 2017–2020; SEO/BirdLife). Habitat quality (0–1) from prior work captured the capacity of urban ecosystems to support wildlife. Vegetation cover (herbaceous and canopy) was mapped via Google Earth Engine using Sentinel-2 (Level-2A) imagery (2017/07/11 and 2017/07/18; bands B2–B8A, B11–B12), with cloud masking (QA60), indices (UI, NDVI), DEM-derived terrain variables, and a Random Forest classifier (n=100 trees) to map urban forests, urban fabric, herbaceous, and water bodies and compute neighbourhood-level vegetation cover. Socio-demographic variables (2015; Basque Statistics Office) included percentage of residents with high educational attainment (proxy for socio-economic status), population density (inhab/ha), and a neighbourhood development age index (legacy indicator). The development age index was computed by summing the age of built transformations across time periods (1800–2015) weighted by the percentage of area built in each period, rescaled to 0–1 (higher = older). Ecosystem services modelling: Regulating ES from public urban trees were quantified using i-Tree Eco v6 for 2015 (inventory of 89,001 trees after data cleaning). ES indicators included transpiration (temperature regulation), avoided runoff (runoff control), air pollution removal (O3, NO2, CO, SO2, PM2.5), and carbon sequestration/storage. Hourly pollution and meteorological inputs were set to 2015 averages from four monitoring stations. ES were aggregated to a single ES index per neighbourhood by min-max scaling each ES to 0–100 and summing (then rescaling 0–100), following Baró et al. 2019. Analyses were stratified at neighbourhood scale (n=28). Statistical analysis: Pairwise Spearman correlations and Global Moran’s I were computed. Generalized linear models assessed biodiversity (species richness), vegetation cover (%), and ES index against explanatory variables: high educational attainment (luxury), neighbourhood age (legacy), habitat quality, vegetation cover (for biodiversity and ES models as applicable), and population density. Where warranted by significant correlations, interactions were tested: high education × habitat quality; neighbourhood age × population density; neighbourhood age × vegetation cover (for ES). Model selection was based on p<0.05, F-test, adjusted R², parsimony, and diagnostics (normality, influential observations). Analyses used R 4.2.2 (tidyverse, performance, sjPlot), ArcGIS 10.7.1 for spatial autocorrelation, and GEE for land cover classification.

• Biodiversity (tree and bird species richness):
  • Positive association with high educational attainment (luxury effect): Adj. R²=0.25, F=10.01, p<0.01.
  • Negative association with neighbourhood development age (inverse legacy effect): Adj. R²=0.22, F=8.61, p<0.01; neighbourhoods developed mainly from late 1970s to late 1990s had higher richness.
  • Positive associations with habitat quality (log response: Adj. R²=0.45, F=22.85, p<0.001) and vegetation cover (Adj. R²=0.16, F=6.35, p<0.05); negative with population density (Adj. R²=0.34, F=14.92, p<0.001).
  • Interactions: High education × habitat quality amplified luxury effect (Adj. R²=0.53, F=31.09, p<0.001); Neighbourhood age × population density strengthened inverse legacy effect (Adj. R²=0.38, F=17.92, p<0.001).
• Vegetation cover (%):
  • No significant relation with high educational attainment (no luxury effect on vegetation cover).
  • Negative relation with neighbourhood age (legacy effect inverse for cover): Adj. R²=0.15, F=5.97, p<0.05.
  • Positive relation with habitat quality (log response: Adj. R²=0.29, F=12.23, p<0.01); negative with population density (Adj. R²=0.39, F=18.36, p<0.001).
  • Interaction: Neighbourhood age × population density: older, denser areas have lower vegetation cover (Adj. R²=0.48, F=26, p<0.001).
• Ecosystem services (ES) from public trees (2015 totals): 185,145 m³/yr transpired water; 30,652 m³/yr avoided runoff; 14.1 ton/yr air pollutants removed; 617.0 ton/yr carbon sequestration.
  • ES index positively correlated with vegetation cover (Adj. R²=0.19, F=7.39, p<0.05).
  • No direct significant correlation with luxury (education) or legacy (age) predictors alone, nor with habitat quality.
  • Interaction indicating legacy effect mediated by vegetation cover: Neighbourhood age × vegetation cover positive (Adj. R²=0.57, F=37.2, p<0.001); older neighbourhoods with more vegetation provide more regulating ES.
• Spatial: Highest ES index in older neighbourhoods surrounding the historic core; lowest in the medieval core (Casco Viejo) and newer, peripheral neighbourhoods with more recent tree infrastructure. Moran’s I for ES index showed no significant spatial autocorrelation (z=0.34, p=0.72).

Findings support the luxury effect for biodiversity: neighbourhoods with higher educational attainment harbor greater tree and bird species richness, especially where habitat quality is high. This aligns with socio-political mechanisms whereby wealthier, better-educated residents influence UGS investments and management, shaping biodiversity patterns. Conversely, an inverse legacy effect for biodiversity was detected: older neighbourhoods showed lower species richness, likely reflecting historical planning and greening priorities in Vitoria-Gasteiz that emphasized preservation of older urban form with more recent biodiversity-promoting management. Population density emerged as a key moderator, directly reducing biodiversity and intensifying the negative association of age with both biodiversity and vegetation cover, consistent with constraints of compact historical fabrics on green space and canopy growth. Vegetation cover did not exhibit a luxury effect, indicating that greenness alone does not track socio-economic status when aggregating public and private vegetation; biodiversity is a more sensitive indicator of luxury effect in this context. The negative age effect on vegetation cover, exacerbated by higher density, suggests prioritizing greening interventions (e.g., green roofs/walls, street trees) in older, denser areas to address climate and environmental injustices. For regulating ES, vegetation structure (coverage and mature canopy) rather than biodiversity per se drove ES provision, and a legacy effect manifested via the interaction between neighbourhood age and vegetation cover: older, greener areas delivered more ES, likely due to larger, older trees and established canopy. This highlights that historical management and planting trajectories influence ES distribution and that newer areas hold potential for future ES gains as canopy matures. Overall, the study demonstrates that socio-economic factors primarily drive urban biodiversity patterns, while management legacies and population density shape vegetation cover and ES, with implications for environmental justice and green infrastructure planning.

The study evidences a luxury effect on urban biodiversity associated with higher educational attainment and identifies a legacy effect mediated by population density for vegetation cover and by vegetation cover for regulating ES. Biodiversity responds strongly to socio-economic gradients and habitat quality, whereas ES provision is mainly governed by vegetation structure and historical greening legacies. Urban planning should: (1) enhance habitat quality and biodiversity particularly in older, denser neighbourhoods; (2) deploy green-grey solutions where space is limited; (3) prioritize equitable distribution of UGS and ES to avoid reinforcing environmental injustices; and (4) plan for long-term canopy development in newer areas to realize future ES benefits. Future research should disentangle species-specific ES contributions to guide tree planting schemes aligned with biodiversity goals and assess how population density, land use, and development age jointly shape biodiversity and ES over time.

• Correlative design limits causal inference; associations do not imply causation.
• ES modelling constrained to 2015 meteorological and pollution inputs due to i-Tree Eco data update limitations; results reflect that year’s conditions.
• Socio-economic data aligned to 2015; median income unavailable, necessitating the use of educational attainment as a proxy for socio-economic status.
• Analyses conducted at neighbourhood scale (n=28), potentially masking finer-scale heterogeneity.
• Biodiversity proxies limited to public tree and bird species richness; other taxa and private vegetation structure were not directly measured in biodiversity models.