Risks posed by invasive species to the provision of ecosystem services in Europe

Since 1970, the capacity of nature to sustain human well-being through biodiversity, ecosystem functions, and services has declined, with invasive species identified as a major driver. Evidence of impacts on ecosystem services is scattered and often species- or site-specific, limiting generalization to continental scales. A key information gap stems from the lack of harmonized accounting and mapping of multiple ecosystem services at large spatial scales. Because ecosystem services vary across ecosystems and the risks from invasions differ among regulating, provisioning, and cultural services, there is a need to determine which services are most threatened and where to prioritize conservation. This study leverages recent advances in invasive species risk assessments, occurrence data, and ecosystem service mapping to quantify current observed risks and map spatial patterns of future risks from 94 invasive species to seven ecosystem services in Europe. It tests the premise that risks vary across service types delivered by ecosystems with differing invasion levels and aims to inform policy, including the EU Regulation on Invasive Alien Species, by integrating exposure to invaders with vulnerability of service provision.

Previous research has extensively documented effects of invasions on native biodiversity, but evidence on ecosystem services has been fragmented and geographically limited. Global and European assessments indicate higher invasion densities and impacts in human-altered habitats (urban, cropland, grassland), where disturbance and propagule pressure are high and service diversity is often lower. Harmonized, large-scale mapping of ecosystem services has been lacking, constraining spatial risk assessments. Prior work suggests regulating and cultural services may be particularly vulnerable in well-conserved areas with lower exposure, while provisioning services can be affected even in highly managed landscapes. These insights motivated a cross-taxon, continental analysis integrating species distribution modeling with standardized ecosystem service accounts.

Study area: Europe (excluding Russia). Species: 94 non-marine invasive species prioritized at EU scale (Union List species plus 13 under consideration): 32 terrestrial plants, 29 terrestrial animals, 20 freshwater animals, 13 freshwater plants. For each species, potential negative impacts on seven ecosystem services were compiled from standardized EU risk assessments; impacts were coded as yes/no due to context dependence and data limitations.

Ecosystem services (ES): Seven services from EEA MAES/INCA datasets (reference year 2012): regulating—habitat maintenance, nitrogen retention, soil retention, flood control; provisioning—ecosystem contribution to crop production, timber provision; cultural—outdoor daily recreation. ES proxies were aggregated to the EEA 10×10 km grid. ES values were classified using percentiles into low (≤20%), medium (20–80%), and high (≥80%) provisioning.

Species data: Worldwide occurrences from GBIF (cleaned with scrubr; filtered coordinate uncertainty ≤50 km), supplemented with EASIN and other compiled sources. Records were thinned/gridded to 10 arc-minute (~11×11 km) resolution using fuzzySim, yielding 310,556 presence points across species. Pseudo-absences equal in number to presences were sampled outside presence pixels but within 200-km buffers to match sampling bias (terra).

Current exposure: For each ES, the number of invaders known to affect that ES and currently present per 10×10 km cell was counted. Differences in exposure among low/medium/high ES provisioning categories were tested with Welch one-way ANOVA and Tukey HSD post hoc comparisons. Sensitivity analyses used alternative ES thresholds.

Potential exposure: Species distribution models (SDMs) were built per species using Bayesian Additive Regression Trees (BART; dbarts) with 21 predictors: accessibility (human travel time), elevation, and 19 CHELSA v2.1 bioclimatic variables, aggregated to 10 arc minutes. Model performance was evaluated with AUC and Miller’s calibration slope via modEvA; cross-validation indicated good discrimination (mean AUC 0.86±0.06) and well-calibrated, non-overfit models (MCS 1.09±0.18). Predictions were converted to environmental favorability (prevalence-corrected) and classified into categories 1 (≤20%), 2 (20–80%), and 3 (>80%). Potential exposure per cell equaled the number of ES-affecting species with favorability >80%.

Range expansion: For each species, the number of grid cells currently occupied was compared to cells with high favorability (category 3) to estimate potential expansion; paired t-tests assessed differences. Alternative high-favorability thresholds were explored.

Risk assessment framework: A 3×3 matrix combined ES provisioning (Low/Medium/High) and invasion favorability (1/2/3) to define nine risk categories. Three focal categories were interpreted: Safe (H1: high ES, low favorability), Critical (L3: low ES, high favorability), and Hotspots (H3: high ES, high favorability). For each species–ES pair (269 combinations), European areas in each category were quantified. Density maps of Hotspots and Critical areas were produced by stacking across species for each ES. Sensitivity analyses assessed alternative ES thresholds. Data and R scripts are available on Figshare.

• Across 658 possible species–service combinations for 94 invaders and seven ecosystem services, 269 potential negative impacts were confirmed (41%). Most affected services: outdoor recreation (70 species), habitat maintenance (57), soil retention (43), crop provision (41), nitrogen retention (31). Fewest: timber provision (14) and flood control (13).
• Current exposure is generally higher in areas with low provisioning of regulating and cultural services (habitat maintenance, flood control, outdoor recreation), but higher in areas with high provisioning for crop provision and nitrogen retention. Differences among low/medium/high ES categories were highly significant for all services.
• Geographically, current exposure concentrates in Western Europe (UK, Netherlands, France, Belgium, Ireland). Future potential exposure increases notably along coasts and in the Atlantic and Continental biogeographic regions.
• Species distribution modeling indicates substantial room for expansion: on average, the area highly favorable for invasion exceeds the currently occupied area by 77% (paired t-test, P<0.001). Using a slightly less conservative favorability threshold (>70%) yields an average potential expansion of 163% beyond current ranges.
• Risk categorization across species and services shows: Safe areas (high ES, low favorability) average 10.2±5.8% of Europe (0–20% across cases), but for nitrogen retention, Safe areas are only 0.4% of Europe. Critical areas (low ES, high favorability) average 1.4±2.0% (0–9%). Hotspots (high ES, high favorability) average 0.8±1.8%, ranging 0–13%, and are particularly important for crop provision, flood control, and soil retention.
• Spatial hotspots vary by service: habitat maintenance risks in Boreal, Alpine, Atlantic regions; flood control hotspots in the Atlantic bioregion; nitrogen retention risks concentrated in central Europe; soil retention hotspots in Boreal, Alpine, Continental regions; crop provision hotspots in Atlantic and Continental regions; timber provision hotspots in Atlantic, Continental, and parts of the Mediterranean; outdoor recreation hotspots widespread with concentration in the Continental bioregion.

The study demonstrates that the spatial distribution of invasive species exposure is mismatched with the locations where regulating and cultural ecosystem services are most strongly provided, helping to explain why observed impacts on these services can be limited at present despite high vulnerability. High-value ES areas often have low accessibility and less favorable climates for many invaders, reducing current exposure. Nonetheless, species distribution modeling reveals large potential range expansions, implying that exposure—and thus risk—could increase markedly, especially in coastal, Atlantic, and Continental regions. By integrating exposure (favorability for establishment) with vulnerability (level of ES provision) into a nine-category risk framework, the study directly addresses where and which services are most at risk, providing actionable spatial information. The results support prioritizing monitoring and management in a small fraction of Europe where Hotspots and Critical areas concentrate, thereby aligning with EU invasive species regulation and broader biodiversity targets. The analysis underscores that while most high-ES areas are currently relatively safe, their conservation value and vulnerability warrant proactive protection against future invasions, and that intermediate-risk areas, covering much larger extents, should not be neglected in management planning.

This work delivers a cross-taxon, continental-scale risk assessment linking invasive species to seven key ecosystem services using harmonized ES accounts and robust SDMs converted to environmental favorability. It quantifies current exposure patterns, projects substantial potential range expansion (average 77%), and delineates small but disproportionately important Hotspots (0–13% of Europe) and Critical areas for targeted action. The outputs—risk categories, maps, and species-by-service risk summaries—can guide policy and management under the EU Regulation on Invasive Alien Species and contribute to the EU 2030 Biodiversity Strategy and the Kunming-Montreal Framework. Future research should incorporate climate change explicitly, quantify abundance–impact relationships, assess positive and negative impacts jointly, and evaluate cumulative and synergistic effects of multiple invaders and cascading effects across services and scales.

• Benefits from invasive species were not assessed due to lack of systematic reviews, so only negative impacts were considered.
• Impact magnitude is context dependent; presence does not guarantee impact. Factors such as invader density, microhabitat, and management can alter outcomes.
• Uncertainties arise from species–service associations, SDM assumptions and predictors, ES proxy maps (reference year 2012), spatial resolution (10×10 km), and thresholds for ES provisioning and favorability.
• Abundance data were not used; exposure was based on species counts per cell, which may over- or underestimate realized impacts.
• GBIF data may include casual (non-established) occurrences, potentially overestimating current spread; SDMs may overestimate suitable areas, especially for habitat- or host-restricted species.
• Interactions among co-occurring invaders (additive or synergistic) were not modeled.
• Climate change effects were not explicitly included, though they may alter both invasion dynamics and ES provision.