Crop and forest pest metawebs shift towards increased linkage and suitability overlap under climate change

The study addresses how climate change will reshape opportunities for crop and forest tree cultivation in Europe and alter risks from invasive insect pests. Managed plants and their pests each have climatic niches; warming is expected to shift suitable areas, potentially enhancing production in northern regions but also enabling pest establishment and spread. The authors hypothesize that climate change will: (1) increase areas suitable for a more diverse array of crops and forest trees, (2) increase pest pressure by creating more potential feeding interactions (more links) between pests and host plants, and (3) increase exposure through greater overlap of climatically suitable areas between pests and hosts. Using a metaweb of known pest–host links combined with species distribution models (SDMs) under future climate scenarios, the work quantifies system-wide changes in linkage and exposure across European regions.

Prior research shows climate change and biological invasions already affect agricultural and forest systems, with expected continued impacts on productivity and biodiversity. While elevated CO2 and longer growing seasons can increase plant productivity, climate extremes and pests can offset gains. Global studies project reduced crop production and food security risks, though parts of Europe may gain productivity and diversification potential, especially at higher latitudes. Invasive pests are rising with trade globalization and warming; many pests are constrained by climate barriers that may diminish, and milder winters enhance survival at high latitudes. Documented cases (e.g., Drosophila suzukii) highlight substantial economic impacts and the challenge of managing novel invaders. Previous modeling indicates poleward movement of pests and higher latitudes facing greater pest pressure, with generalist pests posing particular risks. This study builds on these insights by analyzing an integrated pest–host metaweb under climate change across Europe.

Study scope and data: The authors assembled a metaweb of all known interactions between 126 host plants (96 crops and 30 forest tree species) and 89 EPPO-listed insect pests relevant to Europe. Host plants include economically important crops and managed forest tree species; pests are quarantine or alert-listed species, focusing on those present in fewer than five European countries to emphasize potential future invaders. Occurrence records for hosts (EarthStat rasters for crops; EUFORGEN shapefiles and GBIF occurrences for trees) and pests (databases and literature; vetted against EPPO PQR) were compiled, filtered to reduce sampling bias, and screened for minimum data sufficiency (≥24 occurrences). After QC and model performance filtering, 126 host plants and 89 pests were retained (from initial 128 hosts and 94 pests; five pests and two crops excluded for poor evaluation metrics). Species distribution modeling: SDMs were built as unweighted ensembles of four algorithms (GLM, GAM, GBM, Random Forest) using pseudo-absences (5,000 per species) and prevalence weighting (0.5). Models were evaluated with AUC and TSS via 70/30 split repeated 20 times; only models with AUC > 0.7 and TSS > 0.4 were included. For crops, models were run with temperature-only predictors (given irrigation confounds) after confirming close agreement with temperature+precipitation models. Predictors: for crops/trees—mean annual temperature, temperature seasonality, growing degree-days >5°C (and precipitation variables in tests); for pests—minimum temperature of coldest month, growing degree-days >5°C, annual precipitation, precipitation seasonality (plus temperature seasonality for 15 pests as needed). Binary suitability was set by sensitivity–specificity sum maximization thresholds. Climate scenarios: Historical climate (CHELSA v1.2, ~5 km) and future downscaled projections for 2006–2100 were used from four relatively independent CMIP5 GCMs (CESM1-BGC, CMCC-CM, MIROC5, ACCESS1-3) under RCP4.5 and RCP8.5. Time series were aggregated into 10-year steps; analyses summarized medians across GCMs per scenario. Metaweb and overlap: The EPPO host lists defined all potential pest–host links. A link was considered realizable in a European region and time step if binary suitable ranges overlapped in at least 5,000 grid cells. Exposure was defined as the mean overlap area per realized link. Additional network metrics (modularity, specialization, partner diversity, shared partners) were computed with the R package bipartite. Regional analyses covered Southern, Western, Eastern, and Northern Europe, and the British Isles. Pest pressure was summarized by counts of pests with ≥5,000 suitable grid cells. Range shift analyses: For each species and decade, newly suitable cells were linked to the nearest previously suitable cell to estimate average direction and magnitude of climate suitability shifts per grid cell and period (2020–2060; 2060–2100). Centroid shifts (distance and speed) were computed for hosts and pests. Assumptions and constraints: Models focused on climatic suitability, omitting explicit biotic interactions, dispersal constraints, potential host shifts, evolutionary adaptation, and phenotypic plasticity. A cropland mask constrained crop distributions in main analyses. Temperature-only models were used for crops to reduce irrigation confounding.

- Suitable area expansion for hosts: Median suitable area for crops increased from 1,925,265 km² (2020) to 2,790,484 km² by 2100 under RCP8.5 (+47%) and to 2,487,919 km² under RCP4.5 (+27%). Forest trees increased from 4,225,050 km² to 4,366,851 km² (+3%, RCP8.5) and to 4,561,816 km² (+8%, RCP4.5). - Economic opportunities: Increased suitability projected for 82 species (RCP8.5) and 91 species (RCP4.5) of 126 hosts. Examples: soybean suitable area +190% (RCP8.5; +95% RCP4.5); specialty crops: apple +29%/+47%, grapefruit +756%/+225%, lemon–lime +105%/+70%, melon +87%/+50%, tomato +42%/+23%. - Declines in suitability: Under RCP8.5 by 2100—wheat −9%, maize −14%, oats −44%, rye −76%, potatoes −20%; under RCP4.5—wheat +4%, maize +7%, oats −9%, rye −28%, potatoes +1%. Forest trees under RCP8.5: Abies alba −73%, Fagus sylvatica −12%, Picea abies −77%; under RCP4.5: A. alba −36%, F. sylvatica +8%, P. abies −39%. - Regional opportunities: More species encounter suitable climates in Northern Europe (RCP8.5: +33 species by 2100 over 48 in 2020; RCP4.5: +16) and British Isles (RCP8.5: +28; RCP4.5: +10). Slight increases in Western and Eastern Europe; declines in Southern Europe under RCP8.5. - Increased pest–host linkage: By 2100 up to 80% of all potential links become possible (79% under RCP4.5). Southern Europe already realizes 64% (63% RCP4.5) of links today; Northern Europe increases from 7% to 25% (15% RCP4.5). Links increase markedly in Northern Europe (+166 under RCP8.5; +77 under RCP4.5) and British Isles (+190; +78). Links per species decrease slightly in Southern Europe (6.7→6.1) but increase strongly in Northern Europe (0.8→2.6) and British Isles (1.1→3.2). - Exposure growth: Mean overlap area per link rises by 51% (RCP8.5) and 38% (RCP4.5) in Europe between 2020 and 2100, with largest increases in Northern Europe (+173%/+75%) and British Isles (+165%/+57%); Western (+90%/+43%) and Eastern Europe (+60%/+49%) also increase; little change in Southern Europe (0% under RCP8.5; ~19% under RCP4.5). Despite declines in crop suitability for some staples, overlap with pests still increases (e.g., maize +110%/+39%; wheat +135%/+40%; potatoes +80%/+44%). - Pest suitability expansion: Median suitable area for pests increases by 294,176 km² (+12%) under RCP8.5 and 229,981 km² (+9%) under RCP4.5. Southern Europe is currently suitable for many pests (71). By 2100, pest pressure grows in Northern (RCP8.5: 26→+14; RCP4.5: 27→+5), Western (43→+13; 44→+8), Eastern (47→+11; 45→+6), and British Isles (25→+17; 26→+9). Fall armyworm (Spodoptera frugiperda) suitability area +81% (RCP8.5; +51% RCP4.5). - Network structure: Generalist pests (e.g., Spodoptera frugiperda, Helicoverpa zea) drive increased links and reduce network specialization and modularity while increasing partner diversity and shared partners. - Range shift dynamics: Median centroid shift for pests is 519 km (6.5 km/year) under RCP8.5 and 240 km (3.0 km/year) under RCP4.5; for host plants 588 km (7.3 km/year) and 269 km (3.4 km/year), respectively. Suitability shifts generally trend northward for hosts; pest shifts are more idiosyncratic across Europe.

The results show that climate change will likely enable broader diversification of European agriculture and forestry by expanding climatic suitability for many crops and some trees, particularly in northern regions and the British Isles. However, these opportunities are coupled with increased pest pressure manifested as more realizable pest–host links and larger overlap of suitable areas, especially in northern Europe. Generalist pests expand most, eroding network specialization and modularity and increasing cross-category host use. Even where host suitability declines (e.g., staple crops), overlap with pests can still increase, implying rising management challenges. Regional analyses indicate that Southern and Western Europe already face high immediate risks because many invasive pests can survive under current climates and many host plants are present, while Northern regions will increasingly experience invasions as climates warm. The difference between RCPs suggests substantially lower future pest range expansions and exposure under RCP4.5 than RCP8.5, underscoring the benefits of emissions mitigation. From a management perspective, seizing new production opportunities requires weighing potential gains against costs of higher pest co-occurrence and spillover to other hosts, and strengthening phytosanitary measures, surveillance, and integrated pest management across the supply chain.

Using a continental metaweb linking 126 crops/trees with 89 EPPO-listed pests and high-resolution SDMs under RCP4.5 and RCP8.5, the study demonstrates that climate change will reshape European managed ecosystems toward increased pest–host linkage and exposure while opening new cultivation opportunities, particularly in northern regions. Generalist pests drive reduced network specialization and increased cross-host connectivity. Policymakers and practitioners should prepare to exploit diversification potential while mitigating pest risks through enhanced biosecurity, monitoring, and sustainable pest management. Future research should integrate biotic interactions, potential host shifts, evolutionary adaptation, irrigation and soil constraints, and socioeconomic decision-making to refine forecasts and management strategies.

- SDMs capture climatic suitability but omit explicit biotic interactions, dispersal limitations, and management constraints; realized distributions may be over- or under-estimated. - Potential host shifts, evolutionary adaptation, and phenotypic plasticity under climate change were not modeled. - Crop models used temperature-only predictors to reduce irrigation confounding; water availability and irrigation changes under future climates were not explicitly incorporated. - Overlap/link criteria (≥5,000 grid cells) and regional thresholds may influence link counts and exposure estimates. - Pest selection focuses on EPPO-listed species present in fewer than five European countries; other potential invaders are not included, and greenhouse/stepping-stone occurrences were excluded. - Uncertainties in climate projections are partially addressed via four GCMs, but structural and scenario uncertainties remain.