Landscape features support natural pest control and farm income when pesticide application is reduced

Global biodiversity frameworks and the EU Green Deal target substantial reductions in pesticide use (around 50% by 2030) due to mounting evidence of negative environmental impacts from intensive agriculture and biodiversity loss. Natural enemies provide natural pest control (NPC), potentially reducing reliance on synthetic pesticides. However, general quantification of NPC’s contribution to crop yields and its economic value is limited, often context-specific, and constrained by data. This study aims to quantify how landscape features that enhance NPC (LF-NPC) affect yield gaps between organic (proxy for low pesticide use) and conventional systems across the EU, and to evaluate the associated economic value under a reduced pesticide use scenario.

Prior work shows landscape complexity and semi-natural habitats can enhance abundance of natural enemies and NPC, reducing dependence on chemical inputs. Yet, generalizable estimates linking NPC to yields are scarce due to heterogeneous contexts, data limitations, and ecological complexity. Economic valuations exist but are typically crop- or location-specific, e.g., estimates for soybean aphid or pear production. A broad-scale economic assessment across multiple crops and regions under pesticide reduction targets has been lacking. This study addresses these gaps by integrating EU-wide farm-level yield data with a continent-scale LF-NPC indicator and a regional agro-economic model.

• Data sources: (1) A 100 m resolution EU-wide LF-NPC potential map derived from high-resolution geospatial layers of landscape elements (small woody features, grasslands, forests) combined with field surveys of flying insect predators/parasitoids, further refined using a spatially explicit agricultural intensity indicator and a 10 m crop mask to compute crop-specific regional LF-NPC scores. (2) EU Farm Accountancy Data Network (FADN) farm-level data (2010–2017) on yields, management (organic/conventional), structural and economic variables.
• Yield gap estimation: For each selected crop and FADN region, farm-level log-yields were regressed on covariates controlling for structural/geographical characteristics, management practices, year shocks, and inputs. The organic status coefficient isolates yield differences attributable to pesticide use, controlling for purchased fertilizers/soil improvers, livestock units per ha (proxy for manure), irrigation, specialization, unpaid labor share, subsidies ratio, rented land share, assets/liabilities ratios, assets per ha, capital-labor ratio, farm size (UAA), crop revenue share, seed costs (proxy for pest-resistant varieties), and crop diversity (Shannon index as proxy for rotation). Year dummies and NUTS3/altitude dummies control for local conditions and shocks. Outliers: top/bottom 1% of yields per crop and Member State removed. Regions with fewer than 16 organic farm observations per crop excluded. Yield gap (%Δy) computed from the organic coefficient β2 as %Δy = (e^{β2}−1)*100. Crop-typology specific gaps were averaged to a regional crop estimate when possible.
• Linking LF-NPC to yield gaps: Regional median LF-NPC scores (standardized by biogeographical region; filtered to low/medium intensity areas; crop-specific via crop mask) were paired with estimated regional crop yield gaps. A linear mixed-effects model quantified the effect of LF-NPC on yield gaps, grouping by crop with random intercepts (random slopes dropped after testing). Fixed effects capture the average change in yield gap per one-unit LF-NPC increase; marginal/conditional R2 were computed.
• Economic valuation: The CAPRI partial equilibrium model (EU NUTS2 supply modules linked to a global market module) was parametrized for 2030. Exogenous shocks: (a) yield reductions equal to the estimated regional crop-specific yield gaps as a function of LF-NPC, and (b) an 80% reduction in pesticide costs (reflecting zero synthetic pesticide use but remaining organic crop protection costs). A scenario allowing market price feedbacks was also run (reported in Supplementary Information); main focus is on relative benefits due to LF-NPC.

• Yield gaps: Across 534 region-crop estimations (observations per model ranged from 36 to 20,222 farms), the average EU yield gap attributable to pesticide use differences between organic and conventional farms is −15%, varying by crop (e.g., legumes −8%, barley −17%, rye −23%). Regional heterogeneity is substantial (e.g., fodder corn from −6.5% in Norte e Centro, Portugal to −18.7% in Baden‑Württemberg, Germany).
• Correlation with LF-NPC: Positive Pearson correlations between LF-NPC and (less negative) yield gaps for 9/10 crops. Reported ρ: All crops 0.25; barley 0.26; wheat 0.24; oats 0.23; peas 0.20; corn 0.14; rye 0.05 (no clear trend); potatoes 0.40; legumes 0.32; durum wheat 0.24; fodder corn 0.33. Larger crops by sample size (barley, oat, wheat) often show statistically significant relationships.
• Mixed-effects model: A one-unit increase in LF-NPC reduces the yield gap by about 4.4 percentage points on average. Random intercept variance by crop is significant; random slopes dropped due to limited variation across crops; n=326 region-crop observations modeled.
• Income effects: Across regions, higher LF-NPC is linked to better income outcomes under reduced pesticide use. A simple regression of income change (%) on LF-NPC gives y = 4.20x − 9.85 (p < 0.01), adjusted R2 = 0.46; n = 211. On average, a one-unit increase in LF-NPC corresponds to about a 4.4 percentage point increase in productivity and a similar increase in income.
• Regional crop mix matters: Regions specializing in crops with larger pesticide-related yield gaps (e.g., cereals) face larger revenue losses when reducing pesticides, unless compensated by high LF-NPC and sizable pesticide cost savings. In regions where pesticide costs are a large cost share and LF-NPC is high, cost reductions can partly or fully offset revenue losses.

The study demonstrates that landscape complexity supporting natural enemies mitigates productivity losses when synthetic pesticide use is reduced. By quantifying how LF-NPC reduces yield gaps and improves income, the analysis provides evidence that investing in landscape features can support both environmental targets and farm profitability. While price feedbacks in agri-food markets can influence overall income, the key result is the relative economic advantage of regions with higher LF-NPC potential. The findings align with ecological theory that enhanced habitats for natural enemies reduce pest pressure and associated yield losses, and they offer a generalizable, EU-wide quantification that transcends local variability observed in field studies. However, benefits vary with regional crop mix and cost structures, and LF-NPC’s effectiveness at local scales depends on pest pressures, enemy–pest balance, climate, and management practices.

This work provides an EU-wide quantification of the agronomic and economic benefits of landscape features that support natural pest control under reduced pesticide use. A one-unit increase in LF-NPC is associated with about a 4.4 percentage point reduction in yield gaps and similar gains in farm income, underscoring LF-NPC’s role as a partial substitute for synthetic pesticides. The integrated approach—combining farm-level data, spatial LF-NPC mapping, and a partial equilibrium model—offers actionable insights for policymakers and farmers to guide investments in landscape design to meet pesticide reduction targets. Future research should (i) quantify the costs of landscape redesign, (ii) expand biodiversity metrics beyond flying invertebrates, (iii) incorporate process-explicit biodiversity–pesticide interactions into economic models like CAPRI, and (iv) evaluate complementary on-farm diversification strategies (intercropping, cover crops) alongside landscape features.

• Data constraints: FADN lacks direct pesticide quantity/type and precise farm geolocation; yield gaps inferred from organic status and modeled at regional level. Outlier handling and covariates mitigate but cannot eliminate bias.
• Scope of LF-NPC indicator: Based on flying invertebrates; excludes ground-dwelling natural enemies and other taxa; thus represents a proxy for broader NPC potential.
• Economic valuation scope: Focuses on direct agricultural income; excludes other benefits of reduced pesticide use (e.g., water purification, health) and other ecosystem services from landscape features (e.g., erosion control, carbon sequestration). Costs of implementing landscape redesign not estimated.
• Model structure: Mixed-effects model may understate true yield gaps if some management practices narrowing gaps are insufficiently captured. CAPRI lacks a biodiversity response module to pesticide reduction.
• External validity: LF-NPC’s local effectiveness is context-dependent (pest pressure, climate, management), so regional trends do not guarantee local outcomes.