Blue and green food webs respond differently to elevation and land use

The study investigates whether and how multi-trophic ecological networks respond to key environmental drivers, focusing on elevation (as a proxy for climate, especially temperature) and land use. While climate and land use are known to shape species richness and composition, less is known about their effects on the structure of whole food webs across environmental gradients. Aquatic (blue) and terrestrial (green) food webs are often studied separately and are known to differ structurally—blue webs typically show longer chains and pronounced body-size constraints yielding nested structures, whereas green webs often have shorter chains and higher modularity. However, systematic, landscape-scale comparisons of blue versus green food-web responses to shared drivers are lacking. The research aims to: (i) quantify how multi-trophic food webs vary along elevation and across land-use types, and (ii) test whether blue and green food webs respond differently to these drivers. Using a metaweb approach combined with broad-scale, standardized species occurrence data, the authors infer local food webs and analyze structural and ecological properties along environmental gradients.

Prior work has examined network structural changes along environmental gradients and their implications for persistence and functioning, but typically in simplified or bipartite systems (e.g., plant–herbivore, host–parasitoid) or experimental contexts. Known contrasts between aquatic and terrestrial webs include greater nestedness and body-size-driven interactions in aquatic systems and higher modularity with specialized consumer-resource pairs in terrestrial systems. Despite these insights, cross-biome, multi-trophic comparisons along shared gradients remain rare. The metaweb framework offers a standardizable tool to infer local food webs from regional interaction knowledge and local species co-occurrence, enabling comparisons across disparate taxa and environments.

Design and data: The authors used a metaweb-based inference to assemble local food webs across Switzerland (~42,000 km²) using standardized occurrence data for six focal groups: terrestrial plants, butterflies (larval stage), grasshoppers, birds; and stream invertebrates and fishes for aquatic systems. Occurrence data came from national monitoring programs (BDM/FOEN for plants, butterflies, birds, stream invertebrates; info fauna/CSCF for grasshoppers and fishes; additional fish data from Progetto Fiumi). Species were generally resolved to species level (EPT insects at species; other stream invertebrates at family level), with some multi-species complexes treated as single nodes. Co-occurrence was defined at 1×1 km² grid resolution (5×5 km² for grasshoppers and fishes to match mobility and data sources). Metaweb assembly: Trophic interactions among focal taxa were compiled from literature, datasets, and expert knowledge for Central European taxa, yielding a regional metaweb. Local food webs were inferred by including trophic links among co-occurring taxa under the assumption that metaweb interactions can realize when species co-occur. Isolated nodes (without local trophic links) were excluded. In aquatic webs, basal resources were simplified to three ubiquitous mega nodes (plant material, plankton, detritus) due to limited primary producer occurrence data. Sampling and gradients: The study assembled 462 terrestrial (green) and 465 aquatic (blue) food webs spanning 249–2834 m a.s.l., stratified across Switzerland. Land cover was classified into five dominant land-use types per grid (>50% area): forests, scrub, open spaces, farmlands, urban areas. Elevation was used as the primary predictor (encapsulating temperature variation), supported by analyses showing residual temperature added little explanatory power after accounting for elevation. Metrics: Five properties were quantified: number of nodes (local richness), connectance (realized links / potential links), nestedness (UNODF index, 0–1), modularity (0–1), and consumers’ diet niche overlap (Horn’s index, 0–1). Analyses included: principal component analysis (PCA) of structural metrics; piecewise structural equation modeling (SEM) to assess direct and indirect effects of elevation on metrics and interdependencies among metrics; generalized additive models (GAMs) across all webs to detect nonlinear elevational patterns; linear models within land-use subsets with slope comparisons between blue and green; and two randomization baselines (keep-group randomized preserving focal-group composition and fully randomized preserving size/connectance) to parse drivers of nestedness, modularity, and niche overlap patterns.

- Structural contrasts: PCA revealed clear differences between blue and green webs (PC1 explained ~70%). Across 927 webs, blue webs were smaller (median nodes 35) with higher connectance (0.25) and lower modularity (0.03) than green webs (median nodes 437; connectance 0.06; modularity 0.20). - SEM results: Elevation significantly associated with food-web properties, but with contrasting directions between biomes. In green webs, elevation directly increased modularity (standardized coeff. 0.49) and decreased consumers’ niche overlap (-0.53). In blue webs, the relationships were opposite (modularity: -0.38; niche overlap: +0.10, marginal). Land-use-specific SEMs showed similar contrasts, especially pronounced in farmlands. - Richness (nodes) along elevation: Green webs showed a hump-shaped pattern—nodes increased with elevation up to ~1500–2000 m a.s.l. (tree line), then declined. Blue webs decreased approximately linearly with elevation. Within land-use types, blue–green slope differences were most marked in farmlands: green webs grew with elevation (more plants and butterflies), while blue webs shrank (fish losses outpaced invertebrate gains). - Connectance along elevation: Green webs decreased nearly linearly with elevation. Blue webs decreased up to ~1000 m a.s.l. and then mildly increased. The initial decline reflected gradual loss of fish (generalist diets), reducing links; above ~1000 m (“fish line”), fish rarity and dominance of invertebrates combined with smaller web size and replacement by more generalist invertebrates yielded a mild connectance increase. - Nestedness: Both systems showed decreasing nestedness with elevation. In blue webs, nestedness dropped until ~1000 m and then plateaued, matching the fish-line effect; loss of generalist fish strongly reduced nestedness beyond link-density effects. - Modularity: Green webs exhibited a mild increase in modularity with elevation (particularly below ~2500 m), driven by increased richness of specialist consumers (notably butterflies) at lower elevations until the tree line; above the tree line, opposing effects (declining specialists vs. smaller, sparser webs) produced little net change. Blue webs were less modular than random expectations across elevations. - Consumers’ diet niche overlap: Compared to fully randomized webs, overlap was higher in both systems, but relative to keep-group randomized baselines, green webs had lower and blue webs higher overlap—indicating more diet differentiation among terrestrial consumers and more overlap among aquatic consumers than expected by chance. Elevational patterns were nonlinear and contrasting: green overlap decreased to ~1500–2000 m then increased; blue overlap increased to ~1000 m then stabilized. - Land use interactions: For all five metrics, blue–green regression slopes differed significantly in farmlands, with multiple metrics showing opposite signs (e.g., modularity increasing with elevation in green and decreasing in blue; niche overlap decreasing in green and increasing in blue). This highlights interactive effects of climate (elevation/temperature) and anthropogenic land use, most pronounced in agricultural landscapes.

The study shows that blue and green food webs, though co-occurring within the same landscapes, exhibit divergent structural and ecological responses to elevation and land use. Elevational patterns capture combined influences of temperature and, particularly in streams, topography and hydrology. In terrestrial systems, highly mobile consumers (butterflies, grasshoppers, birds) respond primarily to temperature-driven vegetation changes, producing tree-line-associated nonlinearities. In aquatic systems, a strong topographic/hydrological constraint creates a ~1000 m “fish line,” beyond which fish are rare; this threshold explains rapid changes in connectance and nestedness and a shift toward invertebrate-dominated, more overlapping diets. Inherent biome differences—co-evolved specialization and higher modularity in green webs versus body-size/gape-driven generalist feeding and nestedness in blue webs—underpin opposite elevational trends in modularity and niche overlap. Land use modulates these responses, with farmlands showing the clearest blue–green contrasts; agricultural simplification at low elevations favors generalists and reduces plant and specialist insect diversity, altering web structure. Management implications include maintaining diverse plant resources in terrestrial systems to support specialists at higher elevations, and stabilizing resource quantities in aquatic systems where high dietary overlap can intensify competition as resources decline. The findings suggest that climate warming and land-use change may have opposing or interacting effects across biomes, necessitating coordinated conservation strategies that account for cross-system differences and constraints.

By integrating a metaweb with extensive, standardized occurrence data, the study provides landscape-scale evidence that aquatic (blue) and terrestrial (green) food webs differ fundamentally in structure and respond in contrasting ways to elevation and land use. Key contributions include: (i) demonstrating opposite elevational trends in modularity and diet niche overlap between biomes, (ii) identifying tree-line and fish-line thresholds shaping nonlinear patterns, and (iii) revealing that farmland contexts amplify blue–green divergences. These insights inform biodiversity and ecosystem management under climate and land-use change. Future work should expand taxonomic coverage (including less-studied groups and detailed aquatic basal producers), incorporate cross-biome interactions where co-occurrence permits, refine interaction variability (e.g., ontogeny, adaptive foraging) beyond fixed-diet assumptions, and deploy synchronous blue–green sampling to track temporal dynamics and cross-ecosystem linkages.

- Interaction inference assumes fixed species diets from the metaweb realized upon co-occurrence, potentially overestimating local links and not capturing intraspecific variation (resource availability, predation risk, temperature, ontogeny). - Aquatic basal resources were simplified into three mega nodes (plant material, plankton, detritus) due to limited occurrence data, reducing resolution of basal diversity and interactions. - Taxonomic coverage excludes some less-studied groups; some stream invertebrates resolved to family level and some terrestrial complexes aggregated, potentially smoothing interaction detail. - Isolated nodes (without local trophic links) were removed, possibly omitting transient or non-focal interactions. - Blue–green interconnections (e.g., bird–fish links) were not included due to limited co-occurrence at the grid scale, precluding fully integrated cross-ecosystem webs. - Elevation used as primary predictor encapsulates temperature effects; residual temperature effects beyond elevation were limited in this dataset but may matter in other contexts.