Higher productivity in forests with mixed mycorrhizal strategies

Plant diversity can increase ecosystem functioning such as forest productivity, but the mechanisms—particularly resource complementarity—are difficult to quantify in forests due to trait plasticity and the logistical challenges of measuring functional traits for many tree species across large scales. Mycorrhizal associations represent a key, understudied dimension of plant functional variation. Nearly all tree species form either arbuscular mycorrhizal (AM) or ectomycorrhizal (ECM) associations, which differ in nutrient acquisition strategies, vertical rooting/hyphal distribution, and effects on nutrient cycling. The study asks how tree mycorrhizal dominance (single strategy dominance vs. mixed AM–ECM composition) influences forest productivity across broad spatial scales, and whether these effects depend on tree taxonomic diversity. Two hypotheses guided the work: (1) Mixed mycorrhizal forests exhibit greater productivity than AM- or ECM-dominated forests due to resource partitioning (predicting a concave-negative relationship between AM dominance and productivity). (2) Positive effects of mycorrhizal mixing on productivity are stronger at low tree species richness than at high richness, where other functional differences can promote complementarity.

The paper situates mycorrhizal strategies as a tractable proxy for functional diversity relevant to nutrient acquisition and ecosystem functioning. AM and ECM fungi differ in enzymatic capabilities and nutrient foraging: ECM mycelia often access organic nutrient pools and proliferate in organic horizons, whereas AM fungi rely more on saprotrophic mineralization and occupy upper mineral layers, implying vertical and chemical partitioning of N and P. Tree species associated with ECM typically produce lower-quality, slower-decomposing litter, while AM-associated trees produce higher-quality litter, reflecting conservative vs. acquisitive nutrient strategies. These contrasts suggest potential complementarity when AM and ECM trees co-occur, yet large-scale tests in forests have been limited. Prior work shows mycorrhizal fungi can mediate plant diversity–functioning relationships and influence plant distribution, nutrient cycling, and carbon storage, motivating a continental-scale assessment of mycorrhizal composition and productivity.

Data source and plots: The study used publicly available US Forest Service FIA data, retaining the most recent undisturbed natural forest census per plot following established protocols. Each FIA plot (0.067 ha) consists of four 7.32 m radius subplots (168 m²) spaced 36.6 m apart. Stems with DBH > 12.7 cm were measured; stand age derived from dendrochronology. After excluding records with missing covariates or negative productivity, 74,563 naturally forested plots across the contiguous USA remained. Ecoregions were defined per the National hierarchical framework; 36 were delineated and one with only two plots was dropped (35 ecoregions overall; ecoregion-level analyses excluded regions with <50 plots). Variables: Productivity was estimated as mean annual increment (total aboveground live biomass divided by stand age), representing longer-term growth. Local diversity was tree species richness (also compared with Hill numbers q=1 and q=2 in sensitivity analyses). Mycorrhizal strategy assignment used species-level data (FungalRoot) for 314 species; for 63 species, genus-level assignment was used when >67% of observations agreed. Tree species were categorized as AM or ECM; rare non-mycorrhizal or ericoid species were excluded. Mycorrhizal composition (AM or ECM proportion) was computed as the basal area of AM (or ECM) trees divided by total stand basal area; dual AM/ECM species’ basal area was split 50/50. AM and ECM proportions were near-complementary (96.2% of plots had AM+ECM >0.99). Environmental covariates included mean annual temperature (MAT), mean annual precipitation (MAP), temperature seasonality (WorldClim, 1 km), soil pH (GSDE, 1 km), elevation, slope, and stand age; physiographic class explained little variation and was excluded. Statistical analyses: (1) General linear models (GLMs) modeled log-transformed productivity as a function of ecoregion (factor), AM proportion (linear and quadratic terms), log species richness, AM proportion × ecoregion interactions, and AM proportion × richness interactions, with stand age, elevation, slope, MAT, MAP, temperature seasonality, and soil pH as covariates. (2) To test richness dependence, plots were split into low richness (≤5 species) and high richness (>5 species), and GLMs without richness terms were fit separately; robustness was checked with alternative cutoffs (4, 6, and three-bin splits). Ecoregion-level GLMs (excluding regions with <50 plots) included environmental covariates. Analyses were repeated with ECM proportion instead of AM. All environmental predictors were scaled 0–1. (3) Random forest models assessed relative importance of predictors, including linear and quadratic AM proportion, richness, stand age, MAT, MAP, temperature seasonality, soil pH, elevation, slope, and spatial coordinates (latitude, longitude) to capture spatial structure; an alternative model without elevation and slope yielded similar explanatory power. (4) Piecewise structural equation models (SEMs) tested direct and indirect effects of climate (MAT, MAP, temperature seasonality) and soil pH on productivity via AM proportion (linear and quadratic), fit separately for low- and high-richness plots; model fit was assessed with Shipley’s test of d-separation. Robustness: An independent managed-forest dataset from Indiana’s Continuous Forest Inventory (CFI; 2,771 plots) was analyzed with a GLM relating productivity (periodic annual increment in basal area, 2012–2020) to ECM proportion, yielding patterns consistent with FIA results.

• Across all plots, productivity exhibited a concave-negative relationship with AM proportion (and a similar pattern with ECM proportion), indicating higher productivity in mixed-mycorrhizal stands than in AM- or ECM-dominated stands. Linear and quadratic AM proportion terms were highly significant (p < 0.001). - Ecoregion explained the most variance in productivity, followed by species richness, which had a significant positive effect (p < 0.001). Nevertheless, AM proportion (linear and quadratic) explained more variation than environmental covariates. - Ecoregion-specific analyses: 26 of 34 ecoregions (~76.5%) showed significantly concave-negative relationships between productivity and AM proportion; 4 showed significantly linear-negative relationships; 4 were non-significant. - Random forest models (overall R² = 0.69): After accounting for spatial coordinates, species richness was the most important predictor; linear and quadratic AM proportion had comparable importance. Among environmental variables, MAP and soil pH were most important. - Richness dependence: Interactions of AM proportion (linear and quadratic) with richness were significant (p < 0.001). In low-richness plots (≤5 species), the concave-negative relationship was strong and significant (both linear and quadratic terms p < 0.001), whereas in high-richness plots (>5 species) the relationship was weak (linear p < 0.001; quadratic p = 0.471). Ecoregion-level results: 23/34 low-richness ecoregions showed significantly concave-negative relationships; in high-richness plots, only 2/16 ecoregions did so. These patterns were robust to alternative richness cutoffs and alternative diversity indices (q=1, q=2). - SEMs: Low richness—AM proportion (std. coef. = 0.18) and AM proportion² (std. coef. = -0.39) had stronger effects on productivity; MAP had stronger direct effects on productivity, while MAT and temperature seasonality had stronger direct effects on AM proportion; soil pH had positive direct effects on AM proportion and negative direct effects on productivity (mediated by temperature seasonality). High richness—effects of AM proportion and AM proportion² were weaker (0.08 and -0.08), and climatic variables and soil pH had weaker impacts on AM proportion but MAP and temperature seasonality directly influenced productivity. - Regional exceptions: In several arid western ecoregions with low MAP, AM dominance correlated linearly negatively with productivity, potentially reflecting greater drought tolerance and water-transport advantages in ECM-associated species and limited regional species pools.

The findings support the hypothesis that mycorrhizal mixing enhances forest productivity, consistent with resource complementarity in nutrient acquisition between AM and ECM associations. Contrasting nutrient foraging strategies, vertical soil exploitation, and litter-mediated nutrient cycling likely allow AM and ECM tree species to partition soil resources and increase total community nutrient uptake. Additional mechanisms may include resource enrichment (e.g., AM-facilitated decomposition in ECM-dominated contexts) and complementary biotic defenses against pathogens and herbivores across mycorrhizal types. The diminished effect of mycorrhizal mixing at higher species richness suggests that other dimensions of functional diversity and niche differentiation among many coexisting species can saturate resource space, reducing additional benefits from mycorrhizal complementarity. Context dependence emerged in arid regions, where water-related traits and ECM advantages in water transport may dominate productivity patterns, and limited species pools may accentuate selection effects of particular species. Climatic conditions and soil pH influence both mycorrhizal composition and productivity directly and indirectly, especially in species-poor forests, implying that environmental change can shift mycorrhizal composition and, in turn, forest functioning.

At continental scale, forests with mixed AM–ECM mycorrhizal strategies are generally more productive than those dominated by a single strategy, with the strongest benefits in species-poor communities. In species-rich forests, high productivity appears achievable via multiple complementary trait dimensions irrespective of mycorrhizal composition. The study highlights mycorrhizal strategy as a key axis of functional diversity for predicting and managing ecosystem functioning. Practical implications include promoting mixtures of tree species with diverse mycorrhizal types to enhance productivity, particularly in mesic temperate forests and restoration plantings. Future research should integrate mycorrhizal functional diversity into land surface and Earth system models, leverage emerging remote sensing approaches to map mycorrhizal associations and productivity globally, and resolve contributions of fungal vs. plant trait differences across finer taxonomic and functional gradients.

• Potential selection effects: observed patterns may partly reflect particularly productive ECM species or less productive AM species in some regions rather than complementarity per se, although consistency across most ecoregions argues for a general effect. - Overestimation of ecophysiological differences between AM and ECM associations in prior work that did not account for phylogenetic correlations; caution is warranted in mechanistic interpretation. - Ambiguity in whether patterns derive from mycorrhizal fungal traits vs. correlated suites of host plant traits; disentangling these contributions requires further study. - Grouping all taxa into two broad mycorrhizal types (AM vs. ECM) overlooks substantial within-group diversity in tree and fungal strategies; finer-resolution classifications may improve predictions. - Potential bidirectional causality between mycorrhizal composition and soil pH, complicating causal inference. - Soil pH estimates from harmonized datasets may introduce uncertainty, though applied consistently across analyses; negative-productivity plots were excluded, potentially biasing against recently disturbed sites.