Global systematic review with meta-analysis shows that warming effects on terrestrial plant biomass allocation are influenced by precipitation and mycorrhizal association

The study investigates how climate warming alters plant biomass allocation between roots and shoots (root:shoot ratio, R/S) across terrestrial ecosystems and plant lineages. R/S is a key parameter for estimating terrestrial carbon storage and reflects optimal allocation strategies that shift with resource availability. While global change drivers such as drought and elevated CO2 often increase allocation to roots, the immediate effects of warming on R/S across environmental contexts remain uncertain. The authors hypothesize: (1) warming induces a vertical shift in biomass allocation toward roots (intercept of warmed vs ambient R/S > 0); (2) warming homogenizes R/S across biomes (slope of warmed vs ambient R/S < 1); and (3) mean annual precipitation (MAP) is a principal determinant of R/S responses due to its control over belowground resources. They also test whether plant-mycorrhizal associations modulate these responses.

Background literature indicates that R/S varies with resource availability and global change factors: drought and elevated CO2 tend to increase root allocation, whereas increased precipitation and nitrogen deposition favor shoot allocation. Warming can influence net primary production and potentially create soil nutrient limitations that alter carbon economics and allocation. Climatic antecedents (mean annual temperature and precipitation) are known correlates of R/S at broad scales, but it is unclear if these long-term patterns match short-term warming responses. Belowground ecosystem properties, including root traits and root–soil interfaces, and especially mycorrhizal symbioses, influence plant nutrient foraging and may shape allocation responses. Arbuscular (AMF) and ectomycorrhizal (EMF) fungi differ in physiology and nutrient acquisition strategies and vary in dominance across biomes, potentially explaining biome-level variability in allocation responses.

The authors conducted a global meta-analysis of experimental warming studies (1950–2020) identified via ISI Web of Science using predefined keywords. Inclusion criteria required terrestrial warming experiments reporting at least one of: total biomass (TB), aboveground biomass (AGB), belowground biomass/root biomass (BGB), root:shoot ratio (R/S), water use efficiency (WUE), soil inorganic nitrogen (SIN), soil NH4+, soil NO3−, microbial biomass (MB), or microbial biomass C/N, with clear warming/control temperatures, consistent initial conditions and species composition, duration exceeding one growing season, and extractable means, variation, and sample sizes. In total, 322 studies (280 field, 42 laboratory) were compiled, covering cropland, desert, forest, grassland, tundra, and wetland biomes. Observations for control and warming were digitized (GetData v2.22). Climate data (MAT, MAP) were taken from publications or WorldClim when necessary; soil properties (0–30 cm BD, CLAY, SOC) were sourced from IGBP-DIS and HWSD. Root/BGB measurements included direct harvest, soil cores, and ingrowth bags. Warming magnitude (WM) and duration (DUR) spanned 0.26–12 °C (median 2 °C) and one growing season to 25 years (median 2 years). Biome, plant functional type (PFT), taxonomic categories, and mycorrhizal fungi type (MFT: AMF, EMF, AM-EMF) were assigned using publications and FungalRoot. Effect sizes were computed as natural log response ratios (RR = ln(X/Xc)); subgroup weighted mean effect sizes (RR++) and standard errors were calculated using inverse-variance weighting. Significance was assessed by 95% confidence intervals of RR++; percentage change was [exp(RR++)−1]×100%. Heterogeneity among groups used Q-statistics. Predictors of RR(R/S) were analyzed with stepwise linear regression and model selection via Akaike weights (glmulti). Variance partitioning used nested models (nlme). Relationships between MAP and RR(R/S) were tested using linear regression (metafor). Phylogenetic signal (Blomberg’s K) was tested with picante. Structural equation modeling (lavaan) assessed direct and indirect effects of PFTs, MFTs, climate (MAT, MAP), and warming treatments (WM, DUR) on RR(R/S) via RR(AGB) and RR(BGB).

- Across 94 paired observations with mean warming of 2.50 °C (median 2 °C), R/S increased by 6.1–8.8% (mean +8.1%; p < 0.01). The intercept of warmed vs ambient R/S was positive (0.044–0.107; CI > 0; p < 0.05), indicating greater vertical allocation to roots. The slope was <1 (0.908–0.997; CI < 1; p < 0.05), indicating horizontal homogenization of R/S across biomes. - Biomes with higher ambient R/S exhibited greater variation in R/S responses to warming. - MAP was the most important predictor of RR(R/S); RR(R/S) decreased with increasing MAP (global R² = 0.104), with the weighted RR(R/S) switching from positive to negative at MAP > ~900 mm. - Mycorrhizal fungi type (MFT) was the second most important predictor. The sensitivity of RR(R/S) to MAP was smaller in EMF-dominated biomes than in AMF-dominated biomes (AMF R² ≈ 28.84%, EMF R² ≈ 28.24% for MAP–RR(R/S) regressions). Interactions occurred between MFT and biome, but MFT explained more variance in RR(R/S) than biome alone. - Warming magnitude and duration did not significantly affect RR(R/S) overall. - Biomass components: Overall, TB increased by 10.0%, driven by BGB increase of 13.1%, while AGB decreased by 8.9% (not significant). In AMF systems, TB and BGB did not change, AGB decreased by 13.8%, and R/S increased by 9.9%. In EMF and AM-EMF systems, TB, AGB, and BGB increased; R/S increased significantly only in EMF-associated plants. - Path analysis: MFTs explained 22.9% of variation in RR(AGB); PFTs and MAP were major pathways for RR(BGB) (25.4% explained). RR(AGB) (negative effect) and RR(BGB) (positive effect) jointly explained 46.3% of variation in RR(R/S). RR(R/S) correlated with RR(BGB) (R² = 0.31, p < 0.001) but not with RR(AGB) (R² = 0.01, p = 0.392). - No consistent differences in warming responses of biomass variables by evergreen vs deciduous, broadleaf vs conifer, annual vs perennial, or angiosperms vs gymnosperms when MFT was ignored.

Findings support that warming enhances belowground allocation (higher R/S), particularly in drier habitats where soil moisture limitation intensifies root foraging and resource acquisition, potentially lengthening carbon residence times via slower turnover of root-associated carbon. Biomes with initially low ambient R/S shifted more toward root allocation under warming, leading to homogenization of R/S across biomes. MAP emerged as a key regulator: below ~900 mm MAP, warming likely promotes deeper rooting and water acquisition; above ~900 mm MAP, adequate moisture enhances nutrient turnover, reducing relative investment belowground, especially in AMF-dominated systems. Mycorrhizal associations modulate mechanisms: EMF-associated plants, with enzymatic access to organic nutrients, tended to increase AGB, BGB, and R/S under warming, whereas AMF-associated plants showed decreased AGB and higher WUE under drier, warmer conditions, raising R/S mainly via shoot reduction. PFTs influenced BGB responses, with woody plants showing stronger BGB increases potentially via hydraulic lift, whereas herbaceous plants adjusted root traits rather than biomass. Antecedent MAT also affected responses, with negative relationships to R/S in cooler biomes due to warming-stimulated AGB. Overall, these patterns imply shifts in carbon allocation affecting soil–atmosphere carbon exchange in a warmer world.

A global meta-analysis reveals that moderate climate warming (~2.5 °C) increases plant allocation to roots (mean +8.1% R/S), reduces inter-biome variability in R/S, and that MAP and mycorrhizal association are key determinants of these responses. Below ~900 mm MAP, warming tends to increase belowground allocation; above this threshold, allocation shifts are dampened or reversed, particularly in AMF systems. EMF-associated plants often increase AGB, BGB, and R/S under warming. These results align with projected mid-century warming and underscore the need to incorporate precipitation regimes and mycorrhizal associations into predictions and models of terrestrial carbon dynamics. Future work could prioritize mechanistic experiments across precipitation gradients and mycorrhizal contexts, and assess the roles of warming magnitude and duration within specific biomes and plant functional types.

Effect sizes showed variability across studies and biomes. While overall warming magnitude and duration did not explain RR(R/S) globally, their biome- and PFT-specific effects may differ. Geographic coverage was concentrated in East Asia, North America, and Europe. Biomes dominated by non-mycorrhizal plants were too few for meta-analysis. Differences in experimental methodologies (warming methods, root/BGB measurement) and study settings (field vs laboratory) existed, though they did not consistently explain variation in RR(R/S).