A meta-analysis on decomposition quantifies afterlife effects of plant diversity as a global change driver

The study addresses how changes in plant species diversity influence litter decomposition—an essential process in global carbon and nutrient cycling. Despite extensive research on biodiversity–ecosystem functioning, there has been limited synthesis quantifying the magnitude and direction of plant litter diversity (species mixing) effects on decomposition across biomes. Given accelerating biodiversity loss, habitat simplification, biotic homogenization, and climate change, the research aims to provide a comprehensive, global meta-analysis of litter diversity effects on decomposition and to compare these effects to those expected from climate warming. The central questions are: (1) Do mixed-species litters decompose faster than mono-species litters across ecosystems and climates? (2) How large are these afterlife diversity effects relative to projected climate change impacts on decomposition? This synthesis is intended to inform biogeochemical and Earth system models by quantifying biodiversity’s role as a global change driver.

Prior biodiversity–ecosystem functioning research shows diversity often enhances productivity and can affect decomposition, yet quantitative syntheses specifically targeting litter species-mixing effects across biomes have been scarce. Mechanisms proposed for positive mixing effects include nutrient transfer among litter types with contrasting quality, altered decomposer activity driven by litter traits, and complementarity among decomposers due to resource and habitat heterogeneity. Anthropogenic drivers (e.g., conversion to monocultures, habitat simplification) can reduce plant and decomposer diversity, potentially slowing decomposition. Climate change is known to alter decomposition rates, but models rarely integrate biodiversity-mediated effects. Previous syntheses (e.g., Handa et al. 2014) documented biodiversity–decomposition links, whereas a recent meta-analysis (Porre et al. 2020) suggested small average effects depending on weighting; the current study revisits this at global scale with multilevel meta-analytic methods and expands comparisons across biomes and climates, including an explicit benchmark against climate change projections.

- Literature search and selection: Searched ISI Web of Science (to Dec 2018) using “decomposition AND litter,” refined with terms related to litter diversity/mixing; supplemented by Scopus, Google Scholar, and expert suggestions. From 12,278 initial hits and refined lists (416 and 765), 151 studies met criteria; 131 reported mass loss (or remaining) and 45 reported decomposition rate constant k suitable for effect-size calculation. Grey literature excluded. - Inclusion criteria: Litter-bag experiments comparing mixed- vs mono-species litter; reported means, variance (SD/SE/CI), sample sizes for mass loss or k; same incubation time within comparisons. Where necessary, means/SD derived from medians/boxplots using WebPlotDigitizer; SE/CI converted to SD. Studies without necessary statistics or only reporting additive expectations without measured mass loss/k were excluded. - Dataset composition: Mass loss dataset: 6535 comparisons across 1949 treatments from 131 studies. k dataset: 1423 comparisons across 504 treatments from 45 studies. Ecosystems included forests, grasslands, wetlands, streams, shrubland, desert, seagrass, lakes, and microcosms (terrestrial/aquatic). Litter types included leaves (majority), roots, stems, branches, straw, moss, and macrophytes (aquatic plant litter included in main analysis; leaf-only subsets analyzed separately). Study locations georeferenced; climatic zones categorized (subarctic, boreal, temperate, subtropical, tropical, other). - Effect size and models: Calculated unbiased standardized mean difference (Hedges’ d) for mixed vs mono litter, defined positive when mixed decomposed faster. Addressed non-independence and nested structure (comparisons within treatments within studies) using multilevel random-effects meta-analysis/meta-regression (metafor). Weights combined within- and between-study variance; moderators included ecosystem type and climate zone. Heterogeneity and moderator significance tested with Q statistics (QM). Publication bias assessed via funnel-like plots of effect size vs variance/sample size; no major bias detected. Robustness checked by randomly selecting one comparison per treatment 10,000 times; results remained significantly positive. - Subset and sensitivity analyses: Leaf-only analyses; exclusion of comparisons without species identity; analyses by ecosystem and climate; incubation-time effects (mixed-effects meta-regression on studies with ≥2 retrieval times). Differences in k estimation forms treated as random among-study variance. - Climate-equivalency analysis: To compare litter diversity effects with climate change, constructed standardized climate–decomposition relationships using an independent full reciprocal transplant dataset (Makkonen et al. 2012) spanning subarctic to tropical forests (16 species). Computed Hedges’ d for k relative to the coldest site (subarctic) for identical protocols (same litter identity/origin and decomposer source) to isolate climatic effects. Related effect sizes to bioclimatic variables (WorldClim): (a) annual mean temperature, (b) mean temperature of wettest quarter, (c) precipitation of driest quarter via multilevel mixed-effects meta-regression (protocol as random effect) to obtain slopes. - Using the forest leaf-litter subset of this meta-analysis (57 forest studies), converted observed litter diversity effect sizes into climate-equivalent changes (temperature/precipitation shifts required to match diversity effect magnitude) via the standardized slopes. Compared these to CMIP5 RCP 2.6 and RCP 8.5 projections for the 2070s at study coordinates. - Projected percentage increases in decomposition: Modeled log(k) vs annual mean temperature via LMM (protocol random) from Makkonen et al.; estimated present k at each forest study site; then applied temperature increments equivalent to diversity effects and to projected warming (RCP 2.6/8.5) to obtain projected k and converted to mass loss to report percent increases. - Software: Analyses conducted in R using tidyverse, metafor, lmerTest, effects, maptools, sf, raster; climate data from WorldClim; study coordinates obtained via Google Maps.

- Overall litter diversity effect: Across all studies, mixed-species litter decomposed faster than mono-species litter. Mass loss: significant positive effect (p < 0.0001). Decomposition constant k: significant positive effect (p < 0.01). Results robust for leaf-only data (p < 0.0001 for both metrics) and to sensitivity analyses. - Dataset scope: Mass loss analyses included 6535 comparisons (1949 treatments; 131 studies); k analyses included 1423 comparisons (504 treatments; 45 studies). - Ecosystem-level patterns: Significant positive mixing effects in most biomes (forests, grasslands, wetlands, and others), but not in streams, where effects were non-significant, potentially due to nutrient dynamics, hydrologic constraints on nutrient transfer, enrichment, and disturbance regimes. - Climate and biome interactions: In forests, positive mixing effects tended to be stronger in colder biomes (from tropical to boreal), consistent with increased complementarity under harsher conditions (QM = 8.69, p = 0.069; trend-level support). - Magnitude in forests: Diversifying from mono- to mixed-species litter is estimated to increase decomposition rates by 34.7% (mean) across forest biomes. - Climate-equivalency and comparison to projected warming: The biodiversity-driven increase (34.7%) is comparable to increases projected under climatic warming by the 2070s: 13.6% (RCP 2.6) and 26.4% (RCP 8.5) for forest litter decomposition at the same locations. Climate-equivalency analyses using annual mean temperature, mean temperature of the wettest quarter, and precipitation of the driest quarter corroborated that diversity effects are of similar magnitude to climate-change effects. - Publication bias and non-independence checks indicated minimal bias and robust positive effects (e.g., one-comparison-per-treatment resampling mean Hedges’ d ≈ 0.247 ± 0.045, 95% CI).

The synthesis demonstrates that plant litter species diversity exerts a consistent, positive afterlife effect on decomposition across most ecosystems, addressing the long-standing uncertainty about the magnitude and generality of litter-mixing effects. The findings indicate that biodiversity change is not merely a response variable but a driver of biogeochemical processes, with effects on decomposition comparable to those expected from mid-21st-century climate warming. This emphasizes the need to account for biodiversity in models of carbon and nutrient cycling and climate feedbacks. Biome- and climate-dependent variation—particularly stronger effects in colder forests—suggests interactions between environmental harshness and complementarity mechanisms, aligning with the stress-gradient hypothesis and complementarity theory. The lack of a consistent positive effect in streams highlights context dependence where abiotic constraints or nutrient enrichment may dampen mixing effects, calling for targeted mechanistic research integrating decomposer communities, environmental conditions, and plant functional traits. The results support integrating biodiversity composition into Earth system and land-surface models and suggest that landscape-level decisions (e.g., avoiding extensive monocultures) can have significant consequences for decomposition dynamics and associated carbon fluxes.

This global meta-analysis quantifies strong positive effects of plant litter diversity on decomposition across biomes and shows that, in forests, the magnitude of these effects (≈35% increase) is comparable to decomposition increases projected under climate warming scenarios by the 2070s. By establishing climate-equivalency of biodiversity effects, the study elevates biodiversity change as a key global driver of biogeochemical cycles. The work underscores the importance of incorporating biodiversity, biotic interactions, and vegetation composition into biogeochemical and climate models. Future research should: (1) disentangle mechanisms across environmental gradients with coordinated cross-site experiments; (2) integrate decomposer biodiversity and functional traits into analyses; (3) resolve context-dependent outcomes in aquatic systems; (4) assess effects across broader species richness levels and over time; and (5) leverage remote sensing to map biodiversity composition for model integration.

- Species richness breadth: Most studies involved mixtures of only two or three species, limiting inference to higher richness levels typical of some natural communities. - Temporal and context dependence: Biodiversity–decomposition relationships vary over time, systems, and conditions; incubation period influenced effect sizes, and stream systems showed no general positive effect. - Decomposer biodiversity not explicitly analyzed: The role of decomposer community diversity and composition could not be quantified, despite its known importance for decomposition. - Climate–decomposition model scope: Standardized climate-decomposition relationships (from a four-site transplant) may not capture all climatic gradients or interactions; projections rely on specific bioclimatic variables and assumptions. - Data gaps and heterogeneity: Some comparisons lacked species identities or used different measurement bases (e.g., ash-free dry mass), though sensitivity analyses suggest main conclusions are robust. Potential pseudo-replication was addressed via multilevel models, but inherent non-independence remains a consideration.