The positive effect of plant diversity on soil carbon depends on climate

Abstract

Little is currently known about how climate modulates the relationship between plant diversity and soil organic carbon and the mechanisms involved. Yet, this knowledge is of crucial importance in times of climate change and biodiversity loss. Here, we show that plant diversity is positively correlated with soil carbon content and soil carbon-to-nitrogen ratio across 84 grasslands on six continents that span wide climate gradients. The relationships between plant diversity and soil carbon as well as plant diversity and soil organic matter quality (carbon-to-nitrogen ratio) are particularly strong in warm and arid climates. While plant biomass is positively correlated with soil carbon, plant biomass is not significantly correlated with plant diversity. Our results indicate that plant diversity influences soil carbon storage not via the quantity of organic matter (plant biomass) inputs to soil, but through the quality of organic matter. The study implies that ecosystem management that restores plant diversity likely enhances soil carbon sequestration, particularly in warm and arid climates.

Fig. 1. Two structural equation models depicting the original hypothesis and the new, optimized model.

The initial hypothesis (a) states that plant diversity affects SOC through the quantity of organic matter (plant biomass) inputs to soil, whereas the new model (b) states that plant diversity affects SOC though the quality of organic matter (C:N ratio). Quality of plant organic matter is depicted in the drawing of the grassland in panel b by different colors. Gray boxes show interactions. Black arrows indicate significant regressions. Asterisks indicate the level of significance of the regressions (^*P < 0.05, ^**P < 0.01, ^*** P < 0.001), and (+) and (-) indicate whether the slope of the linear regression model is positive or negative. Blue arrows indicate non-significant regressions. The green boxes display the coefficient of determination (R²) for the endogeneous variables. The orange box displays the Akaike Information Criterion (AIC). The models were fitted to the site-level data. The new, optimized model (panel b)was obtained by increasing the model fit of the initial version of the new model (Fig. S6) by removing non-significant regressions. Plant diversity refers to the Shannon index. SOC stands for soil organic carbon. Plant drawings courtesy of Per-Marten Schleuss, used with permission.

Fig. 2. Relationship between Shannon diversity index and soil organic carbon content.

The relationship is shown across all 84 grassland sites (a) as well as across sites with mean annual temperature (MAT) > 15.58 °C (b), sites with mean annual precipitation (MAP) < 523 mm (c), and arid and semi-arid sites, i.e., sites with an aridity index (AI) < 0.50 (d). The linear models were plotted to the site-level data (and not to the plot data, which is shown to give insight into the within-site variability). The subsets of sites shown in panels b, c, and d are the quartiles of sites for which significant correlations were found between Shannon index and soil organic carbon content (see Table 2). For further information on the relationship between Shannon index and soil organic carbon content depending on climate see Figure S2a, c, and e.

Fig. 3. Relationship between Shannon diversity index and soil C:N ratio.

The relationship is shown across all 84 grassland sites (a) as well as across sites with mean annual temperature (MAT) > 15.58 °C (b), sites with mean annual precipitation (MAP) < 523 mm (c), and arid and semi-arid sites, i.e., sites with an aridity index (AI) < 0.50 (d). Note that by definition the aridity index increases with decreasing aridity. The linear models were plotted to the site-level data (and not to the plot data, which is shown to give insight into the within-site variability). The subsets of sites shown in panels b, c, and d are the quartiles of sites for which significant correlations were found between Shannon index and soil C:N ratio (see Table 3). For further information on the relationship between Shannon index and soil C:N ratio depending on climate see Figure S2b, d, and f.

Fig. 4. Soil organic carbon content as a function of climate.

Soil organic carbon content as a function of mean annual temperature (a), mean annual precipitation (b), and aridity index (c) across 84 grasslands. Note that by definition the aridity index increases with decreasing aridity. The linear models were plotted to the site-level data (and not to the plot data, which is shown to give insight into the within-site variability).