Fig. 1: Hypotheses and actual warming effects on root:shoot ratio.

a–d Hypotheses of the warming effect on root:shoot ratio (R/S). e The actual relationship between log10-transformed R/S at experimental warming and ambient temperature. f The frequency distribution of response ratio of R/S. g The log10-transformed residuals of R/S under the warmed condition from 1:1 line along with log10-transformed R/S under ambient condition. The error bands in panel e represent the upper and lower 95% confidence intervals. Across studies in our meta-analysis, the intercept in panel e was larger than 0, implying greater biomass is allocated (vertically) downwards, i.e., greater R/S under warming than that at ambient temperature (a, e); the opposite scenario of downward allocation (i.e., upward allocation, intercept smaller than zero) is illustrated in panel (b). When the slope is smaller than 1, it would imply that warming decreases the horizontal variability of R/S and thus causes the homogenization of R/S among diverse biomes (c, e); the opposite scenario of homogenization (differentiation, slope >1) is illustrated in panel (d). The size of each dot in panel e indicates the relative weight of the individual response ratio of R/S. The p-value of the normal distribution test of RR(R/S) was 0.3818. The RR₊₊ indicates the weighted response ratio of R/S in panel (f); the ‘n’ in panel (f) is the sample size of the response ratio of R/S.