Are fine roots ‘leaves underground' in terms of allometry? A test in a tropical forest successional series in southwest China

Fine roots have been hypothesized to be ‘leaves underground' in terms of vascular network, but this hypothesis has rarely been tested within the framework of metabolic scaling theory (MST). We measured average fine‐root (diameter < 1 mm) mass (M), surface area (A), volume (V), diameter (D) and length (L) for 216 soil cores from 24 plots across four successional stages in tropical forests of Xishuangbanna (southwest China), and examined eight scaling relationships between these variables at the individual root scale. We tested whether fine‐root allometries conformed to MST's model for leaf (MST l ) or model (MST w ) for woody organs (e.g. trunk). We also assessed the relative effects of environmental factors, tree size, species composition and diversity, and stand structural factors on allometric relationships using structural equation models (SEMs). Our results showed that: 1) fine‐root scaling exponents rarely conformed to MST l 's predictions. 2) The scaling exponents between fine‐root M, A, V and D all conformed to MST w 's predictions in later successional forests, but showed greater deviation towards early successional stage. 3) The scaling exponents associated with fine‐root length differed markedly from MST w 's predictions. 4) Changes of some fine‐root scaling exponents across successional stage were mainly affected by tree size or soil fertility, and species composition affected allometry only indirectly via tree size. Our results suggested that the allometries of individual fine roots largely conform to the scaling rules governing woody organs instead of leaves, probably because leaves are nearly two‐dimensional objects while the other two are three‐dimensional. We showed that MST w can well predict some fine‐root allometries in later successional forests, suggesting great potential of utilizing MST w to better estimate fine‐root biomass and productivity. However, the present MST w still needs to be improved for predicting the scaling relationships concerning fine‐root length, and also for better quantifying allometric exponents in earlier successional forests.