Introduction
Fine roots (<2 mm) are the most metabolically active part of root systems responsible for soil nutrient uptake, mycorrhizal symbiosis and interspecific interactions (Valverde-Barrantes et al. 2015; Eissenstat et al. 2015). In most ecosystems, fine roots are concentrated within the upper soil surface (e.g. 0-20 cm) where there are typically higher nutrient availability and lower physical impedance (Schenk & Jackson 2002). However, even within such a shallow profile, species may differ in their vertical relative root distributions, which is defined as the proportions of roots partitioned among soil zones (Herben et al. 2018; Zhou et al. 2020). Such difference may promote species co-existence through a more complete filling of below-ground space (Ma & Chen 2017) and complementary resource use among species (Blume‐Werry et al . 2019). Nevertheless, very few studies have examined species-specific vertical root distributions in diverse forest communities, primarily due to the difficulty in identifying root materials to species. The lack of such information has hindered our understanding of rooting-depth mediated below-ground niche differentiation, root foraging strategies and species spatial distributions (Gale & Grigal 1987; Klimešová et al. 2016).
As root systems are highly plastic organs, vertical root distributions could express high phenotypic plasticity in response to external factors such as edaphic heterogeneity (Sainju & Good 1993; Case et al.2020) and specific root neighbours (Zhang et al. 2019; Chenet al. 2020). Such adjustments may increase root nutrient uptake efficiency through optimizing root placement according to nutrient availability and neighbouring root identities (Mommer et al.2012; Zhang et al. 2019). While effort is increasingly devoted to elucidating the mechanisms and consequences of root segregation (Holdo 2013; Kulmatiski et al. 2017), most studies are biased to artificial communities (Mahall & Callaway 1992; Belter & Cahill 2015; Zhang et al. 2019) or natural communities in arid and semi-arid regions, typically involving two species (Case et al. 2020; Chenet al. 2020; but see Valverde-Barrantes et al. 2015). As the direction and extent of rooting plasticity may be highly dependent on the number and identities of root neighbours, it remains uncertain if and to what extent relative root distributions would be affected by the presence of multiple interacting neighbours in complex forests. Such information is critical for the prediction of how the change of species composition and edaphic conditions may influence competitive interactions and root neighbourhood stability (Chen et al. 2018a, b).
Within soil zones, the degree of interspecific root segregation could be ultimately determined by their innate difference in root system architecture (Schenk & Jackson 2005) and rooting plasticity in response to edaphic heterogeneity or/and neighbour structure (Zhang et al.2019; Case et al. 2020; Zhou et al. 2020). Despite the proposed mechanisms, the generalization of vertical root segregation in natural communities have not been fully identified, with both apparent (Kesanakurti et al. 2011; Herben et al. 2018) and minor rooting differentiation observed (von Felten & Bernhard 2008; Hoekstraet al. 2015). Whether and how rooting differentiation mediates species co-existence remains a subject of intensive studies. Valverde-Barrantes et al.  (2015) recently reported that root depths of temperate trees in the 0-60 cm soil zone were largely overlapped and exhibited low responsiveness to neighbours. However, root depths alone may fail to fully capture root niches, as species with similar root depths may differ considerably in their relative root distributions (Schmid & Kazda 2001; Bennett et al. 2002; Herbenet al. 2018). It is thus unclear whether relative root distribution represents a root niche dimension independent of root depth that may contribute to vertical root segregation.
In this study, we collected > 4000 root segments from 625 soil samples within the 0-30 cm soil zone in a subtropical forest, China. From these samples, we quantified the relative root distributions of 53 woody species, modeled rooting plasticity of the 29 most common species as a function of soil heterogeneity and root neighbours and finally examined how individual roots of the identified 109 species were vertically placed. We used these data to test three hypotheses. First, we predicted high interspecific differentiation in relative root distributions, with some predominantly placing their roots in specific soil zones whereas some others being homogeneously distributed. Second, we predicted high rooting plasticity in response to either edaphic heterogeneity or/and root neighbours that may optimize root placement according to nutrient availability and neighbour root competition. Specifically, due to the potential pre-emptive utilization of soil resources by some proliferative species (Campbell et al. 1991), we predicted that the increase of one species’ root abundance would reduce other species’ root proliferation, resulting in pervasive negative correlations in relative root abundance among species pairs. Finally, depending on the differentiation in relative root distributions among species complementarily combined with high rooting plasticity, we predicted that heterospecific roots would be vertically segregated either simply by avoiding co-occurring or by avoiding placing similar relative root abundance in the same soil zones.