1 Introduction
The dimensions of rooting systems, indicated by root depth and lateral extent, represent the amount of soil space occupied, which determines the uptake of resources by fine roots (Schenk and Jackson, 2002; Zhou et al., 2020; Freschet et al., 2021). Adjustment of root distribution in soil profiles is crucial for plant resource acquisition and can thus affect plant resilience to environmental stress (Fan et al., 2017; Zhou et al., 2020). For example, deep roots are a key strategy for xerophytes to survive in arid ecosystems (Schenk and Jackson, 2002; Zhou et al., 2020; Fan et al., 2017). Numerous studies have found that fine roots of woody plants are concentrated in shallow soil (Huang et al., 2008; Zewdie et al., 2008; Gwenzi et al., 2011; Gao et al., 2020; Li et al., 2020; Yang et al., 2021), whereas deep roots occur in seasonally dry areas (Schenk and Jackson, 2005). The fine root vertical pattern of many plants exhibits high plasticity under soil heterogeneity (Padilla and Pugnaire, 2007; Li et al., 2020; Luo et al., 2021). Therefore, understanding the pattern of root distribution across environmental gradients and plant ages may provide insight into the adaptive  processes of plant under environmental stress.
Soil heterogeneity is caused by the patchy distribution of coarse particles and nutrients in soil profiles (Huang et al., 2023a, 2023b). Coarse parts of soil (particle size > 2 mm), usually called rock fragments, are widespread in terrestrial ecosystems and their content plays an important role in soil hydrological conditions such as infiltration (Poesen and Lavee, 1994; van Wesemael et al., 2000; Zhou et al., 2009; Zhang et al., 2011; Zhang et al., 2016). Many soils contain rock fragments as a result of processes of soil genesis and human activities such as tillage and resource extraction (Hu et al., 2021). Rock fragment content (RFC) is critical for other soil properties, such as soil physical structure (Xu et al., 2012; Gargiulo et al., 2015, 2016), water and nutrient availability (Rytter, 2012; Qin et al., 2015; Zhang et al., 2016; Ceacero et al., 2020), and microbial composition (Certini et al., 2004; Hong et al., 2021; Huang et al., 2023a). In soils with high RFC (Rytter, 2012; Ceacero et al., 2020; Huang et al., 2023b), growth and biomass accumulation decreases in plants (Mi et al., 2016; Hu et al., 2021), but the allocation ratio of the below-ground parts increases (Hu et al., 2021). The rocky soil, water stress, and infertile environment limits agroforest development of the arid area. However, little is known about the vertical distribution pattern of fine root profiles in response to changes in RFC.
Plants adjust the soil space colonized by roots to obtain sufficient resources to cope with variations in the soil particle composition (Schenk and Jackson, 2002, 2005; Bengough, 2003). A strong sensitivity of rooting depth to local soil hydrological conditions has been identified (Schenk and Jackson, 2005; Fan et al., 2017; Zhou et al., 2020), whereby roots remain shallow in waterlogged land, but usually deepen in arid regions (Schenk and Jackson, 2002; Fan et al., 2017). Soil hydrology is determined by soil texture (Sperry and Hacke, 2002; Fan et al., 2017); coarse-grained (sandy or gravel) soils with low water-holding capacity allow deep infiltration profiles that encourage deep roots (Schenk and Jackson, 2005; Fan et al., 2017; Zhou et al., 2020). The coarse particles present in the soil also increase macropores and reduce mechanical resistance (Xu et al., 2012; Gargiulo et al., 2015, 2016), which is conducive to the penetration of the root system into the deep layer (Clark et al., 2002; Bengough, 2003). However, whether the rooting depth increases when the content of coarse soil particles increases and water and nutrient levels decrease remains to be explored.
The vertical profile of fine roots is also regulated by genotype, presenting as an inherent characteristic of species with wide interspecific variation (Peek et al., 2005; Gao et al., 2020; Li et al., 2020; Zhou et al., 2020; Luo et al., 2021). Simultaneously, stand age determines the spatial distribution of roots (Peek et al., 2005; Zhang et al., 2018). Previous studies have focused on the seasonal dynamics of fine root distribution in the short term, mainly explaining the rooting strategies of plants in response to the dry-wet season (Cheng et al., 2002; O’Grady et al., 2005; Wang et al., 2016). These observations have not yet provided a deep understanding of the dynamic patterns in fine root distribution and resource acquisition of woody plants across different ages and their time cumulative responses to soil heterogeneity.
In the present study, we observed the fine root vertical profiles of four xerophytic species along an RFC gradient (0%, 25%, 50%, and 75%, v v-1) and during three years of growth (the second, third, and fourth years) in the arid valley environment of western China (Hu et al., 2021). These four species are important native species in the arid valleys of the Hengduan Mountain region, with irreplaceable ecological value in vegetation restoration and soil and water conservation (Li et al., 2008, 2009; Wu et al., 2008; Bao et al., 2012). We examined fine root biomass and morphology across the soil layers along the RFC gradient for three years of growth. Our objectives were to investigate: 1) how the vertical pattern of fine root depth and traits varied with the RFC gradient, 2) the temporal pattern of fine root depth and traits along soil profiles and RFC, and 3) interspecific differences in the dynamic response of fine root depth and vertical distribution to the RFC gradient. We hypothesized that: 1) root distribution deepened with the increase in RFC, based on the characteristics of poor water-holding capacity and low strength of coarse soils (Bengough, 2003; Schenk and Jackson, 2002, 2005; Fan et al., 2017); 2) with the increase in age, the response of root distribution to changes in RFC showed a cumulative decreasing effect of time, because younger plants (seedlings) are more sensitive to soil heterogeneity than older ones (Padilla and Pugnaire, 2007).