3.2 The relative importance of resource availability and heterogeneity in plant abundance
The total variation in community abundance explained by both resource availability and heterogeneity was 67% for shrubland and 64% and 63% for shrub layer and tree layer in woodland (Fig. 4). The shared fraction between resource availability and heterogeneity explained a large proportion of the variation in community abundance (48.4%, P =0.002 in shrubland; 33.7%,P =0.004 and 30.7%, P =0.004 in the respective tree layer and shrub layer in woodland). Abundance was also largely explained by the unique contribution of resource availability (16.2%, P =0.003 in shrubland; 19.5%,P <0.001 and 31.8%, P <0.001 in the respective tree layer and shrub layer in woodland), but the unique contribution of resource heterogeneity was marginal (2.8%,P =0.006 in shrubland; 0.9%, P <0.001 and 0.2%,P <0.001 in the respective tree layer and shrub layer in woodland) (Fig. 4).
3.3 Multiple direct and indirect effects of resource availability on species diversity and abundance
To further disentangle the complex relationship between resource availability and plant community characteristics, SEMs were carried out for each vegetation type. Our SEMs (Fig. 5a, b, c, d, e, f) produced good fits with our data for community abundance and species diversity in shrubland and woodland (Table 3). Therefore, these models successfully assessed the extent to which factors influenced species diversity and abundance through direct and indirect pathways.
In shrubland, the standardized effect of environmental factors derived from SEMs showed that soil TN had positive direct effects on both shrub abundance (standardized coefficient, λ=0.67) and species diversity (λ=0.77). Soil moisture had positive indirect effects on both shrub abundance (λ=0.67) and species diversity (λ=0.72) via soil TN. Therefore, the soil TN and soil moisture are the strongest predictor of shrub abundance and species diversity. Additionally, soil depth had a significantly positive effect on species diversity by direct and indirect effects (Fig. 5a, b, Fig. 6a, b, Table A2).
For the shrub layer in woodland, soil depth had positive indirect effect (λ=0.37), and soil pH had negative indirect effects (λ=-0.24) on plant community abundance through TN and TK, Additionally, soil nutrient factors all had negative direct effects (TN: λ=-0.48; TK: λ=-0.36) on abundance (Fig. 5c, Fig. 6c, Table A3). In contrast, soil depth had negative indirect effect (λ=-0.18), and soil pH had positive indirect effects (λ=0.21) on species diversity. Species diversity was mainly positively explained by soil TK (λ=0.41) (Fig. 5d, Fig. 6d, Table A3).
For the tree layer in woodland, plant community abundance was mainly influenced by soil ρb (λ=0.73), followed by soil pH (λ=-0.61), Mg (λ=-0.51) and depth (λ=0.41), among which the negative effect of soil Mg on abundance was through indirectly effecting soil pH. The direct effect of soil depth on abundance was non-significant, but soil depth had positive indirect effect on abundance via ρb (Fig. 5e, Fig. 6e, Table A3). Overall species diversity was primarily driven by soil TN (λ=-0.80), depth (λ=0.61), TK (λ=-0.49) and pH (λ=-0.42). Soil depth and soil pH indirectly influenced species diversity via soil nutrient factors, and soil TN and TK had direct negative effect on species diversity (Fig. 5f, Fig. 6f, Table A3).
Discussion