3.3 Soil phosphorus availability
The available P concentration and P activation coefficient first decreased and then slightly increased in the LDM soil but gradually decreased in other soils as the soil depth increased (Figure 5a–b). Compared with RPM soil, the available P concentration was 17.0%–234.0% and 0.0%–85.1% higher in all soil layers from the LDM and MDM, respectively, with a significant difference at the 10–30 cm and 60–100 cm layers in the LDM and the 0–20 cm layers in the MDM (p < 0.05) (Figure 5a); it decreased by 4.0%–30.7% at the 0–40 cm depth of HDM but hardly changed at the 60–100 cm depth (Figure 5a). The variations in P activation coefficient between the RPM and degraded marsh soils exhibited a trend similar to that of available P (Figure 5a–b). Furthermore, the P activation coefficient in the surface (0–10 cm) soil was lower than the threshold of soil P deficiency of 2.0% (Li and Zhang, 1994).
3.4 Soil available phosphorus related to phosphorus forms
The conceptual structural equation model in Figure 2 was successfully fitted to our data, and the goodness-of-fit metrics were satisfied (χ2/df was approximately 1, NFI > 0.95, CFI > 0.99, and RMSEA < 0.08; Figure 6a). All P forms explained 79% of the variations in soil available P. The non-occluded Ca8-P and Al-P exhibited a strong effect on the variations in soil available P, whereas the effect of non-occluded Fe-P was so minor that it did not exhibit in the fitting structural equation mode (Figure 6a).
Soil organic P had a strong total effect (0.784) on the variations in available P and was the most important positive factor influencing the variations in available P, followed by non-occluded P (0.578), occluded P (0.447), Ca2-P (0.409) and Ca10-P (-0.093) (Figure 6b). This indicated that the variations of available P were mainly related to organic and non-occluded P in alpine wetland soils. The direct effect of soil non-occluded and organic P on available P was slightly less than that of Ca2-P (Figure 6b), suggesting that soil non-occluded and organic P also might be non-negligible direct source of available P. Marsh degradation exhibited a high direct effect (0.455) on the variations in Ca10-P. Meanwhile, Ca10-P presented a high indirect effect (-0.402) on the variations in available P through the effect of organic and non-occluded P in soils (Figure 6a–b). These revealed that marsh degradation significantly influenced soil P availability via the transformation from Ca10-P to organic and non-occluded P.