The correlations between the soil and SMB elemental ratios under the three land-use types were much more informative than the results from the three landscapes. As indicated inTable 4, the strongest correlations (all significant) for the three land-use types were observed between the C:P ratios in the soil and SMB, while the weakest correlations were found between the soil and SMB N:P ratios. All of the correlations between the soil and SMB C:P ratios were significant for the paddy field (r = 0.52,p < 0.001), upland (r = 0.14, p = 0.020), and woodland (r = 0.44, p < 0.001). Additionally, the correlations between the soil and SMB C:N ratios were all significant for the paddy field (r = 0.20, p < 0.001), upland (r = −0.13, p = 0.036) and woodland (r = −0.40, p < 0.001). Regarding the correlations between the soil and SMB N:P ratios, a significant relationship was only found for the woodland (r = 0.59, p < 0.001). There were no significant correlations found for the paddy field (r = −0.08, p = 0.052) or upland (r = −0.08, p= 0.187). Woodland is a special and widely distributed land-use type in southern subtropical China and one where there are generally no external anthropogenic inputs of applied N and P fertilizers or other types of land management practices performed. The significant positive correlation between the atomic C:P ratios in the soil and SMB in the woodland is not surprising because this correlation existed in the paddy field and upland as well. However, the correlation between the N:P ratios in the soil and SMB (r = 0.59,p < 0.001) in the woodland is of particular interest because it is the only significant association detected between the soil and SMB N:P ratios in this study. The two significant positive correlations noted above might indicate that the soil microbes in the woodland in southern subtropical China exhibited weak homeostasis and that their chemical C, N, P composition is dependent on the resources in the soil environment. However, the negative significant correlation between the soil and SMB C:N ratios in the woodland is clear (r = −0.40, n = 214, p < 0.001), but difficult to explain. This finding could indicate the existence of C limitation in the woodland soil. C limitation may occur when woodland ecosystems, dominated by coniferous masson pines in southern China, produce very low quality litter and thus result in a very small amount of litter C contributing to the soil humus C [Yan et al., 2006], which may well be theoretically governed by the SMB ability of adapting to poor nutrient conditions (such as very low P, see Data Set S1 in the auxiliary material) by greatly reducing their C use efficiency [Manzoni et al., 2010]. Additionally, N deposition in the region has been substantially increased either by air pollution or by increased anthropogenic reactive N emissions in recent years [Liu et al., 2011]. However, the actual mechanism underlying this significant negative correlation between the soil and SMB C:N ratios in the woodland remains unclear; to clarify this mechanism, further investigations and more specific experiments are required. Although large amounts of N and P fertilizers were applied to the fields to ensure high yields for crop production, the SMB in the paddy field and upland clearly showed N and P dependence on the soil environment. However, the nutrient removal by grain harvest and a number of process losses (e.g., gas emissions, leaching, soil erosion, etc) from the cropping systems in southern China were substantial, particularly with respect to N [Zhu and Chen, 2002; Zhang et al., 2003]. In the study region, there is usually one crop season each year in the upland, involving crops such as sweet potato or maize, while the paddy field is always double rice cropped. Thus, compared to the paddy field, the upland receives approximately half the amount of N, the same or a greater amount of P fertilizer and very little return of crop residues (Table 1). Consequently, based on the correlations between the soil and SMB C:N and C:P ratios found for these two land-use types, the SMB in the upland exhibited weaker N and P dependence on the soil environment compared to the SMB in the paddy field (Table 4).
 Individually, significant correlations between the soil and SMB C:N:P ratios in the paddy field were only found in Dacai (karst mountain) and Pantang (low hill) and not in Yuanjiang (lowland) (Table 4), where the agricultural activity was much more intensive (Table 1). Because there was a low nutrient demand in the upland, the SMB associated with this land-use type only exhibited N and P resource dependence in Dacai; additionally, SMB only exhibited P resource dependence on the soil environment in Pantang. For the woodland, which is a special category of land-use in the region, significant correlations between the soil and SMB C:P (r = 0.46,p < 0.001) and N:P (r = 0.47, p < 0.001) ratios were observed only in Pantang and not in Dacai. This result could be explained by the differences in nutrient availability and plant nutrient demand between the woodlands in these two landscapes. On average, the SOC (1.894 mol C kg−1), TSN (0.161 mol N kg−1) and TSP (0.0198 mol P kg−1) contents in the woodland in Dacai were almost twice the contents (0.755 mol C kg−1, 0.082 mol P kg−1 and 0.0148 mol P kg−1, respectively) in Pantang (see Data Set S1 in the auxiliary material). The Dacai woodland was covered with lower-biomass shrubs, ferns and masson pines, and consequently it had less nutrient demand. In contrast, the Pantang woodland was associated with higher-biomass masson pines and greater nutrient demand.
 The correlations of the C:N:P ratios between the soil and the SMB varied differently in each landscape. For instance, in Dacai (karst mountain), a descending trend of the correlations between the soil and SMB C:N and C:P ratios was observed in the following order: paddy field > upland > woodland. This trend paralleled the decreasing intensification of anthropogenic activities. However, in Pantang, this trend was only observed in the paddy field and upland, but not in the woodland. There was also no clear trend of the correlations of the C:N:P ratios between the soil and the SMB observed among the land-use types in Yuanjiang.
 In summary, the relationships between the soil and SMB C:N:P ratios under three different land-use types found in the present study were statistically significant, suggesting a possible weak homeostasis in the SMB in the region. This finding might ultimately be related to the balance of nutrients in the soils, which was attributable to litterfall patterns or land management practices (e.g., crop rotations, N and P fertilizer use) applied to these land-use types.