4.3 Effect of sampling depth on the SOC and TN storage
Our meta-analysis showed that the deforestation could lead to SOC and TN losses, whereas land restoration had reverse trend gains (Fig. 2). However, the changes of SOC and TN storages varied with sampling depths (Table 3, Fig. 4, and Fig. 5). With the increasing sampling depth, the response of SOC and TN to LUCC gradually decreased. From shallow sampling depths (< 30 cm) to deep sampling depths (> 30 cm), the corresponding SOC and TN storages under deforestation ranged from - 0.77 to - 0.36, from - 0.57 to - 0.46; the corresponding SOC and TN under land restoration ranged from 0.40 to 0.1, from 0.44 to 0.11, respectively.
The effect of LUCC on the SOC and TN storages in the shallow depth (< 30 cm) were more sensitive than that in the deep sampling (> 30 cm). The results are supported by others (Olson and Al Kaisi, 2015). Till now, shallow sampling (i.e., up to 20 or 30 cm) procedure is still popular in China (Liao et al., 2014). This would undoubtedly produce uncertainties and biases even lead to incomplete conclusions. To accurately assess the impact of LUCC on SOC and TN storage, we should choose deeper sampling method. As such, Lal (2009) recommended that sampling depth should be at least 1 m or 2 m to accurately assess SOC due to land-use changes. Our study highlights that more attention should be paid to the uncertainty of sampling depth in future research. Otherwise, the response of SOC and TN to LUCC would be biased if soil samples were collected only at shallow depth.