Table 1. The coefficients of the soil texture (Sandy vs others), animal (Cattle vs. sheep), grazing period (Period), precipitation (Prcp), temperature (Temp), and sampling year (Year) in the 21 mixed models for the 0–10 and 10–30 cm depths.
Effect sizes | Sandy vs Othersa | Cattle vs Sheep | Period (1/100) | Prcp (1/10000) | Temp (1/100) | Year (1/100) |
---|---|---|---|---|---|---|
0–10 cm depth | ||||||
BD | -0.019* | -0.004 | 0.043 | 0.160 | -0.069 | 0.044 |
SOC | -14.45 | 65.83* | -125.4* | 52.10 | -4.694 | -18.20 |
TN | -0.015 | 5.567* | -6.699 | 76.98* | -13.57 | 44.14 |
C:N | -0.183 | 0.237 | -0.385 | 0.390 | -1.105 | -3.544 |
pH | -0.002 | -0.013 | 0.030 | 0.033 | -0.020 | 0.054 |
P | 0.037 | 0.111 | -0.439 | 3.990* | -2.614* | -0.759 |
WC | 0.021 | 0.069 | -0.472* | 0.150 | 0.296 | -0.553 |
NH4+ | -0.138 | 0.236 | 0.184 | 2.360 | 2.963* | -0.553 |
NO3- | -0.141* | -0.106 | 0.158 | 1.190 | 0.312 | 0.554 |
K | 2.156* | 2.266* | -0.435 | 22.94 | -2.698 | 6.288 |
PR | 0.031 | -0.781* | 0.982 | 6.520* | -4.222* | -4.955 |
EC | -3.177 | 1.855 | -10.60 | 4.140 | 31.00* | 37.53 |
CEC | 0.065 | - | -0.450 | 2.170 | 2.126* | -1.134 |
MBC | 0.046 | 0.026 | 0.272 | -1.900 | 1.531 | 1.251 |
MBN | 0.346 | -0.202 | 1.705 | 1.840 | 0.534 | 0.047 |
10–30 cm depth | ||||||
BD | -0.011 | 0.031* | -0.080* | -0.200 | 0.050 | 0.069 |
SOC | -0.114 | 1.454 | -6.916 | 116.5* | -9.453 | 2.203 |
TN | 1.884 | 1.759 | 17.56 | 75.97 | 40.74 | 1.892 |
C:N | 0.087 | -0.241 | 0.163 | 2.120 | -3.030 | -3.429 |
pH | -0.002 | -0.047 | 0.035 | 0.034 | -0.114 | -0.297 |
P | 0.753* | 0.106 | 2.428* | -0.100 | -0.881 | -5.504 |
aOthers include silty and clayey soils.
*indicates that the independent variables as covariates significantly impacted the effect size of soil property in the mixed model (P<0.10). However, the coefficients of the independent variables cannot be used to interpret how many percentages they can increase or reduce the effect sizes of soil properties because the dependent variables (i.e., effect sizes of soil properties) were transformed using different algorithms based on the Box-Cox method when building these models.