Table 2. Factors associated with changes in percentage tree cover (means with 95% confidence intervals in square brackets) within the Hukaung Valley landscape during two time periods.

Factor (transformation)	Change in percentage tree cover 2000–2005	Change in percentage tree cover 2005–2010
Altitude (metres, log)	0.475 [0.400, 0.550] (16.5%)	1.625 [1.560, 1.689] (84.5%)
Slope (degrees, square root)	−0.018 [−0.043, 0.008] (6.0%)	−0.09 [−0.118, −0.063] (9.5%)
Distance to drainage (rivers and streams, metres, square root)	0.005 [0.004, 0.007] (5.5%)	0.004 [0.002, 0.006] (4.5%)
Distance to roads (metres, square root)	−0.008 [−0.009, −0.008] (25.0%)	−0.001 [−0.002, −0.000] (0.5%)
Distance to communities (metres, square root)	0.008 [0.007, 0.008] (10.5%)	0.009 [0.008, 0.010] (11.5%)
Inside concession (binary: yes/no)	−0.441 [−0.634, −0.248] (2.5%)	−6.489 [−6.220, −6.758] (97.0%)
Distance to gold mine (metres, square root)	0.019 [0.018, 0.020] (90.5%)	0.007 [0.006, 0.007] (1.5%)

Distances are measured in metres. Generalized least squares with constant power of covariance structure for altitude and a rational covariance function to correct for spatial correlation, over 200 repeated runs. Overall model p values were under 0.05 for 80% of models for 2000–2005, and 100% models for 2005–2010. Figures in square brackets show 95% CI of the means for 200 runs. P-values are summarised for each factor in round brackets as the percentage of values across 200 runs less than 0.05.