2.3. Meta analysis The impact of NT in controlling runoff coefficient, sediment concentration and soil losses was estimated by comparing the magnitude of the NT treatment mean X NT to the CT (control) treatment mean X CT . A response ratio (Re) was calculated as the treatment effect size statistic using Eq. (3) . Re ¼ XNT X CT ! ð 3 Þ Values of runoff above 1 indicated that NT treatment had greater mean runoff coefficient, sediment concentration or soil losses than CT. The response ratios were normalised by natural log transformation as shown in Eq. (4) . lnRe ¼ ln X NT X CT ! ¼ ln X NT ln X CT ð 4 Þ MetaWin 2.1 software was used to compute mean effect sizes and generate 95% bootstrapped con fi dence intervals (CIs) using 4999 iterations ( Rosenberg et al., 2000 ). Treatment effect sizes, i.e. NT vs CT, for each category were considered statistically significant if the 95% CIs did not overlap zero ( Borenstein et al., 2009 )