Fig. 1.

Trends in measures of total soil N (to a depth of 0.5 m) in a long-term continuous rice-rice system experiment at the IRRI farm, Philippines comparing a nil-N treatment with the inputs of synthetic N (137 kg N ha^-1 per year), and the inclusion of sources of BNF via azolla or sesbania (adapted from Ladha et al., 2000). Different letters shown between treatments indicate statistical significance (α =0.05) according to Duncan’s multiple range test. * and ** indicate significance at the 0.05 and 0.01 levels, respectively