The high heterogeneity and diverse management effects on trees in agricultural landscapes limit the use of standard allometric equations developed for forests; hence the need to develop robust allometric equations and root:shoot ratios (RS) for more accurate estimation of belowground biomass. Local generic equations were developed from 72 trees (diameter at breast height (dbh): 3-102 cm) destructively sampled across three 100 km 2 benchmark sites in Western Kenya. Belowground biomass of the harvested trees accounted for about 20% of the total tree biomass; yielding an overall RS of 0.26, which varied across an altitudinal and precipitation gradient. The equation based on dbh alone estimated belowground biomass carbon in agricultural mosaics of Western Kenya to be 50.01 Mg C ha -1 with about 90% accuracy. Two of the equations recommended for tropical species produced estimates that were about 35 and 21% lower. The equation with dbh as predictor is therefore reliable for estimating belowground biomass in agricultural landscapes while RS should be used with great care depending on soil and management conditions as shown by the great variability among the blocks evaluated. Equations presented in this study will significantly improve the accuracy of root biomass estimates in agricultural landscape mosaics without the high investments in excavating the root system.
