Subsistence farmers in Africa depend largely on the soil organic matter to sustain crop productivity. Long-term changes in soil organic carbon and nitrogen were measured after woodland clearance for smallholder subsistence farming or for commercial farming. The contents of organic carbon and nitrogen in soil under reference woodlands were largest (53.3 t C ha À1 , 4.88 t N ha À1 ) in a red clay soil ($ 50% clay þ silt), followed by a granitic sand ($ 12% clay þ silt; 22.8 t C ha À1 , 1.47 t N ha À1 ) and least (19.5 t C ha À1 , 0.88 t N ha À1 ) in a Kalahari sand ($ 5% clay þ silt). Organic carbon declined rapidly under cultivation to attain new equilibria within 10 years on all smallholdings. Greatest losses occurred in soils that initially contained most carbon and nitrogen in the order: red clay (22.4 t C ha À1 and 1.0 t N ha À1 ) > granitic sand (13.2 t C ha À1 and 0.8 t N ha À1 ) > Kalahari sand (10.6 t C ha À1 and 0.5 t N ha À1 ). On the clay soil, commercial farming with intensive use of mineral fertilizers and incorporation of maize stover led to more gradual decline: at equilibrium, contents of carbon and nitrogen were 15 t C ha À1 and 1.7 t N ha À1 greater than on smallholdings with similar soil and climate.In the Kalahari sand the 13 C of organic C remained constant after woodland clearance, and maize contributed less than 10% of the total C even after 55 years. The 13 C signature increased slightly with increasing duration of cultivation by smallholders in the granitic sands and red clay soil where maize contributed 29% and 35% of the C at equilibrium. Under more productive commercial farming, the carbon derived from maize accounted for 50% of the total after 10 years of cultivation and 67% at equilibrium. The persistence of woodland carbon in the sandy soil is attributed to chemical stabilization resulting from large concentrations of lignin and polyphenols in the tree litter, or as charcoal.

Direct determination of soil hydraulic properties is often costly and laborious hence the use of indirect methods such as pedotransfer functions (PTFs). Despite progress made in PTF development in general, little evaluation of PTFs has been done for the sandy soils of Niger. We tested the ability of three PTFs, (Campbell, van Genuchten, and Vauclin) to determine soil water retention and unsaturated hydraulic conductivity (K) for sandy soils at two villages (Banizoumbou and Bagoua) in Niger. Modeled K was compared to K estimated from neutron probe readings at 1.4 m; and modeled moisture retention was compared to lab measurements derived from the hanging water column method and pressure plate apparatus for the following depth intervals: 0-30, 30-60, and N 60 cm at Banizoumbou; and 0-30, 30-120, and N120 cm at Bagoua. The Campbell PTF resulted in lower root mean square error (RMSE) (0.05-0.06 m 3 m − 3 ) for soil moisture retention for the three depth intervals at the two sites and performed better than the van Genuchten function (RMSE 0.06-0.07 m 3 m − 3 ) for Bagoua soils which had higher sand content. The van Genuchten PTF consistently overestimated dry regime moisture retention for the three depth intervals especially at Bagoua but performed well for the wet regime. The Campbell and Vauclin PTFs underestimated K (RMSE 0.61-1.01 mm d − 1 ) at both sites whereas the van Genuchten model was close to field measurements (RMSE 0.26-0.47 mm d − 1 ). These results show that the Campbell model is a cheaper alternative to direct measurement of moisture retention and the van Genuchten function can be used to estimate K for Niger's sandy soils with modest accuracy.