Topographic distribution of the soil total carbon content and sulfur deficiency for rice cultivation in a floodplain ecosystem of the Northern region of Ghana

An understanding of the spatial variation in soil fertility is fundamental for extending rice cultivation and developing appropriate management practices in currently under-utilized floodplain ecosystem of West Africa. The spatial patterns in the soil total carbon (TC) contents were analyzed using a Quickbird imagery, a global digital elevation model (ASTER-GDEM) and ArcView GIS functions for a White Volta floodplain of northern Ghana. An irrigated pot experiment was then conducted to relate rice biomass production to the spatial TC distribution and to identify deficient nutrients using different fertilizer treatments.The TC contents showed great variability (2.0–40.2gkg−1) in a lognormal distribution, and the values were logarithmically increased with the proximity to reservoirs, consisting of a main river, back-swamps and ponds. The spatial pattern in the TC contents corresponded to the length of the waterlogging period during the rainy season, which was also logarithmically increased with the proximity to reservoirs, as estimated from the time-series satellite images. The corresponding patterns indicated that the spatio-temporal hydrological variable was a principal factor for the variance in the soil TC contents. Regarding the pot experiment, the amounts of soil mineralizable nitrogen were in accordance with the TC contents and correlated linearly with the plant N uptakes. However, dry matter production was scarcely relevant to the N-supplying capacity of the soils. Limited S concentrations in the plant tissues and a significant response to S application indicated that an inherent S deficiency restricted the dry matter production. NPK application without S increased only the N concentration and N:S concentration ratio in the plant tissues but not the biomass production.The high N-supplying capacity of the soil was related to the seasonal flooding, and the productive land areas for rice cultivation in both indigenous N and water availability were quantitatively predicted as a logarithmic function of the distance to reservoirs. Appropriate S supplementation would be required to utilize effectively the local N supply for rice biomass production in this under-utilized floodplain ecosystem.