|
Author:
|
Ngoze, Solomon; Riha, Susan; Mbugua, David; Shepherd, Keith; Verchot, Lou; Barrett, Christopher B.; Lehmann, Johannes; Wangila, Justine; Pell, Alice |
|
Abstract:
|
"Raising agricultural productivity in smallholder agriculture in sub-Saharan Africa requires an understanding of if and how farm household land use and socioeconomic factors affect soil fertility. Market access, population growth, socio economic characteristics and agro ecological zones have been proposed as important drivers of land use intensity and, consequently, soil fertility. We used diffuse reflectance infrared spectroscopy to measure soil fertility, and multivariate and exogenous switching regression statistical approaches to determine if soil fertility in the smallholder farms of the highlands of Kenya is associated by region, land use categories (cash crop, food crop, fodder and pastures), and selected household socio economic factors (household income, number of adults, farm size and number of cattle). Over 2000 fields on 236 farms were sampled in Embu (eastern Kenya highlands, primarily Andosols) and Madzuu (western Kenya highlands, primarily Ferrasols). Soil fertility variables, including total soil carbon (SC), total nitrogen (TN), pH, available Olsen phosphorous (P), extractable potassium(K), calcium (Ca), and magnesium (Mg), effective cation exchange capacity (ECEC) and texture, were measured using conventional laboratory techniques on 15% of the sampled soils. From these analyses, SC, TN, P and K were all greater in Embu compared to Madzuu soils. Soil fertility variables were significantly higher in pastures compared to other land uses in Madzuu, but were comparable with other land uses in Embu. This soil data was then used to calibrate soil reflectance results in order to predict soil fertility variables for all soil samples. Principle component analysis (PCA) of soil fertility variables developed from the spectroscopy data for each soil sample indicated similarities among sites in the three most important eigenvectors: the first (soil nutrient) vector had high positive loadings for K, Ca, Mg, ECEC and pH; the second (soil organic matter, SOM) vector had high positive loadings for soil organic carbon and total nitrogen; and the third (soil texture) vector had high positive loadings for clay plus silt. However, in Embu, P was associated with the soil organic matter vector while in Madzuu it was associated with the soil nutrient vector. In comparison to pasture all other land uses were associated with lower values of soil nutrient and SOM components in Madzuu, while in Embu, these other land uses were associated with higher values of the SOM component. Number of cattle per farm had no association with any of the three soil fertility components at either site. In Embu, farm income and adult population were both positively related to SOM. In Madzuu, farm size was positively associated with SOM but negatively associated with soil nutrients. More than twice as much P fertilizer is applied on average in Embu compared to Madzuu (27 vs. 11 kg ha-1 season-1). Our study supports the link between poverty dynamics and soil degradation in smallholder agriculture; wealthier households in the eastern Kenya highlands are able to invest in soil fertility management while the poorer households in western Kenya are mining nutrients in soils."
|