Assessment of long-term impacts of organic and inorganic fertilizers on soil extractable phosphorous and carbon in Machang'a, Mbeere disctrict, Kenya

Abstract

Decline in crop yields is a major problem facing smallholder farmers in semi-arid Kenya and the entire Sub-Saharan region. Soil organic matter is rapidly declining and phosphorus (P) and nitrogen (N) deficiencies are important constraint to food production in semi-arid areas. Inorganic fertilisers are unaffordable to most smallholder farmers at current producer prices, limiting their use on farms. While N can be replenished through biological fixation, there is no equivalent process for the introduction of P in soils. Consequently, phosphorus deficiency has become a major constraint to food production. To try to solve this problem, two long-term field experiments were established at Machang'a site to assess long-term effects of manure and inorganic P fertilisers on soil extractable P and organic carbon (OC). The first experiment began in 1989 and treatments were; control, continuous manure at 5 and 10 t/ha/yr, residual manure at the same rates and NP fertiliser at 51 and 12 kg/ha/yr respectively. This trial had 14 years of annual manure application. The second experiment began in 1994 and was planned to supplement the first experiment by giving information on the effects of a one time application of 250 kg/ha TSP on longterm P availability. Diffuse reflectance spectroscopy was used for rapid assessment of soil quality. While common wet laboratory methods for soil nutrient analysis are laborious, time consuming and costly, diffuse reflectance spectroscopy method was explored as an alternative analytical method. Since the current study was part of an on-going project, data accumulated from the project was used for purposes of current study. Soil sampling began in 1993 and samples analysed for OC and extractable phosphorus. Sub-samples of these soils were scanned using reflectance spectroscopy. Samples collected in year 2002 were analysed for sequential P fractions using a modified Hedley's method. Laboratory data was analysed for variation by Genstat (1995), while the near infrared (NIR) spectral data was analysed using multivariate exploratory statistics. Partial least squares regression, principle component analysis and discriminant analysis were used to develop predictive calibration models. Results indicated that repeated manure application at lOt/ha/yr significantly increased the level of plant available P and maintained high OC. Residual manure did not have profound effects on P and OC compared with continuous manure. Though inorganic fertilisers raised the level of bioavialable P fractions, they had negligible influence on OC. The residual value of a one time application of 250 kg/ha TSP was observed to maintain high soil P-test values for close to 10 years. Robust near infrared calibration models developed, well predicted the level of OC by the various treatments. However, poor calibrations were obtained for soil extractable P.1t is not expected that NIR spectroscopy will replace the nutrients analysis used in convention soil fertility C assessment. However, infrared methods have the potential to rapidly and nondestructively predict soil quality changes caused by various management regimes, thus adding considerable value to soil analysis. The method is also economically attractive due to the low costs associated with it.