Diagnosis of soil and plant nutrient constraints in small-scale ground nut (Arachis Hyopaea) production systems of Western Kenya using infrared spectroscopy
Soil fertility degradation is a major problem in Africa leading to food insecurity, ecosystem degradation and poverty. Nutrient depletion and disease epidemics have contributed to a decline in groundnut yields of 25% in the past decade in Sub-Saharan Africa. Studies have demonstrated that infrared spectroscopy (IR) may permit rapid and cost effective analysis of tropical soil nutrients. Trial and error and field observations methods used by farmers are inefficient and have led to inappropriate soil nutrient management strategies and options in small-scale crop production systems. This study sought to survey the prevalence of soil nutrient constraints in the smallscale groundnut production systems of western Kenya and to explore the potential of IR as a diagnostic tool for soil nutrient constraints. The soil properties examined were soil pHw, total carbon (TC), total nitrogen (TN), extractable phosphorus (Ext. P), exchangeable potassium (K), exchangeable calcium (Ca), exchangeable magnesium (Mg) and particle size distribution (PSA) while for plant macronutrients, nitrogen (N), phosphorus (P) potassium (K) were examined and micronutrients, copper (Cu), zinc (Zn) iron (Fe) and manganese (Mn). Reference data that were used for developing calibration models were analyzed using standard laboratory methods widely used for tropical soils and plants. Soil pH was determined using an electrode pH meter for saturated soil paste. Exchangeable Ca and Mg were determined by the use of the atomic absorption spectrometry (AAS). The Olsen method was used to colorimetrically determine Ext.P. Total carbon (TC) and (TN) were determined using the C: N analyzer and particle size analysis (PSA) was determined using the hydrometer method. The plant macronutrients N, P, K, were determined by Kjeldahl distillation method, while the micro nutrients Cu, Zn, Mn and Fe were analyzed by ashing prior to determination by AAS. The reference data were then calibrated to soil and plant reflectance. Stable calibration models were developed for several key soil nutrients using the near-infrared (NIR) and mid-infrared (MIR) diffuse reflectance spectroscopy and partial least square regression (PLSR) statistical analysis. Robust calibration model were obtained; soil pHw (r2=0.85), TC (r2=0.98), TN (;=0.97), Ca (;=0.95) and Mg (r2=0.94), sand r2=0.85) silt ;=0.82 and clay r2 =0.81 for the MIR spectral region. Extractable P and Exch. K had weak calibration models with r2 values of 0.66 and 0.50 for MIR respectively, 0.50 and 0.32 for NIR respectively. Similar results were obtained for above-ground groundnut biomass although P and K yielded good calibration models. Attenuated total reflectance (ATR) yielded robust calibration for TN from saturated soil pastes with r2 values of 0.94. The study demonstrated the utility and potential of IR spectroscopy as a diagnostic screening tool for soil and plant nutrition in small-scale production systems. Principal component analysis (PCA) was used to summarize the variability in soil properties. Soil fertility indicators (SFIs) that were developed from the principal components were then used to evaluate soil nutrient levels based on critical nutrient levels. The SFIs were successfully calibrated to soil reflectance measured in the laboratory with cross validated; values of 0.97 and 0.87 for MIR and NIR, respectively. Groundnut farms were critically deficient in principal nutrients; TN (75%), Ext.P (65%), and Exch. Ca (100%) which fell below the critical nutrient levels based on soil and plant MIR spectral predictions. Replenishment of these principle nutrients is crucial for sustainable groundnut productivity in western Kenya.