Quantification of visual soil erosion indicators in Gikuuri catchment in the central highlands of Kenya

Abstract

Quantification of soil erosion using conventional approaches is hampered by lack of extensive spatial coverage and long duration data. Therefore use of these approaches for land management advisory has tended to result in unsatisfactory landuse plans that are in great disparity to on-site observations by farmers. Farmers have great knowledge of what they perceive as the indicators of soil erosion, which have so far not been empirically linked to actual soil loss or crop yield rates. This study was conducted to attach quantitative values of soil loss and maize crop yields to on-site erosion and sedimentation indicators as perceived by the farmers in the central highlands of Kenya. Soils exhibiting splash pedestals, sheetwash, rills, sedimentation, red colour and stoniness were the erosion indicators selected for quantification. Three soil types and three slope gradients were identified and on each runoff plots were installed to relate the sheet-rill erosion developments to actual soil loss. Whereas the temporal changes of the rills and pedestal height (sheet erosion) were used to quantify erosion rates within nine bounded runoff plots, five erosion indicators were identified within 24–31 farmers' fields with an aim of estimating crop yield gaps. Statistical procedures applied included correlation matrix, linear regressions and analysis of variance using Duncan' multiple range tests within the SPSS program. The study observed that the temporal and spatial dynamics of soil surface level and various dimensions of the rills were influenced by slope length and prolonged rainy days. Soil loss correlated significantly with individual rill dimensions except soil surface level. But when soil surface level, rill depth, width and total length were combined, they were all found to be significant variables influencing the actual soil loss. Two models were constructed relating soil loss rates with both the rill sizes and decline in soil surface levels. Also five widespread erosion indicators were empirically linked to specific crop yield levels. Because of soil erosion a crop yield gap of more than 50% was observed in fields bearing superficial stoniness and sedimentation indicators. On basis of these results rate of soil loss can now be estimated at field scale by fieldworkers in situations where sheet-rill erosion is prone within a rainfall event or season. This would assist in formulating satisfactory and timely advice for the farmers on the effectiveness of soil and water conservation (SWC) measures and also in assessing the rate of soil erosion instead of relying on ill-fitting conventional erosion models. The approach could enable successful planning and implementation of SWC measures whereby priority is given to areas identified with severe erosion indicators. Besides knowing soil loss rates, crop yield decline experienced by farmers could be a useful soil productivity indicator for the on-going soil erosion. Conversely, where progressive evolution of soil erosion indicators corresponded to decline in crop yields then specific indicators could be used to predict crop yields farmers were likely to obtain in a given rainfall season, other land management practices remaining constant.