Abstract: Variation in soil fertility and crop yield in farmers` fields is a factor responsible for the low farm productivity and adoption of agronomic recommendations. A study was conducted to characterize the soil fertility management zones using participatory rural appraisal, conventional survey methods and maize yield. Farmers identified the soil types using colour, texture and productivity. The red soils (Rhodic ferralsols) were rated to be less fertile than darker Humic acrisols and Mollic gleysols. Farmers also identified the soil fertility management niches in terms of topography, physical discontinuities, management and classified them as productive or unproductive. The productive niches occupied between 0.25 to 0.30 ha and were used for maize, bananas and vegetables production. Non-productive niches were between 1.5 to 6.0 ha and were either left fallow or used for maize and sweet potato production. Productive niches had a pH of 5.3% C of 2.3 and silt fraction of 232 g kg-1 and maize yield of 4.3 t ha-1. Non productive niches had a pH of 3.99, % C of 1.9 and a silt fraction of 193 g kg-1 and maize yield of 2.8 t ha-1. Management should target processes that enhance these variables in addition to incorporating the farmers` local knowledge.
INTRODUCTION
Most reports dealing with soil fertility in sub-Saharan Africa indicate an alarming negative nutrient balance[1,2]. While the bleak picture is true, marked differences in soil fertility attributed to inherent soil properties, topographic position and farmer induced management occur pointing to the need for site-specific management[3-5]. Small-scale low resource farmers cherish the soil fertility gradients (homefields in West Africa, banana homegardens in Uganda and darkoa ensete plots in Ethopia) as important tenets of management and food security enhancement[6-8].
In many developing countries such as Kenya, agronomic recommendations are blanketly based on soil types and agro-ecological zones and do not embrace the gradients[9,10]. While site specific management is known to be important for enhancing production efficiency, tools for disaggregation of farms into niches (analysis of soil samples using a grid scheme and topographic position, resource flow maps, partial nutrient balances and evaluation of local soil quality indicators) are costly and slow[2,8]. Individually the approaches give good information, but only the ones that are fast, less costly with ability to quantify the diverse management effects is ideal[11,12]. Participatory Rural Appraisal (PRA) tools have been developed and emphasis is now on integrating the management options into user-friendly decision guides[2,13-15]. One such guide should reflect inclusion of soil fertility variations into gradients related to nutrient use in farmers’ fields. This can however be achieved only if the framework for local soil fertility gradients and their impact on functioning of integrated nutrient management and food security is established. A study was therefore carried out in farmers' fields in Western Kenya to identify and characterize soil fertility management gradients in terms of known soil properties and crop yield.
MATERIALS AND METHODS
Site description: Kabras Division in Kakamega District, Western Kenya lies at latitudes 000 52’ N longitudes 340 52’ E and altitude of 1300 to 1900 m above sea level. The area has a population density of 700 persons km-2, average temperature of 17°C and bimodal rainfall of 2000 mm per annum. The soils are highly weathered clay loams classified as Rhodic ferralsols and Humic acrisols[16]. Most farmers are smallholders with land area varying between 0.2 and 4 ha on which they keep livestock and grow maize, beans, potatoes and bananas for subsistence. Malava forest occupying about 1300 ha of land with both indigenous and exotic trees which forms part of the division provided a reference point for analyzing the impact of management on soil fertility.
Participatory mapping of soil fertility management niches: Two villages (Muhoni and Shitirira) located on a Humic acrisol and Rhodicf ferralsol were purposively selected, physically identified and delineated based on interviews with local leaders and agricultural staff. Participatory soil mapping[17] was used to enlist soil fertility management niches. Six key informants chosen during the village meeting (Liguru’s baraza) drew soil maps, which showed physical boundaries of the village, local soil types, catena and the land use types.
A transect line was established within each village and the soil fertility management niches determined through brief walks at farms. Characteristics of different fields and local categories of the soil types, current and previous crops were noted. At each home, members of the household drew a map to show different soil fertility management niches.
Soil samples (0-20 cm deep) were taken randomly from five spots in the soil fertility management niches and then bulked to form composite samples. A sub-sample (500 g) of the soil from each niche was taken and transported to laboratories of Egerton University’s Department of Soil Science where it was air-dried, ground to pass through a 2 mm sieve and analyzed for soil texture, pH and exchangeable bases. Soil texture was determined by the hydrometer method, soil pH by the soil: water solution ratio of 1:2.5 and total organic carbon (C) by Walkley-Black wet oxidation using a solution of potassium dichromate and concentrated sulphuric acid[18]. Total nitrogen (N) was analyzed by Semi Micro-Kjedhal digestion, P by spectrophotometery, Calcium and Mg by atomic absorption spectrophotometer and K by flame photometry[18]. Exchangeable acidity (Hp) was determined[18] on soils with pH less than 5.5.
A research designed farmer managed experiment, laid out as a Randomized Complete Block Design with farms (eight from each soil type) as replicates, was then carried out. Hybrid maize variety (H614) was planted in 20 m2 plots in the soil fertility management niches. Calcium Ammonium Nitrate (CAN) and Triple Super Phosphate (TSP) at the rate of 75 kg N and 60 kg P ha-1 as recommended[19] was applied. Control plots (no fertilizer applied) were included to facilitate measurement of inherent soil fertility. Grain yield of maize was determined at physiological maturity and related to the soil fertility management niche. The effects of management on soil properties and maize yield were evaluated with analysis of variance (ANOVA) while Pearson correlation and multivariate analysis (Principal Component Regression) were used to determine the joint relationships among variables.
RESULTS
Farmers identified the soil types based on texture (ease of cultivation and consistence), colour and productivity (measured by crop yield and indicator plants). Hierarchically, they divided them into dark red soils that were more fertile and with a high water holding capacity (Olukusi olumali) and the bright red soils that were less fertile with a low water holding capacity (Olukusi olubesemu). The dominant soil type (Olukusi olumali) in Muhoni village was deep (up to 196 cm), acidic (pHkcl between 3.9 and 5.8), infertile (CEC less than 10 cmol kg-1), clayey and dark red brown (7.5YR3/2 and 2.5YR2.5/4 when moist). They were thus classified as Humic acrisols[20]. The dominant soil type (Olukusi olubesemu) in Shitirira village was clayey, dusky red (2.5YR5/4 when moist), acidic (pH between 3.9 and 5.5) and CEC less than 10 cmol kg-1. They were scientifically classified as Rhodic ferralsols. The soil in the valley bottomlands had relatively high soil pH, gleyic properties (water perched at 100 cm) with histic A-horizon and classified as Mollic gleysols[20].
Analysis of the farmers’ diagrams showed a variety of anthropogenic soil fertility management niches generally grouped as productive (homegardens, old kraals, old hutsites and valley bottomlands) and non-productive (anthills and murram soils) fields. The productive fields were usually located far off from the homestead (Table 1). The spatial location of these niches was commensurate with their uses. The productive niches were used mainly for priority (high market value and/or food) crops like maize, tomatoes and vegetable production (Table 2 and 3). All farms had homegardens that were located near homesteads from where they got household wastes and poultry manure.
Old kraal sites resulted from in-situ management of cattle where they were tethered or allowed to roam during the day and restricted to enclosures (kraals) near the homesteads at night. Over 50% of the farmers kept cattle in kraals at night while 30% tethered them within their homesteads.
Manure from kraals was piled in heaps for composting before being transported to the homegardens and sometimes outfields. The non-productive niches (outfields) were used for maize, sweet potatoes, sugarcane and pasture.
| Table 1: | Frequency (%) of the soil fertility management niches in farmers’ fields |
| Table 2: | Average land area (ha) and frequency of crops grown in different soil fertility niches in Muhoni village |
| *Homesteads were not cultivated | |
Variation in management was reflected in differences in soil properties. Productive niches (homegardens, old kraals, valley bottomland and sand forests) had higher soil pH, organic carbon and exchange bases than the non-productive niches (Table 4). Correlation coefficients for the soil properties ranged from 0.3 to 0.85. There was a positive correlation between: maize yield and pH, K, Mg and organic carbon; silt and pH, Ca, Mg and organic carbon; pH and P, K, Ca, Mg and organic carbon; P and K, Ca, Mg and total N, K and Mg and organic carbon (Table 5). A negative correlation however existed between; maize yield and sand fraction and niche; Niche type and silt content, pH and K, sand and clay fraction, silt, pH, Ca and organic C; soil type and silt, pH, P, Ca, Mg and organic C; clay and silt.
Based on Principal Component Regression (PCR) the soil fertility properties were grouped into six factors (Table 6).
Inherent soil fertility properties (maize yield, soil type, silt, available P, pH, Mg and K) characterized factor 1 and explained about 20% of the total spatial variation in soil properties. Rhodic ferralsols had consistently less crop yield than the Humic acrisols hence the negative sign. Productive soil fertility niches had higher crop yield and pH than the non-productive ones that were usually located further away from the homestead.
| Table 3: | Average land area (ha) and frequency of crops grown in different soil fertility niches in Shitirira village |
| *Homesteads were not cultivated | |
| Fig. 1: | Effect of niche and soil type on maize yield (kg ha-1) |
Factor 2 consisted of texture (sand and clay fractions) and the Ca level (inherent physical characteristics and management) and explained 12% of the total spatial variation in soil properties. Total nitrogen, organic matter and exchangeable acidity which are a function of management characterized factor 3 and explained 9% of the total spatial variation in soil properties.
The on-farm experiment showed a significant niche by soil type interaction (Fig. 1).
Crop yield, available P and exchangeable acidity explained about 9% of the total spatial variation in soil properties and represented factor 4.
| Table 4: | Soil properties of the farmer perceived management niches |
| Values followed by the same letter (s) in a column are not significantly different | |
| Table 5: | Partial correlation coefficient of soil properties in the soil management niches |
| *Significant at p≤0.05,**significant at p≤0.005, N = 84 | |
| Table 6: | Rotated factor patterns of the soil properties of the management niches |
| * Letters in bold indicate significant (p≤0.05) relationship | |
Factor 5 characterized clay content and Ca and explained about 8% of the total spatial variation. Factor 6 characterized the soil fertility management niches and explained 6% of the total spatial variations in soil properties.
Productive soil fertility niches had a mean grain yield of 4.3 t ha-1, harvest index of 0.37, stover weight of 23 t ha-1, average 100-seed weight of 45 g and 0.5% sterile ears. The non-productive niches had grain yield 2.8 t ha-1, harvest index of 0.42, stover weight of 11 t ha-1, 100 seed weight of 38 g and 5% sterile ears.
DISCUSSION
A variety of soil fertility management niches ascribed to topography, soil type, physical discontinuities (valleybottomlands, hillsides and swamps) and management (homegardens and/or livestock facilities) existed and influenced the farmers’ decisions to allocate resources. Farmer perception about spatial heterogeneity in soil fertility has been shown to influence management[6]. The niches were classified as productive and non-productive where the former included homegardens, old kraals and valley bottomlands. The productive niches resulted from accumulation of organic resources like Farmyard Manure (FYM) and household wastes and were mainly used for high value food and cash crops. The non-productive niches like murram soils and outfields (normally located far away from homesteads) resulted from nutrient depletion through the grazing, movement of crop produce and residues and erosion. The niches were used for less nutrient demanding crops like sweet potatoes and cassava. As a result of limited productive land and need for more food, maize (a priority crop) is sometimes also allocated to the low soil fertility management niches. Studies at farm and watershed levels in Colombia, Vietnam and Kenya observed similar soil fertility gradients[8,21,22].
The variation in soil fertility management was also reflected in differences in soil properties. A higher amount of clay in home gardens and natural pasture was probably because of human influence (mud from houses) and termite activity. The negative correlation between Ca and clay content is however an indicator that the soil is dominated by the low activity clay minerals that predispose it to leaching of exchangeable bases. As expected, there was a negative correlation between silt content and niche type because of movement of organic inputs to productive niches near homesteads. Conversely, productive niches being near homesteads and used for priority crops were managed more intensively hence less loss of topsoil[15]. Soil texture has been found to be an important indicator of soil fertility and crop yield in farmers' fields in Central and Western Kenya[10,23].
The significant positive correlation between silt, pH and exchangeable bases points to the important role of soil organic matter as a source of nutrients in the productive niches. Soil organic matter has similarly been shown to have a significant influence on the physicochemical properties of highly weathered soils in the tropics[3]. Silt and organic matter content have been found to be important indicators of soil fertility gradients in farmers' fields in Central Kenya[24,25]. In this study, there was a difference of about 19% soil organic carbon between the productive and non-productive niches and over 50% between cultivated and forest niches. High levels of organic input (FYM and household wastes) and reduced nutrient outputs may have contributed to the high soil organic matter in the kraal sites, homegadens and natural pastures. High sedimentation level and low decomposition could have contributed to the high organic matter content in the valley bottomlands. Long-term experiments in agro-ecosystems[26,27] found similar decline in soil organic carbon and attributed it to the cultivation effects. The positive correlation between soil pH and available P, Mg, Ca and K, implies that these soils require liming, soil organic matter and appropriate P fertilizer types if productivity is to be enhanced. High exchangeable acidity and ultimately low pH observed in outfields (non productive niches) could however be attributed to low soil organic matter. The high amount of P in the valley bottomlands relative to other niches could be attributed to deposition of erosion material originating from farms upstream.
Humic acrisols had generally higher maize and kernel weight but productive niches in the Rhodic ferralsols had significantly higher kernel and maize weight. Farmers in the Rhodic ferralsols had large land area and a higher number of cattle per household. The resulting old kraal sites were subsequently more productive because of shorter cultivation periods and higher manure accumulation compared to those in the humic Acrisols. The significant interaction implies that management can increase grain yield irrespective of inherent soil fertility. Humic acrisols had a significantly lower harvest index (0.37) compared to 0.42 in the Rhodic ferralsols. The nutrient stress in Rhodic ferralsols may have forced the crop to partition more dry matter towards the grain and hence the higher harvest index. Humic acrisols had heavier stover than Rhodic ferralsols. The productive niches had higher stover weight (23 t ha-1 fresh weight) than the infertile niches (11 t ha-1). Niche types affected fertility (number of fully formed ears ha-1). Productive niches had less sterile ears (0.5%) compared to 5% in the less productive niches.
Food security an important driver of soil fertility management at farm level has been defined as the capacity to procure adequate food supplies in a stable and sustainable manner[28]. Productive niches account for less than 10% of the total land area per household but give an average yield of 4.3 t ha-1. Given an average family size of 6 members with an annual maize requirement of 900 kg, productive niches could satisfy the food requirements per se. Unfortunately most farmers do not have many sources of income hence rely on the farm to satisfy the food and socio-economic requirements. The varying types and levels of needs make it difficult for productive niche alone to suffice. Hence cultivation is extended to other areas irrespective of the low productivity. In addition to the high yield, productive niches are planted early in season so as to take care of the temporal food deficits that commonly occur at farm level. The cropping system also allows income generating activities like production of tomatoes and vegetables. Although market orientation of the productive niches seem to enhance productivity (crop yield and soil chemical properties) information from the nutrient balance model is however contradictory. This implies that more refined indicators should be sought. A study of the origin, magnitude and importance of soil fertility gradients in Western Kenya has also shown that soil fertility management was related to the wealth status[10].
Farming should therefore be linked to marketing and the farmers’ knowledge. Valley bottomlands, the sinks for alluvial deposits, were also an asset to the low resource farmers. This patch provided diversity in food production and also helped to tap the nutrient lost up-stream through erosion. The non-productive soil fertility niches in Western Kenya are known to be responsible for the low farm productivity[4,29].
CONCLUSIONS
Variation in crop yield attributed to the gradients in soil and crop management are common at farm level. Productive soil fertility niches were attributed to anthropic activities (management of livestock and FYM, household wastes, dwellings and crop residues) while nonproductive niches were caused by nutrient transfer through crop harvests and erosion. Farmers manipulated soil fertility gradients and allocated valuable (food and cash) crops to productive soil fertility niches and left the nonproductive niches fallow or used them for sweet potatoes. The adaptive management was often reactive (based on predetermined soil fertility) and sometimes proactive. The major drivers of the soil fertility gradients were production objectives like food, income and risk aversion. A further analysis of the soil fertility gradients shows that they should form a major part of the framework for soil fertility management. The scientific indicators of productivity included soil organic carbon (important for maintenance of soil quality), pH, exchangeable bases and crop yield. The local indicators soil texture, colour and crop yield were different from the scientific indicators and fell outside the local knowledge. There is need to find a common functional ground for the knowledge systems.
ACKNOWLEDGMENT
I am greatly indebted to the Danish International Development Agency which provided financial support through Tropical Soil Biology and Fertility institute of CIAT for this study.