steen003

Mixed crop-livestock (MCL) systems with small herds of improved dairy cattle breeds produce the bulk of Kenya’s milk. Kenyan dairy policies aim to increase the national milk production by stimulating the milk production at smallholder MCL farming systems through increasing productivity per cow and market orientation. Many development programmes and projects have suggested interventions to achieve increased cow productivity and increased market orientation. However, the adoption of such interventions has been relatively low. This low adoption of interventions could be due to the fact that variation between MCL farming systems is not taken into account in development projects and this could be due to the limited understanding of the variation in MCL farming systems and the context they operate in. Our hypothesis was that variation between MCL farming systems is determined by variation in market quality for inputs and outputs, variation in availability of production factors, and variation in biophysical context. These aspects are associated with spatial variation. Consequently, farming systems in different locations will be different, will have different constraints and different targeted interventions to overcome these constraints. However, the number of scientific studies relating spatial variation to variation in farming system development, constraints for development and targeted interventions is limited.

The aim of this thesis was to understand the variation in farming system development, constraints for development and targeted interventions for development, in order to increase market-orientation and dairy cattle productivity of smallholder MCL system in Kenya.

The first step to study this objective was to get insight into the spatial variation in farming systems. Chapter two had the objective to determine the influence of distance to urban markets on smallholder dairy farming system development. I used a spatial framework, and I selected MCL farming systems in urban location (UL, <15 km from the urban market of Nakuru town), mid-rural location (MRL, in between 20 and 50 km west of Nakuru) and extreme-rural location (ERL, > 50 km west and south-west of Nakuru). I held in-depth interviews with 30 farmers and organized 8 focus group discussions with groups of stakeholders to collect narratives and quantitative data about market quality, availability of production factors, farm performance and functions of dairy cattle. In UL, markets for inputs and for selling of milk were available and functional. Informal market chains with relatively high milk prices were predominant. Inputs like fodder, replacement stock and labour were scarce, whereas high-quality concentrates were costly. Furthermore, the availability of grazing land was limited, and the opportunity costs for family labour were high. Consequently, milk production per cow and per farm were relatively low. In rural location (RL), which comprised MRL and ERL, markets were also functional. Here, I found both formal and informal market chains with relatively low milk prices. Inputs like fodder, replacement stock and labour were available, whereas low-quality concentrates were cheap. Furthermore, the availability of grazing land was adequate, and the opportunity costs for family labour were low. In RL, milk production per cow and per farm were relatively low. I concluded that in UL, farm development was constrained by scarcity of fodder, replacement stock and hired labour, and the limited availability of production factors, such as land and family labour while in RL, farm development was constrained by the low quality of concentrates and low prices of milk.

Second, I looked at how different farmers within a location cope with constraints. I distinguished so-called positive deviants (PDs), which are farmers that overcame constraints and/or were perceived successful for dairy production and non-PDs, that were farmers that had not overcome these constraints. Chapter three had the objective to identify strategies that PD farmers deploy to overcome dairy production constraints. I classified farmers in UL and RL as PDs and non-PDs, based i) on experts’ and/or peers’ perception and ii) on economic performance. All farmers were interviewed to determine farm and household characteristics. In UL, five out of seven (71%) of the perceived PDs were classified as economic PDs. The main factors distinguishing economic PDs from non-PDs were relatively large herd size, high milk yield per cow and good balance between costs and revenues, which enabled PDs to realise higher gross margins than non-PDs. PDs achieved better control of cost with good animal husbandry practices, such as feeding, breeding and veterinary care. Input use (i.e. level, quality and cost), high milk price, good level of knowledge and skills, and financial stability enabled PDs to practice good animal husbandry. In RL, 4 out of 13 (31%) perceived PDs were economic PDs. Economic PDs in RL had large herds, since maintaining a large herd contributes to non-economic functions (particularly store of wealth and insurance) and to economic functions (farm milk output) of the dairy activity of the farm. The cows of RL economic PDs attained low productivity because of low input use. The inputs for increasing productivity, such as high-quality concentrates, were too expensive relative to the low price of milk, presenting economic risks to farmers. Therefore, many of perceived PDs were economic non-PDs because of low productivity and high cost of production. Results suggest that in UL, PDs overcame constraints by increasing herd size and intensity of production, whereas non-PDs lacked the skills and financial stability to increase herd size and milk production per cow. In RL, PDs overcame constraints by increasing herd size, whereas non-PDs lacked the skills and financial stability to increase herd size.

Third, to study variation among farming systems regarding their potential milk production given their biophysical context, accurate estimates of milk production in a lactation (MPL) are essential. Chapter four had the objective to assess this accuracy of MPL-estimates for four alternative data recording methods i.e. a limited number of test-days (TDs) per lactation, a single TD per lactation, a limited number of recall records per lactation, and a single recall record per lactation. At cow level, the relative mean absolute errors (RMAEs) for test-day data were 15% for records of limited TDs per lactation and 20% for records of a single TD per lactation. The RMAEs for recall data were 20% for records of limited recall moments per lactation and 25% for records of a single recall moment per lactation. At herd level, the RMAEs were 13% for records of a single test-day per lactation and 25% for records of a single recall moment per lactation. The results of chapter four suggest that the level of accuracy of estimating MPL based on recall data are acceptable.

Finally, Chapter five had the objective to identify to which extent the various biophysical factors define and limit milk production of dairy cattle in order to explore improvement options that contribute to mitigating yield gaps in smallholder systems in the Kenyan highlands. I selected farms with exotic cattle in UL (ULE) and RL (RLE), and farms with crossbred cattle in RL (RLCB). Actual feed and lactation data were collected through questionnaires administered in 42 farms, 22 in UL and 20 RL. Not all farms met criteria for inclusion and I did the YGA with 22 farms, 10 in UL and 12 in RL. I simulated potential and feed limited MPL, calculated the potential and feed limited yield gaps, and partitioned yield gaps according to biophysical factors. Feed was the most important biophysical constraint for increasing milk production per cow in all farm types and feed quality limitation explained 34% of the yield gap in RLE, 47% in ULE, and 63% in RLCB. Protein deficiency was the most frequently occurring constraining factor during the lactation and supplementing lucerne, with or without concentrates, increased feed-limited milk production by up to 32% for ULE, 45% for RLE and 88% for RLCB. Results of Chapter five suggest that supplementing feeds was necessary to increase productivity of exotic and crossbred cattle. Sourcing affordable protein supplements of good quality is thus a priority for increasing productivity.

In Chapter six, implications of variation in farming systems between and within locations for increasing productivity per cow and market orientation in Kenya are discussed. Difference in constraints for dairy development were found between UL and RL and between PDs and non-PD implying that different interventions should be addressed to the different farming systems. UL farmers and PDs likely are following the stepping-up livelihood strategy while RL non-PDs are hanging-in. For stepping-up, relevant interventions could include: i) developing market-oriented fodder production in RL and market the fodder in UL by strengthening the value chain and markets for fodder, ii) develop infrastructure to develop formal dairy value chains, and iii) develop training and extension aimed at farmers but also at other value chain actors to improve animal husbandry, marketing and entrepreneurial skills. Interventions for RL non-PDs should address both production function and subsistence function of cattle. The reason is that such farmers likely will only gradually transition from subsistence farming objectives to market-oriented farming objectives. Such interventions could include: i) improve the financial stability of farmers, e.g. organise and improve the access to affordable credit facilities, ii) improve farmers access to grazing areas, and iii) to breeding and selection programs with local breeds or crossbreeding, and iv) establish Farmer Field Schools to enhance farmers’ skills for dairy husbandry and fodder production. In summary, interventions for UL and PDs should be tailored for commercial production, since PDs and UL farmers are market-oriented, and if the fodder constraint is elevated, they may have high input use and high output per cow and per farm while interventions for RL non-PDs should be tailored for production and subsistence functions with large herds of dairy cattle.

The future dairy development in Kenya will follow diverse pathways. UL farming systems linked to the informal dairy value chain will persist unless government policies aimed at increasing milk quality for consumers forces UL farmers into formal dairy value chains. If this happens, UL farming systems will be faced with lower milk prices and be likely outcompeted by RL farming systems or may be forced to develop to intensive industrial systems with high productions per farm and as such, their sustainability will require considerations of environmental issues. MRL seems an optimal location for dairy development because land availability allows for fodder production, which could be further developed, formal dairy value chains are present and could be further developed, and also the infrastructure can and will be developed further. In MRL, moreover, livestock is coupled to land, hence nutrients can be recycled between livestock and crop production, and environmental pollution and soil mining can be minimised. In ERL, because of high transaction costs, intensive dairy production will likely not be feasible but the location could be prioritised for production of food, cash and feed crops with a focus on large herds for multiple functions.