The Umande Farmers’ Cooperative Society Limited is located in Laikipia County (Figure 1). The cooperative is located in a semi-arid area receiving less than 700 mm of rainfall per annum. The members of the cooperative are smallholders keeping 2–3 cows on small land holdings ranging from 2 to 4 acres. The average milk output per farmer before the project was 2 liters per farmer during the dry season and 5 L in the wet season. The cooperative was started in 2016 and has 188 registered members who are dairy farmers in Umande Location but only 90 members were supplying milk at the time of data collection. The ad hoc formation of the farmers’ cooperative in 2016, through the amalgamation of crop-based groups had several challenges; not well aligned with Kenya’s milk regulations (Cooperative Act CAP 490), limited skills in commercial dairy farming, low purchasing power, pasture shortages, poor milk handling and hygiene, limited knowledge on good animal husbandry practices, lack of good governance practices, poor methods of milk transportation, lack of cooling facility, limited access to markets, and vulnerability to droughts.

Apart from milk, the farmers in Umande also practice mixed farming which is largely subsistence in nature growing maize, beans, potatoes, and vegetables in addition to raising livestock (cattle, sheep, goats, and poultry) to meet multiple household needs. Some foodstuffs are sold in the local markets making up a significant portion of the local informal trade sector. Other food systems exist in the region including agro-industrial, that produce vegetables for export to European markets and provide employment opportunities for subsistence farmers in the study area. Additionally, the regional food system comprising products such as meat, wheat, and barley value chains occupies a large part of the landscape in Northwest Mt. Kenya region (Jacobi et al., 2020a; Mwangi et al., 2020; Mwangi et al., 2021). The food systems in the area (local, regional, agro-industrial) offer diverse livelihood opportunities, in addition, to sharing scarce resources of water, land, and labor (Peter et al., 2018; Mutea et al., 2019; Mwangi et al., 2020; Mwangi et al., 2021). The resources water and land are scarce in the region because of the following reasons; there is a high demand for water for irrigation by the flower farms, horticulture farms, and smallholders, high population density in the area and high demand for land by multinational companies producing flowers and vegetables for export, commercial ranches, and smallholder farming puts pressure on land resources. On the other hand, labor is very expensive and scarce in the region due to many commercial farms that are preferred employers for wages that support livelihood. This makes it difficult for smallholders to access labor cheaply when the demand on the farms is high.

For objective one, data were collected using 50 members out of the active 90 cooperative members in 2.5-day workshops for the participatory food system assessment. The authors invited all the members of the cooperatives to participate in the workshops, however only 50 out of 90 consistently attended the 2.5-day workshops. For objectives 2 and 3, we used data from two Focus Group Discussions, in-depth interviews with five farmers, and eight key informant interviews (two officials from Laikipia County Government MOALF, two from the Ministry of Cooperatives, and Micro, Small, and Medium Enterprises Development, two staff from a local institution that participated in the knowledge co-production, and two cooperative officials). In total 70 farmers participated in the study exclusive of the Key Informants and in-depth interviews with cooperative leadership and lead farmers. During the workshops we explained the pillars and indicators of food sustainability using Swahili language which is understood by all, second, we agreed on a score of 1–5 (Table 1), and then the farmers scored each indicator through discussions and building consensus. Other activities undertaken during the workshops were sharing experiences from other farmer-led collective actions, discussions on branding of the cooperative, and resource mapping. Postdoc researchers who had been involved in the entire research project right from initiation and had adequate knowledge of the context and the framework facilitated the workshops, shared their experience from other projects in other contexts, explained the framework to the farmers using various drawings, and collected the data from other stakeholders. We applied the FOODSAF tool (Rist and Jacobi, 2016; Rist et al., 2021), to assess to what extent the agropastoral food system in which the milk value chain is embedded, contributed to food security, guaranteed the right to food, supported the reduction of poverty and inequality, promoted social-ecological resilience, and protected the environmental resources (Table 1). The participatory process was made possible by the use of a 5-point Likert Scale. To enhance the consistency of the data collected, all the participants used one venue, and the facilitation was conducted by three senior researchers (two of them postdocs and one co-principal investigator) and two additional staff. A Likert scale is a psychometric scale of agreement applied in social sciences to assess respondents’ opinions and attitudes about a specific subject (Göb et al., 2007; Batterton and Hale, 2017; Gaitán-Cremaschi et al., 2019; Lionello et al., 2021). The framework and scoring process had been validated in food systems in Zambia, Brazil, and Colombia and found to be useful in the diagnosis of challenges facing food systems and designing interventions through a participatory process. The facilitators for the series of workshops had experience in using the framework in other contexts.

The research was a 6-year research project that started from 2015 to 2021. The first year (2015) was used for scoping, mapping of stakeholders, identifying graduate students, and strengthening the methodology. In the first 3 years (2016–2018), empirical research was conducted to understand the food systems in the two countries Kenya and Bolivia. The second phase of the research was action-oriented (Transformative Pilot Actions using the FOODSAF) research (2019–2022). This study was conducted in the second phase, first by following all the activities to understand the process of knowledge co-creation and how it contributes to development, and secondly by conducting research at the end of activities to assess farmers’ and stakeholders’ opinions on benefits. The action-oriented research activities were affected by COVID-19, and hence, there was an extension of 1 year. In the second phase, the research was extended to other countries Zambia, Ghana, Brazil, and Colombia. In total, the research facilitated 15 projects (Transformative Pilot Actions) in six countries using the (FOODSAF). Apart from COVID-19, there was a severe drought that affected the study area, and hence farmers could not participate in any activities rather they were allowed to cope with the drought.

Content analysis was used to analyze data from workshops, FGDs, and in-depth interviews because the data were largely qualitative. The content analysis applied in this study is a naturalistic and interpretive approach and not a quantitative description that relies on reliability, validity, and generalizability (Ahuvia, 2001). This is attributed to the focus of the study which was on the process of coming up with a community-led technology intervention, the perceptions, and opinions of farmers about the benefits of the cooperative, and the milk cooler. Content analysis is used to analyze verbal, audio and video, and visual data (Kleinheksel et al., 2020). The Likert scale data were summarized into tables and then used to draw spider diagrams together with the farmers (Figure 2). The next step was to conduct a rank analysis of the strong and weak indicators to arrive at three agreed-upon indicators (including justifications), to form the next deliberations on the most sustainable intervention that would benefit the majority of the farmers. Data from FGDs and interviews were organized into themes and then ranked by building consensus together with the farmers. Content analysis has been applied in food systems research to analyze qualitative data (Lin and Mao, 2015; Béné et al., 2019).

During the workshops, the farmers participated in the assessment of their food system to identify the strong and weak links. The weakest links in the food system formed the basis of the negotiation of the collective action. The outcome of the assessment showed that the strongest indicator under the food security pillar was power relations with a score of (3). The cooperative members applied mechanisms to manage power within the food system and there were limited power imbalances among the actors. During discussions, one of the members said that ‘they are all equal as subsistence farmers’ meaning there were no power differences among actors in the food system. Household food security and the capacity of the food system to store and process food were given a score of (2) meaning fair/satisfactory. The reasons given were that the area is semi-arid hence during droughts they experience food shortages and some households receive relief food. The farmers have limited access to post-harvest management technologies, making food storage and processing at a household level a challenge.

The right-to-food pillar had the highest score for active participation and non-discrimination (3), while the lowest score was access to information (2). The reasons given for the high score were the perceived high level of social-self organization that led to the formation of the cooperative and limited discrimination within the community. However, they perceived that they did not have adequate information about the food they ate from outside. Information from the government to farmers was also limited due to less contact with extension personnel. The poverty and inequality pillar had strengths access to social-technological infrastructure (3), mainly because they are close to passable roads to transport farm produce to the markets; however, the roads were impassable during the wet seasons making transportation of perishable farm produce difficult. The indicator sources and levels of income and performance of the value chain both scored (1.5) attributed to limited direct access to markets. The majority of the smallholder farmers use middlemen who buy farm produce at low prices and sell the same at high prices in neighboring towns due to farmers’ limited access to transportation means, muddy roads during rainy seasons, as well as farmers having small quantities of farm produce during some seasons. However, large-scale farmers (horticulture, wheat, and livestock for meat) from the same region access markets directly. This means that the middlemen obtain a higher share of the profit than smallholder farmers (Mwangi et al., 2020). The government can incentivize middlemen by improving infrastructure so that the savings on the cost of transport can benefit farmers.

The strongest indicator under the environmental performance pillar was the health impacts of the food system (3). The farmers perceived their food to be of good quality as compared to people living in cities because their food is “natural.” The farmers believed that the quality of their environment was not as modified as urban areas—they had some natural plants and animals. However, the carbon footprint indicator had a score of (2), attributed to increased deforestation and limited efforts to restore degraded areas. The environmental benefits of the food system landscape scored (1), meaning very bad (Figure 2). The horticulture, wheat, and flower farms near the community were perceived to be responsible for the heavy use of agrochemicals some of which ‘contaminated’ the environment, especially water and air. However, there is increasing use of agrochemicals in smallholder agriculture (Jacobi et al., 2019a). For the social-ecological resilience pillar, the indicator social-self organization received the highest score of (3-good) attributed to the ability of farmers to form several groups which were then merged to form a milk cooperative. Social self-organization among smallholders is an important means of overcoming challenges within the food system- challenges that are difficult to tackle at the individual level (Mukhovi et al., 2020a). However, diversity and use of local knowledge both received a score of 2 (fair) because the farmers viewed their farms to be less diverse, used few local seed varieties, and dairy cows were mainly crossbreeds.

The three indicators with the lowest scores were the basis of negotiation for the collective action. These were the environmental benefits of the food system landscape (1), sources and levels of income, (1.5), and performance of the value chain both scored (1.5) (Table 2). After great reflection and negotiation, the farmers prioritized the indicators by building consensus. The performance of the value chain was ranked the highest and hence formed the basis of further deliberations with more stakeholders to initiate a development intervention. Farmers indicated their hard work to produce; however, the farm produce fetched low profit, attributing this to selling through brokers/middlemen. Middlemen play a critical role in the marketing of farm produce in the global south (Abebe et al., 2016; Nguyen Viet and Nguyen Anh, 2021).

The negotiation for the collective action was achieved by several meetings between local institutions and cooperative officials, and later by members of the cooperative. Building trust was not a challenge due to the long relationship between local institutions and the community. Engagement with stakeholders’ right from the beginning of the research was important to understand their interests and perspectives. The reasons why the farmers prioritized improving the performance of the milk value chain were; that it has a higher multiplier effect on household income, 95% of the farmers in the area have a dairy cow, and the milk cooling system would promote direct market access (Table 2). Apart from the improvement of the performance of the milk value chain, the farmers required training on dairy farming, value addition, table banking,³ and cooperative governance. The Ministry officials conducted a training needs assessment and launched a tailor-made comprehensive training together with County government staff from the Ministry of Agriculture, Livestock and Fisheries (MOALF), and Ministry of Cooperatives, and Micro, Small and Medium Enterprises Development.

One of the solutions the farmers identified was capacity development on dairy farming and cooperative governance, value addition, milk handling and safety, and best practices in dairy animal husbandry. This was undertaken by strong collaboration with the MOALF at the county level. The ministry identified personnel in the livestock sector and extension staff who conducted a series of practical training sessions for dairy farmers. In total, 200 farmers attended the tailor-made training sessions. The topics covered were; pasture and fodder establishment, dairy cattle feeding, pasture, and agribusiness, group dynamics and cooperative formation, livestock waste management, disease control, breed upgrading, fodder utilization and conservation, silage preparation, market linkages, cooperative governance, quality milk production and milk handling, milk value addition, and resource mobilization, and table banking as well as building social capital. Additional training was on trainers of trainers (TOT) where farmers learned how to be trainers of other farmers for the sustainability of farmer-to-farmer capacity building in the future. One of the farmers donated a section of his land for the establishment of a fodder farm for demonstration purposes. The farmers were trained on different types of fodder and how to establish them.

During this study, we visited several farmers and observed the establishment of Rhode grass (Chloris gayana), lucerne (Medicago sativa), yellow maize (Zea mays L.), mangels (Beta vulgaris), Sudan grass, Brachiaria, green leaf desmodium (Desmodium intortum (Mill.) Urb), Napier grass, and fodder trees (Marley berry and tree lucerne), which they use to make silage or feed directly to dairy cows. This has contributed to improved feeding and therefore improved milk yield per cow from approximately 3–5.5 L per day (Interview with a female cooperative member who is a retired extension officer). As a result of the training, the number of farmers that supply milk to nearby factories increased from approx. 30–71 (key informant interview). Demonstration farms have proven to be an effective way of social learning by farmers (Ingram et al., 2018;Sutherland and Marchand, 2021; Mukhovi et al., 2020b). The farmers interviewed indicated that as a result of the capacity development, milk production had increased from 180 L to approx. 500–600 L per day due to improved feeding and increased production per cow as a result of improved breeding made possible by artificial insemination (AI), and the purchase of improved bulls (interview with one of the founding members of the cooperative). Zebu breeds in the region include Sahiwal and Boran, while other semen for AI comes from imported breeds such as Charolais, Simmental, and Hereford. Livestock Breeding Regulations of 2023, established Kenya Livestock Breeding Bureau, provides guidelines on animal genetic resources, AI, livestock and livestock inputs, breed society, breeding standards, embryo transfer, experts, genetic materials, importation of animals or genetic material, indigenous livestock breeding, pastoralists, and service providers among other provisions (Government of Kenya, 2023).

Prior to the training, milk quality and safety were poor due to poor handling and hygiene. This resulted in a daily milk rejection of 50 L by the sole buyer. However, after training, the amount of milk rejected was reduced significantly due to improved hygiene and safety as a result of transporting using stainless steel milk churns instead of plastic jericans and improved handling (Interview with one of the officials). Testing of milk quality for somatic cell count to determine subclinical mastitis and quality of milk was done at the factory and other milk cooling plants in the region where the farmers sold their milk. By the time of writing this study, there were several gaps; the machine for milk quality testing to improve safety standards, value addition, employment of experts to support the operation of the cooling machine, and other physical and social infrastructure. However, Zhou and Jin (2009) observed that improvement in food safety standards depends on the size of the cooperative, farmer perceptions and attitudes toward standards, expected market, and anticipated benefits and costs involved. Improved quality of milk is also associated with increased prices. In addition, the cooperative received two motorcycles, one from the Kenya Dairy Board and a second one from a former Member of Parliament to facilitate the transportation of milk safely.

One of the challenges farmers experienced was the marketing of their milk. The second component of the intervention was to construct a cooler house, install a milk cooling system, and establish a transportation system to reduce the middlemen challenge and hence increase the income for farmers (Figure 3). Collaboration with local leaders and county and National governments from the initiation of the project contributed to successful resource mobilization that made the intervention successful. The County government provided land on which the cooler house was constructed, the national government bought the cooling system worth KES. 6M. A local research and training institution—the Centre for Training and Integrated Research in ASAL Development (CETRAD) supported the construction of the cooler house together with the farmers while the community provided labor and locally available materials. The financial resources for the capacity development, construction of cooler house, and purchase of milk cooling and bulking system came from multiple stakeholders; the funder (Swiss National Science Foundation), CETRAD, national government, County government, and the community. Facilitation for other components of the project such as milk transportation and ablution block came from a member of parliament at the time and a Member of County Assembly (MCA).

It was clear from the cooperative members that the livelihood of farmers had improved since the cooperative was formed and they foresee more benefits accruing from the milk cooling system. The farmers never used to sell evening milk;⁴ however, with the cooling system and the infrastructure that will be put in place, this will be achieved. Evening milk is a challenge to many smallholders and has been identified as a major reason for the formation of farmers’ cooperatives and milk-bulking systems in Kenya (Foster, 2015). The challenges of evening milk are limited access to markets, lack of storage, and limited infrastructure. Sell of evening milk had the potential to increase household income and expansion of the market. Improvement of household income has a positive outcome on expenditure on children’s education, food and nutrition, health, and poverty reduction (Cooper and Stewart, 2021).

Milk cooling systems enable farmers to refrigerate evening milk which would otherwise go to waste and also improve milk safety standards hence improving the competitiveness of the cooperative (Foster, 2015; Rojas et al., 2018). Farmers have access to income monthly with extra income generated from other farming activities (poultry and crop farming). Farm performance had a chance of improving as a result of cooperative membership (Verhofstadt and Maertens, 2014) as farmers reinvest income in different dairy and other farm enterprises with the potential to improve the economic and environmental sustainability of the farms (Candemir et al., 2021). Members of the cooperative were also accessing credit and advances to meet emergency household needs. Access to credit by smallholder farmers is a challenge in Kenya as a result of a lack of collateral. This is even worse for women farmers who are constrained by gender rules in African culture (Mukhovi et al., 2020b). Farmers prefer credit from informal sources due to the lack of collateral to obtain from formal institutions (Ullah et al., 2020).