Determinants of On-Farm Diversification Strategies: A Case Study of Smallholder Farmers in Mpumalanga Province, South Africa

Abstract

Promoting resilience, increasing productivity and sustainability, and profit maximization remain key challenges facing farmers globally. These are exacerbated by factors such as climate change, low to no access to technological advancement, financial constraints, poor technical and management skills, inadequate government support, and limited access to resources. However, there are diverse strategies that abound, including on-farm diversification, that farmers could leverage on to address these numerous and complex challenges. This study investigated the determinants of on-farm diversification strategies among smallholders in Mpumalanga Province. The study employed a quantitative approach using closed-ended survey questionnaires to elicit information from a total of 465 farmers who were randomly sampled from a total population of 14,411. The data gathered were analysed using descriptive statistics to determine the on-farm diversification strategies employed by farmers and the factors influencing the use of these strategies. A binary logistic regression model was employed to establish the relationship between on-farm diversification strategies and the determining factors. More than half of the farmers were female (51.8%), with only 48.2% male. The majority (59.1%) of the farmers were between the ages of 36 and 60, with only 20.2% youth participation in farming. Slightly more than half (50.8%) of the farmers practise mixed farming as their on-farm diversification strategy, while only 4.3% of the farmers practise mono-cropping. The study identified significant variables such as level of education (p = 0.001), secondary source of income (p = 0.057), farmland size (p = 0.022), number of farm assistants (p = 0.016), and on-farm diversification awareness as key determinants of on-farm diversification among smallholder farmers in Mpumalanga Province. Therefore, it is recommended that policies within the agricultural sector be revised to encourage on-farm diversification in order to motivate farmers to transition to agripreneurship for poverty alleviation, food security and rural economic development (RED).

1. Introduction

Over the years, diversification has been a common and the most reliable strategy for smallholder farmers to promote resilience, increase productivity and sustainability, and maximize profits [1]. This emanates from the numerous challenges facing smallholder farmers globally. Climate change, low to no access to technological advancement, financial constraints, poor technical and management skills, inadequate government support, and limited access to resources, among others, remain the key challenges impeding the growth of rural economies [2,3,4,5]. These challenges further hinder smallholder farmers from thriving and from experiencing sustainable livelihoods among other things [6]. As a result, diversification of both farming activities and of livelihoods has become necessary [5,7,8].

The contemporary literature maintains that diversification can either be a choice or a necessity for survival, especially among smallholder farmers [1,9,10,11]. These studies suggest that smallholder farmers diversify their farming activities and livelihoods in response to economic and environmental shocks. The economic shocks include issues around farmers’ access to the market, access to agricultural input, funding and infrastructure, and the persistent challenge of unemployment in developing countries [7,11,12]. Moreover, environmental shocks are intricately linked to the ever-changing climatic conditions. The change in climatic conditions comes with numerous challenges such as high infestation of weeds, pests and diseases, insufficiency of irrigation water, extreme weather conditions and patterns (particularly temperature), prolonged drought and flooding seasons, and soil degradation [13,14,15,16]. The implications of the environmental effects of climate change are evident in terms of reduced yields, decline in farm income, soil losses and degradation, intrusion of saltwater, crop losses because of flooding or prolonged drought periods [17,18,19,20].

Therefore, in the quest to remain in business and continue to meet the growing need for the supply of food and fibre, the creation of employment opportunities, continued contribution to the alleviation of rural poverty and food insecurity burden, farmers need to adapt and adjust accordingly. This means that farmers must be innovative and come up with survival strategies. Diversification is one of the key strategies for farmers to adapt and adjust to the harsh realities facing farmers across the world, especially smallholder farmers [21]. Grilli et al. [5] further suggests that farmers dependent on one source of income (single source of livelihood) are limited in their potential for wealth creation, poverty escape and food security. Livelihood diversification, delineated by Chambers and Conway [22], is the group of social and economic activities used by rural communities to cope with shocks. Nevertheless, Chambers [23] posited that sustainable livelihood among farming communities comprises environmental, economic and social activities leading to income security. Hence, livelihood diversification among smallholder farmers is one of the ways to ensure poverty eradication, which in turn contributes towards achieving the United Nations’ Sustainable Development Goals 1, 2, 3, 8, 12 and 13.

It is, therefore, an indisputable fact that diversified livelihood sources afford rural farmers an opportunity to mitigate production risks and improve their livelihoods by increasing their streams of income, thus increasing household income and therefore improving their standard of life [1,7,24,25]. While on-farm diversification involves the diversification of farming activities, off-farm diversification is about increasing the livelihood option sources [26]. Of course, these vary in framework and motivation since on-farm diversification involves crop rotation, mixed farming, double-cropping, agroforestry activities, integration of large and small stock (livestock), and multi-cropping [27]. Off-farm diversification involves combining farming activities with any of the following: civil servanthood, artisanship, rendering services, hunting and gathering as well as tourism [1,5,27].

The novelty of this study lies in its focus on empirically identifying the determinants of on-farm diversification among registered smallholder farmers in Mpumalanga, a context where such analysis has been limited. Unlike existing studies that treat diversification broadly or combine on- and off-farm strategies, this study clearly distinguishes on-farm, off-farm, and general livelihood diversification, isolating farm-level production decisions within the Sustainable Livelihoods Approach (SLA) framework. Using a large, randomly selected provincial sample, it integrates demographic, socioeconomic, and farm-structural variables into a binary logistic regression model to examine how assets such as education, farmland size, farm assistants, secondary income, and diversification awareness influence on-farm diversification. The study further provides new insights, including the unexpected negative associations of education and diversification awareness with on-farm diversification, challenging conventional assumptions that higher knowledge or awareness automatically promotes diversification. Its originality lies in contextual specificity, empirical validation of asset-based determinants, and the generation of nuanced, evidence-based insights for policy and rural development.

1.1. Problem Statement

The high levels of food insecurity, poverty, and unemployment in developing countries are some of the major challenges facing both individuals and governments. In South Africa, poverty levels vary greatly among provinces, with the national poverty rate at 68% and 55.5% of the population thriving under the upper bound of the poverty ratio [28]. Over and above that, 25.2% of the population live in extreme poverty, which means they are below or slightly below the food poverty line of R796.00 (less than sixty dollars) per person per month [28]. This is not just a threat to the country’s economic growth, but a cankerworm to the economy of the country and human well-being, especially in rural areas [29]. In Mpumalanga Province alone, 51% of its 5.14 million residents live in poverty, with 48% of these being unemployed [28]. About half (46%) of the unemployed population are young people, some of whom are also not engaged in education or training.

While the government commits to reducing poverty levels, self-help initiatives (SHIs) are crucial for both sustainable and rural economic growth. Mpumalanga Provincial Government has committed itself to reducing poverty levels to below 35% by 2030 through encouraging youth economic participation and creating more opportunities for young people [30]. Among the SHIs, agricultural activities are key economic drivers for economic emancipation, growth, and stability as well as for employment creation [5,7]. Once society can create employment opportunities and create a conducive environment for businesses to thrive, especially small, medium and micro-enterprises (SMMEs), poverty alleviation can be a possible mission to accomplish. Of course, Mpumalanga Provincial Government has done quite some work; for example, the Department of Agriculture, Rural Development, Land and Environmental Affairs (DARDLEA) supports young people and women in agriculture. The introduction of smart boxes by DARDLEA for climate-resilient crop production further attracted several young people to the agricultural sector [31].

Nevertheless, there is still a lot of work to be done within the agricultural sector. The issues of climate change adaptation, water scarcity adaptation, and sustainable income generation among emerging and smallholder farmers remain among the major challenges for the sector [5]. However, smallholder farmers have a range of options to select for adoption to alleviate the burden of both food insecurity and poverty. Among others, farmers have options to diversify farming activities, adopt Climate-Smart Agriculture Initiatives (CSAIs), implement Water-Smart Agriculture Strategies (WASASs), practise Conservation Agriculture (CA), employ Indigenous Knowledge Systems Approaches (IKSAs), and incorporate Digital Agriculture (DA) into their farming systems. Without further financial investments than the obvious production input, smallholder farmers can diversify farming activities with the aim of reducing financial associated risks, income stabilization, building resilient farming systems, and enhancing the viability of the farm business, as well as optimization of natural, financial, and social resource usages [8,32].

For the context of this study, on-farm livelihood diversification refers specifically to the deliberate combination or expansion of different agricultural enterprises within the same farm unit as a strategy to strengthen resilience, stabilize income, and improve resource use efficiency. It remains strictly within the boundaries of agricultural production and does not extend to non-farm income sources. In practical terms, this includes integrating crops and livestock into mixed farming systems, practising mixed cropping or intercropping, rotating crops across seasons, or incorporating agroforestry into existing production systems. A smallholder who cultivates maize and vegetables while also keeping goats or poultry is diversifying on-farm because all income streams originate from agricultural activities conducted on the farm itself. The intention is usually to reduce production risk, respond to climatic variability, enhance soil fertility through biological complementarities, and maintain more consistent cash flow throughout the year. The defining feature is therefore not the number of activities but the fact that those activities are all embedded within agricultural production.

Off-farm livelihood diversification, in contrast, involves the addition of income-generating activities that occur outside primary agricultural production, even when the household continues farming. These activities may include formal employment, self-employment in small enterprises, artisanal work, tourism services, social grants, pensions, or remittances from family members. A farmer who grows vegetables but also works as a teacher, operates a small retail shop, or receives a pension is engaging in off-farm diversification because part of the household income is derived from non-agricultural sources. General livelihood diversification serves as the broader umbrella concept that encompasses both on-farm and off-farm strategies, reflecting the overall portfolio of activities rural households combine to manage risk and sustain well-being within the framework of the Sustainable Livelihoods Approach. It recognizes that resilience is rarely built on a single income source and that households draw on natural, financial, human, and social assets to construct multiple streams of support. Maintaining a clear distinction between these concepts is essential in empirical research, particularly in studies that aim to isolate determinants of on-farm diversification without conflating them with wider household livelihood strategies.

Therefore, on-farm diversification strategies employed by smallholder farmers in Mpumalanga Province have not been well established in the literature. However, the contemporary literature [1,33] suggests a variety of on-farm diversification options for smallholder farmers. The determinants of on-farm diversification and the benefits of on-farm diversification among smallholder farmers have not been adequately investigated, especially in Mpumalanga Province. This, therefore, puts this paper in the best position to bring forth answers to critical research questions addressing these pertinent issues to enable policy makers, the government and government agencies, as well as development practitioners, to make informed decisions about both policy and rural and agriculture development initiatives and programmes.

1.2. Aim of the Study

The aim of this study was to investigate the determinants of on-farm diversification strategies among smallholder farmers in Mpumalanga Province of South Africa. The study further ascertained the on-farm diversification strategies employed by the smallholder farmers in the study area and their associated benefits among smallholder farmers in the study area.

2. Theoretical Framework

The adopted theoretical framework for the study is the Sustainable Livelihoods Approach (SLA), as shown in Figure 1 [22]. The SLA states that livelihood is a product of people, their abilities and skills, as well as their enabling sources to access food, assets, and income. It values local and indigenous knowledge; it appreciates the role of local indigenes in development as well as the efficient use of resources and assets to maintain a good standard of living [22,34]. This approach analyses the use of assets among rural households and communities towards poverty alleviation. Rural household assets include natural, physical, financial, human, and social assets (see Figure 1). The assumption is mainly based on the fact that rural households know their problems, and they, in a way, identified possible solutions and therefore have a role to play in ensuring the sustainability of their livelihoods. Hence, Sapkota [32] supports the notion of combining human skills, talents, and capabilities, tangible and intangible assets (as defined by Chambers and Conway [22]), and the economic activities within rural communities.

The emphasis is that poverty alleviation, economic and food sufficiency of households, entrepreneurial strategies, employment creation and conservative usage of livelihood assets are to be approached from a more holistic perspective rather than the traditional way [32,34]. This implies that the involvement of rural people, the government, policy, farming systems, and gender issues is key to sustainable livelihoods [8]. It can further be assumed that cultural beliefs and norms play a significant role in either failing or strengthening economic systems in rural areas for economic emancipation, household food security, and poverty alleviation [36]. This holistic approach seeks to ensure that none of the relevant and important stakeholders are left out in the quest for poverty alleviation and rural economic development. Now, within the farming systems as part of the elements of the SLA, it can be assumed that the use of diverse farming strategies leads to better life conditions, such as wealth accumulation and a high quality or standard of living. However, contemporary studies [8,34,36] argue that whether diversification positively or negatively influences wealth accumulation is dependent on several factors. These include but are not limited to the timeframe (period) of on-farm and/or off-farm diversification, household assets, farmers’ choice to either diversify farming activities or livelihood approaches, environmental contexts and factors, such as government policies and market dynamics, as well as cultural and political contexts. Diversification of farming strategies involves the use of all five classified assets (human, social, natural, financial and physical) to improve farming conditions and enhance farm income. These activities may vary from farmer to farmer and rural area to rural area; these are crop rotation, intercropping, mixed farming, agroforestry, eco-tourism, mono-cropping, and cover cropping [1,33].

On the other hand, some farmers may rely on both on-farm diversification and broader livelihood diversification, which encompasses off-farm diversification. Diversification of livelihood is advantageous towards risk mitigation and management, income stability, enhanced quality and/or standard of life, food security, and wealth creation [29,36]. In accordance with Mudzielwana et al. [37], the socio-economic characteristics of smallholder farmers, such as gender, source of income, household headcount (household size), migration (which is another pathway out of poverty), and land ownership, have a positive impact on livelihood diversification. Uddin et al. [38] contends that physical, human and financial resources (capital), when utilized among farmers, improved livelihoods. The SLA therefore assumes that diversification use differs per household, based on circumstances, anticipated outcomes, and capabilities and skills. Lastly, for any livelihood to be sustainable, it must be economically, socially, and environmentally sustainable [22].

3. Methodology

3.1. Description of the Study Site

This study was conducted in Mpumalanga Province in South Africa. Mpumalanga Province is the second-smallest province of South Africa (Figure 2), with a total population of 5,361,261, representing 8.4% of South Africa’s 63,976,884 population [28]. The province is divided into three districts, which are further divided into local municipalities. The three districts are Ehlanzeni, Gert Sibande, and Nkangala (Figure 2). Mpumalanga Province covers 6.5% of South Africa’s total land area and is mostly dominated by diverse agricultural activities, owing to the presence of both temperate and tropical climatic conditions. Hence, agricultural activities include the production of fruits, vegetables, agronomic crops, forestry, agrotourism, as well as animal production. The sub-tropical fruits produced in the province include nuts, citrus, mangoes, bananas, and avocados, while agronomic crops produced include sugarcane, maize, soya beans, dry beans, cowpeas, wheat, and sorghum, as well as medicinal plants such as ginger. The vegetable crops produced include cabbage, spinach, beetroot, onion, tomatoes, and butternut [30].

Considering the importance of the province in the country’s economy, it is worth noting that Mpumalanga Province is rich in diverse economic activities, including agriculture and forestry (commercial and smallholder farmers), mining (gold, coal, manganese, chrome and marble, and limestone), manufacturing, trade and services, construction and tourism [30]. These economic activities contribute to both economic growth and employment generation for the people of the province and the country at large. The population makeup is dominated by Black Africans (91.0%), followed by Whites (8.0%), and Indians and Coloureds together at 1.0%. The most spoken languages vary from district to district, with SiSwati, IsiZulu, XiTsonga, IsiNdebele, and English as the medium of instruction. This is influenced by the culture dominant in each district and influences both farming activities and off-farm livelihood activities. However, there is no documented research outlining the relationship between culture and farming and/or non-farm activities in the literature.

This is one province that was previously disadvantaged in terms of education but has made tremendous progress in transformation with the approval of its first university by the Democratic government in 2013, which opened its doors to higher education in 2014. The province’s population comprises 5.3% elderly (aged 65 and above), 51.3% young people aged 15–34, and 30.1% who are under the age of 15. Be that as it may, the province records a 48% unemployment rate, comprising the young people in the province. Therefore, the government has more to invest in the agricultural sector, which is the sector that employs a greater proportion of the population. Furthermore, policy on SMMEs and eco-tourism needs to strongly support the young people of the province in business start-ups, funding the already-existing SMMEs by young people, and, where possible, create a network of support groups to strengthen linkages among big companies and the SMMEs in the province.

3.2. Sample Method and Size for the Study

The study employed a simple random sampling method to sample the participants who took part in the study. These were sampled from a pool of smallholder farmers in Mpumalanga Province. Simple random sampling afforded all smallholder farmers an equal opportunity of being sampled to participate in the study [40]. The researchers obtained databases from all three districts of Mpumalanga Province and randomly sampled smallholder farmers who would be contacted to participate in the study. It is worth stating that the total target population was not broken into small groups for the purpose of sampling. This was achieved by defining the target population, which in this case is smallholder farmers in the province, whether they engage in livestock, poultry or crop production, as long as production is small-scale. Then, the definition of the target population was followed by the estimation of the sample size using Slovin’s formula, as highlighted below. This step was then followed by combining the three databases from the three districts; then, each member of the target population was assigned a number from 1 to 14,411, which was then used in the selection of smallholder farmers to participate in the study.

In cases where smallholder farmers were not available or unwilling to participate, another number would be randomly selected. However, the study excluded commercial farmers and residents who only had one means of livelihood except for agricultural production. Furthermore, the total number of registered smallholder farmers in the province is 14,411 [31]. Therefore, the sample size for the study was estimated at 389; however, due to the peculiarity of the study, the sample size was then increased to 465 to enhance the accuracy, reliability and validity of the results, to maintain consistency and enable generalization of the findings within the study area. Larger sample sizes in research come with the advantage of increasing statistical power, which enables researchers to detect outliers and errors, while larger sample sizes also serve as a good and strong representation of the target population. This was determined using Slovin’s formula, shown below (with a confidence level of 99% and a margin of error of 0.05):

$$\mathrm{n}={\displaystyle \frac{\mathrm{N}}{1+{\mathrm{Ne}}^2}}={\displaystyle \frac{14411}{1+14411{\left(0.05\right)}^2}}=389.197=\underset{¯}{389}$$

n—sample size (389);

e—margin of error (0.05);

N—total target population.

3.3. Methods of Data Collection

The study employed a quantitative research approach to elicit data from the randomly sampled participants using structured questionnaires. The structured questionnaires comprised questions that sought to extract information that is relevant to the research objectives and responds to the research questions. The quantified data collected were useful in observing patterns and averages in the use of on-farm diversification strategies by smallholder farmers. These further enabled researchers to predict existing relationships between variables (dependent and independent variables), as well as to generalize the findings of the study to the wider population [41].

Prior to the full-scale data collection, the structured questionnaire was subjected to both validity and reliability testing to ensure that it accurately measured the constructs aligned with the study objectives and produced consistent results. Content validity was established through expert review, where academics and practitioners in agricultural economics and rural development examined the instrument to assess the relevance, clarity, and alignment of questions with the determinants of on-farm diversification strategies outlined in the Sustainable Livelihoods Approach (SLA). Their feedback informed revisions to wording, sequencing, and the elimination of ambiguous or leading items. A pilot test was then conducted with a small group of twenty smallholder farmers, who were not included in the final sample, allowing the researchers to assess question comprehension, logical flow, and the appropriateness of response categories within the local context. Reliability was evaluated using internal consistency measures, and responses from the pilot were analysed to determine the stability and coherence of multi-item constructs, with acceptable reliability coefficients confirming that the instrument could consistently capture the intended variables before it was administered to the full sample of 465 respondents.

3.4. Methods of Data Analysis

The IBM Statistical Package for Social Sciences (SPSS) version 27 was used to analyse the data collected (IBM Cooperative, Armonk, New York, NY, USA). Descriptive statistics comprising percentages and frequency tables were employed to present the analysed data. The binary logistic regression equation shown in Equation 1 below was used to determine the determinants of on-farm diversification strategies employed by smallholder farmers. Binary logistic regression was ideal, especially with the presence of both categorical and numerical variables. Moreover, the ‘X’ in the equation below shows the predictor variables, while (Y) represents the dependent and dichotomous variable that was labelled as 1 if the smallholder farmers indicate that they diversify and 0 if they do not diversify on-farm activities. Hence, for objective 2, (Y) will represent the dependent variable that will be labelled as 1 if the farmer does diversify, and 0, if the farmer does not diversify. Therefore, the dependent variables are defined as 0 = No, I do not diversify on-farm activities, while 1 = Yes, I diversify on-farm activities.

However, (X) will be representative of the predictor variables as in objective 2, and the descriptions of all the predictor variables are shown in Equation 1 below. These include the age of the smallholder farmers, their gender, marital status, level of education, farmland size, farming experience, secondary source of income, number of farm assistants, and awareness of on-farm diversification. The use of each of these predictors is described in

Table 1. Independent variables operationalized as used in the binary logistic regression.

Variable	Description	Unit of Measurement	Sign Expected
Age (AGE)	Number of years lived by an individual smallholder farmer	Number in years	-
Gender (GENDER)	The state of being male or female among smallholder farmers	0 = male 1 = female	+
Marital status (MARITS)	The state of being married or not married among smallholder farmers	0 = married 1 = single 2 = divorced 3 = widowed 4 = engaged	-
Level of education (EDUC)	The level of educational attainment among smallholder farmers	0 = no school 1 = adult school 2 = primary school 3 = secondary school 4 = tertiary education	+
Farmland size (FLSZ)	The number of hectares own/utilized by individual smallholder farmers	0 = less than 1 hectare 1 = 1 to 5 hectares 2 = 6 to 10 hectares 3 = 11 to 15 hectares 4 = 16 to 20 hectares 5 = above 20 hectares	+
Farming experience (FMEXP)	Number of years in farming operations among smallholder farmers	0 = 1 to 5 years 1 = 6 to 10 years 2 = 11 to 20 years 3 = 21 to 35 years 4 = more than 35 years	+
Secondary source of income (SSICME)	Means of income except for main income stream (farming)	0 = farming only 1 = employed 2 = self-employed 3 = social grant 4 = pension 5 = remittances
Number of farm assistants	The farm workforce recorded as assistants rather than labourers among smallholder farmers	0 = none 1 = 1 to 3 assistants 2 = 4 to 7 assistants 3 = above 7 assistants	+
Household size	Number of persons per household among smallholder farmers	0 = 1–3 members 1 = 4–7 members 2 = 8–12 members 3 = more than 12 members	-
Diversification awareness	A state of being aware of diversification of farming activities among smallholder farmers	0 = yes 1 = no	+

, which further shows the operationalization of these predictor variables, which are included in Equation (1).

Statistical statement or equation and description of the regression model for objective 2:

Y = β₀ + β₁X₁ + β₂X₂ + Β₃X₃ + β₄X₄ + β₅X₅ + β₆X₆ + β₇X₇ + β₈X₈ + β₉X₉µ

(1)

where:

• Y = smallholder farmers’ choice to diversify or not to diversify on-farm activities, with smallholder farmers having diversified on-farm activities = 1, and 0 otherwise.
• X₁–X₉ represent the predictor variables, which have been demarcated as follows:
• X₁ = age (years lived by the farmer).
• X₂ = gender (farmer being male = 0, and female = 1).
• X₃ = marital status (the state of farmers being married or not married: 0 = married, 1 = single, 2 = divorced, 3 = widowed, and 4 = engaged).
• X₄ = level of education (educational attainment status of the farmer: 0 = no school, 1 = adult school, 2 = primary school, 3 = secondary school, and 4 = tertiary education).
• X₅ = farmland size (number of hectares of land owned by the farmer).
• X₆ = farming experience (number of years in farming operations).
• X₇ = secondary source of income (having other means of income generation except for farming).
• X₈ = number of farm assistants (the farm workforce in headcount).
• X₉ = diversification awareness (the awareness of diversification of on-farm activities).
• ꞵ₀ represents the constant.
• β₁–β₉ represent the standardized regression coefficients.
• µ represents the error term.

4. Results and Discussion

4.1. Demographic Characteristics of the Smallholder Farmers in the Study Area

Table 2. Demographic characteristics of smallholder farmers in Mpumalanga Province.

Variable	Categories	Percentages (%)
Gender	Male	48.2%
	Female	51.8%
Age	18–35 years	20.2%
	36–60 years	59.1%
	Above 60 years	20.7%
Marital Status	Married	29.7%
	Single	53.1%
	Divorced	4.9%
	Widow	10.1%
	Widower	2.2%
Level of Education	No formal school	11.0%
	Adult education	20.4%
	Primary education	9.5%
	Secondary education	49.0%
	Tertiary education	10.1%
Household Size	1–3 members	18.7%
	4–7 members	64.3%
	8–12 members	14.8%
	More than 12 members	2.2%
Secondary Source of Income	Only farming as main source of income	36.1%
	Employed	15.1%
	Self-employed	18.9%
	Social grant	12.0%
	Pension	14.8%
	Remittances	3.0%
Farming Experience	Less than 5 years	27.1%
	5–10 years	36.1%
	11–20 years	21.3%
	21–35 years	11.8%
	More than 35 years	3.7%
Farmland Size	Less than 1 ha	18.9%
	1–5 ha	54.4%
	6–10 ha	14.2%
	11–15 ha	10.1%
	16–20 ha	1.5%
	More than 20 ha	0.9%
Main Water Source	Borehole	11.0%
	River	32.5%
	Dam	38.3%
	Rainwater	7.3%
	Other	11.0%

shows the demographic characteristics of smallholder farmers in Mpumalanga Province. The results show that the majority (51.8%) of the 465 smallholder farmers who participated in the study were female farmers, with 48.2% being male. The decline in male involvement in the agricultural sector over the years may limit the likelihood of on-farm diversification, as women rely more on off-farm diversification than men [36]. However, this is explained by the great work done by Mpumalanga’s DARDLEA to encourage and support female farmers in the province. This is further in keeping with Boakye et al. [42,43], who postulated that female farmers are now dominating the sector. Of course, the South African government has done incredibly well in terms of the redress of gender equality and equity in the country, especially, in the agricultural value chain (AVC). The result shows that 59.1% of the smallholder farmers were aged 36–60 years, while a significant minority (20.7%) were aged 60 years or older. This is a great concern in the sector because it prevents old farmers from accessing funding and reduces the chances of the adoption of digital technologies for both use on the farm and for extension services. This result is in keeping with Zondo and Ndoro [44], who postulated that age is a factor in the adoption of new agricultural digital technologies, hence the persistent low adoption rates of innovation among smallholder farmers, especially in developing countries.

The results show that only 29.7% of the respondents were married, while the majority of the farmers (53.1%) were single. According to Dedieu et al. [45], farming families sometimes depend on family members for farm labour, whether male or female. Now, this result therefore creates concern for the sustainability of the smallholder farm business in terms of the input costs, especially with the spending on labour, because if 53.1% of the smallholder farmers in Mpumalanga Province rely on external personnel for farm labour, then they are not making profit, and there is next to no chance of transitioning to agripreneurship.

The result shows that the majority (49.0%) of the smallholder farmers had only up to secondary education, and only 10.1% had tertiary education. This result aligns with earlier findings by Sithole and Agholor [14], who postulated that the majority of farmers in rural areas have acquired only up to secondary education. This is a growing concern for the sector, especially amid the growing quest for the digitalization of farming, marketing, and extension services. The growth in online platforms necessitates that farmers be upskilled for better and improved technology adoption [46,47].

The findings show that most of the smallholder farmers (64.3%) have at least 4–7 household members, while 2.2% have more than 12 household members. The results show that 36.1% of the smallholder farmers depend on farming as their primary source of income, while 15.1% have alternative employment. At least 26.8% of the farmers rely on government support for a secondary source of income, while only 3.0% depend on remittances from family and relatives. It can, therefore, be argued that the number (26.8%) of the farmers who are dependent on government grants (pensions and child support grants) is attributed to two main factors, namely the farmers’ age and the high unemployment rate of 48.0% in the province (Table 1). The results further reveal that 57.4% have 6–20 years of farming experience, 36.1% have 6–10 years of farming experience and 21.3% have 11–20 years of farming experience, respectively. The findings also show that 11.8% of farmers have 21–35 years of farming experience. This finding has an implication in the adoption of on-farm diversification strategies due to lifetime experiences in the farming business. The majority of the farmers (54.4%) have at least 1–5 hectares of farmland, while 19.1% of the farmers farm on lands of less than 1 hectare. This influences how much of the on-farm diversification strategies a farmer can adopt and employ on their farm. At least 32.5% of smallholder farmers depend on rivers as a source of water, while 32.8% depend on dams as a water source. Of all 465 farmers, 11.0% have other sources of water for their farm operations and activities. These vary from community to community, as some depend on rainfall (dryland farming), and others are a combination of municipal water and boreholes.

4.2. The Smallholder Farmers’ Affirmation of On-Farm Diversification in Mpumalanga Province

Table 3. The smallholder farmers’ affirmation of on-farm diversification in Mpumalanga Province.

	Frequency	Percent (%)
YES	390	83.9
NO	75	16.1
Total	465	100.0

shows the respondents’ confirmation of their involvement in on-farm diversification in the study area. The finding reveals that 83.9% of the 465 respondents who participated in the study were involved in on-farm diversification, while 16.1% of respondents were not involved in on-farm diversification. This confirms the results of the study in

Table 4. The type of farming Activities among smallholder farmers in the study area.

Variable	Frequency	Percent (%)
Mixed farming	236	50.8
Crop production	67	14.4
Animal production	30	6.5
Sole cropping	20	4.3
Mixed cropping	111	23.9
Horticulture	1	0.2
Total	465	100

, which highlighted that the majority (89.1%) of the respondents confirmed that they are involved in various on-farm activities such as mixed farming and mixed cropping. This result agrees with Nyamayevu [48], who postulated that smallholder farmers diversify their on-farm activities to increase their income and resilience to changing climatic conditions.

4.3. Type of Farming Activities Among Smallholder Farmers in Mpumalanga Province

The most common form of on-farm diversification practice, as presented in Table 4, is mixed farming, with 50.8% of smallholder farmers practising it, followed by mixed cropping, which accounts for 23.9%, and crop production, making up 14.4% of activities practised by smallholder farmers in Mpumalanga Province. This result is in keeping with Franke et al. [49] and Nyamayevu et al. [48], who postulated that on-farm diversification such as crop rotation, mixed cropping and mixed farming is common among smallholder farmers. Nyamayevu et al. [48] further argues that on-farm diversification improves soil fertility, reduces weed and disease infestations as well as agrochemical usages, builds climate resilience, and improves crop yields. However, Franke et al. [49] and Yang et al. [50] argue that, while on-farm diversification enhanced soil health and fertility, crop rotation and intercropping must mostly be interchanged with leguminous crops such as beans, cowpeas, and clover.

The other farming activities reportedly used by smallholder farmers in the study area to promote diversification include animal production, crop production and horticulture, accounting for a total of 21.1%. This may be attributed to the fact that the contemporary literature [8,36] suggests that women mainly focus on off-farm livelihood diversification. The majority of the agripreneurs in the informal markets around townships are women. Moreover, farmers in Mpumalanga Province produce different types of crops and livestock. They produce both large and small livestock, ranging from poultry to cattle (both beef and dairy), goats, pigs, sheep and other types of birds [30], while crop production varies from vegetables to fruit and agronomic crops, namely tomatoes, cabbages, potatoes, maize, cowpeas, sweet potatoes, soya beans, okra, mangoes, avocados, litchis, and sugarcane, to mention a few [14,51,52].

The study findings reveal that farming practices, economic motivations, and structural determinants primarily shape on-farm diversification among smallholder farmers in the study area. Hence, the analysis of farming activities shows that mixed farming is the dominant diversification strategy. The pattern in the findings highlights the multifunctional nature of smallholder farming systems, since the integration of crops and livestock offers beneficial effects such as nutrient recycling, risk reduction, and enhanced income security [53]. Livestock can provide manure for soil fertility, while crop residues serve as feed, creating a self-sustaining loop [54]. The other most common strategy, mixed cropping, is an age-old practice that maximizes limited land resources, ensures continuous food availability, and mitigates risks associated with climate variability [50,55]. The relatively low percentage of farmers that engage solely in crop production or specialized enterprises like horticulture or animal production suggests that single-enterprise systems are perceived as less resilient in this context, particularly where resource constraints and market uncertainties are prevalent [56,57].

4.4. The Purpose of On-Farm Diversification Among Smallholder Farmers

The survey findings, as presented in

Table 5. The purpose of on-farm diversification among smallholder farmers in Mpumalanga Province.

Variable	Frequency	Percent (%)
Enhance farm income	144	31.0
Climate change adaptation	32	6.9
Improve food security	51	11.0
Sustainable food production	50	10.8
Escape poverty	86	18.5
Weed management	2	0.4
Pests and disease control	22	4.7
Soil health management	15	3.2
Insure against failure	1	0.2
Other	62	13.3
Total	465	100

, show that on-farm diversification in the study area mainly seeks to improve farm income, with 31.0% of the smallholder farmers surveyed choosing it as a viable rationale towards diversification, and escaping poverty (18.5%) was second behind improving farm income, indicating economic motivations as the dominant drivers of diversification. Other reasons for diversification as chosen by the surveyed smallholder farmers include improving food security (11.0%), sustainable food production (10.8%), and climate change adaptation (6.9%). Reasons that had less impact on the decision-making of smallholder farmers in the study area include concerns involving pests and disease control, soil health, and weed management, which make up 8.3% of the drivers of on-farm diversification. With this in view, one can argue that the focus of the 6.9% of respondents who identified climate change adaptation as the main reason for on-farm diversification corresponds to the level of education, which involves understanding the fact that the impacts of climate change include high levels of pest and disease infestation, soil degradation, and high weed infestation, leading to high production costs incurred by farmers.

The purpose of diversification, as presented in the results above, is strongly tied to economic and socioeconomic imperatives, as also supported by the findings of a study conducted by Vilakazi et al. [58]. The results reveal that increasing farm income and the desire to escape poverty are the principal drivers of diversification. This aligns with the broader reality of rural livelihoods, where farming is often the primary or only source of income, and diversification merely offers a pathway to reduce financial vulnerability [59,60]. For instance, a farmer growing vegetables and staple grains may sell surplus vegetables in local markets to generate cash while relying on grain production for household food security. Beyond economic goals, diversification as revealed in this study is also motivated by food security and sustainable food production, pointing to the dual function of farming as both a livelihood and a food provisioning system. This is supported by the earlier works by Lucantoni and Domarle [61] and Galanakis [62]. Nevertheless, while few farmers identified climate change adaptation and environmental factors like pest control, soil health, and weed management as key reasons for diversification, these aspects remain critical [63]. Thus, such benefits may be secondary or indirect outcomes rather than primary objectives.

4.5. Determinants of Implementing On-Farm Diversification Strategies Among Smallholder Farmers in the Study Area

The parameter estimates for the determinants of implementing on-farm diversification strategies contained in

Table 6. Parameter estimates of the determinants of the use of on-farm diversification strategies by smallholder farmers in the study area.

Parameter Estimates
The Smallholder Farmers’ Choice to Diversify On-Farm Activities	Β	Std. Error	Wald	df	Sig.	Exp (β)	95% Confidence Interval for Exp (β)
							Lower Bound	Upper Bound
Intercept	4.062	1.063	14.594	1	<0.001
Gender	−0.243	0.229	1.126	1	0.289	0.784	0.500	1.229
Age	−0.335	0.228	2.151	1	0.142	0.715	0.457	1.119
Marital status	−0.115	0.119	0.937	1	0.333	0.891	0.706	1.125
Level of education	−0.533	0.115	21.585	1	0.001 *	0.587	0.469	0.735
Household size	−0.015	0.179	0.007	1	0.933	0.985	0.694	1.398
Secondary source of income	−0.174	0.092	3.623	1	0.057 *	0.840	0.702	1.005
Farming experience	−0.017	0.126	0.017	1	0.895	0.984	0.769	1.259
Farmland size	0.328	0.143	5.238	1	0.022 *	1.388	1.048	1.839
Number of farm assistants	0.354	0.147	5.808	1	0.016 *	1.425	1.068	1.901
Diversification awareness	−1.218	0.415	8.604	1	0.003 *	0.296	0.131	0.668

* Significant variable.

show that the level of education (p = 0.001, β = −0.533) decreases the chance of smallholder farmers adopting diversification strategies. The negative coefficient on the level of education of smallholder farmers suggests that the more education a smallholder farmer attains, the less likely they are to prefer on-farm diversification, even though it is significant in the model. This can mean that higher levels of education among smallholder farmers can translate to their preference for strategies to offset risks and challenges other than on-farm diversification. This finding contradicts the studies reported by Tacconi et al. [59] and Sinyolo et al. [64], who postulated that the level of education positively influences farmers’ adoption of technology for poverty and food insecurity alleviation. Additionally, farmland size significantly and positively (p = 0.022, β = 0.328) increases the probability of smallholder farmers adopting on-farm diversification strategies when all variables in the model are kept constant [65]. However, Chmieliński et al. [65] state that the economic viability of a farm is the most important factor in driving farmers’ decisions on whether to adopt on-farm diversification strategies, rather than farmland size. Similarly, when smallholder farmers have farm assistants (p = 0.016, β = 0.354), they are more likely to adopt diversification strategies. Larger farm sizes require more labour, financing, and hands-on management of production processes; thus, diversifying the farming enterprise ensures that the farm can generate enough income to cover operational costs [48,66].

Diversification awareness is another significant predictor variable in improving the adoption of on-farm diversification strategies. However, greater awareness of the benefits of on-farm diversification (p = 0.003, β = −1.218) negatively influences the adoption of on-farm diversification strategies among the smallholder farmers surveyed in the study. The unanticipated result on awareness about on-farm diversification may suggest complexities in comprehending and/or constraints smallholder farmers face in diversifying their enterprises despite awareness of on-farm diversification and reaching higher levels of education [67]. This finding is contrary to the studies of Abera, Yirgu, and Uncha [68], Ragasa et al. [69], and Lokier, Morris, and Thomas [70]. However, other predictor variables contained in the model, such as gender, age, marital status, household size, and farming experience, had no significant influence on the smallholder farmers’ decision to adopt on-farm diversification strategies.

With this result in view, it can be noted that the determinants of on-farm diversification can be classified within the Sustainable Livelihood Approach (SLA). The SLA consists of five types of capital that aid in reducing farmers’ vulnerability by presenting livelihood strategies. These are human, financial, social, natural and physical capital. Age, level of education, farming experience (in years), household size as well as on-farm diversification awareness are classified as human capital. The number of farm assistants is classified as both physical and human capital. The farmland size is classified as natural capital, while the smallholder farmers’ secondary source of income is classified as financial capital, and the gender of smallholder farmers and their marital status is classified as social capital. The result, therefore, points to the fact that the smallholder farmers in Mpumalanga Province have the use of human (p = 0.003, β = −1.218; p = 0.001, β = −0.533), financial (p = 0.057, β = −0.174), natural (p = 0.022, β = 0.328), and social capital (p = 0.016, β = 0.354) to remain resilient to both financial-related and climate change (production) risks. This result is in keeping with Mudzielwana et al. [37], who posited that gender, which is representative of social capital, has a positive impact in the adoption of livelihood diversification of farm workers. Even though gender in this study is not statistically significant, as social capital it is a key factor in the SLA, especially, with the consideration that on-farm diversification has the potential to increase income and helps farmers to better manage production challenges as they present themselves.

4.6. The Model Diagnostic Testing

The model’s performance was evaluated by making use of multiple fit indices, as shown in

Table 7. Model fit test and statistics summary.

Statistic	Value	p-Value
Goodness-of-fit (Pearson)	χ2 = 428.895	0.020
Goodness-of-fit (deviance)	χ2 = 506.073	<0.001
Cox and Snell R2	0.138	-
Nagelkerke R2	0.188	-
McFeddan R2	0.112	-

below, including the Pearson goodness-of-fit test (χ² = 428.895 and p = 0.020) and the deviance chi-square test (χ² = 506.073 and p = <0.001). This test indicated that the model deviated from a perfect fit. As noted by McCullagh and Nelder [71], these measures are highly sensitive to the study’s sample size and lead to the rejection of useful models in complex and real-world datasets. However, given the sensitivity of these tests to the sample size, the pseudo R-square (R²) values were also checked to assess the model’s explanatory power. Of this test, the Cox and Snell R² was 0.138, the Nagelkerke R² was 0.188 and the McFadden R² was 0.112. These values suggest that the independent variables account for approximately 11.2% to 18.8% of the variance in the outcome variable, therefore representing a fair model for the context of the study and the large sample size adopted. Furthermore, pseudo R-square (R²) with values ranging from 0.1 to 0.2 is standard for real-world complex data, especially in social sciences, which makes the values obtained for this study acceptable and provides meaningful insights into predictors of human behaviour [72,73]. Of course, it is an undeniable fact that the Pearson and deviance tests are highly sensitive to large datasets, which is most probably the case in this finding.

4.7. Implications for Climate-Smart Farming and Structured Market Access Pathways

Mpumalanga’s diverse climatic zones, ranging from subtropical lowveld conditions to cooler highland areas [74], provide a strong foundation for bioclimatically adapted production systems that build on existing farming practices rather than replacing them. A strengthened climate-smart mixed farming model is particularly suitable because it integrates crops and livestock in a way that spreads risk and improves internal resource cycling. Combining drought-tolerant staples such as sorghum and cowpeas with vegetables and small livestock such as goats or poultry can enhance soil fertility through manure use, improve household nutrition, and create multiple income streams across seasons. Conservation agriculture practices such as minimum soil disturbance, mulching, crop rotation with legumes, and intercropping can be promoted through practical field demonstrations, allowing farmers to improve soil moisture retention and reduce input costs without investing in expensive machinery [75]. Simple water-smart interventions, including rainwater harvesting, gravity-fed drip irrigation, and improved scheduling of irrigation based on local rainfall patterns, are realistic for resource-constrained farmers and can significantly reduce vulnerability to dry spells [16]. Agroforestry systems that integrate fruit trees with annual crops can further moderate microclimates, protect soils from erosion, and generate higher-value produce suited to the province’s ecological conditions [76]. These models emphasize ecological efficiency and incremental improvement rather than capital-intensive transformation.

For smallholder farmers to access formal national and international markets, production models must be linked to structured aggregation, quality assurance, and value addition mechanisms that reduce individual risk. A cooperative cluster model can enable farmers to pool produce, standardize grading, and meet the volume and consistency requirements demanded by supermarkets and export buyers [77]. Extension support should focus on practical compliance measures such as record-keeping, safe agrochemical use, basic hygiene standards, and improved post-harvest handling, all of which are essential for formal market entry. Contract farming arrangements with established buyers can provide more predictable market access while incentivizing farmers to align production with quality specifications [78]. Incentives may include preferential input support tied to adherence to conservation and climate-smart practices, as well as targeted programmes that encourage youth participation in agripreneurship. Small-scale agro-processing initiatives such as fruit drying, vegetable dehydration, or simple livestock product processing can extend shelf life and create products that meet formal retail standards. When production resilience is combined with structured market linkages, smallholder farmers are better positioned to move gradually from informal sales toward stable participation in higher-value domestic and international value chains without incurring unsustainable financial burdens.

5. Conclusions and Recommendations

Smallholder farmers across the world have used on-farm diversification as one of the key strategies to maximize profits, build resilience, increase productivity and sustainability, mitigate farm financial-related risks, and maintain good soil health. Farmers use various on-farm strategies to diversify their on-farm activities, including crop rotation, mixed farming, mixed cropping, and intercropping, to mention a few. Whether these positively or negatively affect farmers depends on several factors such as farmers’ choice to diversify or not to, environmental contexts, market dynamics, institutional and policy enablement, culture and political contexts, as well as gender, which has over the years proven to be a factor. In Mpumalanga Province, the on-farm diversification strategies were indirectly captured by several studies; however, they were limited in terms of not providing answers to the question of the determinants of on-farm diversification among smallholder farmers. Hence, this study examined the determinants of on-farm diversification strategies among smallholder farmers registered with the Department of Agriculture, Rural Development, Land and Environmental Affairs.

The study employed a quantitative research design, randomly sampling farmers to participate in the study conducted between 2023 and 2024, with data collected primarily using a structured questionnaire distributed to farmers with the aid of trained enumerators. The collected data were analysed using descriptive and inferential statistics, employing tables to present the data for ease of interpretation. The findings of the study reveal that the majority (51.8%) of the smallholder farmers are female farmers, which is a huge achievement in the women empowerment agenda. At least 59.1% are within the age group of 36–60 years, which is key for unemployment reduction; however, youth (age group 18–35 years = 20.2%) participation in agriculture is still very low. The findings further show 49.0% of the farmers have only up to secondary education, which will be a challenge in the future in terms of technology adoption rates, because, when farmers adopt technologies, they must be able to understand and make meaningful use of those technologies and digital platforms for their benefit.

At least 36.1% of the farmers depend primarily on agricultural activities for income with no other source of income. It is assumed that these are males between the ages of 36 and 60, given the difficulty of gaining access to government employment opportunities once one is above the age of 35 in South Africa. A significant number of farmers (36.1%) have 5–10 years of farming experience, which may be a critical reason why some farmers opt not to diversify their on-farm activities. Farming experience plays a huge role in both the adoption of new technologies useful in the sector but also in informing farmers’ choices to diversify their farming activities. Results further show that 50.8% of the farmers practise mixed farming, which involves both livestock and crop production, while 23.9% practise mixed cropping. The use of mixed crops has been seen over the years as not only a way to increase farm income, but also as means to manage soil health and prevent frequent infestations of pests, diseases and weeds. Of course, 3.2% of the farmers confirmed that they diversify their on-farm activities to manage soil health and fertility, while 31.0% diversify to improve income and 18.5% to escape poverty.

On-farm diversification was found to be influenced either positively or negatively by the farmer’s level of education (p = 0.001, β = −0.533), secondary source of income (p = 0.057, β = −0.174), farmland size (p = 0.022, β = 0.328), the number of farming assistants (p = 0.016, β = 0.354) and the farmer’s awareness of on-farm diversification (p = 0.003, β = −1.218). Therefore, this study concludes that smallholder farmers diversify on-farm activities due to pressing need to increase their income, escape poverty, ensure household food security and be climate-change-adaptable. These are influenced by the farmers’ level of education, projecting the understanding of concepts and processes of climate change, and income maximization for sustainable farm businesses. In order to increase smallholder farmers’ market participation to increase their income probabilities, it is recommended that self-help groups be established. The SHGs would raise awareness among their membership of the market requirements and the use of the cost sharing model (CSM). Cost sharing among smallholder farmers can increase income while minimizing production costs. This model would achieve more than the currently existing farmer field schools and farmers’ study groups. Moreover, future studies should investigate the effects of on-farm diversification strategies on the smallholder farmers’ transition to agripreneurship for poverty alleviation, food security, and rural economic development (RED). This will enable deeper understanding of the need for higher income among smallholder farmers, which may then lead to RED and smallholder farmers’ financial independence rather than being dependent on government funding.

Future studies could also investigate the impact of both on-farm and off-farm diversification on the involvement of young people in agriculture or agripreneurship. This will provide an in-depth understanding of the needs of the youth as they venture into farming business operations against the current reports of youth resistance to practise agriculture. Youth involvement in the sector will not only aid farmers’ access to funding but also has the potential of documenting indigenous knowledge, ensuring continuity of the sector with the exit of old farmers as well as creating employment opportunities for both unemployed young people and women. The agricultural policy implications are, to a certain extent, that a deep review into some agricultural policies such Agricultural Development, the Agricultural Products Standards Act of 1990, and the Conservation of Agricultural Resources Act 43 of 1983 is necessary in order to support and promote on-farm diversification among smallholder farmers. Furthermore, policy makers need to look into developing policies to promote young people in agriculture in order to improve the number of farmers adopting on-farm diversification, especially because Mpumalanga Province has a large number of unemployed graduates. The introduction of policies promoting youth involvement and adequate young and emerging farmer support programmes or projects will motivate youth involvement in the sector and create more employment opportunities to curb the growing rates of unemployment in the country, and Africa as a whole.