https://orcid.org/0000-0002-7504-6703
Figures
Abstract
Climate-smart agriculture (CSA) addresses food security issues under climate change. The research examined the effect of adopting CSA practices on food and nutrition security by small-scale urban crop (SSUC) farmers in the eThekwini (ETH) Municipalityusing purposive sampling from 412 SSUC farmers. Results suggest that socio-demographic and institutional factors influence household consumption patterns and dietary status of SSUC farmers. The probit selection model show that the farmer’s age, education, household size, off-farm income, monthly expenditure on food, agricultural training, group membership, and credit access significantly influenced CSA practices adoption decisions. The endogenous switching regression using marginal treatment effects shows that farm income, off-farm income, monthly expenditure on food, group membership, hired labour and distance to the farming site significantly affected household food consumption patterns. Gender, marital status, employment status, age, household size, farm and off-farm income, monthly expenditure on food, group membership, hired labour and number of part-time labourers from households significantly influenced the household dietary diversity status of SSUC farmers. The findings confirm heterogeneity in the effects of adopting CSA practices. Unobserved benefits are prevalent through a positive selection of CSA practices depicted by the Household Food Consumption Score (HFCS) and Household Dietary Diversity Score (HDDS). Adopting CSA practices enhanced the food and nutrition of SSCU farmers, shown by the average treatment effects (ATT) when farmers adopt CSA practices. Adopting CSA practices correlated positively with the food and nutrition security of SSUC farmers, with adopters being 16 and 31 percent more food secure concerning HFCS and HDDS, respectively. Hence, SSUC farmers in ETH Municipality adopting CSA practices were likely better off regarding food consumption patterns and dietary diversity. In light of this, a nexus between SSUC farmers, researchers, and extension services must consider suitable sets of CSA practices of relevant scale chosen and directed toward the welfare of localised contexts.
Citation: Khumalo NZ, Sibanda M, Mdoda L (2025) The effect of heterogeneous adoption of climate-smart agriculture practices on household food and nutrition security of small-scale urban crop farmers in eThekwini Municipality. PLOS Clim 4(1): e0000551. https://doi.org/10.1371/journal.pclm.0000551
Editor: Jamie Males, PLOS Climate, UNITED KINGDOM OF GREAT BRITAIN AND NORTHERN IRELAND
Received: May 14, 2024; Accepted: December 4, 2024; Published: January 10, 2025
Copyright: © 2025 Khumalo et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The research collected data from small-scale urban farmers in EThekwini Municipality. The respondents signed an informed consent and approval to research was granted by eThekwini Municipal Institute of Learning (MILE). The Humanities and Social Sciences Research Ethics Committee (HSSREC) through the University of KwaZulu Natal (UKZN) issued ethical clearance, protocol number: HSSREC/00005367/2023. Per the ethical requirements, data will be stored in the Discipline of Agricultural Economics, School of Agricultural, Earth and Environmental Sciences of at the University of KwaZulu Natal for five (5) years and can be made available through the Research Ethics Committee if needed. The research committee contacts are Telephone: +27(0)31 260 8350/4557/3587; Email: hssrec@ukzn.ac.za; Website: http://research.ukzn.ac.za/Research-Ethics.
Funding: The National Research Foundation (NRF) grant number: NGAP23030380666 provided funding support for this research as a PhD study.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
The issue of food insecurity has garnered significant global attention due to its prevalence in numerous nations worldwide [1]. Natural disasters, climate change, COVID-19 aftermath, economic shocks, conflict, and sharp price increases have substantially impacted global food security and nutrition, disproportionately harming the world’s poorest populations. In 2022, approximately 691 to 783 million people worldwide were affected by hunger, translating to an increase of 122 million [2]. Africa has seen a rise in hunger, with 20 percent of the continent facing chronic hunger in 2021 [3]. Around 61 percent of people in Africa and over 29.6 percent of the world’s population, or more than one-fourth, suffered moderate to severe food insecurity in 2022 [2]. Like many other countries, South Africa is not an exemption. Food availability primarily depends on how well the agriculture sector performs and a nation’s ability to process, import, store, and distribute food [4].
Unlike most other African countries, SA is the second-largest agricultural landholding nation. Nearly 80 percent of the nation’s total land area, or over 96.3 million hectares, was used for agriculture in 2021 [5]. As of 2020, about 87 percent of SA’s land was pastures and meadows, and approximately 12.5 percent was arable [6], which could affect food security. Research indicates that the number of South Africans that are food insecure increased from 17.8 percent (2019) to 20.9 percent (2021) [7]. In 2021, 12.2 percent of people were hungry. In 2021, 17.2 percent of South African households were involved in activities related to agricultural production, with 85 percent of them doing so only to guarantee a surplus of food [7]. There has been a 1.7 percent decrease in crop production volume for the 2021/22 season [8]. Reduced production of winter crops like barley and summer crops like sorghum and maize is the leading cause of this decline. The total value added in 2021 came from agriculture in just 2.3 percent of cases [8]. Due to its primary focus on providing services, the South African economy had one of the lowest shares in Africa.
Although food insecurity in urban areas is increasing due to urbanisation, urban agriculture (UA), unlike conventional agriculture, has the potential for urban households to have enough food to meet their needs. Urban agriculture emerges as a vital solution to the challenges of increasing urbanization, offering a means to enhance food production and sustain local economies [9]. It adapts to urban spaces through innovative practices like rooftop and community gardens, urban greenhouses, and vertical farming, addressing spatial limitations while enriching biodiversity and improving access to fresh food [10]. Urban agriculture is crucial in waste management through composting and mitigates urban heat island effects by promoting greening efforts [11]. Beyond environmental benefits, UA strengthens community ties by uniting individuals to pursue food sustainability and resilience against climate change. It introduces cost-effective, sustainable farming techniques, including aquaponics and hydroponics, which increase production and conserve resources [12]. Furthermore, by incorporating renewable energy, UA reduces the carbon footprint of urban food production. In developing regions, it is especially critical for enhancing food security and reducing poverty, as it lowers food transportation costs and combats food insecurity by providing affordable, nutritious options [13]. In addition, UA protects against global market fluctuations and supply chain disruptions for localised food systems, bolstering urban climate change resilience [14].
The Intergovernmental Panel on Climate Change (IPCC) [15] and the World Bank [16] predicted that sub-Saharan Africa (SSA) would be highly susceptible to climate change. According to the IPCC [17], in SSA, productivity in agriculture will decline by about 11 percent by 2080. Masipa [18] suggests that around 66 percent of Africa’s cultivable land will be gone by 2025 due to insufficient rainfall and drought. Research indicates climate change profoundly influences agricultural land, directly impacting food and nutrition security [19, 20]. The IPCC [17] demonstrates that climate change effects are more prone to Southern Africa and that there would likely be serious consequences that will significantly affect livelihoods. The analysis predicts a rise in temperature and erratic rainfall trends that would lead to a significant decline in stable food production, potentially reaching a reduction of up to 50 percent. South Africa is experiencing droughts, leading to insufficient precipitation, rivers drying up, substantial livestock loss, and a considerable decline in farmers’ production [21]. Due to less rainfall and prolonged drought, SA has been facing heightened levels of food insecurity. Without action, fertile land suitable for agriculture will decrease due to global warming. Therefore, the implementation of CSA practices is paramount.
Given the grave impacts of global climate-related challenges, CSA practices are a promising approach to mitigate agricultural consequences, thereby enhancing food-resilient communities [22]. Implementing CSA practices mitigates greenhouse gas emissions and supports overall food security initiatives while increasing productivity and strengthening household resilience under climate-related stresses [23]. Research has demonstrated that adopting CSA practices can enhance crop productivity, guaranteeing food production [24, 25]. Because CSA practices incorporate cost-effective, farm-centred, sustainable land management practices like agroforestry, effective water use, residue management, and minimum tillage, urban farmers in SA must implement and use them. Furthermore, most CSA practices incorporate conventional knowledge and practices that farmers are familiar with and use for climate change adaptation [26, 27].
Food systems are multisectoral and complex, particularly UA, given the changing climate and climate variability that are more prevalent in SSA. Therefore, the uptake of CSA practices is an appropriate pathway to address sustainable food production in urban areas [28, 29]. Unless governments and municipalities carefully consider urban planning and UA, misallocating human and financial resources in urban areas will continue to yield food and nutrition insecurity resulting from global warnings. Climate-smart agriculture emerges as a solution for a "win-win" outcome as it incorporates adaptation and mitigation necessary to reorient climate policy response to food and nutrition challenges.
Climate-smart agriculture adoption remains relatively low in under-developed countries [30, 31] owing to developmental stresses that include lack of access to agricultural technologies [32], limited research and prioritisation of CSA practices [33], lack of knowledge, the negative perception around new farming practices and sticking to tradition, institutional shortfalls, low human capital, resource constraints, and physical infrastructural barriers [34]. Furthermore, social networks, learning, and environmental consciousness determine decisions to adopt CSA practices [35, 36]. Small-scale urban crop (SSUC) farmers also contend with numerous challenges, such as land availability, lack of capital, and urban pollution [37, 38], likely to impede CSA practices’ adoption and uptake.
Applying CSA practices can enhance urban agricultural foods’ ability to tackle the difficulties associated with global warming. However, in many developing countries, like South Africa, the prospects of adapting to climate variability are complicated by weak institutions and insufficient technology. In light of the significance of peri-urban and UA in supplying ecological and human needs, enacting legislation to guarantee the long-term viability of specific agricultural practices is imperative [37, 39]. Small-scale urban farmers increasingly face challenges concerning household food and nutrition security due to inadequate access to resilient farming practices in the face of climate change and variability. While CSA practices emerge as a global framework to enhance productivity and climate resilience, its adoption among small-scale urban farmers remains heterogeneous. The heterogeneous adoption of CSA practices impacts household food and nutrition security in varying ways, which is still poorly understood for small-scale urban farmers. Again, a gap remains in understanding the decision-making processes of farmers, the factors influencing farmers’ CSA practice choices, and how the adoption of CSA practices affects food and nutrition security. Assessing how CSA practice adoption influences food and nutrition security among small-scale urban farmers is essential to address this knowledge gap and for a nuanced understanding of the role of CSA practice adoption in UA on food and nutrition security and informing targeted interventions. Therefore, it is imperative to comprehend farmers’ CSA practices adoption decisions and their implications on food and nutrition security. This study aims to provide policymakers and development partners insights into the complex dynamics that influence the adoption of CSA practices within SSUC farming systems, which are critical to enhancing sustainable agricultural practices.
This paper, therefore, assesses the heterogeneous outcome (adoption of CSA practices) on food security (including nutrition) by SSUC farmers in the eThekwini (ETH) Municipality. The study 1) determines the socio-demographic and institutional factors of SSUC farmers in ETH Municipality influencing CSA practices adoption and 2) assesses the effect of adopting CSA practices on the food and nutrition security of SSUC farmers in ETH Municipality.
2. Materials and methods
2.1 Ethics statement
Before the study was conducted, it received ethical approval from the Humanities and Social Sciences Research Ethics Committee (HSSREC) through the University of KwaZulu Natal (UKZN), protocol number HSSREC/00005367/2023. The HSSREC reviewed the research protocol and questionnaire tool for its ethical fitness and to ensure that the respondents in the study would be protected against physical and emotional risks or harm. The study collected data from human subjects (SSUC farmers in the ETH Municipality). Therefore, an informed consent form was signed by each respondent willingly, and they were assured that they could discontinue the study at any point if they wished to, without any repercussions. Also, the respondents were assured of anonymity and confidentiality. Additionally, the researcher obtained a gatekeeper’s letter from the eThekwini Municipal Institute of Learning (MILE) before commencing data collection.
2.2 Model specifications
This research adds to existing empirical research on how adopting CSA practices improves food and nutrition security. The main reason for concern with CSA practices is that they impact food production [40]. This paper assesses whether CSA practices enhance food and nutrition security. Previous empirical studies have used analytical techniques like endogenous switching regression or propensity score matching to investigate the effect of adopting agricultural technologies on food security. These studies include Assaye et al. [41], Zegeye et al. [42], Sisang and Lee [43] and Teklu et al. [44]. However, the effect of treatment differs among individuals in a population based on unobserved attributes [45]. Marginal treatment effects (MTEs) pertain to the Average treatment effects (ATEs) observed in individuals who either exhibit a specific resistance to treatment (such as adopting CSA practices) or fall within a particular range of indifference. Therefore, we utilised the MTEs technique in this study to address differential treatment effects from observed and unobserved factors [46]. Marginal treatment effects explain the heterogeneous treatment effects based on the unobserved trait of resistance to treatment [47]. This heterogeneity can result in inconsistency when assessing the estimated adoption impacts.
2.3 Probit selection model specification
A farmer either adopts CSA practices or does not. This study hinges on utility theory as a foundation for this research—modelling the adoption of CSA practices as a random utility function. If CSA practices adoption gains outweigh those of not adopting, the farmer chooses to do so. The expectation is that the adoption decision will influence farmers’ well-being concerning the food and nutrition security status of SSUC farmers. The contextual identity framework captures the socio-demographic and institutional factors influencing a farmer when determining whether or not to adopt CSA practices. Eq 1 represents utility as a function of unobservable components:
(1)
Where Ki is a vector of the farmer and farm’s characteristics (gender, male, female, marital status, employment status, principal economic activity, age, education, household size, farm income, off-farm income, monthly expenditure on food and non-food items, agricultural training received, land ownership, group membership, extension visits, credit access, hired labour, irrigation technology access, number of part-time labourers from household, farming experience and distance to farming site). α is the parameter to be estimated, and j depicts mean derivations, expected to be zero and have a constant variance. The subscript j stands for adoption, where it is equal to 1 (j = 1) will be an adopter, and where it is zero (0) (j = 0), a non-adopter. Farmers with higher expected gains (Qi1) will adopt CSA practices in comparison to non-adopters (Qi0). If is the anticipated net gains from adopting, it is expressed as:
(2)
and 0 if not adopting (otherwise)
Where the latent variable denotes unobserved attributes, while CSAi is the binary dependent variable. An SSUC farmer who adopts CSA practices was coded with a numeric value of 1 and 0 otherwise, where
depicts the latent variable capturing anticipated adoption benefits, Ci are the instruments to detect the model and Hi unobserved heterogeneity for CSA practices adoption propensity. Therefore, adopting CSA practices is non-random and endogenous, with farmers’ selection driven by observable and unobservable attributes. The latent variable CSAi is difficult to predict but can be treated as a function of observable traits.
2.3.1 A description of the explanatory variables inputted in the probit selection model, their measurement and expected outcomes.
The probit selection model was applied to determine factors influencing the adoption of CSA practices by SSUC farmers. The probit selection model is a binary outcome, representing a function of several independent variables, including the socio-demographic and farm characteristics of the SSUC farmers influencing CSA practice adoption decisions. Table 1 describes the explanatory variables inputted in the probit selection model and their measurement and expected outcomes. The gender variable
is binary (males are assigned a numeric value of 1 and females 0). The influence of gender on CSA practice adoption is not straightforward. Therefore, the study expected both males and females to have a higher or lower CSA adoption rate depending on cultural norms in decision-making. Marital status is binary (with wedded SSUC farmers assigned a numeric value of 1 and, for those that are single, 0). The study hypothesised that wedded SSUC farmers are more likely to adopt CSA practices due to stability in households with married couples. Employment status indicates whether the SSUC farmer was formally employed or unemployed, with formal employment coded with a numeric value of 1 and 0 otherwise. The expectation is that formally employed farmers may invest in CSA practices due to access to disposable off-farm income from formal employment. The principal economic activity assessed whether farming or formal employment was the principal livelihood activity for the SSUC farmers. If farming was the primary source of income, the researchers assigned it a numeric value of 1 and 0 for formal employment. The study hypothesised that SSUC farmers with formal employment as a primary source of income may be less inclined to invest in CSA practices than those that solely depend on farming. The age of the SSUC farmer is a continuous variable measured in years. While older farmers might be less adaptive to CSA practices, on the other hand, old age comes with experience, and they might have more resources to adopt and implement CSA practices than their younger counterparts. The study measured education as a continuous variable in terms of the schooling years. Education equips SSUC farmers with the knowledge and skills to comprehend new innovative technologies such as CSA practices. Therefore, more education is likely positively associated with an increased adoption rate of CSA practices due to better awareness of its benefits. Household size reflects the number of household members and potential labour availability to implement labour-intensive CSA practices. Therefore, larger households might encourage SSUC farmers to adopt CSA practices to address their food demands. Annual farm and off-farm income represent the total disposable household income measured as a continuous variable in South African Rands (ZAR). A higher farm and off-farm income represent a financial muscle to invest in CSA practices depending on economic diversity. Increased monthly expenditure on food and non-food items measured in ZAR could drive or encourage the adoption of more CSA practices. For instance, higher food costs may encourage SSUC farmers to adopt CSA practices to meet food security needs. Similarly, increased expenditure on non-food items potentially reflects the financial capacity to invest in CSA practices. Agricultural training capacitates SSUC farmers with skills and keeps them up to date with farming trends and innovations. The study assigned the numeric value 1 to an SSUC farmer who received agricultural training and 0 otherwise. Trained SSUC farmers are likelier to adopt CSA practices because they are technically adept. Land ownership is a binary variable assigned a numeric value of 1 if the SSUC farmer owned the land and 0 otherwise. Land ownership provides the security necessary for long-term investment in CSA practices. Membership in an agricultural group represents collective action by SSUC farmers. The study assigned the numeric value 1 to SSUC farmers who subscribed or belonged to an agricultural-related group and 0 otherwise. Group membership can facilitate knowledge-sharing likely to sustain the momentum to adopt CSA practices. Extension services are critical for farmer support and as a broker to numerous critical institutions. The study assigned a numeric value of 1 to SSUC farmers who received more frequent extension visits and 0 to those who never had extension visits. Frequent extension visits are likely to increase the adoption of CSA practices through support and advice received. Access to credit is critical for acquiring resources and maintaining farm operations. The study assigned a numeric value of 1 to SSUC farmers with access to credit and 0 otherwise. Therefore, credit access allows for capital investment, which is necessary to implement intensive CSA practices. Hired labour is integral in CSA practices to sustainably support and maintain labour-intensive practices. The hired labour variable assessed whether SSUC farmers hired labour for their UA activities, where yes was coded with a numeric value of 1 and 0 otherwise. Use of hired labour positively influences the adoption of CSA practices. Irrigation infrastructure ensures a reliable water supply for UA activities, supports water-intensive crops and enhances crop resilience, thus aiding SSUC farmers in adapting to drought climate conditions. The study assigned an SSUC farmer with access to irrigation infrastructure with a numeric value of 1 and 0 otherwise. Therefore, irrigation access enhances the feasibility of implementing CSA practices. Part-time household labourers provide low-cost labour for implementing CSA practice tasks. The higher the number of part-time labourers in the household, the more flexible it supports CSA practices adoption. Farming experience equips SSUC farmers with the knowledge critical to decision-making concerning adaptive practices such as CSA. Therefore, SSUC farmers with higher farming experience could effectively manage climate-related risks by adopting CSA practices. The distance to an urban farm site determines the travel time and effective maintenance and management of resources and adaptive practices. The study measured the distance to the farming site in kilometres. Therefore, longer distances might deter CSA practice adoption due to accessibility issues impacting sustainable urban food production and resilience.
2.4 Endogenous switching regression model using marginal treatment effects specification
We hypothesised that adopting CSA practices will affect the food and nutrition status of SSUC farmers in ETH Municipality. Hence, the expectation is that adopters will have improved food and nutrition status compared to non-adopters. Considering the outcome variable, HFCS and HDDS, as a linear function of adopting CSA practices (treatment variable) influenced by predictor variables X, the linear equation expresses this function as follows (Eq 3):
(3)
Where Qi indicates a measure of food and nutrition security, Xi are predictor variables, CSAi denotes CSA practices adoption, ℘ and λ are parameter vectors and εi an error term. The parameter λ estimates the influence of adopting CSA practices on the SSUC farmers’ performance, given a random adoption and non-adoption of CSA practices [48]. However, the decision to adopt CSA practices mainly depends on unobservable traits like inherent abilities, motivation and risk attributes. Therefore, adoption may not be random because observed and unobserved traits can correlate with the error term. The situation could lead to a skewed bias if only urban farmers with higher managerial skills or the educated adopt CSA practices, giving them an unequal advantage over their counterparts. This scenario hints that the adoption of CSA practices is potentially endogenous. However, there is strong evidence to support the notion that CSA practices adoption is likely attributed to unobservable characteristics such as entrepreneurship, risk aversion, technical proficiency, and social networking skills, likely to influence the outcome variable, in this case, food consumption and nutrition [49]. Therefore, the impact on the outcome variable’s distribution between the treatment (CSA practices adopters) and untreated (CSA practices non-adopters) may reflect the treatment’s effects and variations brought about by the selection process [50, 51]. Based on Eq 3, the MTEs take a quantile function or the effect of treatment at a specific value of error term [52, 53]. In estimating the marginal treatment model, this study follows the mathematical model employed by Shahzad and Abdulai [54].
2.4.1 A description of the measurement of explanatory variables inputted in the endogenous switching regression model using marginal treatment effects and their expected outcomes.
The study relied on the endogenous switching regression model to evaluate the effect of CSA practice adoption on household food and nutrition security, as already described in the previous section. The endogenous switching regression model was ideal for controlling selection bias in assessing treatment effects and estimating the marginal treatment effects of CSA practices adoption on two outcome variables, household food consumption proxied by the HFCS and household dietary diversity proxied by the HDDS using the same set of explanatory variables as in the probit selection model. The influence of the independent explanatory variables, their measurement, and expected outcomes on the HFCS and HDDS resemble similar outcomes presented in Table 1 based on literature and logical deductions.
2.5 Description and selection of eThekwini Metropolitan Municipality
The study was based in the ETH Metropolitan Municipality in the province of KwaZulu-Natal (KZN) of SA. Johannesburg and Cape Town are the largest metropolitan municipalities in SA, with the ETH Municipality coming third. The coordinates 29.8120° S and 30.8039° E depict the location of ETH Municipality. The ILembe District Municipality borders the ETH Municipality to the north, the uGu District Municipality (south), the uMgungundlovu District Municipality (west) and the Indian Ocean (east) [55]. EThekwini Municipality covers a total area of 2 556 km2 [56]. The largest city in the province is Durban, located within the ETH Municipality [57].
KwaZulu-Natal has a vibrant and diverse agriculture sector in SA. The KZN Province contributes significantly to SA agriculture, including sugarcane, fruits, vegetables, and livestock (dairy and poultry), which are dominant agricultural enterprises [58]. The KZN province is a top producer of sugarcane, accounting for 81 percent of national production [59]. In this study, the KZN Province was first selected purposively based on its agricultural diversity, vibrancy, climate challenges and significance to the South African agricultural sector [60]. Again, the province boasts prominent UA activities and a substantial proportion of SSUC farmers compared to other provinces [61].
Secondly, the researchers selected ETH Municipality from the KZN Province. EThekwini Municipality shows considerable potential among the cities practicing UA in KZN [62]. EThekwini Municipality has six UA hubs: Inchanga, Mariannhill Monastery, Newlands-Mashu Permaculture Centre, Northdene Agroecology Research and Development Centre, Scorpio Place in Mariannridge and Umbumbulu [63]. Besides, ETH Municipality is the economic hub of KZN, yet many people live in abject poverty [64]. Given that it is the economic hub of KZN, it attracts rural-urban migration, exacerbating food and income insecurity issues. For this reason, ETH Municipality was the choice for this study. The ETH Municipality has four agroecological zones (East, Central, South and North) [65]. The current research conveniently chose one ward based on UA activities from the four agroecological zones. These wards are Tongaat Ward 62 in the East agroecological zone. The Central, North, and South agroecological zones included Cato Manor Ward 29, Waterfall Ward 9 and Umbumbulu Ward 109.
The ETH Municipality is a topographically hilly area with many gorges and ravines and does not have a typical coastal plain. A subtropical climate (hot-humid summers and warm-dry winters) characterises ETH Municipality, with temperatures ranging from 16 to 25°C and 23 to 33°C in winter and summer, respectively. The municipality has an annual rainfall of 893mm. The ETH Municipality’s climatic conditions are conducive to agricultural activities. However, as noted by Olanrewaju and Reddy [66], overreliance on rainfall and land degradation suggests exposure to and exacerbated climate change reparations, likely to influence food production negatively and, in turn, urban farmers’ ability to be food and income-secure. In 2019, ETH Municipality had a population of approximately 3 987 648, which has risen by 1.6 and 1.2 percent in 2009 and 2011, respectively [67]. The central and northern parts of ETH Municipality are highly populated, and projections are that the metro’s population will increase to 4 164 503 by 2024 [67], further putting pressure on food demand by the metro’s growing population.
2.6 Sampling procedure and sample size determination
In research, a subset representing a larger population is a sample to be studied [68]. A good sample size makes the research manageable regarding time and resources. Following the multistage sampling, the third phase purposively selected Tongaat Ward 62, Cato Manor Ward 29, Molweni Ward 9, and Umbumbulu Ward 109. The choice of the wards lies in the predominance of SSUC farmers and the climatic conditions of the areas. The last phase of the multistage process involved the selection of the SSUC farmers. The researchers used purposive and snowballing (referral) sampling to select SSUC farmers. The benefits of purposive sampling include time and cost-effectiveness [69, 70]. For a sample to be of a representative size, the researchers employed Cochran’s sampling technique [71] described in Eq 4:
(4)
Where:
n∞ is the sample size
z2 is the standard error associated level of confidence
p is the variability or standard deviation
e is the ideal level of precision (error margin)
p is the estimated population size
Using Cochran’s sampling technique (confidence level of 95%, confidence interval of 5), the required sample size was 384 SSUC farmers (Eq 5).
Based on the required sample size calculation of 384 from Cochran’s sampling technique, the area sampling from the multistage sampling technique implied purposively picking 96 SSUC farmers from each ward. Nonetheless, the researchers were able to collect data as follows: Tongaat (110), Cator Manor (101), Molweni (103) and Umbumbulu (98), totaling 412 SSUC farmers. The researchers included an extra 28 SSUC farmers in the survey as they did not want to be left out, and the researchers could not exclude them. Sampling beyond the required sample size instead of the calculated sample allowed for a more comprehensive dataset, thus enhancing the study’s statistical power and reliability. Again, accommodating additional SSUC farmers eager to participate in the survey fostered goodwill, potentially reflecting a richer dataset from a broader range of respondents and the depth of the research findings.
2.7 Data collection
The study collected data from SSUC farmers through a structured survey method. A structured survey method was ideal for this research as it tests specific phenomena and relational hypotheses using established frameworks. The researchers designed the study’s questions to be closed- and open-ended to allow extensive information gathering and analysis through descriptive and inferential techniques. The questionnaire was translated into the native and most spoken language in the study areas–isiZulu, thus ensuring clarity and comprehension. Before the study commenced, the researchers conducted a pilot test with 10 percent (40 SSUC farmers in ETH Municipality), which did not form part of the analysis to ensure the questionnaire was effective. The scope of the data collection tool included the socio-economic and farm characteristics of the SSUC farmers, insights into CSA practices, and questions addressing the food security status. The researcher utilised trained enumerators to ensure effectiveness in the data collection process. The trained enumerators administered the questionnaire to minimise potential bias. Data collection commenced from 29 May to 26 June 2023, avoiding schedules conflicting with local events such as funerals, weddings, and significant social gatherings in the study areas. Avoiding social gatherings was done to minimise bias on food consumption patterns that fall outside the norm, thus enhancing the accuracy of the data.
3. Results
3.1 Descriptive results of the demographic and farm characteristics of small-scale urban crop farmers in eThekwini Municipality
Table 2 presents the t-test analysis between CSA practice adopters and non-adopters. Table 2 shows that the statistically significant variables (p < 0.05) are marital status, employment status, principal economic activity, age, education, household size, farm income, monthly expenditure on food items, monthly expenditure on non-food items, agricultural training, land ownership, membership in an agricultural-related group, frequency of extension visits, access to credit, hired labour, number of part-time household labourers, farming experience, distance to the farming site, food frequency and nutritional value (HFCS), and dietary diversity (HDDS).
The marital status mean scores of 1.820 and 1.463 among the adopters and non-adopters of CSA practices, respectively, suggest that most adopters were wedded. The t-test p-value of 0.001 indicates a statistically significant difference in marital status between adopters and non-adopters of CSA practices. This finding suggests that married (wedded) SSUC farmers can be positively associated with adopting CSA practices as they might have more household support, shared resources, or motivation to adopt CSA practices than single SSUC farmers, thus likely to be food and nutrition-secure.
Employment status by SSUC farmers plays a critical role in the adoption of CSA practice. The expectation is that SSUC farmers with more stable employment will likely adopt CSA practices. Climate-smart agriculture practice adopters had a slightly higher mean score of 1.972 than non-adopters (1.886). The findings suggest that adopters of CSA practices had a more stable employment status. The t-test p-value of 0.001 indicates a statistically significant difference in employment status between CSA adopters and non-adopters. This finding suggests that employment status is positively associated with adopting CSA practices. Individuals with stable employment may have more financial resources, stability, or support, making adopting new practices like CSA more feasible, which enhances food and nutrition security.
The principal economic activity of the SSUC farmers shows a slightly higher mean score of 1.134 among CSA practices non-adopters than CSA practices adopters (1.047), suggesting a slightly higher engagement in non-agricultural activities among non-adopters. The t-test p-value of 0.002 shows a statistically significant difference in the principal economic activity between CSA adopters and non-adopters. This finding implies that SSUC farmers whose principal economic activity is agriculture are more likely to invest in CSA practices, enhancing their productivity and, thus, food and nutrition security.
The mean age of SSUC farmers is slightly higher among CSA practices non-adopters (57.065) than CSA practices adopters (52.265). The findings suggest that non-adopters of CSA practices were older. The t-test p-value of 0.001 shows a statistically significant difference in age between CSA practice adopters and non-adopters. This finding implies that younger SSUC farmers are more likely to adopt CSA practices, potentially enhancing their productivity and food and nutrition security.
The SSUC farmers’ education (schooling years) shows a higher mean score of 10.473 among CSA practices non-adopters than CSA practices adopters (6.090). The findings suggest that non-adopters had more formal schooling years. The t-test p-value of 0.001 shows a statistically significant difference in educational attainment between CSA practice adopters and non-adopters. This finding suggests that SSUC farmers with fewer years of formal schooling were likelier to adopt CSA practices, potentially improving their productivity and enhancing food and nutrition security. On the other hand, SSUC farmers with more schooling years could be exposed to alternative income sources or inclined to rely on non-agricultural livelihoods, reducing their reliance on innovations like CSA practices.
The household size of SSUC farmers has a higher mean score of 9.948 among CSA practice adopters than non-adopters (6.547). This finding suggests that CSA practice adopters tend to have larger households. The t-test p-value of 0.001 shows a statistically significant difference in household size between CSA practice adopters and non-adopters. This finding confirms that larger households could be more inclined to adopt CSA practices due to the increased availability of family labour, which can support labour-intensive CSA practices, ultimately enhancing food and nutrition security.
The yearly farm income by SSUC farmers shows a higher mean score of ZAR 26 348.56 among CSA practice adopters than non-adopters (ZAR 14 137.75). This finding indicates that CSA practice adopters tend to have higher farm income. The t-test p-value of 0.001 shows a statistically significant difference in farm income between CSA practice adopters and non-adopters. This finding suggests that adopting CSA practices may increase farm income through increased productivity and potentially enhance financial stability, leading to food and nutrition security.
Monthly expenditure on food items by SSUC farmers has a higher mean score among CSA practice adopters (ZAR 3 995.38) than non-adopters (ZAR 3 175.02), with a p-value of 0.001, which shows a statistically significant difference between the two groups. The findings suggest that SSUC farmers who adopt CSA practices tend to spend more on food items than non-adopters of CSA practices, possibly due to larger household sizes and the need for food demand. However, monthly expenditure on non-food items is lower among CSA practice adopters (ZAR 5 636.21) than non-adopters (ZAR 11 009.77), with a p-value of 0.001, showing a statistically significant difference between the two groups. The findings suggest that non-adopters of CSA practices may prioritise other expenses than food-related spending.
Agricultural training by SSUC farmers shows a higher mean score of 3.848 among CSA practice adopters than non-adopters (3.488), with a p-value of 0.001, which shows a statistically significant difference between the two groups. The findings suggest that access to agricultural training positively influences the adoption of CSA practices, improving productivity and enhancing food and nutrition security. Land ownership has a substantially higher mean score of 3.706 among CSA practice adopters than non-adopters (2.060), with a p-value of 0.001, which shows a statistically significant difference between the two groups. The findings indicate that secure land ownership may encourage adopting CSA practices by allowing more control over land use decisions. Therefore, increased agricultural productivity will likely enhance CSA practice adopters’ food and nutrition security.
Membership in agricultural-related groups has a higher mean score among CSA adopters (1.690) than non-adopters (1.353), with a p-value of 0.001, which shows a statistically significant difference between the two groups. The findings suggest that social networks and group support are critical in encouraging CSA practice adoption, likely to enhance food and nutrition security.
The frequency of extension visits shows a substantially higher mean score among CSA practice adopters (1.185) than non-adopters (0.781), with a p-value of 0.001, which shows a statistically significant difference between the two groups. The findings indicate that access to frequent extension services may support adopting CSA practices through ongoing technical guidance, leading to better productivity and improved food and nutrition security.
Access to credit shows a slightly higher mean score for CSA practice adopters (1.825) than non-adopters (1.751). The findings further show that access to credit has a marginally significant p-value of 0.068, suggesting that while credit access is essential, it is not a primary differentiator between CSA practice adopters and non-adopters.
The mean score of using hired labour is substantially lower (0.204) among CSA practice adopters than non-adopters (0.960), with a p-value of 0.001, which shows a statistically significant difference between the two groups. The findings suggest that CSA practice adopters relied more on household labour, likely due to larger household sizes, reducing the need for external labour. While household labour availability is likely to be cost-effective support for CSA practices among adopters, relying solely on household labour could negatively impact food and nutrition security if the household lacks sufficient and skilled labour capacity, leading to reduced productivity or burnout, which could hinder the effective implementation of labour-intensive and complex CSA practices, ultimately compromising food and nutrition security.
The number of part-time labourers from the household by SSUC farmers has a slightly higher mean score of 3.184 among CSA practice non-adopters than adopters (3.081), with a p-value of 0.002, which shows a statistically significant difference between the two groups. The findings suggest that t non-adopters of CSA practices devote slightly more part-time household labour to non-agricultural practices. Allocating part-time labourers to non-agricultural practices instead of CSA practices may negatively impact SSUC farmers’ food and nutrition status due to reduced labour availability, potentially decreasing agricultural productivity and, thus, food and nutrition security.
The farming experience in years of SSUC farmers shows a substantially higher mean score (22.010) among non-adopters of CSA practices than adopters (16.891), with a p-value of 0.001, which shows a statistically significant difference between the two groups. The findings imply that the less experienced SSUC farmers could be more open to adopting CSA practices to improve their yields and fulfil their food and nutrition needs. On the other hand, more experienced SSUC farmers may be less inclined to adopt CSA practices due to a preference for traditional methods they are familiar with and regarding new approaches such as CSA practices to be more risky.
The distance to the farming site is slightly longer for non-adopters of CSA practices, with a mean score of about 4km than adopters (3.6km), with a p-value of 0.024, which shows a statistically significant difference between the two groups. The findings suggest that proximity to the farm may facilitate the adoption of CSA practices because of reduced travel time and logistical, health and security challenges by SSUC farmers. Being closer to the farm by SSUC farmers allows for frequent monitoring and efficient management of CSA practices, improving crop productivity and likely enhancing food and nutrition security.
3.2 Descriptive analysis of the household food consumption score and household dietary diversity score of small-scale urban crop farmers between climate-smart agriculture practice adopters and non-adopters
This section presents the analysis of the household food consumption score and household dietary diversity score of SSUC farmers between CSA practice adopters and non-adopters. Fig 1 demonstrates notable differences in food consumption patterns between CSA practice adopters and non-adopters. The median food consumption for CSA practice adopters is approximately 27.7 HFCS. The interquartile range (IQR) for adopters ranges from 25 to 30 (IQR = 5), reflecting moderate variability. The whiskers suggest some diversity in food consumption extending from about 20 to 35 HFCS among CSA practice adopters. In contrast, CSA practice non-adopters demonstrate a higher median food consumption, 31.6 HFCS. This finding suggests that, on average, CSA practice non-adopters have a slightly better food consumption score than CSA practice adopters. However, the IQR for CSA practice non-adopters is narrower than for the CSA practice adopters, ranging from 30 to 34 (IQR = 4). Again, the whiskers extend from approximately 22 to 37 HFCS, reflecting a smaller food consumption score range than CSA practice adopters. The outlier below the lower whisker shows a deficient food consumption for the CSA practice non-adopters. The findings show distinct food consumption scores between CSA practice adopters and non-adopters. Non-adopters have a higher food consumption and a narrower food consumption range, suggesting more stable food consumption patterns than CSA practice adopters, probably due to more uniform access to or utilisation of food resources. While CSA practices non-adopters exhibit a slightly higher and more consistent food consumption, adopters experience broader variability in food consumption scores. The greater variability in food consumption by CSA practice adopters could depict that factors other than CSA practice adoption influence food consumption, such as diverse access to food resources or differing levels of CSA practice adoption. The findings affirm that CSA adoption is complex and has a non-uniform or no straightforward impact on food consumption, with other factors at play.
Fig 2 reveals a subtle yet meaningful difference in dietary diversity between CSA practice adopters and non-adopters. Climate-smart agriculture adopters have a slightly lower median dietary diversity (7.3 HDDS) than non-adopters (8.7 HDDS). The interquartile range (IQR) for CSA practice adopters ranges from 6 to 9 (IQR = 3), suggesting a moderate spread in dietary diversity. On the other hand, the IQR is more compact for CSA practice non-adopters, with an IQR of 1, ranging from 8 to 9, suggesting a lesser variability in dietary diversity for CSA practice non-adopters. Again, the whiskers for CSA practise adopters suggest more variability in dietary practices (extend from approximately 3 to 12 HDDS). In contrast, the whiskers for CSA practice non-adopters are narrower, indicating a lesser and uniform dietary range (extending from about 5 to 10 HDDS). Interestingly, the data shows outliers among CSA practice non-adopters with high HDDS, which such SSUC farmers could attain through factors unrelated to CSA practice adoption. While there is a relatively similar dietary diversity median for both CSA practice adopters and non-adopters, overall, the broader dietary diversity distribution among CSA practice adopters than non-adopters reflects potential dietary diversity differential benefits or use of diverse food sources by CSA practice adopters.
3.3 Factors influencing decisions to adopt climate-smart agriculture practices by small-scale urban crop farmers in eThekwini Municipality
Table 3 summarises the probit selection model estimates for adopting CSA practices. The chi-square statistic (226.201) is very high, and the associated probability Prob > Chi2 (0.000) indicates that the probit selection model is statistically significant, implying that the model possesses a reliable explanatory capacity for SSUC farmers’ adoption of CSA practices. The pseudo-R2 value is 0.397, suggesting a moderately good fit for the data. This value indicates the probit selection model explains 39.7 percent variation in the binary outcome variable relative to a null model that includes no predictors other than the constant term. In social sciences, human behaviour is complex to predict [72]. Predicting the determinants of adopting CSA practices encompasses explaining social and human factors behaviour, which can be challenging. Therefore, a pseudo-R2 score of 0.397 in the current research is reasonable, indicating a good fit. Most explanatory variables in the model exhibit a 10 percent or lesser statistical significance. The variables of age, education (schooling years), household size, off-farm income, monthly expenditure on food items, agricultural training received, group membership and credit access significantly influenced decisions to adopt CSA practices.
The age of the sampled SSUC farmers negatively and significantly influenced the adoption of CSA practice. Secondly, the probit selection model predicts that each additional year of age for the SSUC farmer in ETH Municipality would decrease by approximately 1 percent, which is the probability of adopting CSA practices. The education (schooling years) of the sampled SSUC farmers shows a negative and significant effect on adopting CSA practices. Secondly, the probit selection model predicts that the likelihood of adopting CSA practices decreases by approximately 5 percent with each additional year of schooling. The study found a positive and significant association between the household size of the sampled SSUC farmers and the adoption of CSA practices at a 1 percent significance level. Further, the probit selection model predicts the likelihood of adopting CSA practices increases by approximately 7 percent with an additional household member. The probit selection model predicted a negative and significant effect between the off-farm income of the sampled SSUC farmers and the adoption of CSA practices at a 10 percent significance level. Secondly, the probit selection model predicts a decrease in adopting CSA practices by approximately 20.8 percent with a rise in off-farm income. The monthly food expenditures by the sampled SSUC farmers show a positive and significant effect on the adoption of CSA practices at a 1 percent significance level. Despite the statistical significance of the monthly food expenditures variable, a coefficient of zero (0.000) implies no practical effect of the changes in monthly food expenditures on CSA practices adoption decisions. Receiving agricultural training by the sampled SSUC farmers negatively and significantly affects adopting CSA practices at a 10 percent significance level. Additionally, the probit selection model predicts that receiving agricultural training is marginally associated with a lesser likelihood of approximately 9.6 percent adopting CSA practices, implying that the probability of adopting CSA practices decreases for the urban farmer who receives agricultural training. The study established a positive and significant effect between the adoption of CSA practices by the sampled SSUC farmers and whether the farmer belonged to a group at a 5 percent significance level. Also, the probit selection model predicts that being a group member bolsters the probability of adopting CSA practices by approximately 12.8 percent. The study affirms that access to credit by the sampled SSUC farmers has a positive and significant effect on the adoption of CSA practices at a 1 percent significance level. Additionally, the probit selection model shows that for farmers who can afford credit and access to financial resources, the chances of adopting CSA practices increase by approximately 26.5 percent.
3.4 Marginal treatment effects of the impact of climate-smart agriculture practices adoption on food and nutrition security
The study assesses the impact of adopting CSA practices on food security, utilising HFCS and HDDS indicators. The MTEs with instrumental variables model was adopted to estimate the marginal treatment effect (Table 4). The estimations provide insight into how much treatment effects differ according to the urban farmers’ observed characteristics. The coefficients measure the impact on the outcome in the state where CSA practices adoption is low–proxied as not adopted (untreated group), depicting the difference in the effects between the states that received treatment and those that did not. Thus, these differences depict variations in the treatment effect depending on different covariate values comparable to the interaction between the treatment effect and a covariate in an ordinary least squares estimation [47].
Table 4 displays the MTEs of adopting CSA practices on food security. In the untreated and treated state, gender shows no significant effect on HFCS. In the treated state, although insignificant, gender exhibits a negative coefficient, suggesting that treatment may decrease HFCS for urban female farmers compared to males. Marital status has an insignificant effect on HFCS while significantly impacting HDDS. However, being a single urban farmer rather than wedded decreases HDDS with treatment (CSA practices adoption). The findings show that the employment status does not significantly impact the HFCS for the treated groups. However, the employment status negatively affects HDDS for unemployed urban farmers in the treated group. In the untreated state, age negatively and significantly impacts HDDS, indicating that as urban farmers age, dietary diversity decreases proportionally. However, the results reveal a positive and significant effect on HDDS for the treated group. Household size shows an insignificant impact on HFCS in the untreated and treated models. At the same time, there is a significant negative and positive relation to HDDS in the untreated and treated models, respectively. The untreated and treated models show that farm income positively and significantly affected HFCS and HDDS. However, a zero coefficient suggests a small effect. Farm income helps urban farmers boost productivity, contributing to household income and food security [38]. Therefore, farm income will likely have a direct and positive association with HFCS and HDDS through urban farmers’ adoption of CSA practices.
Off-farm income highlights a significant positive effect on HDDS for the untreated model, while it exhibits a negative but significant effect on HDDS for the treated group. Monthly expenditure on food items negatively affects HFCS in the untreated group but positively and significantly affects HDDS when treated. Land ownership shows a significant and negative effect on HFCS and HDDS for the untreated group but an insignificant effect on the treated. Membership in an agricultural-related group shows an insignificant effect on HFCS (untreated) but a significant negative effect on the treated group. Membership in an agricultural-related group shows a significant and negative effect on HDDS (untreated) and a significant positive effect on HDDS (treated). The variable “hired labour” shows an insignificant effect on HFCS for the untreated group while a significant negative effect on HFCS for the treated group. The model also indicates that hired labour has a positive but insignificant effect on HDDS (untreated) yet a significant but negative effect on HDDS (treated).
Access to irrigation has a statistically significant and negative effect on HDDS for the untreated group. However, it has a positive but statistically insignificant effect on HDDS. In the treated group, the variable “number of part-time labourers from household” significantly and negatively affects HDDS. In the treated state, the distance to the farming site has a negative and statistically significant effect on HFCS. This finding suggests that increasing the distance travelled to the farm when urban farmers adopt CSA practices could decrease household food consumption. Again, the findings show a positive significant effect on HDDS (untreated) and a positive but insignificant effect on the treated group.
3.5 Propensity scores on the climate-smart agriculture practices adoption status of small-scale urban crop farmers
Fig 3 demonstrates that SSUC farmers share joint support for adopting CSA practices. Fig 3 displays the propensity scores of SSUC farmers’ adoption status. The baseline first-stage probit in the probit selection model predicted the propensity scores. It is evident from Fig 3 that the initial regression stage produces a range of 0.02 to 0.99 for joint support. It exhibits the stable mutual support that arises from differences in the covariates and instruments of the second stage. According to Cornelissen et al. [46] and Shahzad and Abdulai [54], this satisfies the general assumption in MTE estimations that the MTE curve is constant concerning control variables.
3.6 Marginal treatment effects of the adoption of climate-smart agriculture practices on Household Food Consumption Score and Household Dietary Diversity Score
Fig 4 display the MTE curve results. The MTE curves’ 95 percent confidence intervals are from bootstrapped standard errors using 500 replications. Fig 4 clearly illustrates the direct correlation between unobserved characteristics and selection based on gains from HFCS, as higher food consumption score values imply a higher likelihood of adoption. Higher CSA practice values suggest a higher likelihood of adoption and depict an adoption propensity. The MTE curve in Fig 4 illustrates a pattern of direct selection on gains discovered for observed farm household characteristics. It rises with the unobserved increases in adoption. As a result, in terms of reduced food insecurity, farmers who are most likely to implement CSA practices benefit from the pattern of heterogeneity (slope of MTE curve), which is statistically significant at the 5 percent level in the lower half of Table 5 (see the p-value for the test of unobserved (essential) heterogeneity). The curve’s downward slope suggests that adoption benefits decrease as adoption resistance rises. This slope points to a trend of positive selection driven by gains. Thus, SSUC farmers who are more likely to adopt CSA practices stand to benefit more in terms of diversified food.
3.7 Causal effects of the adoption of climate-smart agriculture practices on food and nutrition security
Table 5 shows the results of the causal impact of adopting CSA practices on food and nutrition security from the baseline specification reporting the adoption of CSA practices and its estimated treatment effects on the outcome variables. The findings demonstrate that all outcome variables follow the same selection pattern, with farmers most likely to adopt and benefit from adoption to the greatest extent (treatment effects on the treated (ATT) placed higher than the rest of the parameters). Hence, unobserved gains that indicate positive selection are statistically significant. The average treatment effects (ATE) show a positive and significant effect. According to the ATE results, implementing CSA practices improves food consumption by 16 percent and increases household dietary diversity by 31 percent. These improvements are significant (at the 1 percent level). According to ATT estimates, adoption improves dietary diversity by 76 percent and increases food consumption by 8 percent for an average adopting SSUC farmer.
4. Discussion
The results show that demographics and socioeconomic characteristics, including age, education (schooling years), household size, off-farm income, monthly expenditure on food items, agricultural training received, membership in an agricultural-related group, and access to credit, were statistically significant in the decision to adopt CSA practices. Gender dynamics are critical in adopting CSA practices [73] as they influence resource endowment and decision-making, including shaping the perceived benefits of CSA practices. The results reveal that SSUC farmers in ETH Municipality were predominantly female, as observed by Bisaga et al. [62]. Despite not enough evidence that gender influences the decision to adopt CSA practices by the SSUC farmers in ETH Municipality in the probit selection model estimation, the marginal treatment effect when farmers adopt CSA practices suggests that it increases dietary diversity for urban female farmers than males, whilst it exhibits a decrease in HDDS in the untreated group. This result is not surprising and similar to the findings of Nahar et al. [74] and Mataka et al. [75]. This finding underscores the significance of gender-sensitive UA support for female farmers to enhance their personal, economic and dietary intake benefits, strengthening CSA practices adoption for improved food and nutrition for urban farmers.
Mthethwa [76] asserts that marital status influences the adoption of CSA practices, with wedded farmers less likely to adopt than their single counterparts. The results show that most single SSUC farmers in ETH Municipality could readily adopt CSA practices without the hassle of dealing with family responsibilities and risk-taking behaviours influenced by family arrangements. Although not statistically significant in influencing the decision to adopt, on the contrary, the marginal treatment effects show that marital status has a negative significant impact on HDDS, suggesting that for single urban farmers, HDDS decreases with adoption. This finding indicates that single urban farmers are more prone to lacking familial and social support networks than wedded urban farmers. On the other hand, wedded urban farmers could enjoy resource sharing and economic support, likely to contribute to attaining higher dietary diversity and food security from the adoption of CSA practices [77].
Age significantly influences the adoption of CSA practices, with younger farmers more likely to embrace them than their aged counterparts [78]. The age of the SSUC farmers in ETH Municipality depicts an economically active but, at the same time, ageing farming population. Older farmers could potentially limit the adoption of CSA practices due to being more risk-averse and less open to new technologies [40]. The decision to use new technology might be adversely affected by age, as older farmers tend to exhibit greater risk aversion than their younger counterparts, resulting in a lower likelihood of technology adoption [79]. This assertion aligns with the probit selection model, which shows that age has a negative and statistically significant impact on adopting CSA practices. However, the marginal treatment effects results suggest a positive impact of age on HDDS. This finding indicates that older farmers with the experience and resources to invest in CSA practices will likely increase their dietary diversity levels after implementing them [40, 80]. This finding suggests that implementing CSA practices enables older urban farmers to catch up with young and agile farmers regarding dietary diversity. Again, older urban farmers may prioritise tending to their urban farms as they are less likely to engage in other activities simultaneously [81].
Education has a critical role in decision-making for small-scale urban farmers. Gabriel et al. [82] highlight that education typically positively influences CSA adoption, equipping farmers with the basic literacy and comprehension skills necessary to comprehend CSA practices’ benefits and requirements. The study’s sample (SSUC farmers) generally had some basic education, and the expectation was that they could make sound and informed decisions concerning adopting CSA practices. Surprisingly, contrary to the norm and prior research [83, 84], the probit selection model reveals a negative effect of attaining higher education on CSA practice adoption decisions. While the assumption is to better comprehend climate-related risks and benefits of adaptive practices by the more educated SSUC farmers [85, 86], the study’s results reveal that, paradoxically, higher education might discourage CSA practices adoption by SSUC farmers. This discrepancy could reflect the shift from agriculture and CSA practice-related activities to non-agricultural livelihoods, which are perceived as more stable and lucrative as SSUC farmers attain higher levels of education. Again, highly educated SSUC farmers may be risk averse, given the initial high costs, risks, and complexity of adopting some CSA practices vis-a-vis the short-term benefits, concluding that the CSA practices are less viable for their specific needs and context [87]. Interestingly, the study also found that education level was insignificant concerning household food consumption and dietary diversity, suggesting that factors beyond education influenced food and nutrition outcomes. The findings warrant further research to uncover the intricate intersection between SSUC farmers’ educational backgrounds, risk preferences, economic opportunities, and livelihood prospects.
The results portray large family sizes in the households of SSUC farmers in ETH Municipality. A large household size demonstrates the availability of potential family labour for crop cultivation. This situation could facilitate the adoption of CSA practices by farming households, especially those CSA practices that involve intensive labour [88]. The probit selection model predicts that household size positively and significantly influences the adoption of CSA practices. A larger household size reflects greater labour availability needed to implement labour-intensive CSA practices [89]. This finding was expected. Nonetheless, household size shows an insignificant effect on HFCS but a positive effect on HDDS when farmers adopt CSA practices. The findings suggest that urban farmers with large households could significantly improve their dietary diversity by adopting CSA practices. Though focusing on rural farming households, Ali et al. [78] support this finding that households with larger household sizes potentially increased their dietary diversity intake by adopting CSA practices due to diverse agricultural outputs. The implication could be the same for urban farmers with larger households.
Employment status directly influences the CSA practice’s effectiveness [90]. A substantial proportion of the study’s sample reported being unemployed. Formal and informal economic activities are critical in influencing farmers’ decisions to adopt CSA practices [44]. With limited economic activities for SSUC farmers in ETH Municipality, farming was their dominant livelihood. Nonetheless, employment status was insignificant in influencing the adoption of CSA practices from the probit selection model and did not impact HFCS. However, it negatively affected HDDS for unemployed urban farmers when adopting CSA practices. This finding is expected because adopting CSA practices does not immediately translate into dietary diversity, which can be prolonged over several cropping seasons [91]. Again, unemployed urban farmers could not immediately benefit from adopting CSA practices because of a lack of finance, resources, and knowledge, which could decrease dietary diversity in the short term.
Farm income is critical to adopting CSA practices [92]. Urban farmers with high returns from their UA activities will likely invest capital in sustainable farming, such as CSA practices. Implementing CSA practices is initially costly, and adoption benefits would translate into long-term instead of short-term benefits, requiring financial stability within the transition period [93]. Therefore, urban farmers who have better financial resources could be willing to adopt and implement CSA practices. Conversely, those with lesser income may perceive CSA practice adoption as risky. Farm income is insignificant in shaping urban farmers’ decisions to adopt CSA practices in the probit selection model. However, it displays a positive and statistically significant effect on HFCS and HDDS, suggesting its critical impact on food security. Farm income helps urban farmers boost productivity, thus contributing to household income and food security [38].
Off-farm income also plays a critical role in influencing small farms’ economic behaviour and ability to adopt CSA practices [94]. Although the sampled urban farmers were highly dependent on farming for income, the results suggest a diversified income portfolio, which could influence the adoption of CSA practices. The probit selection model predicts that off-farm income decreases the likelihood of adopting CSA practices. This finding was unexpected because farmers with access to revenue from sources other than farming could be more financially capable and likely to increase their likelihood and level of use of CSA practices [81]. On the contrary, urban farmers with reasonably higher off-farm earnings may not feel pressured to adopt CSA practices due to adequate financial security from off-farm income sources [95]. Similarly, the marginal treatment effects suggest that off-farm income significantly positively affects HDDS for non-adopters. At the same time, it shows a negative impact on HDDS when farmers adopt CSA practices. This finding is similar to Tanimonure et al. [96], who reported a decreased dietary diversity with increased off-farm income. The decline in dietary diversity is explainable. While off-farm income is economically beneficial for farmers, small-scale urban farmers with a higher off-farm income may tend to engage less in farming activities, potentially eroding the consumption of diverse food opportunities that come with their own production.
Farm households’ monthly expenditures on food and non-food items could influence urban farmers’ adoption of CSA practices. Spending more on food items may limit the household’s investment in CSA practices, while spending more on non-food items could indicate financial flexibility and the ability to invest in CSA practices [91]. The findings show relatively high monthly expenditures on food and non-food items by SSUC farmers in ETH Municipality. The higher spending on food and non-food items mirrors the relatively high household sizes, implying more food demand and other household expenses such as schooling and health care. However, the observed high variability in expenditure patterns indicates differential priorities and capability to invest in CSA practices. The probit selection model predicts that the monthly food expenditures positively and significantly affect the decision to adopt CSA practices despite its minimal or no practical effect (zero (0) coefficient). Nonetheless, the marginal effects treatments show that the monthly expenditure on food items positively and significantly affects HDDS when adopting CSA practices. This finding was expected. Innovations in CSA practices that incorporate, for example, diversified cropping and efficient water management enhance food security through access to various foods [97]. Suppose households can spend more on food and adopt CSA practices. In that case, it suggests the household’s resilience and prospects to improve food security when adopting CSA practices.
Diro et al. [98] underscore the significance of agricultural training in enhancing the adoption of CSA practices. Most SSUC farmers in ETH Municipality did not undergo specialised agricultural training, with a few with specialised training. The study hypothesised that farmers with some agricultural training would be more adaptive to CSA practices as they possess the necessary skills and knowledge to implement them effectively. Surprisingly, the probit selection model shows that receiving agricultural training negatively and significantly affects decisions to adopt CSA practices. This finding was unexpected. However, this finding is explainable in that most conventional agriculture may not always align with CSA practices principles and techniques, thus conflicting with innovative, riskier CSA practices that do not offer immediate benefits, leading to reluctance to adopt CSA practices by trained urban farmers. Nonetheless, agricultural training has an insignificant effect on HFCS and the HDDS, suggesting that other factors could be more important in affecting food security for SSUC farmers in the ETH Municipality.
Secure land tenure is critical to adopting CSA practices [99], as it gives farmers the confidence and security to invest in sustainable land management strategies that they might otherwise see as too risky. Based on land ownership, arguably, a fair proportion of SSUC farmers in ETH Municipality have secure land tenure. Farmers with a secure land tenure are likely to adopt CSA practices due to the stability and confidence of their investment in land. On the other hand, farmers’ lack of ownership or rights to land could significantly lessen the adoption of CSA practices, underscoring the need for land tenure security in urban areas. Despite being insignificant in shaping adoption decisions, the study found land ownership to have a significant and negative effect on HFCS and HDDS for the untreated group (non-adopters of CSA practices) but an insignificant impact on the treated. This finding was expected for the non-adopters. Owning land but not investing in effective CSA practices could result in lower food production and lesser dietary diversity [100]. Likewise, the insignificant effect of land ownership on HFCS and HDDS of the adopters of CSA practices suggests the complex dynamics of land ownership, underscoring the significance of other factors apart from land ownership. Therefore, land ownership alone would not merely translate to improved food availability and dietary diversity for urban farmers.
Being a member of an agricultural-related group can significantly enhance the adoption of CSA practices. Farmers’ organisation or membership in agricultural-related groups provides access to knowledge on modern production techniques, innovations, and labour exchange [101]. A fair proportion of the sampled urban farmers had membership in an agricultural-related group, suggesting that they enjoyed the collective benefits of access to shared knowledge, resources, and support systems that come with being a member of an agricultural-related group. The variable membership to an agricultural-related group positively and significantly affects the decision to adopt CSA practices. Belonging to farmer groups offers access to shared knowledge, resources, and learning opportunities likely to spearhead the adoption and implementation of CSA practices, especially where urban farmers struggle with various climate adaptation challenges [102]. The risk is better managed by distributing it among the group and through support systems. Membership in an agricultural-related group shows a significant negative effect on HFCS but a significant positive effect on HDDS when farmers adopt CSA practices. The impact on household food consumption due to CSA practices adoption was unexpected. However, a decrease in household food consumption by urban farmers who belong to farmer groups and adopt long-term CSA practices may not yield immediate food production. Therefore, CSA practices that offer long-term benefits may temporarily affect food availability. On the other hand, the effect of being a member of an agricultural-related group on dietary diversity was expected. Farmer groups improve members’ food consumption and diversity by increasing diverse food production [103, 104].
Extension services are essential to promoting urban farmers’ adoption of CSA practices as they provide critical information, technical support, and training to overcome challenges [81, 105]. Access to extension services remains challenging for SSUC farmers in ETH Municipality, with a relatively large proportion not having access to extension services. Inadequate extension services by a substantial proportion of SSUC farmers in ETH Municipality is problematic as it could hinder the broader uptake of CSA practices for improved food, income and sustainability of urban farms. Nonetheless, this variable did not show statistical significance, suggesting that while it is essential, other factors were of immediate attention in shaping the decisions to adopt CSA practices by SSUC farmers in ETH Municipality and its effect on household food consumption and dietary diversity intake.
According to Wu and Li [106], farmers’ choices to accept innovations are influenced by their access to credit. The results confirm that most of the SSUC farmers in ETH Municipality did not have access to credit. This situation could be attributed to most of them lacking access to credit-related information and collateral. With limited access to credit, the implication is that the SSUC farmers in ETH Municipality could not afford to hire skilled labour or invest the necessary workforce to implement labour-intensive CSA practices for their UA activities [107]. The probit selection model shows a positive and statistically significant influence of credit access on the decision to adopt CSA practices. This finding is not surprising. Access to credit assures urban farmers the ability to deal with or handle challenges associated with financial barriers [78]. For example, access to credit will ensure the timely purchase of necessary inputs and resources required to implement CSA practices effectively. Therefore, financial support smoothens the liquidity flows of urban farmers, allowing investment in longer-term CSA practices. Nonetheless, access to credit did not significantly impact household food consumption and dietary diversity, suggesting other factors are at play.
Linked to the limited financial muscle of SSUC farmers in ETH Municipality, a higher proportion did not hire farm labour for their UA activities but utilised household labour. Hired labour is essential for effectively implementing and sustaining labour-intensive CSA practices over time. Again, skilled labour is vital to implement CSA practices effectively. However, the cost of hiring skilled labour may hinder resource-constrained urban farmers’ adoption of CSA practices. Despite the variable hired labour being insignificant in the decisions to adopt CSA practices, it shows a significant and negative effect on household food consumption and dietary diversity for the treated group (when adopting CSA practices). This finding suggests that hired labour decreases food consumption and dietary diversity when farmers adopt CSA practices. The operational costs of hired labour could temporarily reduce diverse food production or purchases, directly affecting urban farmers’ food consumption and diversity in the short term (during the adoption phase of CSA practices) [108]. Again, the marginal treatment effects show that relying on part-time household labour significantly and negatively affects dietary diversity when adopting CSA practices. The negative effect of household part-time labourers on dietary diversity when urban farmers adopt CSA practices suggests divided attention of labour between UA activities and non-agricultural activities. Household part-time labourers spend more time on out-of-farm employment, thus reducing the time dedicated to diverse crop cultivation critical to dietary diversity [109].
Access to improved water management practices, such as irrigation technology, is paramount for implementing CSA practices [98]. Irrigation technology enhances crop yields, productivity and resilience to variable climate conditions [78]. The findings show that most SSUC farmers in ETH Municipality had access to irrigation technology, suggesting a strong potential for adopting CSA practices. Despite the access to irrigation variable being insignificant in influencing the adoption of CSA practices by SSUC farmers in ETH Municipality, the heightened access to irrigation technology would facilitate the higher adoption of CSA practices, particularly in areas where water is a limiting factor. Concerning the impact on food security, access to irrigation has a positive but statistically insignificant effect on dietary diversity. While irrigation technology may increase agricultural productivity and dietary diversity, its impact is conditional on effective water management and monoculture or cash crop production versus high diet diverse food systems. Although the variable “access to irrigation” is not statistically significant in the treated group (when farmers adopt CSA practices), the positive coefficient suggests that urban farmers with access to irrigation and adopting CSA practices could increase their dietary diversity. This finding is consistent with the notion that dependable water sources for irrigating crops result in increased agricultural output and, thus, enhanced dietary diversity [110].
The results depict that the SSUC farmers in ETH Municipality had substantial experience. Farmers with more expertise could have a positive attitude toward adopting CSA practices since they have more information and knowledge, allowing them to assess the benefits of CSA practices [26]. Nonetheless, farming experience had an insignificant effect on the decisions to adopt CSA practices, household food consumption, and dietary diversity when farmers adopted (treated group) as predicted by the probit selection model and the marginal treatment effects, respectively. This finding suggests that other factors are at play besides farming experience.
The results indicate that while some SSUC farmers in ETH Municipality practised their UA activities at their residence, on average, the distance to the farming site is about 3km. Arguably, this average travel distance to the farming site is relatively short and reasonable in terms of cost and time. However, the impact will rely on transport availability, the urban landscape, and farmer characteristics such as old age and health. Again, related to distance will be factors such as proximity to markets, water sources, and sources of credit concerning the farming site. The longer the distance urban farmers will travel, the lesser the likelihood of adopting CSA practices [102]. Not being near the farming site and necessary sources increases costs associated with transporting materials and accessing the required resources. Despite exhibiting an insignificant effect on decisions to adopt CSA practices from the probit selection model, in the treated state, the distance to the farming site has a negative and statistically significant effect on HFCS, suggesting that an increase in the distance travelled to the farm when urban farmers adopt CSA practices could decrease household food consumption. The finding implies that a shorter distance to the urban farm or farming site is likelier to achieve greater food consumption than those whose farms or farming sites are further away from their residency. When farms or farming sites of urban farmers are closer to their residences, they can utilise CSA practices more efficiently and effectively, with the amount of time spent on the farm or farm site being greater than on farms located further away [111].
The findings overall demonstrate the effects of adopting CSA practices by SSUC farmers in ETH Municipality on food and nutrition security. Variance in the propensity to adopt CSA practices reflects the varying urban farmers’ circumstances and how they respond. The analysis of unobserved farmer characteristics and decision-making towards adopting CSA practices depicts a direct relationship between higher food consumption and unobserved gains (likely farmers’ perceptions and experiences) [78]. The statistically significant MTE curves suggest heterogeneity in benefits from adopting CSA practices, prevalent in farmers with higher baseline food security, affirming the theory of positive selection in CSA practices adoption. The average treatment effects (ATE) and effects on the treated (ATT) reveal statistically significant enhanced outcomes in food consumption and dietary diversity when urban farmers adopt CSA practices compared to non-adopters. The findings underscore the critical role of CSA practices in improving food security among urban farmers as determined by observed and unobserved farmer characteristics [44].
5. Conclusion
In South Africa, food security for small-scale farmers is generally under threat due to climate change and extreme weather events. Adopting CSA practices is one of the strategies to increase food security in a changing climate, particularly in UA. Climate-smart agriculture practices are a group of interventions designed to help farmers better manage risk and adjust their farming systems to climate change while sustainably increasing productivity. The current study assessed the contribution of the adoption of CSA practices to food and nutrition security for SSUC farmers in the ETH Municipality of SA. Four hundred and twelve (412) respondents were selected. A marginal treatment effect model, while controlling for heterogeneity, was employed. The results of the MTE model show that farm income, off-farm income, monthly expenditure on food items, membership in an agricultural-related group, hired labour, and distance to farming site significantly affected household food consumption patterns. In contrast, gender, marital status, employment status, age, household size, farm income, off-farm income, monthly expenditure on food items, membership in an agricultural-related group, hired labour and number of part-time labourers from household significantly factors influenced household dietary diversity status of SSUC farmers in ETH Municipality.
Similarly, the empirical findings demonstrate significant variation in the advantages of CSA practice adoption. Adopting CSA practices across HFCS and HDDS results in a pattern of positive selection on unobserved gains. This observation is explained by the fact that adoption typically positively impacts households more likely to adopt CSA practices. The average treatment effects on the treated (ATT) demonstrate that adopting CSA practices significantly increases farm households’ food and nutrition security. The analysis revealed that adopting CSA practices could effectively raise SSUC farmers’ living standards by enhancing their household’s dietary diversity and food consumption patterns. Adoption of CSA practices was positively associated with household food and nutrition security. Adopters of CSA practices were 16 percent more food secure in terms of HFCS and 31 percent in terms of HDDS. Hence, the probability of farmers who adopted CSA practices were likely to be more food secure in terms of higher household consumption patterns and household dietary scores.
Several targeted policy recommendations emerge in light of the demographic and farm characteristics analysis of SSUC farmers in ETH Municipality. Notably, the predominance of female farmers underscores the necessity for gender-specific interventions that provide access to resources, training, and financial products designed to meet their unique challenges. The evident educational gaps call for enhanced agricultural training programs incorporating CSA practices aligned with sustainable agricultural goals. Addressing the limited access to credit, the study recommends that policies facilitate more accessible access to affordable financial services tailored to urban agricultural needs. Furthermore, given the substantial proportion of SSUC farmers without secure land tenure, implementing policies that provide stable land rights is crucial for fostering long-term investments in CSA practices. Strengthening farmer organizations, expanding extension services, and providing specific support for the ageing farmer population are also vital. Together, these measures can significantly boost the adoption of CSA practices, enhance urban food security, and promote sustainable urban agricultural development.
While the current study presents timely and valuable insights on the heterogeneous adoption of CSA practices in improving food and nutrition security among SSUC farmers in the ETH Municipality, future research should address some limitations. First, the study focused on the ETH Municipality, and future research could include a broader investigation in other regions to allow comparisons of CSA adoption patterns across different urban settings in South Africa, thus validating the findings in a broader context. Again, the current study was a cross-sectional design, and a longitudinal study, on the other hand, would be more beneficial in tracking and providing more long-term insights on CSA adoption and its impact on food and nutrition security over time. Further, a mixed-method study incorporating qualitative insights, such as in-depth interviews or focus groups, would offer deeper insights into farmers’ motivations, perceptions, and barriers to adopting CSA practices. Lastly, delving into a gender-specific study will be critical to understanding and developing gender-sensitive strategies to support small-scale urban farmers to ensure that CSA interventions are inclusive and effective for all farmers.
References
- 1. Wudil AH, Usman M, Rosak-Szyrocka J, Pilař L, Boye M. Reversing years for global food security: A review of the food security situation in Sub-Saharan Africa (SSA). Int J Environ Res Public Health. 2022;19(22):14836. pmid:36429555
- 2. FAO. Africa Regional Review of Food and Nutrition 2023: Trends 2023 [cited 2024 11 March]. Available from: https://www.fao.org/3/cc8743en/online/sofi-statistics-africa-2023/food-insecurity-experience-scale.html.
- 3. Armstrong M. A fifth of people in Africa are suffering from chronic hunger. This map shows where the situation is most severe: World Economic Forum; 2022 [cited 2024 11 March]. Available from: https://www.weforum.org/agenda/2022/08/hunger-in-africa-serious-millions/.
- 4. Kapari M, Hlophe-Ginindza S, Nhamo L, Mpandeli S. Contribution of smallholder farmers to food security and opportunities for resilient farming systems. Front Sustain Food Syst. 2023;7:1149854.
- 5.
Statista. Production-based carbon dioxide (CO2) emissions in Africa in 2020, by country. Statista New York; 2023.
- 6. Statista. Agriculture in South Africa—statistics & facts 2024 [cited 2024 11 March 2024]. Available from: https://www.statista.com/topics/9876/agriculture-in-south-africa/#topicOverview.
- 7. Stats SA. General Household Survey 2021 2023 [cited 2024 11 March]. Available from: https://www.statssa.gov.za/publications/P0318/P03182021.pdf.
- 8. Department of Agriculture Land Reform & Rural Development. Trends in the Agricultural Sector 2022. Pretoria, South Africa: 2023.
- 9. Du Toit MJ, Rendón O, Cologna V, Cilliers SS, Dallimer M. Why home gardens fail in enhancing food security and dietary diversity. Front Ecol Evol. 2022;10:38.
- 10. Sangwan N, Tasciotti L. Losing the plot: the impact of urban agriculture on household food expenditure and dietary diversity in sub-Saharan African countries. Agric. 2023;13(2):284.
- 11. Irfeey AMM, Chau H-W, Sumaiya MMF, Wai CY, Muttil N, Jamei E. Sustainable mitigation strategies for urban heat island effects in urban areas. Sustainability. 2023;15(14):10767.
- 12. Griebel S, Nelle L, Sango E, Wairimu S, Swanby H, Sobgo S, et al. Food and Crisis: the role of Controlled Environment Agriculture in building local food system resilience: TMG Research; 2022 [cited 2024 10 May]. Available from: https://www.researchgate.net/publication/366030327_Food_and_crisis_the_role_of_Controlled_Environment_Agriculture_in_building_urban_food_system_resilience.
- 13. Hawes JK, Goldstein BP, Newell JP, Dorr E, Caputo S, Fox-Kämper R, et al. Comparing the carbon footprints of urban and conventional agriculture. Nat Cities. 2024:1–10.
- 14. Blay-Palmer A, Santini G, Halliday J, Malec R, Carey J, Keller L, et al. City region food systems: building resilience to COVID-19 and other shocks. Sustainability. 2021;13(3):1325.
- 15.
Intergovernmental Panel on Climate Change (IPCC). Africa. In: Climate Change 2022: Impacts, Adaptation and Vulnerability 2022. Cambridge University Press, Cambridge, United Kingdom and New York, United States of America: 2022.
- 16. World Bank. Climate change adaptation and economic transformation in Sub-Saharan Africa Washington, DC, United States of America 2021.
- 17.
IPCC. Africa. In: Climate Change 2022: Impacts, Adaptation and Vulnerability 2022. Cambridge University Press, Cambridge, United Kingdom and New York, United States of America: Intergovernmental Panel on Climate Change, 2022.
- 18. Masipa T. The impact of climate change on food security in South Africa: Current realities and challenges ahead. Jàmbá J Disaster Risk Stud. 2017;9(1):1–7. pmid:29955344
- 19.
Mrabet R. Sustainable agriculture for food and nutritional security. In: Farooq M, Gogoi N, Pisante M, editors. Sustainable Agriculture and the Environment: Academic Press; 2023. p. 25–90.
- 20. Singh BK, Delgado-Baquerizo M, Egidi E, Guirado E, Leach JE, Liu H, et al. Climate change impacts on plant pathogens, food security and paths forward. Nat Rev Microbiol. 2023:1–17. pmid:37131070
- 21. Lottering S, Mafongoya P, Lottering R. Assessing the social vulnerability of small-scale farmers to drought in uMsinga, KwaZulu-Natal. Int J Disaster Risk Reduct. 2021;65:102568.
- 22. Tilahun G, Bantider A, Yayeh D. Analyzing the impact of climate-smart agriculture on household welfare in a subsistence mixed farming system: Evidence from Geshy Watershed, Southwest Ethiopia. Glob Soc Welf. 2023;10(3):235–47.
- 23. Dougill AJ, Hermans TD, Eze S, Antwi-Agyei P, Sallu SM. Evaluating climate-smart agriculture as a route to building climate resilience in African food systems. Sustainability. 2021;13(17):9909
- 24. Belay A, Mirzabaev A, Recha JW, Oludhe C, Osano PM, Berhane Z, et al. Does climate-smart agriculture improve household income and food security? Evidence from Southern Ethiopia. Environ Dev Sustain. 2024:1–28.
- 25. Tesfaye A, Hammond J, Radeny M, Recha JW, Nigussie A, Ambaw G, et al. The role of climate-smart agricultural practices on household income and food security: A propensity score matching approach. Cogent Soc Sci. 2023;9(2):2273959.
- 26. Atta-Aidoo J, Antwi-Agyei P, Dougill AJ, Ogbanje CE, Akoto-Danso EK, Eze S. Adoption of climate-smart agricultural practices by smallholder farmers in rural Ghana: An application of the theory of planned behavior. PLoS Clim. 2022;1(10)
- 27. Baffour-Ata F, Atta-Aidoo J, Said RO, Nkrumah V, Atuyigi S, Analima SM. Building the resilience of smallholder farmers to climate variability: Using climate-smart agriculture in Bono East Region, Ghana. Heliyon. 2023;9(11) pmid:38027792
- 28. Srinivasan K, Yadav VK. An empirical investigation of barriers to the adoption of smart technologies in integrated urban agriculture systems. J Decis Syst. 2023:1–35.
- 29. Vhumbunu CH, Adetiba TC. Climate-smart agriculture in urban farming: Experiences from selected suburbs in Windhoek, Namibia. Afr J Dev Stud. 2023;13(4).
- 30. Ojo TO, Ogundeji AA, Emenike CU. Does adoption of climate change adaptation strategies improve food security? A case of rice farmers in Ogun State, Nigeria. Land. 2022;11(11):1875.
- 31. Ojo T, Kassem H, Ismail H, Adebayo D. Level of adoption of climate-smart agriculture among smallholder rice farmers in Osun State: does financing matter? Sci Afr. 2023;21
- 32. Sisay T, Tesfaye K, Ketema M, Dechassa N, Getnet M. Climate-smart agriculture technologies and determinants of farmers’ adoption decisions in the Great Rift Valley of Ethiopia. Sustainability. 2023;15(4):3471.
- 33. Teklu A, Simane B, Bezabih M. Multiple adoption of climate-smart agriculture innovation for agricultural sustainability: Empirical evidence from the Upper Blue Nile Highlands of Ethiopia. Clim Risk Manag. 2023;39:100477.
- 34. Tesfaye T, Nayak D. Climate change adaptation measures by farm households in Gedeo zone, Ethiopia: An application of multivariate analysis approach. Environ Dev Sustain. 2023;25(4):3183–209.
- 35. Pagliacci F, Defrancesco E, Mozzato D, Bortolini L, Pezzuolo A, Pirotti F, et al. Drivers of farmers’ adoption and continuation of climate-smart agricultural practices: A study from northeastern Italy. Sci Total Environ. 2020;710:136345.
- 36. Savari M, Damaneh HE, Damaneh HE, Cotton M. Integrating the norm activation model and theory of planned behavior to investigate farmer pro-environmental behavioral intention. Sci Rep. 2023;13(1):5584.
- 37. Chitakira M, Ngcobo NZ. Uptake of climate-smart agriculture in peri-urban areas of South Africa’s economic hub requires up-scaling. Front Sustain Food Syst. 2021;5:706738.
- 38. Mutengwa CS, Mnkeni P, Kondwakwenda A. Climate-smart agriculture and food security in Southern Africa: A review of the vulnerability of smallholder agriculture and food security to climate change. Sustainability. 2023;15(4):2882.
- 39. Khumalo NZ, Sibanda M, Mdoda L. Implications of a climate-smart approach to food and income security for urban Sub-Saharan Africa: A systematic review. Sustainability. 2024;16(5):1882.
- 40. Tadesse B, Ahmed M. Impact of adoption of climate-smart agricultural practices to minimize production risk in Ethiopia: A systematic review. J Agric Food Res. 2023;100655.
- 41. Assaye A, Habte E, Sakurai S, Alemu D. Impact assessment of adopting improved rice variety on farm household welfare in Ethiopia. J Agric Food Res. 2022;10:100428.
- 42. Zegeye MB, Fikire AH, Assefa AB. Impact of agricultural technology adoption on food consumption expenditure: Evidence from rural Amhara region, Ethiopia. Cogent Econ Finance. 2022;10(1):2012988.
- 43. Sisang BB, Lee JI. Impact of improved variety adoption on rice productivity and farmers’ income in Cameroon: Application of propensity score matching and endogenous switching regression. J Agric Life Environ Sci. 2023;35(1):26–46.
- 44. Teklu A, Simane B, Bezabih M. Climate-smart agriculture impact on food and nutrition security in Ethiopia. Front Sustain Food Syst. 2024;7:1079426.
- 45. Heckman JJ, Humphries JE, Veramendi G. Returns to education: The causal effects of education on earnings, health, and smoking. J Polit Econ. 2018;126(S1). pmid:30344340
- 46. Cornelissen T, Dustmann C, Raute A, Schönberg U. Who benefits from universal child care? Estimating marginal returns to early child care attendance. J Polit Econ. 2018;126(6):2356–409.
- 47. Andresen ME. Exploring marginal treatment effects: Flexible estimation using Stata. Stata J. 2018;18(1):118–58.
- 48. Awotide BA, Ogunniyi A, Olagunju KO, Bello LO, Coulibaly AY, Wiredu AN, et al. Evaluating the heterogeneous impacts of adoption of climate-smart agricultural technologies on rural households’ welfare in Mali. Agric. 2022;12(11):1853.
- 49. Alhassan U, Haruna EU. Rural farmers’ perceptions of and adaptations to climate change in Sub-Saharan Africa: Does climate-smart agriculture (CSA) matter in Nigeria and Ethiopia? Environ Econ Policy Stud. 2024:1–40.
- 50. Abadie A, Angrist J, Imbens G. Instrumental variables estimates of the effect of subsidized training on the quantiles of trainee earnings. Econometrica. 2002;70(1):91–117.
- 51. Olagunju KO, Ogunniyi AI, Awotide BA, Adenuga AH, Ashagidigbi WM. Evaluating the distributional impacts of drought-tolerant maize varieties on productivity and welfare outcomes: an instrumental variable quantile treatment effects approach. Clim Dev. 2020;12(10):865–75.
- 52. Cornelissen T, Dustmann C, Raute A, Schönberg U. From LATE to MTE: Alternative methods for the evaluation of policy interventions. Labour Econ. 2016;41:47–60.
- 53. Kallus N. Treatment effect risk: Bounds and inference. Manag Sci. 2023;69(8):4579–90.
- 54. Shahzad MF, Abdulai A. The heterogeneous effects of adoption of climate-smart agriculture on household welfare in Pakistan. Appl Econ. 2021;53(9):1013–38.
- 55.
Dawson J, Hutchings K, Sedick S, Clark B. uThongathi Estuary Management Plan Final Report. Tokai, South Africa: KwaZulu-Natal Department of Economic Development, Tourism and Environmental Affairs; 2021.
- 56. Department of Cooperative Governance & Traditional Affairs. eThekwini Metropolitan KZN. Profile and analysis District Development Model 2020 [cited 2024 27 May]. Available from: https://www.cogta.gov.za/ddm/wp-content/uploads/2020/07/Metro-Profile_Ethekwini.pdf.
- 57. eThekwini Municipality. Medium Term Revenue and Expenditure Framework 2022/2023 to 2024/2025. 2022 [cited 2024 11 March]. Available from: https://www.durban.gov.za/storage/Documents/Budget%20Reports/Medium%20Term%20Revenue%20and%20Expenditure%20Framework/EThekwini%20Municipality%20Medium%20Term%20Budget%202022-2023.pdf.
- 58. Tatsvarei S, Mwale-Manjoro M, Mapiye C, Marufu MC, Singh JA, Letty BA, et al. Situational Analysis and Strategic Advisory on Farming and Food Security in South Africa in the Aftermath of Widespread Public Violence and Looting in KwaZulu-Natal and Gauteng between 10–14 July 2021. Sci Advis Group Emergencies. 2021 [cited 2024 10 May]. Available from: https://www.assaf.org.za/files/2021/1%20SAGE%20Food%20Security%20Advisory%20revised%20-%2030%20Aug%202021.pdf.
- 59. Stats SA. Census of Commercial Agriculture COCA 2017. 2020 [cited 2024 11 March]. Available from: https://www.statssa.gov.za/publications/Report-11-02-01/CoCA%202017%20Fact%20Sheets.pdf.
- 60. Sihlobo W. Thinking about KwaZulu-Natal’s agriculture and food production in this time of crisis. 2022 [cited 2024 11 March]. Available from: https://www.agbiz.co.za/blog/details/thinking-about-kwazulu-natals-agriculture-and-food-production-in-this-time-of-cr.
- 61.
Stats SA. General Household Survey. Report: P0318: Pretoria, South Africa, 2019. Pretoria, South Africa: Statistics SA; 2019.
- 62. Bisaga I, Parikh P, Loggia C. Challenges and opportunities for sustainable urban farming in South African low-income settlements: A case study in Durban. Sustainability. 2019;11(20):5660.
- 63. Shezi N, Ngcoya M. Adaptation to the impacts of climate change on agriculture in eThekwini: A literature review. 2016 [cited 2024 10 May]. Available from: https://researchspace.ukzn.ac.za/server/api/core/bitstreams/03b25589-f639-4483-8b75-9a76b51e8d51/content.
- 64. Udo F, Naidu M. Examining the interplay between economic development and local women’s vulnerability to flood impacts in selected local areas in Durban, South Africa. J Asian Afr Stud. 2022:00219096211069646.
- 65.
Chetty D, Bhoola S. The quest for food security in the eThekwini Municipality. In: Moolakkattu JS, Chathukulam J, editors. Challenges to Local Governance in the Pandemic Era: Perspectives from South Asia and Beyond. Newcastle, UK: Cambridge Scholars Publishing; 2022. p. 248–64.
- 66. Olanrewaju CC, Reddy M. Assessment and prediction of flood hazards using standardized precipitation index—A case study of eThekwini metropolitan area. J Flood Risk Manag. 2022;15(2).
- 67. COGTA. eThekwini Municipality KZN. Profile and Analysis District Development Model 2020 [cited 2024 13 March]. Available from: https://www.cogta.gov.za/ddm/wp-content/uploads/2020/07/Metro-Profile_Ethekwini.pdf.
- 68. Lakens D. Sample size justification. Collabra Psychol. 2022;8(1):33267.
- 69. Bhardwaj P. Types of sampling in research. J Pract Cardiovasc Sci. 2019;5(3):157.
- 70. Andrade C. The inconvenient truth about convenience and purposive samples. Indian J Psychol Med. 2021;43(1):86–8. pmid:34349313
- 71.
Cochran WG. Sampling techniques. 3rd ed. New York: John Wiley & Sons; 1977.
- 72.
Ozili PK. The acceptable R-square in empirical modelling for social science research. In: Saliya CA, editor. Social research methodology and publishing results: A guide to non-native English speakers. IGI Global; 2023. p. 134–43.
- 73. Khoza S, van Niekerk D, Nemakonde LD. Gendered vulnerability and inequality: Understanding drivers of climate-smart agriculture dis- and nonadoption among smallholder farmers in Malawi and Zambia. Ecol Soc. 2022;27(4).
- 74. Nahar K, Mahfuzul Haque A, Hossain MM. Adopting a gender-transformative approach (GTA) to increase dietary diversity and decision-making of women: Evidence from a nutrition-sensitive intervention in Bangladesh. 2023 [cited 2024 10 May]. Available from: https://cgspace.cgiar.org/server/api/core/bitstreams/770f524c-56ed-462e-a467-8795c10f9c0a/content.
- 75. Mataka T, Kaitibie S, Ratna NN. Can women’s empowerment in livestock farming improve household food security? Empirical evidence from rural households in Malawi. Agric Food Secur. 2023;12(1):35.
- 76.
Mthethwa KN. The determinants of adoption of climate-smart agricultural (CSA) practices and its effects on smallholder maize farmers’ welfare [Doctoral]. Pietermaritzburg: University of KwaZulu-Natal; 2023.
- 77. Dallmann D, Marquis G, Colecraft E, Dodoo N. Marital transition is associated with food insecurity, low dietary diversity, and overweight in a female population in rural Ghana. Afr J Food Agric Nutr Dev. 2023;23(1):22149–71.
- 78. Ali H, Menza M, Hagos F, Haileslassie A. Impact of climate-smart agriculture adoption on food security and multidimensional poverty of rural farm households in the Central Rift Valley of Ethiopia. Agric Food Secur. 2022;11(1):1–16.
- 79. Duc Truong D, Tho Dat T, Le Huan H. Factors affecting climate-smart agriculture practice adaptation of farming households in coastal central Vietnam: The case of Ninh Thuan Province. Front Sustain Food Syst. 2022;6:790089.
- 80. Matavel C, Hoffmann H, Rybak C, Steinke J, Sieber S, Müller K. Understanding the drivers of food security among agriculture-based households in Gurué District, Central Mozambique. Agric Food Secur. 2022;11(1):1–15.
- 81. Gemtou M, Kakkavou K, Anastasiou E, Fountas S, Pedersen SM, Isakhanyan G, et al. Farmers’ transition to climate-smart agriculture: A systematic review of the decision-making factors affecting adoption. Sustainability. 2024;16(7):2828.
- 82. Gabriel I, Olajuwon F, Klauser D, Michael B, Renn M. State of climate-smart agriculture (CSA) practices in the North Central and Northwest zones of Nigeria. CABI Agric Biosci. 2023;4(1):33.
- 83. Luu TD. Factors influencing farmers’ adoption of climate-smart agriculture in rice production in Vietnam’s Mekong Delta. Asian J Agric Dev. 2020;17(1):110–24.
- 84. Abegunde VO, Sibanda M, Obi A. Determinants of the adoption of climate-smart agricultural practices by small-scale farming households in King Cetshwayo District Municipality, South Africa. Sustainability. 2020;12(1):195.
- 85. Daudu AK, Abdoulaye T, Bamba Z, Shuaib SB, Awotide BA. Does youth participation in farming programs impact farm productivity and household welfare? Evidence from Nigeria. Heliyon. 2023;9(4).
- 86. Mapiye O, Makombe G, Molotsi A, Dzama K, Mapiye C. Information and communication technologies (ICTs): The potential for enhancing the dissemination of agricultural information and services to smallholder farmers in Sub-Saharan Africa. Inf Dev. 2023;39(3):638–58.
- 87. Ejigu AK, Yeshitela K. Exploring the factors influencing urban farmers’ perception and attitude toward the use of excreta-based organic fertilizers in Arba Minch City, Ethiopia. Front Sustain Food Syst. 2024;7:1271811.
- 88. Ngaiwi ME, Molua EL, Sonwa DJ, Meliko MO, Bomdzele EJ, Ayuk JE, et al. Do farmers’ socioeconomic status determine the adoption of conservation agriculture? An empirical evidence from Eastern and Southern Regions of Cameroon. Sci Afr. 2023;19.
- 89. Li J, Ma W, Zhu H. A systematic literature review of factors influencing the adoption of climate-smart agricultural practices. Mitig Adapt Strateg Glob Chang. 2024;29(1):2.
- 90. Hellin J, Fisher E, Taylor M, Bhasme S, Loboguerrero AM. Transformative adaptation: From climate-smart to climate-resilient agriculture. CABI Agric Biosci. 2023;4(1):30.
- 91. Mujeyi A, Mudhara M, Mutenje M. The impact of climate-smart agriculture on household welfare in smallholder integrated crop–livestock farming systems: Evidence from Zimbabwe. Agric Food Secur. 2021;10(1):1–15.
- 92. Agbenyo W, Jiang Y, Jia X, Wang J, Ntim-Amo G, Dunya R, et al. Does the adoption of climate-smart agricultural practices impact farmers’ income? Evidence from Ghana. Int J Environ Res Public Health. 2022;19(7):3804. pmid:35409488
- 93. Autio A, Johansson T, Motaroki L, Minoia P, Pellikka P. Constraints for adopting climate-smart agricultural practices among smallholder farmers in Southeast Kenya. Agric Syst. 2021;194:103284.
- 94. Setsoafia ED, Ma W, Renwick A. Effects of sustainable agricultural practices on farm income and food security in northern Ghana. Agric Food Econ. 2022;10(1):1–15.
- 95. Kifle T, Ayal DY, Mulugeta M. Factors influencing farmers’ adoption of climate-smart agriculture to respond to climate variability in Siyadebrina Wayu District, Central Highland of Ethiopia. Clim Serv. 2022;26:100290.
- 96. Tanimonure VA, Naziri D, Codjoe SNA, Ayanwale AB. Underutilised indigenous vegetables for household dietary diversity in Southwest Nigeria. Agric. 2021;11(11):1064.
- 97. Wekesa BM, Ayuya OI, Lagat JK. Effect of climate-smart agricultural practices on household food security in smallholder production systems: Micro-level evidence from Kenya. Agric Food Secur. 2018;7(1):1–14.
- 98. Diro S, Tesfaye A, Erko B. Determinants of adoption of climate-smart agricultural technologies and practices in the coffee-based farming system of Ethiopia. Agric Food Secur. 2022;11(1):1–14.
- 99. Olumba CN, Garrod G, Areal FJ. Time preferences, land tenure security, and the adoption of sustainable land management practices in Southeast Nigeria. Sustainability. 2024;16(5):1747.
- 100. Maraveas C. Incorporating artificial intelligence technology in smart greenhouses: Current state of the art. Appl Sci. 2022;13(1):14.
- 101. Asprooth L, Norton M, Galt R. The adoption of conservation practices in the Corn Belt: The role of one formal farmer network, Practical Farmers of Iowa. Agric Hum Values. 2023;40(4):1559–80. pmid:37359835
- 102. van Asseldonk M, Girvetz E, Pamuk H, Wattel C, Ruben R. Policy incentives for smallholder adoption of climate-smart agricultural practices. Front Polit Sci. 2023;5:1112311.
- 103. Cele T, Mudhara M. Impact of market participation on household food security among smallholder irrigators in KwaZulu-Natal, South Africa. Agric. 2022;12(2):261.
- 104. Jabbar A, Liu W, Zhang J, Wang Y, Wu Q, Peng J, et al. Enhancing adaptation to climate change by fostering collective action groups among smallholders in Punjab, Pakistan. Front Sustain Food Syst. 2023;7:1235726.
- 105. Vincent A, Balasubramani N. Climate-smart agriculture (CSA) and extension advisory service (EAS) stakeholders’ prioritisation: A case study of Anantapur district, Andhra Pradesh, India. J Water Clim Chang. 2021;12(8):3915–31.
- 106. Wu H, Li J. Risk preference, interlinked credit and insurance contract and agricultural innovative technology adoption. J Innov Knowl. 2023;8(1):100282.
- 107. Amrullah ER, Takeshita H, Tokuda H. Impact of access to agricultural extension on the adoption of technology and farm income of smallholder farmers in Banten, Indonesia. J Agribus Dev Emerg Econ. 2023.
- 108. Ahmed FF, Naphtali J. Socioeconomic characteristics and food diversity amongst high-income households: A case study of Maiduguri Metropolis, Borno State, Nigeria. Am J Soc Manag Sci. 2014;5(1):19–26.
- 109. Bai Y, Zeng X, Fu C, Zhang L. Off-farm employment, agriculture production activities, and household dietary diversity in environmentally and economically vulnerable areas of Asia. J Integr Agric. 2024;23(2):359–73.
- 110. Jambo Y, Alemu A, Tasew W. Impact of small-scale irrigation on household food security: Evidence from Ethiopia. Agric Food Secur. 2021;10:1–16.
- 111. Kehinde A, Adeyemo R, Ogundeji A. Does social capital improve farm productivity and food security? Evidence from cocoa-based farming households in Southwestern Nigeria. Heliyon. 2021;7(3). pmid:33869842