The Contribution of Trees and Green Spaces to Household Food Security in eThekwini Metro, KwaZulu-Natal

: One of the most signiﬁcant issues faced by many low-and middle-income nations, including South Africa, is ensuring access to healthy, affordable, and sustainable food. South Africa is renowned worldwide for its rich biodiversity and a vast body of traditional knowledge among those who consume forest foods. However, despite ecological diversity, frequent barriers remain to getting diversiﬁed household diets. This study sought to investigate the contribution of trees and green spaces to household food security in eThekwini. A total of 280 households met the inclusion criteria and consented to participate in this study by responding to questionnaires. The collected data were analysed using descriptive statistics, the computation of the Household Food Insecurity Access Scale (HFIAS), and the Instrumental Variable Poisson model. The study’s results revealed that only 29% of the respondents were food secure, 36% were mildly food insecure, 27% were moderately food insecure, and 8% were severely food insecure. The Instrumental Variable Poisson model results revealed that cultivated green spaces, wealth index, gender, education level of the head of households, and grants had a negative correlation with household food insecurity. On the other hand, non-cultivated green spaces, local trees, age, marital status, number of dependents, and monthly income positively correlated with food insecurity. Given the existence of trees and green spaces in eThekwini, there is potential for food security solutions to be formed around both cultivated and uncultivated green spaces to promote sustainable access to food and nutritious diets in low-income households. Policy interventions should adopt an approach that encourages the incorporation of foods from both cultivated and uncultivated trees and green spaces in people’s diets.


Introduction
The South African food system faces wicked challenges; the country has the highest rates of wealth inequality in the world, and there are growing concerns about poverty [1,2]. It is difficult to transition to a healthier, more sustainable, and equitable food system in and informal employment; and (iv) ecosystem services necessary for agriculture and food production now and in future. The contributions of trees and green spaces to food and nutrition security should be considered vital and incorporated in the policies that are aimed at achieving Sustainable Development Goals 2 and 15, both of which are linked to food and nutrition security as well as trees and green spaces [18][19][20].
Therefore, there is a need to develop a food system that is good for the people and the ecosystem by formulating holistic solutions to solve the food insecurity issue. One such move is to advance sustainable food systems by including all vegetation-cultivated and uncultivated-in people's diets to reduce the pressure on farming to solve food insecurity. Figure 1 illustrates how trees and green spaces, directly and indirectly, contribute to food security. all. They provide (i) nutritious and diverse foods (which include, nuts, leaves, tubers, insects, and mushrooms) and feed for livestock; (ii) energy for cooking; (iii) both formal and informal employment; and (iv) ecosystem services necessary for agriculture and food production now and in future. The contributions of trees and green spaces to food and nutrition security should be considered vital and incorporated in the policies that are aimed at achieving Sustainable Development Goals 2 and 15, both of which are linked to food and nutrition security as well as trees and green spaces [18][19][20]. Therefore, there is a need to develop a food system that is good for the people and the ecosystem by formulating holistic solutions to solve the food insecurity issue. One such move is to advance sustainable food systems by including all vegetation-cultivated and uncultivated-in people's diets to reduce the pressure on farming to solve food insecurity. Figure 1 illustrates how trees and green spaces, directly and indirectly, contribute to food security. Africa's population is predominantly rural and relies strongly on trees and green spaces for their livelihoods. However, trees and green spaces' role in alleviating poverty and contributing to food security is still under scrutiny and has not been explicitly explored [20]. Scholars have dismissed increasing crop production through conventional means as a critical solution to food security [18,21]. Achieving food security goes beyond placing control in the hands of the farmers [22][23][24]. Most studies [23,25] have focused more on the impact of staple crops on food security; hence, there is limited empirical research on the actual contribution of all trees and green spaces to food security. Incorporating cultivated and uncultivated green spaces into the food system is still emerging, globally. However, in the recent past years, many research publications and two major global reports have drawn attention to the contributions of trees and green spaces to food and nutrition security [26]. This is leading to a better recognition of these roles by international organisations, such as the United Nations Committee on World Food Security (CFS) and the Food and Agriculture Organisation (FAO).
Therefore, this research seeks to contribute to this growing knowledge base on how cultivating and uncultivated green spaces can lead to sustainable food security.
This study aims to assess the impact of local trees and green spaces on household food security. It hypothesises that trees, green spaces, and food and nutrition security are linked. Such linkages help to plan interventions that could improve food security in lowincome and poverty-stricken communities. Africa's population is predominantly rural and relies strongly on trees and green spaces for their livelihoods. However, trees and green spaces' role in alleviating poverty and contributing to food security is still under scrutiny and has not been explicitly explored [20]. Scholars have dismissed increasing crop production through conventional means as a critical solution to food security [18,21]. Achieving food security goes beyond placing control in the hands of the farmers [22][23][24]. Most studies [23,25] have focused more on the impact of staple crops on food security; hence, there is limited empirical research on the actual contribution of all trees and green spaces to food security. Incorporating cultivated and uncultivated green spaces into the food system is still emerging, globally. However, in the recent past years, many research publications and two major global reports have drawn attention to the contributions of trees and green spaces to food and nutrition security [26]. This is leading to a better recognition of these roles by international organisations, such as the United Nations Committee on World Food Security (CFS) and the Food and Agriculture Organisation (FAO).
Therefore, this research seeks to contribute to this growing knowledge base on how cultivating and uncultivated green spaces can lead to sustainable food security.
This study aims to assess the impact of local trees and green spaces on household food security. It hypothesises that trees, green spaces, and food and nutrition security are linked. Such linkages help to plan interventions that could improve food security in low-income and poverty-stricken communities.
The current study report is structured into five sections. The introductory section is followed the methodology section, which includes a description of the study area, data collection methods, and the conceptual framework. The third section presents the research findings, focusing on the contribution of trees and green spaces to household food security. The fourth section discusses the findings of interviews, focus groups, and empirical frameworks. The conclusion and recommendations are presented in the last section.

The Study Setting
This study analyses how trees and green areas contribute to food and nutrition security in communities within the eThekwini Municipal, a portion of the province of KwaZulu Natal [26]. The EThekwini Municipal Area (EMA) encompasses the communities of Osindisweni and Maphephetheni, where data were collected. The two study sites were specifically chosen because they are similar and are thought to be biologically varied locations with a lot of vegetation (forests, grasslands, woods, and bushlands) [27,28]. Africans mostly populate the communities and are associated with severe poverty, high unemployment, ecological damage, and food and nutrition insecurity [29]. This has demonstrated the necessity of several socio-economic programs, which use research to try and eradicate poverty and restore ecosystems while also assuring food security [30,31]. The distribution of green areas at the KwaZulu-Natal, South Africa, study sites of Osindisweni and Maphephetheni is depicted in Figure 2 below.

The Selection of Respondents
For the initial sampling phase, Osindisweni and Maphephetheni Uplands were specifically chosen owing to their levels of poverty, food insecurity, ecosystem projects, and opportunities available to them. A total of 244 households were included in the sample after it was calculated using the Raosoft sample size computation from 2004 (90% confidence level).
Stratified sampling was then done where the population was split into two strata, each represented by a separate area, Osindisweni and Maphephetheni. As a result, the sample size for each study area was calculated depending on its proportion to the overall sample size.
Sample sizes for Osindisweni and Maphephetheni were 41 and 203, respectively, and the sample size derived was a total of 244 households. However, the enumerators managed to collect data from 280 households, due to the willingness of the respondents to participate in the study. The last and final sampling stage was random sampling, each study area had households, and the respondents were selected randomly on the ground.

Research Design
This paper's research data were obtained between June and December 2021. Standardised household questionnaires were used to collect data. The questionnaire included questions regarding specific subjects, such as demographic and employment information. It had open-ended and closed-ended questions to gauge respondents' attitudes and opinions regarding the value of trees, green spaces, and food and nutrition security. Face-to-face interviews were conducted during the surveys-one individual from each household was interviewed by a trained enumerator. All ethical requirements were followed before, during, and after the data collection. Three multilingual (isiZulu and English) enumerators administered the questionnaires.

Data Analysis
The Statistical Package for Social Sciences (SPSS) version 27 analysis was used to manage (capture, code, and clean) the data after they had been collected. The socio-economic makeup of the respondents and the research areas was described using descriptive analysis.
To determine whether there were significant correlations between trees and green spaces, household food security, and nutrition, the Poisson regression analysis was then done. The HFIAS indices were calculated to measure the communities' household food security status.

The Selection of Respondents
For the initial sampling phase, Osindisweni and Maphephetheni Uplands were sp cifically chosen owing to their levels of poverty, food insecurity, ecosystem projects, an opportunities available to them. A total of 244 households were included in the samp after it was calculated using the Raosoft sample size computation from 2004 (90% con dence level).
Stratified sampling was then done where the population was split into two strat each represented by a separate area, Osindisweni and Maphephetheni. As a result, th sample size for each study area was calculated depending on its proportion to the overa sample size.
Sample sizes for Osindisweni and Maphephetheni were 41 and 203, respectively, an the sample size derived was a total of 244 households. However, the enumerators ma aged to collect data from 280 households, due to the willingness of the respondents

Computing the HFIAS
The Household Food Insecurity Access Scale, a globally recognised food measurement tool, was used in the food security assessment (HFIAS). The HFIAS was used to assess the "access component of household food insecurity" using data from the previous four weeks (month). The HFIAS instrument was developed by Coates et al. [32] and this scale contains about nine questions about household food access, and the responses are usually provided by the household head [32,33]. The HFIAS score indicates the extent to which a household experienced food security or insecurity four weeks before the study was conducted. The computation of an HFIAS score involved combining the codes for each item that dealt with the frequency of occurrence for that particular household. The HFIAS score is between 0 and 27. A household with a score higher than the national average may be facing food insecurity. A household with a lower-than-average score may also be considered food secure [32]. The HFIAS score is determined using the equation shown below: Average HFIAS score = Sum of HF IAS in the sample Total number of HFIAS scores in the sample

The HFIAS Survey Questions
The participants were asked the HFIAS's nine questions. The aim of the survey was to determine whether participants had experienced any difficulties in obtaining food in the previous four weeks. The questions were divided into three sections that demonstrated an increasing level of severity of food insecurity (Question 1), inadequate food quality (Questions 2-4), and insufficient food intake (Questions 5-9). The participants were asked to specify how frequently the situation occurred, i.e., if it occurred rarely (once or twice) or never, occasionally (three to ten times), or frequently (more than ten times in the past month). Table 1 below shows the frequencies and percentages of food insecurity situations.

The Instrumental Variable Poisson Model
The HFIAS was used to assess food security. Following that, an Instrumental Variable Poisson model was chosen because it appropriately captures the dependent variable's count nature. The model is widely used to model non-negative outcome variables and to represent count outcomes. The instrumental variable model estimates the parameters of a Poisson regression model with endogenous regressors [34], to estimate the model's parameters, instrumental variables that do not correlate with the error term are used in this case. These instrumental variables are related to the endogenous variables but not the model's error term. Either additive or multiplicative error terms may be used to specify the model. In the error form representation of the exponential conditional mean model, the dependent variable, FS i , is a function of exogenous covariates, X i , endogenous covariates, Z i , and error term, ∈ i . An instrumental variable can enter either additively or multiplicatively, as shown in Equations (1) and (2), respectively [35].
Following Ndlovu et al. [34], three indices, cultivated trees and green spaces, uncultivated trees and green spaces, and local trees, were developed for analysis. The outcome demonstrates that the indicators' dependability is suitable for examining how trees and green spaces affect households' food security. The indices created were then employed as independent variables in an Instrumental Variable Poisson model, among other factors.

Socio-Demographic Profiles of the Respondents
The findings of the study show that more women than men took part in the study, 81.6% and 73.5% in Osindisweni and Maphephetheni, respectively. As a result, women likely head the majority of homes in Osindisweni and Maphephetheni. Most homesteads' household composition was nuclear; over 90% of families in the study area were nuclear and less than 6% of the homesteads were polygamous. In Osindisweni, 65.8% of respondents and in Maphephetheni, 57.4% were married. For both communities, most respondents had completed secondary education. Less than 25% of respondents, however, had obtained tertiary education, either in the form of a certificate or a diploma. Fewer than 10% of respondents said they had no formal schooling. This information is displayed in Table 2 below, and the findings support the view that Osindisweni and Maphephetheni have restricted access to chances for formal education, employment, and income-generating opportunities.
Trees and green spaces contribute indirectly to food security through employment and income generation. However, in the areas covered by the current study, tree-related jobs did not represent the main source of livelihoods. This was not expected given that most households in the community were poor and vulnerable. Most respondents in the study area had one source of income, and only 4% of the households employed several cash income sources for their livelihood. As depicted in Table 1 above, cash income from social grants and employment accounted for the largest contributor to household income, followed by income from family, friends, and jobs. The other income contributors were income from crops, fruit, livestock, traditional medicine, and fuelwood sales. More than 70% of the respondents in Osindisweni and 40% in Maphephetheni had at least one individual receiving social grants. Most included child support and old age grants, which amounted to ZAR 270 monthly. Some households had members involved in both full and part-time employment. However, the proportion of employed individuals was very low. The people employed were engaged in low-paying jobs as suggested by the monthly income bracket of the respondents ranging from ZAR 0-15,000. Only 3% of the respondents earned more than ZAR 9000, and most of them are employed as teachers in schools surrounding the communities. The results indicate that 10% of the respondents had no income, which was concerning because all the respondents were heads of households.

Household Food Insecurity Prevalence
The HFIAS measuring households' access to food revealed that only 29% of the respondents were food secure, 36% were mildly food insecure, 27% were moderately food insecure, and 8% were severely food insecure in the study areas, as illustrated in Figure 3 below. These results show that the majority of the households were food insecure.
These results are in line with the results shown in the HFIAS responses in Table 1, 6.4% to 62.5% of the study participants answered "yes", 11.8% to 31.1% answered "rarely", 2.1% to 25.7% answered "sometimes", and 1.8% to 5.7% answered "often". The affirmative response "yes" had the highest frequency and percentage, which showed mild to moderate forms of food insecurity such as to "worry about not having enough food", "being unable to eat the kinds of food you prefer", and "eating smaller of fewer meals per day". The response "often" had the least frequency and percentage and showed severe forms of food insecurity; it was worrying to note that some households can go a whole day and night without food.
Household food insecurity prevalence.

The Instrumental Variable Poisson Model
According to the Instrumental Variable Poisson model results, eleven variables had a statistically significant influence on household food insecurity, as shown in Table 3 below. Five significant variables were found to be negatively correlated with household food insecurity, implying that increasing either would result in a decrease in food insecurity, suggesting an increase in household food security level. Six of the significant variables, on the other hand, were significant in the positive direction, implying that an increase in any of these variables would be associated with a decrease in food security, suggesting an increase in food insecurity or a less food-secure household.

The Instrumental Variable Poisson Model
According to the Instrumental Variable Poisson model results, eleven variables had a statistically significant influence on household food insecurity, as shown in Table 3 below. Five significant variables were found to be negatively correlated with household food insecurity, implying that increasing either would result in a decrease in food insecurity, suggesting an increase in household food security level. Six of the significant variables, on the other hand, were significant in the positive direction, implying that an increase in any of these variables would be associated with a decrease in food security, suggesting an increase in food insecurity or a less food-secure household.

Discussion
Food insecurity is an issue that many low-and middle-income nations, including South Africa, face [2]. In this study, more than half of the respondents were food insecure as shown in Figure 2. When asked about the sources of their regular meals, more than half of the respondents stated that they had bought from markets and spaza shops. Furthermore, 12% and 28.8% from Osindisweni and Maphephetheni had harvested from home gardens and green spaces around them, and friends and relatives had assisted less than 25% of the respondents. Out of interest, respondents were asked about their most preferred sources of food, and some respondents agreed that trees and green spaces diversify their diets, and they also confirmed that they preferred food harvested from trees and green spaces instead of store-bought food. Thus, it is evident that trees and green spaces contribute to household diets.
The regression analysis results (Table 3) revealed that cultivated green spaces had a negative and significant impact on household food insecurity. The analysis shows that as the use of cultivated green spaces for food increases by a unit, household food insecurity decreases by 0.097, with all other factors held constant. These results also show that households that harvested food from trees and green spaces had more food access than households that did not utilise green spaces for food. The probable explanation for these results might be that 88.4% of the households had trees in their yards and produced food (vegetables, fruits, tubers (sweet potato and amadumbe-Colocasia esculenta), legumes, and maize) in their home gardens and farms, which helped to diversify diets. Some households also indicated that they had livestock in their homesteads which was also the primary source of protein in the form of meat and eggs. In the same vein, the study by Cheteni et al. [36] found that households that grow and cultivate plants for food improved their food security status because of increased production, income, and consumption. Likewise, in South Africa, based on the various studies conducted in different regions of the country, cultivating plants for food through gardening and farming has increased livestock production, crop diversification, and intensification [36][37][38]. These outcomes, in turn, contribute to assured food security.
Uncultivated green spaces and local trees had a negative and significant relationship with household access to food. This implies that food access from uncultivated trees and green spaces (forest foods) increased food insecurity. This result was surprising and unexpected because forest foods provide many benefits to households in their vicinity, such as food, medicine, and fuelwood for food preparation. This result might have been because of a lack of scientific knowledge on the available food resources in forest trees and green space, use potential, changes in diet, and their harvesting and preparation techniques.
The coefficient of the age of the household was positive and statistically significant in influencing the food insecurity status of the households. This result implies that as the age of the household head increases by a unit, the household food insecurity increases by 0.006, with other variables in the model held constant. A decrease in the age of the household head results in a reduction in food insecurity. The younger the household head, the lesser the food insecurity; this might be because younger people have access to more incomegenerating opportunities, which enable them to have funds to access food, especially in markets. However, this finding is contrary to Ndlovu et al. [34] and Poppy et al. [20], who found that household food insecurity decreases as age increases because the older population knows about cultivating their own food and the harvesting and preparation of indigenous foods.
The coefficient of the wealth index for both agricultural and non-agricultural related assets was negative and statistically significant in influencing the food insecurity status of households. This could be attributed to the fact that tools and assets such as the radio and television enhance technology adoption, awareness, and spread of information resulting in increased food production and access to income-generating opportunities which all improve access to food [34,38]. About 20% of the respondents confirmed that they sometimes get information on radio and television that helps them produce more food or access food. Table 3 also show that the coefficient of income was positive and statistically significant. This implied that the higher the income, the higher the household food insecurity. A positive relationship was not expected because income improves access to and food availability. However, this could be attributed to the increase in the cost of living and food in South Africa; hence, more income will not always lead to improved access to food and nutrition. Ndlovu et al. [39] found a positive relationship between income and food access. On the other hand, Gebreyesus et al. [38] conducted food security studies in the Eastern Cape province in South Africa and found that households where the head of household is employed have a much higher income and are more likely to have acceptable diets and a better food and nutrition status. The results show that, to a lesser extent, there is a reliance on trees and green spaces as well as other goods and services that the ecosystem provides. Less than 2% of the households as shown in Table 1 were generating income through the sale of tree products such as herbs, medicine, and fuelwood. However, the total income from those sales was minimal (less than ZAR 1000 a month). This differs from Angelson's [40] study findings which revealed that income from tree products is often higher or equivalent to income from agriculture. This suggests that trees and green spaces can sustain household income and reduce pressure on the government to provide social grants [40][41][42]. The money the government uses for grants can be saved for other development purposes. Hence, emphasis on deriving income from products from trees and green spaces can encourage households to plant more trees on their homesteads and conserve the green spaces surrounding them [36,40]. If this is not applied, the quantity of tree products that households purchase may continue to increase, leading them to higher food insecurity given their low income.

The results in
This study established a negative and significant association between social grants and the insecurity status of the household. This implies that receiving grants reduces food insecurity. This result was expected because social grants in South Africa have improved food consumption and the poverty status of most households. However, poor diets continue to be a massive challenge for South Africans, especially those living in rural communities; this has been further exacerbated by the increased cost of living [36]. Contrary to these findings, Cheteni et al.'s [36] research in eThekwini revealed poor diets and nutrition insecurity among households reliant on social grants for income. Most respondents who had access to social grants were food insecure because the money received was too little to cover the household food needs. Similarly, Kamanga [41] conducted a study on rural livelihoods and reported that most households' main income source was grants. There is a deep dependence on grants, which is explained by the fact that many people in the study are unemployed or earn very low cash income. Withdrawal of social grants would worsen the food insecurity situation of many residents. Table 3 shows that formal education had a negative and statistically significant influence on household food insecurity. The marginal analysis shows that as a household's formal education increases by one unit, household food insecurity decreases by 0.072 while all other variables in the model remain constant. The relationship was anticipated because formal education increases human capital and knowledge, increasing access to opportunities and contributing to the availability and accessibility of diverse foods at all times. This finding is consistent with the findings of a study conducted in South Africa's Limpopo province by De Cock et al. [43], who found that households with educated household heads were less likely to be food insecure, and that promoting formal education in rural households can significantly improve food security levels.
The variable of gender had a negative relationship with a household's food insecurity. This points out that gender affects access to nutritious food. Compared to single-head households, nuclear and female-headed households had better diets and higher incomes. Kennedy et al. [44] notes that this is the case since women are responsible for food preparation and spend the majority of household income on high-quality, protein-rich foods [44]. Additionally, Taruvinga et al. [45] discovered that households with female heads had healthier diets than those with male heads [45] because they were involved in sourcing and preparing food.
The marital status coefficient was both positive and statistically significant in influencing household food insecurity. The result's implication is explained by the fact that spending on nutritious food is inversely correlated to the family head's marital status. This finding conflicts with that of Maziya et al. [46], who discovered that the head of a household's marital status was inversely correlated with food insecurity and that there is a lower likelihood of food insecurity if married people head a household [46]. However, the findings of this study are consistent with those of Ndlovu et al. [39], which found that unmarried household heads are more likely to have food security than married household heads. This could be explained by the fact that married households typically have larger households because they have more children and, therefore, more people to feed. As a result, there is much less food accessible to household members.
The regression analysis results also showed a significant and positive correlation between the number of dependents and household food insecurity. This suggests that food insecurity increases with household size. Large households have six or more dependents, while small households have any number of dependents under six. In this study, each family had an average of eight members; most households were large and food insecure. The fact that most of the household members were not employed may have contributed to these findings. This outcome is consistent with the research of Cheteni et al. [36], who discovered that having many economically inactive family members increases the likelihood of having unhealthy diets and nutrition. A large household size exerts pressure on the availability and accessibility of food because there are more food and non-food expenses.

Conclusions and Recommendations
Trees and green spaces play an important role in sustaining livelihoods and promoting food security in South African communities; however, their role has not been widely accepted. This study's findings indicate that trees and green spaces are contributing considerably to the food and nutrition security of the local households studied. Thus, there is need for deliberate investments and public awareness campaigns recognising the significance of all trees and green spaces to food and nutrition security. That would most likely reduce the pressure put on cultivated cash crops to ensure food security.
There is potential for food security solutions to be formed around cultivated and uncultivated green spaces to promote sustainable access to food and nutritious diets in low-income households. Improvement in education around the benefits of trees and green can also assist in alleviating food insecurity.
Given the foods found in trees and green spaces around the communities, more research is required to develop knowledge in this area and an in-depth evaluation of household diets. This can help understand which nutritious foods can be cultivated and harvested from trees and green spaces, given that they have lower impacts on the ecosystem. Policy interventions should include promotion of incorporation of foods from both cultivated and uncultivated (wild) trees and green spaces in the diets of the local households. That could be done through skills development, raising awareness on the importance of consuming balanced diets, and promoting the sustainable use of natural resources. Further research could also help identify the possibility of cooperation between extension workers, nutritionists, policymakers, and other stakeholders to develop holistic, sustainable solutions to the food and nutrition insecurity problem. This can create an enabling environment for better policies and interventions, and improve households' access to nutritious, high-quality, sustainable foods.

Limitations
The study only collected data that answered the study's objectives; for future research, it is important to broaden the scope and look at how technologies improve food access from trees and green spaces. Furthermore, the study focused on one province, Kwa-Zulu-Natal. To draw lessons from one province to another, the research should be expanded to cover other areas in KwaZulu-Natal and all provinces in South Africa. The current study was conducted to establish the contribution of trees and green spaces on household food security. Further studies could be done to assess the effect of trees and green spaces on household nutrition security and income. Future research could also focus on food preferences, coping strategies, and assisting food-insecure households in accessing nutritious foods from trees and green spaces.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available upon request from corresponding authors.