Description of the study area
The study was carried out in two out of nine provinces of South Africa, Limpopo, and Mpumalanga. These provinces were selected for the study because they consist of smallholder farmers who are heavily involved in the production of indigenous crops. These provinces consist of smallholder communal farmers who depend on agriculture and livestock farming as their livelihood source. Limpopo is located in the Northern part of South Africa, covering 125 754km2 of the area, which is 10, 2% of the total area of the country. Limpopo consist of several ethnic groups distinguished by race, language, and culture: Sepedi (57%), Tsonga (23%), Venda (12%), Afrikaners (2.6%) and the English (1/2%). This province consists of 5, 8 million population situated in five districts known as Mopani, Vhembe, Capricorn, Waterberg, and Sekhukhune [5]. Mpumalanga province consist of different ethnic groups distinguished by culture and language: Siswati (27.67%), Zulu (24.14%), Xitsonga (10.42%), Isindebele (10.10%), Afrikaans (7.24%), Sesotho (3.47%), and English (3.12%) These are the districts where the data collection for the study was carried out as people in these areas are heavily reliant on agriculture. This is also evident in that 89% of the population in this province works within the agricultural sector. Mpumalanga is one of the provinces heavily reliant on agriculture as it produces a wide variety of fruits, vegetables, cereals, tea, and sugar. The production of these crops plays a significant role in the economic growth and development of Limpopo province (Hlatshwayo et al. [20]). It also comprises 167 existing irrigation schemes with small-scale farmers operating on these schemes [8]. These small-scale irrigation schemes have about 10,150 farmers with an average individual land holding of about 1.5 hectares per farmer.
Mpumalanga province is formerly known as Eastern Transvaal. This province is located in the north-eastern part of South Africa. It is bounded by Limpopo province to the North and Swaziland to the east of KwaZulu Natal. It covers about 6.5% of the country’s land area. It also consists of a 4.04 million population where 72% of the population is heavily involved in agriculture [8]. The overall rainfall received by Mpumalanga is 1000 mm annually and also experiences warm weather conditions as it is 665 above sea level. It also produces indigenous crops such as amaranth, cowpea, African eggplant, okra, and pumpkin. The other foods farmers produce in Mpumalanga include corn, sugar, cotton, groundnuts, potatoes, other vegetables, and a wide variety of fruits, including oranges and mangoes in the subtropical Lowveld and peaches in higher elevations. Mpumalanga is also involved in the production of dairy cattle, beef, and wool production.
Data collection method and technique
The study used a quantitative method to collect data on key food and nutrition security indicators. A multistage stratified sampling technique was used to select the participants for the study. This technique divides the population into homogenous, mutually exclusive strata. It also enables individuals in the population to have equal chances of being selected to participate in the study. This method increases the trustworthiness of match rate estimates, inexpensive, and quite easy to implement. It also allows a large sampling of the population, which helps researchers to draw an accurate conclusion about the study, whereas small sampling produces less accurate results, which then lead to wrong conclusion being made about that particular population [21].
In each site, farmers were grouped based on the similar characteristics they share such as socio-economic factors, household size, institutional factors, and sales. The study used secondary data that was collected by the South African Vulnerability Assessment Committee (SAVAC) in 2016. The total number of respondents that participated during the research in Limpopo and Mpumalanga provinces was 1520. Data collected from these two provinces were analysed in a statistical manner using a software program known as Statistical Package for Social Science (SPSS).
The determinants of indigenous leafy vegetables were modelled using seemingly unrelated regression (SUR) model, which assumes that the error terms between components are expected to be correlated. This model is an efficient estimator of coefficients compared with ordinal least square (OLS) especially when the error terms between equations are correlated. SUR model estimates more than two equations simultaneously. The parameters of each equations take information provided by the other equation into account [22]. This model is employed in this study because when interdependence between dependent variables was assumed, the common underlying determinants were well estimated using simultaneous equations of SUR model.
Furthermore, this model was employed in the study because of its three main advantages, firstly, it was used to gain efficiency in estimation by combining information on different equations. Secondly, it imposes test restrictions that involve parameters in indifferent equations. Thirdly, it leads to improved tests of hypothesis of regression coefficient and other parametric values [22]. This model has been used by scholars such as [23,24,25,26] to determine the production and utilization level of indigenous leafy vegetables.
With respect to the study, demographic factors and socio-economic factors were modelled to attain a comprehensive understanding of the extent at which different variables have affected the acceptance of these leafy vegetables. Furthermore, consumers’ determinants of indigenous leafy vegetables are multidimensional; their acceptance relies on a combination of characteristics such as household size, gender of the household head, education level, main economic activity, wage/salary, HIV status, social grants, and irrigation type. Therefore, SUR was seen as ideal model to analyse the study.
The determinants of indigenous leafy vegetables were modelled using seemingly unrelated regression (SUR) model, which assumes that the error terms between components are expected to be correlated. SUR model estimates more than two equations simultaneously. The parameters of each equations take information provided by the other equation into account [22]. This model is employed in this study because when interdependence between dependent variables was assumed, the common underlying determinants were well estimated using simultaneous equations of SUR model. This model has been used by scholars such as [23,24,25,26] to determine the production and utilization level of indigenous leafy vegetables.