The government’s prioritization of the sub-tropical fruit and nut sectors and the promotion of small-scale farmer integration in HVC markets raise concerns for sustainability and require investigation into whether farmers can sustain these HVC systems over time. There is a need for farmers to invest in various land use and management aspects that will affect the long-term sustainability of the farming systems. The study examines the land resource used under different commodities i.e., HVCs in the Vhembe district of Limpopo and how the land characteristics are driving the sustainable production of these commodities under different land ownership and management systems. In order to achieve the overarching goal of sustainability that will ensure long term food security in the country there is need to explore what land characteristics will support production.The two farming systems in the study i.e., small-scale and large-scale are recog-nised as systems due to the multi-variable nature of the processes within the farms and the non-linear interconnectedness that exists between them. The commodities grown in these farming systems are recognized as HVC based on the definition provided by. The four drivers of production i.e., land, labour, capital and enterprise drive the two farming systems and the pathway of agricultural enterprise i.e., production, management, marketing and value adding for each of the systems which have the potential to produce the same outcome in different ways. Future scenarios for sustainable agriculture within the different commodities must consider how production can be sustained under the two main farming systems. The land resource and its use are arguably one of the most important drivers of sustainable agriculture as they highlight numerous environmental interactions that can either be detrimental or beneficial to the sustainability of farming systems. Land is a highly politicized issue in the South African context due to historical allocation of land based on race by the previous government prior to democracy in 1994. There is a need for an emphasis on scale in the analysis of these two main South African farming systems in order to accurately investigate what land variables will drive sustainable agriculture in the country.

Land characteristics namely, farm size and ownership, topography, soil type and fertility, threats and hazards, mobile vertical rack water sources and irrigation, and the impact of climatic and its variability on the farming systems have been selected and are analysed between the two farm sizes and within three different commodities. These land characteristics are further analysed alongside two production characteristics, i.e., income and yield in order to determine to what extent they can drive sustainability.The study took place in the Vhembe district which is the northern most district municipality of the Limpopo Province in South Africa . It shares borders with Zimbabwe and Botswana in the north-east and Mozambique in the south-east through the Kruger National Park. The Vhembe district is one of five district municipalities in the Limpopo Province. It has an area of 2,140,708 hectares of which 247,757 hectares is arable land. The Vhembe district is comprised of four local municipalities: Thulamela, Mutale , Musina and Makhado. The South African governance structure regards the composition of local municipalities as towns and their surrounding rural areas. The main towns within the district are Thohoyandou, Malamulele, Musina and Makhado respectively for the four municipalities Thulamela, Mutale , Musina and Makhado. The district covers a geographical location that is largely rural. According to agriculture is the key contributor to employment and livelihoods in the district. Seventy percent of the farming activities in the district are attributed to smallholder agriculture and the remaining 30% is commercial agriculture. According to the Vhembe District Municipality’s Local Economic Development Strategy in 2019 the district produces 4.4% of South Africa’s total agricultural output, 8.4% of the country’s sub-tropical fruits and 6.3% of its citrus. The district is situated in a semi-arid area, is frequently affected by dry spells that often develop into drought with severe water shortages from May to August. Most commercial farmers in the district rely on irrigation systems for farming whilst the smallholder farmers generally depend on seasonal rainfall which typically falls from November to March..

The average rainfall ranges from 246 mm to 681 mm per annum. Soils in the district are variable and tend to be sandy in the west, but with a higher loam and clay content towards the east. The soils developed on basalt, sandstone and biotite gneiss and some have low inherent soil fertility. Maize is the predominant cereal grain grown in the district among smallholder farmers. Leguminous crops like groundnuts, Bambara nuts and cowpeas are also grown by smallholder farmers as well as vegetable crops which include spinach, cabbage, tomatoes and onions. These are grown for the farmers’ own consumption with any surplussold to neighbours or relatives. Rain-fed crop yields are generally poor due to low and erratic rainfall coupled with poor fertility. Commercial horticulture farming is well established in the south eastern side of the district  and includes stakeholders which grow mangos, litchis, bananas, avocados, citrus, pecan and macadamia nuts.A study was conducted using an analysis of primary and secondary data to identify and characterize large and small-scale farming systems of three tree crops, in the Vhembe district. The analysis was aimed at highlighting the connectivity of interactions between the farming systems in terms of the four drivers of production. The focus of the paper is on land as a driver of production. Secondary data were collected from: the official subtropical crop database obtained from the local Department of Agriculture located in the town of Thohoyandou, climate data from the Institute for Soil, Climate and Water , land type and soils data from the Agricultural Research Council , peer reviewed research papers and related books. The target population was a combination of large-scale commercial and small-scale farmers within the district. Based on the FAO definition of farming systems which informs the study, three different enterprises based on commodities grown at farm sites were chosen: 1) macadamia nut farming systems 2) mango farming systems and 3) avocado farming systems. Farming systems where initially broadly characterised based on available information extracted from the local Department of Agriculture database. The database is comprised of data on the farm location , farm size , gender of farmer, farmer name and telephone number. A purposive sampling method was employed in choosing four criteria for site selection, these were used in the study namely commodity, size of the farm, location of the farm and gender of the farmer. This information was available for six subtropical commodities, namely bananas , litchis , avocados , mangos , macadamia nuts  and citrus . According to the database there are a total of 1121 documented subtropical crop farmers in the Vhembe district. According to the database the three commodities selected in the study were the most commonly grown commodities in the district.

Avocados were selected based on the willingness of the farmers to participate in the study based on a preliminary interaction with the farmers at a local study group meeting. Macadamias were selected based on their significance to the South African agricultural economy as high value export crops. The next selection criterion was size. Farms were selected using a systemic random sampling procedure to ensure that there was equal representation of farms within the size categories that exist in the database, these were namely small-scale  as the study required both farmers with smallholdings and larger holdings.The next selection criterion was location. Farms were selected to ensure that there was equal representation of all 4 local municipalities that comprise the Vhembe district municipality namely Mutale, Makhado, Thulamela and Musina. Lastly, vertical grow table the farmers’ gender was also used as a farm selection criterion. A random number generation method was used to ensure that there was equal representation of both genders across the farms. The process of random number sampling involved allocating a number to the farmers selected from the database based on the above criteria, writing down the numbers and placing them in a container. Numbers were then randomly picked out of the container to make up a total of 12 farms. These 12 farms were comprised of 4 samples for each of the 3 commodities spread across the 4 local municipalities with 2 small-scale and 2 large-scale farms as well as an even mixture of male and female farmers. Once this initial site selection was made, a more detailed characterization of the three farming systems was done based on the significance of the 4 drivers of production i.e., land, labour, capital and enterprise. Primary data were obtained from in-depth interviews that were conducted with farmers in selected farm locations within the Vhembe district. A snowball sampling technique was used in response to this with the aim of maintaining the same sample size initially selected. The results of the snowball sampling produced samples that differed vastly in number to those from the initial sample selection: macadamia nuts , mangos  and avocados . A total of 19 farmers were selected to participate in interviews based on their availability and willingness to participate.Interviews were conducted over two visits to the Vhembe district in October and November 2020. Ethical clearance was obtained from the local Department of Agriculture and the University of the Witwatersrand, protocol number: H19/09/26. The researcher, together with a field assistant, who acted as an interpreter from the Mutale local municipality conducted the interviews. Interviews were conducted in the Vhenda language. Interviews were conducted face-to face with farmers on-site at the farm locations and recorded.A questionnaire was the main instrument of data collection made up of closed and open-ended questions to collect quantitative and qualitative data. Close-ended questions were used to elicit background information and for statistical information regarding the four drivers of production in the context of the selected farm sites. Open-ended questions were used to enable respondents to provide longer answers.

The questionnaire was divided into 4 sections: 1) land 2) labour 3) capital and 4) enterprise as drivers of production.Descriptive statistics were used to analyse quantitative data. This was done by calculating averages, percentages and standard errors. Chi squared and student t-tests were used to compare the means of different farming systems and between the two farm sizes. Pearson Correlation coefficients were used to establish the relationships between selected land and production variables within the two farm sizes and across the three different commodities which were then used to highlight possible relationships. Qualitative data were analysed using thematic analysis using information from participant responses to open ended questions addressing issues relating to land variables between the two farm sizes and across the different commodities. The responses were categorized into predominant themes and percentages calculated. The resulting themes were triangulated with the quantitative data to explain the phenomenon.Data from the 19 participants were collated. Of the 19 participants there were 7  macadamia nut farmers, 4  mango farmers and 8  avocado farmers. Of the 7 macadamia nut farmers, 3  were classified as large-scale and 4  as small-scale. The average farm size amongst large-scale macadamia farmers was 576 hectares compared to 5 hectares amongst small-scale farmers. Of the 4 mango farmers only 1  was classified as a large-scale farmer on a 15 hectare farm and 3  as small-scale farmers. The average farm size amongst small-scale mango farmers was 4.7 hectares. The 8 avocado farmers were comprised of 2  large-scale farmers and 6  small-scale farmers. The average farm size amongst large-scale avocado farmers was 806 hectares compared to 4.9 hectares amongst small-scale farmers. The average tonnage for large-scale macadamia nut farmers was 290 tons compared to 2.7 tons amongst small-scale macadamia farmers while the average yield was 0.5 tons per hectare for both large and small-scale macadamia farmers. The only large-scale mango farmer interviewed had a tonnage of 4.5 tons with a yield of 0.3 tons per hectare compared to an average tonnage of 3.3 tons amongst small-scale farmers and average yield of 1.1 tons per hectare. The average tonnage amongst large-scale avocado farmers was 408 tons compared to 4.9 tons per hectare amongst small-scale farmers. Large-scale avocado farmers had an average yield of 0.7 tons per hectare while small-scale farmers had an average yield of 1.1 tons per hectare. Correlations between farm size and yield will be addressed later in the discussion of results under the heading crop yields.Results revealed that 79% of participants were male while 21% were female.