Competition for the use of public open space in low-income urban areas: the economic potential of urban gardening in Khayelitsha, Cape Town

Abstract

Urban agriculture is a contested issue in the larger South African debate on urban poverty alleviation. This paper investigates the economic viability of urban agriculture and informs the debate on the optimal use of open space in Khayelitsha. It compares the economic performance of the Scaga community garden in the low-income township Khayelitsha, predicted in a 1998 study by Fermont et al., with empirical results of a similar study by Fleming in 2003. It concludes that urban agriculture in Khayelitsha is potentially economically viable, subject to certain conditions being satisfied. However, as a land use, urban agriculture competes with housing, ecological corridors, the stormwater management system and nature areas and reserves.

1. Introduction

This paper reviews an ongoing South African debate on urban cultivation in the context of a City of Cape Town investigation (2002) into the future use of open space in Khayelitsha, Cape Town. Given the designation of Khaylelitsha as a priority Presidential Urban Renewal Project (together with the adjacent Mitchell's Plain), the use of large tracts of open space is an important planning issue in the light of their potential for such competing uses as housing, nature conservation areas and ecological corridors, recreation and the stormwater management system. This paper does not, however, debate the rival claims of these different uses, arguing simply that the economic feasibility of agricultural activities in these relatively large public areas is an important alternative for consideration. Thus far the debate is inconclusive.

It is recognised that by far the greatest proportion of urban agriculture is undertaken as a survival strategy by individual households, generally in backyards to augment household real income. It is also recognised that urban agriculture is but one of many diverse livelihood strategies enabling households to manage risk and reduce vulnerability. This micro-scale agriculture enables the very poor to supplement their diet, reduce dependency on bought food and gain social capital through shared effort. Virtually none of the produce finds its way to local markets. Profit is not the motive. This paper does not engage with the difficulties faced by the marginalised poor in their attempts at such high-intensity urban farming. It argues simply that, the role of micro-scale agriculture notwithstanding, the potential does exist for larger scale commercial or semi-commercial agricultural enterprise to provide employment, cash income, and possibly food security for full-time farmers within low-income urban areas and, further, that such agriculture is a viable use for previously underutilised public open space in a low-income housing area.

The paper analyses the economic performance of existing urban agriculture in Khayelitsha and the potential for dedicating public open space to this activity. It compares the economic performance of the Scaga community garden predicted in a 1998 study by Fermont et al. with the empirical results of a study by Fleming in 2003. Scaga is a formal community garden of approximately 0,25 ha that was initiated by the Cape Town-based non-governmental organisation (NGO) Abalimi Bezekhaya (Xhosa: ‘the planters of the home’) in 1997 and works on organic principles (Small, 2002). A distinction must be drawn between community gardens and home food gardens, which have a different dynamic, being much smaller in scale and more periodic or transitory. Community gardens usually cover a larger planting area, are run by more than one member or household, are more established or permanent, and aim to produce for profit.

The South African Council for Scientific and Industrial Research regards urban agriculture as any form and scale of agricultural activity that happens within the boundaries of the urban environment. It can include horticulture, floriculture, forestry, aquaculture and livestock production (Visser, 2004: 3). Urban agriculture is practised to a varying extent throughout the developing and developed world (de Haen, 2002: 1). It has been a familiar feature of many African cities since the 1970s, but only in recent decades has it become more common in these cities (May & Rogerson, 1995: 167; Rogerson, 2000: 67). It has been advanced as an important contribution towards poverty alleviation in the urban areas of poor countries, notwithstanding extensive debate on the subject.

1.1 The urban agriculture debate in South Africa

This section provides a brief overview of the benefits and problems of urban agriculture in South Africa and a theoretical interpretive framework for analysing urban cultivation in Khayelitsha.

The rise of urban agriculture in South Africa largely coincided with the abolition of influx control in 1986. The resulting sharp increase in poor households migrating into cities and an economy in recession, weakened by sanctions (May & Rogerson, 1995: 168), led to the development of large poor townships in and around South African cities. It is here that urban cultivation predominantly occurs, as the ‘expansion of urban agriculture is inseparable from growing vulnerability by many kinds of households as a result of declining livelihood opportunities in the formal economy’ Rogerson (2000: 68). The South African administration appears to tentatively support urban agriculture (Lachance, 1993: 8), but views diverge among researchers about what role it can play in South Africa.

An important facet of the debate is the potential of urban agriculture to contribute to poverty alleviation. The main argument, as put forward by a number of authors (May & Rogerson, 1994; Rakodi, 1995: 415, 418; Karaan & Mohamed, 1998: 73, 81; Drescher et al., 2000: 6; Visser, 2004: 4), is that urban agriculture can increase the number of livelihood strategies available to the poor. According to this argument, growing food increases a household's food security and allows for savings on food expenditure in often tight budgets and could also improve the nutritional value of the food consumed. However, this has been disputed (Beaumont, 1990: 35; Drescher et al., 2000: 8, 10). Producing a surplus of food and non-food crops and marketing them can, ideally, generate disposable income and business skills, thus economically empowering the poor (Visser, 2004: 15ff.). By implication, urban cultivation is taken to be economically feasible.

Researchers have also identified a large number of benefits not directly relating to poverty, and not necessarily dependent on the economic viability of urban agriculture. These include the recreational potential and aesthetics of green gardens, ecological services to cities, environmental education, social empowerment such as the increased self-esteem a thriving and productive garden provides, and social interaction and the strengthening of community ties (Eberhard, 1989: i; Beaumont, 1990: 14; May & Rogerson; 1995: 166; Karaan & Mohamed, 1998: 81; Slater, 2001: 635, 648; Visser, 2004: 4, 5). Slater (2001: 643, 644, 647) draws attention to the way a food garden can benefit women: as a symbol of stability and an emotional refuge from fear and violence, and by giving them a stronger role in the household through having more control over household food consumption.

However, the debate identifies several problems of urban agriculture's role in poverty alleviation. Webb (1996: 153) cautions against overestimating its potential, arguing that an emphasis on urban agriculture as the main livelihood strategy can lock people into a cycle of poverty by denying them the skills required in a modern economy. Although referring in their report to the rural poor, Berdegué & Escobar's argument is equally valid for the urban poor. They contend that ‘the agricultural path out of income poverty is directly relevant only for a share of all the rural poor: those who have access to sufficient land and other assets relevant to agricultural production, and who operate in contexts that provide the correct incentives for those assets to be sufficiently productive over long enough periods of time, can leverage the household out of poverty. Attempting to force households and communities who lack these resources and contexts to base their development on agricultural production, is to push them deeper into a poverty trap’ (2001: 14).

Significantly, the poorest of the poor rarely engage in urban agriculture (Beaumont, 1990: 34; May & Rogerson, 1995: 171, 173; Webb, 1996: 181; Rogerson, 2000: 68; Castillo, 2003: 342), for a number of reasons. They often lack access to land on which to cultivate, owing to the ‘gatekeeping’ practices of the less poor and more established people, who can control access to existing resources and exclude poor newcomers (Rogerson, 2000: 69). The poorest also have few surplus resources to invest in tools and seeds and necessary household expenditure while the crops are growing, as income from agriculture is not equally distributed across the year. Webb (1996: 273) notes that urban cultivation is generally unable to meet basic household needs and does not constitute the most important livelihood strategy among the urban poor. While this suggests that the potential of urban agriculture as a livelihood strategy for the poorest may be limited, it does not disprove the potential of urban agriculture for the slightly less poor. Indeed, many argue that, while urban agriculture cannot be the sole solution to poverty, it should be considered as one of many possible ones (May & Rogerson, 1995: 172; Rakodi, 1995: 415; Thorgren, 1998; de Haen, 2002: 1; Visser, 2004: 20).

A lack of economic viability is probably the most significant challenge to the poverty alleviation potential of urban cultivation. Several have determined that urban agriculture provides very low returns (Eberhard, 1989: i; Rakodi, 1995: 415; Fermont et al., 1998: 34; Drescher et al., 2000: 6; Slater, 2001: 635). It could become uneconomical for the urban poor if they can access higher alternative incomes or buy food cheaper than they can produce it. The reality is that residents often do not have access to higher alternative incomes, so urban agriculture becomes one of many survival strategies, a viable option particularly in the context of high rates of unemployment when people have excess labour to dispose of. Better-off gardeners might be content to subsidise their food gardens for the social or cultural benefits they derive from cultivation but poor urban farmers cannot indulge these motives. The environmental conditions of the cultivation area also have a major influence on yields and production costs.

The potential of urban cultivation in terms of poverty alleviation and empowerment in South Africa is thus still uncertain. However, it is possible that it may have significant benefits both in economic and social terms if practised under the right circumstances.

Slater (2001: 636) criticises the ‘narrowly defined, utilitarian’ economic frameworks within which some urban agriculture research has been carried out. Generation of money income is not the only test of viability where household total real income would have to be considered. Nevertheless, economic analysis of existing projects can play an important role in establishing the potential of urban cultivation and identify circumstances that can improve or erode its feasibility. Knowing whether it can be economically viable has important implications for policy formulation. If it is not economically feasible per se but is considered to provide other significant benefits to social and urban development, policies subsidising and supporting urban agriculture could be designed to promote it. In this case, urban agriculture would reflect the characteristics of a development programme (Webb, 1998: 105).

2. Economic performance of the SCAGA garden in Khayelitsha

This section provides an overview of conditions in Khayelitsha and research that has been conducted on urban cultivation. It includes a comparison of the Fermont et al. (1998) and Fleming (2003) studies. It determines significant considerations concerning the use of public open space in Khayelitsha for urban agriculture in order to inform the planning debate.

2.1 Urban agriculture in Khayelitsha

Khayelitsha is approximately 30 km southeast of the Cape Town Central Business District within the Cape Metropolitan Area (Figure 1). It covers about 41 km² in the area known as the Cape Flats and is home to an estimated 500 000 residents, many of whom derive from the Eastern Cape (Minor et al., 2004: 1–3).

Figure 1: Khayelitsha in the City of Cape Town. Source: Adapted from City of Cape Town, 2002

Khayelitsha's potential for supporting urban cultivation in community gardens is largely determined by its biophysical and social characteristics, which can be analysed using the following criteria:

•	Availability of land, often deemed the main constraint on urban cultivation (May & Rogerson, 1995: 174; Karaan & Mohamed, 1998: 76; Thorgren, 1998; Drescher et al., 2000: 8; Visser, 2004: 9).
•	Availability of labour and the relative attractiveness of returns from urban agriculture in comparison to alternative livelihood opportunities.
•	Environmental conditions favourable for urban cultivation.
•	Status of social support facilities and networks, and the potential of urban agriculture to substitute for these.
•	Presence of greening programmes and the importance of urban agriculture in greening the area.
•	The remainder of Section 2.1 describes Khayelitsha with respect to these criteria.

2.1.1 Land

Karaan & Mohamed (1998: 80) identified insufficient land available for cultivation as the most significant constraint on urban agriculture in Khayelitsha. Several large areas zoned as ‘open space’, ranging from 1,3 ha to approximately 16 ha, exist within Khayelitsha (Minor et al., 2004: VII). Many of these are associated with the stormwater management system. Studies such as those by the Cape Metropolitan Council (1996: xiii, 52ff.) and OvP Associates (1999: 102–10) have suggested various types of urban agriculture as a possible use of public open space. Additionally, there are approximately 50 vacant school sites in the township, and most of the approximately 50 existing schools do not use their grounds fully (Minor et al., 2004: 71). Urban cultivation on existing or vacant school property is therefore an option. Gardens have been established on church property (Meadows, 2000: 41). On its periphery, Khayelitsha is abutted by approximately 3 000 ha of nature areas, partly used for grazing (Minor et al., 2004: VII, 55–63).

However, the use of open space in Khayelitsha for urban cultivation is restricted by competing demands. Formal housing is urgently needed, as about two-thirds of residents live in informal dwellings (Census, 2001). The degree to which new houses will occupy vacant land in Khayelitsha, thereby decreasing the amount of land available for urban agriculture, is uncertain. And given that the peripheral areas are proclaimed nature reserves or recognised as areas of high ecological significance (Minor et al., 2004: 62), farming on the edges of Khayelitsha might not be viable as it could pose a threat to existing ecosystems.

2.1.2 Relative attractiveness

Several researchers have found that urban agriculture provides very low returns. But it is important to consider the high unemployment in the area and the relative attractiveness of returns from cultivation compared to other income options. Rogerson (1998: 173) shows that, in absolute terms, metropolitan areas carry the greatest burden of poverty in South Africa. This is reflected in Khayelitsha's social conditions. Approximately half of the township's residents live in poverty. According to the 2001 Census, 47 per cent of households had an annual income of R9 600 or less (Census, 2001), and Hauser calculated the poverty line at about R12 000 per household per year (2003: 28). While 70 per cent of residents are of working age, only 35 per cent of them are formally employed (Census, 2001). Employment creation by the informal sector is important, but hard to quantify (Minor et al., 2004: 41). In this light, even low returns from urban agriculture may yield a meaningful additional income.

2.1.3 Environmental conditions

Characteristics such as soil and climate are major determinants of the productivity of cultivation. Khayelitsha has relatively harsh environmental conditions, which have been identified as a major constraint on urban agriculture in the area (Eberhard, 1989: ii; Thorgren, 1998; Fermont et al., 1998: 60). The soils are sandy, nutrient-deficient and do not easily retain water. The area also experiences hot summers and strong winds which increase the water evaporation rate and necessitate extra shelter for plants. As a result, the costs of cultivation are likely to be higher and yields poorer compared to areas with more favourable conditions.

2.1.4 Social conditions

The potential social benefits from urban agriculture are particularly important in an area deprived of social facilities and support. Khayelitsha has a significant shortage of social support facilities (Minor et al., 2004: 32). Partly offsetting this, common ethnic bonds (Census, 2001) provide a form of social cohesion and mutual support, particularly for newly arriving migrants. Urban agriculture networks and groups may be able to exploit these existing bonds and strengthen the community network.

2.1.5 Greening status

Potential greening effects from urban agriculture are particularly important in a harsh physical environment. Khayelitsha lacks greenery in general, and many open spaces within the urban areas are degraded and neglected (Minor et al., 2004: 62). Greening programmes initiated by the municipality seem to have failed in the past (Hauser, 2003: 76). If urban agriculture can fulfil its reputed potential for greening and improving an area (May & Rogerson, 1994: 96; Eberhard, 1989: i), its impact could be significant in Khayelitsha.

Since 1983 (Karaan & Mohammed, 1998: 67), urban agriculture has become established in Khayelitsha. There are a number of studies that focus on urban cultivation specifically in this area, although they draw different conclusions about its feasibility and importance. Karaan & Mohamed present an optimistic view. They found a high demand for urban cultivation in the area and note that ‘with rising unemployment and increasing vegetable prices, home food gardening plays a vital part in securing household food security’ (1998: 70, 80). They concluded that food gardens in Khayelitsha would prove successful as a development strategy if they form part of a broader policy framework.

Other studies are more cautious. Eberhard's (1989: 4) economic analysis suggested that urban agriculture in Khayelitsha could yield a profit of R1,60 per m² per year, with a return for labour of R0,25 per hour. He concluded that urban agriculture was unlikely to become feasible in Khayelitsha, as poverty levels were not severe enough to make its low returns attractive (Eberhard, 1989: i). Supporting this point, Beaumont (1990: ii) found little interest in vegetable cultivation among the community living on the Cape Flats. She attributed this to perceived low economic benefits, the lack of organisational resources, competing pressures for land and other priorities such as employment and housing. Thorgren (1998) concluded in her paper that the ‘future of urban agriculture in Cape Town is uncertain and difficult to predict’ owing to the poor soil and the lack of land and water, especially in informal settlements.

However, the studies by Fermont et al. and Fleming of the economic potential and performance of the Scaga community garden conclude that urban agriculture in Khayelitsha can be viable, despite difficulties. Fermont et al. (1998: 38) recognised that, owing to the unfavourable environmental conditions, a garden will initially make a loss. They estimated, however, that a dedicated gardener could overcome these problems and eventually earn R1 200 per month from a 600 m² mixed vegetable garden. And Fleming (2003) established that the gardeners earned R600 per month, 40 per cent of estimated maximum attainable earnings, confirming Fermont et al.'s thesis that a community garden could be economically viable. Both studies, however, observed that demand for gardening in Scaga was lower than expected and subject to the availability of higher earnings opportunities elsewhere. For instance, members who found paid jobs regularly left the community garden (Fermont et al., 1998: 27; Fleming, 2003: 1). These studies are discussed in detail in Sections 2.2 and 2.3.

In summary, Khayelitsha's difficult social conditions may make larger-scale community urban agriculture a possible alternative livelihood strategy, or at least a major component of such a strategy. The contradictory conclusions of studies conducted on urban agriculture in Khayelitsha show significant uncertainty about the potential of urban cultivation. The harsh environmental conditions appear to be the major constraint, while a potential land shortage may be alleviated through the use of land not suitable for housing development, such as school and church property and land that forms part of the stormwater management system.

2.2 Study by Fermont et al. (1998)

In 1998, Fermont et al. studied the performance of the Scaga garden after one year of operation. At that time, the group of gardeners consisted of 16 members: 15 women and one man. In its first year of operation, three members had left the gardening group and two new ones had joined (Fermont et al., 1998: 23). The study found that the initial costs of setting up a medium-scale garden in Khayelitsha were high, amounting to more than R36 000 for the Scaga garden in 1997 for the installation of basic infrastructure such as fencing, irrigation and tools (Fermont et al., 1998: 32, 38). More importantly, operating costs for water, fertiliser and seeds far outstripped income in the first year (Table 2: year 1). These high costs were mainly due to poor soil fertility, low irrigation efficiency and the need to buy seeds externally. As a result, profits were negative after one year of operation.

As part of their study, Fermont et al. modelled the future economic performance of the Scaga garden. The main drivers of future performance in the model were improving soil fertility and the resulting increase in crop yield and decrease in fertiliser cost as well as own-production of most seedlings, negating the need to buy them at a higher cost. They based their prediction (1998: 32, 40) of the future performance of the Scaga garden (Table 2: years 2–10) on the following assumptions:

•	Cultivated space: 1 800 m2.
•	Crop rotation: two rotations per year.
•	Full potential income level of R48 250 is reached in year 5.
•	Irrigation requirement: 783,1 kilolitre (kl) per year at 100 per cent efficiency.
•	Total water requirement: 1 566,2 kl per year due to 50 per cent irrigation efficiency.
•	Water cost based on 1998 Tygerberg municipality price of R1,75 per kl excluding VAT.
•	Seedlings produced by Scaga from year 3 onwards at a cost of R2 200 per year.
•	Fertiliser: cattle manure at a cost of R85 per ton applied to the soil at the rate specified in Table 1 to first achieve and then maintain the organic matter content of the soil at 2 per cent, the threshold necessary to grow vegetables (Fermont et al., 1998: 62).

The predicted decline in costs for fertiliser and seeds, coupled with increasing yields, was expected to lead to an improvement in the cost margin of 299 per cent in the first year, rising to a forecasted stable 14 per cent in year 10 (Table 2 and Figure 2).

Figure 2: Actual and predicted economic performance for Scaga Garden 1997–2006, Fermont et al. 1998 ). Source: Fermont et al. (1998) author's design

Table 1: Predicted cattle manure requirements in t/ha 1998–2006

	Year 2	Year 3	Year 4	Year 5	Year 6	Year 7	Year 8	Year 9	Year 10
	1998	1999	2000	2001	2002	2003	2004	2005	2006
Fertiliser	280	280	230	160	140	120	120	100	100
Source: Fermont et al. (1998: Figure 6.3), own design.
Note: Units in tons/hectare.

Table 2: Actual and predicted performance of Scaga Garden 1997–2006, Fermont et al. ( 1998 )

	Actual	Predicted
	Year 1	Year 2	Year 3	Year 4	Year 5	Year 6	Year 7	Year 8	Year 9	Year 10
	1997	1998	1999	2000	2001	2002	2003	2004	2005	2006
Income	4 600	12 000	25 000	42 500	48 250	48 250	48 250	48 250	48 250	48 250
Water	750	3 075	3 075	3 075	3 075	3 075	3 075	3 075	3 075	3 075
Fertiliser	6 800	4 284	4 284	3 519	2 448	2 142	1 836	1 836	1 500	1 500
Seeds	6 200	4 500	2 200	2 200	2 200	2 200	2 200	2 200	2 200	2 200
Cost	13 750	11 859	9 559	8 794	7 723	7 417	7 111	7 111	6 775	6 775
Cost margin	299%	99%	38%	21%	16%	15%	15%	15%	14%	14%
Profit	−9 150	141	15 441	33 706	40 527	40 833	41 139	41 139	41 475	41 475
Source: Fermont et al. (1998: Table 4.5, Figure 4.2), author's calculations and design.
Note: All prices are in rand.

Fermont et al. (1998) concluded that the Scaga garden should approximately break even in its second year of operation and become profitable in year 3. Profits were predicted to improve continually until year 9, mainly driven by reduced fertiliser costs. At a garden size of 1 800 m², the profit in year 10 would amount to approximately R2 per m² per month. The forecast is analysed in more detail in Section 2.4.

2.3 Study by Fleming (2003)

Six years after the inception of the Scaga garden, and five years after Fermont et al.'s study, Fleming analysed the garden's performance. He found several changes in its operation. Among others things, the gardening group's composition had changed significantly, suggesting that circumstances in the operation had not been very stable. The group now consisted of eight members: five women and three men. None of the original members remained (Fleming, 2003: 1). Other new conditions were (Fleming, 2003: 1, 2):

•	Cultivated space: 2 480 m2.
•	Crop rotation: three rotations per year on 24 of 31 vegetable beds, two rotations on the remaining beds.
•	Organic matter content of soil: 2 per cent (Fermont et al., 1998: 62; Fleming, 2003: 1).

These data reflect the period 1 November 2002–1 November 2003. During the course of this year, Fleming monitored vegetable output, and the consumption and cost of water, fertiliser and seeds. The economic performance of the Scaga garden is recorded in Table 3. Operating costs absorbed about 16 per cent of income, the biggest cost being water. As a result, the profit of R35 480 translated into earnings of approximately R1,2 per m² per month. The performance is analysed in more detail in the following section.

Table 3: Actual economic performance of Scaga Garden 2003, Fleming ( 2003 )

	Year 7
	2003
Income	41 998
Water	2 839
Fertiliser	2 579
Seeds	1 100
Cost	6 518
Cost margin	16%
Profit	35 480
Source: Fleming (2003), author's design.
Note: All prices are in rand.

2.4 Comparison of the Fermont et al. and Fleming studies

The assumptions and conditions on which Fermont et al. (1998) and Fleming (2003) based their calculations are not identical. In order to make the studies comparable, data have to be normalised. Table 4 specifies the assumptions that required adjustment.

Table 4: Differing assumptions in the Fermont et al. and Fleming studies

Assumption/condition	Fermont	Fleming
Cultivated area	1 800 m^2	2 480 m^2
Crop rotation	2 per year	3 per year on 77% of planted area*
Base year for prices	1998	2003
Source: Fermont et al. (1998), Fleming (2003), author's calculations and design.
Note: *Three crops per year on 24 of 31 vegetable beds.

Fermont et al.'s prediction has been adjusted in Table 5 to account for the following:

•	Increase in size of the planted area from 1 800 m2 in Fermont et al. to 2 480 m2 in Fleming [Fermont's adjusted income (year 7): 48 250*2480/1800 = 66 478, assuming all inputs and outputs increase equally].
•	Increase in crop rotation from two in Fermont et al. to three per year on 77 per cent of planted area in Fleming. This results in a 50 per cent increase in production on this area, or approximately 39 per cent overall production increase 50% more production due to three crop rotations on 77% of area implies a 39% total increase (50%*77% = 38,71%).
•	Inflation of prices for inputs and outputs: • Prices charged for vegetables in Fermont et al. appear to be based on 1998 prices. They were appreciated to 2003 values using the CPI Food Metro (Consumer Price Index Metropolitan areas – Food (CPI Food Metro) (StatsSA, 2004) [CPI Metro Food increased 49,2% from December 1998 to December 2003. Fermont's adjusted income (year 7): 92 211*1,492 = 137 579,8]. • Costs for fertiliser and seed inputs in Fermont et al. appear to be based on 1998 prices. They were appreciated to 2003 values using the CPI Food Metro (StatsSA, 2004) [CPI Metro Core increased 44,5% from December 1998 to December 2003]. • Cost for water in Fermont et al. was based on 1998 Tygerberg Municipality prices of R1,75 per kilolitre excluding VAT (Fermont et al., 1998: 40). These prices were adjusted to the average 2003 City of Cape Town water price of R3,42 per kilolitre (Fleming, 2003: Schedule 2) [Fleming (2003): City of Cape Town water price was R3,09/kl to July 2003 and R3,75/kl thereafter].

Table 5 compares the adjusted forecast from Fermont et al.'s study for year 7, equivalent to 2003 conditions, with the actual performance of the garden in 2003, as found by Fleming.

Table 5: Comparison of predicted and actual performance of Scaga garden in year 7

	Predicted performance – adjusted (Fermont et al., 1998)		Actual performance (Fleming, 2003)
	Year 7 (2003)	% of cost	2003	% of cost
Income	137 579		41 998
Water	10 235	48	2 839	44
Fertiliser	5 070	24	2 579	40
Seeds	6 075	28	1 100	17
Cost	21 381		6 518
Cost margin	16%		16%
Profit	116 198		35 480
Source: Fermont et al. (1998), Fleming (2003), author's calculations and design.
Note: All prices are in rand at 2003 values.

The actual income of approximately R42 000 for the Scaga garden in 2003 constitutes 30 per cent of the income predicted by Fermont et al., adjusted for differences in the size of the garden, crop rotation and inflation. Water accounts for about half the operating cost and is the largest component in both studies. The remaining distribution of expenditure is more heavily skewed towards fertiliser than predicted. The operating cost margin is about 16 per cent in both cases. For a given income, Scaga was therefore as profitable as predicted. However, as the actual income level of Scaga lies significantly below the prediction, the implied earnings per m² per month lie at only R1,2 rather than R3,9 as expected. The predicted and actual results are compared in the following two sections in terms of income and costs.

2.4.1 Income

2.4.1.1 Crop yield

In Table 6, the actual 2003 crop yield from Fleming's study is compared to the assumed crop yield on which Fermont et al. [yields were derived from Appendix 2, i.e. 40 cabbages per bed of 10 m² imply a yield of four cabbages per m²] based their estimates for year 7 income [yields were derived from Schedule 3, i.e. 2 360 cabbages grown on 13 beds of 80 m² each implies a yield of 2 360/13/80 = 2,26 cabbages per m²]. This is then contrasted with the yields recommended in the Vegetable Production Guide Wallchart by Hygrotech. Only crops that appear in both studies were included in this table [crops that appear in only one of the studies are leeks, onions, peas, beans, pumpkin, mealies (corn) and celery]. All yields observed by Fleming in 2003 were below those predicted by Fermont et al. by between 1,6 and 23,9 plants per m², or 37–89 per cent. The average yield of the crops in Table 6 was 60 per cent lower than forecast. Compared to yields suggested by Hygrotech, Fermont et al.'s expectations were in most cases within those boundaries, showing that the assumptions were reasonable. Actual yields only fell within Hygrotech's suggested boundaries for green peppers and cauliflowers, indicating that the yields achieved in 2003 were indeed poor.

Table 6: Comparison of expected and actual crop yields of Scaga garden

Vegetable	Yield: plants per m^2
	Fermont expected	Fleming actual	Difference (%)	Hygrotech suggested
Cabbage	4,0	2,3	−43	3–4
Lettuce	12,0	4,5	−63	5,5–10
Carrots	36,0	13,5	−63	150–250
Onions	27,0	3,1	−89	50–80
Spinach	12,6	1,7	−86	15–34
Beetroot	31,5	7,9	−75	17–35
Green peppers	6,0	3,3	−45	2–5,5
Brinjals	6,0	2,0	−67	2,2–3,5
Turnips	22,5	13,3	−41	20–40
Broccoli	4,3	1,8	−59	2–4
Cauliflower	4,3	2,7	−37	1,6–4
Potatoes	3,4	1,6	−54	Not available
Source: Fermont et al. (1998: Appendix 2), Fleming (2003: Schedule 3), Hygrotech, author's calculations and design.
Note: Units in plants/m^2.

2.4.1.2 Total production

The 2003 total production estimated from Fleming's study [derived from Schedule 3, i.e. total income from cabbage was given as R5 895 for the year, while 2 320 cabbages were linked to a revenue of R5 795. As a result, the average price per cabbage was R5 795/2 320 = R2,498. This implies an approximate total number of cabbages of R5 895/R2,498 = 2 360] is compared in Table 7 to the predicted produce on which Fermont et al. based their estimates for year 7 income [derived from Appendix 2, i.e. 40 cabbages per bed on 30 beds in summer and 25 beds in winter implies 40*30 + 40*25 = 2 200 cabbages in a year]. The Fermont et al. numbers have been adjusted for an increase in plot size and crop rotations from two to three a year in 77 per cent of the garden [total production adjusted for increase in garden and rotation: 2 200*2 480/1 800*1,387 = 4 204,2 cabbages]. The total produce is presented both in saleable units such as fruits or bunches and in kilograms, calculated from the number of produced units and the average weight per unit as observed in a Cape Town supermarket. The estimated total production in terms of marketable units was approximately 30 per cent of predicted produce (45 175 versus 149 479 units). In terms of weight, the actual production was about 25 per cent of predicted produce (15 897 kg versus 63 493 kg), as there are no crops in Fleming's study corresponding to peas, beans, pumpkins and mealies, which are relatively heavy per unit. These low levels of production correspond approximately to the overall income level in 2003 at 30 per cent of predicted income. As vegetables are the only income source of Scaga, the amount of vegetables produced and for sale will directly influence the amount of income. While average yields fell short by 60 per cent relative to forecasts, total production fell short by 70–75 per cent relative to the predicted amount. This suggests that in addition to poor harvests beds have lain idle. Crop shortfalls because of fallow beds would not be reflected in the yield.

Table 7: Comparison of total production

Vegetable	Unit	Unit weight (kg)	Fermont adjusted predicted produce (units)	Fleming produce (units)	Fermont adjusted predicted produce (kg)	Fleming produce (kg)
Cabbage	Fruit	1,50	4 204	2 360	6 307	3 540
Lettuce	Fruit	0,40	459	720	183	288
Carrots	Fruit	0,07	17 200	7 560	1 204	529
Leek/onion	Bunch	0,30	–	480	–	144
Onions	Fruit	0,17	15 996	1 960	2 639	323
Spinach	Bunch	1,00	36 602	4 339	36 602	4 339
Beetroot	Fruit	0,25	10 836	4 402	2 709	1 100
Peppers	Fruit	0,16	1 368	7 137	219	1 142
Brinjals	Fruit	0,28	1 368	2 344	376	645
Turnips	Fruit	0,25	12 040	3 180	3 010	795
Broccoli	Fruit	0,30	414	705	124	211
Cauliflower	Fruit	0,65	740	1 951	481	1 268
Potatoes	Fruit	0,20	27 940	7 484	5 588	1 497
Peas	Kg	1	344	–	344	–
Beans	Kg	1	428	–	428	–
Tomato	Fruit	0,14	17 888	535	2 504	75
Pumpkin	Fruit	2,00	206	–	413	–
Mealie	Fruit	0,25	1 445	–	361	–
Celery	Fruit		–	Some	–	Some
Total Produced			149 479	45 175	63 493	15 897
Source: Fermont et al. (1998: Appendix 2), Fleming (2003: Schedule 3), Hygrotech, author's calculations and design.

2.4.1.3 Prices

The 2003 prices implied by Fleming (derived from Schedule 3) are compared to the crop prices on which Fermont et al. (2003: Appendix 2) based their estimates for year 7 income. The prices have been appreciated to 2003 prices using the CPI Food Metro, which shows an increase of 49,2 per cent from December 1998 to December 2003 [i.e. appreciated price of a cabbage: R1,5*1,492 = R2,238]. Only the vegetables priced in both studies were included in Table 8. On average, the prices in Fleming were 7 per cent higher than expected prices in Fermont et al. after appreciating them to 2003 values. However, with the large crop shortfall identified above, the increase in prices has a small overall impact on the 2003 income of the Scaga garden. It only narrows the gap between estimated and actual income by about 3 per cent [the difference between predicted and actual income is 137 579 – 41 998 = 95 581. Without the 7% price increase 2003 income would have been 41 998 – 7% = 39 058. The difference between prediction and income would then have been 98 521. The price increase has therefore narrowed the difference by (95 581 – 98 521)/ 98 521 = –3%.].

Table 8: Comparison of prices assumed by Fermont et al. and implicit in Fleming

Vegetable	Fermont et al. 2003 prices (R)	Fleming (R)	Change (%)	Vegetable unit
Cabbage	2,24	2,50	11,4	Head
Lettuce	1,50	0,15	−90,0	Head
Carrots	0,56	0,42	−25,7	Fruit
Onions	0,50	0,63	25,4	Fruit
Spinach	1,50	2,59	73,4	Bunch
Beetroot	0,50	0,43	−14,3	Fruit
Green peppers	0,60	0,59	−1,6	Fruit
Brinjals	0,50	0,90	81,5	Fruit
Turnips	1,50	0,38	−74,8	Fruit
Broccoli	5,98	5,87	−1,9	Kg
Cauliflower	1,50	2,30	54,1	Head
Potatoes	0,30	0,45	49,7	Fruit
Source: Fermont et al. (1998: Appendix 2), Fleming (2003: Schedule 3), author's calculations and design.
Note: All prices are in rand at 2003 values.

2.4.2 Costs

2.4.2.1 Water

The actual 2003 water use and price from Fleming's findings is compared to the assumed water use on which Fermont et al. based their estimates for year 7 costs. The water input predicted by Fermont et al. has been adjusted for the larger size of the planting area and increase in crop rotations from two to three per year in 77 per cent of the garden. The water cost in rand per kilolitre in Fermont et al. was based on 1998 prices and has been adjusted to the 2003 City of Cape Town municipal water price (Fleming, 2003: Schedule 2) (Table 9).

Table 9: Comparison of water used

Study	Scaga water used/assumed (kl)	Water use adjusted (kl)	Efficiency (%)	kl 100% efficiency	R/kl 2003 prices	Total cost adjusted (R)
Fermont et al.	1 566	2 993	50	1 496	3,42	10 235
Fleming	8 743 for 9 months	11 657	15–28	1 736–3 224	0,24	2 839
Source: Fermont et al. (1998), Fleming (2003 and personal communication), author's calculations and design.
Note: Fermont et al.'s adjusted kl: 1566 kl*2480/1800*1.387 = 2992.6 kl accounting for increased planting area and crop rotation. Fleming's adjusted kl for a whole year: 8743 kl*12/9 = 11 657.3 kl.

2.4.2.2 Amount of water used

The water use efficiency in the Scaga garden is much lower than assumed in Fermont et al.'s study. The prediction of performance in year 7 assumes that the garden uses approximately 3 000 kl of water, adjusted for a larger planting area and higher crop rotation. This implies an irrigation efficiency of 50 per cent and irrigation requirement of about 90 mm per m² per month (including 500 mm rain per year), in congruence with Scaga's performance in its first year (Fermont et al., 1998: 34, Appendix 3). In reality, however, the garden used more than 11 000 kl of water in 2003, or almost four times the adjusted forecast.

Increasing the efficiency of irrigation, which has dropped to between 15 and 28 per cent, could significantly reduce the use of water. According to Fleming (2003: 3, Schedule 4b), the monthly water requirement of the crop is between 100 and 150 mm per m² per month. This implies that in 2003 between about 360 [with 150 mm or litres/m² water required per m²/month, and 500 mm or litres/m² rain in the year, total irrigation] and 670 per cent excess water was used in the garden. Consequently, at an irrigation efficiency of 100 per cent, between 72 and 85 per cent of water could be saved. Such overwatering not only has cost implications but also rapidly increases the leaching of nutrients from the ground, damaging the crop yield (Fleming, 2003: 3).

2.4.2.3 Cost of water used

At 2003 prices, Fermont et al.'s water price of R3,42 per kilolitre is over 14 times more expensive than the price implicit in Fleming. The reason is that 47 per cent of all water used in 2003 was pumped from the well at the Scaga garden, at a cost of only R0,44 per kWh or R0,52 per kl [Fleming, 2003: Schedule 2: 4 147 kl of ground water pumped at a cost of R2 149, or 0,52 R/kl] for the electricity for the pump. The remaining 53 per cent of water used was city water, for which Scaga did not have to pay (Alan Fleming, Ecoagric, personal communication). The actual cost of the water is therefore unrealistically low in 2003.

2.4.4.4 Fertiliser

The actual 2003 manure input and price from Fleming's findings is compared to the assumed manure requirement and price on which Fermont et al. based their estimates for year 7 costs. The manure input and cost predicted by Fermont et al. have been adjusted for the larger size of the planting area, the increase in crop rotations from two to three per year in 77 per cent of the area [Fermont et al.'s adjusted manure: 21,6t*2 480/1 800*1,385 = 41,2t] and the increase in the CPI Core Metro by 44,5 per cent from December 1998 to December 2003 [Fermont et al.'s manure price of R85/t was increased by 44,5% (CPI Core Metro) to R122,8/t]. The total quantity of manure used in the garden in 2003 was approximately one-third of the adjusted amount predicted by Fermont et al. Although the 2003 cost of manure in rands per ton was about 50 per cent higher than the cost predicted by Fermont et al., the decrease in application of manure outweighs the increase in price, so that the total cost of fertiliser in 2003 is half of that predicted in Fermont et al. after adjustments. The quantity of manure required per bed was lowered in most cases at the beginning of Fleming's study because of the reassessment of fertiliser needs by agriculture expert Alan Fleming. This assessment was based on previous crop performance and the requirements of new crop to be planted. Fleming found that in most cases too much manure had been applied previously (Fleming, personal communication) (Table 10).

Table 10: Comparison of manure used

Study	Recommended manure (t/ha/year)	Recommended manure Scaga (t/year)	Manure use adjusted (t/year)	R/ton 2003 prices	Total cost Adjusted (R)
Fermont et al.	120 for 2 crop rotations	22	41	123	5 070
Fleming	57 for 3 rotations on 77%	14	14	182	2 579
Source: Fermont et al. (1998), Fleming (2003 and personal communication), author's calculations and design.
Note: Numbers calculated on the basis of the size of the garden: 120t/ha*0,18 ha = 21,6t (1800 m^2 = 0.18 ha) and 57t/ha*0,248 ha = 14.1t (2480 m^2 = 0,248 ha).

2.4.4.5 Seeds

The actual 2003 cost of seedlings from Fleming's findings is compared to the assumed cost on which Fermont et al. based their estimates for year 7. The seed cost predicted by Fermont et al. has been adjusted for the larger size of the planting area, the increase in crop rotations from two to three per year in 77 per cent of the garden, and the increase in the CPI Core Metro by 44,5 per cent from December 1998 to December 2003 [Fermont et al.'s adjusted cost of seedlings: R2 200*2 480/1 800*1,387*1,445 = R6 075]. The cost of seedlings again shows a large disparity between prediction and actual result (Table 11). In Fleming, this cost is a sixth of the adjusted forecasted value although both cases are based on the condition that Scaga produces most if not all of its own seeds and seedlings. The cost of own seedling production might be overestimated in Fermont et al. (1998: 34). However, Scaga did not produce all the seedlings needed. Its actual performance appears to be supported by subsidised seeds and seedlings, which the garden receives from Abalimi Bezekhaya at a cost of R5 per pack of seeds or R10 per tray of 200 seedlings (Fleming, personal communication). Fermont et al. (1998: Appendix 5) on the other hand specified the cost of a pack of seeds as R2,40 and the cost of a tray of 120 seedlings as R24. Appreciated to 2003 prices using the CPI Core Metro and adjusted to the same number of seedlings per tray, these numbers compare as in Table 12 [adjusted Fermont et al. price for a tray of 200 seedlings: R24*200/120*1,445 = R57,8]. It is clear that although the Fermont et al. price for a pack of seeds is somewhat less than Fleming's, the cost for a tray of seedlings assumed in Fermont et al.'s prediction far outweighs the cost implicit in the garden's actual 2003 performance.

Table 11: Comparison of seed costs

	Cost study (R)	Cost adjusted (R)
Fermont et al.	2 200	6 075
Fleming	1 100	1 100
Source: Fermont et al. (1998), Fleming (2003 and personal communication), author's calculations and design.
Note: Prices in rands.

Table 12: Comparison of input costs for seeds and seedlings

	Fermont et al. 2003 prices (R)	Fleming (R)
Pack of seeds	3,5	5,0
Tray of 200 seedlings	57,8	10,0
Source: Fermont et al. (1998:Appendix 5), Fleming (personal communication), author's calculations and design.
Note: Prices in rands.

2.4.3 Implications

The above sections show that the 2003 performance of the Scaga garden diverges significantly from the forecast in Fermont et al. (1998). This is mainly due to a significantly lower vegetable output and price variations, as Scaga does not pay a market price for its water and seedlings. Most importantly, it did not reach the predicted income because yield and total output were far below those assumed in Fermont et al.'s model. Actual production reached only about 30 per cent of predicted output. There are several reasons for this shortfall.

•	Yield, and consequently total produce, was depressed because seedlings were planted too far apart (Fleming, personal communication) thus generating fewer plants and fruit per m2.
•	The overwatering of the crops (360–670 per cent too much water) led to the significant leaching of nutrient from the already nutrient-poor soil. As a result, plant growth and yields were prejudiced. This effect was amplified by the choice of manure, which loses nutrients more quickly than compost, as a fertiliser (Fleming, 2003: 3; personal communication).
•	A number of problems not directly related to gardening practice occurred in 2003, which negatively affected crops and therefore yields. Mice damaged the cauliflower seedlings, birds virtually eliminated the pea harvest and an unexpected frost destroyed the bean crop (Fleming, 2003: 3). Nevertheless, these events could not be considered aberrations and would constitute ‘normal’ farming contingencies. The choice of pest control measures is restricted by Scaga's aim of growing organic vegetables.

The shortfall of total production was even more significant than the yield deficit. It coincided with a prolonged period of group conflict and people leaving the group because they found paid work (Fleming, personal communication; Meadows, 2000: 60). These developments at times delayed or prevented planting so that crops were sowed at a less favourable time or beds would lie idle. The loss of potential crops due to fallow beds is not reflected in the yield calculations. The increase in vegetable prices of 7 per cent has a small effect on income, since production levels are low.

The costs of the Scaga garden are considerably at variance with expectations, especially those for water. The irrigation efficiency was significantly below the 50 per cent efficiency assumed in Fermont et al., resulting in excessive use of water. Scaga's cost, however, was much lower than predicted, as the garden project only paid for the electricity needed to pump the ground water that provides 47 per cent of irrigation water. The remaining 53 per cent of water came from the municipal system and was free of charge in the period of Fleming's study.

The calculations in Table 13 show a more sustainable picture of the water cost at Scaga. The first row indicates the current situation. The second and third rows assume current water consumption levels under different payment regimes. Rows four and five assume 100 per cent water efficiency with different irrigation requirements (100 mm and 150 mm per m² per month, respectively).

Table 13: Sensitivity of 2003 profits to the payment for water (rands)

Scenario	Water cost	Total profit	Profit/m^2/month
Pay electricity only (current status)	2 839	35 480	1,19
Pay city water at city prices and electricity	23 969	14 350	0,48
Pay all water at city prices	39 868	−1 549	−0,05
Pay all water (100% eff, & 100 mm) at city prices	5 937	32 382	1,09
Pay all water (100% eff, & 150 mm) at city prices	11 026	27 293	0,92
Source: Author's calculations.

If the garden paid for its city water in addition to the electricity used to run the groundwater pump, the total water cost would reach R23 969 for 2003 [47% of water is groundwater at R0,52 per kl for electricity. Average price for city water in 2003 was R3,42. Thus the cost of 11 657 kl: 47%*11 657,33kl*R0,5181 + 53%*11 657,33 kl*R3,42 = 23 968,8]. This would reduce profits to R14 350 per year or R0,48 per m² per month. Should Scaga pay for all its water at a city rate, as assumed in Fermont et al.'s prediction, the water cost would reach R39 868 for 2003 11 657,3*3,42 = 39 868. In this case, profits would be negative, minus R1 549 per year, or minus R0,05 per m² per month.

It is therefore clear that Scaga's profit in 2003 was artificially bolstered by the extremely (and unsustainably) low cost of water. However, if Scaga had achieved a 100 per cent irrigation efficiency, even at a higher assumed water requirement of 150 mm per m² per month, water cost in 2003 would have been R11 026 3224*3,42 = 11 026, resulting in a positive profit of R27 293 per year or R0,92 per m² per month. This demonstrates that Scaga can be profitable at the current production level even if it pays municipal water prices, as long as the resource is used more efficiently. Paying for water might prove a powerful incentive for a more stringent irrigation regimen.

The costs for seeds and fertiliser were both below predictions. This may partly be due to the low total production and beds lying idle, so that less fertiliser and seedlings were needed. It would therefore be realistic to expect some increase in these costs with an increase in production although the effect on overall profitability may be negligible.

The above analysis shows that the Scaga community garden yielded a profit of R1,2 per m² per month to its members in 2003. Even though the current conditions under which this result is achieved, such as free water provision, appear unsustainable, the general profit level of approximately R1 per m² per month appears sustainable with relatively simple measures such as greater water efficiency. Fleming (2003: Schedule 4b) argues that Scaga could achieve a profit of R2,37 per m² per month. Fermont et al. (1998: 38) assumed a profit of R2 per m² per month for two crop rotations, which increases to R2,6 with an increase in crop rotations and to R3,9 per m² per month when taking into account inflation between 1998 and 2003.

2.5 Market gardening in Khayelitsha

This section outlines the main considerations concerning the use of public open space in Khayelitsha for urban agriculture.

2.5.1 Relative economic attractiveness of urban agriculture

In a study of eight community garden projects in the Cape Flats, Meadows (2000) found that six of the eight projects stated that income generation and food production were their primary motivation for urban agriculture. This section discusses urban agriculture as a means of income generation only and is not to intended to skew the analysis by denying the significance of food for direct consumption. Obviously income lost through food withheld from the market would have to be costed on a case-specific basis. The economic attractiveness of urban cultivation in Khayelitsha then depends to a large extent on its income earning potential relative to alternative incomes in the area.

The average monthly household income in Khayelitsha in 2001 was about R500 per month (an approximation calculated from Census 2001 income and population statistics), equivalent to R576,5 in 2003 prices, using the CPI Core Metro. This approximate value may therefore serve as a guideline for required gardening income. It is important to note, however, that in contrast to wages or pensions, income from agriculture is not equally distributed across the year. Also cultivation usually requires that the gardener invest money ‘upfront’, while profits only materialise later. Moreover, the income from gardening carries a greater risk than wage or pension income because of the potential of crop loss from pests, bad weather and theft (of fencing, tools, crops), and so on. Gardeners may therefore require a higher return from cultivation than from wage labour in order to offset risk, before they consider the income opportunities as equivalent.

A community garden's performance depends to a large extent on the good management of operations by a stable group. If the garden can retain its members, it can benefit from their skills improving over time, the accumulation of support and business networks, and experience in group and leadership dynamics. Thus, a sine qua non for viability is to guarantee members of the gardening group a sufficient income, as too low an income from gardening means that members who can find paid jobs will leave, potentially damaging every member's productivity. This view is supported by Matschke (2002: 10). An average monthly income of R500 may be the lowest acceptable income limit, and the circumstances under which urban agriculture is practised should offer gardeners the potential to significantly improve this.

From the analysis in the previous section, four possible scenarios and profit levels can be established, as shown in Table 14:

•	R0/m2/month if paying full water prices at current operational (especially irrigation) levels;
•	approximately R1/m2/month with free water (current case); or
•	approximately R1/m2/month with full water payment combined with higher irrigation efficiency; or
•	approximately R2–R2,50/m2/month in Fleming's best case (and about R4/m2/month in Fermont et al.'s adjusted best case scenario).

Although each garden will experience different circumstances and unique problems, it is argued here that these scenarios can be generalised across community gardens in the Khayelitsha area. The problems at Scaga that have led to its profits being far below expectations, such as inefficient planting and resource use and group conflict, are likely to also affect other community gardening projects in a similar environment at some stage of their existence.

Table 14: Profit scenarios and land requirements

Scenario	R/m^2/month	m^2 required to equal R500
Current operation pay all water – worst case	0	No income possible
Current operation or pay and reduce water	1	500
Fleming best case	2–2.50	200–250
Fermont et al. adjusted best case	4	125
Source: Author's calculations.
Note: All numbers are in rands.

Meadows compares a range of community gardens on the Cape Flats. She notes problems such as inefficient planting practices and irregular attendance in the Masizakhe Gardening Group, New Crossroads; lack of motivation and excessive water use in the Acacia Community Development Project, Parkwood; group conflict in the Sinethemba Garden and Catering Project, Langa; limited participation at the Manyano Support Group, Khayelitsha; and group conflict and potentially excessive water use in the Masibambane Neighbourhood Gardening Group, Philippi (Meadows 2000). Another parallel with Scaga established by Meadows (2000) was that at the time of her study most if not all of the community gardens in the Cape Flats did not have to pay for their water.

2.5.2 Required size of gardens

At R2 profit per m² per month, a gardener would need to cultivate 250 m² of land to achieve the average income in Khayelitsha of approximately R500. Table 14 shows how much land would need to be cultivated in the other scenarios in order to yield that average income. At current output levels of the Scaga garden a gardener thus needs 500 m² of land to earn an income of approximately R500 per month. As mentioned above, similar scenarios may apply to other community gardens in the area. It could be argued, therefore, that in order to maximise the likelihood of a minimum income of R500 a month in the presence of obstacles to cultivation, a gardener needs a plot of 500 m². Moreover, allocating a plot size on the basis of a low profitability gives the gardener the potential to improve her or his income by improving productivity.

Gardeners in a community garden can benefit from economies of scale. They can, for example, share tools, a water pump or water and electricity connection, facilities and networks among themselves. It thus appears sensible to combine individual farmers' plots, creating larger gardens. According to Eberhard (1989: 5), the cultivation of a food garden of 30 m² requires about 30 minutes of work a day. For 500 m² this implies approximately eight hours of work a day – the equivalent of a full-time job. An area of 500 m² would therefore represent an upper limit of what one person is capable of cultivating. This assumes that the gardener has only very basic tools at her or his disposal. An improvement in equipment could increase the amount of land one person is able to tend but would, of course, increase capital costs.

It is important to note that gardeners are not necessarily interested in income generation but may instead aim for subsistence gardening or indulge in gardening as a social activity. Two of eight Cape Flats gardening projects studied by Meadows (2000) gave social reasons, such as community development, as their main motivation for gardening. In such a situation, plot sizes could be smaller and more flexible depending on the local circumstances. A further potentially negative factor of 500 m² plot sizes is that many of the gardeners are women, who are also often the head of their household, with the myriad other responsibilities this implies (Fermont et al., 1998; Meadows, 2000). These women may not be able to devote a whole day's work to the garden unless it yields a substantial income.

It is noteworthy, however, that even in the absence of significant financial returns many garden groups are acutely aware of the limiting factor of land size. The Masibambane Neighbourhood Gardening Group in Philippi operates on about 1 500 m². This group said it would like to reduce its membership from 12 to eight members, all women, owing to the limited amount of available land, thus allocating about 190 m² per person (Meadows, 2000: 76). Members of the Quaker Peace Centre in Nyanga receive plots of about 18 m², which is considered insufficient (Meadows, 2000: 66). One member observed: ‘All the people are greedy for this land, but there is no land’ (Meadows, 2000: 70). The Manyano Support Group in Khayelitsha repeatedly expanded its activities onto new lands. The group's total plot size was 2 000 m² in 2000. With ten active members, each gardener (all women) cultivated on average 200 m² (Meadows, 2000: 71, 72). The Phatisanani Women's Gardening Group in Philippi was allocated 1 500 m² of land and had 12 members in 2000 (Meadows, 2000: 79), which equates to 125 m² per gardener.

In summary, it appears that a minimum plot size of 500 m² is required for gardeners who want to derive a sustainable survival income from urban agriculture. The benefits from economies of scale suggest that farmers could benefit from the establishment of larger, aggregated community gardens. Combining between five and ten gardeners, each with a plot of about 500 m², would therefore require an area of between 2 500 and 5 000 m², or a quarter to half a hectare of land. In a high-density township this requires a considerable area of unfettered land, which is invariably difficult to consolidate and release. Gardening groups pursuing more social benefits and not willing or able to devote as much time to their garden would require less land. Existing groups on the Cape Flats appear to allocate between 18 m² (Quaker Peace Centre) and 310 m² (Scaga) per person.

2.5.3 Community gardens competing for land

As mentioned earlier, there are a number of competing demands on land in Khayelitsha, leading to possible tension and conflict. These include housing, recreation and public open space, social consumption goods (e.g. schools, clinics, community halls), commercial and industrial activity, wetlands and stormwater management. Different values, not all of which are monetary, inhere in each of these land uses. For instance, equity issues in terms of access to basic living conditions would play an important role. If the land is used for income-generating urban agriculture, then the profitability of the land should be an important factor in deciding on the most appropriate use. Urban cultivators could potentially avoid land that is suitable for housing and locate on grounds belonging to schools, churches and other such institutions. However, this would incur a range of opportunity costs and a full cost–benefit analysis would be necessary to incorporate both monetary and non-monetary costs and benefits. These include the inherent value of open space as a social good and structuring device, something that provides aesthetic stimuli, formal and informal recreation, and the evocation of a ‘sense of place’ and community identity. It is beyond the scope of this paper to analyse in depth the economic viability of market gardening against such competing land uses.

3 Conclusions

The experience in Khayelitsha reveals that the potential of urban cultivation in low-income townships in South Africa in terms of poverty alleviation and empowerment is still uncertain, although evidence suggests it may have significant benefits both in economic and social terms if practised under the right circumstances.

The difficult socio-economic conditions of poverty and unemployment in Khayelitsha and the generally well-developed social support facilities and available open space appear to make urban agriculture conditionally viable as a source of livelihood, particularly given a minimal scale of operation. The harsh environmental conditions seem to be the major constraint. A potential land shortage, because Khayelitsha lies adjacent to ecologically sensitive nature areas, is a further problem. Land use tensions occur mainly between urban agriculture and the development of housing and other potentially incompatible uses such as ecological corridors, public open spaces and stormwater management areas. The land shortage may possibly be alleviated by using school and church property, mixed land use in parts of the ecological corridors or the stormwater management system, and also through different forms of housing design, such as cluster developments, terrace housing and apartments. These possibilities require further investigation, however, in view of the contradictory conclusions of studies in Khayelitsha as to the potential of urban agriculture

Economic analysis of the Scaga garden indicated that the activity was initially loss-making, mainly because of the low initial soil fertility and the correspondingly low yields. However, even after the improvement of the soil fertility and six years of existence of the garden, production and earnings at Scaga fell far below expectations. There are a number of explanations for this. Gardening practices were inefficient, as massive overwatering of 360–670 per cent leached essential nutrients from the ground. (The effect of overwatering was restricted to a decline in yields, as the water was free and did not affect cost.) Prolonged group conflicts seem to have hindered planting activities. The instability of the group appears to be exacerbated by the low income derived from urban agriculture, as members who find paid jobs leave the garden, and this further destabilises the operation The resulting earnings of R1,2 per m² per month fell about 70 per cent short of expectations.

The findings that gardening is initially loss-making and that profitability only rises slowly thereafter is likely to apply to other community vegetable gardens in Khayelitsha, as the problems experienced are inherent in the area and the type of gardening. The difficulties are thus likely to be replicated in similarly structured community gardens in a similar social environment.

Generally the analysis in this paper has shown that the most important issues that need to be addressed in order to improve gardening outcomes are:

•	Establishment of stable gardening groups who can benefit from improving skills and tried-and-tested group structures.
•	Efficient gardening procedures, including irrigation, fertiliser application and pest control and prevention.
•	Payment for resources, especially water, to induce efficient management and to make people aware of the costs.
•	Own production of compost and seeds.

Owing to the currently poor economic performance of gardening, the potential of urban cultivation to overcome household poverty in Khayelitsha is limited at best. No evidence was found of a person in Khayelitsha deriving a full income from urban agriculture that was sufficient for subsistence. However, in the face of widespread unemployment and deprivation, agriculture could certainly contribute in other ways to expanding household real income and to reducing vulnerability.

There was some evidence of social benefits materialising through community gardens. These include communication and motivation within the community, pride, skills training and networking outside of the community (Meadows, 2000: 42; Fleming, 2003: 5). It is noted frequently, however, that community interest in urban agriculture is not universal (Fermont et al., 1998: 27; Meadows, 2000: 43, 49). Moreover, virtually all the Cape Flats community gardens assessed in Meadows (2000) are dependent on NGOs for support and the free water from the City of Cape Town. Empowerment through urban cultivation is still a distant goal. On the other hand, the analysis also indicates that, with better resource and human management, urban agriculture could yield a substantial profit. This study established that an individual needs a minimum of 500 m² of land for cultivation to derive an income of between R500 and R2 000 per month, depending on productivity and assuming the present Scaga productivity as minimum level.

The question of whether urban cultivation should be subsidised on account of its social benefits is therefore twofold. First, the social benefits deriving from urban agriculture versus other development projects need to be assessed. Secondly, the number of people who are to be reached through urban agriculture determines the average plot size that can be allocated per person. If a person requires a minimum of 500 m² to make a living from gardening, the total number of potential gardeners will be limited by the total amount of defined open space available and competing demands for that land. Against this, were urban gardening to play a more modest role in a complex vulnerability strategy the land needs could be reduced considerably.

Any urban agriculture development policy should preserve the equity of benefits from public open space usage by permitting individuals interested in commercial urban farming to do so on larger plots that are rented. Besides such farmers, smaller-scale community gardens still using tracts of public land could cater for a larger number of people who are not interested in high-intensity urban farming but rather desire to produce a modicum of food and to enjoy the social benefits. These smaller operations could be supported by public means in terms of both finance and structured training if they are considered suitable vehicles for empowerment. It is emphasised that both types of urban agriculture contrast with individual, private, often impermanent backyard endeavours. The fact that a significant number of community gardens have been established on the Cape Flats indicates that some of the residents do consider them worthwhile.

A number of other issues require further study. These include the social and development benefits that can be derived from urban agriculture compared with other initiatives, priorities among rival demands on land in Khayelitsha, how benefits from urban agriculture compare with benefits from other land uses such as housing in the aggregate, the impact on profitability of the use of grey water, special water tariffs or higher value crops, the amount of land suitable for cultivation in Khayelitsha, the ecological impact of urban agriculture on ecological corridors in Khayelitsha and how to allocate land and rent for land equitably.