Homestead greening is widespread amongst the urbanpoor in three medium-sized South African towns

Humphrey Kaoma & Charlie M. Shackleton

# Springer Science+Business Media New York 2014

Abstract Trees in urban areas are important sources of ecosystem services and benefits. Inmost towns the bulk of urban biodiversity, and trees specifically, are found in homesteadgardens. But there is only limited understanding of the tree holdings in such gardens, and howthey vary within and between towns, especially for developing countries where rapid urban-isation and high poverty influence the use of and reliance on land and local resources. Wereport on the nature of tree holdings in private gardens of poorer suburbs in three medium-sized towns along a gradient of decreasing mean annual rainfall in northern South Africa. Atotal of 3 217 trees were enumerated across 450 randomly selected homesteads. Most (90 %)households had at least one tree on their homestead, with an average of 7.7±6.1 trees. Mosthouseholders had planted the trees themselves. The density of trees declined along themoisture gradient. Within towns, tree density was positively related to garden size, which inturn was related to relative affluence and age of the suburb. Newer and poorer suburbs had thefewest trees per household. Sixty-two tree species were recorded, which were dominated byalien species, especially fruit trees. There was no relationship between the moisture gradientand tree species richness per household, but within towns there was a difference betweensuburbs, being lowest in the newest suburbs. Numbers of trees and species per household waspositively related to age of the household head.

Keywords Alien . Density . Gardens . Richness . Trees . Towns

Introduction

Global urbanisation increased rapidly during the 20th century resulting in the majority of theworld’s population now living in urban areas (Grimm et al. 2008; Singh et al. 2010). Whilst inmost developed countries more than 80 % of the population is urban, developing countries stilllag far behind. Nonetheless, they are experiencing unprecedented shifts in population distri-bution in favour of urban areas, with millions of people moving to urban areas every year(Montgomery 2008).

Urban EcosystDOI 10.1007/s11252-014-0362-3

H. Kaoma : C. M. Shackleton (*)Department of Environmental Science, Rhodes University, Grahamstown 6140, South Africae-mail: c.shackleton@ru.ac.za

The increase in urban populations is a consequence of in-migration from rural areas andinternal population growth (Montgomery 2008). Migrants are attracted to the economicopportunities and employment security offered in towns and cities as well as the higher levelsof and proximity to services such as education, sanitation, health care and basic infrastructure.Yet, the rapid pace of urbanisation in developing countries, South Africa included, poseschallenges as urban planning and service delivery often struggle to keep up with the magnitudeof the influx (Mosha 2012). Thus, new migrants do not always find better living conditions,but may end up in slums and informal shantytowns with limited services and opportunities(Hunter and Posel 2012). Under such circumstances, new migrants may make use of naturalresources from the surrounding commons or remnant lands within urban spaces to providebasic needs such as firewood, timber for shelter, medicinal plants, fruits and browse forlivestock (Davenport et al. 2011; Kaoma and Shackleton 2014). This, coupled with landtransformation to infrastructure and housing developments which reduces resource abundance,can put heavy pressure on local biological resources, especially trees (e.g. Nkambwe andSekhwela 2006), with some seeing this as the urban deforestation frontier (Ahrends et al. 2010;DeFries and Pandey 2010). Institutional interventions to address declining tree resources inand around urban areas in developing countries include establishment of woodlots, introduc-tion of biomass stoves, development of markets in tree products, land zonation, punitiveregulations and the like (Bansal et al. 2013; Long and Nair 1999).

Trees in and around urban areas supply a multitude of tangible and intangible benefits, andhence urban forestry and greening is a growing institutional strategy in promoting urbansustainability (Cilliers et al. 2004). Tangible goods include consumptive products such asfruits, fuelwood, medicinal bark and gums, and wood for building and fencing; whilstintangible goods include non-consumptive benefits such as habitat provision, recreationalspace, shade, soil protection, carbon sequestration, and many more (Adekunle and Agbaje2012; Fuwape and Onyekwelu 2011; Kuchelmeister 2001). Components of the urban forestinclude trees on private property, in public green spaces and parks, street trees and trees on theedges of towns (Bentsen et al. 2010; Fuwape and Onyekwelu 2011; Horst 2006). Yet mostresearch and interventions in this regard have been orientated towards public spaces. Treeplanting is implemented by municipal authorities along sidewalks, in parks and aroundgovernment buildings and bulk infrastructure. However, recently it has been revealed that ata city or town scale the bulk of urban biodiversity is found on private space, mainly gardens inresidential areas and, to a lesser extent, gardens around office and corporate buildings. Forexample, Iverson & Cook (2000) found in metropolitan Chicago (USA) that approximately71 % of green areas were located on private land. The corresponding figure for León inNicuragua was 86 % (Gonzáles-Garcia and Gómez Sal 2008). These private areas are thussignificant contributors to urban biodiversity and non-consumptive ecological functionssustained by trees (Bigirimana et al. 2012; Smith et al. 2006). They are also potentiallyimportant in providing consumptive products (Alvey 2006), especially in poorer towns orsuburbs (Kaoma and Shackleton 2014), but this has been little studied, particularly indeveloping countries (Shackleton 2012). In a recent review Shackleton (2012) revealed thatonly 12 % of articles on urban forestry published in the peer-reviewed literature since 2001came from developing countries (excluding China).

As with most cities and towns of the world, South African towns are spatially differentiated.Historically this was on racial grounds, and only more recently along socio-economic lines.During the former apartheid period, black South Africans were restricted to living in poorlyserviced and high density areas called townships, whilst their white counterparts lived in moreleafy and well serviced suburbs (Wilkinson 1998). From the early 1990s, the racially enforcedseparation was repealed and the ANC government initiated a large-scale low-cost housing

Urban Ecosyst

programme as part of the national Reconstruction and Development Programme. These housesand suburbs are now called RDP houses, and are allocated to indigent people who haveregistered with their municipality (Wilkinson 1998). However, with the increasing migrationrates, housing backlogs have accumulated, resulting in many new and poorly skilled migrantsliving in informal settlements with houses made from cheap or scavenged materials and withlimited or no access to electricity, water or sanitation services (Hunter and Posel 2012; StatsSA 2001). Thus, the poorer areas of most South African towns can be differentiated into atleast three zones, which may influence the abundance of and practices relating to trees andbiodiversity (Shackleton et al. 2014); (i) townships which are generally several decades old,(ii) government developed RDP areas which are typically less than 15 years and are reservedfor the indigent, and (iii) informal settlements which are of varying age and are occupied bypoor new migrant families waiting for RDP housing.

In summary, current understanding of the benefits of trees in urban livelihoods is foundedon work focussed on public green spaces of towns and cities rather than homestead areas, andthat the majority of scientific understanding emanates from developed countries. The role andimportance of private green spaces is poorly understood generally, and is practically absentfrom the limited literature from developing countries. Yet, the urbanisation pressures andpoverty profiles in such areas potentially have significant implications for environmental andlivelihood quality. Within the context of the above, this study sought to characterise treeholdings in private spaces of poorer suburbs in small to medium-sized towns in a developingcountry, i.e. South Africa. Specifically we sought to assess the nature and extent of trees inprivate residential gardens across suburbs of differing wealth and age. We hypothesised thattree species and densities would be lowest in the most transitory or poorest suburbs, especiallythe informal settlements.

Study areas

This study was conducted in three medium-sized South African towns in the savanna biome ofthe Limpopo and North West provinces (Fig. 1). The three towns span a precipitation gradientof relatively high to low rainfall. Tzaneen receives approximately 850–900 mm p.a., BelaBela, 650 mm p.a. and Zeerust 550 mm p.a. (Mucina and Rutherford 2006). Each has apopulation of approximately 25,000–35,000, and each also serves as a commercial centre forlarge rural populations relatively close by. Socio-demographic statistics therefore refer to thewhole municipality, rather than the towns specifically.

Tzaneen

Tzaneen (23° 50′ S; 30° 10′ E) is located 800 m above sea-level in the north-east of LimpopoProvince (Fig. 1). The study was conducted in Nkowankowa, RDP and Lusaka (informal)residential areas. The geology consists of the potassium poor gneisses of the Goudplaatsgneissand the Archaean grantite dyke, there are also some shales and quartzite rocks in the region(Mucina and Rutherford 2006). The dominant soil types are Mispah and Glenrosa, which areshallow to deep sandy and gravelly and well drained (Mucina and Rutherford 2006).Climatically, the region enjoys a sub-tropical climate, with hot wet summers (Oct – May) andmild dry winters. The mean monthly maximum temperature is 36.4 °C experienced in Januaryand a minimum of 3.9 °C in June (Mucina and Rutherford 2006). Frost is rare. The vegetation isbroad-leaved savanna of the Tzaneen Sour Bushveld (Mucina and Rutherford, 2006).

Agriculture accounts for 40 % of the employment followed by manufacturing (38 %)(Greater Tzaneen Local Municipality 2011). Poverty levels are high; about 29 % of households

Urban Ecosyst

report no cash income sources and the income levels for most households (70 %) are less thanapprox. US$1,200 per annum (Stats 2008). The employment status is characterised by 20 %unemployed, 28 % employed and 49 % not economically active (Stats 2008). Less than 5 % ofhouseholds live in informal housing. Many houses in the RDP suburb are not connected to thetown’s electricity supply. Literacy levels are low, with 37 % of the people having little or non-schooling (Stats 2008).

Bela Bela

Bela Bela (24° 54′ S 28° 20′ E) is located at approximately 1 000 m above sea level in thesouth of the Limpopo Province (Fig. 1). The topography is relatively flat on the west andsouth, however, it is mountainous on the eastern and northern parts. According to Mucina andRutherford (2006) the geology is of the Karoo super group volcano-sediments. The mostcommon ones are the mafic volcanic rocks, mudstones and shale. The characteristics of thesoils in this region are “red-yellow apedal, freely drained with high base status and self-mulching, black, vertic clays” (Mucina and Rutherford 2006). The principal characteristic ofvertic clay soils is that they expand during summer (November to March) and contract duringthe dry season resulting in soil cracks (Mucina and Rutherford 2006). Mean annual rainfall istypically 550 to 650 mm, received as convectional thunderstorms during the summer. Theaverage monthly maximum temperature is 35.2 o C, and the minimum temperature −2 °C inJuly (Mucina and Rutherford 2006). Winter frost is common. The vegetation of this region isthe Springbokvlakte Thornveld (Mucina and Rutherford 2006).

Fig. 1 The three study towns (Tzaneen, Bela Bela and Zeerust) in South Africa

Urban Ecosyst

In Bela Bela Local Municipality 31 % of the labour force is unemployed and 11 % ofhouseholds receive less than US$180 per month (Bela Bela Local Municipality 2010). Theeconomy is based mainly on the tourism sector (34 %), construction (21 %) and agriculture(15 %) (Bela Bela Local Municipality 2010). Sixteen percent of people in the local munici-pality have never been to school. Between one-quarter and one-fifth of households live ininformal structures (Stats 2008). As yet there is no electricity supply in the informal settlement.

Zeerust

Zeerust (25° 32′ S 26° 6′ E) is found in North West province at an altitude varying between 1000 and 1 200 m above sea level (Fig. 1). The local geology consists of the Pretoria Groupsediments, with Silverton and Rayton rocks composed of mainly shale and small amounts ofquartzite and conglomerate (Mucina and Rutherford 2006). Soil characteristics include deep,red-yellow, apedal, well drained and base rich. Other soil types are vertic or melanic clays,which are characterised by expansion and contraction during the wet and dry seasons,respectively (Mucina & Rutherford, 2006). Zeerust receives summer rainfall ranging between450 and 600 mm per annum, with frequent droughts (Mucina and Rutherford 2006).According to Mucina and Rutherford (2006), Zeerust experiences a mean monthly maximumtemperature of 36.7 °C in January and the coldest month is June with a minimum of just belowthe freezing point (−0.4 °C). Frost is common. The vegetation is Zeerust Thornveld (Mucinaand Rutherford 2006).

Unemployment is high at 54 %, approximately 60–80 % of the population lives in povertyand are sustained mostly by government welfare grants (Ngaka Modiri Molema DistrictMunicipality 2007; Stats 2008). The major employer in Zeerust is government services whichaccounts for 37 %, followed by the wholesale and retail industry (13 %) and agriculture,forestry and fisheries account for 6.4 %. Approximately 10 % of the population live ininformal housing, and there is no electricity in the informal settlement. Just over one-thirdof the population has no formal education (NNgaka Modiri Molema District Municipality2007).

Methods

Sampling

Using aerial photographs (scale 1: 5 000) the RDP, township and informal settlementareas were identified in each town. Within these three types of residential areas ineach town, 50 households were randomly selected from the aerial photographs forsampling. At each randomly selected house, we undertook an inventory of treespresent and conducted an interview to capture household information relevant to thisstudy.

Trees in homesteads

At each selected household, the area of the homestead plot was measured and all trees on theplot were recorded with respect to species and basal circumference (at 35 cm above groundlevel). Specimens were collected, labelled and pressed for identification at the SelmarSchonland herbarium in Grahamstown. Young trees below 40 cm tall were only countedand the species noted.

Urban Ecosyst

Interviews

The household head was interviewed if present, but where absent, any adult member of thefamily was interviewed. The interviewees were encouraged to ask other family members asthey wished. Data collection included weekends and public holidays to accommodate peoplewho work. If nobody was present at a selected household, or if the household head refused tobe interviewed (8 households), the next randomly selected household on the sample frame wasinterviewed. Interviews were conducted in English or translated into a preferred local lan-guage. Each interview lasted approximately 1 hour. Core topics related to how the trees wereestablished on their plot and perceived trends in trees abundance over the recent past (5 years)

Data analysis

Prior to data analysis the answers to the structured interview schedule were coded to facilitatenumeric analysis (Neuman, 2003). Coded qualitative and quantitative data were entered intoMicrosoft Excel and were analysed using the Statistica 10. Initial data exploration via principlecomponents analysis (PCA) was used to portray the relationships between household data andtree data. For normally distributed data, a two-way ANOVA was used to analyse continuousdata and compare town and suburbs simultaneously, and a non-parametric test (Kruskal-Wallistest) was used for data which were not normally distributed. A Chi square (χ2) in the form of 2× 2 contingency tables were used to compare the percentages of indigenous and alien treespecies and linear regression was used to test for associations between plot area and number oftree species per household and tree density. In all analyses a significance level of 5 % wasadopted.

Results

Homestead plot size

The size of homestead plots varied between towns and one residential area to another(Table 1). Tzaneen and Bela Bela had significantly larger (H=18.0; p<0.0001) homesteadplots (1 876±1 153 and 1 762±829 m2) than Zeerust (1 508±1 511 m2). Across residentialareas, the township and informal residential areas had significantly larger (H=20.6; p<0.0001)homestead plots than the RDP residential areas (Table 1).

Distribution and abundance of trees on homesteads

Most households (90 %) had at least one tree on their homesteads. On average, Zeerustrecorded the highest percentage of households with trees, followed by Tzaneen, whilst BelaBela had the least (Table 1). Comparing residential areas, the townships recorded the highestpercentage (93 %) of households with trees and the RDP areas had the least (86 %). A total of3 217 trees were counted across the 450 households. Most trees were encountered in Tzaneenwhich had 1 494 (46.4 %) trees, followed by Bela Bela 884 (27.5 %) and Zeerust 839(26.1 %). The informal residential areas recorded more (42.8 %) trees than the townshipwhich was also greater (32.9 %) than the RDP residential areas.

The mean number of trees per household in Tzaneen was significantly greater (H=19.0;p<0.0005) (10.7±9.1) than in Bela Bela and Zeerust (Table 1). Considering residential areas,the RDP residential areas had significantly fewer (H=11.8; p<0.005) trees per household

Urban Ecosyst

Tab

le1

Hom

estead

plot

andhomestead

tree

attributes.(Com

paring

towns,unlik

esuperscriptsrepresentsignificantdifferenceswithin

columns.Note:

The

numberof

speciesand

percentage

ofindigenous

treesdo

notincludetheunknow

nspecies)

Town

Resident-ialarea

Attributes

Plot

size

(ha)

%hh

with

trees

Density

oftrees

Meancircum

.(cm

)Carbonstock

Speciesrichness

%indigenous

Perhh

Perha

kg/hh

kg/ha

Persuburb

Perhh

Species

Trees

Tzaneen

Inform

al0.3±0.1a

9817.9±14.7a

71.7±72.8a

46.5±34.2a

43.5±54.2

170.3±243.9

385.6±2.9a

50.0

8.6

RDP

0.1±0.1a

785.8±7.0

a40.9±56.2

a10.6±12.3a

1.3±1.3

13.1±13.5

192.6±1.5

a36.8

5.3

Township

0.2±0.1a

948.3±5.7a

65.2±16.1a

70.7±47.6a

46.2±29.5

314.8±413.7

193.7±1.9

a20.0

6.9

All

0.2±0.1

a90

10.7±9.1a

59.3±16.2

a42.6±31.4a

30.3±28.3

166.1±223.7

424.0±2.1a

47.6

7.7

BelaBela

Inform

al0.1±0.03

b82

4.9±4.4b

41.0±41.3

b22.4±18.7b

2.6±4.5

24.0±39.2

292.9±2.2b

48.3

30.7

RDP

0.2±0.1b

805.5±4.7

a27.9±30.8

a14.1±14.9a

1.9±3.1

12.1±19.9

323.2±2.0

a46.7

27.7

Township

0.3±0.1b

100

9.2±8.0a

37.0±33.6

a53.7±51.9b

43.6±41.5

175.0±182.2

505.8±3.8b

43.5

16.5

All

0.2±0.1

a87

6.5±5.7b

35.3±7

b30.1±28.5b

16.0±16.4

70.3±80.4

664.0±2.7a

54.5

22.6

Zeerust

Inform

al0.2±0.04

b96

6.2±3.7b

37.3±26.4

b22.7±29.6b

7.7±10.5

44.7±61.8

353.6±1.8a

b43.8

54.1

RDP

0.1±0.04

b100

6.7±4.4

a57.4±58.9ab

29.5±29.3b

7.7±8.3

74.4±126.6

263.5±2.0

a40.0

34.4

Township

0.1±0.3c

864.9±2.8b

47.1±67.9

a33.2±32.9c

8.2±9.3

132.6±397.2

222.8±1.4

a23.8

8.6

All

0.2±0.1b

945.9±3.6b

47.3±10.1

a28.5±30.6c

7.9±9.3

83.9±195.2

423.3±1.7a

42.9

34.9

Com

bined

Inform

al0.2±0.1a

929.7±7.6a

53.0±17.4a

30.5±27.5a

53.7±23.1

79.7±115.0

734.0±2.3b

58.9

30.7

RDP

0.1±0.1b

866.0±5.4b

42.1±14.8b

18.1±18.8b

11.0±4.2

33.2±53.3

523.1±1.8a

50.0

24.0

Township

0.2±0.2a

937.5±5.5a

49.8±14.3b

52.5±44.1c

98.1±26.8

207.5±331.0

604.1±2.4b

45.0

10.8

All

0.2±0.1

907.7±6.1

48.3±5.6

33.7±30.2

54.3±18.0

106.8±166.4

623.8±2.2

55.4

18.7

Urban Ecosyst

(6.0±5.4) than the township and informal residential areas. There was a positive correlationbetween homestead plot size and the number of trees per homestead in the informal residentialareas (r=0.5; p<0.05) and township areas (r=0.2; p<0.05). Tzaneen and Zeerust had signif-icantly higher (H=14.0; p<0.001) tree densities (59.3±16.2 and 47.3±10.1 trees per ha) thanBela Bela. Across residential areas, the informal residential areas had significantly (H=6.3;p<0.05) higher tree density (53.0±17.4 trees per ha) than the township and RDP residentialareas.

Tree size and carbon content

Trees on Tzaneen homesteads were significantly (H=170.8; p<0.0005) larger (42.6±31.4 cm)than trees in Bela Bela, which in turn, were larger than trees in Zeerust. The older theresidential area, the larger the tree circumference, therefore, trees in townships had signifi-cantly (H=357.3; p<0.0001) larger circumferences than trees in the informal residential areas,which also had larger mean tree circumference than the RDP residential areas (Table 1).

A large proportion of trees were relatively small (Fig. 2), signifying that most of the treeswere recently established on homesteads. The size class profiles portrayed a classical reverse J-shape, indicative of expanding populations. Trees in the RDP area, the youngest suburbs, weresignificantly smaller (H=49.2; p<0.0001) than those in the informal areas and township.

Informal RDP Township

Tzaneen

Bela Bela

Zeerust

0

50

100

150

200

250

300

No.

of

tree

s

020406080

100120140160180

No.

of

tree

s

0

10

20

30

40

50

60

70

80

90

No.

of

tree

s

020406080

100120140

No.

of

tree

s

020406080

100120140

No.

of

tree

s

020406080

100120140160180

No.

of

tree

s

0

50

100

150

200

250

1 2 3 4 5 6 7 8 9 10111213

No.

of

tree

s

Circumference (cm)

020406080

100120140160180200

1 2 3 4 5 6 7 8 9 10 11 12 13

No.

of

tree

s

Circumference (cm)

0

20

40

60

80

100

120

140

1 2 3 4 5 6 7 8 9 10111213

No.

of

tree

s

Circumference (cm)

Fig. 2 Size class profile of trees (Circumference size classes in 25 cm interval)

Urban Ecosyst

Tree species richness

The mean number of tree species per household ranged between 2.6±1.5 and 5.8±3.8(Table 1). Considering towns, each averaged three to four species per household and therewere no significant differences among them. Across the residential areas, the township andinformal residential areas had significantly (H=9.6; p<0.01) more tree species per household(4.1±2.4 species) than the RDP residential areas (Table 1). Different tree species dominated ateach town. The notable ones were M. indica in Tzaneen, Tecoma stans and Prunus persica inBela Bela and in Zeerust P. persica was widespread (Table 2). Considering the top seven treespecies, there were more fruit tree species (61 %) than non-fruit trees.

Overall, there were more alien than indigenous tree species. However, the informalresidential areas had more indigenous trees species than the RDP residential areas whichhad also more than the townships (Table 1). This suggests that the residents in the informalresidential areas retained most of the tree species during the establishment of their home. Thetownship in Tzaneen had significantly (χ2=13.2; df=1; p<0.0005) more alien trees (80 %)than the one in Bela Bela, similarly the one in Zeerust had significantly (χ2=8.9; df=1;p<0.005) more alien trees than Bela Bela (Table 1).

Alien tree species dominated the top seven tree species in each residential area and theywere mainly fruit tree species (Table 2). Across all towns, the informal residential areas had thelargest percentages of indigenous trees whilst the township recorded the lowest proportions.

Tree establishment

Most of the trees in all towns were planted by the current occupants (71.1 %), while some werealready on the homesteads when the respondents moved onto the plots, and others were self-seeded (Table 3). More respondents collected propagation material from their neighbours,friends, relatives and urban green patches in their neighbourhoods than bought from retailoutlets. Others retained trees on their homesteads when they constructed their houses, espe-cially over half of households in the informal residential area and RDP in Zeerust retainedtrees.

Perceived trends in tree abundance over the past 5 years

More respondents in all towns and suburbs felt that the numbers of trees on homesteads wereincreasing than those who thought they were decreasing (Table 4). The respondents attributedthis to people planting more trees complimented by self-seeding recruits. Those who perceiveda decrease in tree abundance stated it was because trees were frequently cut down to allow forexpansion of houses and of residential areas generally. Some respondents expressed unhappi-ness that sometimes they had to remove trees on their homesteads, and wished that the plotswere larger so that they could plant more trees. A few ascribed the perceived decreases inabundance to a lack of water, poor soil conditions and browsing by livestock.

Tree attributes and socio-economic characteristics

There were relatively few strong relationships between tree attributes and the profiles of therespondent households (Fig. 3) as evident from the first two axes, which together explained33 % of the variance. The primary axis draws out a temporal dimension as related to years inthe town and years in the specific house. The second axis relates to wealth attributes withwealthier households in the lower hemisphere and poorer households in the upper one

Urban Ecosyst

Tab

le2

The

sevenmostcommon

tree

speciesforthreetowns

(Indigenoustree

speciesarein

bold

type;usecodesarec,cultu

ral;d,

decorativ

e;f,fruit;h,

hedge;s,shade)

Town

Residentialarea

Inform

alMainuses

%hh

%trees

RDP

Mainuses

%hh

%trees

Township

Mainuses

%hh

%trees

Tzaneen

Mangifera

indica

f,s

9866.6

Mangifera

indica

f,s

7445.1

Mangifera

indica

f,s

9454.2

Caricapapaya

f42

7.1

Caricapapaya

f30

25.7

Caricapapaya

f40

13.6

Persea

americana

f38

3.8

Persea

americana

f18

5.8

Persea

americana

f32

4.6

Psidium

guajava

f26

2.0

Psidium

guajava

f14

4.4

Palm

d30

5.9

Citrus

limon

f22

2.3

Thujaoccidentalis

d14

4.4

Citrus

limon

f24

4.1

Sclerocaryabirrea

subsp.

caffra

f,s,c

202.3

Citrus

limon

f10

2.2

Citrus

sinensis

f24

3.1

Prunus

persica

f18

1.5

Malus

domestica

f6

2.2

Litchi

chinensis

f16

2.3

Total

85.6

89.8

87.8

BelaBela

Prunus

persica

f40

14.4

Prunus

persica

f44

24.5

Prunus

persica

f58

13.9

Tecomastans

d40

22.8

Acaciakarroo

s30

15.0

Citrus

limon

f44

8.7

Morus

nigra

f34

12.9

Mangifera

indica

f,s

269.1

Morus

nigra

f40

6.1

Acaciatortilis

s12

8.4

Persea

americana

f24

5.9

Mangifera

indica

f,s

388.0

Citrus

limon

f12

3.5

Citrus

limon

f18

5.0

Caricapapaya

f36

6.3

Persea

americana

f10

3.5

Psidium

guajava

f14

5.0

Jacarandamim

osifolia

d,s

346.1

Dodonaeaviscosa

h8

8.4

Tecomastans

d12

4.5

Persea

americana

f30

5.6

Total

73.9

69.0

54.7

Zeerust

Searsialancea

s56

18.9

Prunus

persica

f78

30.2

Prunus

persica

f54

24.9

Prunus

persica

f48

19.3

Acaciasp.(Mosu)

s36

14.7

Ligustrum

lucidum

h42

32.1

Ziziphusrivularis

s46

11.0

Ligustrum

lucidum

h32

14.1

Morus

nigra

f40

12.9

Morus

nigra

f36

8.1

Searsialancea

s32

6.0

Vitisvinifera

f,s

226.2

Acaciasp.(Mosu)

s26

8.4

Morus

nigra

f22

4.8

Ficussp.(Feiye)

s,f

164.3

Ficussp.(Feiye)

s,f

205.4

Vitisvinifera

f,s

183.0

Citrus

limon

f12

3.3

Malus

domestica

f16

5.1

Ziziphusrivularis

s16

3.0

Citrus

sinensis

f8

2.9

Total

76.2

75.8

86.6

Urban Ecosyst

Table 3 Modes of tree establishment (Some respondents gave more than one answer; therefore totals can exceed100 %)

Town Residential area Mode of tree establishment (%) Propagation material (%)

Planted Already there Self-seeded Bought Collected

Tzaneen Informal 92 4 12 46 80

RDP 76 4 2 8 66

Township 78 16 8 46 60

All 88.7 8.0 7.3 33.3 68.7

Bela Bela Informal 62 20 20 14 52

RDP 60 26 22 24 36

Township 74 30 20 18 42

All 65.3 25.3 20.7 18.7 43.3

Zeerust Informal 68 54 26 4 64

RDP 78 50 36 2 76

Township 52 30 20 6 46

All 66.0 44.7 27.3 4.0 62.0

Combined Informal 74 26 19.3 21.3 65.3

RDP 71.3 26.7 20 11.3 55.3

Township 68 25.3 16 23.3 49.3

All 71.1 26.0 18.4 18.6 56.6

Table 4 Percentage of responses on trends in tree abundance in the past 5 years (Note: Due to rounding off,some figures do not add up to 100 %)

Town Res. area Change over the past five years Type of change Causes of change

No change Change No response Increase Decrease Plant more Cut down Die

Tzaneen Informal 24 74 2 50 24 46 22 0

RDP 72 4 24 2 2 2 0 2

Township 28 66 6 24 42 24 42 0

All 41.3 48.0 10.7 25.3 22.7 24.0 21.3 0.7

Bela Bela Informal 24 56 20 40 16 42 6 8

RDP 40 38 22 26 12 28 0 12

Township 46 50 4 38 12 38 8 4

All 36.7 48.0 15.3 34.7 13.3 36.0 4.7 8.0

Zeerust Informal 80 16 4 12 4 10 4 2

RDP 72 28 0 24 4 24 4 0

Township 70 16 14 14 2 14 2 0

All 74.0 20.0 6.0 16.7 3.3 16.0 3.3 0.7

Combined Informal 42.7 48.7 8.7 34 14.7 32.7 10.7 3.3

RDP 61.3 23.3 15.3 17.3 6 18 1.3 4.7

Township 48.0 44 8.0 25.3 18.7 25.3 17.3 1.3

All 50.7 38.7 10.6 25.5 13.1 25.3 9.8 3.1

Urban Ecosyst

characterised by a high number of social grants and larger numbers of people per bedroom.The proportion of indigenous trees was negatively associated with suburb, being highest in theinformal settlement and lowest in the township. This was also weakly negatively related tohousehold income (r=−0.10; p<0.05). Tree size (circumference) was associated with suburb,being highest in the township and lowest in the RDP suburbs. The PCA suggested that genderof the household head had a weak correlation with tree species richness and density, both beinghigher in female-headed households. However, subsequent direct analysis showed that thedifferences were not significant for species richness (female=3.5±0.20; male=3.1±0.15species per household; t=1.5; p>0.05) or density (female=50.1±6.28; male=43.2±3.17; t=1.04, p>0.05). The number of species was positively associated with age of the householdhead (r=0.18, p<0.001), as was the number of trees per household (r=0.11, p<0.05).

Discussion

Tree holdings on urban homesteads

This study clearly shows that urban residents of poorer suburbs in small towns of a developingcountry actively plant and manage trees on their homesteads, even though this practice isfrequently ignored by researchers, governments and international forestry organisations(Kuchelmeister 1999; Shackleton 2012). This tempers the perspective that urban expansion

SuburbCircumf

Gender

Hh income

No. of grants

-1.0 -0.5 0.0 0.5 1.0

Factor 1 : 18.12%

-1.0

-0.5

0.0

0.5

1.0

Fac

tor

2 : 1

5.07

%

SuburbCircumf

Gender

Hh income

No. of grants

EducationMeat/mth

% indigenous

People/bedroom

Town

Yrs in house

Yrs in town

AgeNo. of spp

No. of trees

Density

Fig. 3 PCAvector plot of tree and household attributes (Axis 1=18.1 % of variance; Axis 2=15.1 % of variance)

Urban Ecosyst

by new migrants in informal settlements, drives urban forest degradation (Nkambwe andSekhwela 2006). We have shown that urban residents plant and maintain trees on theirhomesteads to benefit from the various consumptive and non-consumptive products provided,despite being limited by inadequate space, finance and in some instances, adverse conditions(Lubbe et al. 2010). However, this varies through time as some subsequently fell trees in theiryards if the size of their dwelling increases. This parallels processes in even large world cities,where trees and patches of greenery change in size and location through opposing deforesta-tion and reforestation processes (Zipperer et al. 2012).

The majority of trees on homesteads had been directly planted, with the remainder beingretained when the homestead was first built and those which had self-seeded. This showed thatmost of the residents have an interest in planting and maintaining trees on their homesteads.This is contrary to the argument of Cilliers et al. (2004) who stated that new migrants clearland for housing. Thus, a more nuanced view is necessary. It is clear that some trees may haveto be cleared to build household infrastructure, but that often large and significant trees alreadyon the land are retained provided that they are not directly where the house or otherinfrastructure is to be erected, and once the household is established, additional trees may beplanted or nurtured from wild-seeded recruits. Most urban residents were aware of thecontribution of trees by retaining some in their homesteads. Planting material was typicallyaccessed from own homesteads, friends, family and neighbours. It could also be that plantingmaterial is not readily available at commercial nurseries or it is too expensive hence residentswould rather collect than buy.

Overall, township homesteads had the largest plots, the RDP houses the smallest, and thoseof the informal residential were intermediate between these two. McConnachie and Shackleton(2010) also reported that township homestead plots were larger than the RDP residential areaplots. Whilst both are relatively poor, the low cost and mass development planning model ofRDP residential areas has dictated that housing density must be high to help reduce associatedinfrastructure and bulk supply costs per unit (Hunter and Posel 2012). This potentiallyconstrains greening opportunities, as larger plots typically have more trees (Tratalos et al.2007; Uddin and Hasan 2001; Venn and Niemela 2004). This was echoed by some respon-dents who bemoaned the limited size of homestead plots, especially that trees were cut to allowfor expansion of houses, as well as others who wished that homestead plots were larger so thatthey could plant more trees.

The lowest proportion of households with trees was recorded in the RDP areas of Tzaneenand Bela Bela. This is probably because they were only constructed within the last 5 years orso and the residents recently started living in them. In contrast, the RDP residential area inZeerust was established over 10 years ago and all homesteads had at least one tree. Therefore,it possible that the proportion of households with trees in the RDP areas of Tzaneen and BelaBela will increase with time, except where expansion pressures result in tree removal. Yetchanges through time will also be influenced by changes in household circumstances, such asincreasing wealth if currently poor households secure reasonable employment. For example,Barbosa et al. (2007) and Zhang et al. (2008) reported that wealthy households typically havemore trees, which is echoed in the greater densities found in the township areas during thisstudy. However, in a developing country context, Gangopadhyay and Balooni (2012) revealedthat as households in Kozhikode, India, grew wealthier, the cover of trees declined. Elsewherein South Africa, Lubbe et al. (2010) reported increasing plant species richness in domesticgardens in Tlkowe city with increasing affluence. However, they also noted that the moreaffluent households were situated on better soils.

Within towns, homesteads in the informal residential areas had significantly more trees (9.7±7.6 trees) that the townships and RDP areas (6.0±5.4 trees). These numbers are towards the

Urban Ecosyst

lower end of the range (4 – 38) for rural homesteads in South Africa reported by Paumgartenet al. (2005). The RDP residential areas had more trees which were less than 40 cm tall,probably a reflection of the relative newness of these suburbs. It can be expected that aftersome years these small trees will mature and these residential areas will be leafier. In all areas,the size class profile of trees was dominated by small stems, which is similar to rural areas(Paumgarten et al. 2005). The few relatively large circumference trees in informal residentialand RDP areas were not planted but were retained when the plots were established. Since thetownships were relatively older than the other two residential areas, mean tree circumferenceswere greater.

The density of trees per household declined along the climatic moisture gradient fromTzaneen through to Zeerust showing the influence of background environmental gradientseven in markedly modified human environments. It may also relate to the absence of otherunfavourable conditions at Tzaneen. For example, some respondents in Bela Bela complainedabout poor soil conditions constraining the growth of trees. The soils in Bela Bela and Zeerustcrack as they swell and shrink during the wet and dry season (Mucina and Rutherford 2006).Lack of water was the other reason listed by some respondents at Bela Bela and Zeerust. Thus,the distribution and abundance of trees on homesteads is affected by both micro- and macro-environmental conditions (Uddin and Hasan 2001).

Species richness

Unlike tree density, species richness per household did not vary significantly along the rainfallgradient, being approximately four in Tzaneen and Bela Bela and 3.3 in Zeerust. On a townbasis, approximately 60 % more species were found at Bela Bela than the other two towns.More than half of the total numbers of trees in Tzaneen were M. indica (an exotic fruit tree),while plots in Bela Bela and Zeerust had several tree species per household. Hence, highabundance of trees does not indicate high tree species richness due to preference by householdsfavouring specific tree species (Uddin and Hasan 2001). Within towns the informal andtownship residential areas had similar species richness of trees per household, which wassignificantly higher than found in the RDP areas. This may be the effect of plot size. Bernholtet al. (2009) showed that larger homestead plots had higher species richness and tree densitythan smaller ones. Similarly, rural households with larger homestead plots had higher speciesrichness (Paumgarten et al. 2005).

The most common trees were fruit trees (making up 61 % of the top seven common trees)mirroring findings by Alam and Masum (2005), Bernholt et al. (2009), Lubbe et al. (2010) andNdaeyo (2007). Chibende (2009) reported that urban residents in Zambia planted more fruittrees than non-fruit trees with a ratio of 3:1 and these were mainly exotic fruit trees. The mostcommon fruit species in Tzaneen was M. indica while in Bela Bela and Zeerust it wasP. persica, both of which are exotics. Fruits probably contribute to the nutrition and healthof the residents and can also be used to generate income (e.g. Termote et al. 2012).

Most trees in Tzaneen and Zeerust were alien species, whereas in Bela Bela most wereindigenous. The majority of residents prefer alien tree species because they favour conven-tional fruit species. Previous work has shown that both rural and urban residents favour alientree species (Bernholt et al. 2009; Ndaeyo 2007; Paumgarten et al. 2005). The greatestproportion of indigenous species was found in the informal settlements in each town, becauseof retention of existing trees when new migrants established a new homestead. The townshipareas generally had the lowest proportion of indigenous trees, reflecting importation of exoticspecies over time and perhaps due to greater relative affluence. The indigenous tree speciesthat were among the top seven included: Sclerocarya birrea subsp. caffra, Ficus sp., Acacia

Urban Ecosyst

sp., Ziziphus rivularis, Searsia lancea, Acacia karroo and Acacia tortilis. Among the alientrees recorded, T. stans, P. guajava, L. lucidum and J. mimosifolia are regarded as invasive inSouth Africa (Nel et al. 2004).

The nature of tree holdings per household is clearly influenced by a number ofecological and social factors. The mean annual rainfall influenced tree density, but notspecies richness. Local effects of soil type and water provision also influenced whatpeople planted and in what densities. Stronger influences were evident with some ofthe social variables such as plot size, which affected both density and speciesrichness. Higher relative affluence and age of the suburb also resulted in greaterdensities of trees, trees of larger size and a greater proportion of alien species.Most attributes of the household head, such as gender, education and duration ofliving in that house or town did not have any strong influence on homestead treeholdings. In contrast, in rural households Shackleton et al. (2008) found that thegender of the household head was linked to the density of trees on homesteads.However, age of the household head was positively associated with both the numberof trees and number of species per household. This may be a reflection of older headshaving more time to establish gardens and trees, as there was some associationbetween age and length of time in the specific town and particular homestead.

Conclusion

Residents in the poorer suburbs of medium-sized towns undoubtedly have an interestin planting and managing trees on their homesteads. These trees offer a variety ofbenefits to residents, such that some residents, without running water on their home-steads, walk long distances to fetch water for young trees. The township and informalresidential areas had larger plots compared to the newly established RDP residentialareas, which was a major factor affecting the number of trees and species onhomesteads. These findings confirm the appreciation of trees by poor urban residentsirrespective of the macro and local level environmental and socio-economic differ-ences, and that this can be a viable foundation for promotion of urban biodiversity.

Acknowledgments HK and the field work were funded by SANPAD (project 10/58), for which we are grateful.We are also indebted to Tony Dold at the Selmar Schonland Herbarium (Grahamstown) for his help in identifyingsome of the tree species.

References

Adekunle MF, Agbaje BM (2012) Public willingness to pay for ecosystem service functions of a peri-urbanforest near Abeokuta, Ogun State, Nigeria. J Dev Agric Econ 4:45–50

Ahrends A, Burgess ND, Milledge SAH, Bulling MT, Fisher B, Smart JCR, Clarke GP, Mhoro BE, Lewis SL(2010) Predictable waves of sequential forest degradation and biodiversity loss spreading from an Africancity. Proc Natl Acad Sci U S A 107(33):14556–14561

AlamMS, Masum KM (2005) Status of homestead biodiversity in the offshore island of Bangladesh. Res J AgricBiol Sci 1:246–253

Alvey AA (2006) Promoting and preserving biodiversity in the urban forest. Urban For Urban Green 5:195–201Bansal M, Saini RP, Khatod DK (2013) Development of cooking sector in rural areas in India: a review. Renew

Sust Energ Rev 17:44–53

Urban Ecosyst

Barbosa O, Tratalos JA, Armstrong PR, Davies RG, Fuller RA, Johnson P (2007) Who benefits from access togreen space? A case study from Sheffield. UK. Land Urban Plan 83:187–195

Bela Bela local municipality IDP (2010) Integrated development plan 2010/2011. Bela Bela local municipality.South Africa. 1–185

Bentsen P, Lindholst AC, Konijnendijk CC (2010) Reviewing eight years of urban forestry and urban greening:taking stock, looking ahead. Urb For Urb Green 9:273–380

Bernholt H, Kehlenbeck KK, Gebauer J, Buerkert A (2009) Plant species richness and diversity in urban andperi-urban gardens of Niamey, Niger. Agrofor Syst 77:159–179

Bigirimana J, Bogaert J, De Canniere C, Bigendako M-J, Parmentier I (2012) Domestic garden plant diversity inBujumbura, Burundi: role of the socio-economical status of the neighbourhood and alien species invasionrisk. Land Urban Plan 107:118–126

Chibende C (2009) To investigate the effects of selling government houses on the conservation of urban forests,Special project report. The Copperbelt University, Kitwe

Cilliers SS, Muller N, Drewes E (2004) Overview on urban nature conservation: situation in the western-grassland biome of South Africa. Urb For Urban Green 3:49–62

Davenport N, Gambiza J, Shackleton CM (2011) Use and user of municipal commonage around three smalltowns in the Eastern Cape, South Africa. J Environ Mgmt 92:1149–1460

DeFries R, Pandey D (2010) Urbanization, the energy ladder and forest transitions in India’s emerging economy.Land Use Policy 27:130–138

Fuwape JA, Onyekwelu JC (2011) Urban forest development in West Africa: benefits and challenges. J BiodiversEcol Sci 1:77–94

Gangopadhyay K, Balooni K (2012) Technological infusion and the change in private, urban green spaces. UrbanFor Urban Green 11:205–210

Gonzáles-Garcia A, Gómez Sal A (2008) Private urban greenspaces or “patios” as a key element in the urbanecology of tropical central America. Hum Ecol 36:291–300

Greater Tzaneen Local Municipality IDP (2011) Integrated development plan 2011/2012. Greater Tzaneen LocalMunicipality. South Africa. 1–366

Grimm NB, Faeth SH, Golubiewski NE, Redman CL, Wu J, Bai X, Briggs JM (2008) Global change and theecology of cities. Sci 319:756–760

Horst A (2006) Rehabilitation of urban forests in Addis Ababa. J Dry 1:108–117Hunter M, Posel D (2012) Here to work: the socioeconomic characteristics of informal dwellers in post-apartheid

South Africa. Environ Urban 24:285–304Iverson LR, Cook EA (2000) Urban forest cover of the Chicago region and its relation to household density and

income. Urban Ecosys 4:105–124Kaoma H, Shackleton CM (2014) Collection and use of urban tree products by households in poorer residential

areas of three South African towns. Urban For. Urban Green. doi:10/1016/j.ufug.2014.02.002Kuchelmeister G (1999) Urbanisation in developing countries-time for action for national forest programmes and

international development cooperation for the urban millennium. Forest policy research forum: the role ofnational forest programmes to ensure sustainable forest management. Joensuu, Finland, pp 1–17

Kuchelmeister G (2001) Urban forestry. CTA, Wageningen, pp 1–7Long AJ, Nair PKR (1999) Trees outside forests: agro, community and urban forestry. New For 17:

145–174Lubbe CS, Siebert SJ, Cilliers SS (2010) Political legacy of South Africa affects the plant diversity patterns of

urban domestic gardens along a socio-economic gradient. Sci Res Essays 5:2900–2910McConnachie MM, Shackleton CM (2010) Public green space inequality in small towns in South Africa. Habitat

Intn 34:244–248Montgomery MR (2008) The urban transformation of the developing world. Sci 319:761–764Mosha AC (2012) Low income access to urban housing and land in Botswana. Urban Forum (in press)Mucina L, Rutherford CR (eds) (2006) The vegetation of South Africa, Lesotho and Swaziland. Strelitzia 19.

South African national Biodiversity Institute, PretoriaNdaeyo NU (2007) Assessing the contributions of homestead farming to food security in a developing economy:

a case study of south-eastern Nigeria. J Agric Soc Sci 1:11–16Nel JL, Richardson DM, Rouget M, Mgidi TN, Mdzeke N, le Maitre DC, van Wilgen BW, Schonegevel L,

Henderson L, Neser S (2004) A proposed classification of invasive alien plant species in South Africa:towards prioritising species and areas for management action. S Afr J Sci 100:53–64

Neuman WL (2003) Social research methods: qualitative and quantitative approaches, 4th edn. Mass: Allyn andBacon, Boston, pp 1–600

Ngaka Modiri Molema District Municipality IDP (2007) Integrated development plan 2007–2011. Ngaka ModiriMolema District Municipality. 1–137

Urban Ecosyst

Nkambwe M, Sekhwela MB (2006) Utilization characteristics and Importance of woody biomass resources onthe rural–urban fringe in Botswana. Environ Mgmt 37:281–296

Paumgarten F, Shackleton CM, Cocks M (2005) Growing trees in home-gardens by rural households in theEastern Cape and Limpopo provinces, South Africa. Intn J Sust Dev World Ecol 12:365–381

Shackleton CM (2012) Is there no urban forestry or greening in the developing world? Sci Res Essays 7(40):3329–3335

Shackleton CM, Paumgarten F, Cocks ML (2008) Household attributes promote diversity of tree holdings inrural areas, South Africa. Agrofor Sys 72:221–230

Shackleton CM, Hebinck P, Kaoma H, Chishaleshale M, Chinyimba A, Shackleton SE, Gambiza J, Gumbo D(2014) Low-cost housing developments in South Africa miss the opportunities for household level urbangreening. Land Pol 36:500–599

Singh VS, Pandey DN, Chaudhry P (2010) Urban forests and green spaces: lessons for Jaipur. RSPCB occasionpaper 1/2010. Rajasthan, India. Rajasthan State Pollution Control Board (RSPCB), India, pp 6–17

Smith RM, Thompson K, Hodgson JG, Warren PH, Gaston KJ (2006) Urban domestic gardens (IX): compo-sition and richness of the vascular plant flora, and implications for native biodiversity. Biol Constr 129:312–322

Stats SA (2008) Community survey 2007 basic results: municipalities. Statistics South Africa, Pretoria, pp 1–28Stats SA (2001) Concepts and definitions. Report no. 03.02.26. Statistics South Africa, Pretoria, pp 1–19Termote C, Everaert G, Meyi MB, Djailo BD, Van Damme P (2012) Wild edible plant markets in Kisangani,

democratic republic of Congo. Human Ecol 40:269–285Tratalos J, Fuller RA, Warren PH, Davies RG, Gaston KJ (2007) Urban form, biodiversity potential and

ecosystem services. Land Urban Plan 83:308–317Uddin MJ, Hasan MK (2001) Economics of homestead agroforestry in Char areas at Noakhali District.

Proceedings of National Workshop on Agrofestry Research. Bangladesh. 37–43Venn SJ, Niemela JK (2004) Ecology in a multidisciplinary study of urban green space: the urge project. Boreal

Environ Res 9:479–489Wilkinson P (1998) Housing policy in South Africa. Habitat Intn 22:215–229Zhang Y, Tarrant MA, Green GT (2008) The importance of differentiating urban and rural phenomena in

examining the unequal distribution of locally desirable land. J Environ Mgmt 88:1314–1319Zipperer WC, Foresman TW, Walker SP, Daniel CT (2012) Ecological consequences of fragmentation and

deforestation in an urban landscape: a case study. Urban Ecosyst 15:533–544

Urban Ecosyst