Patterns of enclosure in the urban morphology : a comparative case study of Accra and Pretoria

3

Summary

In the urban studies literature, residential enclosure is described from the perspective of the upper class who are inhabiting gated communities, enclosed- securitized urban areas and enclaves. The increase in levels of enclosure are attributed to the power of these urban elites in the city. However, dynamics of enclosure are not limited to these traditional urban forms. This comparative case study, extends the knowledge level by researching residential enclosure beyond its traditional forms. The analysis is focused on multiple growing neighbourhoods since 2000 in Accra (Ghana) and Pretoria (South Africa), two major cities in the ‘Global South’. With the use of satellite imagery, a unique method is developed that produces maps which indicate physical walls constructed in multiple growing neighbourhoods. The characteristics and dynamics of enclosure between the two cities are compared with the use of morphological analytical tools. In both cities, enclosure seems to increase in the urban structure. The dominant role of Chiefs and families in the land-tenure system in Accra results in the development of more unplanned unequal subdivided urban blocks around pre-existing villages. In Pretoria, the local government is more involved in city development, resulting in more structured, equally subdivided blocks. In both cases this is reflected in the character of residential walling, enclosed villa-type plots are distinguished from non-enclosed small shacks. The shift in focus from enclosure in its traditional forms towards enclosure in the wider urban structure shows that urban exclusion takes place within multiple scales. With the use of residential walls, visual separation takes place on a more individual level in growing neighbourhoods.

Key words: urban enclosure, urban morphology, residential walling, Accra, Pretoria

and Global South.

5

Content

1.

Introduction

2.

Theoretical Framework

2.1. Enclosure …. page 8

2.2. Dynamics contributively to enclosure … page 9

2.2.1. Danger and. safety measures …. page 9 2.2.2. Privatization …. page 9

2.2.3. Territoriality …. page 10 2.2.4. Land Ownership …. page 10

2.3. Urban morphology …. page 11

2.3.1. Categorization of morphologies …. page 11 2.3.2. Neighbourhoods and plot patterns …. page 15

2.4. The context of Accra …. page 16

2.5. The context of Pretoria …. page 17

2.6. Expectations …. page 18

3.

Methodology

3.1. Research Aims …. page 19

3.2. Case Study Considerations … page 19

3.2.1. The multiple case design: why Accra and Pretoria… page 20 3.2.2. Units of analysis … page 20

3.3. Preparation of the analysis … page 27

3.3.1. Data … page 27

3.3.2. Enclosure analysis … page 28

3.3.3. The case studies and comparison … page 29

4. Case Study I: Accra, Ghana

4.1. Presentation of the morphology and level of enclosure per neighbourhood; Accra …. page 31

6

5. Case Study II: Pretoria, South Africa

5.1. Presentation of the morphology and level of enclosure per neighbourhood; Accra …. page 44

5.2. Patterns of enclosure in growing neighbourhoods; Accra… page 58

6. The comparison: Accra versus Pretoria

6.1. The overview …. page 60

6.2. The expectations … page 66

6.2.1. Hypothesis 1; level of enclosure …. page 66 6.2.2. Hypothesis 2; subdivision of plots …. page 68

6.2.3. Hypothesis 3 & 4; unplanned vs. planned urban blocks …. page 72 6.2.4. Hypothesis 5; a matter of time …. page 74

7. Conclusion

7.1. Results …. page 75

7.2. Limitations and strengths… page 76

7.3. Contributions and future research… page 77

8. Bibliography

7

1. Introduction

Urban residential areas are becoming increasingly enclosed. Whole neighbourhoods are physically walled or fenced off. This process is driven by the fear of who pose a threat, the securitization of private property and the desire to control and exclude the criminalised poor (Dovey, 2000; Fauveaud, 2016; Hodkinson, 2012: 500-501). The first gated neighbourhoods were developed in North America in the 19th _{century. Over the past 50 years the number of}

enclosed neighbourhoods in the world increased significantly, although the exact numbers are unknown. Since the 1990s, walled neighbourhoods have developed in cities across the globe. The characteristics of enclosed neighbourhoods differ depending on the national context and existent historical patterns of walling (Atkinson & Blandy, 2012; Le Goix, 2015). At the same time urban inequality and fragmentation are increasing in the urban landscape. Jeffrey et al (2012: 1251) argue that the neoliberal reworking of the city initiates new urban forms. Examples of these urban forms are the demolition and renovation of informal settlements, the growth of special economic zones (SEZs) or the creation of urban areas into world class cultural, knowledge, business and health centres.

The current body of literature refers to enclosure as gated communities, enclosed or securitized urban areas, enclaves and even gentrified areas. The increase of enclosure is thereby attributed to the power of new elites in the city. However, dynamics of enclosure are not limited to these kinds of urban forms. Residential enclosure is not a homogeneous process. Its form is dependent on national and historical context and it is largely different in local situations (Feaveaud, 2016; Hodkinson, 2012; Rosen & Grant, 2011). This thesis extends the knowledge level in contemporary urban literature by researching enclosure within the urban morphology beyond its traditionally described forms. In a great number of cities in the Global South, urban walling is observable throughout large parts of the urban form. On top of this, enclosed residences are not only occupied by urban elites, but also by other social classes. This thesis aims to extend the knowledge of the current field of literature with an analysis of the patterns of urban enclosure in growing neighbourhoods in cities in the Global South. The thesis focuses on growing neighbourhoods since 2000 by the means of a comparative case study focused on Accra in Ghana and Pretoria in South Africa. The following research questions guide this thesis in the aim to find out more about enclosure in both cities:

What are the main characteristics of enclosure in Accra and Pretoria in growing neighbourhoods since 2000? And how can the dynamics of enclosure in both cities be explained in their local context?

This thesis examines these questions in chapter 2 to 7. Chapter 2 first describes how to understand the concept and dynamics of enclosure. Second, it provides an understanding and categorization of the urban morphology. Additionally, it provides an overview of the context of Accra and Pretoria in which patterns of enclosure take shape. Finally, it presents the main expectations of the analysis. Chapter 3 describes the methodological considerations. A unique method is introduced for the creation of enclosure maps, to image patterns of residential walling on a neighbourhood level. Chapter 4 and chapter 5 provide an analysis of both cases individually. Chapter 6 provides a comparison and it tests the main

8

hypotheses. Finally, chapter 7 presents the main results, limitations and contributions of this research.

2. Theoretical Framework

This chapter introduces the main theoretical concepts and their particular use in this research. Section 2.1. describes the conceptualization of ‘enclosure’ in contemporary literature. Section 2.2. explains the role of multiple dynamics, both global and local in nature, and its contribution to enclosure in the urban realm. Section 2.3. describes elements of the study of morphology and the urban form in order to interpret different neighbourhoods in the urban context. The local conditions in both Accra (section 2.4.) and Pretoria (section 2.5.) are sketched, related to urban growth and development of enclosure. Finally, section 2.6. poses the main expectations, related to enclosure in neighbourhoods in Accra and Pretoria, in the form of hypotheses.

2.1. Enclosure

‘Enclosure’ is the central concept in this research. The conceptualization and measurement model from Adcock & Collier (2001) (appendix, figure 108) clarifies how the concept supports this research. At the background-level, enclosure is generally explained as:

“… the transformation of commonable lands into privately owned hands, and the concomitant extinction of common rights to land and resources ….”

(Jeffrey et al, 2012: 1247).

In recent years, the concept is seen as a key process of neoliberal globalization, which produces certain forms of spatiality and subjectivity. Enclosure is not only focused on the physical–legal process of enclosing land, but also on dispossession. This means that certain groups are excluded from land, services and knowledge they used to have access to. In sum, the multiple senses of enclosure described in the literature are: privatisation, physical fortressing and control, displacement and exclusion (Hodkinson, 2012: 506-509). When entering level 2 in the Adcock & Collier’s model, the concept enclosure is formulated more explicitly, suitable for the aims of the research. With a spatial-geographical dimension, enclosure is best described as the way in which property is territorialized. This process of territorialisation is seen as a dynamic process (Blomley, 2016).

Entering level 3 in the step-approach (indicators) makes the concept enclosure more tangible. Physical walls and boundaries constitute the most rudimentary and geographically obvious form of enclosure (Jeffrey et al, 2012: 1250). Other indicators like gentrification, applicable in the Global North, and its related dynamics are seen as a strategy of enclosure. In the Global South, demolition and renovation of informal settlements can have the same effects (Blomley, 2008; Jeffrey et al, 2012). In both cases, economic and social hierarchy within society plays a role in initiating enclosure. This research focuses on the most basic forms of enclosure: walling as a process. Level 4 of the step-approach is marked by deciding what kind of scores are representative for the indicators that are measured. Based on the method to mark walls in the urban structure, explained in section 3.3.2., multiple

9

neighbourhoods within Accra and Pretoria are mapped based on the level and character of enclosure.

2.2. Dynamics contributively to enclosure

Enclosure cannot be observed as a unitary phenomenon. The characteristics of enclosure vary across local contexts and its form is dependent on national and historical context (Rosen & Grant, 2011). To better understand the dynamics which contribute to forms of enclosure, the role of ‘danger and safety measures’, ‘privatization’, ‘territoriality’ and ‘land ownership’ is explained.

2.2.1. Danger and safety measures

The current field of literature recognizes a strong relationship between urban design and public behaviour. Urban forms mediate certain distribution of safety and danger (Dovey, 2000: 10). An attempt is made to control levels of crime and violence in the city by interventions in the urban design, in order to create ‘successful’ urban environments. There are several approaches to create a safer environment. This research focuses on the fortress approach, which involves walls, barriers, gates, physical segregation, privatisation and control of territory and strategies of exclusion (Carmona et al, 2003: 124). Physical interventions like walls, are seen as a response to the fear of crime and distance oneself from this perceived threat. The appeal of the fortress approach lies in its visibility. It is an activity towards creating a safer environment that can be directly observed.

In the second half of the 20th century, enclosure became a way to get a sense of control over criminalised spaces in the inner-city. The retreat from the inner-city to the suburbs can be seen as a form of enclosure. It is a way to separate oneself from that which is strange or different (Dovey, 2000: 11; Newman, 1972). Despite the idea that physical interventions may result in positive effects by redistributing crime and violence, Dovey (2000) claims that the vital elements of the urban public realm (safety, diversity, accessibility, vitality, creativity, and democracy) are not sustained by enclosure, because the rich and the poor become more segregated. Minton (2018: 86) contributes to this claim by arguing that the increasing levels of security and enclosure may be counterproductive. It arguably creates a heightened sense of fear and it limits the expression of the democratic society. Enclosure affects both the lives of the urban poor and urban elites by creating forms of separateness. Their way of life and interactions in public- or private spheres are transformed as a result of physical walling (Caldeira, 2000).

2.2.2. Privatization

Urban enclosure can be linked to strategies of capitalist growth focused on the process of privatization and dispossession of resources. Simply put, privatization in this context means the transformation of public space into private hands. The growing levels of privatization in the urban structure since the 1980s can thereby be seen as an outcome of the broader

10

pattern of regulation and controlled interventions in the urban design, initiated by neoliberal globalization (Dovey, 2000; Hodkinson, 2012; Minton, 2018).

With the aims of privatization, former public spaces and services are transformed for the benefit of capital. Streets become designed for vehicles only, without sidewalks. Shopping areas got internalized and wealthy areas got isolated. As a result, the urban poor are displaced and excluded from particular spaces in the city. Privatization is thereby focused on walling or fencing off certain spaces to secure separation between the urban poor and urban elites, who can realize exchange value and profit from it. This development intensifies urban fragmentation and the lack of usage of public space in the city (Caldeira, 2000: 307; Hodkinson, 2012; Jeffrey et al, 2012).

2.2.3. Territoriality

Different approaches exist to describe ‘territoriality’. This research takes a non-psychological, geographical approach, because of its applicability at any scale level:

“By human territoriality I mean the attempt to affect, influence and control actions and interactions (of people, things and relationships) by asserting and attempting to enforce control over a geographic area.” (Sack, 1983: 55).

This research analyses territoriality at the neighbourhood level. Territoriality is seen as a biological drive or instinct. Despite the variation in scale and size of what people may consider as their natural home or territory, they may feel more comfortable when they are in control of their territory. Territoriality must be seen as something that is socially constructed and territorial relationships are defined within a specific social context (Sack, 1983; Vollaard, 2009). “In other words, a territory does not exist without human activity regulating access to a purposefully delimited geographical area (Vollaard, 2009: 26).”

Enclosure is seen as a result of territoriality. Territorial boundaries (symbolic or physical) are created to enforce the control over geographic areas. The development of physical walls is seen as a way to exercise power over space. These boundaries give strength to the sense of ownership and it facilitates economic, political and social life. Simultaneously, boundaries discourage strangers and criminals to enter, creating safe spaces. The emphasis on territory and individual ownership in urban policies can lead to the development of high-security residences. In short, territoriality is described as a specific kind of power that uses space as its medium (Kärrholm, 2007; Newman, 1972; Rosen & Grant, 2011; Sack, 1983 & 1986). 2.2.4. Land Ownership

Finally, enclosure is linked to the concept of land ownership. The global rush for land is caused by economic globalisation and the extended reach of international capital and markets to societies in the Global South. This results in a rapid growth of large-scale land acquisitions by a great variety of (new) actors (Cotula, 2013: 1610; Zoomers, 2011: 12). The

11

increase of enclosure in the urban structure is seen as a result of this rush for land ownership:

“Large land deals involve the economic transacting of land, which leads to the fencing off of areas that had previously been used as common property resources and according to non-commercial criteria. (Cotula, 2013: 1610)”

Land is conceptualized as a place to live and being important for shaping personal identity. It provides access to natural resources and it can provide food, income and employment. Securing access to land is seen as a crucial condition for promoting local development and to combat poverty (Zoomers, 2011: 12). Just like other commodities, land is an asset for investment. The global rush for land ownership led to an increase in land commodification and competing land claims in the Global South (Cotula, 2013). This eventually leads to the displacement of the local population. Especially for people who do not own legal property rights and do not enjoy much protection from the law. The urban poor are the first ones to sell their land in return for capital. The increase in enclosure in the urban structure is seen as a result of this change in land ownership. Local groups become excluded from certain areas (Zoomers, 2011).

2.3. Urban Morphology

To interpret multiple neighbourhoods within Accra and Pretoria, the study of urban morphology is introduced. Urban morphology is the study of the form and shape of settlements. Appreciation of morphology helps to gain awareness of local patterns of development and processes of change in the urban form. Morphologists describe settlements in terms of several key elements of which land use, building structures, plot pattern and street pattern are considered crucial (Carmona et al, 2003: 61). The interpretation of plot patterns and the phenomenon of enclosure within the urban structure are elements of morphology used in this research. In order to gain a better understanding of the characteristics and dynamics of enclosure, the urban form of various neighbourhoods in Accra and Pretoria is studied. The ‘urban form’ can be defined as:

“The physical patterns, layouts, and structures that make up an urban center …. As the most basic canvas upon which settled human societies are built, urban forms are critical to both our daily lives right now and our interpretations of past cultures. …. Urban forms are ever changing, adapting with every new building, park, sidewalk, road, or gate that's erected.” (Muscato, 2018).

The urban form within and between cities is very different. Section 2.3.1. presents a rough categorization of morphologies to create some order in the wide variation of urban forms. This categorization, based on the work of Taubenböck et al (2018), aids the comparison between the morphologies of neighbourhoods within and between different areas and cities. Section 2.3.2. makes use of the study of morphology by defining the concept of ‘neighbourhood’ and explaining how ‘plot patterns’ can serve as an analytical tool.

12

The presence of physical walling to indicate levels of enclosure in the urban form is the main focus in this research. Patterns of enclosure can be traced back in the built structure of a neighbourhood (Hillier, 2013). Before enclosure is studied within the neighbourhood, it is necessary to interpret the morphological character of different neighbourhoods and to be aware of the variety of urban forms between and across cities. Although someone might expect strong similarities between neighbourhoods that for example are known to inhabit urban poor. These neighbourhoods often differ in its physical characteristics. In an attempt to distinguish different types of morphologies in the expanding cities of Accra and Pretoria and to make various neighbourhoods in both cities more comparable, a categorization of the built morphologies of different neighbourhoods is made.

The first step, is to be aware of the rough division made between formal planned areas, which are designed and coordinated, and the more organic, complex and dense areas which are developed without centralized planning (Muscato, 2018; Taubenböck et al, 2018). To differentiate organic neighbourhoods from planned neighbourhoods scholars focus on “the physical features characterizing the built environment by indicators such as high heterogeneity in building alignments, irregular street networks, small building sizes or high building densities (Taubenböck et al, 2018: 152).” A precise categorization is complex because building densities, building sizes and the geometric organization of urban blocks show a huge variety within neighbourhoods itself. Both large and small buildings, high versus comparatively low building densities, and complex and ordered alignments of buildings can be present in the same area. To avoid the complexity of categorising the morphology of neighbourhoods as a whole, this research focuses on categorising the morphology of urban blocks within neighbourhoods. Based on a simplified morphologic categorization, developed by Taubenböck et al (2018: 162), a distinction between 1) structured, 2) mixed structured-unstructured and 3) structured-unstructured morphologies is made. Based on the interpretation of satellite images in Accra and Pretoria, the unstructured and structured morphology are built up in two sub-categories. Each of the five morphological categories are illustrated by examples of both Accra and Pretoria. A first impression of the level and form of physical walls per category is taken into account in the description. Set up in this manner, it becomes clear how each category relates to a certain type of enclosure.

1a. The structured villa-type standardized morphology: contains planned, formal, structured blocks which contain geometric alignments, large/high buildings and are relatively low in density.

1b. The structured residencies with variation in plot size and empty plots morphology: has similarities with the previous category. The urban blocks are structurally designed in nature, the difference is the variation in plot size and the presence of empty plots that characterizes this morphology. This morphology frequently occurs in relatively new developing, growing areas.

13

Figure 1 to 4 image structured urban blocks in Accra and Pretoria. In structured blocks, plot sizes are relatively large, and the subdivision of the blocks is equal in nature. The level of enclosure seems high at first glance, as walls secure the villa-type residences. The structured

morphology with variation in plot size and empty plots within urban blocks (1b) is only

identified in Accra (figure 2). This morphology typifies growing peri-urban areas. Urban blocks contain new built houses in various stages of completion. Some plots are surrounded by walls only. Other plots contain initial structures awaiting further consolidation. Residential walls are built to safeguard the property from potential thieves (Gough & Yankson, 2009).

2. The mixed structured-unstructured morphology: in which both patterns of formal, structured building blocks (e.g. geometric alignments, frequent spatial transition of buildings and open spaces) and features of complex, dense small shelters are intertwined. The morphology is by trend closer to structured, formal neighbourhoods (Taubenböck et al, 2018: 162).

Figure 4 to 8 represent various mixed structured-unstructured urban blocks (2) in Accra and Pretoria. The urban blocks tend to be structured in nature, but the plots contain houses of varying size and type. Relatively large structured residences alternate with a more complex shaped small type of shelters. Fully enclosed plots can be distinguished from the small type of shelters which are not physically walled.

14

3a. The unstructured, demarcated morphology: in which urban blocks are characterized by informal or unplanned plots. Demarcation of plots in the form of fences or walls is recognized in these urban blocks.

3b. The unstructured slum-like morphology: in which the physical appearance is dominantly high in density with a complex pattern of deprived building types. The morphology of this type of neighbourhood features at least one extreme physical indicator of the built environment: high heterogeneity in building alignments, irregular street networks, small building sizes or high building densities (Taubenböck et al, 2018: 152).

The unstructured blocks in Accra and Pretoria contain a heterogenic building style with a high building density (figure 9 to 12). The unstructured, demarcated blocks (3a) in Accra differ from the mixed structured-unstructured blocks (2) in the sense that the plots are unplanned in nature. Demarcation in the form of walls is still recognized around the villa-type residences (figure 9). In Pretoria, unstructured, demarcated blocks (3a) are more structured in nature. The relatively small houses of a deprived building type characterize the area. These buildings are surrounded by fences that indicate the plots (figure 11).

15

The unstructured slum-like blocks (3b) in Accra contain relatively small, deprived and high dense built structures. Demarcation of property in the urban structure appears to be absent (figure 10). Similar urban blocks in Pretoria contain more dense, small, deprived built structures in which demarcation cannot clearly be recognized compared to category 3a (figure 12).

Using this morphological categorization, urban blocks within various neighbourhoods can be defined. Different neighbourhoods within and between cities can be compared based on its morphology. The next step is to analyse patterns of enclosure within these morphologies. This categorization is a simplified way to divide urban morphologies. Figure 1 to 12 provide a first impression of the level of enclosure in different types of morphologies.

2.3.2. Neighbourhoods and plot patterns

In the cases of Accra and Pretoria, I select multiple units of analysis at the neighbourhood level. Usually, neighbourhoods are defined in terms of its population, area or as a derivation of one or more social communities found in an area. Created as a top-down construction, neighbourhoods are conceptualized as discrete units with clear boundaries and a distinct territory. City plans are often established bounded to these territories. Neighbourhoods can be classified in the following five types (Carmona et al, 2003: 115):

16

1. Arbitrary neighbourhoods: the only common feature is spatial proximity.

2. Ecological/ethnological neighbourhoods: a common environment and shared identity. 3.Homogeneous neighbourhoods: inhabited by a socio-economic or ethnic group. 4. Functional neighbourhoods: related to shared service provision in the area.

5.Community neighbourhoods: a socially homogeneous group engages in primary contact. The neighbourhood level is the most suitable scale level for the selection of the units of analysis, because it is the most convenient way to select an area based on its geographical boundaries. Secondly, neighbourhoods are suitable for researching the urban form, as it is not necessary to focus on social or economic interaction within the city. Furthermore, demographical data provided by the city’s municipalities is related to the neighbourhood level. This offers the ability to explain characteristics of a particular area in more detail. I select neighbourhoods in Accra and Pretoria based on their geographical boundaries traced in Google Earth Pro (2019). Satellite images are generated and compared by analysing plot patterns and the phenomenon of enclosure within the urban structure. The method to analyse urban walling is explained in section 3.3.2. The use of plot patterns as an analytical tool is explained below.

Urban blocks or cadastral units are generally subdivided into plots or lots. In structured morphologies plots are set up back to back. Each having a frontage onto the street and a shared boundary at the rear. In some cases, plots also have service alleys at the rear. Plots with a frontage onto a main street at each end are rarer. Over time, the form and boundaries of plots within urban blocks may change. Plots can be bought or sold, subdivided or amalgamated and thereby increase or decrease in size. In general, buildings change more rapidly than plot patterns. But when plot patterns change, it is more often the case of amalgamation (Carmona et al, 2003: 63). The size of plots and the way urban blocks are subdivided says something about the way a neighbourhood is structured. Structured urban

blocks (1ab) consist of relatively equal subdivided urban blocks that contain relatively large

plots often surrounded by walls. While the unstructured areas (3ab) are known for a more uneven subdivision of plots within urban blocks that contain smaller plots where walling seems absent (section 2.3.1.). Although this is a rough categorization, it is helpful in categorizing various morphologies and its relation to enclosure within the city.

2.4. The context of Accra

Urban development in Ghana is characterized by rapid growth in peri-urban areas. The movement of people in their search for cheap residential accommodation in the city is the cause for high levels of population growth. They move away from densified, more expensive, inner-city areas. Accra is by far the largest city in Ghana, both in size and in numbers. The number of inhabitants and density increased heavily in the past decades. According to the Atlas of Urban Expansion (2016), the population between 2000 and 2014 increased from 2,513,026 to 4,429,649. The urban extent is 87,212 hectares. The density measured in 2014 is 51 people per hectare. The urban growth pattern reveals outward urban sprawl in peri-urban areas in combination with densification in the inner-city. Peri-peri-urban residential development, in various stages of completion, tend to be concentrated in and around

17

historic pre-existing villages. They contain a variety of advantages like basic levels of critical services (Owusu et al, 2011; Yeboah, 2003; Doan & Oduro, 2012).

From a local-historical perspective, land management in the largest part of Ghana is driven by indigenous tenure systems. Land is controlled by Chiefs who are trustees on behalf of the people. Chiefs constitute political and territorial authority. Property rights to land are thus quite limited for the government (Lund, 2013: 18; Gough & Yankson, 2000). Families and Chiefs hold majorities of land in Accra. Many families make claims to the same land which results in a low value of land in peri-urban Accra. Cheap land encourages residential development. Conflicts over land between planning institutions and Chiefs are the result of this land-tenure system. While the registration, planning, and subdivision of land in the city is the responsibility of state agencies and local authorities, Chiefs grant rights to any person who pays the right price for the land. In these cases, local authorities have no clue as to what is going on in their jurisdiction. This results in unplanned spontaneous residential development (Yeboah, 2000: 83-84; Yeboah, 2003: 117).

2.5. The context of Pretoria

South Africa’s political history still strongly influences contemporary urban life. On 26 May 1948 the National Party won the elections under the slogan of Apartheid. The Apartheid ideology forced different ethnic groups to live in designated, separate areas. From the 1940s to the 1970s, over a million housing units were built in the form of public housing estates or townships. They were located at the urban edge on cheap land, a safe distance from the ‘white’ city centre. Racial segregation was enshrined into the urban fabric (Landman, 2017; Mabin, Butcher & Bloch, 2013: 174; Ross, 2008: 122-123). Since the formal end of the Apartheid regime in 1994, cities in South Africa transformed spatially. In the 1990s, the ANC won the elections and Nelson Mandela became the political leader of South Africa. The ANC promoted the spatial integration and social inclusion of a different ethnic classes and they introduced measures to improve the welfare of the poor. The free movement of people led to an increase in urbanisation and crime in South African cities (Clarno, 2013: 1193; Landman, 2017). Pretoria, as part of the city region Tshwane, deals with an increase in urbanisation. According to Wazimap (2016), the population between 2011 and 2016 increased from 2,921,488 to 3,275,152. Due to the growing levels of insecurity and major changes within South African cities, enclosure in the urban structure increased. The racial division and urban inequality in the urban landscape still persist (Landman, 2017).

In comparison to the urban development in Accra, the local governments in South Africa are more involved in the structured planning of major cities. From as early as the beginning of the 20th _{century, planning authorities are involved in coordinating and servicing Pretoria’s}

expanding suburbs and townships. During the 20th century, the introduction of certain laws empowered the position of planning authorities even more (Mabin, 1995). Post-Apartheid residential development was initially focused on discouraging urban sprawl at the urban edges and the contribution towards the development of compact cities. More recently, residential development policies focus on achieving social justice and spatial equality in different neighbourhoods. Since the mid-1990s, local governments are granted powers and responsibilities to provide basic services to all residents. However, because of time- and

18

budgetary issues, the local government is not able to provide sufficient housing and services for low-income groups at the urban edges (Horn, 2018).

2.6. Expectations

This section presents five expectations in regard to enclosure and the characteristics of urban growth in Accra and Pretoria. These expectations are based on the theory and urban context described in the sections above. The hypotheses are tested in chapter 6.

Crime and violence can be controlled via interventions in the urban design. With the fortress approach, physical interventions like walls are used to create a safer environment. Relatively high levels of crime and violence in Pretoria, compared to Accra, would enhance the use of the fortress approach in the urban design (table 1; section 3.2.1.). The expectation is that the quantity of walls in the morphology of Pretoria is higher than in Accra.

Hypothesis 1: The heavy presence of safety issues in Pretoria results in a

relatively high level of urban residential walling compared to Accra.

In Accra, growth is particularly high in the peri-urban areas. Predominantly around pre-existing villages. In the contemporary land-tenure system, urban growth in the peri-urban areas is not controlled by local authorities. Chiefs and families have the authority to sell small pieces of land to private (individual) developers. This system results in the development unplanned urban blocks. In Pretoria, significant growth is high at the urban edges. Local authorities are granted with more power and responsibility to develop compact, structured neighbourhoods which are more socially just. The urban form of growing neighbourhoods in Accra and Pretoria is expected to be different.

Hypothesis 2: Plot sizes in growing neighbourhoods in Pretoria are more equally

subdivided than in growing neighbourhoods in Accra.

Hypothesis 3: The lack of top-down (governmental) planning in combination with

the dominant role of Chiefs in Accra results in a more unstructured, unplanned and unequally formed neighbourhoods in the urban growing areas.

Hypothesis 4: The more dominant role of the local municipality in Pretoria results

in more structured and equally subdivided urban blocks within neighbourhoods.

Residential enclosure is evident in various morphologies in the urban structure of Accra and Pretoria (section 2.3.1.). Walls are mainly built to secure and control property. In Accra, walls surround new developed plots in peri-urban areas. Even before the full completion of houses, walls are built to protect the property. In Pretoria, it is expected that residential walls are built around existing plots as part of the fortress approach. At the same time, the number of new developed residential walls would exceed the number of new developed plots.

19

3. Methodology

This chapter presents a justification of the research methods. Section 3.1. focuses on the research aims. Section 3.2. discusses the case study considerations and research design. Finally, section 3.3. focuses on the elaboration of the analysis: 1) The types of data that are used, 2) how the enclosure analysis method is performed and 3) the way in which the units of analysis are compared.

3.1. Research Aims

Traditionally, ‘enclosure’ is studied from the perspective of urban elites. Linked to that, the concept is described in its traditional forms such as gated communities and specified enclosed areas (Blomley, 2008; Hodkinson, 2012). To ensure that theorization of enclosure is not pre-set, the diversity of theoretical starting points must be taken into account (Robinson, 2015). The dynamics of enclosure in cities in the Global South can be better understood by analysing the phenomenon within the wider urban structure, instead of focussing solely on its traditional forms.

There is an indisputable need for new geographies of urban theory (Peck, 2015: 179). In the search to generate new global understandings of the urban, this research focuses on the dynamics of walling throughout the urban structure by analysing growing neighbourhoods in two expanding cities in the Global South (Robinson, 2015). Growing neighbourhoods are defined as the city neighbourhoods that have been (re)shaped in the past 20 years as a result of urban expansion. These particular neighbourhoods serve as a place of arrival, for living and working (Taubenböck et al, 2018). The selection of growing neighbourhoods in Accra and Pretoria is substantiated in the next section. A unique method is introduced to perform an enclosure analysis. The results are used to uncover and explore patterns of enclosure and to illustrate the structure of walled residences in the urban landscape. This research aims to contribute to the existing field of literature of enclosure, morphology and urban design. The focus on enclosure throughout the city beyond its traditional researched forms can provide new insights, because it is more open-minded and less based on pre-set theorization. Different groups produce, organize, reproduce, consume and interpret space in their own way. By looking both at diversity and similarity in the practices of enclosure between neighbourhoods within and between Accra and Pretoria, new insights about residential development forms are developed (Rosen & Grant, 2011: 779).

3.2. Case Study Considerations

This section contains two parts: 1) it presents the case study design and 2) it presents the selection of the multiple units of analysis with the use of maps and imagery. Starting from the city-level as a whole towards a neighbourhood scale selection. Within these neighbourhoods, satellite snapshots are selected for a closer enclosure analysis.

20

3.2.1. The multiple case design: why Accra and Pretoria?

A multiple case design with embedded units of analysis is used for this research. The cities of Accra in Ghana and Pretoria in South Africa are selected for a comparative case study, making it a cross-national comparison (Anderson, 2016; Yin, 2009). Both cases are selected to extend the knowledge about enclosure in cities in the Global South. Accra and Pretoria are similar in a way that they both experienced recent urban growth (see section 2.4. and 2.5.). Residential walling beyond its traditional forms can be found throughout neighbourhoods in both these cities.

What makes the cases particularly interesting for a comparison is the difference in their political history and perception of safety. Since its independence in 1957, civilian and military governments alternated in Ghana. In 1993, a multi-party democracy officially entered into force. In South Africa, the multi-party democracy became effective in 1997. Tensions between ethnic groups are still very present as a result of the Apartheid policies which ruled the country between 1961 and 1994 (CIA, 2019ab). The perception of safety in Accra and Pretoria is also differently interpreted by its citizens (table 1). Based on filtered surveys from the past 3 years, Numbeo (2019ab) provides a crime and safety index which are both the result of combined variables. These indices are well suited for usage in a cross-national comparison. The crime index in Accra is seen as moderate while the crime index in Pretoria is seen as very high. The safety index in Accra is seen as moderate while in Pretoria the situation is very unsafe. The particular history of South Africa contributes to this

perception of crime and safety.

To provide complementary knowledge about enclosure in an African context, Accra and Pretoria are selected with the aim of theoretical replication. Theoretical replication means that different results within both cases are expected based on contrasting local conditions (Yin, 2009). In section 2.6., these expectations are presented in the form of five hypotheses. 3.2.2. Units of analysis

Within the case studies of Accra and Pretoria, I selected multiple units of analysis in the form of neighbourhoods. This section substantiates the set-up of this selection procedure. Three scale levels are used to select comparable neighbourhoods: 1) the city as a whole, 2) city districts and 3) the neighbourhood (Jacobs, 1961: 117). A fourth step is necessary to select snapshots within the selected neighbourhoods. These snapshots are turned into maps based on the level of enclosure in the urban structure. First, the procedure to select satellite snapshots is explained in general (figure 13). Second, the procedure to select comparable

21

neighbourhoods is performed for the case of Accra and for the case of Pretoria. Finally, the selection procedure for the satellite snapshots is further explained.

1. City level: I provide a rough overview of the urban morphology in both Accra and Pretoria from a city-level scale. This overview is primarily based on the impressions generated by studying satellite images of both cities. During the Urban Lab project in 2018, I did explorative research on multiple neighbourhoods in Accra and Pretoria. This resulted in the development of the ‘enclosure analysis method’, a method that filters physical walling out of satellite images. Via Google Earth Pro (2019), it is possible to compare satellite images over time. Based on the first impressions of satellite images in combination with the categorization of morphologies (section 2.3.1.), I provide a rough description of the urban structure on a city-level scale.

2. District level: I motivate why certain city districts or greater neighbourhoods in Accra and Pretoria are interesting for an enclosure analysis. Comparable areas within both cities are selected based on the patterns of urban growth since 2000.

3. Neighbourhood level: Comparable neighbourhoods or wards within the city districts or greater neighbourhoods are selected for a further enclosure analysis. This selection is based on areas that experienced urban growth since 2000.

4. Satellite snapshots: Within the selected

neighbourhoods or wards, multiple

snapshots of satellite images are gathered, based on the representation of the urban form of each neighbourhood. For example, when the morphology of the selected neighbourhood contains structured blocks (1ab) and unstructured blocks (3ab), both categories are selected for a further enclosure analysis. An estimation of the morphological structure within the selected neighbourhoods is presented, besides the enclosure maps, to get a sense of the urban form of the neighbourhood as a whole (table 8 to 15; appendix).

Accra

1. City level: the overview of Accra on a city scale is based on the urban extent of the city within the Greater-Accra region (figure 14 & 15). The urban extent of the Ghanaian capital reaches far beyond its central area. It is made up of a combination of multiple districts within the Greater-Accra region. Between 2000 and 2014, Accra became more dense (pink areas) and its territory extended and merged towards pre-existing urban clusters (orange areas) (figure 15). Structured (1ab), mixed structured-unstructured (2) and unstructured

morphologies (3ab) make up the urban form of the city. Available data lacks in terms that it

22

structure. What is clear is that the urban form of the inner-city as well as peri-urban areas contain elements of each morphological category, there is no heavy concentration of one morphological category. The three main categories tend to be dispersed across the city (Weeks et al, 2009).

23

2. District level: First, I select comparable city districts (figure 14). With the use of parliamentary constituency areas, two sub-districts are selected as an intermediate step before the neighbourhood selection (GhanaDistricts, 2006; GhanaDistricts, 2007 & Wikipedia, 2018).

City Districts: The ‘Accra Metropolitan District’ experienced growth in its western area. The

area is divided in multiple sub-metropolitan districts that contain even more neighbourhoods. The ‘Ga East Municipal District’ is mainly built up of urban clusters connected to pre-existing villages. The district contains huge new developed residential areas.

Sub-Districts: ‘Ablekuma’ is a sub-metropolitan district on the west-side of the Accra

Metropolitan District, it is split in four constituency areas. The Ga East Municipal District contains huge growing areas, the ‘Dome-Kwabenya area’ which is in the South of the Ga East district is further researched. This sub-district is created as a parliamentary constituency in 2004 (Wikipedia, 2018). The two research areas are indicated with the red frames in figure 15.

3. Neighbourhood level: I select four neighbourhoods within the two sub-districts for a further enclosure analysis. Within Ablekuma, I select the neighbourhood ‘Tunga’ and ‘Mallam’ (figure 16 & 17). Both neighbourhoods are situated in the western part of the sub-district. Within the Dome-Kwabenya constituency, I select the neighbourhood ‘Kwabenya’ and ‘Taifa’ (figure 18 & 19). In total, four neighbourhoods are selected within the parts of the city districts that experienced recent urban growth (Atlas of Urban Expansion, 2016). Table 2 presents the surface area of the selected areas. Figure 24 provides an overview of this selection procedure itself.

25

Pretoria

1. City level: Pretoria is situated within the greater urban area of Tshwane in the Gauteng province (figure 20). The former Apartheid policies and the role of the planning authorities still strongly influences the urban structure of the city (section 2.5.). The lay-out of the city is dominated by traditional grid patterns and square urban blocks can be identified in between. The centre of the city is known as a working environment characterized by its high-rise, structured, formally planned neighbourhoods. Structurally designed gated communities are found in most of its suburbs. Suburban expansion accelerated since the introduction of Apartheid policies in the 1960s. Mostly the wealthy, white citizens moved to these areas. On the urban edges, townships were developed by South African planning agencies. These areas were meant as residencies for black communities. Recent urban growth takes place at the edges of the city. A mix of structured (1ab), mixed structured-unstructured (2) and

unstructured morphologies (3ab) characterize these areas (Mabin, 1995 & Google Earth Pro,

2019).

2. District level: In comparison to Accra, similar sized city-districts are not found in Pretoria. Instead, I select two greater neighbourhoods at the western and eastern edge of Pretoria which have a comparable size as the sub-districts in Accra. The two sides experienced recent residential development, both in the form of structured and unstructured morphologies.

Sub-Districts: ‘Mamelodi’ is situated at the east-side of the capital, the latest extensions of

the area are shaped in eastern direction (figure 21). Mamelodi is built up in six sub-areas. Mamelodi East and Nellmapius are two areas situated at the eastern edge of the district that experienced urban growth since 2004. ‘Atteridgeville’ is situated on the west-side of Pretoria. At the western edge, growth takes shape in the form of informal settlements. The core area, in which the township was originally developed, is named Saulsville.

26

3. Neighbourhood level: The city of Pretoria is subdivided in wards. Wards are created to divide the City of Tshwane in measurable units, they are well-suited to serve as research areas (Wazimap, 2016). The wards in Pretoria are similar in size as the selected neighbourhoods in Accra. Ward 17 and ward 10 are selected within ‘Mamelodi East’. ‘Nellmapius’ is situated within ward 86 and ward 15 (figure 22). Table 3 presents the surface area of the selected areas. Figure 24 provides an overview of this selection procedure itself. Within ‘Atteridgeville’, ward 107, ward 68 and 63 (Saulsville) and ward 72 are selected for an enclosure analysis (figure 23).

27

4. Satellite snapshots: To perform the enclosure analysis, I analyse multiple snapshots of satellite images within each selected neighbourhood or ward. In order to get a representative analysis, the various morphologies within the selected neighbourhoods need to be captured. A morphological map of each selected neighbourhood is drawn to get an overview of its urban structure. Urban blocks within each neighbourhood are classified based on their morphological character. With the use of the categorization of the morphologies (section 2.3.1.) and Google Earth Pro (2019) I created these maps. Next, I select snapshots within the selected neighbourhoods with the aim to analyse all the various morphologies found in one neighbourhood. Table 8 to 15 (appendix) present an estimation of the distribution of morphologies in a neighbourhood. I select snapshots of (+/-) 0.2 km² with an circumference of (+/-) 1.85 km to create a comparison on an equal scale-level. With the use of the enclosure analysis it is imaged how each morphology relates to urban walling.

3.3. Preparation of the analysis

This section focuses on the elaboration of the analysis. The different types of data, the method to perform the enclosure analysis and the comparison of the units of analysis are discussed.

3.3.1. Data

Multiple types of data sources are used in the analysis (chapter 4, 5 and 6). Both primary and secondary data are gathered to explain the dynamics of enclosure in the urban morphology.

28

Satellite images, gathered via Google Earth Pro (2019), Google Maps (2019) and PlanetGIS (2016), form the main source for developing the analysis. The enclosure analysis is used to compare the multiple units of analysis within and between Accra and Pretoria. Via Google Street View, imagery of neighbourhoods is used to get a better view of the characteristics of the selected neighbourhoods.

The case study of Accra is conducted as a desk-based orientation. Secondary sources in the form of academic articles are used to provide the narrative. An implication of this is that the comparative analysis is mainly focused on the enclosure analysis, performed with the use of satellite images. For the case study of Pretoria, I conducted fieldwork in South Africa. This fieldwork was first focused on the gathering of primary sources of the selected neighbourhoods, in the form of photo-material. The photos are focused on 1) the form/type of enclosure, 2) the differences within neighbourhoods related to the size of the plots and presence of enclosure and 3) imaging the ambience in a neighbourhood. Secondly, with the use of key-interviews it was possible to gather background information about 1) the characteristics of the selected neighbourhoods in Pretoria and 2) about the motivations for residents to enclose their property. Thirdly, secondary data in the form of policy documents focused on the four selected neighbourhoods in Pretoria is collected from the municipality. These documents help to better understand residential developments in the selected neighbourhoods.

3.3.2. Enclosure Analysis

In order to filter physical walls out of the urban structure, I developed a method during the Urban Lab project at the University of Amsterdam in 2018. With the use of satellite imagery, the presence of walls can be filtered out of the urban morphology. Satellite images are generated by Remote Sensing and used to get a better view of the surface of the earth. The analysis is aimed at the selected neighbourhoods in both Accra and Pretoria. ‘Sebrepor’ and ‘Tunga’ in Accra, and ‘Nellmapius’ in Pretoria, were analysed during this Urban Lab project. In the form of satellite images, parts of these neighbourhoods were printed on A3-format. In these printed images, walls are marked with the use of a black coloured marker. I make use of Google Street View to create more clarity on the exact presence of walls and the characteristics of a neighbourhood. In some cases, the visibility of walls in satellite images is impeded by the presence of vegetation or by the low resolution of the satellite images. Google Street View helps to lower the amount of error within the analysis and to get an overview of the characteristics of a neighbourhood. For instance, the height and characteristics of buildings and the quality of physical boundaries. Google Street View also gives an impression where walling is not present. This information is used to find out more about the dynamics and the possible patterns of enclosure. When the process of marking is done, the ink of the marker is visible on the backside of the printed satellite images. After redrawing these lines, a map of the structure of walls is created. These maps need to be mirrored with the use of Paint, because the created map is visible on the backside of the printed satellite images. After mirroring the created maps, the structure of walls in the selected neighbourhood can be linked to the original satellite images. This method is used to retrieve the level of enclosure in neighbourhoods in Accra and Pretoria in the form of a map.

29

3.3.3. The case studies and comparison

Theoretical replication is the aim of this research, as different results within both cases are expected based on contrasting local conditions (Yin, 2009). Both similar and contrasting features of neighbourhoods within and between Accra and Pretoria provide new insights of urban theory. On one hand, the similarities of the dynamics of enclosure and urban forms in both Accra and Pretoria are analysed in order to come up with new insights related to cities in the Global South. On the other hand, divergent circumstances, localized complexity and un-patterned diversity are taken into account by focussing on the differences between neighbourhoods in the two cities. The role of uniqueness and particularity in urban theory has gained value in the current field of urban studies, therefore context-specific information is filtered for the cases of Accra and Pretoria (Peck, 2015: 161; Robinson, 2015; Anderson, 2016).

Case studies

The analysis first focuses on each case study individually. Chapter 4 and 5 present the characteristics of the selected neighbourhoods related to the morphology and enclosure in the urban structure. I create two types of maps for the selected neighbourhoods in order to analyse the patterns of enclosure per morphology.

1. Morphological maps: For each of the selected neighbourhoods, I create a map that captures the morphological character of the urban blocks. Urban blocks are classified along six morphological categories. Five categories are presented in section 2.3.1., the

unstructured mixed slum-like and territorialized urban blocks category is added in the

maps. This category is placed in between the presented unstructured categories (3a and 3b). Larger non-residential areas are indicated with a yellow marker. The snapshots selected for a further enclosure analysis are indicated with a black frame, just as the boundaries of the selected neighbourhood.

2. Enclosure Analysis: Based on the morphological maps, I select snapshots within the neighbourhoods. I select the snapshots in such a way that I capture the variety in morphologies in the enclosure analysis. This way, the level of walling in a neighbourhood can be interpreted for every morphology. The performance of the enclosure analysis is described in section 3.3.2.

3. Level of enclosure per morphology: The previous steps are combined to present an overview of the level of enclosure in growing neighbourhoods within each case. The level of enclosure is linked to the 5 main morphological categories.

30

The Comparison

The analysis mainly focuses on the comparison between the two cities. Chapter 6 presents a comparison between Accra and Pretoria. This chapter 1) explains the development of urban growth related to the urban morphology and pattern of enclosure per case. 2) It analyses the main differences and similarities for each morphological category (section 2.3.1.) linked to patterns of enclosure within. 3) It uses four different methods to test the hypotheses introduced in section 2.6.

1. Hypothesis 1; level of enclosure: I use a quantitative method to calculate the number of enclosed plots per morphological category for each city. First, I calculate the total surface area of each morphology over the selected neighbourhoods. Second, I come up with a percentage of enclosed plots per morphology. Third, I combine step 1 and 2 in the form of a score for each morphology. Finally, I link the scores to the surface area of residential built environment only. The result is a percentage of enclosed plots in the built environment in the selected neighbourhoods.

2. Hypothesis 2; subdivision of plots: I use a qualitative approach to analyse the design and subdivision of urban blocks. I analyse representative blocks within structured (1ab),

mixed structured-unstructured (2) and unstructured morphologies (3ab) in detail to

distinguish various sized and shaped plots in the neighbourhood structure. 3. Hypothesis 3 & 4; unplanned vs. planned urban

blocks: I use a quantitative approach to combine the distribution of the morphologies (step 1, hypothesis 1) with the design of urban blocks (hypothesis 2). First, I quantify the design of the urban blocks with the use of a point system (table 4). Points are awarded based on the morphology in combination with the subdivision of blocks, which results in a score. Second, this score is linked to the distribution of morphologies over the residential area to calculate the level of structured, equal subdivided blocks over the selected neighbourhoods.

4. Hypothesis 5; a matter of time: Via Google Earth Pro (2019), the development of enclosure is analysed over time with a qualitative approach. Satellite images from the 21th century (since 2000-2003) are compared to recent satellite images or with the enclosure maps. It may reveal something of the motivations for the development of walls in a particular neighbourhood.

31

4. Case Study I: Accra; Ghana

This chapter provides an interpretation of the performed enclosure analysis for the case of Accra. The provided coloured maps present the morphology of the four selected neighbourhoods, Tunga, Mallam, Kwabenya and Taifa. Selected satellite snapshots within these neighbourhoods are turned into maps that show the level of urban walling. The interpretation of both type of maps provide a sense of the characteristics of enclosure in the growing neighbourhoods in Accra. In this chapter I discuss 1) the characteristics of each neighbourhood related to its morphology and level of enclosure and 2) I explain the main observations related to enclosure for growing neighbourhoods in Accra.

4.1. Presentation of the morphology and level of enclosure per

neighbourhood; Accra

This section presents the morphological maps, enclosure maps and main observations for the selected neighbourhoods in Accra. For each neighbourhood, three types of maps are combined to come up with an interpretation of the characteristics of the neighbourhood. One map presents the morphology of the neighbourhood structure. A second map presents the level of enclosure within the neighbourhoods selected snapshots. A third map presents original satellite images of the selected snapshots as a frame of reference.

Tunga

Tunga is situated at the western edge of the sub-district Ablekuma. Closely adjacent to the Pambros Salt Ponds, an area non-suitable for urban development. The structure of urban blocks in Tunga was already evident in 2000. Since then, residential development increased at its western edge while densification took place in the rest of the neighbourhood. The morphology of Tunga is dominated by structured villa-type standardized blocks (1a). The majority of the urban blocks in the core area of the neighbourhood are equally subdivided and most plots are enclosed. The morphology of the western edge and a number of blocks in the core area contain structured residencies with variation in plot size and empty plots (1b). This morphology typifies blocks that experienced more recent residential development. Plots within both structured morphologies are dominantly enclosed.

Three urban blocks in the heart of the neighbourhood stand out as unstructured morphologies. Two of them are characterized by a slum-like morphology (3b), shaped as a core area of pre-existing villages. Small shacks or houses which lack residential walls are built towards these cores. The rims of these blocks contain enclosed villa-type residences, built in unstructured fashion (figure 25, 26 and 27). Because the structure of Tunga is dominated by enclosed plots, it seems that empty spaces within urban blocks lack housing. This

32

assumption is weakened by showing the original satellite images besides the enclosure maps (figure 28).

33

Mallam

Mallam is situated on the border between the Accra Metropolitan District and the Ga South Municipal District, 2 kilometres away from Tunga. The current urban structure of the neighbourhood is formed around a huge pre-existing village. Since 2000, the core area densified and new residential urban blocks developed around the village. Residential development in southern direction is limited because of the Pambros Salt Ponds. The morphology of Mallam is dominantly unstructured in nature. The pre-existing village in the heart of Mallam has a slum-like morphology (3b). Small shacks shape the high dense blocks, enclosure is highly absent. At the rims of the former village, some walled plots are randomly situated within the blocks. Unstructured, demarcated blocks (3a) surround this area. Villa-type, walled plots are built in an unequal manner. Some of the unstructured blocks contain enclosed plots. Other plots are separated from each other by a wall at one or two sides of the plot. The number of walled plots increases further away from the former village. The southern part of Mallam contains recent developed urban blocks with structured residencies

with variation in plot size and empty plots (1b) and industrial sites. The northern part of

Mallam contains structured and mixed structured-unstructured (2) blocks, small non-enclosed shacks alternate with non-enclosed villa-type houses (figure 32). The housing blocks in structured areas are clearly distinct in the enclosure map, walled plots are dominant and shape the housing blocks in these areas (figure 29, 30 and 31).

35

Kwabenya

Kwabenya is a peri-urban area in the South of the Ga East municipal district. In 2000, huge parts of the area were covered with forests and grasslands. Initial structures of urban blocks were only visible in in the southern part of the neighbourhood. Small villages are recognized in the area. Since 2000, residential development started to take place around these initial structures and pre-existing villages. Kwabenya spread its territory in northern direction. In 2019, the whole area of Kwabenya is filled with residential housing blocks. Villa-type

structured residencies with variation in plot size and empty plots (1b) dominate the structure

of Kwabenya. This variation is expressed in the form of residential walls in its urban blocks. Some plots are walled but do not contain houses, other plots contain initial structures of houses without enclosure (figure 38). This indicates that the built environment of Kwabenya is still in development. The size of plots in this morphology differentiates highly.

In the southern part of Kwabenya, structured villa-type standardized urban blocks (1a) occur. These blocks were part of the initial structures found in 2000, plots within this morphology are more equally subdivided and dominantly enclosed. I expect that urban blocks of category 1b can turn into category 1a over time. Mixed structured-unstructured urban blocks (2) are identified in the southern part of Kwabenya. Non-enclosed small shacks are built in between enclosed villa-type residences. In the north-east of Kwabenya, three unstructured areas with a slum-like morphology (3b) are distinct in the urban structure. Non-walled small shacks dominate the cores of these areas while the rims of these blocks contain larger enclosed plots. This part of Kwabenya is part of pre-existing villages (figure 33-37).

39

Taifa

Taifa is situated south from Kwabenya in the Ga East municipal district. Since 2000, densification took place in the central and southern area of the neighbourhood. The western flank is part of more recent developed urban blocks. The morphological structure of Taifa is variated. Two large areas with structured villa-type standardized urban blocks (1a) are found in the core and southern part of Taifa. Two areas with mixed-structured-unstructured urban

blocks (2) occur in the central and western part of the neighbourhood (figure 39). The

central area of Taifa, contains a high number of fully enclosed plots. The central situated blocks of morphological category 2, contain empty plots, plots with initial housing structures and plots filled with small shacks. The level of enclosed plots here is relatively high (figure 40, 41, 44 and 45).

Two areas with a slum-like morphology (3b) are identified at the eastern and western boundary of Taifa. Structures of pre-existing villages remain visible in these areas. The western part of Taifa is part of more recent residential development. The number of enclosed plots increases further away from the pre-existing structures of small shacks. Unstructured urban blocks with villa-type residences surround this area. The mixed

structured-unstructured morphology (2) in the western part of Taifa contains villa-type plots

which have a high variation in size. Non-enclosed empty plots or plots filled with small shacks are identified in between the villa-type residences (figure 42 to 45).

42

4.2. Patterns of enclosure in growing neighbourhoods; Accra

The statements made in this section are based on the presentation of the maps of the morphologies and level of enclosure in Tunga, Mallam, Kwabenya and Taifa, four growing neighbourhoods in Accra since 2000 (section 4.1.). Walls dominate the structure of blocks within the four selected neighbourhoods. In the analysis it becomes clear that the patterns of enclosure are closely related to the morphology of urban blocks. The morphological distribution in the selected neighbourhoods is presented in the appendix (table 8 to 11). The patterns of enclosure in Accra are now linked to the three main morphologies.

1ab. Structured morphologies: The structured villa-type standardized urban blocks (1a) exist in all of the four neighbourhoods. Blocks within these areas are fully subdivided by plots that all seem to be walled at one side at least, most of the plots are enclosed. This type of morphology (1a) dominates in Tunga, the more recent growing neighbourhoods contain smaller areas with this particular morphology. What is striking in Accra is the high