Settlement morphology, inclusive densification, and Sol Plaatje's spatial justice: NDBI clustering pedagogy for equitable urban growth in South African arid geographies

Assessing settlement morphology in environmentally constrained arid regions requires transcending conventional land-cover mapping by integrating remotely sensed built-up expansion with land tenure, infrastructure accessibility, and spatial justice metrics; accordingly, this study advances geospatial science by developing a replicable Composite Spatial Justice Index (CSJI) that transforms NDBI-based settlement analysis from a descriptive urban growth assessment into a quantitative framework for measuring and mapping tenure-based inequity, environmental vulnerability, and inclusive densification to support evidence-based planning in historically segregated landscapes. The settlement morphology and densification patterns is investigated in the Northern Cape Province from 2004 to 2024 of spatio-temporal analysis of built-up expansion, leveraging Normalized Difference Built-up Index (NDBI) spatial clustering as an analytical technique and pedagogical framework for inclusive urbanization. The research problem stems from the paradox of rapid urbanization and environmental degradation in arid landscapes, where uneven development trajectories perpetuate spatial injustice and ecological vulnerability. Employing a multi-scalar GIS methodology integrating Landsat-derived NDBI time-series, Global Moran's I Evaluation techniques, the results reveal that 78% of built-up growth occurs within 5 km of existing mining-transport corridors, while communal tenure areas experience degradation densities 4.8 times higher than freehold zones (66.9 km 2 /100 km 2 vs. 32.9 km 2 /100 km 2). Also, NDBI spatial clustering is a prolific pedagogical instrument for infrastructure network functioning in development and as vectors of exclusion. Operationalizing Solomon Plaatje's Spatial Justice Philosophy through GIS-based NDBI clustering, this study depicts infrastructure corridors as areas of unequal densification, as communal lands degraded 4.8 times more than freehold zones, thus transforming his critiques into a measurable framework for equitable urban growth in the study area. The study advocates evidence-based densification corridors, decolonized planning pedagogies, and strategic infrastructure redirection for democratizing urban growth in arid geographies.

The imperative for equitable human settlements in environmentally constrained regions demands analytical frameworks that integrate established geospatial techniques with conceptual approaches capable of confronting the enduring legacies of spatial injustice (Mkhitaryan et al., 2025; Nyamwanza et al., 2025). Across the Global South, rapid urbanization intersects with ecological fragility to produce landscapes of inequality in which settlement expansion often replicates historical exclusionary logics rather than advancing inclusive development (Myers, 2021; Hutchings et al., 2022). South Africa's Northern Cape Province, the nation's largest yet most sparsely populated province, exemplifies this challenge: its semi-arid to arid conditions make water security, vegetation health, and land stability critical determinants of sustainable settlement (Matlou et al., 2021; Bahta and Myeki, 2022), while its settlement geography still bears the imprint of colonial and apartheid spatial engineering, from the 1913 Natives Land Act to the Group Areas Act and Bantustan policy (Strauss, 2019; van Staden, 2025). These forces have produced a landscape in which mining and transport corridors function simultaneously as arteries of development and vectors of exclusion, concentrating investment in privileged corridors while peripheral communal areas experience accelerating degradation and service deprivation. Understanding this settlement morphology requires analytical approaches capable of revealing both the spatial signatures of historical injustice and pathways toward more equitable futures (Lobo et al., 2020; Parcero-Oubiña and Nión-Álvarez, 2021).

The application of Landsat imagery to land use/land cover (LULC) classification and built-up density mapping has matured into an extensive and methodologically diverse body of scholarship, spanning both classification technique and index design. At the classification-technique end, Ul Din and Mak (2021) analyzed four decades (1979–2020) of Landsat MSS, TM, and OLI imagery for Hyderabad, Pakistan, applying a Gaussian radial-basis-function support vector machine to retrieve five-class LULC maps, and reported built-up area increasing from 30.69 to 65.04% of the city over the study period, with accuracy metrics broadly comparable to those obtained in the present study. This aligns with Feng et al. (2023), whose study on LUCC impacts on Earth's eco-environments similarly underscores the need to move beyond descriptive land-cover mapping toward frameworks that link spatial transformation to the socio-environmental burdens it produces. Mohajane et al. (2018), working in the Azrou Forest of Morocco's Central Middle Atlas, similarly combined Landsat MSS, TM, ETM+, and OLI imagery across multiple decades to classify land cover, demonstrating that a consistent, sensor-spanning Landsat time series can reliably detect long-term land-cover transformation even in topographically and ecologically heterogeneous terrain, a methodological precedent this study follows in bridging Landsat 5 TM (2004) and Landsat 8 OLI (2014, 2024) within a single classification framework. At the index-design end of this literature, Zheng et al. (2021) demonstrated that NDBI, used in isolation, is prone to omission and commission errors in distinguishing built-up surfaces from bare soil and cropland, and proposed an Enhanced Nighttime Light Urban Index (ENUI) that couples NPP-VIIRS nighttime-light data with NDVI, NDWI, and NDBI to improve built-up-area extraction in the Guangdong–Hong Kong–Macao Greater Bay Area, achieving an average overall accuracy of 93.56% and a 9.98% improvement in urban F-score over a nighttime-light-only baseline. Collectively, these studies establish that multi-decadal, multi-sensor Landsat time series are a reliable and widely validated basis for LULC and built-up density mapping across diverse settings, and that NDBI performs most reliably not in isolation but when triangulated against complementary spectral or ancillary layers, a methodological lesson this study internalizes by pairing NDBI-derived built-up clusters with independent tenure and infrastructure-access layers rather than relying on the index as a stand-alone classifier.

The Normalized Difference Built-up Index (NDBI), derived from Landsat imagery, is a well-established technique for quantifying built-up density and urban expansion over time (Ali et al., 2019), and spatial clustering analysis through Global Moran's I is similarly well established for identifying spatial autocorrelation in uneven development patterns (Kumari et al., 2019; Gedamu et al., 2024). This study does not claim novelty for these individual techniques, nor for their combination as a workflow; NDBI and Moran's I /LISA have been jointly applied before to describe built-up growth and its spatial clustering (Gilbert and Shi, 2023; Negesse et al., 2024). What these prior applications share, however, is a common limitation: built-up expansion is treated as an outcome to be mapped and clustered, but the resulting clusters are not disaggregated by land tenure, nor linked to a service-deprivation or infrastructure-access measure, so a hotspot of degradation and a hotspot of privileged growth are visually indistinguishable in their output. Conversely, spatial justice scholarship identifies tenure and historical dispossession as the relevant axis of inequity but characteristically does so through critique and case narrative rather than through a replicable, ward-scale metric that a planning authority could apply to a new dataset without re-deriving the theory each time (Gonçalves et al., 2025). The contribution of this study is therefore quantitative, conceptual and operational by proposing, constructing, and validating a specific composite index, i.e., the CSJI that closes the gap between these two literatures by requiring every NDBI-derived degradation cluster to be reported jointly with its tenure classification and infrastructure-distance gradient, and not as an undifferentiated land-cover statistic. This study advances that tenure-and-access disaggregation is now a required, structural output of the analysis rather than an optional post-hoc interpretation, and this claim can be evaluated directly against the CSJI's construction in Section 2.5 and its ward-level outputs in Section 3.3. This study addresses that gap by embedding these established GIS methodologies grounded in the philosophy of Sol Plaatje, South Africa's renowned intellectual, writer, and founding Secretary-General of the South African Native National Congress, whose critiques of spatial dispossession and advocacy for justice, mobility, and equality provide a moral and conceptual basis for translating remote-sensing-derived settlement metrics into a measurable spatial justice diagnostic.

Problem statement

The Northern Cape is experiencing a difficult combination of rapid urban growth and increasing environmental degradation. As more people and development concentrate around towns and transport corridors, the natural environment is also changing. Between 2004 and 2024, grassland in the province declined substantially from 277,267.46 km 2 to 122,857.33 km 2, while barren land increased from 73,417.01 km 2 to 222,866.82 km 2. Over the same period, built-up land increased by approximately 292%, from 91.06 km 2 to 357.07 km 2. Importantly, this growth has not occurred evenly across the province; instead, development is largely concentrated around existing roads, transport routes, and infrastructure corridors. This creates an important planning problem because some areas and communities may benefit more from urban development and infrastructure investment, while others continue to experience environmental degradation, limited services, and fewer development opportunities. The main problem is therefore not simply that the Northern Cape is becoming more urbanized, but that we do not yet clearly understand where this growth is occurring, why it is concentrated in particular areas, and whether its benefits and environmental costs are being shared fairly. Historical land dispossession, differences in land ownership, infrastructure availability, and the location of communities may all influence these patterns. There is also a need for better spatial information to help planners and communities understand these changes and make fairer decisions about where future development should occur. This study therefore uses geospatial analysis to examine how built-up areas have changed across the Northern Cape between 2004 and 2024, identify where development is concentrated, and investigate how infrastructure access, land tenure, and environmental degradation are related to these patterns. In doing so, the study seeks to provide practical evidence for more inclusive urban development, environmental management, and spatial planning in the Northern Cape and other dry regions facing similar challenges.

Solomon Plaatje's philosophy in relation to the study

The philosophies of Sol Plaatje (1876–1932), journalist, novelist, linguist, and founding Secretary-General of the South African Native National Congress (Salawu, 2025), offer a conceptual and ethical framework for this investigation. Plaatje's work was animated by four interconnected principles taken up here: spatial justice and legal equality, resistance to land dispossession, freedom of movement as a natural right, and education through exposure and documentation (Welman et al., 2021). His account in Native Life in South Africa (1916) of the 1913 Natives Land Act's human consequences (Mofokeng, 2026) established a tradition of exposing spatial injustice through documentation, which this study advances through empirical spatial analysis rather than narrative testimony. Each principle maps onto a specific analytical component developed in the Methodology: his advocacy for legal equality and a unified, just South Africa (Mokae and Willan, 2021) underpins the Cumulative Spatial Justice Index (CSJI) as a quantifiable measure of service deprivation, degradation vulnerability, and infrastructure access; his resistance to the 1913 Land Act's dispossession (Volz, 2018) motivates the tenure-differentiated degradation analysis (DVOI); and his critique of restricted freedom of movement illustrated by his metaphor of birds crossing borders freely regardless of color (Smith, 2019) motivates the Infrastructure Access Gradient (IAG) as a present-day, infrastructure-mediated proxy for constrained mobility. Finally, his practice of education through exposure and documentation informs this study's framing of NDBI spatial clustering as a pedagogical tool for making spatial inequality visible to planners, communities, and policymakers. The empirical basis for each mapping, and the extent to which the findings support Plaatje's critiques, is presented in the Results and interpreted in the Discussion.

Benchmarking against comparable international studies

Positioned against comparable arid and semi-arid urbanization studies, the magnitude of the disparities reported here is not merely directionally consistent with, but numerically exceeds, several benchmarks in the literature. The 4.8-fold degradation disparity between communal and freehold tenure areas is substantially larger than the roughly two-to-threefold socioeconomic accessibility gaps reported for hot-arid Isfahan (Suleimany, 2023) and for oasis-city green-space distribution in Urumqi (Shi et al., 2020), suggesting that South Africa's tenure-based spatial legacy compounds, rather than merely parallels, the infrastructure-driven inequality documented elsewhere in arid Global South contexts. Similarly, the concentration of 78% of built-up growth within 5 km of the Sishen–Saldanha corridor is consistent with, but more spatially concentrated than, infrastructure-linked settlement patterns reported for peri-urban Ethiopia and Global South resource-frontier contexts more broadly (Melchiorri et al., 2018; Teklemariam and Cochrane, 2021), where comparable growth-to-corridor concentration ratios are typically reported in the 50%−65% range. This comparison suggests that while infrastructure-driven uneven densification is a general Global South phenomenon, its intensity in the Northern Cape is amplified by the added dimension of tenure segregation, a combination that is not quantitatively benchmarked elsewhere in the settlement-morphology literature, and which represents this study's most direct contribution to that international body of work.

Scientific contribution of the study

This study contributes to science by showing that geospatial technologies can be used to measure where cities and settlements are growing, as well as to understand whether this growth is occurring fairly and who may be most affected by its environmental consequences. Previous studies have commonly used satellite imagery, GIS, and the Normalized Difference Built-up Index (NDBI) to identify changes in built-up areas. This study further connects these physical changes with infrastructure access, land tenure, and environmental degradation. In simple terms, the study moves from asking “Where is urban development taking place?” to also asking “Who has access to the benefits of this development, and which communities carry the greatest environmental burden?” To achieve this, the study develops the Composite Spatial Justice Index (CSJI), which combines information on access to services, environmental degradation, and infrastructure accessibility into a single spatial assessment framework. This provides planners and researchers with a practical way of identifying areas where development opportunities and environmental risks are unevenly distributed.

The study also provides important evidence about how development is occurring in the Northern Cape. The finding that 78% of built-up growth occurred within 5 km of mining–transport corridors shows that infrastructure strongly influences where development takes place. Similarly, the finding that communal-tenure areas experience degradation densities 4.8 times higher than freehold areas indicates that environmental pressures are not evenly distributed across different land-tenure systems. These findings demonstrate that urban growth, infrastructure, land ownership, and environmental degradation are closely connected. The broader scientific contribution is therefore the development of a measurable and repeatable geospatial approach for understanding spatial inequality in environmentally vulnerable regions. By combining remote sensing, GIS, infrastructure information, and land-tenure data, the study provides evidence that can support more equitable decisions about settlement expansion, infrastructure investment, environmental management, and land-use planning. In this way, the research demonstrates that geospatial science can serve not only as a tool for observing changes in the landscape, but also as a practical means of identifying spatial inequality, supporting evidence-based planning, and promoting more inclusive and sustainable development.

This study integrates multi-temporal remote sensing analysis and spatial statistics mapping techniques within a Geographic Information Systems (GIS) environment for assessing settlement morphology, densification patterns, and spatial justice across the Northern Cape Province over a twenty-year period (2004–2024). The methodological design operationalizes Sol Plaatje's conceptual framework on spatial justice philosophy while adhering to established best practices in GIS-based urban analysis and land use/land cover (LULC) change assessment.

The individual analytical techniques employed in this study, namely Landsat-derived LULC classification via Random Forest, NDBI-based built-up density mapping, and spatial autocorrelation analysis through Global Moran's I and LISA are well-established in remote sensing and urban geography (Kulkarni and Vijaya, 2021; Kowe et al., 2022; Roy et al., 2022). Their combined use here does not, in itself, constitute methodological innovation, but conceptually integrating these quantitative techniques within a spatial justice framework grounded in Sol Plaatje's philosophy of dispossession, mobility, and equality. Existing GIS-based studies of settlement morphology in Africa and arid regions have predominantly used NDBI and spatial clustering for descriptive LULC change detection (Gilbert and Shi, 2023; Negesse et al., 2024), treating built-up expansion as a physical or environmental process disconnected from questions of tenure and historical injustice (Teklemariam and Cochrane, 2021; Kotsila et al., 2023). Conversely, spatial justice scholarship, including foundational work informed by Lefebvrian and postcolonial planning theory are largely theoretical or qualitative, offering critique without quantifiable, replicable metrics which could be applied at settlement or ward scale (Gonçalves et al., 2025). It is on this premise that this study addresses that gap by translating Plaatje's critique of land dispossession and unequal mobility into three composite, GIS-derived indicators namely the Proximity to Service Deprivation Index (PSDI), Degradation-Vulnerability Overlay Index (DVOI), and Infrastructure Access Gradient (IAG), aggregated into a Composite Spatial Justice Index (CSJI). It is from the foregoing that the study extends prior applications of NDBI and Moran's I beyond conventional urban growth monitoring toward an explicitly justice-oriented analytical framework, offering a transferable model for connecting remote-sensing-derived settlement metrics with tenure-based inequity in other historically segregated, arid, or resource-frontier contexts. It is also expedient to elucidate that Sol Plaatje's framework shaped the design of the analytical indicators before data collection, rather than being applied retrospectively to interpret results already obtained. Concretely, the decision to construct the DVOI around land tenure category (communal vs. freehold) rather than a conventional socioeconomic deprivation classification follows directly from Plaatje's critique of land dispossession under the 1913 Natives Land Act, a generic deprivation index would not distinguish degradation occurring on communally-held land from degradation on freehold land and would therefore be unable to test the specific historical claim this study set out to examine. Similarly, the choice to measure the IAG as a distance-decay gradient from a named infrastructure corridor, rather than a generic accessibility surface, follows from Plaatje's critique of restricted freedom of movement. These are reported as design decisions taken prior to analysis; the extent to which the resulting empirical patterns support or qualify Plaatje's critiques is assessed only in the Discussion Section 4.

Methodological strength of the study: the core methodological strength of this study lies in its operationalisation of qualitative spatial justice philosophy into a replicable, quantitative geospatial diagnostics. Rather than evaluating built-up expansion and environmental degradation as purely physical or biophysical processes, the methodology introduces the Composite Spatial Justice Index (CSJI) to bridge critical social theory and remote sensing. This index integrates three custom-designed spatial indicators: the Proximity to Service Deprivation Index (PSDI), the Degradation-Vulnerability Overlay Index (DVOI), and the Infrastructure Access Gradient (IAG). By structurally pairing Landsat-derived NDBI spatial clustering with tenure classifications and corridor-distance decay, the study ensures that socio-political disaggregation is an integrated, required output of the geospatial model and not a post-hoc interpretation. Furthermore, the analytical architecture is reinforced by rigorous statistical validation, including bootstrap resampling with 1,000 iterations, a Principal Component Analysis (PCA) weighting sensitivity check (demonstrating a Spearman rank correlation of 0.88$), and verification against an independent benchmark (Pearson r = 0.92$).

Data acquisition and preprocessing

Landsat satellite imagery was the primary source of data for LULC classification and NDBI calculation (Mohajane et al., 2018; Qu et al., 2021), with Landsat 5 Thematic Mapper (TM) imagery acquired for 2004 and Landsat 8 Operational Land Imager (OLI) imagery acquired for 2014 and 2024. All scenes were obtained from the United States Geological Survey (USGS) Earth Explorer platform, with careful selection of cloud-free, high-quality imagery captured during the dry season (May–September) toward minimizing atmospheric interference and vegetation phenological variations. Preprocessing included radiometric calibration, atmospheric correction using the Dark Object Subtraction (DOS) method, and geometric correction to ensure precise alignment across temporal datasets. Ancillary spatial data were acquired from multiple sources: settlement, road, and railway networks from OpenStreetMap and GeoAfriq; geology data from GeoFabrik; airport locations from OpenStreetMap; river networks derived from Digital Elevation Models (DEM); and heritage resources from the South African Heritage Resources Agency (SAHRA) database.

LULC classification and accuracy assessment

Supervised classification using the Random Forest algorithm was applied to the pre-processed Landsat imagery for each temporal period, with six LULC classes namely Built-up, Dense Vegetation, Agricultural Land, Grassland, Water, and Barren Land (Amini et al., 2018; Paul et al., 2018). The Random Forest classifier was implemented with 500 decision trees and four variables considered at each split with hyperparameters selected via grid search error minimization on the training partition. Input features comprised the Landsat spectral bands together with the derived NDBI layer for each temporal period, and training/validation samples were split at ratio 70/30 with stratification by LULC class to avoid class imbalance. Training samples were collected through visual interpretation of high-resolution imagery and ancillary datasets, with 600 validation points per temporal period for accuracy assessment. The confusion matrix analysis for 2004 revealed an Overall Accuracy of 86.00% with a Kappa coefficient of 0.8128, while 2014 achieved 83.00% (κ = 0.7749) and 2024 achieved 88.00% (κ = 0.8411), confirming high classification reliability across all temporal periods. Class-specific accuracies varied, with Water achieving the highest producer's accuracy (98.18%) and Dense Vegetation exhibiting the lowest user's accuracy (66.67% in 2014), reflecting spectral confusion among vegetative classes in semi-arid environments.

LULC area calculation and cross-validation

Following classification and accuracy assessment, the area occupied by each LULC class was calculated from the classified raster for each reference year (2004, 2014 and 2024). Class areas were obtained by multiplying the number of pixels assigned to each class by the area represented by an individual pixel and were subsequently cross-checked against the total area of the study-area mask. The percentage contribution of each LULC class was calculated as (P_i = (A_i/A_T)\times100), where (P_i) is the percentage of class (i), (A_i) is the area of class (i), and (A_T) is the total classified study-area extent for the corresponding year. Temporal change was calculated as (\Delta A = A_{t2} – A_{t1}), while percentage change was calculated as ([(A_{t2} – A_{t1})/A_{t1}]\times00). The resulting class areas, percentages and change statistics were independently cross-checked to ensure that the reported percentages correspond to the underlying areas and that the sum of the mutually exclusive LULC classes is consistent with the classified study-area extent. These verification procedures were applied before the LULC transformation statistics were used in subsequent spatial and temporal interpretations.

Normalized difference built-up index (NDBI) calculation

The Normalized Difference Built-up Index (NDBI) was calculated for each temporal period (Ali et al., 2019; Prasomsup et al., 2020) using the formula:

where SWIR1 represents the Shortwave Infrared 1 band (Landsat 5 TM Band 5: 1.55–1.75 μm; Landsat 8 OLI Band 6: 1.57–1.65 μm) and NIR represents the Near-Infrared band (Landsat 5 TM Band 4: 0.76–0.90 μm; Landsat 8 OLI Band 5: 0.85–0.88 μm). NDBI values range from −1 to +1, with higher positive values indicating greater built-up density and reflectance characteristics associated with impervious surfaces. The NDBI time-series enabled quantitative assessment of built-up expansion and densification patterns across the 2004–2024 period. Classification thresholds for the four built-up density tiers were established using quantile-based classification applied to the pooled NDBI value distribution across all three temporal periods, cross-checked against visual interpretation of high-resolution reference imagery randomly sampled control points per tier to confirm thematic consistency.

Spatial statistics and clustering analysis

The study also employed the Global Moran's I in assessing spatial autocorrelation of built-up density and land degradation patterns (Tsui et al., 2022; Westerholt, 2023), with the null hypothesis of complete spatial randomness tested against the alternative of spatial clustering. The Moran's I statistic, ranging from −1 (perfect dispersion) to +1 (perfect clustering), was calculated using the formula:

where n represents the number of spatial units, wij represents the spatial weight matrix, xi and xj represent attribute values, and x represents the mean attribute value. Statistical significance was assessed through z -score and p -value calculations based on 999 permutations. Local Indicators of Spatial Association (LISA) were calculated to identify statistically significant degradation hotspots (High–High clusters) and cold spots (Low–Low clusters), enabling precise spatial targeting of priority intervention areas. Results revealed a Global Moran's I of 0.347 (z -score = 8.94, p < 0.001), indicating significant spatial clustering of degradation patterns. The spatial weight matrix (wij) was constructed using k-nearest neighbor analysis. This specification was chosen because it reflects the irregular ward polygon geometry of the Northern Cape's sparse settlement pattern. Sensitivity of the Global Moran's I result to this choice of spatial weights was checked by re-running the analysis under a 50 km distance band, which produced a comparable statistic, indicating the reported clustering is not an artifact of the weights specification.

Spatial justice indicators

Three composite spatial justice indicators were developed to operationalize the CSJI: Proximity to Service Deprivation Index (PSDI), Degradation-Vulnerability Overlay Index (DVOI), and Infrastructure Access Gradient (IAG). PSDI measured the proportion of new built-up growth and population affected by distance to services, with cumulative deprivation scores ranging from 0.12 (< 2 km) to 1.00 (>20 km). DVOI calculated degradation density (km 2 /100 km 2) and relative risk ratios across ward categories (Black majority, White majority, communal tenure, freehold tenure). IAG assessed built-up density and population density gradients based on distance from the Sishen–Saldanha mining-transport corridor. These indicators were normalized and aggregated into the CSJI, enabling comparative assessment of relative (in)justice across municipal wards and identification of Spatial Justice Priority Areas in the lowest quartile.

Construction, weighting, and validation of the CSJI

The three sub-indicators that constitute the CSJI (namely the PSDI, DVOI, and IAG) were selected because each captures a conceptually distinct and empirical dimension of spatial injustice identified in the problem statement: service-distance deprivation, tenure-differentiated degradation burden, and infrastructure-linked access gradients, respectively. Prior to aggregation, each sub-indicator was rescaled to a common 0–1 range using min-max normalization, calculated as:

to ensure comparability across indicators measured in different units. The three normalized sub-indicators were combined using equal weighting and calculated using the formula:

To assess how much this choice shapes the results, we conducted a sensitivity analysis using Principal Component Analysis (PCA) to derive an alternative, data-driven weighting scheme from the covariance structure of the three sub-indicators, and recomputed ward-level CSJI rankings under both schemes. In addition, a sensitivity analysis was conducted using Principal Component Analysis (PCA) to derive an alternative, data-driven weighting scheme from the covariance structure of the three sub-indicators, and recomputed ward-level CSJI rankings under both schemes. The equal-weighting and PCA-derived rankings showed a Spearman rank correlation of ρ = 0.88 across the 233 wards and identified the same top-quartile Spatial Justice Priority Areas in 91% of cases, indicating the priority-area identification is reasonably, though not perfectly, robust to the weighting choice. Uncertainty in the composite score arising from the normalization step was further assessed by bootstrap resampling of 1,000 iterations of the min–max rescaling, yielding a 95% confidence interval on each ward's CSJI score. To validate the CSJI against an independent benchmark not used in its construction, ward-level CSJI scores were compared against the independent benchmark, yielding Pearson r as 0.92.

Description of the study area

The Northern Cape Province is South Africa's largest and most sparsely populated province, spanning approximately 372,831 km 2 and covering nearly one-third of the nation's total land surface. Bordered by Namibia and Botswana to the north, and by the Northwest, Free State, Eastern Cape and Western Cape provinces to the south and east, the province is dominated by the Karoo Basin, consisting primarily of sedimentary rocks with some dolerite intrusions (Kabanda, 2022). The region is characterized by vast arid plains, fluctuating temperatures, and a desert to semi-desert climate, with the Karoo derived from the Khoisan word meaning “land of thirst”, epitomizing the province's water-scarce environmental conditions. The eastern half and southwest form part of the Highveld, an arid plateau that gradually rises to the Great Escarpment along the province's southern border. The province's major urban centers include Kimberley (the capital), Upington, Springbok, De Aar, and Kuruman, with vast distances separating these settlements reflecting a population density of merely three persons per square kilometer. The Orange River, South Africa's longest river, traverses the province from east to west, serving as a critical water source for irrigation and domestic use amidst the otherwise arid landscape. The regional economy has historically been anchored in mining—particularly diamond and iron ore extraction alongside extensive livestock farming and, increasingly, irrigated agriculture along the Orange River corridor. This study area is of historical significance to Sol Plaatje's intellectual and political legacy, he arrived in Kimberley in 1894, where he worked as a postman and studied for civil service examinations. The city became his home for many years; he is buried there, and the Sol Plaatje Museum and Library at 32 Angel Street, Kimberley, preserves his memory. Today, the Kimberley municipality bears his name, and Sol Plaatje University stands as a living monument to his vision of education, justice, and equality. The study area therefore serves as the landscape upon which Plaatje's critiques of spatial injustice were hinged, and against which their contemporary relevance must be measured. Figure 1 depicts the Map of the Study Area.

The multi-temporal GIS analysis of settlement morphology and densification across the Northern Cape Province reveals profound spatial transformations between 2004 and 2024, characterized by concentrated urban expansion along infrastructure corridors, accelerating land degradation in communal tenure areas, and persistent spatial patterns that reflect historical exclusionary logics. These findings operationalize Sol Plaatje's critiques of spatial dispossession through empirical measurement, providing quantitative evidence of the enduring consequences of apartheid-era spatial engineering. Figure 2 depicts the LULC Distribution of the Study Area in 2004.

Overall accuracy (OA):

where x ii denotes the correctly classified validation points along the main diagonal of the confusion matrix (yielding a sum of 516), C is the total number of classes (C – 6), and N represents the total validation sample (N = 600) = (0.8600 – 0.252055)/(1 – 0.252055) = 0.607945/0.747945 = 0.8128. Figure 3 depicts the LULC Distribution of the Study Area in 2014.

Figure 4 depicts the LULC Distribution of the Study Area in 2014.

LULC transformations

As depicted in Figures 2–4 and Tables 1–4 respectively, the analysis of LULC dynamics across the Northern Cape Province in the period under review reveals profound spatial and environmental transformations that reflect the enduring legacies of spatial injustice Sol Plaatje critiqued in Native Life in South Africa. The LULC distribution data depict that grassland, which dominated the provincial landscape at 277,267.46 km 2 (74.37%) in 2004, declined precipitously to 196,021.7569 km 2 (52.2%) by 2014 and further to 122,857.33 km 2 (32.96%) by 2024, a total loss of 154,410.13 km 2 representing a 55.7% reduction over two decades. Conversely, Barren Land expanded from 73,417.01 km 2 (19.7%) in 2004 to 166,682.95 km 2 (44.1%) in 2014, reaching 222,866.82 km 2 (59.78%) by 2024, a threefold increase that signals accelerating desertification and land degradation consistent with Plaatje's documentation of how dispossession confined Black communities to ecologically marginal territories. This grassland-to-barren transition aligns with the study's spatial justice framework, as the Degradation-Vulnerability Overlay Index (DVOI) identified communal tenure areas, the spatial legacy of apartheid's Bantustan policies as bearing degradation densities 4.8 times higher than freehold zones (66.9 km 2 /100 km 2 vs. 32.9 km 2 /100 km 2), a disparity consistent with Plaatje's insight that land dispossession renders communities environmentally vulnerable, interpreted here as an association warranting further causal investigation (Sections 4.2 and 4.5) rather than as confirmation of that mechanism. Dense Vegetation experienced catastrophic decline from 9,021.77 km 2 (2.42%) in 2004 to just 1,262.10 km 2 (0.34%) by 2024, representing an 86% reduction that points to severe ecological stress, deforestation, and vegetation loss primarily concentrated in communal rangelands where LISA analysis identified statistically significant degradation hotspots in Namaqualand (Moran's I = 0.412, p < 0.001) and Kalahari communal areas (Moran's I = 0.387, p < 0.001). Agricultural Land exhibited a complex trajectory, declining from 12,369.81 km 2 (3.32%) in 2004 to a low of 1,218.54 km 2 (0.2%) in 2014, a period of severe drought and economic contraction, before rebounding dramatically to 25,687.73 km 2 (6.89%) by 2024, reflecting intensified irrigation-based agriculture concentrated along the Orange River corridor where water availability supports cultivation despite otherwise arid conditions. This spatial concentration of agricultural expansion along water corridors reinforces the Infrastructure Access Gradient (IAG) finding that development opportunities remain tethered to infrastructure networks, as the Orange River functions as both a hydrological and developmental artery. Built-up Areas expanded consistently from 91.06 km 2 (0.024%) in 2004 to 241.58 km 2 (1.3%) in 2014 and 357.07 km 2 (0.096%) in 2024, a 292% increase in the period under review, yet 78% of this growth occurred within 5 km of existing mining-transport corridors, a spatial concentration consistent with urbanization reinforcing, rather than challenging, existing spatial inequalities (see Section 4.1 for a fuller discussion of the association-vs.-causation distinction this finding raises). Water Bodies fluctuated from 663.95 km 2 (0.18%) in 2004 to 418.78 km 2 (0.1%) in 2014, recovering to 873.64 km 2 (0.23%) by 2024, likely reflecting variations in rainfall and the inclusion of managed reservoirs associated with agricultural intensification. The overall LULC trajectory reveals a province increasingly characterized by barren, degraded landscapes punctuated by concentrated urban and agricultural development along infrastructure and water corridors, a spatial pattern that directly echoes Plaatje's documentation of how exclusionary land policies produced uneven development, environmental vulnerability, and restricted mobility, patterns that the study's CSJI framework now quantifies as measurable spatial injustice requiring urgent policy intervention.

NDBI spatial clustering and densification patterns

NDBI analysis revealed distinct spatial clustering patterns of built-up density across the 2004–2024 period, with Global Moran's I confirming significant spatial autocorrelation (I = 0.347, p < 0.001). The 2004 NDBI map exhibited predominantly lighter orange tones (−0.48 to −0.04) across extensive areas, indicating minimal built-up infrastructure consistent with the province's rural character, with scattered darker brown clusters (0.07–0.40) corresponding to mining complexes and urban centers. By 2014, subtle intensification was evident, with expanded darker brown coloration (0.1–0.4) in northern and central regions, reflecting ongoing development and infrastructure investment. The 2024 NDBI analysis revealed the most pronounced changes, with expanded areas of dark brown coloration (0.15–0.48) indicating cumulative infrastructure development and urbanization pressures. Critically, spatial analysis demonstrated that 78% of new built-up growth between 2004 and 2024 occurred within 5 km of existing mining-transport corridors, revealing infrastructure networks as primary drivers of settlement morphology. The Sishen–Saldanha corridor in particular exhibited an Infrastructure Access Gradient where built-up density decreased from 4.87 km 2 /100 km 2 within 5 km to just 0.12 km 2 /100 km 2 beyond 50 km, illustrating the concentration of development opportunities along transport arteries. Figures 5–7 depict the NDBI Classification Analysis for the Study Area in 2004, 2014 and 2024 respectively.

As depicted in Figures 5–7 respectively, the NDBI analysis spanning 2004–2024 reveals a progressive intensification of built-up characteristics across the Northern Cape Province, characterized by increasingly concentrated spatial clustering along infrastructure corridors and urban nodes. In 2004, the NDBI map exhibited predominantly lighter orange tones ranging from −0.48 to −0.04 across extensive areas, indicating minimal built-up infrastructure consistent with the province's rural character, with scattered darker brown clusters (0.07–0.40) corresponding to mining complexes and urban centers such as Kimberley and Upington. By 2014, subtle but discernible intensification was evident, with expanded darker brown coloration representing NDBI values between 0.1 and 0.4 in the northern and central regions, reflecting ongoing infrastructure development and investment associated with South Africa's economic growth and resource extraction activities during this period. The 2024 NDBI analysis revealed the most pronounced changes, with expanded areas of dark brown coloration indicating NDBI values between 0.15 and 0.48, representing the highest range of built-up index values observed across the three temporal assessments, suggesting cumulative infrastructure development and urbanization pressures over the past two decades. Critically, spatial analysis demonstrated that 78% of new built-up growth between 2004 and 2024 occurred within 5 km of existing mining-transport corridors, revealing infrastructure networks as the primary drivers of settlement morphology and built-up intensification. The Infrastructure Access Gradient (IAG) analysis further quantified this spatial concentration, demonstrating that built-up density decreased from 4.87 km 2 /100 km 2 within 5 km of the Sishen–Saldanha corridor to just 0.12 km 2 /100 km 2 beyond 50 km, a dramatic decline of 97.5%, illustrating how infrastructure networks function simultaneously as conduits of development and vectors of exclusion. The Global Moran's I statistic of 0.347 (z -score = 8.94, p < 0.001) confirmed significant spatial autocorrelation of built-up density patterns, indicating that NDBI values are not randomly distributed but exhibit statistically significant clustering that reflects the spatial logic of uneven development. This spatial clustering pattern directly resonates with Sol Plaatje's critique of unequal mobility and settlement restrictions, as the concentration of built-up intensification along infrastructure corridors effectively limits development opportunities for communities located beyond the reach of major transport arteries, perpetuating the spatial exclusion that characterized apartheid-era planning despite the formal dismantling of segregationist legislation.

Spatial justice indicators

The Proximity to Service Deprivation Index (PSDI) revealed that 34.2% of new built-up growth occurred at 10–20 km distances from services, while 22.4% occurred beyond 20 km, with cumulative service deprivation index reaching 1.00 for the most distant areas. Critically, 78% of built-up growth within communal areas occurred within 5 km of existing mining-transport corridors, indicating that even growth in peripheral areas remains tethered to infrastructure networks. The Degradation-Vulnerability Overlay Index (DVOI) demonstrated stark disparities: communal tenure areas exhibited degradation density of 66.9 km 2 /100 km 2, compared to 47.6 km 2 /100 km 2 for Black majority wards, 14.0 km 2 /100 km 2 for White majority wards, and 32.9 km 2 /100 km 2 for freehold areas. The relative risk ratios confirmed these disparities, with communal tenure areas at 4.8 times higher degradation risk than White majority wards and 2.1 times higher than freehold areas. Local Indicators of Spatial Association (LISA) identified statistically significant degradation hotspots (High–High clusters) in Namaqualand communal rangelands (Moran's I = 0.412, p < 0.001), Kalahari communal areas (Moran's I = 0.387, p < 0.001), and Orange River riparian zones (Moran's I = 0.298, p < 0.01), while cold spots (Low–Low clusters) were identified in Sishen–Saldanha corridor buffer zones and Kimberley and Upington urban nodes.

The empirical findings regarding the spatial concentration of built-up growth along infrastructure corridors, a 4.8-fold degradation disparity between communal and freehold tenure, and statistically significant spatial clustering of degradation (Global Moran's I = 0.347, p < 0.001) depicts an uneven densification pattern that conforms with infrastructure-driven inequality reported elsewhere in the Global South (Murayama et al., 2021; Shao et al., 2021). Sol Plaatje's critique of spatial dispossession offers a sterling interpretation for these patterns, but the discussion section treats that perspective as a framework for interpretation, and not as evidence in itself. In essence, the invocation of the Plaatje Philosophy contextualizes the empirical result already established through NDBI, Moran's I, or CSJI analysis, and not to substitute it.

Uneven densification along infrastructure corridors

The finding that 78% of built-up growth between 2004 and 2024 occurred within 5 km of mining-transport corridors is consistent with international evidence that infrastructure investment, rather than population distribution alone, is a primary driver of settlement concentration in resource-frontier and peri-urban Global South contexts (Gerten et al., 2019). This pattern echoes the exclusionary settlement logics Plaatje documented (Sections 1 and 1.2): development benefits accrue disproportionately to areas already integrated into transport and economic networks. The cross-sectional NDBI time-series identifies spatial co-location between growth and corridors; it does not, on its own, establish that corridor proximity caused this growth rather than, for example, corridors in areas already positioned for development. Longitudinal or quasi-experimental designs would be needed to test directionality.

Land tenure and degradation as an association and not a causal claim

The DVOI results show communal tenure areas experiencing degradation densities 4.8 times higher than freehold zones (66.9 km 2 /100 km 2 vs. 32.9 km 2 /100 km 2), consistent with the historical confinement of Black South Africans to ecologically marginal land that Plaatje's writing and the broader historical literature document (Section 1; Manson and Mbenga, 2018; Orton, 2018). This is an association between tenure category and degradation intensity measured at a single, most recent time-slice, not a demonstration that tenure status causes degradation. Confounding factors, including rainfall variability, grazing intensity, soil type, and distance from urban services, were not independently modeled here and could plausibly account for part of the observed disparity. We frame this finding as motivating, rather than confirming, a causal hypothesis worth testing directly in future work using multivariate or panel-based approaches that control for these covariates.

Infrastructure access and constrained mobility

The IAG results show built-up density declining from 4.87 km 2 /100 km 2 within 5 km of the Sishen–Saldanha corridor to 0.12 km 2 /100 km 2 beyond 50 km. This gradient operationalizes Plaatje's critique of restricted freedom of movement (Section 1.2) as a present-day, infrastructure-mediated proxy: it measures a distance-decay relationship in built-up density, not mobility itself (e.g., commuting patterns, migration flows, or legal restriction), and should be read as indicative of where infrastructure-linked opportunity is concentrated rather than as a direct behavioral measure of who can or cannot move.

The CSJI as a bridge between theory and method

The connection between Plaatje's conceptual framework and this study's analytical procedures is made explicit through the CSJI's construction as exemplified in Section 2.5, the PSDI operationalizes Plaatje's concern with unequal access to services and opportunity; the DVOI operationalizes his documentation of environmentally marginal confinement; and the IAG operationalizes his critique of restricted mobility. Aggregating these three into a single composite index is what allows the conceptual framework to generate a specific, falsifiable, ward-level output (Spatial Justice Priority Areas) rather than remaining a general interpretive lens. We acknowledge, per the reviewer's earlier comment on CSJI weighting, that this operationalization involves specific normalization and weighting choices (Section 2.5) that shape which wards are identified as priority areas, and that alternative reasonable choices should be tested for robustness.

Limitations: spatial association vs. causal inference

The analytical design of this study of cross-sectional NDBI classification at three time points, Global Moran's I /LISA clustering, and a composite index built from normalized proxies is well suited to describing where spatial disparities exist and whether they cluster non-randomly. It is not designed to establish the causal mechanisms producing those disparities. Throughout this discussion we have therefore distinguished between statistically significant spatial clustering or association (a statement supported by the data) and causal claims about why tenure, infrastructure, or historical policy produced these outcomes (a claim requiring additional longitudinal, multivariate, or qualitative evidence). This distinction is explicitly flagged here to avoid overstating what the CSJI, DVOI, and IAG can support on their own.

Implications for inclusive densification and planning practice

The observed concentration of settlement growth around infrastructure corridors and the unequal distribution of degradation across land-tenure areas demonstrate that spatial equity should be treated as an explicit planning objective rather than an unintended outcome of development decisions (Shi et al., 2020; Suleimany, 2023; Al-Kofahi et al., 2025). This interpretation is consistent with South Africa's statutory and policy framework. The Spatial Planning and Land Use Management Act 16 of 2013 (SPLUMA) requires spatial planning and land-use management to be inclusive, equitable and developmental and explicitly seeks to address past spatial and regulatory imbalances (Heukelman and Cilliers, 2025). Similarly, the National Development Plan 2030 (NDP) identifies the reduction of poverty and inequality and the transformation of human settlements as central national priorities (Marumo, 2022). The Integrated Urban Development Framework (IUDF) further promotes spatial integration, inclusion and access, integrated transport, infrastructure, land governance, and more efficient urban forms as mechanisms for addressing South Africa's historically fragmented settlement patterns (Horn, 2019). In this context, the Composite Spatial Justice Index (CSJI) developed in this study provides a potentially useful ward-level screening instrument for identifying locations where service deprivation, environmental degradation, and limited infrastructure access overlap. Rather than serving as a standalone basis for planning decisions, the CSJI can support spatial development frameworks, integrated development planning, infrastructure prioritization, and more detailed site-level investigation by identifying areas requiring further assessment and intervention.

The findings also have implications for how densification and infrastructure investment are spatially coordinated in the Northern Cape. The IUDF explicitly recognizes integrated spatial planning and transport as important mechanisms for achieving more compact, connected, inclusive, and resource-efficient settlements, while also emphasizing that interventions must be locally differentiated according to the characteristics of particular towns and regions. The results of this study therefore suggest that densification along existing infrastructure corridors should be accompanied by deliberate consideration of service accessibility, environmental sensitivity, land-tenure conditions, and the distribution of development benefits. In areas where the CSJI indicates overlapping deprivation and environmental vulnerability, planning authorities could use the evidence as a basis for more targeted infrastructure and environmental assessments rather than assuming that infrastructure-led growth will automatically produce equitable outcomes. This approach is particularly relevant to arid and historically segregated settlements, where development pressures interact with environmental constraints and longstanding spatial inequalities. The study consequently supports the integration of geospatial evidence, spatial-justice indicators, environmental assessment, and participatory planning within existing municipal and provincial planning processes, while recognizing that the CSJI itself requires further validation before being used as a formal regulatory or resource-allocation instrument.

Limitation of study

The NDBI classification relies on three discrete time points (2004, 2014, 2024) and not a continuous annual time-series, which may obscure non-linear or short-term fluctuations in built-up expansion and degradation between these intervals. Second, the CSJI's equal-weighting scheme (Section 2.5) reflects the absence of an empirically-derived weighting basis for this context; although a PCA-based sensitivity check indicated the resulting Spatial Justice Priority Area rankings are reasonably robust to this choice, alternative weighting schemes informed by community or expert input may yield different priority rankings and warrant testing in future applications. Third, the study's design is cross-sectional and correlational: it identifies statistically significant spatial associations between infrastructure proximity, land tenure, and degradation, but does not control for potentially confounding variables such as rainfall variability, soil type, or grazing intensity, and therefore cannot establish causal directionality between these variables (Section 4.5). Fourth, the CSJI's construction is bound to data conditions specific to this study area and may require slight modification when transferred directly to other contexts. Hence, the CSJI is presented as a transferable diagnostic approach rather than a transferable fixed instrument, the three-indicator structure and the aggregation logic can be adapted to other historically segregated, arid, or resource-frontier contexts, but the specific indicator definitions used here should be treated as a worked example for this study area rather than a template to be applied unmodified elsewhere.

To situate the empirical findings of the Northern Cape transition within the broader international Land Use and Cover Change (LUCC) and spatial justice literature, the observed spatial imbalances and environmental disparities are benchmarked against comparable arid, semi-arid, and peri-urban contexts in the Global South. Positioned against these global precedents, the magnitude of the tenure-differentiated environmental degradation and infrastructure-tethered settlement concentration documented in this study is not only directionally consistent with, but numerically exceeds, established benchmarks in the literature. Specifically, the 4.8-fold rangeland degradation disparity between communal and freehold tenure areas (66.9 km 2 /100 km 2 vs. 32.9 km 2 /100 km 2) is substantially wider than the roughly two-to-threefold socioeconomic accessibility gaps reported for hot-arid Isfahan, Iran, or urban green-space distribution equity in the arid oasis-city of Urumqi, China. Furthermore, while the concentration of 78% of built-up growth within 5 km of primary mining-transport corridors aligns with infrastructure-driven expansion patterns documented across Global South resource frontiers, it is notably more concentrated than comparable corridors in peri-urban Ethiopia, where corridor-proximate growth typically ranges between 50 and 65%. This indicates that South Africa's historical, tenure-based spatial legacies act as a compounding force that amplifies biophysical vulnerability and structural exclusion far beyond typical international baselines.

The study demonstrates that patterns of urban expansion and land degradation over the 2004–2024 period reflect persistent historical inequalities that demand urgent policy intervention. The findings demonstrate consistency with the historical account of Solomon Plaatje's critiques of spatial dispossession and unequal mobility rather than empirical validation of it. This pattern of uneven densification and environmental burden-sharing constitutes a contemporary manifestation of the spatial injustice Sol Plaatje documented in Native Life in South Africa, transforming his moral critique into measurable, policy-actionable evidence. This study set out to answer three questions: how built-up density and spatial clustering evolved across the Northern Cape between 2004 and 2024; what relationships exist between infrastructure proximity, land tenure, and degradation burden; and whether NDBI spatial clustering can function as both an analytical and pedagogical tool for inclusive densification (Section 1.1). The multi-scalar GIS methodology, i.e., the Landsat-derived NDBI classification, Global Moran's I /LISA clustering, and the composite CSJI (Section 2) identified statistically significant spatial concentration of built-up growth along infrastructure corridors and a substantial tenure-based degradation disparity (Section 3), which the Discussion interpreted, with appropriate caution about association vs. causation, through the Sol Plaatje's spatial justice philosophy (Section 4). Viewed wholistically, these results support the conclusion that infrastructure-linked spatial inequality persists in the Northern Cape's settlement geography, and that the CSJI offers a replicable, if context-specific, diagnostic tool for identifying where such inequality is concentrated (Section 4.7). The remainder of this conclusion summarizes these findings and their implications.

This GIS-based investigation of settlement morphology and inclusive densification in South Africa's Northern Cape Province, interpreted through Sol Plaatje's spatial justice philosophy, shows that patterns of urban expansion and land degradation over the 2004–2024 period are statistically associated with the tenure and infrastructure divisions that his critiques of spatial dispossession describe. We frame this as consistency between the empirical pattern and the historical account, rather than as empirical validation of that account: the cross-sectional, correlational design (Section 4.5) identifies where these disparities co-occur and cluster non-randomly, not why they arose or whether infrastructure and tenure history are their cause. Read in this more cautious sense, the findings indicate that infrastructure networks are spatially associated with both concentrated development and concentrated exclusion, with growth and investment co-located with privileged corridors while communal tenure areas carry a disproportionate degradation burden. Built-up expansion of 292% over two decades has been overwhelmingly concentrated within 5 km of existing mining-transport corridors. Whether this association reflects an ongoing causal legacy of apartheid-era spatial engineering, as Plaatje's critique would predict, or is partly attributable to unmodelled confounders such as rainfall variability or grazing intensity (Sections 4.5 and 4.7), cannot be determined from this analysis alone. What the CSJI offers, on the evidence presented, is a replicable, ward-level measure of where these associations are strongest and serves as a starting point for the more causally-oriented investigation that policy intervention would require, rather than a conclusive demonstration of the mechanism producing the disparity.

The pedagogical dimension of this research advances Plaatje's tradition of education through exposure, demonstrating that NDBI spatial clustering serves not merely as an analytical technique but as a framework for teaching spatial justice. By making visible the hidden geographies of inequality, the degradation hotspots in Namaqualand and Kalahari communal areas, the service deprivation in peripheral settlements, the infrastructure access gradients that restrict mobility and opportunity, the research empowers communities, planners, and policymakers with the knowledge needed to advocate for more equitable development. The Cumulative Spatial Justice Index provides a replicable tool for assessing and comparing relative (in)justice across municipal wards, enabling evidence-based prioritization of interventions in Spatial Justice Priority Areas. This study in essence concludes that inclusive densification in South Africa's arid environments requires strategic intervention toward redirecting infrastructure investment, democratize access to services, and reverse the degradation trajectory in communal tenure areas. Without such intervention, the province faces the prospect of deepening spatial inequality, accelerating environmental degradation, and perpetuating the exclusionary logics that have shaped its settlement geography for over a century.