Land-use and land cover change dynamics in urban Ghana: implications for peri-urban livelihoods

ABSTRACT

This paper examined the dynamics of urban land use and land cover change, and their implications for livelihoods in peri-urban Tamale, Ghana. The study employed household data and Landsat Thematic Mapper (TM) for 1986, Enhanced Thematic Mapper Plus (ETM +) for 2004 and Landsat 8 Operationalised Landsat Imager/Thermal Infrared Sensor (OLIS/TIRS) for 2019. The findings show that the urban expansion process witnessed a transition from agricultural livelihood to a more complex and monetised urban economy, which presented mixed impacts on the livelihoods of the peri-urban households. We argued that the horizontal expansion of urban areas into prime agricultural lands could be halted by promoting vertical development in the form of multi-storey buildings rather than the horizontal multiplication of individual housing units. Supporting the peri-urban households to diversify their livelihood portfolios by venturing into beekeeping, poultry farming, and livestock keeping, which do not require large tracts of land is recommended.

KEYWORDS:

• Ghana
• land cover change
• livelihoods
• peri-urban
• Tamale
• urban expansion

1. Introduction

Urbanisation is fast becoming an impediment to sustainable urban planning, particularly as the unplanned expansion of towns and cities is now ubiquitous in almost all developing countries (Saha et al. 2022; Wang et al. 2022, 2023; Zhang et al. 2023). The natural increase in population, uncontrolled rural-urban migration, the concentration of higher-order facilities, and better jobs opportunities in cities are some of the factors driving urbanisation in most third world countries (Satterthwaite et al. 2010; Adarkwa 2012; Wu and Zhang 2012; Siiba et al. 2018; UN 2019; Ibrahim and Siiba 2020). Ghana’s urban population has more than tripled between 1921 and 2021. At the time of independence in 1957, less than 30% of Ghanaians lived in urban areas but Ghana now has about 56.7% of its population in urban areas (Ghana Statistical Service 2022).

The effects of urbanisation on land use and land cover changes (LULCs) have been widely documented (Al-Hameedi et al. 2021; Dey et al. 2021; AlDousari et al. 2022; Rahaman et al. 2022; Saha et al. 2022; Wang et al. 2022, 2023; Zhang et al. 2023). The most acknowledged general consequence of urbanisation is the conversion of agricultural land surfaces into alternative land uses which has long-term impacts on peri-urban livelihood (Song et al. 2015; Mishra and Rai 2016; Pal and Ziaul 2017; Tran et al. 2017). Urbanisation and uncontrolled urban expansion do not only negatively affect the agricultural-depended livelihood of peri-urban households, but also present food security threat in urban areas by creating imbalances in the food supply and demand landscape (Gomes et al. 2019; Bonye et al. 2021). As cities expand spatially, large portions of peri-urban agricultural lands get converted to urban land use forms, thereby depriving farmers of their arable lands (Kuusaana and Eledi 2015; Oduro et al. 2015; Abass et al. 2018; Bonye et al. 2021). Although urbanisation can serve as a positive development factor (Alaci 2010; Afriyie et al. 2020), it can also be detrimental to development if its impacts are not well managed (Ode and Fry 2006).

Méndez-Lemus and Vieyra (2017) asserted that urban expansion into peri-urban areas sometimes means the migration and restructuring of social groups and livelihoods. Urbanisation of rural areas thus provides families with a range of agricultural and non-agricultural livelihood options. This means that families can participate in various livelihood activities to increase their income while sharing new resources (Satterthwaite et al. 2010; Hussain and Hanisch 2014; Afriyie et al. 2020). Despite competition from non-agricultural land users, farmers in the peri-urban interface often adjust their traditional agricultural livelihood styles to embrace appropriate livelihood strategies to sustain their lives in their new locations (Hussain and Hanisch 2014; Afriyie et al. 2020).

A livelihood strategy is an activity that individuals and households undertake to improve or maintain their well-being or to cope with inherent impoverishment using available assets (Oduro et al. 2015). Although various livelihood strategies can be carried out to mitigate the negative effects of urbanisation (Brook and Dávila 2000; Farrington et al. 2002; Afriyie et al. 2020), the frequently cited livelihood strategies are the ones explained by Scoones (1998), which are agricultural intensification, livelihood diversification, and migration strategies.

Agricultural intensification strategies involve continuous dependence on agriculture irrespective of how much the farm possessions of affected persons have been hit (Afriyie et al. 2020). Intensification involves the use of fertiliser (organic and inorganic), irrigation as well as pests, diseases, and weed control to increase yield per unit area. A great deal of capital and management skill is required for the successful application of irrigation, flood, drainage, and erosion control, fertiliser, and biocide use. However, the high demand for land, which has necessitated the commodification of lands in peri-urban makes it difficult for affected farmers to continue cultivating on large scales as a livelihood strategy (Kuusaana and Eledi 2015). Livelihood diversification means expanding livelihood options, including agricultural and non-farm income activities. The reasons for diversification may vary from the desire to accumulate, invest, and diversify risk to the need to adapt to survive in erosive environments (Adepoju and Obayelu 2013). Nevertheless, limited non-farm opportunities for new urban settlers could make it difficult for large-scale diversification among affected peri-urban dwellers (Adepoju and Obayelu 2013; Kuusaana and Eledi 2015). Migration, on the other hand, means moving temporarily or permanently in search of new ways of life (Afriyie et al. 2020). Migration is an adaptive strategy to alleviate poverty, risk and vulnerability, and also a strategy for navigating changing trends of modern cities (Zhang et al. 2021, 2023; Wang et al. 2022, 2023). All the above livelihood strategies are not necessarily mutually exclusive, but there is a trade-off between alternative types and the possibilities of combining different selected components (Ford et al. 2009).

There is a significant number of research on the impact of LULCs on the development of future cities (Weng et al. 2007; Sun et al. 2016; Wu et al. 2016). In the global context, the majority of the studies are based in Asia and are focused on modelling the impact of future LULCs on land surface temperatures, seasonal land surface temperature changes, impact of vegetation cover loss on surface temperature, and carbon emission. The few studies on Ghana (Cobbinah and Amoako 2012; Abass et al. 2018; Afriyie et al. 2020) have largely concentrated on the impact of urbanisation on arable land and agricultural production in southern Ghana. This leaves a gap in understanding the livelihood effects of land use and land cover change dynamics in Ghana, particularly northern Ghana.

In the present study, we aim to use household data and satellite imagery to examine the impact of urbanisation and urban expansion on peri-urban livelihoods in Tamale, Ghana. By understanding the key impacts of urbanisation-induced LULCs and the peri-urban households response mechanisms. The findings can be useful for urban planners, government, and non-governmental organisations in designing future interventions towards promoting sustainable urban development and livelihood strategies. The remaining sections of the paper discussed the conceptual framework, followed by the methodology of the study. Furthermore, the results are presented, followed by the discussion and the conclusion and policy implications.

2. Conceptual framework

Our proposed conceptual framework (Figure 1) highlights the issues of urbanisation, urban expansion and livelihood strategies, and their interrelationships. It builds on the Sustainable Livelihood Framework (DFID 1999), encompassing five interacting elements, namely effects of urban expansion, livelihood assets, livelihood strategies, livelihood outcomes, and institutions, policies and processes. A key component of the framework is the five livelihood assets – natural, physical, human, financial, and social – on which peri-urban livelihoods are built.

Figure 1. Conceptual framework.

Sources: Adapted from DFID’s Sustainable Livelihood Framework (1999).

Natural capital assets include land, forests, water, and mineral resources that can be consumed, sold, or converted directly into consumer goods or tradable products (Oduro et al. 2015). However, urbanisation resulting in fierce competition for resources in the peri-urban zones adversely affects the natural capital base available to peri-urban households. Human capital assets also refer to the quantity and quality of labour that households can use to perform their productive tasks. They include knowledge, skills, health and physical abilities, which are essential for the implementation of various livelihood strategies. Physical capital assets, on the other hand, consist of tangible assets created by people who directly or indirectly contribute to their livelihoods. These include assets such as houses, tools, and equipment, as well as public infrastructure that they can use (Farrington et al. 2002). The availability and access to urban facilities such as better roads and transport systems, often free up better job opportunities for people, thus increasing the possibility for higher income. Financial capital assets also include personal savings, cash, credit/debit cards, remittances or regular pensions, and other economic assets (DFID 1999). The urban poor may survive by engaging in commercial activities, while rural dwellers generally have better access to land for subsistence farming. Even though obtaining financial capital through various cash-income jobs enables people to pursue different livelihood strategies, peri-urban dwellers are considered vulnerable if their survival depends on cash-income jobs (Kutiwa et al. 2010). Social capital assets include the social support networks that people seek when pursuing different livelihood strategies. Heterogeneous lifestyles in cities undermine social ties and erode communal spirit, reinforcing people’s tendency to neglect traditional responsibilities (Adomako 2013).

We argue that in the process of horizontal expansion and continuous conversion of agricultural lands into non-agricultural uses, assets are lost, but at the same time, new ones are created. This creation-destruction phenomenon is reflected in the various restrictions and benefits generated by horizontal urban expansion, including changes in land use, loss of cultivated lands, access to urban market and access to employment opportunities in urban areas (Oduro et al. 2015). All these are external forces that directly limit or enhance the status of household assets. For example, agricultural lands are rapidly converted into urban land use forms and the surrounding natural resources including land, water and air are used as a sink for urban garbage, which directly destroys natural capital assets such as the land on which peri-urban residents depend for survival (Brook and Dávila 2000). But peri-urban areas tend to benefit as they are exposed to urban economic activities. Households can give up their traditional livelihood activities and engage in alternative income-generating activities. Households can also exchange natural capital for financial capital when farmlands are sold out for alternative uses. Nevertheless, problems arise when the profits from the sale of farmlands are not used by the displaced people themselves for productive purposes. The extent of the impact of urban expansion on peri-urban livelihoods depends on the sources of livelihoods and the combination of assets available to household members and how they can effectively use these assets (Oduro et al. 2015; Afriyie et al. 2020). Peri-urban households rely on various livelihood resources to formulate livelihood strategies in response to the dynamics of urban growth and its horizontal expansion. The assets that people have today help them deal with the opportunities and constraints brought about by urban expansion (DFID 1999).

Livelihood strategies and outcomes are not only determined by access to capital assets or limited by households’ vulnerability, but also by structures and processes (Serrat 2008). The institutions also play an important role in creating and shaping the contexts, assets, and outcomes of vulnerability. At the peri-urban interface, different institutions operate at different levels of decision-making. The institutions enable people to achieve positive livelihood outcomes by creating an enabling environment where people can implement their livelihood strategies. This is achieved by introducing and implementing useful plans and providing structures such as markets to transform one asset into another.

Although the sustainable livelihood framework is suitable for the analysis of the livelihood outcomes of urbanisation, it presents a simplistic description of the outcomes of this phenomenon. To suggest that urban expansion creates solely positive livelihood outcomes is a simplification of an otherwise complex interaction between assets, vulnerability contexts and livelihood strategies. The outcomes of the livelihood strategies can also be negative, positive, or neutral (where no change occurs). Besides, it is relatively ambiguous as regards what transforming structures are and how the vulnerability is influenced by livelihood outcomes (Afriyie et al. 2020). This notwithstanding, the framework provides an analytical structure that enhances a broad and logical understanding of the different factors that limit or promote livelihood opportunities, and demonstrates how they are interrelated (Addinsall et al. 2015). It provides a valuable conceptual starting point to enhance a people-centred, all-inclusive, and multilevel understanding of the dynamic ways in which traditional landholders are able to adapt their livelihood strategies in response to the changes that come with modernisation (Addinsall et al. 2015).

3. Methods

3.1. Study setting

Our study is focused on the Greater Tamale Metropolitan Area (GTaMA). The GTaMA consists of the Tamale Metropolitan Assembly (TaMA) and the Sagnarigu Municipal Assembly (SaMA). The GTaMA is located within latitudes 9°16′N and 9°34′N and longitudes 0°34′W and 0°57′W (Figure 2). It covers a total area of 922 km², lying approximately 180 metres above sea level with a generally rolling topography (Tamale Metropolitan Assembly 2013; Ghana Districts 2022; Ghana Statistical Service 2022). The GTaMA is also characterised by two distinct seasons, the rainy and the dry seasons. The rainy season predominantly occurs between April and October annually while the dry season predominantly occurs between November and March per annum. The seasonal cycle of the city affects agricultural productivity, resulting in only one cropping season and limited feed for livestock in the area.

Figure 2. The study setting.

As the administrative and commercial centre of the Northern Region, Tamale has diverse job opportunities and social services, making it attractive to potential settlers from other parts of Ghana (Ghana Statistical Service 2013). Until the early 1980s, the proportion of the metropolis’ population engaged in agriculture (both crop farming and livestock rearing) was around 70%. The local economic structure of the metropolis has changed considerably with the proportion of the population engaged in agriculture-related activities falling to around 42% in 2008, while employment in the service sector such as trade, banking and non-banking financial institutions, transport and NGOs rose to around 58% of the metropolitan population (Tamale Metropolitan Assembly 2013). The 2010 population and housing census of Ghana revealed that about 81% of those employed in the Tamale metropolis were engaged in private informal activities, compared to 13% in the public formal sector and 5% in private formal activities.

3.2. Data collection

We purposively selected the peri-urban communities based on proximity to Tamale (i.e. at least 10 kilometres away from the city centre) (Cobbinah & Amoako 2012) and the presence of multiple livelihood portfolios in response to the impact of urban expansion. Four peri-urban communities – Adubiliyili, Kakpagyili, Banvum, and Zuo – were studied (Figure 2).

We presented information sheet explaining the objective of the study to each participant and gave them sufficient time and the necessary assistance to study and understand the content of the information sheet to enable them take informed decisions to participate in the study. We asked participants who finally decided to participate in the study to sign or thumbprint a consent form before the data collection started. The consent form informed the participants of their rights in the study – to participate or not and to end participation at any stage during the research. We made it clear to the participants that participation in the study was voluntary and that they could choose not to take part in the study or withdraw from the study at any time. Where necessary, we made provisions for the information letter and consent form to be read to the potential participants in the most dominant local language (i.e. Dagbani) of the study setting.

The household survey was conducted using interview schedules. These were administered to the participants in a cross-sectional study using the face-to-face approach. We employed the systematic random sampling strategy to select the participants from within the selected communities for the study. In all, 270 participants, involving only adults (men and women) aged 18 years and above who were heads of their respective households were selected. We collected the background information of the participants in the first section of the research instrument and information about their economic activities and livelihood strategies in the second section. The research instruments were administered by the lead author with the support of two trained research assistants who were all familiar with the communities and could speak Dagbani.

The qualitative data were obtained using interview guides. This involved in-depth interviews involving 56 purposively selected heads of households. The interview centred on the livelihood impact of urban expansion and household livelihood responses. We conducted all the interviews in the local dialect (Dagbani) to ensure adequate comprehension and effective participation by the participants. Each interview lasted for about 40 minutes and the responses were audio recorded.

We downloaded the satellite imageries from the United States Geological Survey (USGS) after fulfilling the initial requirement for registration on the webpage. The main satellites imageries were Landsat Thematic Mapper (TM) for 1986, Enhanced Thematic Mapper Plus (ETM +) for 2004 and Landsat 8 Operationalised Landsat Imager/Thermal Infrared Sensor (OLIS/TIRS) for 2019. Information on the satellite imageries is shown in Table 1.

Table 1. Characteristics of satellite imageries.

Year	Landsat sensor	Bands used	Spatial Resolution
1986	Thematic Mapper (TM)	1, 2, 3, 4, 5 and 7	30m×30m
2004	Enhanced Thematic Mapper Plus (ETM+)	1, 2, 3, 4, 5 and 7	30m×30m
2019	Operationalised Landsat Imager/Thermal Infrared Sensor (OLIS/TIRS)	1, 2, 3, 4, 5, 6 and 7	30m×30m

3.3. Data analysis

We processed the quantitative data using the Statistical Package for Social Sciences (SPSS, v27) alongside Microsoft Excel to generate descriptive statistics about the characteristics of the participants. The qualitative data were analysed using the thematic analytical framework (Nowell et al. 2017). There were various steps in this analytical process. The audio recordings were initially translated into English. The transcripts were then carefully read after this. After this, we thoroughly studied the field interview transcripts to familiarise ourselves with the data. The transcribed interviews were then exported to the NVivo 12 analytical software where data were coded. Coding continued until a point of theoretical saturation where further coding of the data yielded no new concepts. A further step entailed grouping and organising all of the possibly pertinent coded data extracts into themes. We defined and named these themes. Theme-specific stories were isolated and followed by a thorough analysis of each of these themes. Where relevant, direct quotes from interview transcripts were used to highlight pertinent themes.

Analysis of the satellite imageries went through image pre-processing, image classification, and evaluation of map classification accuracy. The need for pre-processing of remotely sensed data resulted from frequent atmospheric and geometric noises that characterise satellite data (Lillesand et al. 2015). Fundamentally, satellite-based imageries are constantly confronted with atmospheric effects such as scattering by aerosols and other cloud cover, or absorbed by suspended water vapour (Kafy et al. 2020). We corrected the atmospheric effects on the satellite imageries and enhanced the visualisation of the imageries by converting the pixel values to Top of the Atmospheric Reflectance (TOA). In the level-1 of the Landsat products, all data were precision registered using Ground Control Points and Digital Elevation Model (DEM).

To facilitate the band combination process, we merged all the individual bands into a composite band using the raster composite algorithm in ArcGIS 10.8. Afterwards, we undertook the band combination task through both the true and the false colour composition. For the true band under the Landsat 8 OLIS, we used band combinations 4, 3, and 2 for true colour composition and band combination 3, 2, and 1 for the Landsat 5 TM. However, due to the visual inability of humans to distinguish extremely fuzzy related features, we used a false colour combination to bring out certain features. With respect to distinguishing urban features from the rest of the features, band combination of 7, 6, 4 under Landsat 8 OLI, and for vegetation cover, bands 5, 4, 3 were used. Whereas in Landsat 5 and 7, we used bands 4, 3, 2 for vegetation, we used bands 7, 5, 3 for urban area emphasis.

We used the supervised classification method to group pixels in the data set into classes corresponding to user-defined training classes (Saha et al. 2022). Supervised classification is carried out in three steps: training, classification and output data set. Training samples were collected for each type of land use/cover change in the image to be classified. For us to obtain a true representation of the cover classes, the training samples were repeatedly selected, evaluated and analysed by either merging or deleting them. The classification was done using a Maximum Likelihood Classifier (MLC) (Dey et al. 2021; Naim and Kafy 2021). The MLC uses the pixel value to evaluate which class the pixel is most likely to belong to. As shown in Table 2, using the Anderson classification system, a single image was divided into five different land cover categories – urban/built-up, bare lands, agricultural land, water bodies and vegetation.

Table 2. Land use land cover classification scheme.

Land use/cover type	Description
Build-up/urban	It consists of dense land use areas, most of which are covered by buildings. Land uses include residential, industrial and commercial units, road and rail networks, and other related land uses.
Bare lands	It is made up of all vacant spaces, rocky areas, sand and cleared lands.
Agricultural lands	Consists of irrigated and rain-fed agricultural land, agricultural land under permanent cultivation, agricultural land and fallow land.
Water bodies	Consists of all areas of open water, including lakes, rivers, streams and canals, reservoirs and ponds.
Vegetation	Consists of a mixture of light forest groves, grasslands and other green spaces. These include natural and manmade forests, natural grasslands and sparsely planted trees.

4. Results

4.1. Background profile of the participants

In general, 57% of the respondents were males, with 88% aged between 25 and 54 years. The highest level of education among the respondents was tertiary education, with 41.5% of them not having any form of formal education. Farming was self-reported as the most predominant occupation and primary source of income for over half of the study participants (61.5%), followed by trade/commerce, services and artisanal work representing 22.6%, 6.3% and 5.2%, respectively. This generally shows that the communities are predominantly farming communities with some non-farming activities occurring in off-farming seasons. The reported monthly income of the participants ranged between GH¢100 and GH¢1500. About 38.9% of the respondents earned between GH¢ 501 and GH¢ 1000 as monthly gross income, and only 7.4% had a monthly income above GH¢1500. Farmers in the study communities were engaged in crop farming and animal husbandry. Some of them too engaged in the cultivation of maize, sorghum, rice, tubers, cassava, sweet potatoes, and groundnuts. Vegetable production was, however, only visible in the immediate outskirts of Tamale, where many of the participants were engaged in irrigation farming. The background characteristics of the participants are shown in Table 3.

Table 3. The demographic and socio-economic data of respondents.

Variable	Frequency	Percentage
Gender
Male	154	57.0
Female	116	43.0
Age
19–24	27	10.0
25–34	93	34.4
35–44	95	35.2
45–54	50	18.5
55 and above	5	1.9
Level of Education
No formal education	112	41.5
Primary	39	14.4
JHS	58	21.5
SHS/SSS	31	11.5
Technical/Vocational	13	4.8
Tertiary	17	6.3
Marital status
Single	34	12.6
Married	222	82.2
Divorced/Separated	7	2.6
Widow/Widower	7	2.6
Household size
1–5	54	20.0
6–10	148	54.8
11–15	47	17.4
16 and above	21	7.8
Main occupation
Farming	166	61.5
Manufacturing	12	4.4
Trading/Commerce	61	22.6
Service provision	17	6.3
Artisan	14	5.2
Monthly income (GH¢)
100–500	101	37.4
501–1000	105	38.9
1001–1500	44	16.3
Above 1500	20	7.4

4.2. Accuracy assessment

We utilised the maximum likelihood classification for the 1986, 2004 and 2019 LULC maps and obtained an overall accuracy of 90.23%, 89.76% and 84.41% respectively. Randomly selected points were generated from the reference data as ground truth to verify the classification accuracy.

The producer’s, user’s and overall accuracies, and Kappa coefficients are shown in the confusion matrix (Table 4). Confusion matrix and Kappa statistics are among the best measures of image classification accuracy (Pontius and Millones 2011; Kafy et al. 2020). In general, the maps conformed to the minimum accuracy stipulated by the Anderson classification scheme (Anderson 1976). The image processing methods are considered effective in producing compatible land use/cover data regardless of the differences in spatial, spectral, and radiometric resolution of satellite data. In Table 4, for example, the producer’s accuracies for urban, cropland, vegetation, water, and bare land for 1986 are 100%, 78.84%, 95.12%, 95.45%, and 85.71% while the user’s accuracy for urban, cropland, vegetation, water and bare land are 83.72%, 95.34%, 90.69%, 97.67%, and 83.72% respectively.

Table 4. Confusion matrices.

Classified Data	Built-up	Cropland	Vegetation	Water	Bare land	Total	Producer’s Accuracy	User’s Accuracy
1986 – Overall Classification Accuracy = 90.23% Kappa Coefficient 87.79%
Built-up	36	3	1	0	3	43	100.00%	83.72%
Cropland	0	41	1	0	1	43	78.84%	95.34%
Vegetation	0	0	39	2	2	43	95.12%	90.69%
Water	0	1	0	42	0	43	95.45%	97.67%
Bare land	0	7	0	0	36	43	85.71%	83.72%
Total	36	52	41	44	42	215
2004 – Overall Classification Accuracy = 89.76% Kappa Coefficient 87.72%
Built-up	32	5	2	0	4	43	94.11%	74.41%
Cropland	1	41	1	0	0	43	80.39%	95.34%
Vegetation	1	2	38	0	2	43	90.47%	88.37%
Water	0	1	0	42	0	43	100.00%	97.67%
Bare land	0	2	1	0	40	43	86.95%	93.02%
Total	34	51	42	42	46	215
2019 – Overall Classification Accuracy = 84.41% Kappa Coefficient 80.23%
Built-up	32	3	1	0	7	43	86.48%	74.41%
Cropland	1	39	0	0	3	43	79.59%	90.69%
Vegetation	3	2	33	0	5	43	89.19%	76.74%
Water	0	0	1	42	0	43	100.00%	97.67%
Bare land	1	5	2	0	35	43	70.00%	81.39%
Total	37	49	37	42	50	215

The omission error for urban, cropland, vegetation, water, and bare land are 0.0%, 21.16%, 4.88%, 4.55%, and 14.29% respectively. The computed error of commission for urban, cropland, vegetation, water, and bare land in percentages terms are 16.28%, 4.66%, 9.31%, 2.33%, and 16.28%, respectively. The 100% (producer’s accuracy) for urban/built-up have been correctly identified as urban/built-up, but only 83.72% (user’s accuracy) of the areas identified as urban are actually urban. This indicates that there is a commission error of 16.28%. For water bodies, although 95.45% have been correctly identified as water bodies, 97.67% of the areas are actually water bodies, implying an omission error of 4.55%. In the case of vegetation, 95.12% have been correctly identified as vegetation but only 90.69% are actually vegetation. For cropland, 78.84% have been correctly identified as cropland but 95.34% are actually cropland. In the case of bare ground still using the 1986 image, 85.71% of the land area covered by bare ground has been correctly identified as bare ground but only 83.72% of the area is actually bare ground.

4.3. Land use and land cover changes

The land use and cover classifications for the various years 1986, 2004 and 2019 are shown in Figures 3a, 3b and 3c. The results show that all the land use types experienced one form of change or the other during the 34-year period. Specifically, built-up/urban lands gained more in terms of spatial coverage over the years, while agricultural land, bare land and water bodies lost in spatial coverage. The composite statistics, in terms of hectares and percentage of land cover change, are also shown in Tables 5 and 6 respectively. In this regard, the results show that build-up areas steadily increased from 1277.31 hectares in 1986 to 1444.41 hectares in 2004 and further to 3637.74 hectares in 2019 (see Table 5). Urban/built-up areas recorded the highest percentage of change in spatial coverage (+184.78) (Table 6). Agricultural lands, on the other hand, experienced a reduction from a land cover size of 54,493.88 hectares in 1986 to 50,448.32 hectares in 2019. Vegetative cover decreased from an initial 921.21 hectares in 1986 to 883.12 hectares in 2019. Water bodies also declined from 54.13 hectares in 1986 to 45.01 hectares in 2019.

Figure 3. Land use/cover of tamale metro in 1986.

Figure 3. Land use/cover of tamale metro in 2004.

Figure 3. Land use/cover of tamale metro in 2019.

Table 5. Composite area statistics (Hectares).

Year	1986		2004		2019
LULC	Area (ha)	% Total	Area (ha)	% Total	Area (ha)	% Total
Urban/Build-up	1277.31	1.97	1444.41	2.23	3637.74	5.62
Croplands	54493.88	84.14	56681.16	87.52	50448.32	77.90
Vegetation	921.21	1.42	775.69	1.20	883.12	1.36
Water bodies	54.13	0.09	37.65	0.05	45.01	0.06
Bare grounds	8019.55	12.38	5827.98	9.00	9751.91	15.06

Table 6. LULC percentage change trend (1986–2019).

Year	1986–2004		2004–2019		1986–2019
LULC	Area (ha)	% Change	Area (ha)	% Change	Area (ha)	% Change
Urban/Build-up	167.1	13.08	2193.33	151.85	2360.43	184.78
Croplands	2187.28	4.01	−6232.84	−11.00	−4045.56	−7.42
Vegetation	−145.52	−15.80	107.43	13.85	−38.09	−4.13
Water bodies	−16.48	−30.44	7.36	19.55	−9.12	−16.84
Bare grounds	−2191.57	−27.32	3923.93	67.33	1732.36	21.60

4.4. Effects of urban expansion on households’ agricultural livelihood

As Tamale expands spatially, it affects the livelihoods of peri-urban households in different ways. Findings show that the horizontal expansion of Tamale is both a beneficial and detrimental. It is detrimental because it has led to a loss of agricultural land, and a decline in food crop production and household income. About 62% of the respondents expressed worry about how urban residential development has taken up their farmlands. We found that the farm size per head in the study communities reduced from about six to three acres for 55.6% of the participants. Twenty-eight percent of the respondents suffered a reduced arable land from about four to two acres; 4.3% lost the entire land under cultivation while 9.1% did not suffer any loss of land. This means that 90.9% either lost part or entire parcel of the land which they hitherto cultivated. Consequently, the agricultural livelihoods of the majority were affected negatively. Respondents indicated a reduction in their agricultural output due to agricultural land depletion occasioned by the urban expansion. The general position among the study participants was hiking food prices. A participant said in an in-depth interview:

Tamale has grown rapidly in size leading to encroachment of arable lands. A big chunk of what used to be cultivable land has been converted to residential facilities. I lost the land that I used to cultivate forcing me to move farther outward. This really has somehow affected my output and income negatively. (Male respondent, Kakpagyili)

Another participant at Adubiliyili also expressed that:

When my husband and I first moved to this community some 25 years ago, there was a lot of land for farming. In fact, that was the main reason why we moved to this place. However, things have changed in the last few years because most of the farmlands have now been taken up for the construction of houses. We have to move outward for land to cultivate. The major problem is that it is a bit far. (Female respondent, Kakpagyili)

I have lost my land entirely due to encroachment by residential use. People put up houses anyhow while the city authorities do nothing about it. Although I am now into petty trading, which is a bit lucrative, I have to spend so much of my income to buy food. (Male respondent, Zuo)

Conversely, the expansion of Tamale from the perspective of some respondents is beneficial. These respondents constituted 25% of the study participants. They mentioned increased market for their agricultural produce as the peri-urban areas get exposed to urban influence, easy access to the city as a result of improvement in road networks and other forms of infrastructure. But as the respondents noted, the negative effects of the expansion of the city on their economic wellbeing far outweigh the benefits. They explained that limited capital was a major constraint in diversifying their agricultural livelihood activities and expanding their operations to meet the increased demand by the expanding urban market for their agricultural output.

4.5. Urban expansion and household livelihood responses

Due to the unfavourable pressures of horizontal expansion, our participants employed both agricultural and non-agricultural related strategies to sustain their livelihoods. The results show that arable and pastoral farming remain the dominant livelihood activities with 95.7% of the respondents engaged in these. However, there was evidence of households diversifying their agricultural livelihoods and going into non-agricultural livelihood activities to supplement their income. Agricultural livelihood strategies took the form of either changing within the same livelihood activity (e.g. from growing corn to the cultivation of soybeans because there was more demand for soybeans) or from one occupation to the other (e.g. agriculture to commercial activity). Agricultural response strategies involved agricultural intensification and diversification of crop production. About 16% of the participants practised agricultural diversification such as animal rearing under the semi-extensive system of farming while the majority of crop farmers tried to diversify their agricultural livelihood portfolio by adopting mixed cropping.

Participants also engaged in non-agricultural livelihood activities. Among these non-agricultural activities were commercial activities, manufacturing (soap, locally brewed drink, etc.), artisanal works, construction activities, and charcoal production. The reasons for engaging in these economic activities, according to the participants, included the need to augment the household income in the face of dwindling agricultural fortunes resulting from loss of arable land to urban expansion. More than half of the participants (66.3%) reported resorting to trading/business alongside other livelihood activities. About 15% of the respondents were into artisanal works as supplementary to farming; 4.4% were into manufacturing; 5.2% were engaged in the services sector including mobile money service and the sale of call credit cards for some telecommunication companies. Charcoal production was becoming an important livelihood activity, particularly during the off-farming season. Remittances from household members living elsewhere were seen as an important part of the participants’ livelihood. About 50.4% of respondents relied on the benevolence of their family members. Thus, our participants asserted that the adoption of alternative livelihoods in trade/business and other non-agricultural activities was a good opportunity for them to compensate for their shortfalls in the agriculture sector. The following are some excerpts from the study participants:

For years, I have been engaged in food crop farming to meet my household livelihood needs. Since farming is no longer profitable due to reduced arable land, I have moved into buying and selling of domestic birds and charcoal burning to complement my livelihood activity. (Male Respondent, Banvim, IDI)

… We had to adapt when we realised that farming as a sole livelihood activity was no longer fetching us much. I am a tailor, a farmer and a businessman. What we have been doing provide us a guaranteed income and a secured future. These days things are hard and so you can’t depend on one job.(Male participant, Kakpagyili, IDI)

As a matter of fact, I used to farm on different pieces of land miles away from the city but all such lands are currently absorbed into the city as a result of its expansion. I am now farming on the little plot behind my house as you can see. I cultivate different types of crops and vegetables on the land such as maize, tomatoes, pepper and okra. Even though I do not get much from it, it is better than doing nothing. (Male respondent, Zuo, IDI)

The outcome of these strategies was varied. To the majority (75%), the expansion of Tamale has affected their livelihoods and their economic wellbeing adversely. About 20% said their wellbeing has improved while 5% claimed they neither experienced improvement nor fall in their economic wellbeing.

5. Discussion

Our study shows a horizontal expansion of Tamale, largely along the north and the south-eastern stretch of the city, resulting in a decrease in agricultural lands over the 34 years period. As urban centres expand both in numbers and spatial extent, demand for land for residential and other urban land use forms inevitably rises (Abass et al. 2018, 2019). The increase in the extent of the built-up area in this study occurred at the expense of arable land and vegetative cover which, to some extent, also explains the decrease in the size of vegetation cover (Kafy et al. 2021). A key issue that has been observed in our study is the rapid population growth coupled with the uncontrolled horizontal expansion which makes land within the city centre more scarce and costly relative to the demand for it (Afriyie et al. 2020). Afrane and Amoako (2011) have argued that rising prices of urban land push developers outward to the peri-urban areas where land prices are comparatively lower. This explains why city outskirts and peri-urban areas have become hotspots for developers and other investors. Consequently, hitherto arable land and natural vegetation in the outskirts and peri-urban areas get rapidly converted to housing and commercial development, dispossessing farmers of their lands in the process (Abass et al. 2018, 2019; Afriyie et al. 2020; Adu et al. 2023).

In 1986, agricultural land constituted the largest land use cover, representing more than half (84.14%) of the total land cover for the period. However, by the year 2019, agricultural land lost 6.24% of its area to urban/built-up areas. Indeed, the conversion of agricultural lands and vegetative cover to urban uses manifests in the expansion of road networks for public use and increases in the construction of new social amenities such as health facilities, schools and commercial enterprises. These findings support that of Naab et al. (2013) in northern Ghana and Afriyie et al. (2020) in southern Ghana. Abass et al. (2018) also noted that rapid growth of urban population was associated with the increasing demand for urban land, especially for housing, which invariably affected agricultural lands in peri-urban communities. Similarly, Kpienbaareh and Oduro-Appiah (2019) observed that urban expansion in the Wa Municipality of the Upper West region occurred at the expense of agricultural lands. While the above, in essence, supports the argument that agricultural lands in peri-urban areas are persistently lost to urban land uses, as conceptualised in Figure 1, our study is inconsistent with the findings of Stow et al. (2016), who indicated that city expansion occurs alongside agricultural expansion.

The results showed that the rapid physical transformation of Tamale had both positive and negative implications for the agricultural livelihoods of peri-urban households. Positively, the horizontal expansion of Tamale offered opportunities for access to the urban market and the corresponding expanded market for peri-urban and rural agricultural output. It is expected that as a city expands into hitherto rural areas, it will lead to improved connectivity through improved road networks. This will make it not only easy for farmers to transport their produce to the urban market but will also provide the stimulus for farmers to expand their output through access to agricultural inputs. But our findings suggest that the peri-urban households studied could not grab the opportunities presented by the expanding city to improve their economic conditions.

The study has revealed that urban growth will not automatically impact positively the lives of rural or peri-urban dwellers. They must be prepared for it and take advantage of it. This underscores the need for a mix of livelihood assets (espoused in our conceptual framework) at the disposal of households. From the extant literature, the gains or the losses to the households depend on how they were able to effectively respond to the emerging urban ways of life. As Abass et al. (2013) and Afriyie et al. (2020) noted, how effectively households respond to their livelihood challenges in the face of urban expansion not only depends on the vulnerability context but also on the household’s command of livelihood assets and livelihood strategies adopted. The ability to increase the area under crop cultivation when arable land is under siege of urban expansion for example will depend on the ease with which households are able to access alternative cultivable lands within or beyond the peri-urban interface. This access may be influenced by a number of factors including but not limited to the level of income of households, social networks, land ownership, and tenurial arrangements (Afriyie et al. 2020). Livelihood gains may also depend on the background training, talents, and skills of the individual, which present households with the needed human capital to make a living from varied livelihood portfolios.

On the other hand, the horizontal expansion of Tamale has been associated with negative consequences such as loss of agricultural land, reduction in agricultural-based incomes, and a rise in prices of foodstuffs. The high demands for the peri-urban lands coupled with the emerging commercialisation of farmlands made large-scale farming difficult. Thus, the immediate outcome of the shrinkage of farmlands is reduced crop yield. The reduction in yields further translates into reduced household income and this may signal the emergence of poverty in such households. This development, as the respondents indicated, has increased their expenditure on food items and poses food security threat to the households. This, from the perspective of our study participants, undermines their economic wellbeing. This finding aligns with Méndez-Lemus and Vieyra (2017) who argued that urban growth dynamics adversely impact the livelihoods and wellbeing of people in peri-urban areas. Therefore, urban expansion into peri-urban landscapes may initiate the gradual transition of farm households into poverty. When livelihoods come under threat from shocks and stressors, households would have to adopt different livelihood coping and adaptive strategies to survive (Abass et al. 2013; Elhadary et al. 2013). These strategies may include agricultural intensification, diversification and extensification, and the adoption of multiple livelihood portfolios. Our findings show that the response strategies of the households in the peri-urban Tamale included agricultural and non-agricultural ones. Strategies such as shifting from the cultivation of one crop to another (horizontal mobility), moving from one occupation to a different one (vertical mobility) and adopting multiple income activity as found in the current study have been reported in related studies (Abass et al. 2013; Oduro et al. 2015; Afriyie et al. 2020). Faced with limited suitable arable land, individuals will seek non-farm jobs over agricultural activities (Atamanov and Berg 2011). As Tacoli (2004) pointed out, it is a strategic survival response by vulnerable households who have been forced out of their traditional occupations to reduce risks and satisfy their needs.

Given that income diversification is key to household risk management (Afriyie et al. 2020), the adoption of multiple livelihood portfolios was justified. A study in Nairobi by Thuo (2010) showed that most peri-urban households who previously depended on their farms for food and income resorted to non-farm occupations when opportunities in the agricultural sector declined following population pressure and land conversions. As Afriyie et al. (2020) asserted, the reliance of peri-urban households on a multitude of livelihood activities, resulting from urban expansion, is an important way of risk reduction and livelihood security improvement.

The outcome of these strategies, as illustrated by the underlying framework, was varied. In relation to the current findings, the majority indicated that their economic wellbeing had declined owing to the negative consequences of the horizontal growth of Tamale. This finding consistent with both the underlying theoretical framework and the findings of Afriyie et al. (2020).

In fact, this study is one of the few studies to analyse the intricate link between urban expansion, household livelihoods and response strategies. Our findings, however, should be interpreted in the context of some important limitations. First, the cross-sectional study design did not afford us the opportunity to examine the effects longitudinally. Future studies should employ longitudinal design to examine how the growth of the city impacts the livelihoods of households. Second, the data on the livelihood effects of peri-urbanisation on households was retrospectively self-reported which may potentially be infused with social desirability and recall bias that potentially may undermine the veracity of our findings. The use of member checks and triangulation in the data collected ensured the rigour and trustworthiness of the data collected. We also shared the study’s findings with the participants to make sure they were accurate and consistent with their experiences. Finally, we used satellite imageries with a cloud cover of not more than 10%, and we also corrected all imageries for potential atmospheric effects by converting the pixel values to Top of the Atmospheric Reflectance (TOA). The study may still be improved by using better remotely sensed satellite imageries as the accuracy of classified maps largely depends on the quality/resolution of the imageries used.

6. Conclusion and policy implications

Our study assessed the impact of the horizontal expansion of Tamale on the livelihoods of peri-urban households using the Sustainable Livelihood Framework. The results showed that urban expansion provided livelihood opportunities for people living in peri-urban areas in the form of new jobs such as trade and services. At the same time, it presented some threats to livelihoods in the form of reduced agricultural lands. In essence, the urban expansion simultaneously destroyed and created different livelihood opportunities. Our finding validates key aspects of the SLF. In light of the rapid rate of horizontal expansion of Tamale and its consequences on agricultural lands and the livelihoods of peri-urban dwellers, interventions are required to avert further horizontal expansion and to create decent livelihood opportunities for affected persons in the peri-urban areas.

To address the negative consequences of the urban expansion, urban planning authorities need to control the physical growth of cities into the outlying peri-urban and rural areas. Thus, vertical development of Tamale in the form of multi-storey buildings should be encouraged rather than the horizontal multiplication of individual housing units. This will prevent the sprawl of the city into prime agricultural lands. In relation to this, the spatial planning unit of the Tamale Metropolitan Assembly should integrate remote sensing and GIS techniques in their planning for the purposes of identifying the various land uses, and patterns of urban expansion, monitoring spatial changes over time and maintaining reliable land cover/use database systems. This will enable near real-time decisions on land to be compared as well as monitoring peri-urban livelihoods that depend on the natural landscape.

Socially, farmers in the peri-urban areas should be supported by key stakeholders such as the Ministry of Food and Agriculture (MoFA) to diversify their farming practices. Most farmers in the study communities practised food crop farming, which requires large tracts of arable lands and relies on rainfall for survival. With education and support from MoFA, households can diversify their activities to include other livelihood activities such as beekeeping, poultry farming and livestock keeping, which do not require large arable land. This will reduce the over-dependence on rain-fed agriculture thereby providing a more sustainable way of enduring the threats of urban expansion on livelihoods.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Suale Iddrisu

Suale Iddrisu is a Master of Philosophy student in Geography and Rural Development at the Department of Geography and Rural Development, Kwame Nkrumah University of Science and Technology (KNUST), Kumasi, Ghana. He holds Bachelor of Arts degree in Geography and Rural Development from KNUST. His research interest includes, land-use/cover change, urban planning, human-environment interactions and remote sensing.

Alhassan Siiba

Alhassan Siiba is with the Renaissance Research and Development (RReD), Ghana. His research interests revolve around the intersection of transportation and urban Planning, focusing on issues related to active transport (i.e., walking and cycling), health promotion and transport and land-use integration for sustainable development.

Joseph Alhassan

Joseph Alhassan is a Master of Philosophy student in Geography and Rural Development at the Department of Geography and Rural Development, Kwame Nkrumah University of Science and Technology (KNUST), Kumasi, Ghana. He holds Bachelor of Arts degree in Geography and Rural Development from KNUST. His research interest includes, Climate change, Food security, Water resources and Human-Environment interactions.

Kabila Abass

Kabila Abass is an Associate Professor in the Department of Geography and Rural Development, Kwame Nkrumah University of Science and Technology (KNUST), Kumasi, Ghana. He holds Doctor of Philosophy in Geography and Rural Development from Kwame Nkrumah University of Science and Technology, Master of Philosophy and Bachelor of Arts degrees in Geography and Resource Development from University of Ghana, Legon. His research interest includes: urban land use and environment, problems of urbanisation and peri-urbanisation, hazard studies, geography of health, environment, health and development