Urbanization in Small Cities and Their Significant Implications on Landscape Structures: The Case in Ethiopia

Abstract

Fundamental ideas concerning urbanization are primarily based on studies performed in large cities. It is of interest to study whether or not similar phenomena take place in smaller cities. Small cities are an inherent component of urbanization, and in the future, the majority of globalization is expected to occur in small and mid-sized cities. Understanding the effects of small cities on landscape structures is, therefore, an essential component in planning city land expansion. Accordingly, this study focused on six towns of the Oromia Special Zone Surrounding Finfinnee, Ethiopia, which is broadly known to be experiencing dramatic growth. Time-series Landsat images from 1987 to 2019 with an integrated method, landscape metrics, and built-up density analysis were employed to characterize and compare the dynamics of landscape structures, urban expansion patterns, process, and overall growth status in the towns. The results highlight that all the towns experienced accelerated growth in the built-up areas and highly scattered nature in spatial growth. Landscape ecology analysis confirmed a highly fragmented urban landscape, a significant loss of natural land covers, and disconnected and complicated agro-vegetation patches in all towns, suggesting a lack of rigorous implementation of the master plan. Results also indicated that the Oromia Special Zone surrounding Finfinnee has failed to control urban sprawl to surrounding ecological sensitive areas. The study results, more broadly, highlight that the small cities would have a limited physical and demographic footprint and relatively less contribution to the national economic agglomeration; nonetheless, they can have a notable and important impact in terms of their ecological and environmental influence. Hence, the study suggests policies for monitoring such dynamics and protecting agro-environmental connectivity with particular focus on the small cities.

1. Introduction

Rapid population growth and the related need for housing and other amenities have resulted in an increasing urbanized land cover, mostly along the major roads and rural countryside, which is typically scattered and inefficient in resource utilization [1,2]. The scattered urban growth and disaggregated landscape structures as a result of human activities also influence the function and structure of the ecosystem [3,4,5]. Dispersed urban development can pose a challenge for planning infrastructure and services. Whereas a compact spatial form can assist in implementing an enhanced transportation system as well as minimize the consumption of traffic energy-efficiency and resource utilization such as water pipe networks and drainage system [6,7]. However, some studies argue that the potential for damage to urban green spaces [8], increased traffic congestion, pollution, and crowded services [9] are notable shortcomings of compacting urban development. Even though there is some controversy regarding spatial growth types, the general agreement is that the urban spatial morphology can determine the social, environmental, and ecological situations of a city. Consequently, exploration of the urban spatial dynamics and landscape structures analysis continues to of interest due to its high importance in creating sustainable cities.

Several studies have investigated the effects of urban expansion on the landscape composition and configuration [10,11,12], investigating the existing and projected urban land dynamics [13], and exploring the patterns and possible drivers [14,15]. However, most of these studies have concentrated on large metropolitan regions due to their significant influences on socio-economic and environmental conditions [14,16]. In contrast, the smaller cities and towns, where scattered urban growth is more prominent [17,18,19], have been given little attention [20,21,22]. As a result, the conceptual ideas regarding urbanization have been primarily influenced by the studies in large cities. With the dominance of small cities expected to lead future urbanization globally, it is important to study smaller cities. In 2018, for instance, about 26.5% of the world inhabitant lived in cities with less than 0.5 million dwellers [23]. Similarly, from 2018 to 2030, the number of cities with 0.5 million people or more is expected to increase by 23% in Asia and by 57% in Africa [23]. This scenario has a likelihood to cause fast urban expansion and amplify the stress on neighboring ecosystems. Hence, understanding the rapid urbanization and its implication, the dynamics of landscape patterns in the small cities can play a notable role in sustainable development especially considering the issues of ecosystem services and changing climate while offering common services and infrastructures for the growing population. In addition, such studies can help guidance concerning the actions that could slow down the negative impacts of urbanization that have already been witnessed for large cities.

By analyzing the urban landscape structures, the landscape heterogeneity, and the impact of different ecological services associated with cities can be ascertained [24,25]. Landscape metrics can be useful to reveal the general circumstances of the urban landscape pattern [10,26,27], triggers of urban land use, and change analysis [11,14,15,21,28,29,30,31,32], as well as assessing the projected future urban spatial form [13]. Landscape metrics studies are also primarily applied to detect the relationship between landscape patterns and land surface temperature [33,34,35,36]. Thus, the studies on urban land use dynamics have not only presented temporal and spatial characteristics of urban development but have also offered evidence to address severe ecological and environmental concerns threatening urban ecosystems [37,38,39]. Referring to the land use maps and changes, policymakers, and urban planners can often identify measures to minimize adverse effects related to urban growth.

This study concerns urban expansion in Ethiopia. The urban growth in Ethiopia, and likely in many African cities, is generally characterized by a lack of adequate planning or lack of implementation of plans, resulting in cities that often have a shortage of necessary infrastructure and services [40,41,42]. The absence of urban planning or enforcement may also have contributed to fragmented urban landscape patterns along major roadways to its rural fringes especially in the regional cities. Since the past couple of decades, regional cities in Ethiopia have grown faster than the capital city, Addis Ababa [43]. This feature can likely be explained considering the national economic development [44] and urban improvement policies [45] that fostered the rapid urban expansion in secondary and tertiary cities.

Some studies have explored urban expansion in Ethiopia [15,19,42,46,47], primarily focusing on the analysis of urban growth and urban land changes in the regional or capital cities. However, the literature on the trend of urban spatial growth and its influences on other natural landscape configurations is quite limited with regards to small cities and towns. Additionally, there is a general lack of studies considering the detailed spatial and temporal urban landscape patterns and urban spatial growth trends analysis at different distances from the town center.

The Oromia Special Zone surrounding Finfinnee/Addis Ababa is one of the most rapidly growing towns in Ethiopia. This Zone in general and the towns, in particular, have experienced a high rate of urban expansion. The towns experienced rapid urbanization as an industrial and residential hub. The expansion was propelled by the large migration from various rural areas and from Addis Ababa to the surrounding towns. The Special Zone with is proximity to Addis Ababa, the national capital city also likely contributed to the migrant influx. As a result, the peri-urbanization development continues and the area is also highly susceptible to land grabbing [48]. The spatial growth of the towns in the Special Zone presents an excellent scenario for studying rapid expansion from the inner to the outskirts and from the periphery to the centers of towns. Such trends also reflect the characteristics of most satellite cities globally in general and for the large African cities, in particular.

Accordingly, in this study, we postulate that investigating and comparing the detailed urban landscape structures and dynamics of the towns is important for understanding the historical changes of urban regions, and for guiding future sustainable development. In particular, this study addresses the following questions: Are the urbanization processes in small cities similar to the large and medium-sized cities? In what way do smaller cities affect landscape structures, and what are the possible drivers for urbanization? These questions are addressed by focusing on six towns of the Oromia Special Zone surrounding Finfinnee, Ethiopia. The assessment included the dynamics of landscape structures, the trend in urban growth, the urban spatial patterns in multi-buffer rings analysis, and its possible drivers, from the year 1987 to 2019, as discussed in the following section.

2. Materials and Methods

2.1. Study Area

The study was conducted in the six towns: Burayu, Sebeta, Sululta, Lege-Tafo, Gelan, and Dukem, in the Special Zone surrounding Finfinnee, Oromia regional state, Ethiopia (Figure 1). Finfinnee, which established in the late 18th century, is the capital city of the country as well as the regional state of Oromia. Since its foundation, it has expanded rapidly by conjoining the surrounding rural areas, which is outside of its legal administrative boundary. Within the past 30 years, the physical growth of the city tripled itself, from 99 km² area in 1987 to 284 km² in 2017 [15]. Recognizing the need to contain such a sprawl, Oromia regional state created the Oromia Special Zone surrounding Finfinnee in 2008 [48]. The Special Zone consists of six towns mentioned above having self-administrative powers. Their geographical locations, climate, and socio-economics conditions are summarized in

Table 1. Geographical position, population, and climate of the six cities.

Town	Latitude - Longitude	Population (2017)a	Elevation (m.a.s.l.b)	Mean-Temperature (°C)c
Burayu	9°01’00’’- 9°06’00’’N, 38°35’30’’- 38°42’00’’ E	92,331	2712	14.7
Sululta	9°06’00’’- 9°12’00’’N, 38°42’00’’- 38°47’00’’E	55,358	2730	15
Lege-Tafo	9°01’30" - 9°07’30"N, 38°51’00’’- 38°57’00’’E	27,636	2453	15.6
Sebeta	8°52’30’’- 8°59’30’’N, 38°34’00’’- 38°42’30’’E	167,127	2346	17
Gelan	8°47’30’’- 8°53’00’’N, 38°47’00’’- 38°53’00’’E	59,817	2215	18.5
Dukem	8°45’30’’- 8°50’30’’N, 38°51’30’’- 38°56’00’’E	40,180	2059	19.5

^a estimated from the Central Statistical Agency (CSA) and Bureau of Oromia Finance and Economic Development. ^b Meter above sea level. ^c Source: Ethiopia Meteorological Agency.

and Figure 1. All selected towns, except Dukem, share common boundaries with Addis Ababa, the national capital city. The population of each town ranged from 27,600 (for Lege-Tafo) to 167,127 (for Sebeta) in 2017 (Table 1). The total area of each city is less than 100 km². The elevation ranges from 2976 m (in Burayu) to 2059 m (in Dukem) above sea level (Figure 1). The mean annual temperature ranges from 14.5 °C to 19.5 °C (Table 1).

2.2. Remote Sensing Data

Cloud-free, Landsat Thematic Mapper (TM) and Landsat 8 (OLI) images were used in this assessment. The Landsat images were acquired on the 09, 10, and 01 of February 1987, 1999, and 2019, respectively. All Landsat images (path 168, row 54) were obtained from the United States Geological Survey and Global Land Cover Facility websites. Image classification was done using ENVI version 5.3, with accuracy assessments using ERDAS Imagine version 2014, and ArcGIS version 10.4 to perform the image data processing which includes projection correction and producing maps. The land cover types were classified into five categories: (1) built-up area; (2) agricultural land; (3) vegetation cover; (4) water body; and (5) grassland. They were classified based on knowledge of the areas, the spectral responses of features on the Landsat images, the use of higher spatial resolution imagery, and visual analysis of different remote sensing products. Descriptions of these five categories are summarized in

Table 2. Land use/cover (LULC) classes and descriptions.

LULC Classes	Description
Urban/built-up areas:	Residential, commercial, industrial, transportation, services, communication, and utilities.
Agricultural areas:	Cropland, horticultural farms, irrigation farms, and other agronomic regions.
Vegetation cover:	Land covered by forest patches, woodland, shrubs, scattered trees mixed with grass, and perennial crops.
Water bodies:	Lakes, rivers, and ponds.
Grass:	Herbaceous cover with a minor proportion of trees and shrubs, lawns, parks, and grasses mixed with shrubs and scrubs.

. Supervised image classification was done using the Support Vector Machine (SVM) Classifier algorithm which is widely applied for the urban area analysis [26,49].

Furthermore, Google Earth Pro® (GE) was applied to verify the accuracy of the classified satellite images [50]. The Google Earth images acquired in 2019 were used to verify the accuracy of the classified results. Due to the lack of high-resolution historical satellite images in the study site, the classification accuracy results of 1987 and 1999 were validated by generating more than 140 random points for each land use/cover (LULC) class from the areas where the land cover remained unchanged.

2.3. Analysis of Land Use/Cover and Urban Expansion

Figure 2 shows the general workflow for urban landscape analysis adopted in the study. Generally, this technical workflow presents four major activities: (1) Remote sensing data and processing, (2) classification and validation of the classified images, (3) landscape structure analysis, and (4) spatial patterns of urban growth analysis.

Expansion percentage of change (PC), annual increase (AI), and annual expansion rate (AER) indicators were employed to analyze the LULC changes and calculate the area and rate of annual urban expansion. The PC indicator is used to calculate the proportion of LULC changes, while the AI is employed to reflect the speed of urban areas growth across the study periods. The AER index is applied to show the rate of urban land expansion within each study period, as shown in Equations (1-3), which follows [15,19,51], with some modification.

$$PC= 100\%\times \frac{A_e- A_i}{A_i}$$

(1)

$$AI = \frac{U_e-U_i}{T}$$

(2)

$$AER = 100\% \times \left[{\left(\frac{U_e}{U_i}\right)}^{\frac{1}{T}}-1\right]$$

(3)

where, A_i and A_e represent the area coverage of each land-use land-cover type (km²) at the initial and end of the monitoring period, respectively. Positive percentage values suggest an increase whereas negative values imply a decrease in area coverage. The U_i and U_e represent the same type of built-up area at the initial and end of the monitoring period, respectively, and T is the period from the time i to e. When T is 1 year, the AER is the annual built-up expansion rate.

2.4. Landscape Metrics Analysis

Five landscape metrics were selected for studying the landscape changes, namely the percentage of the urban area (UA), the number of patches (NP), edge density (ED), mean nearest neighbor distance (MNN), and mean patch size (MPS) (

Table 3. Selected landscape metrics adopted from McGarigal [52].

Landscape Metrics	Abbreviation	Unit	Description
Urban land percentage	UA	%	The proportion of the landscape occupied by urban patch in the buffers
Number of patches	NP	number	The total number of urban patches in the landscape
Mean patch size	MPS	ha	Mean urban patch size
Edge density	ED	m/ha	The total length of all edge segments per hectare urban patches
Mean nearest neighbor distance	MNN	m	Urban patch edge-to-edge distance

[52]). These metrics were selected by considering: (i) relevance to the study objectives; (ii) significance in terms of theory and practice; (iii) minimal redundancy; and (iv) ease in calculating and interpreting the results [33,53,54] to characterize landscape dynamics due to the effects of urbanization. The UA landscape metrics quantify the spatial composition while the last four metrics calculate urban spatial configuration.

In addition, the built-up density (BD) analysis has been employed to explore the compactness of the newly growing urban condition using Equation (4), following [25,49]. The BD analysis provides information regarding whether the new urban growth was high or low density, which identifies the occurrence of urban sprawl. Patch Analyst version 5.2 (Fragstats Interface) extension in ArcGIS with the 8-neighbor rule was employed to compute the value of the landscape metrics [55]. The study further explored the detailed landscape alterations across the concentric distances. A series of concentric buffer circles within every one-kilometer distance was generated. In the concentric rings, the urban-rural gradient appeared to be more effective in illustrating the landscape pattern due to the unchanged sampling size [56,57]. Finally, the BD values within each buffer circle were calculated as:

$$BD = \frac{Urban area per buffer ring}{Total buffer area of ring} \times 100\%$$

(4)

where BD represents the built-up density. The value of BD lies between 0 and 1, with 0 representing the low density, whereas 1 signifying high density.

3. Results

3.1. Accuracy Assessment

Table 4. Accuracy assessment of the classified images by study periods (in percent).

Class	1987		1999		2019
	Producers Accuracy	Users Accuracy	Producers Accuracy	Users Accuracy	Producers Accuracy	Users Accuracy
Agricultural land	86.76	77.63	79.14	84.87	82.72	88.86
Built-up area	84.51	91.60	89.39	87.41	92.44	84.18
Vegetation	88.46	85.82	90.48	92.12	91.15	84.16
Water body	100.00	100.00	99.23	100.00	99.50	100.00
Grassland	80.00	85.50	84.13	76.81	82.80	89.87
Overall Accuracy	87.88		88.18		89.4
Overall Kappa	0.85		0.85		0.88

summarizes the LULC classification accuracy assessment results. Overall classification accuracy was found to be 87.9, 88.2, and 89.4 percent for the year 1897, 1999, and 2019, respectively. The kappa coefficient values of the LULC classification were about 0.85–0.88. These results indicate that the classified maps’ truthfulness was greater than 87%. Thus, the image processing method implemented in this study was effective in producing compatible LULC data over time.

3.2. Urban Expansion and Its Response to Other Natural Land Cover Patterns

Figure 3 and Figure 4 display the overall changes in the land use pattern of the study towns over the 32 years study period. Each of the cities witnessed rapid and substantial changes in their land-use configuration. In 1987, the built-up areas were in a relatively small region and concentrated around the city centers. The green area (vegetation and grass) was the largest land-use type in towns of Burayu, Sululta, and Lege-Tafo (Figure 3a,c and Figure 4a), whereas, agriculture was the predominant one in Sebeta, Gelan, and Dukem (Figure 3b and Figure 4c,d). In 1999, the built-up areas stretched out gradually and significantly increased while replacing the vegetated and agriculture area through 2019. Such alterations, caused by urbanization, have notably affected vegetation coverage and agricultural lands in the areas.

As shown in

Table 5. Land-use land-cover (LULC) area in km² between 1987 and 2019.

Classes	Burayu			Sebeta			Lege-Tafo			Sululta			Dekem			Gelan			Total Changes
	1987	1999	2019	1987	1999	2019	1987	1999	2019	1987	1999	2019	1987	1999	2019	1987	1999	2019	1987–2019
Built-up	1.25	1.88	32.34	1.66	2.45	34.50	0.09	0.14	18.88	0.21	0.29	12.06	0.39	0.42	21.38	0.13	0.24	12.20	127.63
Agriculture	15.73	22.89	17.47	50.50	52.27	33.88	14.39	29.29	23.61	10.24	10.75	8.88	34.36	34.02	13.87	63.51	63.58	50.03	−41.00
Vegetation	41.83	34.06	21.11	17.79	16.90	10.91	4.29	3.35	1.99	13.31	9.36	6.42	0.57	1.30	0.83	3.72	3.21	3.36	−36.90
Grass	41.05	23.44	11.18	13.97	12.28	4.62	31.42	17.41	5.70	19.52	22.78	15.61	0.82	0.40	0.06	0.88	1.21	2.65	−67.84
Water	1.06	1.06	1.06	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
Total	83.34	83.34	83.34	103.78	103.78	103.78	50.18	50.18	50.18	62.85	62.85	62.85	36.14	36.14	36.14	68.24	68.24	68.24

and

Table 6. Percentage of LULC class changes for 1987 to 2019.

Classes	Burayu		Sebeta		Lege-Tafo		Sululta		Dukem		Gelan		Total
	1987–1999	1999–2019	1987–1999	1999–2019	1987–1999	1999–2019	1987–2019	1999–2019	1987–1999	1999–2019	1987–1999	1999–2019	1987–2019
Built-up	50.5	1616.9	47.6	1306.9	59.8	13434.8	40.3	3958.3	0.3	5390.7	90.5	2153.7	3421.7
Agriculture	45.5	−23.7	3.5	−35.2	103.6	−19.4	5.0	17.34	−2.4	−58.6	0.1	−20.9	−21.7
Vegetation	−18.6	−38.0	−5.0	−35.4	−22.0	−40.6	−29.7	−31.4	11.3	−81.4	−13.7	4.5	−45.3
Grass	−42.9	−52.3	−12.0	−62.4	−44.6	−67.2	16.7	−31.47	−4.0	−42.0	36.8	120.1	−63.0
Water	0.0	0.0

, the towns under the study area experienced large changes in LULC over the 32-year study period. The highest loss of green areas occurred during the 1999–2019 period in all towns, where the vegetation area increased by 4.5% and grass class increased by 120.1%, except for Gelan (Table 5). Increase in the agricultural regions was noted in all towns except Dukem from the year 1987 to 1999; however, the values were significantly decreased from 1999 to 2019. Conversely, the area of urban land showed a significant increase in all towns during the last decade, in which the highest expansion was observed in Lege-Tafo (13,434.8%) and smallest in Sebeta (1306.9%) (Table 6). In general, due to urbanization, 41 km² ($-$21.7%) area of agricultural land, shifted into urban areas, whereas, a loss of 36.9 km² ($-$45.3%) area of vegetation cover and 67.84 km² ($-$63%) area of grasslands was found (Table 5 and Table 6). The area occupied by the water bodies was relatively stable and showed no changes during the entire study periods (Table 5). To generalize, the total urban area of all towns in the study increased to 127.63 km² (+3422%) in 2019 from the baseline area of 3.73 km² in 1987, leading to a continuous fragmentation and degeneration of natural components of the ecological network.

3.3. Magnitude and Rates of Urban Expansion

The urban expansion maps for the six towns across the study periods are displayed in

Table 7. Annual intensity (AI) and annual expansion rates (AER) of the built-up area from 1987 to 2019.

Town	AI(ha/year)			Urban Expansion Rate (AER) ( % per year )
	1987–1999	1999–2019	1987–2019	1987–1999	1999–2019	1987–2019
Burayu	5.27	152.30	97.15	4.20	81.01	77.73
Sebeta	6.59	160.25	102.63	3.97	65.41	61.82
Lege-Tafo	0.44	93.70	58.73	4.63	669.29	652.43
Sululta	0.71	58.85	37.02	3.17	202.93	176.34
Gelan	0.29	104.80	65.60	0.64	249.52	168.19
Dukem	0.96	59.80	37.72	7.05	249.17	290.14
Total	14.26	629.70	398.85	3.78	116.18	106.93

. With the exception of Burayu and Gelan, the built-up area was small and mainly situated in the central parts of each town in the year 1987, whereas Burayu exhibited the scattered forms in the eastern and south-eastern direction parts of the town. Gelan was rural and had a small and dispersed built-up area during the same time period. The expansion of the built-up area class is evident in the 2019 map. The evaluation of class statistics displays that there has been a noticeable urban area expansion within a span of 32 years.

As shown in Table 7, all towns experienced a fast increase of built-up areas, resulting in dramatic transformations in spatial patterns from 1987 to 2019. The highest and the lowest AI occurred from 1999 to 2019 and between the period 1987 and 1999, respectively (Table 7). Sebeta, which has the biggest city, had the highest AI (103 ha) during the year 1987–2017. However, the town experienced the least annual expansion rate (61.8%) during the same period (Table 7), whereas the smallest town (Lege-Tafo) experienced the highest AER (652.4%). This indicated that small towns have higher expansion rates. Other towns showed rates of annual expansion ranging from 77.7% (Burayu) to 290.1% (Dukem).

More generally, the total annual increment was 398.9 ha/year for the study period from 1987 to 2019. Thus, annually on average about 67 ha of non-urban land from each city was converted to built-up land per annum for the past 32-years. From 1987 to 1999, built-up areas of the Special Zone increased by about 14.26 ha per year. More recently, between the years 1999 and 2019, the study site annually shows 629.7 ha of the non-urban area (Table 7). Thus, the Special Zone transformed about 630 ha of natural land covers per annum during the second study period. Indeed, if the trend continues, it would be equivalent to losing 1.73 ha of agro-vegetation areas per day.

3.4. Dynamic Change of Landscape Pattern During 1987–2019

As shown in Figure 5a–c, the analysis of urban fragmentation metrics calculated from the LULC maps at the urban scale revealed significant changes throughout the study period. The landscape of all towns became more fragmented and complex in pattern as indicated by the constant increase in the NP, ED, and decrease in MPS values. However, the immense increase of fragmentation observed from 1999 to 2019 is a consequence of intensive sprawling processes.

Between 1987 and 1999, the NP shows a gradually increasing trend except for agriculture across the towns, and vegetation in Sebeta, Lege-Tafo, and Sululta (Figure 5a). The results of ED also follow similar trends as of NP over the periods except for Burayu and Lege-Tafo in which the gradual increasing patterns of agriculture were recorded. In contrast, the NP and ED values of built-up, vegetation, agricultural, and grass showed significantly increased during the second study period (1999–2019) in whole towns. At the same time, the densification of the urban core is confirmed by a decrease in MNN over the periods (Figure 5d). For instance, it is marked that in the year 1987, the urban patches of Burayu, Gelan, and Lege-Tafo towns were relatively more distant (370 m, 312 m, and 302 m, respectively) from each other. In 1999, the average distance reduced persistently and was the lowest in 2019 for Burayu, Gelan, and Lege-Tafo towns: 59 m, 80 m, 62 m, respectively. Thus, the urban core is becoming compact due to the development of new built-up patches near and between existing built-patches. In contrast, connectivity of the other LULC patches revealed decreasing trends, suggesting that the natural landscape has become more fragmented across each of the cities over the last 32 years.

In comparison, over the entire study period, the highest fragmented vegetation patches were observed in Burayu and Sebeta, as evidenced by the near-continuous increase in the NP values. Furthermore, the agriculture patches were considerably disconnected in Gelan and Dukem, as indicated by significantly decreased MPS. In 1987, the MPS in Dukem was 486 ha while Gelan had 634 ha (Figure 5). These numbers significantly declined to 27 ha and 8.5 ha, respectively, in 2019. This implies that the towns grew at the expense of farmlands. In general, the landscape metrics analysis indicated that all the studied towns of the Special Zone surrounding Finfinnee had a more fragmented and complicated urban landscape due to the scattered nature of urban growth during these respective periods.

3.5. Built-Up Density Analysis

Figure 6 shows the spatial patterns of the built-up density of each town. Based on the BD values, Sululta and Gelan revealed a similar trend. For both these towns, between the years 1987 and 2019, the BD values steadily increased within 1-2 km distance and the decreasing trend was seen from 2 km onwards (Figure 6c,f). That is, dense built-up and reduced outskirts of the towns are evident. An inverse trend was observed in Burayu and Sebeta. The BD values decreased within 1–5 km in Burayu and 1–6 km in Sebeta from their downtowns and increased at fringes of their towns (Figure 6a,b). This indicates that the newly developed areas on the outskirts are relatively dense. The BD results of Lege-Tafo and Dukem showed that the outward moving of the reducing trend from a distance of 1 to 6 km. The BD values increased within 6–8 km in Lege-Tafo, and finally, the decreasing trend was observed from the distance 8 km (Figure 6d,e). In general, the BD analysis showed that over the past 32 years, except for Burayu and Dukem, the built-up density revealed a complex distribution of clusters growing in patches across each of the towns.

4. Discussion

The study provides clear evidence that the urban expansion rate of towns and small cities can be very high (and larger than those reported for large cities in the literature). Most of the large cities are growing at a rate of less than 10 percent per annum. The annual expansion rate of Addis Ababa (Ethiopia), Tshwane (South Africa), Beijing (China), and Kolkata metropolitan (India) were, for example, 3.4%, 3.52%, 3.46 %, and 5.4%, respectively between the year 1984 and 2017 [11,15,58,59]. This value, however, can exceed 100% in the towns. In this study, for instance, the annual spatial growth rates of Lege-Tafo was 652.43% (see Table 7). The landcover change due to the pressure of urbanization can affect the ecology and environmental conditions.

Most of the large and medium-sized cities, for example, Wuhan [60], Cairo, and Kigali [25] have grown in a compact form with an inner core and minimally scattered manner on the outskirts. However, in small cities, such characteristics were not uniformly noted and much more irregular pattern was found. Among the six towns, in this study, only Dukem showed as such a trend, while all the rest revealed irregular patterns. Dukem showed growth from the inner core to its periphery in all directions (see Figure 6e and Figure 7f). This town has no shared boundary with Addis Ababa, and could be the reason for its expansion in all directions. Other towns have shared a common boundary with a capital city, and they showed a tendency to expand towards Addis Ababa, following the main road, suggesting a lack of rigorous implementation of the master plan.

The outcome from prior studies [10,18,37,38,46] indicated that rapid urbanization and lack of good monitoring and managing urban development, lead to a shift in the natural ecosystem of the area and the formation of a more fragmented landscape pattern. In this study, ecological service can be deteriorated due to the considerable decrease in the grassland, agriculture, and vegetation ecosystem (

Table 8. The total percentage of LULC changes from 1987 to 2019 (Percent).

Classes	Burayu	Sebeta	Lege-Tafo	Sululta	Dekem	Gelan
Built-up	2487.20	1978.31	20877.78	5642.86	9284.62	9284.62
Agriculture	11.06	−32.91	64.07	−13.28	−21.23	−21.23
Vegetation	−49.53	−38.67	−53.61	−51.77	−9.68	−9.68
Grass	−72.76	−66.93	−81.86	−20.03	201.14	201.14

). Such a situation (of loss in natural landscape) could be even more aggravated in smaller cities than the other medium-sized cities. For instance, Mekelle and Hawassa, are the third and fourth-largest cities, respectively, of the country. These cites lost 20.8% [19] and 48.5% [46] of vegetation cover between the years 1984 to 2014. However, this value was higher in this study. Between 1987 and 2019, coverage of vegetation in Burayu, Sululta, and Lege-Tafo towns were, for example, reduced by 49.5%, 51.8%, and 53.64%, respectively (see Table 8). This suggests that the small cities in the vicinity of large cities could be strongly impacted in terms of the ecological connectivity than that of the medium cities.

4.1. Urbanization and Its Influences on the Landscape Compositions

Study results revealed a negative correlation between urban expansion, and green (vegetation and grass), and agricultural area coverage. The rapid increase in the urban area from 1987 to 2019 has resulted in the loss of 21.7% of cropland across the six towns in the study domain, which can decrease the local agricultural products. A comparable pattern was observed in a similar study made on other regional cities of the country, Bahr Dar [40] and Mekelle [19] though the rates differ. Both studies highlighted population growth as a major influencing factor. Indeed, loss in agricultural farmland initiated by urbanization is widespread across the globe. Similar studies made in Pune Metropolis, India [3] and Benin metropolitan area and Nigeria [61] noted a loss of extensive farmlands due to urbanization. The loss of such large agricultural areas might be due to the urban land-use tends to have a strong revenue per unit of land than agricultural products. However, such kind of trend can affect food securities in countries like Ethiopia with high population growth.

Furthermore, the most significant dynamics related to urban growth is the loss of vegetation cover around the periphery due to the replacement of natural land cover by impervious surfaces. Nevertheless, several studies have suggested that urban expansion is also attributed to a reduction of vegetation cover in the urban center, because of the steady rise in urban density and paving of green spaces [49,62,63]. In this investigation, the loss of about 45.3% of vegetation cover including natural forest, could contribute to the deterioration of the environmental services of a city. An increase in the informal settlement could also be one of the main reasons for the degeneration of green vegetation. The decrease in vegetation covers is mainly because of ineffective land management exercise and less soil protection criteria [38]. This situation highlights the need to consider the balance between urban expansion and green vegetation conservation.

4.2. Urban Landscape Dynamics and Associated Triggers

The technologically driven spatial and temporal analysis of this investigation also clearly showed the intensity of urban landscape changes in the study sites. The landscape ecology method implemented in this study contributed to a better understanding of the spatial and temporal dynamics of urban landscape patterns. These landscape metrics permitted a better calculation of the dynamics in spatial composition and configuration inside the urban region and revealed that fragmentation of natural land resources has amplified in all towns over the past 32 years.

According to the landscape metrics analysis, the agro-vegetation patches reduced considerably in composition and became more fragmented in the configuration. The built-up patches grew significantly; however, their structures were more dispersed in all towns, particularly post 1999, which suggests increasing urban sprawl. A similar trend has been observed over many small cities around the world such as Josi, Nigeria [32], Kurukshetra, India [64], and Idaho, USA [22], which is indicative of a dispersed urban growth. As compared with the latest study periods, a relatively stable landscape fragmentation was recorded in the year between 1987 and 1999. Eventually, landscape fragmentation significantly increased during the second study period of urbanization. Thus, a significant fragmentation and increased complexity of the natural land covers/urban landscape distribution is evident due to the influence of isolated and unstructured urban growth. This is indicative of the development of small urban patches around the outlying of the city and along the major roads. Such ad-hoc development also highlights unregulated urban expansion patterns [15,19,21,42,57,65]. Enaruvbe and Atafo (2018) suggested the loss of farming land in recent years might also be a sign of the periurban land becoming urban due to the higher land price in the city center and more accessible urban fringes. Thus, unless appropriately managed, fragmented urban growth is inefficient economically well as ecologically unfriendly [2,18,22].

The rapid growth of urban areas and the fragmented urban landscape of the six towns since 1999 is also indicative of the policy shift towards urbanization and industrialization that has significantly contributed to the rapid transformation of the Ethiopian economy [43]. The increase in the urban area is also associated with population growth. For instance, the population of Sebeta increased by about 612%, from 24,924 in 1999 to 177,370 in 2019, and the built-up area increased by about 1307% during the same study period (see Table 6). This suggested that the urban expansion rate follows the population growth and could be much larger than the population growth rate. Similar trends have been reported for other Ethiopian cities as Addis Ababa, Mekelle, and Bahir Dar [15,19,40,42]. Hawassa city has shown that population growth exceeds the urban expansion rate [15]. This is likely associated with informal settlements, as evidenced by increasingly fragmented and complicated patterns of the urban landscape at the fringes of the city (Figure 6). Furthermore, the towns are closer to the capital city, Addis Ababa, and have the advantage of access to trained labor, infrastructure, and financial services. As a result, businesses tend to establish in these cities. As feedback, this led to the immigration of employment seekers from other parts of the country, resulting in further rapid and scattered urban growth and high fragmented natural landscape patterns.

The urbanization process in the Oromia Special Zone surrounding Finfinnee towns is clearly influenced by their distance from Addis Ababa. The towns with the shortest distance from Addis Ababa have a relatively high stage of urban agglomeration. Except for Dukem, all towns expanded surrounding Addis Ababa. Interestingly, Dukem also has the smallest size relative to the other towns. In addition, the expansion of these towns was influenced by ‘pulling’ forces such as governmental institutions and policies and ‘pushing’ forces such as industrial investment, and access infrastructures (e.g., along with Gelan and Dukem towns), and market for the urban land, in which the price of land is lower at the outskirt. This resulted in the intensified trend of dispersed urban land expansion in these towns. The socioeconomic, cultural, political, and biophysical reflections are the main drivers of land-use trends in several countries and could not be explicitly analyzed [5,42].

4.3. Patterns of Urban Spatial Growth

The built-up density (BD) analysis with distance buffer zones suggests that the built-up areas continue to be dense in the core town, related to infill and extension growth. Nevertheless, the class of built-up areas became disconnected and dispersed on the periphery, attributable to outward and leapfrog development. The dispersion of the urban landscape may be a result of new urban land development around the fringes of the town. However, in most African cities, particularly in Ethiopia, dispersed urban growth is a result of unplanned and uncontrolled manners of newly developed urban areas, which is connected with the informal settlements [15,42].

5. Conclusion

Controlling the fragmentation of landscape in ecologically sensitive regions is essential to conserve agro-vegetation ecosystems. A detailed understanding of such change provides to design holistic strategies intended at controlling the landscape fragmentation and enhancing agro-vegetation connectivity in urban outskirts areas. Understanding urban growth processes and identifying these patterns at a different distance from the city center is vital to manage and plan urbanization processes. Thus, the study comprehensively characterized and compared spatiotemporal landscape dynamics structures and urban expansion patterns in six towns of the Oromia Special Zone surrounding Finfinnee, Ethiopia, using spatially explicit Landsat imagery with combined multiple approaches over the 1987–2019 period.

The results revealed that: (1) all towns under the study experienced accelerated growth in the built-up area, scattered urban spatial growth along the outskirts, and irregular patterns of built-up density. Such an increase may indicate a lack of rigorous implementation of the urban master plan. (2) A significant loss of natural land covers, fragmented, and complicated agro-vegetation patches. In that, it could be inferred that the intensity of human activities influences land-use dynamics, ecological change, and landscape fragmentation. (3) The role of urban planning and management, population, policy, and distance to the capital city emerge as driving forces for the expansion of towns. (4) It also found that the highest urban development, the most dispersed urban growth and the landscape fragmentation period were observed during the year 1999-2019. Concurrently, the rapid increase in the urbanized areas has resulted in a loss of 38.45% of natural land cover. This is indicative that the recently created zone, the Oromia Special Zone surrounding Finfinnee did not control or contain the rapid urban sprawl. It can be postulated that the corresponding reduction in food production might affect and shift to nearby areas, and there might be a probability of flood risk and environmental deterioration due to the significant loss of natural land cover. This situation sounds an alarm warning to consider the balance between urban expansion and green vegetation conservation for improved urban ecosystem services and enhanced liveability. Thus, urban changes need to consider and evolve with green areas, agricultural modernization, and new rural development following the policies based on scientific planning and steady progress. Indeed, sustainable urban growth not only can successfully improve land-use efficiency and promote regional economic, but also plays a vital role in social harmony.

More broadly, the study concluded that even though small cities would have limited footprint and demographic size and have less contribution to the national economic agglomeration, they can still be important for the region’s ecological and environmental services. To curtail the continuous degradation of land that appears to be underway, the study suggests that policies controlling influences of human disturbance and ecological barriers should develop robust conservation of landscape and in particular, seek to reduce-edge effects by controlling urban sprawl in the agro-vegetation system.

The proposed methods in this study can help in identifying the priority areas to be selected for conservation, ecological improvements, urban densifications, and agro-forest system restoration. The findings from the study can help planners and decision-makers in developing sustainable strategies for landscape planning and management. The outcomes from such research could also help to postulate theories and generalize the state of urbanization process and dynamics of the landscape patterns with particular focus on small cities and towns.