Introduction

Urbanization is a complicated process that causes profound changes in rural landscapes in the vicinity of towns and cities (Antrop and Van Eetvelde 2000). The cities have been spreading in a scattered way throughout the countryside by following the transportation lines and the coastlines (EEA 2006: Hepcan et al. 2011). The result is fragmentation of the existing rural landscapes around the urban areas (Antrop 2000), a matter of great concern from an ecological point of view. Consequently, recent studies and efforts (see Esbah et al. 2010a and Hepcan et al. 2011) have been mostly involved in monitoring and understanding landscape changes due to urbanization. Studies for coastal landscapes have been particularly valuable in detecting and solving land degradation issues (Esbah et al. 2010b) because highly populated coastal strips have been in the center of urban development (EEA 2006).

Interpreting how landscape structure and function are influenced by landscape changes is essential to developing sustainable planning strategies in order to minimize fragmentation. For this purpose, landscape metrics can be employed as a useful tool to apply landscape ecological concepts to planning (Botequilha Leitao and Ahern 2002). Connectivity is an important landscape property that illustrates the relationship between landscape structure and function (Botequilha Leitao et al. 2006). Connectivity refers to the degree to which the landscape facilitates or impedes the flow of energy, materials, nutrients, and species throughout the landscape (Marulli and Mallarach 2005; Botequilha Leitao et al. 2006).

Although each urban settlement presents its own challenge, most Turkish (metropolitan) cities have been grappling with massive expansion of built-up areas that have caused profound landscape alterations in their vicinities (Evrendilek and Doygun 2000; Eşbah 2007; Doygun et al. 2008; Hepcan et al. 2011). A recent study confirmed that massive spatial changes from a predominantly rural landscape to a sprawled urban morphology primarily took place in the last decades in the vicinity of Izmir, a metropolitan city in Turkey, along the area’s transportation network and coastline (Hepcan et al. 2011). Likewise, some patterns of Anglo-Saxon suburban development were reported to have re-appeared around the peripheries of Izmir, especially along the growth axis of the city to the west, around the district of Urla (Sönmez 2009). This district, therefore, represents a good example of re-emerged suburbanization and presents a distinctive potential of quantifying impacts of dispersed urban development upon natural landscape pattern.

Up until now, little has been done to understand the impacts of suburban development that are likely to affect natural and rural landscapes in the region. This study was targeted at analyzing and interpreting changes in landscape pattern and connectivity as a key landscape property in the Urla district, using core landscape metrics based on a 42-year data (1963–2005). The objectives were to investigate if and how suburban development affected natural landscape pattern as well as landscape fragmentation, and to explore sustainable development strategies for the study area.

Study area

Landscape characteristics of the Urla district

The Urla district is one of the 29 districts in the Izmir metropolitan area. It comprises an area of 444 km2 and is located between the coordinates 38° 26′ 41″ and 38° 11′ 51″ North, 26° 29′ 46″ and 26° 55′ 50″ East on the southern coast of Izmir bay. It lies in one of the major growth axes of the Izmir metropolitan area to the west (Fig. 1). Urla has 14 villages with a population of 52,500 people (TurkStat 2011).

Fig. 1
figure 1

The study area

Full size image

The district is very rich in terms of cultural and architectural assets. It has two exceptional archeological sites named Limantepe and Klazomenai. The Limantepe mound is the core of the settlement. The earliest evidence of its ancient history dates back to the sixth millennium bc. Klazomenai was one of the 12 ancient Ionian poleis. Excavations have exposed the presence of a settlement from the Neolithic to the Hellenistic period (Klasp 2011). The history of the modern town dates back to the eleventh century. The central part of the town has an organic settlement pattern with narrow and dead-end streets that is typical in most of the Ottomans cities (Sönmez 2009). This is interwoven with a pattern dominated by linear streets (Şengün 2007).

The fundamental landscape elements are agricultural lands, olive plantation, shrubland, and Mediterranean forests dispersed between hill slopes and lowlands. The study region has a Mediterranean climate with hot and dry summers and mild and rainy winters with a mean annual temperature of 16.8°C.

Mediterranean shrubland and coniferous forest are the main components of vegetation cover. Forest is represented by Pinus brutia and Pinus halepensis. Mediterranean shrubland is represented by maquis, phrygana, and maquis–phrygana communities. Maquis is dominated by such species as Arbutus andrachne, Arbutus unedo, Cercis siliquastrum, Laurus nobilis, Myrtus communis, Nerium oleander, Olea europea, Pistacia terebinthus, Pistacia lentiscus, Quercus coccifera, Quercus ilex, and Vitex angus-castus. The main elements of phrygana vegetation are Cistus creticus, Lavandula stoechas, Sarcopoterium spinosum, and Verbascum sinuatum (Görk et al. 1989; Semenderoğlu 1999; Durmuşkahya 2006).

A brief description of suburban development in the Urla district

The first emergence of suburban areas in the city of Izmir goes back to the seventeenth and nineteenth centuries for the higher income people, especially foreign merchants in the city. At the time, suburban development appeared in the Bornova and Buca districts. Especially after the 1980s, transformations in the global economy caused changes in the urban growth dynamics of mega-cities in developing countries. An example of such a change was suburbanization at the peripheries of the Izmir metropolitan area. Suburban settlements sprung up for the city’s upper and upper–middle income groups, primarily along the western growth axis of Izmir toward the Urla district. Urla and its vicinity have been developing as a residential (usually second or vacation homes) and recreation area for Izmir’s dwellers since the 1950s. However, what happened in the Urla district after the 1980s was an accelerated process of urban development. It is important to note that the construction boom in Urla was most likely caused by the construction of the highway that links Urla to Izmir’s central area and the Cesme district (Sönmez 2009). In terms of suburbanization, another important period was the 2000s when the encroachment of luxury tract homes into forested areas and shrubland along the Çeşme–İzmir highway began to impact the landscape (Hepcan and Ozkan 2007). Besides the construction of the highway, other factors that triggered suburbanization in the district were the attractive natural landscape characteristics of the region such as a long coastline, the relatively intact forested areas, the spatial planning decisions by the municipality of Urla, and close proximity to the city center of Izmir.

In recent years, the Urla district has seen a somewhat an unconventional development of the suburban areas. There has been a growing interest among the upper/middle-income people from different cities in Turkey to develop greenhouses, organic farming, vineyards and wineries, and animal husbandry and to convert agricultural and shrubland into sports fields for different recreational activities such as horse riding, paint ball competition, and other activities. Some of the occupants prefer to stay in Urla all year round, but the summer period is still the most popular season for most of the owners.

Method

Data preparation

The data used in this study came from 1963 CORONA and 2005 ASTER satellite images and ten 1/25,000 topographical maps. 1963 land use map was derived from CORONA (1963) satellite images. Three scenes of CORONA (1.83-m ground resolution) black and white images were rectified using the rubber sheeting method (spline). Control points were collected from 2005 IKONOS (2005) and Landsat (1980). The land use map of 1963 was developed by screen digitizing using ArcGIS 9.2 (ESRI 2006). The black and white aerial photographs from 1957 were used for the accuracy assessments of the CORONA images.

In order to develop land use map of 2005, supervised classification was performed to ASTER 1B (2005) image (15-m spatial resolution; 3,2,1 VNIR band combination) using the maximum likelihood parametric rule provided by ERDAS Imagine Professional 9_1 (Leica Geosystems 2006). Accuracy assessment was performed based on the stratified random sampling method where 500 points were selected from IKONOS. Results of the user’s and producer’s accuracy were used to compute overall accuracy and kappa values.

Land use classes of land use maps were determined using the CORINE land cover nomenclature (Table 1) (Bossard et al. 2000).

Table 1 Land use and land cover classes
Full size table

Data analysis

Landscape pattern changes

In order to explore landscape characteristics of the study area, nine landscape composition and configuration metrics were (Table 2) as follows: class area (CA), percentage of landscape (PLAND), number of patches (NP), patch density (PD), largest patch index (LPI), landscape shape index (LSI), mean patch size (AREA_MN), perimeter area fractal dimension (PAFRAC), and connectance index (CONNECT). These metrics were selected on the basis of relevant literature (McGarigal and Marks 2003; Botequilha Leitao et al. 2006);

Table 2 Explanations of landscape metrics (McGarigal and Marks 2003)
Full size table

Class area is a measure of landscape composition, specifically, how much of the landscape is comprised of a particular patch type. CA approaches zero as the patch type becomes increasingly rare in the landscape. CA = TA (total area) when the entire landscape consists of a single patch type (McGarigal and Marks 2003).

Percentage of landscape quantifies the percentage of each patch type in the landscape. When PLAND approaches zero, patch class becomes increasingly rare in the landscape and when it approaches 100, the entire landscape consists of a single patch type (McGarigal and Marks 2003).

Number of patches equals the number of patches of the corresponding patch type. It is a simple measure of the extent of subdivision or fragmentation of the patch type. “Number of patches” was utilized as a measure of the degree of fragmentation of land uses. Thus, land uses with a larger number of patches are characterized by small, highly isolated patches with high edge and low structural connectivity (McGarigal and Marks 2003).

Patch density represents the number of individual patches of a particular type per 100 ha. Patch density is a measure of spatial heterogeneity (McGarigal and Marks 2003).

Largest patch index quantifies the percentage of total landscape area comprised by the largest patch. As such, it is a simple measure of dominance. LPI approaches zero when the largest patch of the corresponding patch type is increasingly small. It equals 100 when the entire landscape consists of a single patch of the corresponding patch type; that is, when the largest patch comprises 100% of the landscape (McGarigal and Marks 2003).

Landscape shape index provides a simple measure of class aggregation or clumpiness. LSI equals zero when the landscape consists of a single square or maximally compact (i.e., almost square) patch of the corresponding type. LSI increases without limit as the patch type becomes more disaggregated. The landscape shape index is a standardized measure of total edge and is essentially a descriptor of connectivity, insularity, and spatial heterogeneity in the landscape (McGarigal and Marks 2003).

Mean patch size is simply the average size of patches of particular land cover types (class types) or across the entire landscape (landscape level). It is a measure of subdivision of the class or landscape (Botequilha Leitao et al. 2006).

Perimeter area fractal dimension of a patch mosaic provides an index of patch shape complexity across a wide range of spatial scales (i.e., patch sizes). Specifically, it describes the power relationship between patch area and perimeter, and thus describes how patch perimeter increases per unit increase in patch area. If, for example, small and large patches alike have simple geometric shapes, then PAFRAC will be relatively low, indicating that patch perimeter increases relatively slowly as patch area increases. Conversely, if small and large patches have complex shapes, then PAFRAC will be much higher, indicating that patch perimeter increases more rapidly as patch area increases, reflecting a consistency of complex patch shapes across spatial scales. The fractal dimension of patch shapes, therefore, is suggestive of a common ecological process or anthropogenic influence affecting patches across a wide range of scales, and differences between landscapes can suggest differences in the underlying pattern-generating process (e.g., Krummel et al. 1987). PAFRAC requires patches to vary in size. Thus, PAFRAC is undefined and reported as “N/A” in the “basename” land file if all patches are the same size or there is <10 patches (McGarigal and Marks 2003).

Connectance index is defined by the number of functional joining between patches of the corresponding patch type, where each pair of patches is either connected or not connected based on a user-specified distance criterion. CONNECT equals zero when either the focal class consists of a single patch or none of the patches of the focal class are “connected.” It equals 100 when every patch of the focal class is “connected (McGarigal and Marks 2003).

The metrics were computed using FRAGSTATS 3.3 at both the class and landscape level (McGarigal and Marks 2003).

Landscape change matrix

In order to evaluate landscape changes between 1963 and 2005, GIS Analysis Matrix function was performed using thematic maps with 10-m resolution by ERDAS. A “from–to” landscape change matrix was extracted which shows the list of landscape classes in 1963 in rows and the list of landscape classes in 2005 in columns. It describes the extent (in hectares) of land use/land cover changes between 1963 and 2005 (Table 5).

Results and discussion

Landscape characteristics in 1963

In 1963, the Mediterranean shrubland, including maquis–phrygana, phrygana, and maquis vegetation, covered two thirds of the study area. The maquis–phrygana covered 35% of the entire area. While maquis–phrygana was mostly spread out in the northwest, west, and south in the form of larger patches, it was scattered across the remaining parts of the study area in smaller patches (Fig. 2). Phrygana, which mostly occurred within maquis–phrygana and maquis, occupied 21% of the Urla district. The largest phrygana patches were distributed in a north–south direction in the western part of the study area. The maquis distribution was larger and continuous in the south and the north than in the east and the northwest. It had the lowest proportion with 13% within the entire Mediterranean shrubland. Based on the highest NP and low LPI values, it was also the most fragmented and disconnected land cover type in the landscape. LPI value of maquis was nearly one third of the maquis–phrygana value. The high LSI value also revealed that maquis patches had the second most complex shapes among the Mediterranean shrubland. The connectivity of maquis vegetation patches was two times lower than maquis–phrygana (Table 3).

Fig. 2
figure 2

Land use/land cover maps of 1963 and 2005

Full size image
Table 3 Class level metrics values of the Urla district
Full size table

Agricultural land covered 16% of the entire landscape. The larger agricultural lands were situated in the lowlands and on the hillsides that run in a northeast-to-southwest direction. The smaller patches were scattered among the northern area (Güvendik, Balıklıova), the western area (Gülbahçe, Barbaros), and the eastern area (Ovacık, Bademler). The low NP and LSI, and the high LPI indicated that there was low fragmentation between agricultural land patches. Connectivity was the highest in the agricultural land in 1963 (Table 3).

Coniferous forest constituted 8% of the study region (Table 3). Large and small forest patches were scattered over the south slopes of the hills around Demircili and Yağcılar. There were medium-sized patches to the north of Güvendik and to the east of Gölcük (Fig. 2). Table 3 shows that the coniferous forest patches were well connected compared to the other natural land cover types (maquis, maquis–phrygana, and phrygana) in the region.

Results showed that olive plantations covered only 3% of the landscape and were mostly located outside of the large agricultural land patch of Urla in the form of small- and medium-sized patches (Table 3 and Fig. 2). Additionally, some olive plantations were located to the east around Bademler, and to the north in the vicinity of Balıklıova and Maksut. NP and LPI showed that the olive grove patches were smaller and disaggregated.

The built-up areas with only 1% of the landscape were located in the central and northern parts of the study area. There were some districts that contained second houses/vacation homes along the coastal areas such as Zeytinalanı, İskele, Malkaça, Gülbahçe, and Balıklıova to the east and west (Fig. 2). Results of landscape metrics indicated that built-up areas in 1963 were scattered across the study region in the form of small patches with very low connectivity scores. With regard to the patch shape of the maquis–phrygana, olive plantation, agricultural areas, and built-up areas had similar average values as shown in the perimeter area ratio index (PAFRAC), while barren and forest land had slightly lower values than maquis and lake areas (Table 3). In conclusion, the landscape of the study area in 1963 can be described as predominantly rural (natural and agricultural), and the existing urban settlements could easily be ignored because of their significantly smaller size than the other land cover types.

Landscape characteristics in 2005

The landscape configurations in Urla changed significantly by 2005. The total area of agricultural, maquis–phrygana, and forest areas decreased, while the built-up olive plantation and phrygana areas increased (Table 3).

Table 3 describes the significant change of AREA_MN between 1963 and 2005. The number of patches (NP) of maquis–phrygana increased from 236 to 313, while the PLAND declined from 35 to 32. Significantly, the AREA_MN value dropped from just over 66 to 45 ha. These numbers indicate that the larger maquis–phrygana patches became smaller and the connectivity declined.

Coniferous forest decreased almost 50,000 ha (Table 5). It was mainly replaced with 22,111 ha of maquis, 18,003 ha of maquis–phrygana, 2,768 ha of olive plantation, 2,049 ha of built-up areas, and 2,029 ha of afforestation. Results in Table 3 show that NP increased, PD and LPI remained the same, and LSI increased. The coniferous forest patches were fragmented, while their shapes became more complicated. The connectivity between existing patches significantly declined.

NP of agricultural land increased from 89 to 115, but the total area of agricultural land decreased to 9% of the entire landscape. LPI decreased from 8.3 to 2.0. AREA_MN decreased dramatically by more than 50%. This meant that agricultural land patches became more divided. However, surprisingly, the agricultural patches seemed well connected in 2005 by the CONNECT values (Table 3). These contradicting results can be explained with the alterations of the configuration of the agricultural land patches in 2005. Namely, there was a large agricultural patch in 1963 as well as several smaller patches around the town of Urla. Over the years, while this large patch was fragmented into pieces, most of the smaller agricultural patches near the neighborhoods of Bademler, Güvendik, Gülbahçe, and Balıklıova completely disappeared. Since the fragments of the previous large agricultural patch remained spatially close to each other, the connectivity between these patches was found to be higher than it was in 1963. However, this higher connectivity in 2005 may be misleading because it gives the impression that no significant fragmentation occurred by 2005.

The olive plantation almost doubled due to new plantations in the region (Table 3, 4, and 5). Some agricultural lands such as tobacco fields were converted to olive plantation mostly around the town of Urla. Additionally, the olive plantations near the rural settlements in the east (Ovacık and Bademler), the west (Barbaros), and north (Balıklıova) expanded (Fig. 2). As the olive plantation patches spread throughout the study area, the connectivity of the patches naturally decreased.

Table 4 Landscape level metrics values of the Urla district
Full size table
Table 5 From–to land use/land cover change matrix, 1963–2005 (hectares)
Full size table

The largest changes were found in agricultural land use. The total amount of agricultural land use and olive plantation was almost 20% of the entire study area in 1963, but 42 years later, this amount declined to 16%. During this time, the fertile agricultural lands were converted into other land uses (Table 5).

Built-up areas constituted 7% of the landscape in 2005 (Table 3). The town of Urla and the small villages in its vicinity expanded into agricultural land, phrygana, and maquis, but the most remarkable changes occurred on the coastal zones of the Zeytinalanı, İskele, and Güvendik districts on the northeast coast of the study area and the Maksat, Malkaça, Gülbahçe, and Balıklıova districts along the coast of Gülbahçe Bay (Fig. 2).

Table 3 shows that the mean patch size increased from 6.8 to 39.2, while LPI increased almost ten times. These values illustrate the increase of built-up areas in the Urla district. The small settlements across the study area significantly increased their peripheries that resulted in larger urban areas. The connectivity value of built-up patches doubled. This can be explained by the land use changes and urbanization in the region. Likewise, Sönmez (2009) stated that “the number of building constructions have increased dramatically since the 1980s.” Regarding the qualitative classification of urban areas by Bierwagen (2005), the process of (sub)urbanization in the study area can be defined as a path evolving from a rural, monocentric urban typology to a more suburban, polycentric morphology.

Fruit trees, which were mostly citrus trees, were planted in the vicinity of the Güzelbahce and Bademler districts. They were located in groups only in this particular part of the study area. This explains why the connectivity value of these patches of 10.9 was the highest in these districts (Table 3).

The PLAND value of phrygana cover increased from 21.26% to 22.57%. New patches emerged especially in the northwest due to transformations from maquis–phrygana. A similar transformation was also seen in the areas where the highway cut through the agricultural and natural lands. As a result, new phrygana patches appeared along the highway because of man-made regressive succession driven by construction of the highway. Increasing NP, PD, and LSI and decreasing LPI and AREA_MN reflected the increasing fragmentation. Connectivity was almost the same as it was in 1963 (Table 3).

Table 3 reveals that maquis declined slightly to 13.2%. But the increasing NP and LSI and the decreasing PLAND, PD, LPI, and AREA_MN indicate that the maquis became more fragmented. The continuous patches in the southern part of the study area were divided into smaller patches over time that resulted in lower connectivity.

The values of PAFRAC in maquis–phrygana, agricultural land, phrygana, and maquis declined slightly, while the values in olive plantation, built-up, and lake decreased significantly in 2005 (Table 3). A decrease in the perimeter area fractal dimension indicates that the patches have less complex shapes.

New mineral extraction sites replaced some of the maquis–phrygana and phrygana areas over this period. Because these new sites were located close to each other, the CONNECT values increased (Tables 3 and 4).

It is important to note that the connectivity of the natural landscape (forest, maquis, and maquis–phrygana) and olive plantation declined by 2005 (Table 3). This is consistent with the emergence of suburban settlements at the peripheries of the Urla district. This led to the appearance of more fragmented and isolated natural land cover patches that were mostly cut off by built-up areas and roadway networks including the highway. Some parts in the study area such as Zeytinalani, İskele, and Güvendik were completely transformed into suburban settlements with low-density houses scattered across the landscapes (Fig. 2).

Results of the landscape level metrics suggest that the landscape composition of Urla was turned into a very patchy and scattered configuration over the years. Increasing PN, PD, LSI, and decreasing LPI and AREA_MN indicate that the landscape of Urla became more fragmented. In this study, the perimeter area fractal dimension index slightly decreased in 2005. This also suggested that large patches were divided into smaller ones with simple forms. Consequently, the existing landscape was transformed into a very patchy and scattered configuration accompanied by slightly declining connectivity in the entire landscape.

In the metropolitan context, built-up areas expanded primarily along the coastline of Izmir Bay, parallel to the development of the transportation network. The period from 1963 to 2005 was an important era for the urbanization of the Izmir metropolitan area because there was a dramatic decline in some land cover types such as agricultural areas. This was accompanied by a threefold increase in the urban fabric (from 8.18% to 28.18%) (Hepcan et al. 2011). It is possible to follow a similar path in the present study area where more than sixfold increase in the built-up areas (from 1.16% to 7.24%) was observed. In the meantime, the population of Urla more than doubled with an increase from 20,625 to 52,500 people between 1965 and 2005 (TurkStat 2011). Bierwagen (2007) stated that the size of the urban area influences the connectivity. For instance, while small urban areas spreading through disaggregated habitats may have a limited impact on connectivity, large urban areas predominantly decrease connectivity. In the present study, even though Urla expanded its size more than six times during a 42-year period, it did not witness a dramatic decrease in the connectivity of the entire landscape. However, the result was quite the opposite on a class level evaluation in that the connectivity of natural land cover such as forest and shrubland significantly declined by 2005 (Table 4).

Conclusions

Using landscape metrics, this study was undertaken to explore how urban development, including recent suburbanization, has affected landscape pattern and connectivity in the Urla district. The Urla district significantly expanded over its suburban area and took over agricultural lands and natural areas at its peripheries between 1963 and 2005. The landscape configuration became significantly fragmented and the connectivity of the natural landscape declined. There was more than sixfold increase in the built-up areas during the study period.

To prevent further fragmentation, it is critical to keep the existing natural land cover types and agricultural areas intact. It is possible to take advantage of the natural sites that already exist in the region. The natural site is a legal nature protection category in Turkey. Existence of the natural sites in the region somewhat affected trajectory of (sub)urbanization in Urla by delimiting further encroachments of the built-up areas over natural habitats. However, this protection category has always been controversial because of its weaknesses in terms of implementation and low-level acceptance by the local people. In this way, the revised version of natural sites could really contribute to the protection of native vegetation in the region.

More importantly, as stated by Botequilha Leitao et al. (2006), a sustainable development scenario towards conservation and rehabilitation of the intact and damaged natural resources for the study area is required. This concept includes creating a green infrastructure or an ecological network plan (i.e., core areas and ecological corridors) that is a key planning approach to support the ecological functioning of the natural landscape in the long term. For instance, in the Netherlands, ecological networks are as well a pro-active strategy and the core of national conservation strategy. As one of the worlds’s densely populated countries, this is required due to the intensive interaction with nature (Jongman et al. 2004). The ultimate goal of the National Ecological Network is to create a network of natural habitats of sufficient ecological quality. When it is fully completed in 2018, more than half of it will consist of large areas of connected nature areas (greater than 2,000 ha in size) (PBL 2005). Another example to support the ecological functioning of the landscapes can be given from the continental level. Pan-European ecological networks aim at conserving the full range of ecosystems, habitats, species, and landscapes of European importance (Jongman et al. 2011).

It should be emphasized that there is already a defined core area that covers almost 45% of the entire study area. This core area is dominated by the natural areas such as maquis–phrygana, maquis, and forest vegetation to the north and south of the highway and west of the Gülbahçe Bay and in the vicinity of the Malkaca area and the Kuscular village (Hepcan and Özkan 2011). In addition to the existence of a core area, two Key Biodiversity Areas (KBAs) (Eken et al. 2006) are located on the northern and southern parts of the highway within the boundaries of the study area. Both the core area and KBAs can be used as the starting point for future green infrastructure or ecological network planning in the region.

In ecological terms, as a countermeasure to fragmentation of natural habitats in the urbanizing areas, Forman (1995) emphasized maintaining large patches of native vegetation as well as providing connectivity between important resource patches. The higher priority lies in developing and/or rehabilitating connectivity between the components of the natural landscape in the study area. Establishing connectivity is an integral component of a green infrastructure and ecological network planning, and may be dealt with in different ways such as using indicator species. Moreover, this planning approach must be a multipurpose one that considers both cultural and natural aspects of the region. But it is important to exercise caution in the planning process and not compromise sustainability.

It is recommended that alternative planning scenarios are prepared for the region using landscape metrics. Plan alternatives provide flexibility to land planners and managers, thus allowing local people with different priorities to examine alternatives.

Agricultural patches for food production and recreation should be protected and integrated into the green infrastructure or ecological network plan. These patches can also serve as stepping stones for recreational activities (Botequilha Leitao et al. 2006). In the case of Urla, further fragmentation of and decrease in the existing agricultural patches should be avoided using the framework of the sustainable development strategy. In recent years, there has been a growing interest for agricultural production such as organic farming in the Urla district. This recent trend may help to avoid a further decrease in the existing agricultural patches by making “these areas competitive with the urbanized land” (Catalan et al. 2008).

Finally, the success of preventing further fragmentation and restoring connectivity can be achieved using green infrastructure or ecological network planning. The planning should be prepared with the consensus of local people and land managers. Otherwise, as with the experience in the Izmir metropolitan area in the past, the conventional spatial planning approach will not be effective in managing changes in the region in a sustainable way because “it lacks an ecological vision” (Hepcan et al. 2011).