The ‘Sprawl Divide’: Comparing models of urban dispersion in mono-centric and polycentric Mediterranean cities

Abstract

While recent studies have illustrated how European cities are shifting towards exurban development, the influence of urban form on sprawl models has been poorly investigated at both the regional and local scales. The present study introduces an original sprawl index and analyses its spatio-temporal dynamics between 1960 and 2010 in two Mediterranean regions (Athens and Rome) characterized by contrasting urban morphology. Different models of urban dispersion have been identified during the investigated period by correlating variables such as population density, land-use and territorial characteristics to the sprawl index. This index significantly diverged in the two regions, suggesting that recent urbanization patterns are influenced by the city’s form. The results contribute to the debate on sustainable urban form for expanding cities in Europe.

Keywords

Athens mono-centric model Rome sprawl index urban form

Introduction

Patterns and processes of urban expansion are extremely fascinating but, at the same time, complex and difficult to analyse. Although urban development involves multiple territorial and socioeconomic factors, compactness and dispersion were and are the two main schemes with which cities have evolved. For these reasons, themes such as urban morphology, exurban development, changes in the use of land and land consumption have and are attracting the interest of many researchers from all over the world, having the objective to define the nature, dynamics and consequences that the phenomenon of low-density urban expansion is having on capital cities at vastly different spatial scales (Bruegmann, 2005; Gill et al., 2008; Johnson and Lewis, 2007; Kahn, 2000; Richardson and Chang-Hee, 2004).

Urban sprawl has been defined as ‘a pattern of urban and metropolitan growth that reflects low-density, automobile-dependent, exclusionary new development on the fringe of settled areas often surrounding a deteriorating city’ (Squires, 2002). While in the last century sprawl has been a normal trend in developed countries, including the United States, Canada and north-western Europe (Glaster et al., 2001), it represents a relatively novel phenomenon in southern Europe (Antrop, 2004; Gargiulo Morelli and Salvati, 2010; Kasanko et al., 2006). While Mediterranean cities have been considered, for a long time, examples of mono-centric regions, new spatial organization models, sometimes oriented towards polycentricism, have emerged since the 1990s due to diffused economic growth and the consequent population relocation in the neighbouring rural areas (Longhi and Musolesi, 2007; Schneider and Woodcock, 2008; Turok and Mykhnenko, 2007).

One of the most widely recognized traits of Mediterranean cities is compactness. Jenks et al. (1996) examined the claim that the compact city is a sustainable urban form. A great deal of complexity was found and the study did not conclude with a ringing endorsement of the compact city model, at least as conceived in the context of developed countries. Questions were raised about the extent to which urban form could achieve sustainability (Breheny, 1997; Burton, 2000; Ewing, 1997; Hofstad, 2012; Soliman, 2004). There were benefits in relation to the saving of agricultural, forest and other valuable land, but there were also problems concerning environmental quality and local acceptance of a more compact form of urban living. Williams (2000) conclude that there is no single sustainable morphology, but rather a variety of urban forms that are more sustainable than recent development patterns. These depended crucially on the characteristics of an area and the local and strategic pathways chosen for sustainability.

Understanding the impact of urban form on the recent sprawl processes therefore requires sophisticated decision-making processes that are multidimensional and adaptive (Burton, 2002). A number of indicators have been proposed to monitor sprawl in peri-urban areas (Arribas-Bel et al., 2011; Hasse and Lathrop, 2003; Ioannidis et al., 2009; Sutton et al., 2010; Weber et al., 2005). A common way to measure the extent of sprawl in a metropolitan area is the ratio of two growth rates: the rate at which land development near the outer suburbs has increased divided by the rate at which the population living in the metropolitan area has grown (Couch et al., 2007). Measuring sprawl in this way, a recent study notes that ‘if land is being consumed at a faster rate than population growth, then a metropolitan area can be characterised as “sprawling”. If population is growing more rapidly than land is being consumed for urbanization, then a metropolitan area can be characterised as “densifying”’ (Burchell et al., 2005). A large imbalance between a place’s spatial expansion and its population change (where the former increases much more rapidly than the latter) is not unusual in large urban regions. Growth of this sort produces a lower density outcome, with people and their residential and commercial buildings taking up more space at the expense of agricultural and natural land (Salvati et al., 2013a).

This article illustrates a procedure based on indicators to analyse sprawl and long-term landscape transformations observed in large urban regions with different morphologies. The study tests if the form of cities influences selected socioeconomic features of sprawled urban development. By identifying long-term expansion paths using population and land-use data, the procedure was applied to two expanding Mediterranean cities (Athens and Rome) respectively taken as examples of compact mono-centric and partly polycentric, fragmented cities undergoing spatial restructuring towards a more diffused spatial organization of residential settlements and economic activities. Previous studies focused on urban expansion in both Rome (e.g. Munafò et al., 2010; Salvati, 2013; Salvati and Sabbi, 2011) and Athens (e.g. Arapoglou and Sayas, 2009; Chorianopoulos et al., 2010; Salvati et al., 2013a, 2013b; Sayas, 2006). As an original contribution to urban studies, the present work specifically explores the role of topography, land-use and demographic factors characterizing the sprawl model observed in the two cities and questions if recent low-density expansion has been influenced by the initial urban form.

Methods

The investigated regions

Two southern European urban regions are considered in this paper: Rome in Italy and Athens in Greece. For each city, the investigated area corresponds to the prefecture administered by that city. The selected areas encompass (or are a little larger than) the boundaries of the related ‘Urban Atlas’ (UA) region (see European Environment Agency (EEA), 2010). The rationale of this choice was to have an area from which a significant share of the residents commute into the city. Table 1 reports selected characteristics of the two regions. In order to collect comparable spatial data, the same elementary domain was used, that is, municipalities corresponding to the LAU-2 level of the European Nomenclature for Territorial Statistics. This domain allows for a detailed analysis of urban trends covering a relatively long time period and based on official statistics and cartographic analysis (Salvati and Sabbi, 2011).

Table 1.

Selected characteristics of the Rome and Athens regions.

Variable	Rome	Athens
Urban morphology	Dispersed and moderately polycentric form	Compact and mono-centric form
Settlement characteristics	Medium-density settlements mixed with low- and medium-low-density settlements	Very dense and moderately dense settlements intermixed
Demographic dynamics	Weak population decline after a continuous increase (1950–1980)	Stability after considerable population growth (1950–1990)

The investigated area in Greece covers a large part of the Nuts-2 region of Attica. A total of 114 local municipalities (of which 58 belong to the Athens’ urban area and extend for 430 km²) have been considered in this study involving a surface area covering more than 3000 km² (Figure 1). The city-region mostly consists of mountains bordering the urban conurbation of Athens, which occupies a relatively flat area (Salvati et al., 2013a). The Italian study area covers the province of Rome (121 municipalities with a surface area of 5355 km²). The region is characterized by a complex topography consisting of 30% lowlands, 50% uplands and 20% mountains. The municipality of Rome is the largest one (1285 km²) and it was further divided into 115 urban districts with a size comparable to that of the remaining municipalities of the prefecture and the municipalities of the other study area (Athens). Urban districts follow an official land classification and topographic system developed by the municipal government of Rome and are administered by local councils. The investigated area was therefore partitioned into a total of 235 spatial units.

Figure 1.

Geographical location of Athens and Rome and the administrative boundaries of the related urban regions (left: Athens; right: Rome).

Land-use data

The surface area of three basic land-use classes (urban areas, cropland and forests) was quantified at three dates (1960, 1990 and 2010 in Athens; 1960, 1990 and 2006 in Rome) at the spatial scale described before (Salvati and Sabbi, 2011). This spatial resolution allows a relatively detailed analysis of land-use changes for the whole time period considered (Salvati et al., 2013a). In Athens, data were obtained from (i) the land-use census undertaken by the Greek National Statistical Office (EYSE) (available years: 1960, 1970, 1980, 1990; ESYE, various years), (ii) the LaCoast (LC) project database (available year: 1975), (iii) the Corine Land Cover (CLC) cartography (available years: 1990 and 2000; EEA, 2006) and, finally, (iv) the UA map (available year: 2010; EEA, 2010). In Rome, data were obtained from elaboration of three compatible digital land cover maps: the CLC-like ‘Land Cover Map of Italy’ (scale 1:200,000) produced by the National Research Council in the early 1960s, and two CLC land cover maps (scale 1:100,000) respectively dated 1990 and 2006. Since the land cover classification used in the three maps is comparable, land categories were reclassified into homogeneous classes (Salvati et al., 2013a): croplands including pastures and fallow land (AGR), forested land (FOR) and urban areas (URB). The remaining land, including water bodies, was not considered in the analysis. The procedure carried out to produce comparable data across the analysis period was as follows. In Athens, seven classes were surveyed by the ESYE census: agricultural areas, private pastures, common pastures, forests, inland water, urban areas and other unproductive lands. In this study, the two categories of pastures were analysed together. Moreover, since ‘urban areas’ and ‘other unproductive lands’ were recorded in 1960 into a unique class, we estimated separately the surface area of the two classes by multiplying the 1960 unique value (urban + unproductive) by a weight corresponding to the ratio between urban (or unproductive) areas and the total municipal area as observed in 1970 (Salvati et al., 2013a). Finally, agricultural and pasture areas were combined, obtaining the four land-use classes previously described.

Notably, although the number of municipalities was relatively stable during the investigated period, eight small municipalities were formed during 1950–1970. The value of the considered variables (surface area by each land-use class) for each newly formed municipality was imputed through the average of the three neighbouring municipalities. Finally, the land-use classes available in the LC and CLC maps in 1975, 1990 and 2000 and the UA map in 2010 were reclassified in order to make them comparable to the four basic ESYE classes. The first level of the CORINE nomenclature system (classes 1, 2 and 3) was used for both LC and CLC maps. Land-use data collected in 1975 (LC)–1980 (ESYE) and 1990 (ESYE and CLC) and elaborated at the municipal scale were used to test the readability of these two data sources. A non-parametric statistical test (Mann–Whitney U-test) was carried out to compare the distribution of the four land-use classes in the investigated municipalities. Results of the pair-wise test indicated that the surface area of each class is comparable between LC (1975) and ESYE (1980) land-use census, as well as between CLC (1990) and ESYE (1990) land-use census (Mann–Whitney U-test, p > 0.05 in both comparisons).

In Rome, the surface area of the selected land-use classes was calculated in each spatial domain on the basis of a spatial intersect between each land-use map and a shapefile containing the boundaries of the municipalities. The CLC project (coordinated by the EEA) aimed to produce diachronic land-use and land cover maps based on satellite images. The choice of scale (1:100,000), minimum mapping unit (25 ha) and minimum width of linear elements (100 m) for CLC mapping represents a trade-off between production costs and land cover information details (Salvati and Sabbi, 2011). The first level of the CORINE nomenclature system (classes 1, 2 and 3) was used for CLC maps. The soundness of the land-use figures derived from these maps was checked by comparison with independent data obtained from the National Census of Agriculture (various years) and the basic land-use map of Italy prepared by the Istituto Geografico Militare in 1949. In both the analysed regions it should be noted that the spatial resolution of land-use data allows for a trend analysis at the municipal scale. Specific patterns in settlement distribution over time (especially out-of-plan settlements and detached, single buildings) should be in-depth investigated using additional, high-resolution data sources.

Population data

Population data were made available in 1961, 1991 and 2011 (2009 in Rome) at the municipal and urban district scale. The primary data source was the General Census of Population carried out by the national statistical authority (ISTAT in Italy and ESYE in Greece). The digital cartography depicting the boundaries of municipalities and urban districts in the two investigated areas (Figure 1) was obtained from ISTAT and ESYE respectively for Rome and Athens at 1:50,000 scale. To derive population density, the surface area of each spatial unit was calculated.

Statistical analysis

Data were analysed in order to quantify the available land per capita by three classes (URB, AGR and FOR) for each of the three investigated years. The surface area of each class was calculated and assigned to each territorial unit on the basis of a spatial intersect (Salvati and Sabbi, 2011). Based on the change over time in per capita urban land at the municipal scale, a standardized sprawl index (SI) was developed separately for two sub-periods (1960–1990 and 1990–2010) by dividing the percent change of per capita built-up areas (_pURB) in each municipality by the standard deviation of the spatial series measured in the same observation time period:

S I = \frac{(_{p} U R B_{t + 1} -_{p} U R B_{t}) / (_{p} U R B_{t}) * 100}{s t . d e v [(_{p} U R B_{t + 1} -_{p} U R B_{t}) / (_{p} U R B_{t}) * 100]}

(1)

In this way, the SI is calculated as a standardized variable with zero average. A SI zero score indicates no trends in compactness or sprawl over time. Each municipality is classified into one of seven classes based on the deviation (s) of the SI from 0: (1) markedly densifying municipalities (s > 2); (2) densifying municipalities (1 < s < 2); (3) moderately densifying municipalities (0.1 < s < 1); (4) stable municipalities (–0.1 < s < 0.1); (5) moderately sprawling municipalities (–1 < s < –0.1); (6) sprawling municipalities (–2 < s < –1); and (7) markedly sprawling municipalities (s < –2). The index quantifies the increase in per capita land take as a proxy of sprawled urban development. Conversely, a reduction in the values of the indicator illustrates a process of urban densification. The procedure modifies and enriches the formulation of the SIs illustrated by Terzi and Bolen (2009) and Jaegera et al. (2010). Due to the standardization procedure, the index proposed here allows comparisons among different areas and territorial contexts that appears as an important tool in urban studies (e.g. Kasanko et al., 2006). The relationship observed at the municipal scale in the per capita SI observed during 1960–1990 and 1990–2010 was evaluated using a Pearson correlation coefficient and a best-fit linear equation. Correlation significance has been tested at p < 0.05.

The ‘sprawl trend’ in each investigated spatial unit was determined by analysing jointly the value of the s parameter in the two sub-periods; nine categories were identified on the basis of the seven classes described above (Table 2): (i) stable trend over the whole period (–); (ii) stability and then densification (-C); (iii) densification and then stability (C-); (iv) densification over the whole time period (CC); (v) densification and sprawl (CS); (vi) stability and sprawl (-S); (vii) sprawl and stability (S-); (viii) sprawl and densification (SC); and (ix) sprawl over the whole period (SS). Classes (1)–(3) and (5)–(7) described above were pooled together respectively in two categories (densification (C): 1–3 and sprawling (S): 5–7). Based on this classification, 11 supplementary variables (listed in Table 3) have been used to profile the investigated municipalities according to their topography, land-use and demographic patterns.

Table 2.

The classification of sprawl trends in each investigated spatial unit.

Trend in 1960–1990	Trend in 1990–2010	General trend (1960–2010)	Class
Stable (–0.1 < s < 0.1)	Stable (–0.1 < s < 0.1)	Stable trend over the whole period	–
Densifying (s > 0.1)	Densifying (s > 0.1)	Densification over the whole period	CC
Stable (–0.1 < s < 0.1)	Densifying (s > 0.1)	Stability and densification	-C
Densifying (s > 0.1)	Stable (–0.1 < s < 0.1)	Densification and stability	C-
Densifying (s > 0.1)	Sprawling (s < –0.1)	Densification and sprawl	CS
Sprawling (s < –0.1)	Densifying (s > 0.1)	Sprawl and densification	SC
Sprawling (s < –0.1)	Stable (–0.1 < s < 0.1)	Sprawl and stability	S-
Stable (–0.1 < s < 0.1)	Sprawling (s < –0.1)	Stability and sprawl	-S
Sprawling (s < –0.1)	Sprawling (s < –0.1)	Sprawl over the whole period	SS

–: stable trend over the whole period; CC: densification over the whole time period; -C: stability and then densification; C-: densification and then stability; CS: densification and sprawl; SC: sprawl and densification; S-: sprawl and stability; -S: stability and sprawl; SS: sprawl over the whole period.

Table 3.

The supplementary variables considered in this study by theme.

Variable	Methodology	Statistical source^a
Land-use
Urban area (%)	Spatial analysis using GIS tools (see the Land-use data section)	Various (see the Land-use data section)
Agricultural areas including pastures (%)
Forests (%)
Topographic and territorial variables
Resident population	Extracted from population Census data	National Statistical Office
Municipal surface area (km²)	Elaboration on a municipal boundary map	Cartographical Service of the National Statistical Office
Distance from the inner city (km)	Elaboration on a municipal boundary map
Proximity to the sea	Dummy for coastal municipalities (= 1; the remaining municipalities = 0)
Lowlands (%)	Elaboration on a Digital Elevation Model raster file and a municipal boundary map	Cartographical Service of the National Statistical Office and ASTER DEM
Uplands (%)
Population
Population distribution (%)	Elaboration on Population Census data	National Statistical Office
Population density (inhabitants/km²)	Elaboration on Population Census data	National Statistical Office

Further information and freely available digital maps can be downloaded from the websites of the national statistical authorities in Greece (www.statistics.gr) and Italy (www.istat.it).

Results

Trends in land take in Athens and Rome

Based on population dynamics and land-use changes, Figure 2 shows the distribution of per capita urban land in the two investigated cities. Although increasing in both cities, per capita land take1 was more rapid in Rome during 1960–1990 (from 64 to 168 m² of impervious land per inhabitant) than in the subsequent period (from 168 to 169 m² per inhabitant), while in Athens per capita land take increased at a relatively stable rate in both periods (from 196 to 207 m² and to 223 m² of impervious land per inhabitant respectively in 1960–1990 and 1990–2010).

Figure 2.

Per capita land take (m²) in the two regions by year.

Classification of local municipalities based on the sprawl index

The proposed SI classified local municipalities in Rome and Athens in a different way during the two periods examined here (Table 4). Sprawling municipalities were more common in Rome (78%) than in Athens (51%) during 1960–1990, while the opposite pattern was observed in 1990–2010 with more than half of the investigated municipalities (55%) being classified as sprawling in Athens and only 13% in Rome. Interestingly, markedly sprawling municipalities increased in number especially in Athens (+22%) during the most recent period, while municipalities characterized by a stable trend in the SI over time were more common in Rome. Municipalities experiencing urban densification were only 4% in Rome during 1960–1990, but increased to 56% in the subsequent period. In Athens the proportion was quite stable over time: 35% in 1960–1990 and 37% in 1990–2010. This reflects the compact tradition characterizing the urban development of the Greek capital.

Table 4.

Percent distribution of the investigated municipalities in Athens (n = 114) and Rome (n = 235) according to per capita sprawl index classification by time period (see the Statistical analysis section for further details).

Class	Athens			Rome
Class	1960–1990	1990–2010	% change	1960–1990	1990–2006	% change
Markedly densifying	7.0	14.8	7.8	0.4	3.6	3.1
Densifying	6.6	10.4	3.7	0.3	8.6	8.3
Moderately densifying	21.1	12.1	−9.1	2.7	42.9	40.2
Stable	14.1	7.9	−6.1	18.7	32.2	13.4
Moderately sprawling	32.2	11.9	−20.2	60.4	10.7	−49.7
Sprawling	7.2	8.7	1.5	11.9	0.4	−11.5
Markedly sprawling	11.8	34.2	22.4	5.5	1.6	−3.8

The spatial distribution of the SI diverged in the two time periods in both cities (Figure 3): in Athens sprawling municipalities were concentrated in the eastern part of the urban agglomeration during 1960–1990, while they shifted to the north and west directions in the subsequent period. Municipalities experiencing a densification process moved from the fringe around the compact urban fabric (1960–1990) to the eastern part of the region (1990–2010). Sprawling municipalities in Rome concentrated in the urban fringe south of the city in 1960–1990 while moving at larger distances from the inner city in 1990–2010. Municipalities experiencing densification show the reverse pattern, suggesting that sprawl and densification processes have followed during the last 50 years. Taken together, densification follows sprawling processes, especially in the western part of Rome’s province and in the eastern part of Athens’ metropolitan region.

Figure 3.

Spatial trend in per capita sprawl index by time period and city (left: Athens; right: Rome).

The distribution of municipalities in the seven categories defined by the SI diverged in the two cities in both examined periods (Mann–Whitney U-test, p < 0.05). The relationship observed in the value of the SI in each municipality between the two periods was also different in Athens and Rome (Figure 4): a significant negative trend with a steep slope has been observed in Rome (R² = 0.64), while a moderately negative correlation (R² = 0.25) has been observed in Athens. A non-parametric correlation analysis using a Spearman rank test confirmed these results. Based on this evidence, the next section introduces an in-depth analysis of the profile of different municipalities in both cities.

Figure 4.

The relationship in per capita sprawl index observed at the municipal scale in the two investigated time periods (x-axis: 1960–1990; y-axis: 1990–2010; left: Athens; right: Rome).

Profiling municipalities with sprawl

Figure 5 illustrates the spatial distribution of per capita SI over the whole period under investigation in Athens and Rome. Municipalities around the consolidated urban area (at distances <10 km from the inner city) were mainly classified as compacting and dense. This may confirm the assumption about the mono-centric spatial organization of Athens’ urban region. Sprawling municipalities were concentrated at larger distances from the city centre. In Rome, sprawling municipalities are intermixed with stable or dense municipalities. The most widespread class of the SI was SC; this suggests that settlement dispersion and compaction followed in Rome’s fringe. This class indicates the prevalent development pattern in the whole peri-urban region around Rome over the last 50 years with no evident relation with the distance from the inner city.

Figure 5.

Classification of the investigated municipalities in the two regions according to per capita sprawl index (left: Athens; right: Rome).

Topographic, land-use, population and other territorial variables have been used for identifying the ‘sprawl model’ characterizing each investigated municipality in both cities (Table 5). In Athens, urban municipalities with percentage of impervious land >60% always show stability or densification trends in the SI. Sprawling municipalities are characterized, on average, by a negligible percentage of impervious land in 1960 (2%) and a rapid increase in the subsequent years (8% in 1990, 18% in 2010). To the contrary, a clear differentiation in the SI between strictly urban and peri-urban municipalities based on the percentage of impervious land was not found in Rome. As a matter of fact, both sprawling (SS) and compacting and dense (CC) municipalities show similar proportions of impervious land over time and a moderate increase in urban areas.

Table 5.

Classification of the investigated municipalities in Athens and Rome according to per capita sprawl index by land-use variable. Each class is represented by symbols (–: stable trend over the whole period; CC: densification over the whole time period; -C: stability and then densification; C-: densification and then stability; CS: densification and sprawl; SC: sprawl and densification; S-: sprawl and stability; -S: stability and sprawl; SS: sprawl over the whole period) according to the trends observed in 1960–1990 and 1990–2010 (see the Statistical analysis section for further details).

Class	Urban areas (%)					Agricultural areas (%)					Forests (%)
Class	1960	1990	2010	Δ(60–90)^a	Δ(90–10)^a	1960	1990	2010	Δ(60–90)^a	Δ(90–10)^a	1960	1990	2010	Δ(60–90)^a	Δ(90–10)^a
Athens
–	72.6	86.3	88.0	0.6	0.1	12.1	11.2	10.0	−0.2	−0.5	8.1	5.7	1.9	−1.0	−3.3
-C	44.3	79.1	65.3	2.6	−0.9	36.9	22.9	29.8	−1.3	1.5	14.3	4.8	4.0	−2.2	−0.8
C-	64.9	79.2	82.4	0.7	0.2	10.8	12.2	12.9	0.4	0.3	24.3	18.1	4.5	−0.8	−3.8
CC	45.7	68.8	64.1	1.7	−0.3	32.5	24.2	32.6	−0.8	1.7	20.4	16.0	3.2	−0.7	−4.0
CS	20.0	22.1	37.5	0.3	3.5	45.1	50.9	56.1	0.4	0.5	33.6	35.2	6.4	0.2	−4.1
-S	6.0	10.3	26.9	2.3	8.1	57.8	62.1	66.9	0.3	0.4	26.2	29.0	5.9	0.4	−4.0
S-	33.7	83.0	98.1	4.9	0.9	63.0	27.0	1.6	−1.9	−4.7	0.0	0.0	0.0	–	–
SC	2.6	32.1	27.2	37.4	−0.8	61.4	45.5	65.5	−0.9	2.2	33.2	29.6	6.9	−0.4	−3.8
SS	1.9	8.3	17.9	11.5	5.8	61.5	68.4	72.4	0.4	0.3	22.6	21.9	9.4	−0.1	−2.8
Rome
–	37.6	44.8	46.6	0.6	0.3	51.8	33.4	32.9	−1.2	−0.1	10.6	20.0	20.5	3.0	0.1
C-	17.6	31.1	30.5	2.6	−0.1	69.1	45.3	47.0	−1.1	0.2	11.7	21.4	22.5	2.8	0.3
CC	23.3	26.4	27.9	0.4	0.4	62.7	53.8	55.2	−0.5	0.2	14.0	19.7	16.9	1.4	−0.9
CS	3.2	0.0	3.7	−3.3	3.7	90.1	58.3	53.8	−1.2	−0.5	6.7	41.7	42.5	17.5	0.1
-S	12.8	13.9	16.8	0.3	1.3	71.4	63.5	66.3	−0.4	0.3	11.4	16.8	16.9	1.6	0.0
S-	25.9	43.1	45.1	2.2	0.3	65.6	42.8	41.5	−1.2	−0.2	8.4	13.3	13.4	1.9	0.0
SC	5.6	20.3	21.8	8.9	0.5	79.6	65.1	65.1	−0.6	0.0	13.4	12.9	13.1	−0.1	0.1
SS	14.3	30.1	35.0	3.7	1.0	72.8	53.6	50.5	−0.9	−0.4	12.5	15.0	14.5	0.7	−0.2

Annual percent rate of change between the indicated years.

A comparable pattern was observed for agricultural and forest classes. In Athens, agricultural areas were concentrated in sprawl municipalities in 1960 and decreased rapidly in the following years, while forests declined, especially during the last 20 years, due to clearcutting and recurrent fires with no evident differences between compacting and sprawling municipalities. In Rome, agricultural areas decreased homogeneously in 1960–1990 in both compacting and sprawling municipalities, with a more stable pattern in 1990–2010. Forests were generally stable or moderately increasing over the study period; however, in contrast to Athens, a marked increase in forest land was observed in compacting municipalities compared to sprawling ones.

As far as topographic and territorial variables are concerned (Table 6), the surface area classified as sprawling was larger in Rome than in Athens (84% versus 77%), although the SS class area undergoing continuous sprawl processes over the period was found to be higher in Athens (29%) than in Rome (11%). In both cities, the municipalities experiencing sprawl processes administrate a larger surface area compared to the remaining municipalities; this is especially evident in Athens, where the municipalities belonging to the SS class administrate more than 108 km² each compared to the 26 km² average municipality size in the region. The distance from Athens was larger in sprawling municipalities than in compacting or stable municipalities, while no differences have been recorded in Rome, partly confirming the findings obtained from the land-use distribution analysis.

Table 6.

Classification of the investigated municipalities in Athens and Rome according to per capita sprawl index by topographic and territorial variables.

Class	Surface area (%)	Average municipal area (km²)	Average population				Distance from the inner city (km)	Coastal municipalities (%)	Proportion of land (%)
Class	Surface area (%)	Average municipal area (km²)	1961	1991	2011	n	Distance from the inner city (km)	Coastal municipalities (%)	Lowlands	Uplands
Athens
–	5.3	6.7	63,426	86,643	81,777	24	6.0	16	77	23
-C	3.6	27.0	11,173	19,442	23,655	4	14.0	90	90	10
C-	2.3	6.3	15,813	50,165	53,591	11	6.7	7	26	74
CC	11.5	13.9	4175	15,078	22,270	25	14.6	27	63	37
CS	21.0	42.3	2291	6568	8938	15	23.1	19	17	79
-S	5.2	26.2	4337	6892	8301	6	30.1	63	63	37
S-	0.3	5.0	9578	17,635	22,212	2	7.5	0	100	0
SC	22.2	35.3	2153	7257	11,870	19	25.9	36	50	33
SS	28.7	108.3	7231	11,397	14,925	8	29.9	72	35	37
Rome
–	13.0	13.5	27,576	27,818	24,034	52	22.9	0.1	1.9	7.7
C-	1.8	11.9	4126	13,873	13,068	8	29.4	0.1	0.4	0.7
CC	1.6	14.1	2062	8802	16,157	6	22.3	0.6	1.0	0.6
CS	0.2	8.2	719	819	958	1	38.6	0.0	0.0	0.2
-S	5.7	34.0	7846	9026	11,340	9	28.2	2.5	3.5	1.9
S-	17.4	16.7	10,278	15,753	16,793	56	20.0	0.0	3.9	10.4
SC	49.6	32.1	5095	10,543	15,890	83	23.7	9.9	24.2	22.1
SS	10.9	29.2	11,374	15,755	14,824	20	24.0	1.3	4.0	6.0

Closeness to the sea seems to be a variable not directly influencing the sprawl trend in Athens, while in Rome coastal municipalities experienced sprawl more frequently than internal areas. Taken as a proxy of the availability of land to edification, the proportion of flat and hilly surface area did not characterize sprawling from compacting municipalities in Athens, while in Rome flat areas were found to be more common in sprawling than in compacting municipalities.

As far as population distribution is concerned (Table 7), 91% of Athens’ population resided in compacting municipalities in 1961 and 85% in 2011, while the percentage was lower and considerably decreasing over time in Rome (53% in 1961 and only 35% in 2011). Municipalities with population density <100 inhabitants/km² in 1961 are mainly classified as sprawling in Athens (apart from the class S-, which shows a relatively high density even in 1961), while sprawling municipalities in Rome are characterized, on average, by higher population density.

Table 7.

Classification of the investigated municipalities in Athens and Rome according to per capita sprawl index by population variables.

Class	Population distribution (%)			Δ(61–91)^a	Δ(91–11)^a	Population density			Δ(61–91)^a	Δ(91–11)^a
Class	1961	1991	2011	Δ(61–91)^a	Δ(91–11)^a	1961	1991	2011^b	Δ(61–91)^a	Δ(91–11)^a
Athens
–	75.2	59.6	52.0	−15.6	−7.6	9502	12,980	12,251	1.2	−0.3
-C	2.2	2.2	2.5	0.0	0.3	414	720	876	2.5	1.1
C-	8.6	15.8	15.6	7.2	−0.2	2492	7905	8445	7.2	0.3
CC	5.2	10.8	14.7	5.6	3.9	300	1082	1598	8.7	2.4
CS	1.7	2.8	3.5	1.1	0.7	54	155	211	6.2	1.8
-S	1.3	1.2	1.3	−0.1	0.1	166	263	317	2.0	1.0
S-	0.9	1.0	1.2	0.1	0.2	1911	3518	4431	2.8	1.3
SC	2.0	4.0	6.0	1.9	2.0	61	206	336	7.9	3.2
SS	2.9	2.6	3.2	−0.2	0.5	67	105	138	1.9	1.5
Rome
–	51.6	38.4	30.4	−13.2	−8.0	2043	2061	1780	0.0	−0.8
C-	1.2	2.9	2.5	1.8	−0.4	348	1169	1101	7.9	−0.3
CC	0.4	1.4	2.4	1.0	1.0	146	625	1148	10.9	4.6
CS	0.0	0.0	0.0	0.0	0.0	87	100	116	0.5	0.9
-S	2.5	2.2	2.5	−0.4	0.3	231	265	333	0.5	1.4
S-	20.7	23.4	22.9	2.7	−0.6	616	944	1007	1.8	0.4
SC	15.2	23.2	32.1	8.0	8.8	159	328	494	3.6	2.8
SS	8.2	8.4	7.2	0.2	−1.2	390	540	508	1.3	−0.3

Annual percent rate of change between the indicated years.

2009 in Rome.

Discussion

Urban sprawl has been monitored at different spatial scales using statistical data, field surveys and remote sensing approaches (e.g. Hasse and Lathrop, 2003; Kasanko et al., 2006; Schneider and Woodcock, 2008). Although sprawl has been considered a crucial issue in the European policy agenda since the early 2000s (EEA, 2006), its socioeconomic and territorial drivers have been explored only partially in European countries (Couch et al., 2007). In particular, the relationship between urban form and sprawl models has been poorly investigated. The present study contributes to this issue by assessing the spatial evolution of compact and sprawling areas in two expanding urban regions through the long-term analysis of population and land-use data. The comparative analysis of the development trajectories of cities that differ in urban form, population trends (growing versus declining), topographic variables (e.g. availability of land to edification) and socioeconomic characteristics is of interest because of the current debate on urban form and the sustainable spatial organization of peri-urban regions (Alphan, 2003; Catalàn et al., 2008; Chorianopoulos et al., 2010; Munafò et al., 2010; Terzi and Bolen, 2009).

Two distinct ‘sprawl models’ have been identified in Athens and Rome, possibly influenced by the different urban morphology: a profile of these two models based on the variables analysed in this work is summarized in Table 8. Processes of exurban development reflect, in the two cities, the progressive decoupling of population increase and the city’s expansion from the largest urban centres (in Rome) and the strong dependence on the past, partly unplanned, development (in Athens). The prevailing ‘sprawl model’ in Rome was characterized by medium-density settlements and moderate population increase and was supported by urban plans (Munafò et al., 2010). Sprawled settlements were concentrated on flat and coastal areas with large availability of land to edification and high pressure on land prices due to second home diffusion, determining a land take process primarily at the expenses of (abandoned) cropland and pastures (Salvati and Sabbi, 2011). In Athens, while compact growth is associated with fringe areas similarly to what observed in past decades (Leontidou, 1990), sprawl phenomena are more evident outside the consolidated urban agglomeration, and are concentrated in peri-urban and rural municipalities with low or medium-low population density, determining relevant impacts on the natural (mainly forest) ecosystem (Salvati et al., 2013a). Regional planning contributed to shape the geography of low-density settlements in the area by introducing different land development coefficients according to the closeness to the inner city or to the coastline (e.g. Gospodini, 2009). Changes in the use of the existing building stock revealed also an important factor of sprawl (Salvati et al., 2013b). For instance, the transformation of second homes to primary residences in recent periods has been a widespread phenomenon influencing coastal areas (Grekousis et al., 2013). This influences not only the SI of the municipalities where this occurs, but also the travel to work flows and thus the compactness or diffusion of the local labour market (Sayas, 2006).

Table 8.

A summary profile of compacting and sprawling municipalities in Rome and Athens based on the analysis of the three groups of supplementary variables.

Variables	Athens’ municipalities		Rome’s municipalities
Variables	Compacting	Sprawling	Compacting	Sprawling
Land-use	Built-up is the prevalent use of land; relict forests decline over the study period; cropland increases moderately	Natural and semi-natural land are the dominant use of land; forests strongly decline over the study period; mixed trends for cropland	Mixed land-use with moderate increase in impervious land; strong decline in cropland concentrated in 1960–1990; forests increase in 1960–1990	Mixed land-use with increasing impervious land, decline in cropland and stable (or weakly declining) forest land
Topography	Large-size municipalities with >100,000 residents in 1990 concentrating along the urban fringe (<15 km from the inner city)	Small-size municipalities with <10,000 residents in 1990 located at distances >25 km from the inner city	Small- and medium-size municipalities concentrated at distances >20 km from the inner city	Medium- and large-size municipalities placed along the coastline or in areas with large availability of land to edification at distances >20 km from the inner city
Population	Concentrate more than 80% of resident population with a moderate decrease over time and population density always >500 inhabitants/km²	Concentrate less than 20% of resident population but with a considerable increase in population density over time	Concentrate less than 40% of resident population with a moderate decrease over time	Concentrate more than 60% of resident population with average density >300 inhabitants/km²

Results highlight the role of urban form determining the prevalent ‘sprawl model’ in the two cities: while sprawl has modified only partly the mono-centric model traditionally observed in Athens (Salvati et al., 2013a), the process of spatial diffusion of population and economic activities created new functional relationships between coastal and internal areas in Rome, leading to the moderately polycentric spatial organization that has characterized the city’s expansion since the 1960s (Salvati, 2013). However, it should also be recognized that changes in the economic structure (e.g. from construction or industry to advanced service activities) may be a major factor influencing population growth in certain areas, and this may not be a proxy for long-term expansion (or sprawl), but rather an indicator of changes in the spatial organization of the urban region (Sayas, 2004).

Interestingly, the strategic master plans enforced in law in the two cities seem to exalt, rather than mitigate, the environmental impact of the development path observed in the recent past. Athens’ master plan (Organismos Rithmistikou Schediou Athinon, 2012) is oriented towards the spatial rebalancing of the huge density disparities occurring between urban and rural areas in Attica through the development of sub-centres specializing in services and located at both the fringe of the consolidated city and far away from the main urban agglomeration. The conservation of natural and open spaces is promoted via development of a green belt that will protect four mountains surrounding Athens. However, cropland in lowland and hilly areas where sprawl processes, as clearly shown by Chorianopoulos et al. (2010), concentrated, are currently excluded from the green belt. Moreover, general statements on the protection of the natural environment and the preservation of cultural landscapes appear as rhetorical and poorly effective without an efficient system of environmental controls incorporated into strategic planning. But more importantly, compactness, although recognized as a strength of Athens’ urban form, is not pursued through processes of urban regeneration and re-use of brownfield sites (except in limited circumstances). Indeed, the master plan could indirectly stimulate new urbanization waves decoupled from the main urban centres (Arapoglou and Sayas, 2009).

In Rome, the Strategic Master Plan has been pursuing a polycentric spatial organization in the last 15 years by promoting new sub-centres, strengthening metropolitan poles, stimulating new economic activities at the fringe and re-localizing urban functions (Gemmiti et al., 2012). According to the planners’ vision, these measures will contribute to the shift from the traditional Mediterranean city’s organization towards polycentrism with the undesired consequence of the conversion of traditionally agricultural land to urban uses (Munafò et al., 2010). As a matter of fact, although polycentric growth has become a ‘normative agenda’ in European regional policy (Davoudi, 2003), the peculiar characteristics of Rome’s region are different from those observed in western and central Europe both in terms of forms and functions (Salvati and Sabbi, 2011), possibly making the polycentric development a driver of (instead a solution to) sprawl (Leontidou, 1996).

Conclusions

Over time, different forces are involved in shaping cities’ development processes, while different phases of urbanization, related to the production of space mechanisms and their relation to the organization of space and land-use patterns, as well as their effects on urban sustainability are taking place. Choosing the right policies and strategies requires an in-depth understanding of the situation and the phase of development in which the city is involved (Richardson and Chang-Hee, 2004).

A new planning philosophy capable of addressing the specificity (and, in some ways, the uniqueness) of the Mediterranean urban regions is needed to face the recent sprawl process in connection with the increasing loss of environmental values and ecosystem services (e.g. Faludi, 2006). Regional planning should incorporate measures promoting the environmentally sustainable and socially balanced development of peri-urban regions according to the prevailing territorial characteristics and urban form (Davoudi, 2003).

Although Burton (2000), among others, provided evidence that a compact urban form is more sustainable than a sprawled one, results of the present study corroborate the ‘compact city’ paradox illustrated by Neuman (2005) and suggest that conceiving the city in terms of form is neither necessary nor sufficient to achieve the goals ascribed to the compact city. Instead, conceiving the city in terms of processes instead of patterns (Couch et al., 2007) holds more promise in attaining the elusive goal of a sustainable city. Based on sustainable land management principles, adapting a moderate polycentric development (Herrschel, 2009) incorporating the environmental sustainability targets of traditionally compact cities to the specific socioeconomic traits of each urban region seems to be the key issue in the control of dispersed city expansion in the Mediterranean region. But primarily, according to the opinion of Delladetsima (2012), sustainability should be recognized in southern European countries as a socially acceptable target with effects at the local scale rather than the objective of top-down policies decided further away from the affected land.

Footnotes

Notes

References

Alphan

(2003) Land use change and urbanisation of Adana, Turkey. Land Degradation & Development 14(2): 575–586.

Antrop

(2004) Landscape change and the urbanization process in Europe. Landscape and Urban Planning 67(3): 9–26.

Arapoglou

Sayas

(2009) New facets of urban segregation in southern Europe. European Urban and Regional Studies 16(4): 345–362.

Arribas-Bel

Nijkamp

Scholten

(2011) Multidimensional urban sprawl in Europe: a self-organizing map approach. Computers, Environment and Urban Systems 35(4): 263–275.

Breheny

(1997) Urban compaction: feasible and acceptable. Cities 14(4): 209–217.

Bruegmann

(2005) Sprawl: A Compact History. Chicago, IL: University of Chicago Press.

Burchell

Downs

McCann

Mukherji

(2005) Sprawl Costs: Economic Impacts of Unchecked Development. Washington, DC: Island Press.

Burton

(2000) The compact city: just or just compact? A preliminary analysis. Urban Studies 37(11): 1969–2007.

Burton

(2002) Measuring urban compactness in UK towns and cities. Environment and Planning B: Planning & Design 29(1): 219–250.

10.

Catalàn

Sauri

Serra

(2008) Urban sprawl in the Mediterranean? Patterns of growth and change in the Barcelona Metropolitan Region 1993–2000. Landscape and Urban Planning 85(3–4): 174–184.

11.

Chorianopoulos

Pagonis

Koukoulas

Drymoniti

(2010) Planning, competitiveness and sprawl in the Mediterranean city: the case of Athens. Cities 27(2): 249–259.

12.

Couch

Petschel-held

Leontidou

(2007) Urban Sprawl in Europe: Landscapes, Land-Use Change and Policy. London: Blackwell.

13.

Davoudi

(2003) European briefing: polycentricity in European spatial planning: from an analytical tool to a normative agenda. European Planning Studies 11(8): 979–999.

14.

Delladetsima

(2012) Sustainable development and spatial planning: some considerations arising from the Greek case. European Journal of Spatial Development Refereed article No. 46: 21 pp.

15.

European Environment Agency (EEA) (2006) Urban sprawl in Europe – the ignored challenge. EEA Report no. 10. Copenhagen: EEA.

16.

European Environment Agency (EEA) (2010) Land in Europe: prices, taxes and use patterns. Technical report no. 4. Copenhagen: EEA.

17.

Ewing

(1997) Is Los Angeles-style sprawl desirable? Journal of the American Planning Association 63(1): 107–125.

18.

Faludi

AKF

(2006) From European spatial development to territorial cohesion policy. Regional Studies 40(6): 667–678.

19.

Glaster

Hanson

Ratcliffe

Wolman

Coleman

Freihage

(2001) Wrestling sprawl to the ground: defining and measuring an elusive concept. Housing Policy Debate 12(4): 681–717.

20.

Gargiulo Morelli

Salvati

(2010) Ad hoc Urban Sprawl in the Mediterranean City: Dispersing a Compact Tradition? Rome: Nuova Cultura.

21.

Gemmiti

Salvati

Ciccarelli

(2012) Mind the gap! ‘Global’ and ‘Ordinary’ cities in the planning perspective – Rome as a case study. International Journal of Latest Trends in Finance and Economic Sciences 2(2): 91–98.

22.

Gill

Handley

Roland Ennos

Pauleit

Theuray

Lindley

(2008) Characterising the urban environment of UK cities and towns: a template for landscape planning. Landscape and Urban Planning 87(4): 210–222.

23.

Gospodini

(2009) Post-industrial trajectories of Mediterranean European cities: the case of post-Olympics Athens. Urban Studies 46(1): 1157–1186.

24.

Grekousis

Manetos

Photis

(2013) Modeling urban evolution using neural networks, fuzzy logic and GIS: the case of the Athens metropolitan area. Cities 30: 193–203.

25.

Hasse

Lathrop

(2003) Land resource impact indicators of urban sprawl. Applied Geography 23(3): 159–175.

26.

Herrschel

(2009) City regions, polycentricity and the construction of peripheralities through governance. Urban Research & Practice 2(3): 240–250.

27.

Hofstad

(2012) Compact city development: high ideals and emerging practices. European Journal of Spatial Development Refereed article no. 49: 23 pp.

28.

Ioannidis

Psaltis

Potsiou

(2009) Towards a strategy for control of suburban informal buildings through automatic change detection. Computers, Environment and Urban Systems 33(2): 64–74.

29.

Jaegera

JAG

Bertillerb

Schwickc

Cavensd

Kienast

(2010) Urban permeation of landscapes and sprawl per capita: new measures of urban sprawl. Ecological Indicators 10(2): 427–441.

30.

Jenks

Burton

Williams

(1996) The Compact City: A Sustainable Urban Form? London: Spon.

31.

Johnson

Lewis

(2007) Land Degradation – Creation and Destruction. Lanham, MD: Rowman & Littlefield.

32.

Kahn

(2000) The environmental impact of suburbanization. Journal of Policy Analysis and Management 19(4): 569–586.

33.

Kasanko

Barredo

Lavalle

McCormick

Demicheli

Sagris

Brezger

(2006) Are European cities becoming dispersed? A comparative analysis of fifteen European urban areas. Landscape and Urban Planning 77(1–2): 111–130.

34.

Leontidou

(1990) The Mediterranean City in Transition. Cambridge: Cambridge University Press.

35.

Leontidou

(1996) Alternatives to modernism in (Southern) urban theory: exploring in-between spaces. International Journal of Urban and Regional Research 20(2): 180–197.

36.

Longhi

Musolesi

(2007) European cities in the process of economic integration: towards structural convergence. Annals of Regional Science 41(1): 333–351.

37.

Munafò

Norero

Sabbi

Salvati

(2010) Urban soil consumption in the growing city: a survey in Rome. Scottish Geographical Journal 126(3): 153–161.

38.

Neuman

(2005) The compact city fallacy. Journal of Planning Education and Research 25(1): 1–26.

39.

Organismos Rithmistikou Schediou Athinas (2012) Strategic Plan for Athens/Attica 2021. Athens: Ministry of the Environment, Energy and Climatic Changes. Available at: www.organismosathinas.gr (accessed May 2013).

40.

Richardson

Chang-Hee

(2004) Urban Sprawl in Western Europe and the United States. London: Ashgate.

41.

Salvati

(2013) ‘A chronicle of a death foretold’: urban shrinkage and land consumption in Rome, Italy. European Planning Studies 21(8): 1176–1188.

42.

Salvati

Sabbi

(2011) Exploring long-term land cover changes in an urban region of southern Europe. International Journal of Sustainable Development and World Ecology 18(4): 273–282.

43.

Salvati

Sateriano

Bajocco

(2013a) To grow or to sprawl? Land cover relationships in a Mediterranean city region and implications for land use management. Cities 30(2): 113–121.

44.

Salvati

Zitti

Sateriano

(2013b) Changes in the city vertical profile as an indicator of sprawl: evidence from a Mediterranean Region. Habitat International 38(1): 119–125.

45.

Sayas

(2004) An exploration of the social and spatial division of labour in the Athenian urban space. The Greek Review of Social Research 113: 167–206.

46.

Sayas

(2006) Urban sprawl in the periurban coastal zones of Athens. The Greek Review of Social Research 120: 71–104.

47.

Schneider

Woodcock

(2008) Compact, dispersed, fragmented, extensive? A comparison of urban growth in twenty-five global cities using remotely sensed data, pattern metrics and census information. Urban Studies 45(3): 659–692.

48.

Soliman

(2004) Regional planning scenarios in South Lebanon: the challenge of rural-urban interactions in the era of liberation and globalization. Habitat International 28(2): 385–408.

49.

Squires

(2002) Urban Sprawl: Causes, Consequences, and Policy Responses. Washington, DC: The Urban Institute Press.

50.

Sutton

Goetz

Fildes

Forster

Ghosh

(2010) Darkness on the edge of town: mapping urban and peri-urban Australia using nighttime satellite imagery. Professional Geographer 62(1): 119–133.

51.

Terzi

Bolen

(2009) Urban sprawl measurement of Istanbul. European Planning Studies 17(10): 1559–1570.

52.

Turok

Mykhnenko

(2007) The trajectories of European cities, 1960–2005. Cities 24(3): 165–182.

53.

Weber

Petropoulou

Hirsch

(2005) Urban development in the Athens metropolitan area using remote sensing data with supervised analysis and GIS. International Journal of Remote Sensing 26(4): 785–796.

54.

Williams

(2000) Achieving Sustainable Urban Form. London: Spon.