Keywords

1 Introduction

Suburban landscapes include a variety of land-use types reflected in fragmented morphologies and heterogeneous functions (Antrop, 2004; Duvernoy et al., 2018). Recent landscape transformations determined new elements and landscape structures superimposed upon the traditional (urban–rural) landscape mosaics (Alberti, 2010; Allen, 2003; Coluzzi et al., 2022; Delfanti et al., 2016Mc Donnell et al., 1997). These phenomena may reduce the intrinsic environmental quality of entire districts, determining an increased fragility of habitats, especially when located in ecologically sensitive areas (Foster et al., 2003; Imbrenda et al., 2022a; Jim, 2004; Johnson, 2001; Nickayin et al., 2022; Salvati et al., 2008; Zambon et al., 2018).

Mediterranean metropolises were affected by intense landscape transformations at the fringe, thanks to the rapid expansion of settlements observed since the 1950s and following the post-war demographic boom (Carlucci et al., 2018; Salvati, 2014; Salvati & Carlucci, 2016). These trends determined (and sometimes consolidated) a land take causing an irreversible loss in soil resources and cultural/natural heritages (Antrop, 2004; Atmis et al., 2007; Bianchini et al., 2021; Catalàn et al., 2008; Chorianopoulos et al., 2010; De Marco et al., 2019; Economidou, 1993; Nickayin et al., 2021; Salvati et al., 2012a; Santarsiero et al., 2022). Urbanization of already natural land with olive groves, vineyards, annual crops or shrublands, pastures and woodlots was largely documented along the fringe of Mediterranean cities (Barbati et al., 2013; Biasi et al., 2015; Cecchini et al., 2019; Paul & Tonts, 2005; Salvati et al., 2013a, Simoniello et al., 2022).

Less knowledge is however available on how landscape elements evolve, creating new structures and original patterns of functions (Imbrenda et al., 2022b; Kosmas et al., 2016; Marull et al., 2009), that may impact environmental quality at large (Falcucci et al., 2007; Salvati & Zitti, 2007, 2009; Zambon et al., 2017). In such contexts, wildfires shape the ecological fragility of Mediterranean fringe landscapes determining environmental degradation (e.g., on sloping/rocky/arid land, see Kosmas et al., 2000a, 2000b; Simeonakis et al., 2007; Salvati & Bajocco, 2011; Pignatti et al., 2015; Santarsiero et al., 2020; Nolè et al., 2020; Samela et al., 2022). Based on these premises, we assume land-use change as exerting a negative impact on the ecological fragility of a given territory fueling the intrinsic vulnerability to wildfires, likely determined by, e.g., modifications in plant cover and vegetation composition (Bajocco et al., 2015, 2016; Fares et al., 2017; Smiraglia et al., 2015).

A complete survey of land-use changes over 43 years (1975–2018) allowed us to estimate the spatiotemporal variation in a composite index of vulnerability to wildfires in a metropolitan region of Southern Europe (Athens, Greece) experiencing comparatively high exposure to fires than other metropolises in the macro-region, thus representing a sort of ‘worst scenario’ for other socioeconomic contexts in the same European quadrant. The explicit knowledge stemming from our study may support strategic planning containing wildfires and mitigating soil degradation and habitat fragmentation along urban fringes.

2 Methodology

The investigated area covers more than 3000 km2, a large part of the administrative region of Attica, Central Greece (Fig. 1). The area was partitioned into 115 mainland municipalities (58 forming Athens’ conurbation that extends 430 km2), including two municipalities on Salamina island close to Piraeus harbor, and excluding the remaining municipalities in the other islands of Argosaronic Gulf, Aegean Sea. The area consists of mountains with the highest elevation in Mount Parnitha, 25 km far away from downtown Athens (1413 m on the sea level). Three coastal plains (Messoghia, Marathon and Thriasio) are located immediately outside Athens’ conurbation (Ciommi et al., 2019; Di Feliciantonio et al., 2018; Salvati et al., 2012b, 2013b). The population living in the area amounted to 1.6 million people in 1951, increasing to 2.7 million people in 1971 (approximately 900 inhabitants/km2) and reaching 3.7 million people in 2021 (nearly 1200 inhabitants/km2). Urban population in total population declined from the peak of 92% observed in 1971 to 80% in 2021 (Gavalas et al., 2014).

Fig. 1
Two maps depict the study area of Attica, emphasizing the prefectures of Eastern Attiki, Athens, Western Attiki, and Piraeus. The second map displays the Digital Elevation Model ranging from 0 to 1400.

Study area of Attica (Greece): a location of four prefectures and b Digital Elevation Model from EU-DEM dataset

Full size image

The spatial distribution of selected land-use classes was investigated over two years (1975 and 2018) based on (i) the LaCoast (LC) digital cartography available for 1975 at 1:100,000 scale and covering the European coastal areas (Perdigao and Christiansen, 2000) and (ii) the Corine Land Cover (CLC) pan-European digital cartography available at the same spatial scale for 2018 (Büttner et al., 2017).

These two geospatial sources were regarded as comparable digital maps covering the study area homogeneously in the two-time points, in line with the nomenclature of the CLC project. Being coordinated by the European Environmental Agency (EEA), this initiative provided land-use maps multiple times for the whole of Europe with an inventory based on satellite images as the primary information source, with spatial scale (1:100.000), geometric resolution (minimum mapping unit of 25 ha), and minimum width of linear elements (100 m) reflecting the trade-off between production costs and level of detail of land cover information. By providing a comprehensive description of the landscape in the study area (Economidou, 1993), the standard CLC nomenclature includes 44 land classes (Salvati & Bajocco, 2011) grouped into a three-level hierarchy (1: artificial surfaces, 2: agricultural areas, 3: forests and semi-natural areas, 4: wetlands, and 5: water bodies).

The vulnerability of plant cover to fires was derived, composing the information from the two maps described above following the rules dictated by a quantitative approach developed in the Environmentally Sensitive Area (ESA) methodology of the Medalus II (Mediterranean Desertification and Land Use) international cooperative research project (Kosmas et al., 1999). Being explicitly validated on the field in several target sites (Brandt, 2005), this framework was applied at both the regional and local scale in Mediterranean regions displaying locally differentiated environmental conditions (Imbrenda et al., 2021; Lanfredi et al., 2022; Pace et al., 2023).

Vegetation vulnerability to fires was estimated by attributing a score between 1 and 2 to each land-use class and leading to a final indicator named FR (‘Fire Risk’: Salvati & Bajocco, 2011). The approach was supported by a preliminary analysis (Kosmas et al., 2000a) developed to define the correlation between each land-use type and fire vulnerability based on literature review and field research mostly collected in the framework of Medalus projects (e.g., Basso et al., 2000; Brandt, 2005; Kosmas et al., 2000b). The results of a sensitivity analysis and a focus group allow for confirming the most valid, low-cost, and efficient score set (Kosmas et al., 1999). Ranging from 1 (the lowest vulnerability to wildfires) to 2 (the highest vulnerability to wildfires), FR indicated zero values assigned to land-use classes that were excluded from the analysis, namely compact urbanization (Salvati & Zitti, 2012).

The spatial distribution of the FR index was mapped for 1975 and 2018 using the same spatial resolution of CLC maps. Land changes and the related variations in FR were studied at the first CLC-class level. This analysis gave information on the positive (or negative) contribution of each class to FR. Changes over time (1975–2018) in the average FR score were quantified separately for each municipality of the study area (n = 115) by using the ‘intersect’ tool provided with ArcGIS software (ESRI Inc., Redwoods, USA) after the overlap between the FR map in 1975 (or 2018) and the shapefile depicting the municipality’s boundaries.

3 Results

Basic land-use changes between 1975 and 2018 were delineated in Table 1 and indicate that built-up areas and cropland expanded into other lands, although with different rates of increase, whereas forest/semi-natural landscape matrices decreased slightly. These changes resulted in a global rise of the FR index (0.1% per year – from 1.42 in 1975 to 1.44 in 2018). ‘Complex cultivation patterns’ contributed the most to the increase of the FR index (4.3% more than the average increase); ‘land principally occupied by agriculture, with significant areas of natural vegetation’ was the second contributor (2.0%). The highest negative contribution to FR came from ‘sparsely vegetated areas’ (−3.4%), ‘olive groves’ (−3.1%) and ‘mixed woodlands’ (−2.5%). At the aggregate level, however, urbanization (e.g., discontinuous, low-density settlements) contributed to FR more (0.75%) than cropland (+0.3%).

Table 1 Surface area of selected land-use classes in metropolitan Athens by year, and the consequent variation in the FR index
Full size table

The spatial distribution of the FR index in 2018 was illustrated in Fig. 2. The districts with a high vulnerability to wildfires are concentrated in the Athens fringe. At large, plant cover vulnerability does not seem correlated with the distance from downtown Athens (Spearman rank correlation test, p > 0.05). Changes in the FR index along the 43 observation years were mapped at the municipal scale (Fig. 3). The increase in the FR index was spatially heterogeneous, concentrating in suburban municipalities North and East of Athens; these areas experienced low-density urban expansion. Few municipalities at the fringe also showed a moderate reduction in the FR index over time possibly due to the loss in forest cover caused by recurrent fire events in the last decades.

Fig. 2
A spatial distribution of the study area in Attica illustrates the categorization of the fire risk index into Not Evaluated, Low, Medium Low, Medium High, and High.

Spatial distribution of Fire Risk index in the study area (2018)

Full size image
Fig. 3
A map of the study area in Attica illustrates the Fire Risk Index in the municipalities of metropolitan Athens between 1975 and 2018, categorized into Decreasing, Stable, and Increasing.

Discretized changes over time in FR index in the municipalities belonging to metropolitan Athens between 1975 and 2018

Full size image

4 Discussion

Transforming fringe landscapes into low-quality cropland and fragmented forest mosaics is detrimental to environmental quality, exalting the ecological fragility of land (Corona, 2018; Corona et al., 2016; Nocentini et al., 2017; Recanatesi et al., 2016). As a possible response to urban sprawl, a rising vulnerability to wildfires was recorded in Attica, although with a mostly heterogeneous spatial pattern that suggests how suburbs between 15 and 30 km from downtown Athens are mostly endangered. Results also support the relevance of an ecological spiral of human pressure resulting in the increase of local vulnerability to wildfires. The recurrent fires in Attica (Salvati et al., 2012a, 2012b) have likely represented an additional engine of land-use changes. Coupled with the increasing events of droughts and the asynchronous distribution of rainfall across large areas (Coluzzi et al., 2020; Lanfredi et al., 2020), the synergistic impact of wildfires and landscape transformations have consolidated the local vulnerability to wildfires (Cillis et al., 2022; Salvati et al., 2013a, 2013b).

These trends can be assessed continuously by integrating remote sensing, geospatial information sources (e.g., Corine Land Cover maps), and field surveys estimating the environmental impact of urban sprawl (European Environment Agency, 2016). Altering landscape patterns and fragmenting high-quality vegetation covers, urban sprawl fueled the intrinsic divergence in ‘extensive’ and ‘intensive’ land uses, exacerbating the spatial polarization in high- and low-quality vegetation areas and thus increasing the local vulnerability of vegetation cover to wildfires (Pickett et al., 2001). The increase of vegetation vulnerability to fires because of inherent homogenization in species composition, disruption of the hydrological systems, and modification of energy flow and nutrient cycling (e.g., Alberti, 2010; Foster et al., 2003; Johnson, 2001) should be also monitored extensively in the framework discussed in our work.

5 Concluding Remarks

Sustainable land management in fringe districts is made more urgent with climate change and requires appropriate assessment methodologies and conservation strategies focusing on relict, high-quality vegetation cover (Chelleri et al., 2015; Ciommi et al., 2018; Perrin et al., 2018). Efforts are finally needed to effectively integrate ecological studies and socioeconomic disciplines in a comparative, local-scale perspective informing fire science and the prevention/suppression cycle in suburban areas.