Exploratory Spatial Analysis of Social Vulnerability and Forest Fire Risk in the Pinhal Interior Sul (Central Portugal)

: Mainland Portugal is a Mediterranean country, particularly known for the recurrence of forest ﬁres in terms of burnt areas, losses, and damage. The central Portugal administrative area has been in a “stage” of prominence for this dangerous phenomenon, with the sub-region Pinhal Interior Sul being one of the most critical territories. In addition to the manifestation of the dangerous process itself, it is important to state the importance of vulnerability, which has been relatively ignored in the deﬁnition of public policies in terms of spatial planning. Vulnerability concerns the way people and their assets are exposed to forest ﬁres (as they represent the main targets of their harmful consequences), as well as how they are affected by, resist, and recover from them. This paper investigates only one of the components of vulnerability—social vulnerability—in terms of demographic, cultural, socio-economic, and infrastructural conditions to establish a comparative analysis among the parishes of the sub-region Pinhal Interior Sul (in particular, between the more urban and rural ones in each municipality). In the studied area, the high values of social vulnerability reinforce the high hazard level. At the municipality level, it is possible to conclude that the population living in more markedly rural and peripheral parishes is more vulnerable to forest ﬁres than those living in the more urban conditions.


Introduction
Due to the Mediterranean climate in which mainland Portugal is located [1][2][3][4], it is one of the most affected southern European countries in terms of forest fires [5][6][7][8][9][10][11][12][13]. The multi-decadal evaluation of forest fires evolution indicates an annual average of burnt area of 138,083 ha for Portugal, during the period from 2010 to 2019 [14]. From a national perspective, the central region possesses a considerable probability of forest fires [15,16], usually resulting in substantial losses (e.g., primary and second homes, businesses, forests, and agriculture), in addition to the loss of human lives. This scenario proves to be increasingly worrying in the context of climate change which, for the near future, indicates a significant increase in temperatures and a reduction in rainfall for the Mediterranean regions wherein Portugal resides [13,17].
For the central Portuguese administrative area, it is important to mention the fires of 2017 where 49,890 ha was burnt in the sub-region known as the Pinhal Interior Sul [18][19][20][21], promoting the biggest fire in terms of dimension ever registered (29,758 ha) in the Sertã municipality and in the parish of Várzea dos Cavaleiros [22]. The high vulnerability of its territory, along with a decreasing and constantly ageing population framed in rural and abandoned landscapes with an essential forest cover [12,23,24], has transformed Pinhal Interior Sul into one of the most problematic areas in terms of the consequences of forest fires in Portugal.
Forest fires are nothing more than the existence of uncontrolled and undesired fires within a forested perimeter [53]. Their genesis originates from the ignition of fire in a forested perimeter under favourable conditions to their propagation. Among them, we can highlight the lack of forest management, low moisture content of fine combustibles, steep slopes, insolation in southern-facing slopes (lower air humidity levels), etc.
It should be noted that forest fires often spread beyond the forest perimeter, covering other types of land cover such as agricultural areas and even built spaces. Therefore, the forest urban interface plays a major role in terms of management and action in the protection of individuals and their assets/goods [42].
In short, forest fire risk cannot only be seen through the probability of occurrence, but also through its potential adverse consequences for society [28,54,55]. This is to say that it implies, necessarily, an approach to vulnerability, or the probability of occurrence of the negative repercussions of fires (material or human losses).

Study Area
The sub-region Pinhal Interior Sul is situated in the central Portugal administrative area, integrating five municipalities, namely, Oleiros, Proença-a-Nova, Sertã, and Vila de Rei, in the district of Castelo Branco, and, lastly, the municipality of Mação, in the district of Santarém. Forty-three parishes cover the entire study area (Figure 1).   In Portugal, the rural territories of inland, as the study area, are nothing more than the "mirror" of the "demographic transition phenomena", which resulted in the economic and social transformations that have occurred since the 1960s and 1970s (i.e., migrations to coastal areas, to larger urban centres, and abroad, and even the declining birth rate) [56]. When it comes to the inland municipalities, we have been witnessing a strong decrease in the population in recent decades, explained not only by "inter-municipal movements" but also by the general abandonment of rural areas. Depopulation is nothing more than the reality of traditional rural areas, which is also described by its immediate consequences, such as agricultural land abandonment due to the lack of workforce and low income on the part of this field of activity. Figure 2 shows that the parishes of the study area corresponding to the seats of the municipalities (no. 9, 17, 26, 33, and 40) register higher numbers of residents per km 2 , highlighting the parishes of Sertã (no. 17) and Mação (no. 40). This is justified since it is in the seat of the municipalities, as in some parishes around them, in which a higher amount of goods and services associated with the economic activity are concentrated (with the predominance of the tertiary sector).
Alvito da Beira, (20)  In Portugal, the rural territories of inland, as the study area, are nothing more than the "mirror" of the "demographic transition phenomena", which resulted in the economic and social transformations that have occurred since the 1960s and 1970s (i.e., migrations to coastal areas, to larger urban centres, and abroad, and even the declining birth rate) [56]. When it comes to the inland municipalities, we have been witnessing a strong decrease in the population in recent decades, explained not only by "inter-municipal movements" but also by the general abandonment of rural areas. Depopulation is nothing more than the reality of traditional rural areas, which is also described by its immediate consequences, such as agricultural land abandonment due to the lack of workforce and low income on the part of this field of activity. Figure 2 shows that the parishes of the study area corresponding to the seats of the municipalities (no. 9, 17, 26, 33, and 40) register higher numbers of residents per km², highlighting the parishes of Sertã (no. 17) and Mação (no. 40). This is justified since it is in the seat of the municipalities, as in some parishes around them, in which a higher amount of goods and services associated with the economic activity are concentrated (with the predominance of the tertiary sector).  In contrast, the more rural and peripherical parishes, such as Marmeleiro (no. 30), Amêndoa (no. 37), São João do Peso (no. 34), São Pedro do Esteval (no. 36) and Isna (no. 16), are characterized by low dynamism of economic activities associated with the tertiary sector where there are low rates of job offers, making them unattractive to young people (as well as resulting in poor agroforest activities dynamism).
Regarding the forest fires of the study area, in the period from 2010 to 2018 [58], we can see that the greater extent of the burnt area (49,890 ha) was created in 2017 ( Figure 3), due to the particularly adverse weather conditions that affected every parish belonging to the sub-region Pinhal Interior Sul.
In contrast, the more rural and peripherical parishes, such as Marmeleiro (no. 30), Amêndoa (no. 37), São João do Peso (no. 34), São Pedro do Esteval (no. 36) and Isna (no. 16), are characterized by low dynamism of economic activities associated with the tertiary sector where there are low rates of job offers, making them unattractive to young people (as well as resulting in poor agroforest activities dynamism).
Regarding the forest fires of the study area, in the period from 2010 to 2018 [58], we can see that the greater extent of the burnt area (49,890 ha) was created in 2017 ( Figure 3), due to the particularly adverse weather conditions that affected every parish belonging to the sub-region Pinhal Interior Sul.

Materials and Methods
The statistical data used to evaluate social vulnerability for the study area were collected from the Geographical Database for Referencing (BGRI database http://mapas.ine.pt/download/index2011.phtml; accessed on 15 October 2020), as well as from the Contemporaneous Portugal Database (PORDATA-https://www.pordata.pt/; accessed on 15 October 2020). Both data sources refer to the year 2011, and all of the information was collected at the parish scale. The BGRI database was chosen since it is easy to use and for its articulation with georeferenced spatial data (e.g., limits of municipalities and parishes). In addition, due to the general scarcity of data available in Portugal at the parish scale, more specifically regarding determining the support capacity, information on the 24-h service chemists' website (https://www.farmaciasdeservico.net/;

Materials and Methods
The statistical data used to evaluate social vulnerability for the study area were collected from the Geographical Database for Referencing (BGRI database http://mapas. ine.pt/download/index2011.phtml; accessed on 15 October 2020), as well as from the Contemporaneous Portugal Database (PORDATA-https://www.pordata.pt/; accessed on 15 October 2020). Both data sources refer to the year 2011, and all of the information was collected at the parish scale. The BGRI database was chosen since it is easy to use and for its articulation with georeferenced spatial data (e.g., limits of municipalities and parishes). In addition, due to the general scarcity of data available in Portugal at the parish scale, more specifically regarding determining the support capacity, information on the 24-h service chemists' website (https://www.farmaciasdeservico.net/; last accessed on 15 October 2020) was also gathered from municipal social maps that are made up of three distinct parts: (i) the social diagnosis; (ii) the social development plan; and (iii) the action plan. Social diagnosis is used as an instrument for identifying priority needs and problems, as well as the resources, potentialities, and constraints of each territory. The social development plan, based on the priorities defined in the social diagnosis, determines intervention axes and strategic objectives, which are in turn operationalized through an annual action plan. Now, many of the municipalities in Portugal have a social map as a tool to aid in municipal planning. The road density was obtained from primary, secondary, and tertiary road-length line-vector data (in a geographic information system-GIS).
The reduced amount of information that could be gathered at the parish scale meant that it was not possible to carry out a separate calculation for criticality and support capacity, but to include all of the data regarding both in the calculation of a general index of social vulnerability that would take into consideration the integration of these two features.
Considering the available statistical data, a set of variables was chosen to assess the social vulnerability of the study area. To avoid redundancy in the information of the variables, the initial number was reduced using a correlation matrix containing 35 variables, of which only 31 were selected ( Table 1). The SoVI method [25,26,41], which was adapted according to the type of risk (forest fire), the nature/scale of the study area, and the data availability, was then applied. This set of thirty-one variables, distributed among six main categories (Table 1), contribute in different ways to increasing the negative effects and economic losses due to forest fires, thus raising the vulnerability of the individuals and communities. In some cases, the chosen variables concern more general aspects related to the exposure of people and goods, such as population density and building density. In other cases, these are variables that concern the way in which the population can face forest fire situations, contributing to their early warning, to a first confrontation of the flames, and to the direct protection of their goods-this is the case for variables such as the proportion of women, the youth de-pendency index, and the proportion of the employed population, among others. Other chosen variables are related to the age, poverty, and illiteracy conditions of the population, which makes it especially difficult for them to recover after economic losses related to forest fires. Examples of these types of variables are the proportion of the aging population, the number of old or recent buildings, and the proportion of employees and retirees, as well as the proportion of individuals who cannot read or write. However, the general state of preparation of the territories seen, for example, through the number of pharmacies or roads, can also contribute to raising social vulnerability when they are insufficient and/or poorly spatially distributed.
After the selection of these 31 variables, the factorial analysis on SPSS 25 software was performed, from which the KMO (Kaiser-Meyer-Olkin measure-sample suitability) value was 0.61. This value, between 0.6 and 0.7, falls under the category of reasonable significant values, and therefore, principal component analysis could be carried out [59]. The Bartlett test was also performed, which returned good results, in accordance with the communality one [59]. Using factorial analysis, ten factors were obtained, from which we excluded the last one from the calculation of vulnerability once it presented a weak correlation with any of the selected variables.
Subsequently, the social vulnerability index (SVI), for each territorial unit under consideration (parish), was applied. This was based on the sum of the values of the scores of each territorial unit for each factor, weighted by the percentage of the explained variance by each of the factors and signalled positively or negatively according to their relevance (positive or negative in the increment in social vulnerability), as shown in the following expression: where F n = principal components analysis resulting factors, such as F n = 1 . . . 9; Var (F n ) = percentage of the variance explained by the factors, such as F n = 1 . . . 9. Finally, after applying the expression (1), a cartography regarding the spatialization of social vulnerability, sorted by parishes, was made using GIS tools (namely ESRI's ArcGIS 10.7.1 software). The natural-breaks classification procedure was used to create the intervals between classes of social vulnerability: very high (red), high (orange), medium (yellow), low (light green), and very low (dark green). These five classes were based on natural groupings inherent in the data (GIS software identified break points by picking class breaks that best grouped similar values and maximised the differences between classes), and the features were divided into classes with boundaries that were set when there were relatively large jumps in the data values.
It is also worth noting that a dispersion graph regarding the crossing of the principal factors (Factors 1 and 2) was made using a Microsoft Excel spreadsheet. Table 2 shows the results of factorial analysis where the nine factors (percentage of the total explained variance), the correlation with the dominant variable, and the direction of this correlation are shown. These nine factors, which explain 79.2% of the total variance, were considered for the study of social vulnerability in the sub-region Pinhal Interior Sul, among which Factor 1 is worth highlighting, which explains 27.6% of the variance; has a −0.9 correlation with the dominant variable (the proportion of the elderly population); and relates negatively to the percentage of the retired population. Factor 2 is related to families with no unemployed members which, naturally, suggests a better economic condition and therefore less vulnerability. Factor 3 is related to the percentage of women in the population and reflects both the aging process (women have a longer life expectancy) and internal/foreign migratory processes. The greater the proportion of women, the more fragile the family will be, both in the process of resistance and in the recovery from forest fires. Factor 4 is directly related to the health conditions of the inhabitants, but it is also a sign of the social and economic development of the parishes, which reduces the vulnerability of the populations. Furthermore, regarding the comparison between Factors 1 and 5 (Table 2), these two variables should be strongly correlated. However, the ageing index refers to the proportion of the elderly population (aged 65 years and over) per 100 individuals younger than 14 years old in a specific population. Practically, it is a tool to measure the untapped potential of older people for active and healthy aging in a society. The difference between the percentage of elderly people and the ageing rate of a population may seem subtle, but they have different meanings. In the case under study, the numbers of the ageing index are more discrepant than the numbers of the percentage of elderly people in relation to the average values of central Portugal and the country as a whole.

Results and Discussion
Hence, Factor 1 concerns demographical characterization where the young population means a contribution to the generational turnover, the interior being "known" by its ageing population. The higher scores in this factor occur, logically, in the parishes of Sertã (no. 17), Proença-a-Nova (no. 26), and Oleiros (no. 9), as well as in other less rural parishes (i.e., with better infrastructures and more developed from a socio-economic point of view). Cernache de Bonjardim (no. 20) and Cabeçudo (no. 23) (Figure 4) are, in the context of the study area, the parishes with a higher young population percentage, which can be explained by the existence of a set of conditions regarding education, health, and housing.
Regarding Factor 2, it can be observed that the parishes that are the seat of municipalities, such as Oleiros (no. 9), Proença-a-Nova (no. 26), and Sertã (no. 17), along with other less rural parishes, such as Amieira (no. 3), Cambas (no. 1) and Isna (n. 16), have a lower percentage of unemployed individuals (Figure 4), which means that there is a better development of economic activities, especially in the tertiary sector (the main employment creators). However, the rural parishes have high levels of unemployment, which reveals the weak dynamism of the activities of the primary sector.  It can be observed in Figure 5 that the seats of the municipalities have the lowest values of social vulnerability in relation to the other parishes. In a decreasing order, we can highlight Sertã (no. 17), Oleiros (no. 9), Proença-a-Nova (no. 26), Mação (no. 40), and Vila de Rei (no. 33). Despite not being the seat of a municipality, Cernache do Bonjardim (no. 20) has a very low value of social vulnerability, which can be explained by the fact that it is a town with characteristics, functions, and equipment identical to the town of Sertã (no. 17), which is even better positioned than Vila de Rei (no. 33). Other less rural parishes, such as Mosteiro (no. 12), Troviscal (no. 13), Cabeçudo (no. 23), and Cumeada (no. 28) also have very low levels of social vulnerability, followed by low levels in the parishes of Estreito (no. 7), Castelo (no. 15), Várzea dos Cavaleiros (no. 24), São Pedro do Esteval (no. 36), and Sobreira Formosa (no. 18). In the case of the parish of Várzea dos Cavaleiros (no. 24), the low level of social vulnerability is tied to the fact that, on the other hand, there are several kindergartens which are not present in other rural parishes (more favourable levels of social vulnerability, e.g., Palhais-no. 29). This is explained by the fact that it is in the seat of the municipalities in which the main economic activities, but also the main social infrastructures (i.e., the nursing homes, the chemists, the schooling centres, the kindergartens and the pre-schooling centres, the fire stations, etc.) are found. Combined with the main elements that characterize the support capacity, there are local features when it comes to criticality that justify these levels, such as the greater concentration of the young population and the existence of services tied to the tertiary sector, which are areas that attract people from peripherical parishes who end up finishing their studies in the seat of the municipality where they live (or even out of them in medium-sized cities in the case of higher education). In short, it can be observed that the parishes that are the seat of the municipalities, such as Sertã (no. 17), Oleiros (no. 9) and Proença-a-Nova (no. 26), are better positioned when related to the two most important factors once they register a better socio-economic dynamism. On the other hand, parishes such as Palhais (no. 29) and Vilar Barroco (no. 4), the most peripherical and truly rural parishes, register a least a favourable position given the low economic power, decayed housing stock, aged population, and active population professional structure (which is largely based on the primary sector).
It can be observed in Figure 5 that the seats of the municipalities have the lowest values of social vulnerability in relation to the other parishes. In a decreasing order, we can highlight Sertã (no. 17), Oleiros (no. 9), Proença-a-Nova (no. 26), Mação (no. 40), and Vila de Rei (no. 33). Despite not being the seat of a municipality, Cernache do Bonjardim (no. 20) has a very low value of social vulnerability, which can be explained by the fact that it is a town with characteristics, functions, and equipment identical to the town of Sertã (no. 17), which is even better positioned than Vila de Rei (no. 33). Other less rural parishes, such as Mosteiro (no. 12), Troviscal (no. 13), Cabeçudo (no. 23), and Cumeada (no. 28) also have very low levels of social vulnerability, followed by low levels in the parishes of Estreito (no. 7), Castelo (no. 15), Várzea dos Cavaleiros (no. 24), São Pedro do Esteval (no. 36), and Sobreira Formosa (no. 18). In the case of the parish of Várzea dos Cavaleiros (no. 24), the low level of social vulnerability is tied to the fact that, on the other hand, there are several kindergartens which are not present in other rural parishes (more favourable levels of social vulnerability, e.g., Palhais-no. 29). This is explained by the fact that it is in the seat of the municipalities in which the main economic activities, but also the main social infrastructures (i.e., the nursing homes, the chemists, the schooling centres, the kindergartens and the pre-schooling centres, the fire stations, etc.) are found. Combined with the main elements that characterize the support capacity, there are local features when it comes to criticality that justify these levels, such as the greater concentration of the young population and the existence of services tied to the tertiary sector, which are areas that attract people from peripherical parishes who end up finishing their studies in the seat of the municipality where they live (or even out of them in medium-sized cities in the case of higher education). Sustainability 2022, 14, x FOR PEER REVIEW 12 of 16  Figure 5). These values, which reflect high social vulnerability, and the differentiation of its levels among the parishes of the sub-region Pinhal Interior Sul (Figure 5), show the fragility of the population against natural forest fire events, which are the main environmental risk at the regional scale, especially in the most peripherical and rural parishes. This fragility of the population means that the resistance to forest fires, as well as the recovery after them, is therefore seriously compromised both at the socioeconomic and territorial levels.
In terms of the methodological approach used, it allows us to effectively describe the levels of social vulnerability that affect one of the most impoverished, depopulated, and aged territories in the inland of Portugal. The cartographic results achieved ( Figure 5), These levels can be mostly explained by the increase in the ageing index along with the fact that the economic activities are based on the primary sector. However, the low income and the lack of state support for agroforestry activities mean that most adults and the young population are not interested in these activities, seeking their sustenance in other activities and in other territories.
Some parishes of the Oleiros municipality, such as Sobral (no. 8), Álvaro (no. 6), Amieira (no. 3), and Cambas (no. 1) have a very high level of social vulnerability, being the most vulnerable compared to the other ones despite Palhais (no. 29), São João do Peso (no. 34), and Peral (no. 35) also having highly significant values ( Figure 5). These values, which reflect high social vulnerability, and the differentiation of its levels among the parishes of the sub-region Pinhal Interior Sul (Figure 5), show the fragility of the population against natural forest fire events, which are the main environmental risk at the regional scale, especially in the most peripherical and rural parishes. This fragility of the population means that the resistance to forest fires, as well as the recovery after them, is therefore seriously compromised both at the socioeconomic and territorial levels.
In terms of the methodological approach used, it allows us to effectively describe the levels of social vulnerability that affect one of the most impoverished, depopulated, and aged territories in the inland of Portugal. The cartographic results achieved ( Figure 5), especially at the comparative scale of parishes, will allow local/regional public authorities to intervene differently and to target the specific problems of each area. However, it is also important to highlight that the used method has some limitations. Indeed, starting from an initial set of variables, the factor analysis seeks to synthesize and identify a smaller set of hypothetical variables (factors) [59], although these factors have a much more complex perception than the initial variables, making their practical use in solving problems difficult. The absence of some data sources for specific variables considered important (these being the most expressive example of per capita income) should also be mentioned. In fact, in Portugal there are few data at the parish scale (and more specifically regarding the territorial conditions), which make it impossible to calculate social vulnerability in the two separated parts (i.e., criticality and support capacity). On the other hand, the detailed cartography carried out at the parish scale establishes territorial generalizations that are sometimes difficult to apply to a phenomenon such as a forest fire, which is relatively discontinuous from a spatial point of view.
Finally, the transformations registered in the last 50/60 years in Portugal, namely in the rural areas of its inland-such as the sub-region Pinhal Interior Sul-were responsible for agricultural abandonment, as well as for the proliferation of forest and bush spaces, which are almost always poorly organized and poorly managed. This transformation is related to depopulation and the progressive aging of the population, which is increasingly exposed and more vulnerable to forest fires. In this scenario, this investigation made it possible to develop an applicative study on social vulnerability for a critical area in terms of forest fires at an analytical and spatial scale (parishes) never used before. Furthermore, this study provides the possibility of verifying how this vulnerability increases the fire risk in the territory under investigation, as well as defining new proposals of public policies based on a reduction in vulnerability.

Conclusions
The current study focuses on the relevance of social vulnerability at the parish scale, from which it is possible to highlight its spatial differentiation in a way that enables adequate social intervention when it comes to public policies.
The methodological approach used here to analyse the social vulnerability of the sub-region Pinhal Interior Sul (central Portugal) at the parish scale, and the cartographic results obtained, are aimed at helping to define and put into practice policies that envisage the mitigation of risk in this territory. Thus, in envisaging addressing the high and very high levels of social vulnerability at the parish scale, some medium-to long-term measures are proposed in the context of public policies targeting risk mitigation (namely the risk of forest fire, probably one of the most significant risks in the study area). Among the measures proposed are the following: (1) Creating and installing companies to manage forest combustibles, which would, in turn, create job opportunities, as well as creating professionalizing courses associated with this sector of economic activity; (2) Reinforcement of the municipal capacity to intervene in the management of the combustible containment strips on private pieces of land; (3) Promote a decrease in the unemployment rate and promote access to human resources; (4) Re-installation or replacement of infrastructure, as well as social and economic services that were being removed as the population declines; (5) Promote technical and academic studies on forest fire risk, which are of economic value to forests and help to create measures to reduce and mitigate the risk.
These measures should be integrated into public policies at the regional and municipal levels to promote forest management and develop prevention and rescue services. The foreseeable reduction in hazards, which is necessary in the context of climate change that we are experiencing, will have to be accompanied by measures to reduce vulnerability, such as the installation of services, job creation, population retention, and the development of rural tourism. Only in this way will it be possible to reverse a process of territorial