Assessing deserti ﬁ cation sensitivity map under climate change and agricultural practices scenarios: the island of Crete case study

The aim of this study is the assessment of deserti ﬁ cation risk for a typical Mediterranean island, in the frame of climate change and the application of good agricultural practices. Based on the MEDALUS Environmentally Sensitive Area Index (ESAI) approach, the sensitivity in deserti ﬁ cation is estimated by employing 15 quantitative parameters divided in to four main quality indices: climate, vegetation, soil, and management quality. The methodology applied for a baseline scenario (current conditions), two future climate change scenarios (RCP 4.5 and RCP 8.5) and a soil quality improvement scenario. According to the results, more than 13% of the island ’ s area is characterized as critically sensitive to deserti ﬁ cation in the current conditions. This percentage will increase in the future under both the RCP 4.5 and the RCP 8.5 climate scenarios, where the critical areas will rise above to 15%. By applying the soil quality improvement scenario simultaneously with the climate change scenarios, a slight mitigation of deserti ﬁ cation risk in the future could be achieved. The methodology developed in this study may be used to assess deserti ﬁ cation process under various climate, soil, and land use management scenarios in regions of the Mediterranean Sea. deserti ﬁ cation scenarios of GAPs and


GRAPHICAL ABSTRACT INTRODUCTION
Land degradation is the 'reduction or loss in the biological or economic productive capacity of the land caused by human activities and often magnified by the impacts of climate change' (UNCCD ). Desertification is a type of land degradation that occurs mainly in arid, semi-arid, and dry sub-humid environments, where water is the main limiting factor of land use performance in ecosystems.
Environmental systems are generally in a state of dynamic equilibrium, thus a small change in climate or at the land use intensity can affect the desertification risk of an area (Kosmas et al. ; Baartman et al. ).  (Montanaro et al. ). On the other hand, soil erosion, especially in arid and semi-arid lands, contributes to soil degradation, which in turn affects the sustainable agricultural land use and productivity (Lal ).
Over the past decades, various models and methodologies have been proposed to assess the desertification sensitivity of an area (Oldeman & (Boudjemline & Semar ). In Greece, the methodology has been applied in Lesvos island by Kosmas et al. () and Symeonakis et al. () and in Crete by Morianou et al. (). The majority of the aforementioned studies assess the desertification risk in a one-time-step static system. Symeonakis et al. () noted that the land degradation process is dynamic due to the time dimension and the indicators should, therefore, derive from a sequence of temporal steps. To fill in this gap, they developed a continu-

Case study
Crete, the study area, is an island in the Mediterranean Sea and is the largest island in Greece (35:20:27 N,25:07:46 E).
Crete is an island of the Mediterranean Sea with a total area of about 8,265 km 2 . The western part of Crete includes the Chania and Rethymnon prefectures, while the eastern includes the prefectures of Heraklion and Lassithi. The elevations in Crete range from 0 m to 2,400 m mean sea level. In Crete, lowlands (<200 m) cover an area of 2,165 km 2 , 26% of the total area, semi-mountainous areas (200-800 m) cover an area of 4,627 km 2 which is 56% of the total area, while the mountainous area (>800 m) is approximately equal to 1,473 km 2 which is 18% of the total area.
The climate of Crete is sub-humid Mediterranean with humid and relatively cold winters, and dry and warm summers. The annual rainfall ranges from 300 to 700 mm/year in the lowlands, from 700 to 1,000 mm/year in the semimountainous areas, while in the mountainous areas it reaches 2,000 mm/year. Significant rainfall differences are recorded between the western and eastern areas of the island during the wet seasons (Kourgialas et al. ).

ESAI methodology
The areas environmentally sensitive to desertification can be assessed in relation to various parameters such as landscape, soil, geology, erosion, vegetation, climate, and human action (Kosmas et al. ). Each of these parameters is reclassified according to its behavior on desertification and weighting factors are assigned in each class.
In this study, the modified ESAI methodology proposed by Morianou et al. () is used. This modification of the MEDALUS approach includes two additional parameters related to soil quality (water erosion and soil organic matter) compared to the original one. Fifteen layers belonging to four main sensitivity groups (climate, soil, vegetation management, and land management) were collected from various sources and their values were standardized between 1 (low sensitivity) and 2 (high sensitivity) (Tables 1-4). The four sensitivity groups were estimated using the following formula (Equation (1)): (1) where: sensitivity x , represents the computed value of each sensitivity and n represents the number of sub-indicators (layers) used to calculate each Sensitivity Index (SI).
Also, in this study, climate change scenarios (RCP 4.5 and RCP 8.5) as well as GAP scenarios were incorporated into climate and soil main sensitivity groups to investigate their effects on the final desertification sensitivity map. A flow chart of the proposed methodology is presented in Figure 1. In the following sections, analytical information for each involved sensitivity group index/parameter is given.

Climate sensitivity
In this study, the Climate Sensitivity Index was calculated from three climate indicators: rainfall, evapotranspiration, and aspect. Rainfall amount and its spatial distribution as well as the hydrological extreme events in the semi-arid and arid zones of the Mediterranean are the main climatic   Table 1.

Vegetation sensitivity
Vegetation cover is a very crucial factor affecting desertification and land degradation. The key indicators of desertification related to the existing natural or agricultural vegetation can be considered in relation to: (i) fire risk and ability to recover, (ii) erosion protection offered to the soil, (iii) drought resistance, and (iv) percentage plant cover.
The vegetation sensitivity map was produced using the vegetation cover data according to the CORINE Land Cover 2012 (CLC ). The input data that were used in the vegetation sensitivity group index are presented in Table 2.

Soil sensitivity
The soil sensitivity map was produced using the following soil data: texture, slope, parent material, soil depth, organic, and soil erosion caused by water. For the purpose of this study, soil data was derived from the European Soil Database (ESDAC). Soil erosion caused by water were derived from the study of seasonal and annual erosion assessments in Mediterranean agricultural areas using the G2 model (Panagos et al. ). The input data that were used in the soil sensitivity group index are presented in Table 3.

Management sensitivity
The definition of ESAs to desertification requires key indicators related to the physical environment and to the human-induced stress. The intensity of land use factor was produced according to the CORINE Land Cover 2012 (CLC ) and the protection policies factor that was produced using data from EU Directive 92/43 (NATURA ). The combination of these two parameters gives the final Management Sensitivity map. The input data that were used in the management sensitivity group index are presented in Table 4.

Climate change scenarios
Regarding the two studied climate change scenarios, RCP 8.5 scenario is characterized by increasing greenhouse gas    (Table 5).
Analytically, in the last ESAI stage, the final sensitivity to desertification of an area is evaluated from the

RESULTS AND DISCUSSION
The results/maps for the four sensitivity groups are shown in Figures 3-6. In terms of climate sensitivity (Figure 3), the majority of the island was mapped as having low to moderate climatic sensitivity, except for a part of the eastern island that was mapped as having high sensitivity under the current conditions (Figure 3(a)). This could be attributed to the fact that eastern areas (south Heraklion and Lasithi prefecture) receive less precipitation than the western part of the island (Koutroulis et al. ). Comparing the climate maps of the current situation (Figure 3(a)) with the two climate scenarios, RCP 4.5 (Figure 3(b)) and RCP 8.5 (Figure 3(c)), a significant part of the island seems to become more sensitive to desertification in terms of climate. It can also be noted that under the climate scenario RCP 4.5, there is higher sensitivity than under the scenario RCP 8.5. This is due to the fact that the climate model used gives less rain under the scenario RCP 4.5 compared to RCP 8.5 for the period 2031-2060. Since RCP 8.5 is a no-mitigation scenario, while RCP 4.5 is an average mitigation scenario, it is considered that, in general, the amount of rainfall for the Mediterranean region under the RCP 8.5 scenario is higher compared to the RCP 4.5. Nevertheless, this corresponds to an average trend which can differentiate depending on the RCM used and the specific area under study.
The amount of precipitation also affects the aridity index (precipitation/evapotranspiration), which gives lower values under climate scenario RCP 4.5. Specifically, at the eastern part of the island, in contrast to the general depiction presented by the climate maps, there seems to be an improvement in terms of climate sensitivity under both future climate scenarios. This is because, in recent years, the region of eastern Crete has long periods of drought, and in both climatic scenarios this region shows an increase in precipitation amounts.
Regarding the Vegetation Sensitivity Index, the areas with low drought resistance (e.g. agricultural areas), high fire risk (e.g. pine forests), and low percentage of vegetation cover are identified as highly sensitive. As Figure 4 indicates, the largest part of the island is characterized by moderate vegetation sensitivity except for some areas of Heraklion prefecture, which are characterized by high vegetation sensitivity. This is due to the fact that Heraklion prefecture includes large urbanization areas in the north and areas with intensive agricultural activity across the prefecture.
Considering also that vegetation cover is a crucial factor for soil erosion control in sloppy areas, a considerable part   Comparing the soil result maps, in the current condition ( Figure 5(a)) and after the application of GAPs for 40 years ( Figure 5(b)), a reduction of high sensitivity areas and an increase of the moderate sensitivity areas is observed. This is due to the increase of organic content in agricultural areas and the reduction of water erosion in areas with a slope of more than 10%. Therefore, it is observed that by the improvement of even two of the six soil sensitivity factors, a slight improvement in soil quality could be achieved.
In terms of the management policies dimension, Figure 6 shows that the north-western part of the island and the The environmental sensitivity to desertification of the Crete island according to the ESA approach in the current conditions (baseline scenario) is shown in Figure 7(a). According to these results and Table 6, the vast majority of the island is By applying the ESAI methodology under two climate change scenarios, this study indicates that, in the future, a greater extent of the island will be characterized as critical to desertification rather than fragile and potential. This Comparing the effects on desertification risk between the two future climate change scenarios (RCP 4.5 and RCP 8.5), it is observed that under the climate scenario RCP 4.5 more areas are characterized as critical (15.76% VS 15.09%) and less as non-affected and potential to desertification than under the scenario RCP 8.5 (3.97% vs 4.28% and 8.60% vs 9.05%, respectively) ( Figure 8). This is because the estimated amount of precipitation is lower in scenario RCP 4.5 compared to RCP 8.5.
Finally, the results obtained by the incorporation of the soil improvement scenario with the climate change scenarios showed that with the application of GAPs, the risk of desertification is slightly reduced. More specifically, the critical to desertification areas are reduced and, correspondingly, the percentage of areas not affected by desertification has increased with the application of GAPs. The fragile areas remain at the  . These sources of uncertainty could be reduced for small-scale field measurements, capturing as far as possible the heterogeneity of the studied system, as well as detailed climatological information.
These approaches are included in the present study and contribute to overcoming the limitations of the proposed method.

CONCLUSIONS
The present study is an application of the MEDALUS -ESAI methodology in a typical Mediterranean island (Crete) for the assessment of the desertification risk under different climate and agricultural practices scenarios. As a step forward from the previous desertification study in Crete by Morianou et al.
(), two future climate change scenarios were used to dynamically study the evolution of desertification sensitivity in the future. Going one step further, a scenario of soil quality improvement using GAPs is incorporated for the mitigation of the phenomenon.
In the current conditions (baseline scenario), a significant part of the island is characterized as fragile or critically sensitive to desertification. Human activities are the main causethe non-proper changes of land use, overgrazing and deforestation are only some of these activities.
The eastern part of Crete is more sensitive to desertification than the western part mainly due to the drier climate conditions.
By applying the ESAI methodology for two climate change scenarios, this study shows that the desertification risk of the island warrants attention for the future. The sensitivity of the island will increase greatly due to the increasing temperature and decreasing precipitation. Therefore, in this study a soil quality improvement scenario was also applied in combination with the two future scenarios.
The scenario was based on the results of GAPs applied in the frame of LIFE þ AgroClimaWater, to reduce the desertification sensitive areas in Crete. The results of this scenario show a slight mitigation of desertification risk in the future.
However, it seems that good agricultural practices, individually, are not enough to mitigate the risk of desertification in the context of climate change. Therefore, the problem of land degradation must be addressed comprehensively by the authorities by applying policies for the prevention and mitigation of desertification in the Mediterranean region.