Risk Assessment of Irrigation-Related Soil Salinization and Sodiﬁcation in Mediterranean Areas

: Salinization and sodiﬁcation are important processes of soil degradation a ﬀ ecting irrigated lands. A large proportion of the global irrigated area is a ﬀ ected by some degree of soil salinity or sodicity caused by the intensiﬁcation of irrigation. The increase of the frequency of adverse climatic conditions, like high temperatures and variations in precipitation patterns caused by climate change, will potentially amplify these processes in arid, semi-arid, and Mediterranean areas. The use of integrated approaches for the spatial and temporal prediction of the risk of salinization and sodiﬁcation in irrigated areas is of great value, helping in the decision-making regarding land uses and choice of more suitable agricultural practices. In this study, based on key criteria for the assessment of irrigation-related salinization processes (e.g., climate, topography, soil drainage, water quality for irrigation, and crop irrigation method), we developed a methodology for the prediction of soil salinity and sodicity risk in irrigated lands, using two composite indices, the Salinization Risk (RSA) index and the Sodiﬁcation Risk (RSO) index. The application of these indices to a real scenario (a Mediterranean area in Southern Portugal) showed that 67% of the potentially irrigated area presented a low risk of salinity development, 68% had a moderate risk of sodiﬁcation, and 16% was of high risk of sodicity development. Areas under moderate risk of salinization (26%) were mostly characterized by low slopes and ﬁne-textured soils, like Luvisols and Vertisols, with limited drainage conditions. Areas with high risk of soil sodiﬁcation presented a large incidence of low slope terrain, moderate-to-restricted soil drainage, in high clay content Luvisols, Vertisols and Cambisols, and land use dominated by annual crops irrigated with surface or sprinkler systems. These risk prediction tools have the potential to be used for resource use planning by policymakers and on-farm management decision by farmers, contributing to the sustainability of irrigated agriculture in Mediterranean regions.

models, aiming to extend their prediction capabilities (for e.g., [15,[26][27][28]). According to Zhou et al. [29], the use of risk indices or frameworks for soil salinity assessment may facilitate communication among scientists, policymakers, and practitioners by providing a basis for evaluating the environmental impact of agricultural practices and also government programs and policies. Frequently, these risk indices are GIS-based in order to ascertain priority areas, predict where it may occur in the future, and determine appropriate management options to mitigate salinity risk [30]. One of the earliest indices for the assessment of soil salinity risk was the Salinity Risk Index (SRI), proposed by Eilers et al. [31], later updated by Wiebe et al. [32] to evaluate the salinization risk trends on the Canadian Prairies. The SRI was composed of five risk factors: the current presence and extent of salinity, topography, soil drainage, aridity, and land use [31,32]. A four-component risk assessment framework for salinity prevention was proposed by Grundy et al. [33] for Queensland, Australia, where the components (current salinity stage, current land management, value of the assets under threat, and biophysical risk factors) interacted to give an overall salinity risk index. From this framework, Biggs et al. [30] developed a GIS-based system to assess the risk of salinity in the Condamine basin, Australia. Zhou et al. [29] presented a soil salinity risk assessment methodology by selecting a set of risk factors based on the conceptual Pressure-State-Response (PSR) sustainability framework of the Organisation for Economic Co-operation and Development (OECD) [34], where a set of fourteen risk factors were selected to develop a composite risk index for soil salinity in the Yinchuan Plain, China. Masoudi et al. [35] proposed a model for risk analysis of soil salinization for the Payab Basin, Iran, using GIS with the overlay of hazard maps of nine indicators as the causes of soil salinity (depth of water table, soil texture, slope, irrigation water quality, groundwater quality, efficacy of surface geology, climate, aridity index, status of soil salinity) to obtain a final risk map. Specifically for irrigated lands, similar salinization risk assessment methodologies were developed, using GIS, and considering irrigation water quality, soil properties, and climate indicators [15,24]. These methodologies have a dominant focus on the prediction of the development of saline soils, and the assessment of the risk of soil sodicity alone has generally received less attention [36]. Regardless of the modeling perspective or the soil deterioration problem to be predicted, the use of these integrated approaches evidenced its value for helping in decisionmaking regarding land resource management, allowing for the adoption of reclamation actions and the prediction of the spatial variation of irrigation-induced soil salinity and sodicity.
In this study, we aimed to: (i) present an overview of major criteria and respective indicators related with soil salinity and sodicity induced by irrigation; (ii) propose two composite risk indices, for soil salinization and soil sodification assessment in large areas, based on climate, irrigation water quality, soil information, and land-use data; (iii) present the risk indices when applied in a real scenario in a Mediterranean area in Southern Portugal.

Criteria and Indicators of Soil Salinization and Sodification
Indicators to predict soil salinization problems can be grouped following the common criteria for risk area identification according to soil threats [16] (Table 1), that considers a two-tiered approach, from a lower spatial resolution, less detailed and supported in available information, to a higher spatial resolution, requiring more detailed data and in situ measurements. Table 1. Common criteria for risk area identification according to soil threats considering a two-tiered approach (Tier 1 for National/Continental level; Tier 2 for regional level) (adapted from [16]).

Criteria
Indicator Data Source/Type of Information The most common soil salinity indicator is the electrical conductivity, usually measured in a saturation paste extract (ECe; dS m −1 ). Yield reduction of many crops can occur for ECe > 4 dS m −1 , a condition that defines a saline soil [37]. In saline soils, the large osmotic potential values on soil solution hinder plant transpiration, diminishing water availability to plants, which results in physiological deteriorations. Crop salt tolerance is the ability of plants to survive and produce profitable yields under the adverse conditions caused by soil salinity [38]. Data of tolerance to salinity were first reported by Maas and Hoffman [39] for numerous crops, by quantifying salt tolerance using the threshold (or breakpoint) model, that can be presented in terms of percentage of the maximum yield, known as the relative production function Equation (1): where Yr is the relative crop yield, with 100 being the maximum yield, a is the threshold (breakpoint) salinity (dS m −1 ), beyond which yield reduction will occur, ECe is the salinity of saturated soil extract, b is the slope, that represents the percentage of yield expected to be reduced for each unit of added salinity above the threshold salinity, and c is the level of soil salinity (dS m −1 ), above which the yield is zero. Values of a and b, and relative tolerance ratings for several crops, can be found in Maas and Grattan [40], indicating, e.g., that while wheat (Triticum aestivum L.) or olive (Olea europaea L.) are classified as moderately tolerant, grapevine (Vitis vinifera L.), almond (Prunus dulcis (Mill.) D.A. Webb) and a majority of vegetable crops are moderately sensitive to salinity, which increases the restrictions for their cultivation [41] and the need for appropriate irrigation management measures for these crops [42]. For sodic conditions, in addition to salinity diagnosis based upon ECe, the Sodium Adsorption Ratio (SAR) of the soil solution or of the irrigation water Equation (2) can be calculated [38]: ] are the concentrations (meq L −1 ) in the saturated soil paste extract or in the irrigation water. A sodic soil presents a SAR value of 13 or higher. When ECe > 4 dS m −1 and SAR > 13, the soil is classified as saline-sodic [37]. In sodic soils, the high exchangeable sodium in the soil complex causes dispersion of its colloids, structural destruction, and aggregate failure, resulting in unfavorable physical properties, like low infiltration rate and low hydraulic conductivity [9,13]. When the salinity of the soil water is insufficient to counteract the negative effects of adsorbed Na + on soil structure, a "rainfall effect" [43] can occur. More specifically, the leaching of salts by precipitation or excess irrigation water reduces the salinity of the soil solution, while the reduction in exchangeable Na + will be far smaller due to a buffer effect, performed by the exchangeable cations adsorbed in the soil matrix, causing, therefore, a reduction in the stabilizing effects of salinity on aggregate stability [5,22,44]. When dry, sodic soils become denser and structureless since natural aggregation is destroyed. At the soil surface, dispersed clay particles can act as adhesive, forming relatively dense crusts that impede seedling rooting and emergence. The degree of crusting depends on soil texture, the mineralogy of the clay, the exchangeable sodium content, the energy of raindrop impact, and the rate of drying [22]. At lower SAR levels, when chemical bonding is weakened, but no spontaneous dispersion takes place, inputs of energy are required for actual dispersion. While in drip irrigated fields this is hardly a problem, given the low flow rates of the drippers, the use of surface or sprinkler irrigation can increase the probability of soil structure degradation and surface crusting due to the physical disruption caused by runoff or by the water drops as they impact the soil surface aggregates [45].
In addition to soils that are classified as saline or sodic due to their genesis conditions, like Solonchaks and Solonetz, there are other soils whose characteristics make them more susceptible to the development of secondary salinization. This is the case of poorly drained soils (soils with permanent or temporary water table close to the surface) or soils with unfavorable hydrophysical properties (low permeability, low porosity, fine-textured layers or horizons). A systematized list of saline or saline prone soils, according to the World Reference Base of Soil Resources [46], was presented by Daliakopoulos et al. [22] (Table 2). Many of the laboratory and field methods used to determine soil hydraulic properties are time-consuming, laborious, and limited to the sampling quantity and location, which restricts their use due to soil heterogeneity [47,48]. In fact, texture variability is one of the determining factors of this heterogeneity, and several methods have been proposed to establish a relation between soil texture and soil hydraulic properties [47][48][49]. Following a methodology similar to de Paz et al. [15], a classification of soil drainage can be made considering its relationship with textural classes (Table 3). Drainage conditions are also affected by groundwater levels. Areas of low relief and particularly closed depressions encourage lateral translocation in surface and groundwater and salt accumulation [50]. Table 3. Drainage classes in relation with textural class and saturated hydraulic conductivity (systematized after [15] and [47][48][49]). For the risk assessment of irrigation-induced salinization and sodification, the evaluation of irrigation water quality is of primary importance since it often constitutes the main source of salts. Some of the parameters used to evaluate soil salinization problems are the same that are used to evaluate the irrigation water quality, namely, electrical conductivity, in this case, Water Electrical Conductivity (EC W ; dS m −1 ), Total Dissolved Salts (TDS; %), SAR, and the concentration of some specific ions that might cause toxicity to crops.
Although the suitability of a saline water for irrigation depends on different conditions of use, like crop, climate, soil, irrigation method, and management practices, only very tolerant crops can have satisfactory yields if irrigated with waters with EC that exceeds 10 dS m −1 . In fact, few normally used irrigation waters exceed EC of about 2 dS m −1 [51]. It is consensual the use of the Food and Agriculture Organization of the United Nations (FAO) water quality guidelines for irrigation [52], which considers three levels of restriction for the use of an irrigation water: none, slight to moderate, and severe. In these guidelines, salinity is assessed from EC W , and sodicity (more specifically, infiltration rate of water in the soil) is assessed using EC W and SAR together (Table 4). 1) Based on [52]: None-no soil or cropping problems are experienced; Slight to moderate -gradually increasing care in selection of crop and management alternatives is required if full yield potential is to be achieved; Severe-there will be soil and cropping problems or reduced yields, but even with cropping management designed especially to cope with poor quality water, a high level of management skill is essential for acceptable production; 2) Affects crop water availability; 3) Affects infiltration rate of water into the soil.
Irrigated agriculture develops primarily in areas where the interannual and seasonal quantity and distribution of precipitation and evapotranspiration are irregular. This leads to permanent (aridity) or temporary (drought) imbalances in water availability consisting in low average annual rainfall, with high spatial and temporal variability resulting in overall low moisture [53]. These features contribute to unfavorable salt balances, which can be intensified by irrigation with poor-quality water [54]. Climate plays a key role in the process and, whenever data are available, should be considered when defining a set of indicators for salinization/sodification risk prediction. A well-known indicator used as a numerical expression of dryness for climate classification and for the identification of regional patterns of aridity is the United Nations Environmental Programme (UNEP) Aridity Index (AI) [55], a mean accumulated annual precipitation/potential evapotranspiration ratio (P/PET). Considering the AI scale, the classification of climate types is hyper-arid (<0.05), arid (0.05-0.2), semi-arid (0.2-0.5), dry sub-humid (0.5-0.65), and humid (≥0.65). For dryland climate subtypes, AI values are below 0.65, and, according to Schofield and Kirkby [50], values between 0.05 and 1 (arid to moist sub-humid) describe climatic conditions suitable for salinization to occur.

Study Area
The study region selected for this work, which we will refer to as Inland Alentejo, is part of a province in Southern Portugal, with a Mediterranean-type climate, named Alentejo ( Figure 1). The region has a total area of 11,260 km 2 , of which 10,114 km 2 are predominantly of agriculture and/or agroforestry use (excluding urban areas, artificial zones and rocky terrain). Large areas of traditional rainfed agriculture, conditioned by water availability constraints, have been gradually replaced by irrigated agriculture [56]. While there are some private irrigation systems, the main source of irrigation water is public. In this area, small public irrigation schemes with a total area of approximately 30,000 ha in operation in 2018 [57] were implemented in the last century. However, the largest public scheme, which started to operate in 2002, is the Alqueva irrigation plan, that presently covers nearly 120,000 ha and is expected to expand to 170,000 ha by 2022, totaling 15% of the study area. The irrigation plan is a part of the Alqueva Multi-Purpose Development Project (EFMA-Emprendimento de Fins Múltiplos de Alqueva). The EFMA is centered in the Alqueva reservoir (Guadiana River Basin), the largest artificial lake in the Iberian Peninsula, with a total storage capacity of 4150 hm 3 , providing water for public supply, irrigation, industry uses, energy production, and tourism. Studies carried out in the water and sediments of this reservoir have reported ecotoxicological risk to aquatic organisms, induced mostly by pesticide and metal pollution, an increment over time of organic matter and nutrients in the water body, and of trace elements in the sediments [58][59][60][61]. More recently, a temporal trend of degradation in water quality for irrigation was reported [11,12], that could be attributed to the expansion of the irrigation area, with the concomitant intensification of agriculture, and to the increase of drier conditions resulting from climate change. These studies pointed to a potential increased risk of soil infiltration problems and of salinity-induced yield reductions in sensitive crops, especially in drought years, and during periods of high atmosphere evaporative demands.
The climate in the region is classified as temperate with hot and dry summer (Csa in Köppen Classification), with a small area of mid-latitude steppe (Bsk). The long-term means for the period 1981-2010 of annual precipitation and annual mean temperature are, respectively, 558 mm and 16.9 • C, for Beja, a main town located approximately in the center of the study area [62]. During 2018, the annual precipitation in Beja was 653 mm, mean temperature was 16.2 • C, and reference annual evapotranspiration totaled 1193 mm [63].
Predominant soils in the study area (approximately 81% of total area) are Luvisols, Leptosols, Regosols and Vertisols. For the purpose of this study, a map of Reference Soil Groups (RSG) was built for the study region, based on the World Reference Base for soil resources [46] and on previous studies [64,65] ( Figure S1). , was used to obtain the land occupation and use map of the study area ( Figure S2). The minimum cartographic unit of COS is of 1 ha, with a minimum distance between lines of 20 m, and it comprises 83 classes, from which 11 distinguish different agricultural types. According to COS, the main agricultural land uses in the region in 2018 were temporary rainfed and irrigated crops, and olive groves, respectively 22% and 15% of the study area. When considering only the Alqueva irrigation area, the main irrigated crops grown in 2018 were (by descending order of importance): olive; fruit trees, mainly nuts like almond and walnut (Juglans regia L.), but also pear (Pyrus communis L.) or peach (Prunus persica (L.) Batsch); grapevine; maize (Zea mays L.); forage crops; oil crops; horticultural crops; cereals other than maize [67]. The non-irrigated land is occupied mostly by rainfed crops, pastures and agroforestry systems of holm oak and cork oak.
Given the importance of the EFMA irrigation area in the study region, the water quality data used for the study were collected in the Alqueva reservoir at four sampling sites in floating platforms, as shown in Figure 1 . The Al site was chosen because it was near the Álamos pumping station, an inlet where water is elevated to part of the EFMA conveyance network. The Mn site was selected to ensure representativeness of the reservoir area. Mr and Lf sites were chosen considering previous results for the reservoir water quality assessment [12,58].

Data Collection and Mapping Methodology
In order to compute a spatial distribution of the Aridity Index, ERA5 dataset of monthly mean averages values was used. ERA5 [68] is the fifth-generation reanalysis product provided by the European Centre for Medium-Range Weather Forecasts (ECMWF). Reanalysis combines observations and a numerical weather prediction model to produce complete atmospheric fields over a regular 3D atmospheric grid. In the present study, the monthly mean values of total precipitation, 2 m air temperature (T 2m ; • C), and 2 m dew-point temperature (d 2m ; • C) for the 30 year period 1990-2019 were downloaded from the ERA5-Land monthly averaged dataset with a horizontal resolution of 0.1 • (~10 km) [69]. The monthly accumulated PET values were computed using Equation (3), the practical and well-known formulation of Romanenko [70]: where T a ( • C) is the monthly mean air temperature and RH (%) is the monthly mean relative humidity calculated from the ERA-5 T 2m and d 2m values. Finally, the 30 years (1990-2019) mean annual PET and P were computed for the study area to access the AI as defined by the UNEP [55].
The map with the Reference Soil Groups (RSG) in the study area is presented in Figure S1. Slope classes were obtained based on the slope of the predominant soil units [71].
From the textural classes of the surface horizons/layers of the predominant soil units [72], a map with drainage classes was built after the classification presented in Table 3, with the addition of a restricted drainage classification whenever soil use capacity was severely to very severely limited due to excess water.
Within the study area, a potentially irrigated area was established for the application of the risk indices. For this purpose, the following crops/type of crops from the COS classes were considered as possibly irrigated: a-temporary dry and irrigated crops; b-vineyards; c-orchards; d-olive groves; e-complex cultivation patterns. A most likely irrigation method was assigned to each crop/group of crops as follows: I-Drip irrigation, in the case of b, c, and d; II-Surface or Sprinkler irrigation, in the case of a and e.
The study covered six water sampling campaigns in 2018, performed in January (Jan), March (Mar), May, July (Jul), September (Sep), and November (Nov). At each sampling location, 2 L of surface water were collected, at a depth of 50 cm, using a Van Dorn bottle. The water samples were stored in polyethylene bottles and transported to the laboratory, in a cooler at 4 • C, where they were conserved and stored, following the requisites for water conservation for each parameter [73]. The concentrations of Ca 2+ , Mg 2+ , and Na + , were determined by ionic chromatography methodology adapted from APHA [73]. Water EC W was measured in situ on the same dates (except November due to a temporary equipment failure) at a 50 cm depth using a multiparametric probe YSI 6820 MPS probe ® . Values of SAR were calculated using Equation (2). For the purpose of this work, ECW and SAR were evaluated after the FAO guidelines presented in Table 4.

Risk Indices
To obtain the composite risk indices of salinization and sodification, two sets of indicators were adopted, based on a tier 2 approach [16], previously presented in Table 1, and based on Masoudi et al. [35]. The Salinization Risk (RSA) index includes a set of four criteria and respective indicators, represented in classes (I to V or I to III, depending on the range of values or the classes considered), to which a rating score (1 to 5) was assigned (Table 5): (i) Climate, quantified with the AI scale from [55]; (ii) Topography, represented by slope intervals obtained from the Portuguese soil and land-use capacity map [71]; (iii) Soil drainage, characterized by drainage classes according to Table 3; (iv) Irrigation water quality, from in situ measurements, classified according to the degrees of restriction to its use with respect to salinity, as presented in Table 4. The Sodification Risk (RSO) index comprises a set of five criteria, and respective indicators, to which a rating score (1 to 5) was assigned ( Table 6): AI-Aridity Index; EC W -Irrigation water electrical conductivity; SAR-Sodium Adsorption Ratio. 1) According to the Portuguese soil and land-use capacity map [71]. 2) According to Table 3. 3) According to Table 4.
(i) to (iii) equal to the ones considered in the RSA index; (iv) crop, represented by a most likely irrigation method, considering two classes (I-drip irrigation and II-surface or sprinkler irrigation); (v) Irrigation water quality, from in situ measurements and sample analyses, classified according to the degrees of restriction to its use with respect to infiltration, as presented in Table 4.
The RSA and RSO indices incorporated the effect of the indicators in each set, obtained by their score sum, using Equations (4) and (5): where S i is the rating score assigned to each of the indicators class. From the RSA and RSO values, five risk classes were considered (Very Low, Low, Moderate, High and Very High), each corresponding to a level of management practice recommendation (Table 7), an adaptation of the classification described in Wiebe et al. [32]. In the case of RSA, a condition was added to assign a Very High risk class (RSA > 18) whenever soils were Solonchaks, regardless of the remaining indicators scores. Likewise, in the case of RSO, a condition was added to assign a Very High risk class (RSA > 22) whenever soils were Solonetz, regardless of the remaining indicators scores.    [75], evaluating climate change potential impacts on irrigated agriculture in the Guadiana River Basin, projected a general increase in crop irrigation requirements, questioning the sustainability of irrigation area expansion in the region. Godinho et al. [76] reported the risk of increased aridity in the area resulting from changes in land surface albedo and land surface temperature caused by land cover transformations, such as deforestation to promote agricultural lands. In relation to the topography criterion, more than half of the area (60%) presented slopes < 8%; thus, was grouped in slope classes II and III (Figure 3). Considering the drainage class, 73% of the area was in class II and 7% in class III, which related with the prevalence of medium-to-fine textured soils in the region (Figure 4). According to the assumptions of the model, the potentially irrigated A study by Valverde et al. [75], evaluating climate change potential impacts on irrigated agriculture in the Guadiana River Basin, projected a general increase in crop irrigation requirements, questioning the sustainability of irrigation area expansion in the region. Godinho et al. [76] reported the risk of increased aridity in the area resulting from changes in land surface albedo and land surface temperature caused by land cover transformations, such as deforestation to promote agricultural lands.
In relation to the topography criterion, more than half of the area (60%) presented slopes < 8%; thus, was grouped in slope classes II and III (Figure 3). Considering the drainage class, 73% of the area was in class II and 7% in class III, which related with the prevalence of medium-to-fine textured soils in the region (Figure 4). According to the assumptions of the model, the potentially irrigated area totaled 401,408 ha, corresponding to about 40% of the agricultural and forestry land. In this area, 43% is most likely irrigated with drip systems and the remaining with sprinkler or surface systems ( Figure 5). When considering only the potential irrigated area, some of the percentages previously presented had some significant variation: (i) the percentage of Leptosols varied from 17% in the total area to 3% in the potentially irrigated area, indicating a low agricultural, and accordingly, low irrigation aptitude of this reference soil group; (ii) slopes assigned to class II and III increased, respectively, from 46% to 49% and from 14% to 17%; (iii) the areas with restricted drainage (class III) increased from 7% to 12%; (iv) temporary crops and olive groves remained the predominant COS classes but with higher occupation within the potentially irrigated area of 57% and 37%, respectively. However, it may be that the percentage of temporary crops was overestimated since the COS class includes rainfed and irrigated crops and, from the information reported in EDIA (Empresa de Desenvolvimento e Insfraestruturas do Alqueva) [67], olive groves occupy a large proportion of the irrigated lands in the region.   When considering only the potential irrigated area, some of the percentages previously presented had some significant variation: (i) the percentage of Leptosols varied from 17% in the total area to 3% in the potentially irrigated area, indicating a low agricultural, and accordingly, low irrigation aptitude of this reference soil group; (ii) slopes assigned to class II and III increased, respectively, from 46% to 49% and from 14% to 17%; (iii) the areas with restricted drainage (class III) increased from 7% to 12%; (iv) temporary crops and olive groves remained the predominant COS classes but with higher occupation within the potentially irrigated area of 57% and 37%, respectively. However, it may be that the percentage of temporary crops was overestimated since the COS class includes rainfed and irrigated crops and, from the information reported in EDIA (Empresa de Desenvolvimento e Insfraestruturas do Alqueva) [67], olive groves occupy a large proportion of the irrigated lands in the region.

Water Quality for Irrigation
The temporal variation observed for ECW and SAR for the four sampling sites is presented in Figure 6a,b. In the case of ECW, the highest average values were measured in September (0.56 ± 0.015 dS m −1 ) and July (0.49 ± 0.020 dS m −1 ), during the dry season. As for SAR, the highest values were registered at Álamos in January (1.64 ± 0.000), which was probably due to the 2017 and early 2018 drought that lead to low water storage and, possibly, reduced salt dissolution. The lowest SAR value occurred at the same site in July (1.12 ± 0.010). There were no relevant differences between sites.  Table 4 (green dashed line is the limit value for salinity problems and purple lines are the limit values for infiltration problems). Al-Álamos; Mn-Montante; Mr-Mourão; Lf-Lucefécit. Jan-January; Mar-March; Jul-July; Sep-September; Nov-November.
According to the methodology of salinization and sodification risk prediction presented, the irrigation water quality indicators fall in the same classes for every date and site. ECW values showed no degree of restriction of use, thus were grouped in class I ( Table 5). The combined assessment of ECW and SAR showed a slight-to-moderate degree of restriction of use in all water samples analyzed. Therefore, these sodification indicators fall within class II for every date and site (Table 6). In this When considering only the potential irrigated area, some of the percentages previously presented had some significant variation: (i) the percentage of Leptosols varied from 17% in the total area to 3% in the potentially irrigated area, indicating a low agricultural, and accordingly, low irrigation aptitude of this reference soil group; (ii) slopes assigned to class II and III increased, respectively, from 46% to 49% and from 14% to 17%; (iii) the areas with restricted drainage (class III) increased from 7% to 12%; (iv) temporary crops and olive groves remained the predominant COS classes but with higher occupation within the potentially irrigated area of 57% and 37%, respectively. However, it may be that the percentage of temporary crops was overestimated since the COS class includes rainfed and irrigated crops and, from the information reported in EDIA (Empresa de Desenvolvimento e Insfraestruturas do Alqueva) [67], olive groves occupy a large proportion of the irrigated lands in the region.

Water Quality for Irrigation
The temporal variation observed for EC W and SAR for the four sampling sites is presented in Figure 6a,b. In the case of EC W , the highest average values were measured in September (0.56 ± 0.015 dS m −1 ) and July (0.49 ± 0.020 dS m −1 ), during the dry season. As for SAR, the highest values were registered at Álamos in January (1.64 ± 0.000), which was probably due to the 2017 and early 2018 drought that lead to low water storage and, possibly, reduced salt dissolution. The lowest SAR value occurred at the same site in July (1.12 ± 0.010). There were no relevant differences between sites.
According to the methodology of salinization and sodification risk prediction presented, the irrigation water quality indicators fall in the same classes for every date and site. EC W values showed no degree of restriction of use, thus were grouped in class I ( Table 5). The combined assessment of EC W and SAR showed a slight-to-moderate degree of restriction of use in all water samples analyzed. Therefore, these sodification indicators fall within class II for every date and site (Table 6). In this case, low-salt water may reduce infiltration even for a low SAR, given that the salinity of the applied water is insufficient to counteract the negative effects of adsorbed sodium on soil structure [51,52,77].
The temporal variation observed for ECW and SAR for the four sampling sites is presented in Figure 6a,b. In the case of ECW, the highest average values were measured in September (0.56 ± 0.015 dS m −1 ) and July (0.49 ± 0.020 dS m −1 ), during the dry season. As for SAR, the highest values were registered at Álamos in January (1.64 ± 0.000), which was probably due to the 2017 and early 2018 drought that lead to low water storage and, possibly, reduced salt dissolution. The lowest SAR value occurred at the same site in July (1.12 ± 0.010). There were no relevant differences between sites.  Table 4 (green dashed line is the limit value for salinity problems and purple lines are the limit values for infiltration problems). Al-Álamos; Mn-Montante; Mr-Mourão; Lf-Lucefécit. Jan-January; Mar-March; Jul-July; Sep-September; Nov-November.
According to the methodology of salinization and sodification risk prediction presented, the irrigation water quality indicators fall in the same classes for every date and site. ECW values showed no degree of restriction of use, thus were grouped in class I ( Table 5). The combined assessment of ECW and SAR showed a slight-to-moderate degree of restriction of use in all water samples analyzed. Therefore, these sodification indicators fall within class II for every date and site (Table 6). In this Figure 6. Spatial and temporal variation of EC W (a) and SAR (b), during 2018, at the Alqueva reservoir. Dots represent single in situ measurements. Bars represent mean and whiskers represent the standard deviation for three repetitions Dashed lines correspond to the limit values from FAO water quality standards, as presented in Table 4 (green dashed line is the limit value for salinity problems and purple lines are the limit values for infiltration problems). Al-Álamos; Mn-Montante; Mr-Mourão; Lf-Lucefécit. Jan-January; Mar-March; Jul-July; Sep-September; Nov-November.

Salinization Risk Areas
The risk map of soil salinization in the potentially irrigated lands of Inland Alentejo shows that there was a predominance of low (67%) and moderate risk (26%) areas and that there were no areas within the higher risk classes (Figure 7). case, low-salt water may reduce infiltration even for a low SAR, given that the salinity of the applied water is insufficient to counteract the negative effects of adsorbed sodium on soil structure [51,52,77].

Salinization Risk Areas
The risk map of soil salinization in the potentially irrigated lands of Inland Alentejo shows that there was a predominance of low (67%) and moderate risk (26%) areas and that there were no areas within the higher risk classes (Figure 7). Moderate risk of salinization was predominantly related to low slope areas (Figure 8a). Moderate risk of salinization was predominantly related to low slope areas (Figure 8a). Also, moderate and restricted soil drainage, a characteristic of fine-textured soils, like Luvisols or Vertisols, (Figure 8b,c) lead to the augmentation of soil salinization risk. Similar relationships between slope and salinization-prone areas were obtained by other authors [78] when mapping the irrigation salinity risk in an arid region. Seydehmet et al. [78] found high salinization risk associated with flat and low terrain, high salt content in the subsoil, high salinity of groundwater and small depth of the water table. The development of soil salinity as a consequence of poor drainage conditions, fine texture and low saturated soil hydraulic conductivity was reported by de Paz et al. [15] and Bouksila et al. [79], in their works about salinization risk in semi-arid irrigated land. The influence of soil water storage capacity was referred to as the most important soil indicator affecting soil salinization by Kairis et al. [80], indicating that moderately fine and fine-textured soils were more likely to be affected by salinization.
As expected, the main types of land use where this level of risk appeared were olive groves and annual crops (Figure 8d), which was mainly a consequence of their extensive distribution in the area. In the case of irrigated high-density olive groves, largely grown in the region, Ramos et al. [81] found that the long-term (30 years) impact of current irrigation practices (including deficit irrigation, i.e., seasonal irrigation volumes corresponding to a fraction of the seasonal crop evapotranspiration) showed reduced risks of soil salinization, with simulated average root zone salinity values at the end of the irrigation seasons (normally, late summer or early autumn) remaining below the salinity threshold limit for olive groves. Additionally, these authors stated that the use of deficit irrigation practices in olive groves has the potential to improve soil salinity management by a better control of rising water tables and by a reduction in the import of salts by irrigation water, but it does not provide the same degree of leaching as full irrigation.

Sodification Risk Areas
The results obtained for the RSO index showed that most of the potentially irrigated area in Inland Alentejo is classified as moderate risk (68%), followed by high risk (16%) and low risk (10%) of sodification (Figure 9). Predictions were decidedly influenced by the indicators slope, drainage, water quality, and irrigation method. Moderate risk areas comprised mainly medium slopes (Figure Also, moderate and restricted soil drainage, a characteristic of fine-textured soils, like Luvisols or Vertisols, (Figure 8b,c) lead to the augmentation of soil salinization risk. Similar relationships between slope and salinization-prone areas were obtained by other authors [78] when mapping the irrigation salinity risk in an arid region. Seydehmet et al. [78] found high salinization risk associated with flat and low terrain, high salt content in the subsoil, high salinity of groundwater and small depth of the water table. The development of soil salinity as a consequence of poor drainage conditions, fine texture and low saturated soil hydraulic conductivity was reported by de Paz et al. [15] and Bouksila et al. [79], in their works about salinization risk in semi-arid irrigated land. The influence of soil water storage capacity was referred to as the most important soil indicator affecting soil salinization by Kairis et al. [80], indicating that moderately fine and fine-textured soils were more likely to be affected by salinization.
As expected, the main types of land use where this level of risk appeared were olive groves and annual crops (Figure 8d), which was mainly a consequence of their extensive distribution in the area. In the case of irrigated high-density olive groves, largely grown in the region, Ramos et al. [81] found that the long-term (30 years) impact of current irrigation practices (including deficit irrigation, i.e., seasonal irrigation volumes corresponding to a fraction of the seasonal crop evapotranspiration) showed reduced risks of soil salinization, with simulated average root zone salinity values at the end of the irrigation seasons (normally, late summer or early autumn) remaining below the salinity threshold limit for olive groves. Additionally, these authors stated that the use of deficit irrigation practices in olive groves has the potential to improve soil salinity management by a better control of rising water tables and by a reduction in the import of salts by irrigation water, but it does not provide the same degree of leaching as full irrigation.

Sodification Risk Areas
The results obtained for the RSO index showed that most of the potentially irrigated area in Inland Alentejo is classified as moderate risk (68%), followed by high risk (16%) and low risk (10%) of sodification (Figure 9). Predictions were decidedly influenced by the indicators slope, drainage, water quality, and irrigation method. Moderate risk areas comprised mainly medium slopes (Figure 10a), moderate drainage soils included in the Luvisols RSG (Figure 10b,c), annual crops irrigated by surface or sprinkler systems and, to a lesser extent, olive orchards irrigated by drip systems (Figure 10d,e).
Water 2020, 12, x FOR PEER REVIEW 16 of 23 10a), moderate drainage soils included in the Luvisols RSG (Figure 10b,c), annual crops irrigated by surface or sprinkler systems and, to a lesser extent, olive orchards irrigated by drip systems (Figure  10d,e).   The analysis of the relationships between the RSO index and its indicators in high-risk areas showed a strong prevalence of flat and low slope terrain (Figure 11a), moderate-to-restricted soil drainage in Luvisols, Vertisols and Cambisols (Figure 11b,c), and annual crops irrigated with surface or sprinkler systems (Figure 11d,e). The results were in accordance with Sentís [36], who, in their study to evaluate the development of secondary sodicity in the surface soil, found that it was strongly affected by inadequate drainage and inefficient irrigation. This risk prediction could indicate that even non-sodic soils with ESP (or SAR) < 3 can have a tendency for swelling, aggregate failure, and increase in dispersion rates, thus behaving like sodic soils at very low electrolyte concentration [82]. As a consequence, soil infiltration rates decline and surface crust formation problems can occur, conditions that may worsen due to the physical disruption of soils particles caused by the use of sprinkler irrigation systems. Although clay soils are more affected by sodicity, it should be considered that there are some soil properties, other than texture, that are significant in influencing the behavior of soils under sodification risk, namely, depth and thickness of the sodic horizons and clay mineralogy. For instance, the presence of high-activity clay minerals, like smectite, is likely to increase the soil susceptibility to sodification [83]. The analysis of the relationships between the RSO index and its indicators in high-risk areas showed a strong prevalence of flat and low slope terrain (Figure 11a), moderate-to-restricted soil drainage in Luvisols, Vertisols and Cambisols (Figure 11b,c), and annual crops irrigated with surface or sprinkler systems (Figure 11d,e). The results were in accordance with Sentís [36], who, in their study to evaluate the development of secondary sodicity in the surface soil, found that it was strongly affected by inadequate drainage and inefficient irrigation. This risk prediction could indicate that even non-sodic soils with ESP (or SAR) < 3 can have a tendency for swelling, aggregate failure, and increase in dispersion rates, thus behaving like sodic soils at very low electrolyte concentration [82]. As a consequence, soil infiltration rates decline and surface crust formation problems can occur, conditions that may worsen due to the physical disruption of soils particles caused by the use of sprinkler irrigation systems. Although clay soils are more affected by sodicity, it should be considered that there are some soil properties, other than texture, that are significant in influencing the behavior of soils under sodification risk, namely, depth and thickness of the sodic horizons and clay mineralogy. For instance, the presence of high-activity clay minerals, like smectite, is likely to increase the soil susceptibility to sodification [83].

Conclusions
Negative impacts of soil salinity and sodicity in Mediterranean regions include decreased food production, deteriorating water quality in surface streams and groundwater, loss of biodiversity, and increased risk of flooding and desertification. The present study aimed to develop two risk indices for better prediction of soil salinity and sodicity in large areas, in order to provide farmers, rural planners, and resource managers with a tool to help identify susceptible areas, to apply appropriate agronomical strategies and management options and to avoid soil degradation in irrigated lands.
The application of the Salinization Risk (RSA) index to Inland Alentejo indicated that areas under moderate risk, where land-use practices should be monitored carefully, showed a prevalence of low slopes and fine-textured soils like Luvisols and Vertisols with limited drainage conditions. The Sodification Risk (RSO) index calculations estimated a predominance of moderate-and high-risk potentially irrigated lands, corresponding to management practices which leave land exposed to agri-environmental degradation or are conductive to agri-environmental degradation. High-risk areas had a large incidence of low slope terrain, moderate-to-restricted soil drainage in high clay content Luvisols, Vertisols and Cambisols, and annual crops irrigated with surface or sprinkler systems.

Conclusions
Negative impacts of soil salinity and sodicity in Mediterranean regions include decreased food production, deteriorating water quality in surface streams and groundwater, loss of biodiversity, and increased risk of flooding and desertification. The present study aimed to develop two risk indices for better prediction of soil salinity and sodicity in large areas, in order to provide farmers, rural planners, and resource managers with a tool to help identify susceptible areas, to apply appropriate agronomical strategies and management options and to avoid soil degradation in irrigated lands.
The application of the Salinization Risk (RSA) index to Inland Alentejo indicated that areas under moderate risk, where land-use practices should be monitored carefully, showed a prevalence of low slopes and fine-textured soils like Luvisols and Vertisols with limited drainage conditions. The Sodification Risk (RSO) index calculations estimated a predominance of moderate-and high-risk potentially irrigated lands, corresponding to management practices which leave land exposed to agri-environmental degradation or are conductive to agri-environmental degradation. High-risk areas had a large incidence of low slope terrain, moderate-to-restricted soil drainage in high clay content Luvisols, Vertisols and Cambisols, and annual crops irrigated with surface or sprinkler systems.
Although the RSA and RSO indices have been proposed for a risk prediction purpose, and despite the fact that their validation with soil sampling or with in situ measurements with proximal sensors was outside the scope of this work, more precise knowledge is very important, especially in areas identified as of moderate and high risk. In this sense, it is our goal to develop work to validate the methodology presented in order to identify assumptions that must be adjusted or to integrate indicators of the current state of soil salinity and sodicity. Further possible developments of this methodology also include the application of these indices to different climate change scenarios, assessing the impact of expected increased aridity and decreased irrigation water quality on the risk of development of soil salinization and sodification processes in irrigated Mediterranean areas.
The proposed risk indices integrated climate, topography, soil, water quality, and crop indicators that are easily available for large scale assessments and territorial planning but can be applied at the farm level scale, encouraging farmers to get to know their land better, implement soil monitoring and adopt agronomic strategies that contribute to the sustainability of irrigated agriculture.