Nutrition Quality Parameters of Almonds as Affected by Deficit Irrigation Strategies

The influence of full irrigation, double-regulated (RDI) and sustained deficit irrigation (SDI) treatments on almond quality was assessed by analyzing different parameters: sugars, organic acids, antioxidant activity, total phenolic content (TPC), and volatile compounds. Almond quality studies for plants submitted to water stress are scarce, and it is essential to understand the biochemical responses of plants to water stress in maintaining fruit yield and quality. Citric acid, sucrose, antioxidant activity, and TPC were not affected by the application of studied deficit irrigation strategies (DI). An increase in malic acid and a decrease in glucose was observed for stressed samples (T3 and T4), while a higher number of total volatiles compounds was found for moderate RDI (T2). Using deficit irrigation strategies, the almond yield and quality was not changed, and in fact, some parameters, such as glucose and key volatile compounds, slightly increased under moderate RDI. This finding might encourage farmers to implement these strategies and contribute to sustainable agriculture.


Introduction
Fresh (non-salty and adequate for irrigation) water is a limited resource and the uncertainty of the remaining amount for the next generation has it this a dramatic global risk factor [1,2]. Agriculture is the primary use of fresh water worldwide because more than 40% of food production comes from irrigated fields, and it is in a weak position due to its susceptibility to weather (temperatures and precipitation) changes [3]. Mediterranean agriculture is a perfect model of arid and semiarid farming, The implementation of deficit irrigation DI strategies was possible by controlling the crop water status through midday stem water potential. As observed in Figure 1b,c, T1 showed the lowest levels of stress with a value of −1.55 MPa for the minimum stem water potential (SWP) and a stress integral (SI) of 54.2 MPa × day, while T3 (severe RDI) displayed the highest levels of stress for both parameters (MinSWP = −2.08 MPa; SI = 94.9 MPa × day).

Sugars and Organic Acids
Sucrose was the main soluble sugar found in almonds, representing 85-91% of the total carbohydrates, due to its accumulation during ripening and because many other reducing sugars are substrates for its synthesis [25]. Table 1 shows that the results of the sucrose content ranged from 33 to 35 g kg −1 , being significantly similar among treatments. Previous studies reported a sucrose increase with irrigation (total water supply of ~10 m 3 tree −1 during summer) in almonds from "Ferragnes", "Texas", and "Guara" genotypes, but different irrigation strategies were applied [20,25]. On the contrary, other authors observed a higher sucrose content in non-irrigated almonds from "Marta" variety and linked it to the plant adaptive physiological responses to water stress, while a significant similarity between those irrigated at full (110%) ET, 50% ET periodically supplied to only one side of the root system for the entire year (PRD), and RDI (irrigation at 100% of ET for the entire year and at 30% of ET from early June until harvest) was observed [26]. Almond yield (a), minimum stem water potential (b), and water stress integral (c) for each treatment. Bars with the same letter were not significantly different ((a) p < 0.001; (b) p < 0.01; (c) p < 0.001), according to Tukey's least significant difference test.

Sugars and Organic Acids
Sucrose was the main soluble sugar found in almonds, representing 85-91% of the total carbohydrates, due to its accumulation during ripening and because many other reducing sugars are substrates for its synthesis [25]. Table 1 shows that the results of the sucrose content ranged from 33 to 35 g kg −1 , being significantly similar among treatments. Previous studies reported a sucrose increase with irrigation (total water supply of~10 m 3 tree −1 during summer) in almonds from "Ferragnes", "Texas", and "Guara" genotypes, but different irrigation strategies were applied [20,25]. On the contrary, other authors observed a higher sucrose content in non-irrigated almonds from "Marta" variety and linked it to the plant adaptive physiological responses to water stress, while a significant similarity between those irrigated at full (110%) ET, 50% ET periodically supplied to only one side of the root system for the entire year (PRD), and RDI (irrigation at 100% of ET for the entire year and at 30% of ET from early June until harvest) was observed [26]. Glucose, fructose, sorbitol, raffinose, and inositol were other monosaccharides found previously in almonds [25], however in this study, glucose and fructose were the only reducing sugars identified. While fructose was found only in trace amounts (data not shown), glucose content presented statistically significant higher values for T1 and T2 (moderate RDI) treatments. Sánchez Bel et al. [25] in studies about almonds ("Guara" genotype), and Nahar et al. [27] in tomatoes, concluded that water stress enhanced the sweetness of tomatoes by increasing their glucose content. These results might be related to the osmotic adjustment, which can be activated by accumulation of solutes (rich in hydroxyl (-OH) groups, such as sugars, proline, etc.) in the cytoplasm under stress conditions. This biochemical mechanism aids plants in naturalizing to dry and saline conditions by protecting the cellular membrane, protein, and enzymes against dehydration [28]. Consequently, the osmotic adjustment enhances the capacity to maintain positive turgor, increasing the sugars and organic acid [27]. Egea et al. [6] and Cornacchia et al. [26] obtained no statistically significant differences among fully irrigated and DI strategies in almond trees.
The main identified organic acids (Table 1) were citric and malic. Citric acid was not affected by DI, as other authors have previously reported [6]. On the contrary, 0.4% increase of citric acid content was observed in drip-irrigated "Guara" almonds (16.8 m 3 year −1 ) with respect to non-irrigated almonds [25]. An increase of 2.4 and 4.2 g kg −1 of malic acid content for T3 and T4 (SDI), respectively, was observed; this experimental finding agreed with previous studies on tomatoes and grapes, in which water stress enhanced their quality by raising the concentration of important organic acids [27]. However, a decrease of malic acid content in "Guara" almonds under non-irrigated conditions [11] and in "Marta" almonds under partial root zone (PRD) conditions (irrigation supplied at 50% ET during the whole growing season) was observed [25]. Malic acid was stable for moderate RDI and similar results were reported in "Marta" almonds under RDI (trees were irrigated at 50% ET during kernel filling and at 100% ET for the rest growth period) [6]. Studies in organic acid content in almonds under water stress conditions are scarce, although it seems that organic acids may play a limited role in the quality of almonds.
The TPC found in this study for raw kernel (5.5 g gallic acid equivalents (GAE) kg −1 , mean value of all treatments) was higher than those previously reported by other authors (range from 0.6 to 1.9 g GAE kg −1 ) but lower than that reported by Lin et al. [29] (7.5 g GAE kg −1 ) for the almonds from Almond Board of California. The difference could be attributed to factors such as variety, geographical area, or agricultural practices [21,30]. Blanched kernel TPC (0.5 g GAE kg −1 ) was similar to that reported in literature (0.7 g GAE kg −1 ) for almonds in general [30]. Finally, a mean TPC value of 13 g GAE kg −1 was found on almond skin and similar values were reported for American almonds (11-17 g GAE kg −1 ), but lower values were found for Spanish almonds (26 g GAE kg −1 ) [22].
The total dietary intake of polyphenols is estimated to be around 1 g [31]. The results of the present study showed that, for example,~30 g of almonds (~20 kernels of "Vairo" genotype cultivated in Sevilla, Spain) provides~170 mg of total phenols and represents almost 20% of the total dietary polyphenols intake.
Antioxidant activity and TPC were not significantly affected by the irrigation treatment in the raw almond. Regarding the blanched almonds, T4 was found to have the lowest antioxidant activity measured with ABTS and the lowest TPC. Blanched samples from the moderate RDI (T2) showed the highest TPC and the antioxidant activity was similar to the control (for the ABTS and DPPH methods). Regarding the almond skin, the TPC and antioxidant activity (ABTS and DPPH) were not affected by the irrigation treatments, except for FRAP method, which showed that T2 was similar to the control and T4 had the lowest antioxidant activity. These results agreed with Cano Lamadrid et al. [32] and Sánchez-Rodríguez et al. [12] in studies investigating effects of deficit irrigation on quality and functional profiles of table olives (hydroSOStainable olives) and pomegranates, respectively [9,12,33]. An increase, however, in polyphenols and antioxidant activity was observed for lettuce growing under water stress conditions [34] and grapes from clusters exposed to RDI (compared with SDI), which might happen due to the reduction of canopy leaf area of vines under RDI conditions [35].

Volatile Compounds
Twenty-six compounds were identified and quantified in the volatile profile of "Vairo" almonds. A total of 10 alcohols, 9 alkanes, 3 aldehydes, 1 terpene, 1 ketone, and 1 organic acid were identified and are presented in Table 3, together with their retention time, retention indices, and their odor descriptors, while Table 4 presents the contents of the volatile compounds. Alcohols, which were the most abundant volatiles (0.82 mg kg −1 ), are released by enzymatic reactions in raw almond and contribute to the characteristic sweet aroma and to the consumer acceptance [23]. Most of the alcohols were not affected by the studied treatments, except 1-hexanol, which had the highest content, 0.40 mg kg −1 , in T2 almonds. High levels of hexanol content was also found by other authors in Nonpareil almonds extracted with a similar method [23]. It is known that hexanol increases with the almond ripening and is associated with herbal odor (fruity, alcoholic, sweet, green notes) and green flavor (fruity, apple skin, oily) [11,36]. Garcia Esparza et al. [11] in studies about DI in grapes concluded that watering during post-veraison (the change of grapes color) at 75% of the crop ET (irrigation was applied to replace 75% of crop ET) compared to rain fed decreased the alcohol content (from 4.08 to 3.91 mg g −1 ) and increased that of the aldehydes (hexanal from 0.43 to 0.50 mg g −1 ), which produced herbaceous (non-desirable) aromas in wines [11].   From the alkanes (2.2 mg kg −1 mean values), statistically significant differences among studied treatments were found for pentamethyl heptane and tridecane. Pentamethyl heptane increased in T3 and T4 samples and tridecane in T2 and T3 samples. Previous studies showed that tridecane was formed by decarboxylation of myristic fatty acid (C14:0) [39] and this fact could be explained by the higher content of tridecane in T2 and T3, which can be induced by the water stress produced in the almond trees by RDI and could affect the plant metabolism.
Limonene was the only terpene found, and statistically significant differences were observed among the samples. Moderate RDI (T2) increased the limonene content, which is associated with fresh, citrus, and sweet notes. This result agreed with Carbonell et al. [10], who observed a higher content of limonene (from 12.4% to 14.8%) in pistachio under moderate RDI [10].
Benzaldehyde, which is the major volatile compound in bitter almond, was not predominant in these almonds, and this might be attributed to the low content of amygdalin [23]. Amygdalin is a cyanogenic glycoside naturally produced in almond, and it is the benzaldehyde precursor [23]. No statistically significant differences between the control and DI treatments were observed, implying that the sensory quality might not be affected by DI strategies. The finding agreed with other authors in studies of pistachio under RDI [10].
The total content of volatile compounds was significantly different among samples. Almonds from moderate RDI showed a higher total content (4.39 mg kg −1 ) as compared to the control samples (2.50 mg kg −1 ). Similar results (4.36 mg kg −1 ) were observed for raw almonds (mixture of "Butte" and "Padre" varieties) [40]. The aroma of raw almonds is, in general, weak, and a low total content of volatile compounds is usually expected. However, this content can be increased during roasting (6.17 mg kg −1 after 28 min of roasting; 11.4 mg kg −1 after 33 min of roasting, and 16.0 mg kg −1 after 38 min of roasting) [40] due to the Maillard and lipid oxidation reactions [40][41][42].
In this study, a reduced number and content of aldehydes was found, confirming the freshness of the studied almonds. In previous studies, aldehydes (e.g., hexanal) levels significantly increased, from 0.42 mg kg −1 in raw Butte and Padre almonds to 0.98 mg kg −1 in almonds roasted for 28 min at 138 • C and to 1.63 mg kg −1 after 24 weeks of storage at 35 • C [43]. During storage, and due to lipid oxidation reactions [44], hexanal increased, and this is why this compound (hexanal) is considered an indicator of oxidation or rancidity (low degree of freshness) in nuts and nut oils. The freshness (low level of rancidity) of all four almond samples was confirmed by the low content of hexanal and nonanal found [44,45]. Both hexanal and nonanal values showed no statistical significant differences among treatments, as well as a low content of hexanal and a similar content of nonanal when compared to the literature [46]. Table 5 shows the Pearson's correlation coefficients among studied variables with significant differences among treatments. A positive and significant correlation was observed between stress integral (SI) and (i) TPC (R = 0.72; p < 0.01), (ii) 2-butanol (R = 0.57; p < 0.05), and (iii) total volatile content (R = 0.75; p < 0.001). A positive correlation between the TPC and water stress levels in leaves of Solanum villosum and roots of Solanum scabrum has been reported [47]; this same positive correlation was also observed in tomatoes plants and maize [48]. Water stress can create damage in plants due to formation of reactive oxygen species (ROS) and to the alteration of the water-plant relationship [48]. The degree to which the plant can avoid or soften the physiological processes determines the resistance degree to water stress of each plant species [48]. Volatile compounds, which confers the typical almond aroma, were also found to be positively correlated with water deficit in different plants, such as grapevines, apples, tomatoes, and strawberry [49]. The increase in the total volatile compound value might be associated with the metabolic responses that deal with (i) high levels of light energy and (ii) formation of oxidative compounds under drought [49]. Note: *, **, ***, significant at p < 0.05, 0.01, and 0.001, respectively.

Pearson's Correlation Coefficients
A negative correlation was found between the contents of glucose and malic acid (R = −0.89; p > 0.05). Partially similar results were reported in other studies on plums, apricots, and apples [50,51]. This negative correlation is normally observed during the ripening process, because glucose and fructose increased due to the reaction of invertase enzyme through glycolysis, while the organic acids decreased because they are used in respiration and are converted into sugars [52].
A significant correlation was observed between the TPC and the total content of volatile compounds (R = 0.76; p < 0.001); this same trend was also found in olives [53].

Plant Material, Growing Conditions and Experimental Design
The experiment was performed during the 2017 season at the commercial farm "La Florida" (37.23 • N, −5.91 • W, Dos Hermanas, Seville, Spain). The almond (Prunus dulcis) orchard was 7 years-old at the beginning of the experiment. There were 2 almond cultivars in the orchard, "Guara" and "Vairo", and the tree spacing for both cultivars was 6 m × 8 m. The experimental plots had 4 lines of 3 trees and measurements were performed in the central trees of the "Vairo". The trees were irrigated with a line of drip emitters (3.8 L h −1 ) separated by 0.4 m. Irrigation scheduling was performed daily.
The seasonal weather data were obtained from the "Instituto de Investigación y Formación Agraria (IFAPA) Los Palacios" station, in the Andalusian weather stations network (Figure 2). This station is about 6 km away from the experimental orchard. The data for 2017 was typical of Mediterranean zones, with null rainfall during the summer period and warm winters.
with a line of drip emitters (3.8 L h −1 ) separated by 0.4 m. Irrigation scheduling was performed daily.
The seasonal weather data were obtained from the "Instituto de Investigación y Formación Agraria (IFAPA) Los Palacios" station, in the Andalusian weather stations network (Figure 2). This station is about 6 km away from the experimental orchard. The data for 2017 was typical of Mediterranean zones, with null rainfall during the summer period and warm winters.  The irrigation was scheduled according to measurements performed using a pressure chamber (PMS Instrument Company, Albany, OR, USA) and the threshold values of midday stem water potential (SWP) were measured to evaluate the level of plant stress. Three irrigations treatments were established together with a control treatment:

•
Full irrigation (T1), to assure the estimated ET during the entire growing season.

•
Moderate RDI (T2); in the period of kernel filling, almonds were irrigated when SWP < −1.5 MPa, and for the rest of the time, trees were irrigated to keep a SWP as the baseline proposed by McCutchan and A Shackel [54].

•
Severe RDI (T3); the same as T2, except trees were irrigated when SWP < −2 MPa during kernel filling. • SDI (T4); a lower amount of water was distributed uniformly throughout the year.
Equation (1) was used to calculate the stress integral (SI) and min ψ stem represented the average of minimum SWP for any interval, while n was the number of days interval: The harvest was done in August using a self-propelled trunk shaker with collector. Each treatment was harvested separately, and almonds were sun-dried to reach a moisture content below 5%, and then delivered to Miguel Hernández University for quality analyses. Almonds were shelled, and the kernels were ground, vacuum packed, and frozen until analysis.

Sugars and Organic Acids
High-performance liquid chromatography (HPLC) equipment was used to identify and quantify the sugars and organic acids as previously described by Lipan et al. [2] with some modification. Almond finely ground (1 g) in a Moulinex grinder (AR110830) for 10 s was homogenized with 5 mL of phosphate buffer 50 mM (pH = 7.8) with an homogenizer (Ultra Turrax T18 Basic) during 2 min at 11,300 rpm, while the tube was maintained in an ice bath, then was centrifuged (Sigma 3-18 K; Sigma Laborzentrifugen, Osterode and Harz, Germany) for 20 min at 15,000 rpm and 4 • C and was filtered (0.45 µm Millipore membrane filter). The filtered supernatant (10 µL) was injected into a Hewlett Packard (Wilmington DE) series 1100 (HPLC) using 0.1% ortophosphoric acid elution buffer. Sugars were measured using a Supelcogel TM C-610H column (30 cm × 7.8 mm) with a pre-column (Supelguard 5 cm × 4.6 mm; Supelco, Bellefonte, PA) and the detection was carried out with a refractive index detector (RID). Organic acids were separated in the same HPLC condition as sugars and absorbance was measured at 210 nm with a diode-array detector (DAD). Calibration curves were run in triplicate injection using standards of different organic acids and sugars provided by Sigma (Poole, UK). Analyses were run in triplicate and results were expressed as g kg −1 dw.

Fatty acids Analysis
Fatty acids methyl esters (FAMEs) were prepared as described by Lipan et al. [2] with some modification, while identification and quantification were done according to Tuberoso et al. [55]. Briefly, 40 mg of ground almond were saponified with 100 µL of dichloromethane (Cl 2 CH 2 ) and 1 mL of sodium methoxide solution and kept for 10 min at 90 • C. Boron trifluoride (BF 3 ) methanolic (1 mL) was added, followed by 30 min of reaction in darkness. FAME extraction from the mixture was done with 1.5 mL hexane. The separation of FAMEs was conducted using a Shimadzu GC17A gas chromatography coupled with a flame ionization detector and a DB-23 capillary column (30 m length, 0.25 mm internal diameter, 0.25 µm film thickness) J&W Scientific, Agilent Technologies. Helium gas was used as the carrier with a flow rate of 1.1 mL min −1 , and 35 mL min −1 at the make-up point. The temperatures of the injector and detector were 240 and 260 • C, respectively. The injection volume was 0.8 µL (split ratio 1:20). Finally, the temperature program was: initial temperature 100 • C held for 1 min, temperature gradient of 3 • C min −1 until 220 • C, followed by a gradient of 5 • C min −1 until 245 • C, and kept at 245 • C for 1 min. FAME peaks identification was performed by comparing the retention times of the FAME Supelco MIX-37 standards and the results were expressed quantitatively as g kg −1 concentration using methyl nonadecanoate as internal standard.

Antioxidant Activity and Total Phenolic Content
The antioxidant activity and total phenolic content was carried out not only for raw kernel but also for blanched almond and its skin. The blanching process consisted of almond immersion in boiling water (100 • C) for 2 min, followed by manual skin removal. The method of extraction consisted of 0.5 g of sample being sonicated with 10 mL of extractant (MeOH/water (80:20, v/v) + 1% HCl at 20 • C) for 15 min and stored for 24 h at 4 • C. The next day, the mixture was sonicated again under the above-mentioned conditions, then, it was centrifuged at 10,000 rpm for 10 min.
The antioxidant activity of the obtained extract was measured using 3 methods: DPPH , ABTS + , and FRAP. DPPH (2,2-diphenyl-1-picrylhydrazyl) free radical was used to determine the radical scavenging activity of the sample as described by Brand-Williams et al. [56]. A brief description of the process is that 10 µL of the sample were mixed with 40 µL of MeOH and 990 µL of the free radical solution, shaken, and placed in darkness for 15 min. Later, the sample absorbance was measured at 515 nm. Moreover, 10 µL of the sample supernatant was mixed with 990 µL solution of ABTS + (2,2-azinobis-(3-ethylbenzothiazoline-6-sulfonic acid)) or FRAP (ferric reducing antioxidant power) free radicals to determine the free radical scavenging capacity of the sample. After 10 min of reaction, the sample absorbance was read at 734 nm for ABTS + method and 593 nm for FRAP. All measurements were carried out in an ultraviolet-visible (UV-vis) spectrophotometer (Helios Gamma model, UVG 1002E; Helios, Cambridge, UK). The quantification was done according to the calibration curve of Trolox, prepared in a concentration ranging from 0.5 to 5.0 mmol Trolox L −1 . The linearity was above R 2 = 0.998 and results were expressed in mmol Trolox kg −1 [9,56].
Total polyphenolic content (TPC) was determined using the Folin-Ciocâlteu colorimetric method, in which 100 µL of sample supernatant were mixed with 200 µL Folin-Ciocâlteu reagent and 2 mL of H 2 O 2 . This mixture was stored at 22 • C for 3 min and 1 mL of 20% sodium carbonate was added, followed by incubation for 1 h at room temperature. Later, the mixture absorbance was measured at 765 nm in the above-mentioned equipment. The results were calculated according to the gallic acid calibration curve and were expressed as gallic acid equivalents (GAE), g GAE kg −1 [9,56].

Volatile Compounds Analysis
Volatile compounds were extracted using headspace solid phase microextraction (HS-SPME). For the extraction, 2 g of grinded almond and 50 µL of β-ionone (100 mg L −1 ) were placed in a hermetic vial with polypropylene cap and PTFE(polytetrafluoroethylene)/silicone septa and was used as an internal standard; this internal standard was used for the semi-quantification of the volatile compounds, as no calibration curve was done for each of the compounds reported in this study. The vial was placed in a water bath at controlled temperature to assure the vial was 40 • C, which was needed to simulate the mouth temperature when chewing almonds. Once the temperature was reached and was constant, a 50/30 µm Divinylbenzene/Carboxen/Polydimethylsiloxane (DVB/CAR/PDMS) fiber was introduced in the headspace of the vial for 50 min. This fiber is characterized by high capacity of trapping volatile compounds from fruits and nuts. Moreover, the fiber was desorbed for 3 min in the injector port of a gas chromatograph Shimadzu GC-17A (Shimadzu Corporation, Kyoto, Japan) coupled with mass spectrometer (MS) detector Shimadzu GC-MS QP-5050A used for the volatile compounds identification. The GC-MS was equipped with a SLB-5ms Fused Silica Capillary Column of 30 m × 0.25 mm × 0.25 µm film thickness, 5% diphenyl, and 95% dimethyl siloxane (Supelco Analytical). For the analyses, helium was used as gas carrier at a flow rate of 0.7 mL min −1 in splitless mode. The oven program was: (a) initial temperature 40 • C, (b) rate of 2.0 • C min −1 to 145 • C, (c) rate of 25 • C min −1 from 145 to 300 • C and hold for 90 s. In addition, the injector was kept at 230 • C, while the detector at 320 • C. The volatile compounds identification was done by using 3 methods: (a) retention indices, (b) Gas Chromatography -Mass Spectrometry (GC-MS) retention times of authentic chemicals, and (c) mass spectra (authentic chemicals and NIST05 spectral library collection) [38].

Statistical Analysis
The statistical analyses were done by using one-way analysis of variance (ANOVA), and data were submitted to Tukey's multiple range test to compare means. Statistically significant differences were considered when p < 0.05 and were studied using XLSTAT Premium 2016 (Addinsoft, New York, USA). Pearson's correlation was carried out with the same program in which data were subjected to Correlation Tests. For figures preparation, Sigma Plot 11 was used.

Conclusions
Based on the literature search, this study was the first to evaluate the quality parameters (sugars, organic acids, antioxidant activity, and total contents of phenolic and volatile compounds) of hydroSOStainable "Vairo" almonds under regulated (RDI) and sustained (SDI) deficit irrigation. Almond yield was not affected either by RDI or SDI. Almonds are a very good source of polyphenols (5.5 g GAE kg −1 ; mean values for all treatments), and regarding the antioxidant activity and total phenolic compounds, there were no significant differences among irrigation treatments. Moderate regulated deficit irrigation led to almonds with high glucose content (potentially linked with almond sweetness) and total content of volatile compounds (potentially linked with almond odor, aroma, and flavor), implying a high sensory quality. Consequently, almond quality can be improved if the water stress in the plant is induced in a controlled way, contributing to sustainable agriculture with greater benefits for the farmers through commercialization of these almonds as hydroSOS products. Funding: The study has been funded (Spanish Ministry of Economy, Industry and Competitiveness) through a coordinated research project (hydroSOS mark), including the Universidad Miguel Hernández de Elche (AGL2016-75794-C4-1-R, hydroSOS foods) and the Universidad de Sevilla (AGL2016-75794-C4-4-R) (AEI/FEDER, UE). Marina Cano-Lamadrid was funded by a Formación de Profesorado Universitario (FPU) grant from the Spanish Ministry of Education (FPU15/02158).

Conflicts of Interest:
The authors declare no conflict of interest.