Two Fixed Ratio Dilutions for Soil Salinity Monitoring in Hypersaline Wetlands

Highly soluble salts are undesirable in agriculture because they reduce yields or the quality of most cash crops and can leak to surface or sub-surface waters. In some cases salinity can be associated with unique history, rarity, or special habitats protected by environmental laws. Yet in considering the measurement of soil salinity for long-term monitoring purposes, adequate methods are required. Both saturated paste extracts, intended for agriculture, and direct surface and/or porewater salinity measurement, used in inundated wetlands, are unsuited for hypersaline wetlands that often are only occasionally inundated. For these cases, we propose the use of 1:5 soil/water (weight/weight) extracts as the standard for expressing the electrical conductivity (EC) of such soils and for further salt determinations. We also propose checking for ion-pairing with a 1:10 or more diluted extract in hypersaline soils. As an illustration, we apply the two-dilutions approach to a set of 359 soil samples from saline wetlands ranging in ECe from 2.3 dS m-1 to 183.0 dS m-1. This easy procedure will be useful in survey campaigns and in the monitoring of soil salt content.


Introduction
Salts commonly occur across the Earth's surface. Most salts needed for life are imbibed by plants from the soil. Thus, fertile soil provides an adequate content of salts needed by plants. Contrariwise, highly soluble salts are biocides at high concentrations. Few organisms, called extremophiles, are adapted to hypersaline conditions. Soil salts at the Earth surface are dissolved and redistributed across the landscape, leading either to salt leaching or accumulation at specific geomorphic positions. Saline soils are more frequent in regions where evaporation exceeds rainfall. Together with natural factors, human actions can salinize or desalinize soils, sometimes in only a few years. Examples of such include the clearing of lands in Australia, or irrigation with brackish water pervasive in some countries. In some cases, changes in soil salt contents have been measured or surmised at several temporal scales [1][2][3][4].
The conservation of saline enclaves comes up as the demands for environmental protection become more elaborate, e.g., [5], especially in arid lands as reviewed by Williams [6][7]. Many saline wetlands around the world are included in the Ramsar Convention signed by 167 conductivity of the saturated paste extract (ECe) for expressing soil salinity, with the extracts at other fixed soil to water ratios considered auxiliary or less important. The amount of water needed to prepare the saturated paste is reported as cm 3 of water per 100 g of soil, i.e., weight to weight (w/w). This amount, i.e., the saturation percentage (SP), depends on the textural composition of the soil, clay mineralogy, and organic matter contents. Saturated paste was intended to enable soil solution to be extracted with simple equipment, overcoming the practical impossibility of extracting the actual soil solution used by crop roots. ECe has become a standard measure of soil salinity, widely used to compare the salt tolerance of cultivated plants, and for irrigation and drainage engineering.
The purpose of soil salinity monitoring in hypersaline soils of protected wetlands is not to determine the effects of dissolved salts on plant growth, as is the case in agricultural or in plant physiology research. The monitoring of protected wetlands will watch and ward the salt contents associated with the prevalence of the halophytes and other valuable organisms adapted to hypersalinity. The available methods to measure the electrical conductivity (EC) in-situ are influenced by temperature and soil moisture, and are unpractical in these intermittently inundated wetlands. Measures of soil salinity by a lumped parameter, EC, at standard and repeatable conditions of temperature and dilution will be needed for years to come.

3.1
Extracts at fixed soil to water ratios, with 1:5 and 1:10 as the more popular, have been used in soil science. The disclosure of the relationships of their EC with ECe [15] has garnered the attention of many scientists using different approaches as shown by [16]. However, the United States Salinity Laboratory Staff ( [14], page 13) recommended fixed ratios for determining the change in salinity with time; in the same way, [17] recommend 1:10 for evaluation of the total soluble salts of soil and for the assessment of reclamation procedures.
The search for such an equation can be unpractical for soil survey or monitoring operations due to the need for either supplementary data to be included in the equation or chemical considerations often applied with ionic speciation software. However, if the interest is focused on the off-site effects of soil salinity (i.e., salt discharge to surface and underground waters, rather than on the effects of salinity on plants), a fixed soil to water ratio more diluted than the saturation extract can better estimate the content of salts in soils.

3.2
The non-occurrence of ion-pairing is a condition to have a simple relationship of EC with the content of salts in solution. If ion-pairing occurs, the measure of the salt content would require either gravimetry of the total dissolved salts or the individual titration of soluble ions. Ion-pairing does not occur in saturation extracts of low EC; e.g., in the experiments of [35] with a nonsaline soil of ECe = 0.66 dS m -1 . However, the reservations [36][37] for soil salinity estimates from ECe in non-saline calcareous soils can be extended to gypsiferous soils [29].

3.3
The correspondence between ECe and salt content diminishes at high salinities, with the classical olifant-shaped silhouette in the scatter diagrams of ions vs. ECe [34] needing logarithmic transformations to approach linear relationships. Many soil scientists [38,39] have shown the problems of relating ECe with total salt content for hypersaline soils. Similar concerns were expressed by [40] for ECe > 8 dS m-1 when relating ECe with the electrical conductivity measured in the saturated paste. Moreover, [36] found that for non-saline calcareous soils, saturation extracts require dilution by a factor of 1000 to accurately predict soil salinity. The aforementioned considerations jeopardize the meaning of ECe and also the additivity of this magnitude from different samples, and can question the meaning of averaged ECe for pedons sampled at several depths, or for multiple samples when estimating soil salinity of broad extensions. These constraints can be dramatically reduced if no ion-pairing happens, as would be the case for extracts more diluted than those from saturated paste.

3.4
Most ECe measurements and subsequent determinations in the saturation extracts are used for agriculture or for plant salt-tolerance appraisal e.g., [41][42][43], and for the study of saline habitats, e.g., [44,45]. Most of these works do not report saturation percentage (SP). However, when ECe is to be used for comparison of salinity in the same soil, evaluation of the deviations in SP between different times or laboratories [27,34,46] is compulsory. Yet often SP is not reported in comparisons of salinity, e.g., [2,47,48]. A 5% SP deviation has been suggested as an allowable difference ( [49], page 265). This is a key issue for comparisons involving different technicians or laboratories, especially for long-term monitoring. The measurement of changes in ECe and ion concentrations can be biased not only because of the different criteria in the end-point of the paste but also by the grinding or extraction methods [50][51].
Moreover while centrifugation of the saturated paste is used by many labs for obtaining the extract, many others use vacuum extraction, a technique that concentrates the extract, as evidenced by the descent of temperatures in the flasks produced by evaporation under vacuum. In practice, extract concentration achieved by evaporation cannot be compensated by calculations, and will variably affect the extraction depending on the intensity and time of vacuum, lab temperature, or the whole volume of air evacuated by the pump. The total effects will be associated with the granulometry and other properties of the soil sample, affecting each sample at different degrees even within the same extraction set.

3.5
The saturation extract cannot be obtained in field labs. Moreover, saturated paste preparation is made by hand, with no prospects for automatization, and the extraction of the solution by vacuum can last more than one hour. The process is lengthy and hard to perform under the current conditions of dismantlement of many soil labs. This, plus the scarcity of technicians trained in the preparation of saturation extracts endorses the use of extracts at fixed water to soil ratios for surveys or other works of great spatial or temporal span needing serialized determinations.

3.6
Natural hypersaline soils occur both in the coasts and inland with salinity levels and other circumstances precluding the growth of common profitable crops. Moreover environmental regulations of some countries forestall agriculture in these wetlands. Very often, the appraisal of soil salinity by EC in such environments is needed for applications other than agriculture. Most of these soils are intermittently inundated, or at least saturated, by salty water or brine. For salt-tolerance evaluation, it seems trivial to relate ECe or EC at other dilution ratios with the natural composition of the soil solution that would limit the growth of plants. Under these conditions, to appraise the content of salts is more interesting than to approach the salinity affecting plant growth as intended by the saturated paste method. Thus, application of a fixed ratio should be better than saturation extracts.

3.7
Very often, the saturation extracts from saline wetlands make ion-pairs, while in most cases the 1:5 and 1:10 extracts will not do that. Thus, the relationships between the ionic concentration and EC will be simpler than the saturation extract. A noteworthy advantage of these extracts for hypersaline soils would be to start in the lab from concentrations more convivial with the conventional chemistry labs titrating ions with methods or equipment not devised for high ionic concentrations. It will reduce or eliminate the need for time consuming dilutions; a classical source of mistakes and errors in serial analytical determinations.

3.8
Extracts at 1:5, 1:1, or other fixed soil to water ratios are often used in different scientific domains . Regrettably, sometimes the dilution ratio is missing or not clearly stated in reports or articles [78][79][80][81] or maybe this ratio, saturation or other, is assumed, e.g., [82][83][84][85][86], compromising or precluding future comparisons and generalization. As saturated paste is unlikely to be adopted as a standard in many of these scientific domains, and the SP is very often not reported, fixed soil to water ratios would be easier and more repeatable. In short, should the methods of soil science depart from other scientific disciplines like biology or ecology that also conduct studies on soil salinity?

3.9
Even though in our experience no ion pairing happens in 1:5 extracts, this circumstance merits checking in hypersaline soils. The method of diluting the saturation extract until EC reaches a value between 0.1-0.3 dS m -1 [87] overcomes ion-pairing to achieve a linear relationship with the salt content of the saturated extract. This method could be used for extracts at fixed ratios (e.g., 1:5), where the occurrence of ion-pairing was suspected, but regression with a lower ratio extract (e.g., 1:10) is easier and ancillary to a duplicate determination of the EC1:5. [88] successfully used paired 1:5 and 1:10 soil to water extracts in seven saline wetlands. Should the relationships between EC1:5 and EC1:10 depart from the linearity pointing to ion-pairing, an extraction at more diluted ratios would be needed.

The Two Dilutions Extract Approach
Our proposal for soil salinity expression for monitoring purposes is to prepare both 1:5 and 1:10 extracts. The regression between EC1:5 and EC1:10 will be a check for the absence of ionic pairs. This easy procedure is free of chemical models and permits a choice of the extract to be used for further titration of ions. The second extracts can be considered as surrogates for duplicated determinations. Depending on the desired confidence level, the second dilution can be limited to a reduced subset of soil samples if only a statistical check of no ionic-pairing is desired.
The effects of the preparation procedures of the 1:5 extract using a set of 20 samples ranging in ECe from 0.96 to 21.20 dS m -1 , and in SP from 32.4% to 68.1%, were studied by [89]. After selecting the appropriate procedure, they stressed the need to report its detailed description.
Our approach involving two different soil to water ratios can be transposed to the methods based on volume ratio extracts instead of weight ratios, as proposed by [90] for glasshouse soils. More recently, upon the establishment of the suborders Wassents and Wassists in US Soil Taxonomy [48], a method for measuring the EC of subaqueous soils has been defined by Soil Survey Staff ( [91], page 292). This method measures the EC from a fresh, field wet sample using a mixture (not extract) of soil to water at the ratio of 1:5 by volume (EC1:5 vol), instead of the conventional 1:5 ratio by weight, and measures the EC of the supernatant. If, for taxonomic or characterization purposes the salinity is to be calculated from EC1:5 vol, the eventual occurrence of ion-pairing could be checked with the extract of 1:10 vol.

Materials and methods
We studied several hypersaline wetland "saladas" protected under Spanish environmental regulations and located in the Monegros desert, Spain (Fig 1). The geological framework of the distribution of these wetlands was addressed by [12]. The soils of the saladas were Gypsic Aquisalids and Typic Haplogypsids per US Soil Taxonomy [49]. Permissions for sampling were granted by the Authority in charge of protected areas: Instituto Aragonés de Gestión Ambiental (www.aragon.es/inaga). The soils at the floors of ten of these saladas: Amarga Alta, Amarga Baja, Camarón, Gramenosa, Guallar, Muerte, Pez, Piñol, Rebollón, and Rollico (Fig 1), were hand-augered at 59 sites, taking cores by depth increments of 20-25 cm. The resulting 359 soil samples were air-dried first at room temperature and then in a ventilated oven for 2 weeks at 40°C in order to avoid the destruction of gypsum (CaSO 4 •2H 2 O). Samples were ground to pass  Table. The insert shows the location of the Monegros area within the Ebro Basin, Spain. a 2-mm sieve; no coarse fragments occurred. The saturated pastes were prepared with their saturation percentage (SP) recorded. The electrical conductivity (EC) was measured from the saturation extract (ECe) and from extracts with soil to water ratios in weights of 1:5 (EC1:5) and 1:10 (EC1:10). All the electrical conductivities were expressed in dS m -1 at 25°C.
All samples contained gypsum, a common mineral in saline wetlands. Then, sufficient time was allowed to guarantee the maximum dissolution of gypsum to attain stable and reproducible EC measurements. The saturated paste was left to stand overnight before extraction. For the 1:5 and 1:10 extracts, we applied 30 min of reciprocal shaking at 175 oscillations per minute followed by overnight standing before filtration for EC measurement.
The Clwas titrated in 288 samples from eight of the studied saladas on extracts at 1:10 dilution using a potentiometer, and expressed in milliequivalents per liter (meq L -1 ). Gypsum content was determined by thermogravimetry per [92] and calcium carbonate equivalent was measured by gasometry with a Bernard calcimeter.
After exploratory data analysis of the data shown at S2 Table, ordinary least squares (OLS) regression was applied with the coefficient of determination (R 2 , expressed as %) and the standard error (SE) calculated.

Results
The sampled soils were either bare or supported halophytes at the less inundable positions. This is in agreement with the hypersalinity of the studied soils expressed by the mean ECe = 72.3 dS m -1 of the 359 samples studied and their range from 2.32 dS m -1 to 183.00 dS m -1 (Table 1). Ten samples were below the ECe threshold 4 dS m -1 for saline soils, while 300 samples had ECe > 16 dS m -1 ,5 the threshold for very strongly saline soils. Only 14 samples had EC1:10 2.25 dS m -1 , the approximate electrical conductivity produced by calcium sulfate saturation marked with a dashed line in  Table 1 shows the high contents of gypsum and calcium carbonate. The gypsum content ranged from 1.9% to 96.4%, with a mean of 49.3%. Fig 3 shows the samples ranked by gypsum content, with 180 samples having > 50% gypsum. Only three samples with gypsum ranging from 1.9% to 2.1% had their gypsum content below the thresholds for saturating the 1:5 or the 1:10 dilutions, i.e., 1.2% and 2.4% gypsum, respectively, assuming that saturation is attained at 2.4 g of gypsum per liter of pure water, a solubility that increases if Clor other non-common ions are present. Most samples qualify as gypseous [93], but their high salinity supersedes other limitations to life [94]. Fig 3 also shows the contents of calcium carbonate equivalent (CCE). The sum of CCE plus gypsum ranged from 37.8% to 99.6%, with a mean of 71.2%; this sum was > 50% for 338 samples of the 356 having both gypsum and CCE titrated. The average SP = 39.1% falls within the range of the SP reported for coarse or sandy textures [14,95]. It agrees with the field textures of these soils, controlled by the abundance of visible-sized gypsum crystals. The Clconcentration (Cl1:10) in the 1:10 extracts of the 288 samples analyzed showed a linear distribution against the EC of the same extracts, in spite of the inflection in the scatterplot at EC~2.25 dS m -1 (Fig 4).
The OLS adjustment is given in Eq 1: The relationship between the EC at the two soil to water dilutions of 1:5 and 1:10 was studied by scatterplot (Fig 5) and by OLS regression using the 359 soil samples.
The OLS adjustment is given in Eq 2:

Discussion
We have reviewed the shortcomings of ECe for the study of soil salinity in hypersaline environments with no agricultural purposes, and the advantages of using more diluted extracts. Our proposal of assessment of salinity is based on two extractions at different soil to water ratios, 1:5 and 1:10. The preparation of these extracts is much easier, less time-consuming, and equipment-demanding than saturation extracts. Also, further titration of individual ions would be easier than from the saturation extract, which is much more concentrated. Another advantage of the 1:5 extract is the coincidence with the common practice in Australia of expressing salinity by EC1:5 [31,96] and by the Clconcentration in this extract [97]. The quality of the analyses in our example is supported by the scatterplot of Clon EC in the 1:10 extracts (Fig 4) and by the linear relationship between these determinations shown by Eq 1 with R 2 = 98.8%. The same support is provided by the scatterplot of EC1:5 on EC1:10 ( Fig 5) and by the R 2 = 99.1% attained by the OLS regression (Eq 2). Straight line adjustments indicate that both extracts have the same degree of ionization, i.e., no ion-pairing happens.   Systematic departures from the adjusted lines occur for the few soil samples which lack sufficient gypsum or other soluble minerals for achieving an EC1:10 of 2.25 dS m -1 . If these samples are eliminated from the adjustments (Eqs 1A and 2A) the slight improvements are below the average allowable errors of most routine lab analyses.
The ubiquity of gypsum in the studied samples, a frequent setting in athalassohaline wetlands, and their specific ionic composition prevents comparisons of the above adjustments with other soils. The purpose of the adjustments was not predictive, but a check of both the full ionization in the 1:5 dilution and the analytical quality. Thus, depending on circumstances like the confidence in the lab performance, the allowable lab workload, and the number of samples, the determinations of EC1:10 could be limited to a reduced number of samples covering all range of salinity determined at 1:5 dilution.
The scatterplot of EC1:5 over EC1:10 ( Fig 5) shows the classical inflection due to gypsum. The samples having either EC1:5 or EC1:10 2.25 dS m -1 are located under the inflection in the figure, showing in this region a systematic departure from the OLS regression line.
The presence of gypsum was early recognized as a source of troubles in the conversion of EC between extracts at different soil to water ratios ( [98], page 14). That notwithstanding, the effect of gypsum producing an EC of~2.25 dS m -1 at saturation, can be neglected in these conversions for hypersaline soils if the gypsiferous and non-saline samples are a minority. These samples, grouping around the point ECe = EC1:5 = 2.2 dS m -1 [99], can compromise the homocedasticity of the distribution if they are numerous.
If a significant number of extracts with EC < 2.25 dS m -1 occurs, or if specific consideration is wanted for these extracts, a separate regression could be undertaken. However, categorizing soils into groups with or without any measurable amount of gypsum to improve the prediction of ECe [29] with a separate determination of gypsum would be cumbersome for long series of determinations and for simple labs. A similar procedure based on a qualitative assessment of gypsum by precipitation with acetone in 410 soil extracts at 1:5 dilution was used in gypseous soils by [99].
Several authors have used EC1:5 as an auxiliary variable for soil mapping with electromagnetic induction (EMI) measurements [100], or have calibrated EMI measurements with EC1:5, as is the case of [52,[101][102] and the examples mentioned by [103]. In our experience, the calibration of EMI measurements with EC1:5 was possible, even if the attained coefficients of determination were lower than with ECe [99,104]. EMI devices, as other sensors, respond to both variable and invariable soil characteristics and their distribution along the soil profile such as moisture, temperature, mineralogy, pore size and architecture, etc. Therefore, it is requisite to take some soil samples for calibrating the sensor signal with the target characteristic. EMI has been used to map soil salinity in areas with shallow water tables in the central Ebro valley, e.g. [105][106][107], and in saline coastal wetlands, e.g. [33,108,109]. The adoption of EC1:5 as the standard measure of soil salinity would simplify the calibrations and eliminate the time consuming preparation of the saturation extract.
Other developing technologies will be able to measure salinity in the field. One example is portable x-ray fluorescence spectrometry whereby Clor other elements are used as a proxy for soil salinity, but it also requires correction for soil moisture when levels are above 20% [110,111]. It shows promise for soil salinity appraisal, and could overcome the measurements of ECe on water extracts. These matters fall out of the scope of this article and merit discussion and effort both for comparisons and for long term monitoring of soil salinity.
The multitude of models and adjustments proposed in the literature for predicting ECe from EC at a fixed soil to water ratio do not provide a conclusive, universal, and non-sophisticated procedure. For the studied soils, Fig 6A shows the inflection due to gypsum plus the classical dispersion at higher values; both features entangling the response model. For large sets of samples, one can undertake an adjustment of EC1:5 or other fixed ratios versus a reduced number of ECe determinations. However, it must be noted that the "gypsum inflection" lasts ( Fig  6B) even after the transformation that includes SP [46,112] and enables for decreasing the dispersion. Moreover, the shortcomings mentioned at Sections 3.3 and 3.4 strongly question the estimation of total salt content from ECe, at least for soils containing calcium carbonate or gypsum as well as for hypersaline soils.

Conclusions
The study of inland wetlands located in arid climates requires the development of new concepts to be incorporated into mainstream wetland science. With the proposed approaches, we try to simultaneously assess "agricultural" and "environmental" methods that are studying the same natural objects, saline wetlands.
The degrees of salinity of the studied soils plus frequent flooding preclude their agricultural use. This setting is frequent in saline wetlands around the world that are often protected by environmental rules. Conservation of these valuable habitats requires monitoring the salt contents of the soils, diminishing the usefulness and meaning of saturated paste. The expression of total salt content by the electrical conductivity in the extracts at 1:5 dilution, with checking for no ion-pairing with the 1:10 extract, has shown to be easy, unsophisticated, and robust. Our two dilution approach bypasses the agricultural salinity expression by saturation extract by adopting the 1:5 extract as a standard for soil salinity expression instead of the saturation extract.
Supporting Information S1 Table. UTM coordinates (European Datum ED 50) of the 59 sampling sites with their vegetation and the depth of the augerings. (PDF) S2 Table. Analytical data of the soil samples. (PDF)