Salinity in Calcarosols occurs through the presence of sodium, chloride, bicarbonate and sulfate ions, is caused by sodicity, and leads to decreased osmotic potential

Measuring exchangeable cations

One critical concern facing researchers who wish to measure the concentrations of adsorbed cations in a sodic soil that also accumulates salt is that the conventional method of extracting cations using BaCl₂/NH₄Cl extraction (Rayment and Lyons 2010; Method 15E1) mixes the cations from the soluble and adsorbed pools together. This has the effect of increasing the calculated ESP, leading to the suggestion that the soil may be more sodic than it is. Some researchers overcome this issue by applying an alcohol ‘pre-wash’ to the soil, discarding this, and then extracting the remaining adsorbed ions using the BaCl₂/NH₄Cl extraction method (Rayment and Lyons 2010; Method 15E2). However, successful use of this method requires that none of the adsorbed cations are leached from the exchange surfaces by the alcohol pre-wash. To overcome this problem, we became interested in the method for determining adsorbed cations used by So et al. (2006). In their work the concentration of adsorbed cation is calculated as the concentration obtained using the BaCl₂/NH₄Cl extract less the concentration obtained in a water-soluble extract. In the present work we have compared this method of calculating exchangeable cations with that obtained using the alcohol pre-wash method.

Depth changes in Calcarosols

Calcarosols are defined in the ‘Australian Soil Classification’ as soils that are calcareous because of in situ soil formation processes (Isbell 2016). In a survey of 150 different Calcarosol profiles over 15 sites in Victoria, Nuttal et al. (2003) found average pH_1:5 values increased from 7.5 (0–0.1 m) to 8.6 (0.6–1.0 m), EC_e values increased from 2.0 dS m⁻¹ (0–0.1 m) to 8.4 dS m⁻¹ (0.6–1.0 m), and ESP values increased from 7.5 (0–0.1 m) to 19.1 (0.6–1.0 m). (It should be noted that in this study, exchangeable cations were determined using the alcohol pre-wash method, influencing the determination of ESP.) These effects might be expected to have adverse effects on the rooting depth, growth, and yield of crops.

Principle soluble cations and anions leading to increased EC_1:5 values in affected soils

In general, there has been little consideration of the ions that cause salinity in sodic soils. The ions responsible for the high EC_e values in the study from Victoria referred to above were not determined, but studies from North America, India, Africa, and Hungary suggest that soil alkalinity can be associated with soluble Na⁺, SO₄²⁻, Cl⁻, HCO₃⁻, and CO₃²⁻ (Szabolcs 1989).

Factors causing increased EC_1:5 in soils

One of the curious features of salinity is that although causal links to sodicity have been claimed (Rengasamy 2002), correlations describing this cause and effect are not widespread in the literature. An exception is in the study of Nuttal et al. (2003). In their survey of Calcarosol profiles in Victoria, a simple linear correlation between EC_e and ESP had a correlation coefficient (r) of 0.71 and the correlation was significant at P < 0.01. Our current working hypothesis for salinisation due to sodicity is that sodicity causes soils to disperse which clogs soil pores and decreases soil hydraulic conductivity (Frenkel et al. 1978). Salt is continuously added to soils in small amounts in aerosols and rain (Hingston and Gailitis 1976), but if this salt is unable to leach through the soil profile (due to the combination of low rainfall, high evaporation and sodicity), then it may accumulate where it can have adverse effects on crop growth (Barrett-Lennard et al. 2016). In addition to this effect, it is possible that several other factors could also play a role, including high soil pH values, high soil cation exchange capacity and high clay content.

Impacts of salt accumulation on osmotic potential

It is not the salt concentration in the bulk soil (of which the EC_e and EC_1:5 are measures) that adversely affects plant growth; rather, it is the salt concentration in the soil solution that affects growth. For soluble salts like those mostly present in saline soils, ion concentrations in 1:5 extracts are more dilute than ion concentrations in the soil solution by the factor 500/W where W is the gravimetric water content of the soil expressed as a percentage (Rengasamy 2010). The EC of the soil solution can therefore be estimated as:

Rather than express salinity in this way, we prefer osmotic potential (Ψ_s; a negative number) as an index of the salinity stress of plants (Rengasamy 2002). Expressing the salinity of the soil solution in this manner enables us to estimate the total water potential (Ψ) of the soil because Ψ is the sum of soil matric potential (Ψ_m) and Ψ_s. If the salt makeup of the soil is known, then Ψ_s can be estimated from measurements of soil EC_1:5 values and gravimetric soil water content values.

The paper applies these themes to a set of soil surveys conducted to characterise four field experiments near Merredin and Moorine Rock in Western Australia in July 2022. These soils have been used for annual cropping for about 100 years. Nevertheless, we make the case here that the levels of salinity in these soils are of such magnitude that they are likely to impact adversely on crop growth, particularly in dry years.

Location of soils sampled

Soil samples were collected from four experimental sites in July 2022 (Table 1). Three of these sites were located on the Merredin Research Station of the Department of Primary Industry and Regional Development (31.5072°S, 118.2177°E), and the fourth site was located near Moorine Rock, about 86 km east of Merredin (31.3453°S, 119.2625°E). The sites were located on Calcic Calcarosol (Isbell 2016, Australian Soil Classification) or Vertic Calcic Calcisol (Sodic) soils (IUSS Working Group WRB 2014).

The soils from the Moorine Rock site are commonly called ‘morrel’ soils after the dominant tree ‘red morrel’ (Eucalyptus longicornis (F.Muell.) Maiden) that originally occurred on this land; these soils are known to be naturally saline (Teakle and Burvill 1938; Barrett-Lennard et al. 2003). The Merredin sites would have had an original vegetation dominated by salmon gum (Eucalyptus salmonophloia F.Muell.), remnants of which can still be seen in roadside vegetation.

Soil samples were collected from the furrow of agronomic trials planted to wheat (Triticum aestivum L.) or barley (Hordeum vulgare L.) using a percussion drill rig with a 50 mm diameter soil core to depths of 1.0–1.4 m, sampling at the intervals given in Table 1.

Analysis of soils

The soils were initially weighed, then oven dried for 1–2 weeks at 40°C and reweighed. The difference between wet and dry weights was used to calculate gravimetric soil water. Once dried, the samples were crushed and sieved to produce a soil sample with a particle size of less than 2 mm.

The sampled soils were analysed in a commercial laboratory (Eurofins APAL, Hindmarsh, South Australia) using the methods summarised in Table 2. Each oven-dried sample was analysed for pH, EC_1:5, total cations (exchangeable plus soluble), exchangeable cations with alcohol pre-wash, soluble cations, soluble anions, and particle size (sand, silt and clay). For each cation, exchangeable cation concentrations were calculated as the difference between total cation concentration (determined from the BaCl₂/NH₄Cl extract) and the water-soluble cation concentration (see Table 2 and Section 1 of Results).

Exchangeable sodium percentage (ESP) was calculated from exchangeable cation concentration values (units of cmol⁺ kg⁻¹) as:

Dispersive charge (Rengasamy et al. 2016) was calculated from the exchangeable cation concentration values (units of cmol charge kg⁻¹) as:

Estimation of osmotic potentials

Soil osmotic potential is estimated from the concentrations of the different salts causing the EC_1:5 and the gravimetric soil water content. Our soil analyses reported concentrations of Cl⁻, HCO₃⁻, SO₄²⁻ and CO₄²⁻ in units of cmol kg⁻¹. We converted these into salt concentrations in the soil solution assuming the anions to be paired with Na⁺ (see Section 3 of Results) and that the salts were dissolved in the soil water or in a standardised amount of water (30% dry mass). This provided us with concentrations of each salt in the soil solution (units of mmol L⁻¹).

Using the data of Wolf et al. (1986) we established calibration curves between salt concentrations in solution (mmol L⁻¹; x-variable) and osmotic potentials (MPa; y-variable) (see Supplementary Fig. S1). Wolf et al. (1986) report data relating molar salt concentrations, molar water concentrations and the osmolality of solutions of different salts at 20°C, however the temperature compensation standard for most EC meters is 25°C.

Osmotic potential (MPa) can be calculated (LabXchange 2023) as:

where i is the number of ions that a salt dissolves into (the van’t Hoff factor), C is the concentration of the solute (moles L⁻¹), R is the universal gas constant (0.00831441; L.MPa moles.K⁻¹), and T is the absolute temperature (°K).

The product of i and C is the osmolarity of each salt (O) and the data published by Wolf et al. (1986) enable its calculation at 20°C as:

The osmotic potential of each salt (Ψ_s; MPa) at 25°C (298°K) was calculated as:

Using the calibration curves for the four salts we estimated osmotic potentials for each salt dissolved in the soil water. The Ψ_s of the soil solution was calculated assuming it to be the sum of component osmotic potentials from each salt.

Statistical methods

The statistical analyses reported here were conducted using GENSTAT 22nd edition (VSNi).

1. Determining exchangeable cations as the difference between total and soluble cations

Soils in Australia have been regarded as being sodic and strongly sodic if they have exchangeable sodium percentages >6 and >15, respectively (Northcote and Skene 1972). More recently, Rengasamy et al. (2016) have proposed that sodicity be expressed in terms of dispersive charge, a variable derived from the concentrations of exchangeable Na⁺, K⁺, Mg²⁺, and Ca²⁺. Calculation of each of these variables requires the use of a reliable method for determining exchangeable cations.

Measures of exchangeable cations are widely determined based on the concentrations obtained after soil is extracted with BaCl₂/NH₄Cl (Rayment and Lyons 2010; method 15E1). This aggregates all the cations, soluble and exchangeable, into one extraction. For soils affected by salinity, this approach has the disadvantage of also attributing soil dispersion to any cation (but particularly Na⁺) that might be present in the soluble as well as the exchangeable pools.

An alternative to this approach is to apply an ethanol/glycerol pre-wash to the soil, to remove (and discard) the soluble cations, and then further extract the soil with BaCl₂/NH₄Cl to remove (and measure) the remaining exchangeable cations (Rayment and Lyons 2010; method 15E2). This approach assumes that the pre-wash only removes soluble ions and does not strip off any of the cations adsorbed to the soil exchange surfaces. If this approach is valid then for any cation, the adsorbed concentration after pre-wash should be the same as the difference between the total concentration of ion after BaCl₂/NH₄Cl extraction and the concentration of ion in a water extract (c.f. So et al. 2006).

We tested this proposition for the Na⁺, K⁺, Mg²⁺, and Ca²⁺ contained in our soil collection in Fig. 1. This figure compares (for each cation) the concentration of that cation persisting after the alcohol pre-wash with the concentration of the total cation concentration minus the soluble concentration. For each soil sample these points were graphed against the total cation concentration (x-variable).

Fig. 1.

Cations in different adsorbed cation pools: (a) sodium; (b) potassium; (c) magnesium; and (d) calcium. The x-variable is the total cation concentration (obtained with the BaCl₂/NH₄Cl extract alone). Cation pools were either: the difference between total and soluble concentration (closed circles), or the cation concentration after use of an alcohol pre-wash (open circles). The 1:1 relationship is indicated by the dashed line.

The results show that the two methods (use of alcohol pre-wash or difference between BaCl₂/NH₄Cl extract and soluble extract) gave similar results for K⁺ and Mg²⁺, somewhat different results for Ca²⁺ but very different results for Na⁺. For Na⁺ there was a divergence between the two groups of points across the entire range of total cation concentration tested (Fig. 1a). The result indicates that the use of the alcohol pre-wash stripped about 50% of the adsorbed Na⁺ from the exchange surfaces of the clay. For Ca²⁺ there was a divergence between the two groups of points particularly at moderate to high (8–12 cmol kg⁻¹) total cation concentrations (Fig. 1d), indicating that the use of the alcohol pre-wash stripped Ca²⁺ from exchange surfaces at high total Ca²⁺ concentrations. By contrast, for K⁺ and Mg²⁺, the points for the pre-wash and the difference between BaCl₂/NH₄Cl extract and soluble extract largely overlayed each other (Fig. 1b, c).

Another issue that can be seen from Fig. 1 is that for K⁺, Mg²⁺ and Ca²⁺ the black-filled points (total minus soluble) fell close to the 1:1 line (dashed line) showing that total cations (x-axis) were nearly all in the adsorbed pool (y-axis) (Fig. 1b–d). However, this was not the case for Na⁺ (Fig. 1a). Here the black filled points (total minus soluble; y-axis) were approximately half the values on the x-axis, showing that about half the Na⁺ was in the adsorbed pool and the other half was soluble.

Fig. 2 shows several relationships between different methods of quantifying sodicity. For each of the three graphs, the x-variable is the ESP based on the truly adsorbed cations (BaCl₂/NH₄Cl extract minus soluble extract). In Fig. 2a, the x-variable was correlated with the ESP based on total cations (from the BaCl₂/NH₄Cl extract alone). The composite data set from Moorine Rock and Merredin produced a significant (P < 0.001) line of best fit with a slope of ~1.4, suggesting that about 40% of the Na⁺ in the BaCl₂/NH₄Cl extract was soluble. In Fig. 2b, the x-variable was correlated with the ESP based on the BaCl₂/NH₄Cl extract after alcohol pre-wash. The composite data set produced a significant (P < 0.001) line of best fit with a slope of 0.6, suggesting that about 40% of the exchangeable Na⁺ was lost because of the alcohol pre-wash. In Fig. 2c, the x-variable was correlated with dispersive charge (c.f. Rengasamy et al. 2016) also calculated based on the truly adsorbed cations (BaCl₂/NH₄Cl extract minus soluble extract). The composite data set from Moorine Rock and Merredin produced a significant (P < 0.001) line of best fit with a slope of ~11.

Fig. 2.

Relationships between ESP (based on difference between BaCl₂/NH₄Cl extract and soluble extract; x-variable) and either: (a) ESP based on BaCl₂/NH₄Cl extract alone (y-variable); (b) ESP based on alcohol pre-wash then BaCl₂/NH₄Cl extract (y-variable); or (c) dispersive charge (based on difference between BaCl₂/NH₄Cl extract and soluble extract; y-variable). Although different symbols have been used for data from Moorine Rock and Merredin, the lines of best fit have been drawn using composite data.

Given the outcomes of this section, we have calculated the ESP and dispersive charge assuming each exchangeable cation had to be determined as the total concentration (cmol kg⁻¹ with a BaCl₂/NH₄Cl extract) minus the soluble concentration (cmol kg⁻¹ with a water extract).

2. Characterising the soils: clay type, alkalinity, sodicity and salinity

A database of the soil physical and chemical variables measured on each soil sample is in the Supplementary Material, Tables S1 and S2. The analyses presented below show that the soils at the Moorine Rock and Merredin locations fell into two groups, which affected their alkalinity, sodicity, cation exchange capacity, and salinity.

Clay types

X-ray diffraction spectrometry (XRD) and semi-quantitative analysis were used to determine the clay mineral concentrations for four soil samples from the soil layers (0–0.1 m and 0.4–0.5 m) at Moorine Rock and Merredin (trial ME11). Kaolinite was the dominant clay (66–91%) at each site and depth (Table 3). However, the secondary clays at Moorine Rock were vermiculite (8–16%) and illite (13% at 40–50 cm), whereas the secondary clay at Merredin was illite (29–34%). Vermiculite has a higher cation exchange capacity (100–150 mmol of charge per 100 g) than illite and kaolinite (10–40 and 3–15 mmol of charge per 100 g, respectively; Grim 1968).

Table 3.Clay mineral composition of soils at Moorine Rock and Merredin.

Location	Depth (m)	Clay minerology (%)
Moorine Rock	0–0.1	Kaolinite (91%), vermiculite (8%)
	0.4–0.5	Kaolinite (71%), vermiculite (16%), illite (13%)
Merredin (ME11)	0–0.1	Kaolinite (66%), illite (34%)
	0.4–0.5	Kaolinite (71%), illite (29%)

While a wide range of variables were measured (Supplementary material), variation in cation exchange capacity, pH, sodicity (ESP) and salinity were pivotal. Fig. 3 shows the relative frequency distribution of these four variables at Merredin and Moorine Rock, and Fig. 4 shows the variation in these four variables with average soil depth.

Fig. 3.

Relative frequency distribution of variation in four soil variables: (a) cation exchange capacity (CEC); (b) pH; (c) exchangeable sodium percentage (ESP); and (d) EC_1:5. The data from Merredin and Moorine Rock were separated into different categories.

Fig. 4.

Variation in four soil variables with average depth in the soil profile: (a) cation exchange capacity; (b) pH; (c) exchangeable sodium percentage; and (d) EC_1:5. Each point is the mean of 4–32 replicates. Error bars denote the standard error of the mean. Soils were from Moorine Rock or Merredin; the legend refers to the experiment numbers for the Merredin-based trials (see Table 1).

3. EC_1:5 increases mostly because of one soluble cation (Na⁺) and four soluble anions (Cl⁻, SO₄²⁻, HCO₃⁻, and CO₃²⁻)

This section examines the degree to which EC_1:5 was affected by and correlated with the concentrations of soluble cations and anions (Fig. 5).

Fig. 5.

Relationships between EC_1:5 and soluble ion concentrations: (a) Na⁺, K⁺, Ca²⁺ and Mg²⁺; (b) Cl⁻ and SO₄²⁻; and (c) HCO₃⁻ and CO₃²⁻. The indicated lines of best fit for Na⁺, Cl⁻, SO₄²⁻, and HCO₃⁻ were all significant at P < 0.001.

A useful quality assurance test for datasets like ours is to determine whether the sum of charge from soluble cations equals the sum of charge from soluble anions. The correlation of these two variables was significant (P < 0.001; r² = 0.986) with a slope close to 1.0 (0.92; see Supplementary material, Fig. S3). We are therefore relatively confident that our methods adequately accounted for the cations and anions that could be extracted in water.

4. Cause of high EC_1:5 values: sodicity

Given the high pH, sodicity and EC_1:5 (Fig. 4) that occurred in these soils, it was appropriate to determine whether the evidence pointed to high EC_1:5 values being caused by variation in soil sodicity (or related variables) and/or pH.

Simple linear regression was conducted to determine the effects of a range of variables on EC_1:5 using data aggregated across all sites (results summarised in Table 5). Initial results showed that the incorporation of data from shallow soils added strongly to the residuals in correlations. Therefore, Table 5a shows results for samples from depths greater than 0.2 m and Table 5b shows results for samples from all depths. In both situations, EC_1:5 was most strongly positively correlated with variables indicative of increasing sodicity (dispersive charge, exchangeable Na⁺ and ESP; adj R² values = 65–73% in Table 5a, and 46–59% in Table 5b). EC_1:5 was also negatively correlated with exchangeable Ca²⁺ (adj. R² = 48% in Table 5a, and 17% in Table 5b). The relationship incorporating pH was significant but was not as strong as for other variables (adj. R² = 6% in Table 5a, and 20% in Table 5b).

Fig. 6 shows a scatter-graph of the most significant positive relationship, that between EC_1:5 (y-variable) and dispersive charge (x-variable) for soil samples deeper than 0.2 m, with those from Moorine Rock and Merredin indicated by different symbols. Despite the differences in clay mineralogy, the points from Moorine Rock and Merredin appeared to be co-linear and well-described by the line drawn in common between all points. Based on the line of best fit, EC_1:5 values of 1, 2, and 3 dS m⁻¹ occurred with dispersive charge values of ~240, 450, and 660 cmol charge kg⁻¹ of soil, respectively.

Fig. 6.

Relationship between EC_1:5 (y-variable) and dispersive charge for soils deeper than 0.2 m from Moorine Rock (closed symbols) and Merredin (open symbols). The line of best fit was significant at P < 0.001.

5. Impacts of salts on osmotic potential

Fig. 5 shows that the bulk of the salinity hazard in these soils occurs because of the presence of NaHCO₃, NaCl and Na₂SO₄, with Na₂CO₃ making up a small proportion of the total. Each of these salts is highly soluble, which means that their concentrations in the soil solution can be estimated from their concentrations in 1:5 soil:water extracts (c.f. Table 2). It also means that their contribution to the soil osmotic potential (Ψ_s) can be calculated based on previously determined relationships between each salt and Ψ_s (c.f. Wolf et al. 1986).

Osmotic potentials were estimated based on four assumptions. First, the osmotic potential of each soil sample was due to the presence of four anions: (1) Cl⁻; (2) SO₄²⁻; (3) HCO₃⁻; and (4) CO₃²⁻, with Na⁺ as cation (Fig. 5). Second, the four sodium salts of these anions were dissolved either in the soil water content at the time of drilling or in a standard amount of water (30% soil dry weight). Third, the osmotic potential due to each salt could be calculated based on relationships derived from the data of Wolf et al. (1986). Fourth, total osmotic potential was the sum of the osmotic potentials of each individual component salt

Soil water content

Prior to sampling (in July 2022), the soils at Moorine Rock and Merredin received between ~65 mm and 69 mm of rain in May–June, respectively (based on nearby Bureau of Meteorology rainfall stations). With this early season rain, the soils at Moorine Rock contained an average of 11–14% gravimetric water in the upper 0.2 m of the soil profile, and 14–17% water in the deeper soil. The soils from Merredin contained an average of 13–18% water in the upper 0.2 m of the soil profile, and 12–17% water in the deeper soil (Fig. 7a).

Fig. 7.

Variation in gravimetric soil water content and osmotic potential with average depth in the soil profile: (a) soil water content at the time of sampling; (b) osmotic potential estimated assuming the soil water content of (a); and (c) osmotic potential estimated assuming a gravimetric soil water content of 30%. Each point is the mean of 4–32 replicates. Error bars denote the standard error of the mean. Soils were from Moorine Rock or Merredin; the legend refers to the experiment numbers for the Merredin-based trials (see Table 1).

Osmotic potentials

Soil osmotic potentials in the different soil layers were estimated assuming that the salts were either dissolved in the soil water described above (Fig. 7b) or were dissolved in the water of soils hydrated to 30% moisture (Fig. 7c). The relative frequency of osmotic potentials was also determined for soil samples from the surface (0–0.2 m) and subsoil at Merredin and Moorine Rock; here the soils were separated into four categories with osmotic potentials between 0 and −0.5 MPa, −0.5 and −1.0 MPa, −1.0 and −1.5 MPa, and less than −1.5 MPa. (Fig. 8). For crops, osmotic potentials less than −0.5 MPa are likely to jeopardise the production of an economic yield (see Discussion).

Fig. 8.

Relative frequency distribution of osmotic potentials for Merredin and Moorine Rock at two levels of soil water content: the low water contents at sampling (a, c); and 30% gravimetric soil moisture (b, d). The data have been categorised according to soil depth: surface soil is at 0–0.2 m; subsoil is at depths >0.2 m.

When osmotic potentials were calculated based on the water content at the time of sampling, there was strong stratification of osmotic potentials at Merredin (less negative in the surface soil, more negative in the subsoil) (Fig. 7b). At Merredin, most (88%) surface soil samples had osmotic potentials between 0 and −0.5 MPa, but there were few (only 9%) of subsoil samples in this class. Most (56%) subsoil samples had osmotic potentials between −1.0 and −1.5 MPa (Fig. 8a). By contrast, at Moorine Rock this stratification was not evident (Fig. 7b). Most surface soils (75%) and subsoils (92%) had osmotic potentials less than −1.5 MPa, and there were no soil samples in either depth range in the least severely affected class (from 0 to −0.5 MPa) (Fig. 8c).

Increasing soil water content to 30% made osmotic potentials less negative (Fig. 7c) which affected the distribution of soil samples across osmotic potential classes. At Merredin, the stratification of osmotic potential with depth was still evident (Fig. 7c), but 100% of surface soils were in the least severely affected class (from 0 to −0.5 MPa); there there were many (40%) subsoil samples in this class, but most (58%) subsoil samples were in the class from −0.5 to −1.0 MPa (Fig. 8b).

For Moorine Rock, increasing soil water content had a strong effect on the distribution of samples by osmotic potential (Figs 7c, 8d). In the surface soil, some (19%) samples were in the least severely affected class (from 0 to −0.5 MPa) and many (50%) were in the next class (from −0.5 to −1.0 MPa). However, the subsoil was still strongly osmotically affected with most samples (66%) in the class from −1.0 to −1.5 MPa (Fig. 8d).

Measurement of exchangeable cations in soils affected by salinity

The idea that special care needs to be taken in the measurement of exchangeable cations in soils affected by salinity is not particularly new, but this key message is frequently ignored. Bower et al. (1952) proposed that exchangeable cations be determined from the difference between total cations (determined in an ammonium acetate extract) and soluble ions (determined in a saturation extract), and this approach was then widely promoted through USDA Handbook 60 (United States Salinity Laboratory Staff 1954). More recently, this case has been remade for Australia by So et al. (2006) using clay soils from Queensland with dominant clay species smectite and kaolinite.

The erroneous inclusion of soluble Na⁺ into the exchangeable component can have adverse consequences for the classification of soils. In the present work, using soils with EC_1:5 values between 0.1 and 19.6 dS m⁻¹ we found that the use of BaCl₂/NH₄Cl extractions over-represented adsorbed Na⁺ by ~50% (Fig. 1). This had consequences for the assessment of sodicity, exaggerating values of ESP by ~40% (see slope of line of best fit in Fig. 2a).

When ESP was calculated based on total cations after alcohol pre-wash, values were ~40% lower than when calculated with the truly adsorbed cations (i.e. BaCl₂/NH₄Cl extract minus soluble extract) (see slope of line of best fit in Fig. 2b). This shows that about 40% of the truly adsorbed Na⁺ is weakly held on the exchange surfaces of the clay and is easily leached off. A similar conclusion was reached by So et al. (2006).

In their seminal work ‘Australian soils with saline and sodic properties’, Northcote and Skene (1972) proposed that sodic soils in Australia should be defined as having ESP values >6 and highly sodic soils should be defined as having ESP values >15. At the present time, there are no published critical values defining sodic and highly sodic soils based on dispersive charge. It is unclear whether the original Northcote and Skene assessment was made using soils affected by salinity, but if we take their critical values for ESP at face value then our correlation between ESP and dispersive charge (Fig. 2c) suggests that the critical dispersive charge limit could be 115 cmol charge kg⁻¹ for ‘sodic’ soils and 215 cmol charge kg⁻¹ for ‘highly sodic’ soils (c.f. Fig. 2c). It would be useful to see these values confirmed using reliable ESP and dispersive charge data from other regions.

Ions causing salinity

One feature of our dataset is the diversity of ions responsible for salinity (Na⁺, Cl⁻, HCO₃⁻, SO₄²⁻, CO₃²⁻). Because Cl⁻ is the dominant anion in seawater (accounting for ~90% of the anionic charge; Lyman and Fleming 1940), there has been a tendency to think that Cl⁻ should be the dominant anion in saline soils in Australia. Our dataset shows that this should not be assumed. HCO₃⁻ was at higher concentrations than Cl⁻ in 167 (43%) of the 388 soil samples measured (see Supplementary material Table S2).

Causes of salinity

Our data lead us to suggest that there are three drivers of salinity in Calcarosols: (1) sodicity decreases the permeability of the soils, resulting in the accumulation of salts from rainfall and the atmosphere; (2) soil alkalinity leads to increased bicarbonate and carbonate; and (3) soil salinities are impacted by rainfall and evaporation.

Impacts on osmotic potential

Salt affects crop growth for two major reasons: (1) adverse effects on plant water relations; and (2) specific ion effects (i.e. toxicities) that directly adversely affect cell metabolism (Greenway and Munns 1980; Munns and Tester 2008). The former effects occur because salinity decreases the osmotic potential of the soil solution; this lowers the water potential of the soil and places plants under osmotic stress. The latter effect occurs because, when taken up, the ions adversely affect metabolism. For example, Na⁺ can adversely affect many of the metabolic functions in plants that are normally facilitated by K⁺ (Munns and Tester 2008; Shabala and Cuin 2008).

Total water potential is the sum of the soil’s osmotic potential (Ψ_s) and matrix potential (Ψ_m), both negative numbers. At the water contents reported in Fig. 6a, Ψ_m values would presumably also have been low (negative) but we are not able to quantify these effects using our data. However, we can calculate Ψ_s values using soil ion concentration and water content measurements, which provides a sense of the magnitude of crop water stress.

What do the osmotic potential values reported in Figs 7 and 8 mean for crop growth? Published data (Maas and Hoffman 1977; Steppuhn et al. 2005a, 2005b) suggest that the yields of most crops can be expected to become unprofitable as osmotic potentials decrease from 0 to −0.5 MPa and plant survival will be threatened at more negative osmotic potentials. In addition, osmotic potentials in the range from 0 to −0.5 MPa can substantially decrease germination (for our summary see Supplementary material, Appendix S2). We conclude that for grain crops in Australia, profitability will be negligible at osmotic potentials less than −0.5 MPa.

In this work, we have calculated soil osmotic potential (MPa) as the sum of the individual osmotic potentials due to NaCl, NaHCO₃, Na₂SO₄, and Na₂CO₃. We need to ask whether there is there any point in doing this, or whether the results are about the same as would be achieved by multiplying the EC_1:5 (dS m⁻¹) by the dilution factor of the 1:5 extract relative to the soil (i.e. 500/W; Rengasamy 2010) and a general factor for converting EC to Ψ_s (i.e −0.036; Marschner 1995). We have compared the two methods of calculating osmotic potential in Supplementary material Fig. S4 for osmotic potentials calculated using the measured gravimetric soil water content in the soil (Fig. 7a). The correlation between the two methods had a slope of 1.07 and an r² value of 0.98. This outcome supports the view that the two methods give similar results, and it might not be necessary to go to the trouble of measuring soluble anions for the estimation of osmotic potential in future work.

One of the strong messages from this paper is that higher soil water content is important in increasing (making less negative) osmotic potentials (c.f. Figs 7b, 8a, and c with low water content, and Figs 7c, 8b, and d with 30% soil moisture). At Moorine Rock, in surface soils osmotic potentials were occasionally greater (less negative) than −0.5 MPa with 30% soil water but were never greater than −0.5 MPa at low water content. By the contrast, the effects of moisture were less severe at Merredin. When osmotic potentials were calculated based on the soil water contents at sampling, osmotic potentials were mostly greater than −0.5 MPa in surface soils and were all greater than −0.5 MPa in surface soils with 30% soil moisture. However, this differed in the subsoil. With the soil water at sampling, most (~90%) of soil samples had osmotic potentials less than −0.5 MPa, and even with the higher level of soil moisture about 60% of soil samples had osmotic potentials less than −0.5 MPa (Fig. 8).

Three conclusions emerge from these data: (1) we can expect that crop growth at Moorine Rock will be severely affected by low osmotic potentials; (2) it is likely that crop rooting depths will be constrained by low osmotic potentials in subsoils at Merredin; and (3) the adverse effects of salinity will be more severe in dry than wet growing seasons.

We conclude with a final point. Our measurements of the salinity of the soil solution and osmotic potential have been estimated for the bulk soil. However, it needs to be acknowledged that in clayey soils plant roots will tend to grow mostly through macropores. Salt concentrations in the soil solutions lining macropores might differ from those of the bulk soil either because of accelerated leaching during periods of rainfall (c.f. Allaire et al. 2009), or because of salt accumulation in the root-zone caused by the mass flow of the soil solution to the root surface and the uptake there of water but not salt (c.f. Alharby et al. 2014). There is therefore a need for work to be conducted to examine the variation in estimated salinities in exposed soil faces at millimetre-scale paying particular attention to the salinity of the soil solution in massive peds and macropores.