Short- and Long-Term Effects of Lime and Gypsum Applications on Acid Soils in a Water-Limited Environment: 3. Soil Solution Chemistry

: Aluminum (Al) toxicity imposes a signiﬁcant limitation to crop production in South Western Australia. This paper examines the impact of surface-applied lime and gypsum on soil solution chemistry in the short term (1 year) and the long-term (10 years) in water limited environments. In the experiments, we measured soil solution chemistry using a paste extract on soil proﬁle samples collected to a depth of 50 cm. We then used the chemical equilibrium model MINTEQ to predict the presence and relative concentrations of Al species that are toxic to root growth (Al associated with Al 3+ and AlOH 2 or Toxic-Al) and less non-toxic forms of Al bound with sulfate, other hydroxide species and organic matter. A feature of the soils used in the experiment is that they have a low capacity to adsorb sulfate. In the short term, despite the low amount of rainfall (279 mm), sulfate derived from the surface gypsum application is rapidly leached into the soil proﬁle. There was no self-liming effect, as evidenced by there being no change in soil solution pH. The application of gypsum, in the short term, increased soil solution ionic strength by 524–681% in the 0–10 cm soil layer declining to 75–109% in the 30–40 cm soil layer due to an increase in soil solution sulfate and calcium concentrations. Calcium from the gypsum application displaces Al from the exchange sites to increase soil solution Al activity in the gypsum treatments by 155–233% in the short term and by 70–196% in the long term to a depth of 40 cm. However, there was no effect on Toxic-Al due to Al sulfate precipitation. In the long term, sulfate leaching from the soil proﬁle results in a decline in soil solution ionic strength. Application of lime results in leaching of alkalinity into the soil proﬁle leading to a decreased Toxic-Al to a depth of 30 cm in the long term, but it did not affect Toxic-Al in the short term. Combining an application of lime with gypsum had the same impact on soil solution properties as gypsum alone in the short term and as lime alone in the long term. − 2 , and AlHSO + 2 were generated from the MINTEQ model. We then grouped these species as follows: (1) Org-Al, (2) Toxic-Al as Al 3+ (3) OH-Al as AlOH 2 + , Al ( OH ) + 2 , Al ( OH ) − 4 , Al ( OH ) 3 , and (4) SO 4 -Al as AlSO + 4 , Al ( SO 4 ) − 2 , and AlHSO + 2 . We present the results as Al species activities ( µ mol L − 1 ). The ﬂuoride and phosphate concentrations were below the detection limits. We did not use the concentration


Introduction
Aluminum (Al) toxicity results in reduced crop yields when subsoil layers, layers below 10 cm, have 0.01 M CaCl 2 extractable Al (Al CaCl 2 ) concentrations greater than 2.5-4.5 mg Al kg −1 [1]. The use of lime (L, CaCO 3 ) and gypsum (G, CaSO 4 ) applications reduces subsoil Al toxicity resulting in increased crop grain yield [2][3][4][5]. Gypsum use can be more profitable than L application in overcoming subsoil Al toxicity in the South Western Australia agricultural region's low rainfall zone in the short term. This is due to G having comparatively higher solubility than L and the low sulfate (SO 4 -S) sorption properties of some soils in this region [5]. The limited SO 4 -S sorption combined with high June and July when the Al Act is greater than 480 µmol L −1 [38], while for less tolerant crops, there is a lower critical value of 100 µmol L −1 for mung bean [39] and for 9-15 µmol L −1 for white clover [40].
The paper aims to examine the impact of a factorial combination of L and G applications on soil solution chemical properties and Al speciation in the short term (ST, one year) and the long term (LT, 10 years). The soil profiles have a low SO 4 -S adsorption capacity or a sulfate buffer index (SBI) of <35 [5,6]. Previous papers in this series have highlighted the impact of these treatments on crop grain yield, nutrient concentration and soil chemical properties [5,8]. We hypothesize that soil solution measurements will provide greater insight into the soil chemical reactions than soil test measurements, which occur due to G and L application to soils with a low SBI and hence explain the crop yield responses observed by [5].

Field Experimental Sites
This paper reports soil solution chemistry obtained from the ST and LT, which have received a factorial combination of L and G application presented by [5]. The first experiment was a ST trial located near South Burracoppin (−31 • 30 9" S latitude, 118 • 38 50" E longitude), which had 2-4 t L ha −1 applied in the 15 years prior to the experiment. At Bonnie Rock (−30 • 37 3" S latitude, 118 • 14 9" E longitude), the second experiment was a LT experiment on a site with no previous L use. The soil type for both sites was a Tenosol according to the Australia Soil Classification System or Arenosols under the World Reference Base for Soil Resources [41]. At the Bonnie Rock site, the gravel content increased with increasing soil depth [5]. Both experiments consisted of small plots of 1.8 m by 20 m arranged in a complete randomized block design with three replications.

Soil Measurements
We measured soil solution chemical properties on soil samples collected to a depth of 50 cm, in increments of 10 cm on 20-22 March 2018. For the ST experiment, we collected soil samples from the control (C), 2.5 t L ha −1 (L2.5), 2.5 t G ha −1 (G2.5), 2.5 t L + G ha −1 (L2.5 + G2.5) treatments applied on 31 March 2017. For the LT experiment, we used the control (C), 4 t L ha −1 (L4), 2 t G ha −1 (G2), and 4 t L ha −1 + 2 t G ha −1 (L4 + G2) treatments with half of these rates (2 t L ha −1 and 1 t G ha −1 ) in March 2008 and the other half in March 2013. When we present the two experimental sites, the factorial treatment abbreviations used are C, L, G, and L + G. More details of the soil sampling procedures are giving in [5,8].
Previous researchers extracted soil solution from air-dry soil that had been wet to a tension of 0.1 bar and centrifuged at a force 9009× g for 45 min [42]. Our initial aim was to remove the soil solution using this approach. However, we were only able to extract a small volume of soil solution, which was insufficient to undertake all of the measurements. Also, we had a limited amount of soil sample available to do the soil solution extractions. Hence, we used a saturated paste extract with a solution to soil ratios (ml water to g soil) of 0.4:1.0 for the 0-10 and 10-20 cm layers and 0.35:1.0 for the other sampling three depths between 20 and 50 cm. We used different ratios due to greater water retention by the 0-20 cm samples compared to the 20-50 cm samples. We acknowledge the best method to extract soil solution is provided by [42]. However, we note other Al Soln chemical species studies have used water extracts using a range of solution to soil ratios of 1:1, 0.40:1, 0.20:1, and 0.10:1 [42][43][44] because it is difficult to measure the true soil solution [42].
The water extraction procedure involved weighing 200 g of soil into a 500 mL beaker. We then added 80 gm of Milli-Q ® water to the 0-10 and 10-20 cm and 70 gm to the below 30 cm soil samples. We then covered the beakers with glad wrap to stop evaporation loss and incubated them for 16 h at 25 • C. We then extracted the soil solution by centrifuging the soil for 10 min at 9223× g, followed by filtration using a 0.45 µm cellulose membrane.
We then measured the following chemical properties of the saturated paste extract solution which from this point onwards we refer to as the "soil solution". The solution's alkalinity (Alk Soln ) is the amount of acid (mg L −1 ) required to lower pH CaCl 2 to 4.3 [45]. We used ICP-AES, which measures the total concentration (mg L −1 ) of the following ions in solution; Al Soln , Ca Soln , iron, potassium, magnesium, sodium, phosphate, SO 4 -S Soln , silicon. We measured redox potential (mV), pH Soln , and fluoride concentration using an ion-specific electrode. Nitrogen concentration (mg L −1 ) as ammonium and nitrate was measured colorimetrically [45]. Dissolved organic carbon in the soil solution (DOC Soln ) concentration (mg L −1 ) was determined using a modified [46] method for soil solution using a sample size of 5 mL. We have compared soil solution measured (pH Soln , Al Act , Ca Soln and SO 4 -S Soln ) to the corresponding soil chemical measurements pH CaCl 2 , Al CaCl 2 , (pH and Al extracted by 0.01 M CaCl 2 ) and exchangeable Al Ex and Ca Ex measured by [47], and SO 4 -S removed by 0.25 M KCl heated at 40 • C for three hours (S KCl40 ) [48].
Using a greater solution to soil ratios than the method recommended by [42] dilutes the extracted ion's concentration. In acidic soils, water stress resulted in the greater appearance of Al toxicity symptoms in roots than for the treatments without water stress [43], suggesting a rapid increase in the Al concentration in the soil solutions with decreasing soil water [49]. We confirmed this observation in a small incubation experiment using five solutions to soil ratios ranging from 0.2:1.0 to 0.5:1.0 decrease Al concentration from 1.6 to 0.9 mg L −1 . Hence, we calculated ion concentrations at a water content of 0.14 L kg −1 for the 0-10 cm layer and 0.10 L kg −1 for the layers below 10 cm, or the drainage-upper limit observed in Tenosols [50].

Ionic Strength
The IS Soln (mmol L −1 ) was calculated from the electrical conductivity (EC) measurements using the formula of [42]: where EC is the electrical conductivity (dS m −1 ) measured in a 1:5 soil to water solution. Al concentration is expressed in soil solution studies as the Al ion activity in the soil solution [51]. The activity of Al in the soil solution (Al Act ) is the product between Al Soln and activity coefficient (y i ): Al Act = Al Soln × y i , Debye-Huckel's equation is used to calculate the activity coefficient (y i ) [27]: where A = 0.509 for pure water at 25 • C; Zi = ion valency charge; µ = IS Soln .

Aluminum Species
The measured values for DOC Soln , Alk Soln , pH Soln , Al Soln , Ca Soln , and SO 4 -S Soln were used as inputs for the chemical equilibrium model, MINTEQ version 3.1 to predict the composition of chemical species [25]. Since soil solution Al was measured using an ICP-AES, it represents the total concentration in the soil solution (Total-Al) or inorganic Al (Inorg-Al) plus organic Al (Org-Al) [36]. To account for this effect, we calculated Org-Al using the equation of [35]. We then calculated Inorg-Al as Total-Al minus Org-Al. We then entered Inorg-Al into MINTEQ, as the input Al concentration input and used the Gaussian dissolved organic matter model to calculate Org-Al [25,52]. The model inputs predicted a charge balance difference of less than 10%, indicating that MINTEQ provides an accurate prediction of Al species. The Al species, Org-Al, Al 3+ , AlOH 2+ , Al(OH) + 2 , Al(OH) − 4 , Al(OH ) 3 , AlSO + 4 , Al(SO 4 ) − 2 , and AlHSO + 2 were generated from the MINTEQ model. We then grouped these species as follows: (1) Org-Al, (2) Toxic-Al as Al 3+ (3) OH-Al as AlOH 2+ , Al(OH) + 2 , Al(OH) − 4 , Al(OH ) 3 , and (4) SO 4 -Al as AlSO + 4 , Al(SO 4 ) − 2 , and AlHSO + 2 . We present the results as Al species activities (µmol L −1 ). The fluoride and phosphate concentrations were below the detection limits. We did not use the concentration (mg L −1 ) of iron, potassium, magnesium, sodium, ammonium, nitrate, and silicon to run the model because these ions are not associated with Al species and including these measurements resulted in a charge difference of greater than 10%. We then present Org-Al, Toxic-Al, OH-Al, and SO 4 -Al species in the paper's Results and Discussion sections.

Statistical Analysis
Analysis of variance (ANOVA) to determine the significance of treatments effects on the soil solution chemical properties within soil layers of the soil profile to a sampling depth of 50 cm was done using Genstat ® [53]. We report treatment responses if the treatment resulted in a change in the soil solution measurement of greater than the least significant difference values determined at p = 0.05.
We used the SigmaPlot ® version 12.5 [54] regression wizard to define relationships between soil and soil solution measures. Linear equation (y = a + [b × x]) were fitted for most relationships. The exception was the use of exponential equations to define the relationship between S KCl40 and SO 4 and between pH CaCl 2 and Al Soln (Al Soln = a + b × exp −c × pH CaCl2 .

Soil Solution Measurement
The C treatment's soil profile SO 4 -S Soln values decreased with increasing depth in both the ST and LT experiments (Figure 1a (Figure 1a). In the LT, the G2 treatment increased SO 4 -S Soln by 681% in the 0-10 cm layer, which was 123% greater than SO 4 -S Soln for the L4 + G2 treatment ( Figure 1b). The G2 and L4 + G2 treatments also increased SO 4 -S Soln by 244-408% in the 10-30, which declined to 92-118% in the 40-50 cm layer. SO 4 -S Soln was lower for the L + G treatment than the L treatment in the 0-10 cm layer.
For the C treatment, Ca Soln decreased with increasing depth for both the ST and LT experiments (Figure 1c (Figure 1e,f). Only G application increased IS Soln in these experiments. In the ST, IS Soln was greater for the G2.5 treatments than the C by 524-681% in the 0-10 cm layer, declining to a 75-109% increase in the 30-40 cm layer (Figure 1e). The IS Soln for the L2.5 (1.4-3.1 mmol L −1 ) was the same as the C treatment for all layers. In the LT, IS Soln for the G2 treatment was 192% greater than the C in the 0-10 cm layer (Figure 1f). The IS Soln for the G2 and L4 + G2 treatments was 41-99% greater than the C in the 10-50 cm layer. For the L4 treatment, IS Soln was the same as the C treatments in all layers. In the ST experiment, Al Act for the C treatment was 4 µmol L −1 in the 0-10 cm but much greater (65-83 µmol L −1 ) in the 10-50 cm layer (Figure 2a). In the LT experiment, the highest soil Al Act for the C treatment of 65 µmol L −1 occurred in the 10-20 cm layer (Figure 2b). Al Act was also relatively high (36-33 µmol L −1 ) in the 0-10 and 20-30 cm layer but lower (13 µmol L −1 ) in the 30-50 cm layer. The application of G increased Al Act in both the ST and LT, while L application resulted in a decrease in Al Act in the LT. In the ST, Al Act for the G2.5 treatment was 233% greater than the C in the 20-30 cm layer (Figure 2a). Also, Al Act for the G2.5 and L2.5 + G2.5 treatments was 155-157% greater than the C in the 30-40 cm layer. In the LT, Al Act for the L4 and L4 + G2 treatments was 67-96% lower than the C treatment in the 0-20 cm layer. In contrast, for the G2 treatment, Al Act was greater by 70-89% in the 0-20 cm layer and by 100-196% in the 20-40 cm layer (Figure 2b). Al Act for the L4 + G2 treatment was the same as the L4 treatments in the 0-50 cm layer.
In the ST experiment, pH Soln for the C treatment was 6.7 in the 0-10 cm. The pH Soln was lower (4.4-4.9) in the 10-50 cm layer (Figure 2c). In the LT experiment, pH Soln for the C treatment ranged within 4.3 to 4.6 in the 0-50 cm layer with the lowest pH Soln existing in the 10-30 cm layer (Figure 2d). Only L application increased pH Soln in these experiments. The L and L + G treatments increased pH Soln from 5.4 to 6.9-7.4 in the ST and from 4.6 to 6.9 in the LT (Figure 2c,d). In the LT, pH Soln in the 10-20 cm layer for the L4 and L4 + G2 treatments was 5.7 compared 4.3 for the C treatment, which not significant due to the high variability in pH Soln observed for this layer. In contrast, G2 application did not affect pH Soln in either experiment.
The Alk Soln value for the C treatment was 3.3 mg L −1 in the 0-10 cm layer for the ST experiment. Lower Alk Soln values (0.2-0.6 mg L −1 ) existed in the 10-50 cm layers of the ST experiment and the 0-50 cm layers of the LT experiment (Figure 2e,f). In these experiments, L and L + G application increased Alk Soln . In the ST, the Alk Soln for the L2.5 and L2.5 + G2.5 treatments was 275-388% greater in the 0-10 cm layer than the C treatment (Figure 2e). In the LT, Alk Soln for the L4 and L4 + G2 treatments was 275-388% greater than the C treatment in the 0-10 cm layer (Figure 2f). In contrast, G application alone did not affect Alk Soln in either experiment.
The highest DOC Soln (14-31 mg L −1 ) appeared in the 0-10 cm layer with lower (4-9 mg L −1 ) DOC Soln in the 10-50 cm layer for both experiments (Figure 3a,b). Application of L and G increased in DOC Soln in the ST experiment. In the ST, the L2.5 and G2.5 treatments increased DOC Soln by 65% in the 0-10 cm layer, while the L2.5 + G2.5 treatment increased the DOC Soln by 110% compared to the control treatment (Figure 3a). In the LT, the L4, G2, and L4 + G2 treatments may have increased DOC Soln in the 0-20 cm layer, but this difference was not significant due to the large variation in DOC Soln at the site (Figure 3b).   Figure 4a presents the relationship between soil pH CaCl 2 and pH Soln . As pH CaCl 2 increased, the difference between pH CaCl 2 and pH Soln increased, as indicated by the linear regression slope, which was equal to 1.54 (Table 1). Table 1. Regression coefficients for the relationship between soil and soil solution measurements of pH, SO 4 -S, Al and Ca across the control (C), lime (L), gypsum (G) and L + G treatments in the short term (ST) and long-term (LT). The soil measurements examined were pH, and Al were measured using 0.01 M CaCl 2 (pH CaCl 2 and Al CaCl 2 ), exchangeable Al (Al Ex ) and SO 4 -S extracted by 0.25 M KCl heated for 3 h at 40 • C. The soil solution measurements using a paste extract to measure pH (pH Soln ), SO 4 -S (SO 4 -S Soln ), Ca (Ca Soln ), and Al activity (Al Act ). See Figure 4 for plotted relationships.   Table 1 for regression equations for fitted lines.

Comparison
In the ST, for the L2.5 + G2.5 treatment, pH CaCl 2 was lower with an associated greater pH Soln compared to the other treatments. Soil S KCl40 was related to SO 4 -S Soln (r 2 = 0.59) for both experiments across all treatments (Figure 4b). Soil Al CaCl 2 was related to Al Act (r 2 = 0.71), but only for the C and L treatments (Figure 4c). Al Act concentrations were greater for the L2.5 + G2.5 and L4 + G2 treatments compared to the C treatment in both experiments. There was a poor relationship between soil pH CaCl 2 and Al Act (r 2 = 0.42) across all treatments with the highest Al Act observed for the ST trial for the G2 treatments (Figure 4d). Soil Al Ex was related to Al Act (r 2 = 0.64), but only for the C and L treatments (Figure 4e). The Al Act concentration for a measured Al Ex was greater for the L2.5 + G2.5 and L4 + G2 treatments than the C and L treatments in both experiments. Soil Ca Ex was related to Ca Soln (r 2 = 0.87) for the C and L treatments with greater concentrations occurring for the G2.5 treatment in the ST experiment (Figure 4f).

Soil Solution Al Species
In the ST, Org-Al ranged within 0 to 16 µmol L −1 in the 0-50 cm layer of all treatments (Figure 5a). The L2.5, G2.5, and L2.5 + G2.5 treatments had 48-75% lower Org-Al than the C treatment in the 10-30 cm layer. Org-Al was variable among treatments (8-15 µmol L −1 ) in the 30-50 cm layer. The OH-Al decreased with increasing depth from 4 µmol L −1 in the 0-10 cm layer to 1 µmol L −1 in the 20-50 cm layer (Figure 5b). The treatments did not affect OH-Al in 0-50 cm layer. The Toxic-Al increased with increasing depth from 0 µmol L −1 in the 0-10 cm layer to 9 µmol L −1 in the 30-50 cm layer regardless of treatments (Figure 5c). There was no SO 4 -Al in the 0-10 cm layer of all treatments (Figure 5d). Application of G increased SO 4 -Al by 174-1039%, 432-666%, and 299-301% in the 10-20, 20-30, and 30-40 cm layers, respectively, compared to the C and L2.5 treatments. In the LT, the L4 and L4 + G2 treatments had 100% lower Org-Al than the C treatment in the 0-10 cm layer (Figure 6a). Org-Al was 258-405% greater for the L2 and G2 treatments than the C treatment in the 10-20 cm layer. Org-Al was low (1-7 µmol L −1 ) for all treatments in the 30-50 cm layer. The L and G treatments did not affect OH-Al with concentrations ranging between 0.2-2.5 µmol L −1 in the 0-20 cm layer (Figure 6b). The Toxic-Al was lower for the L4 and L4 + G2 treatments by 100%, and 87-90% (at p = 0.053) and 32-33% in the 0-10, 10-20, 30-40 cm layers, respectively, than the control treatment (Figure 6c). In contrast, Toxic-Al was the same in the soil profile for the C and G2 treatments. The application of G2 resulted in greater SO 4 -Al by 330%, 137%, 188%, and 397% in the 0-10, 10-20, 20-30, 30-40 cm layers, respectively, compared to the C treatment (Figure 6d). SO 4 -Al was lower for the L4 + G2 treatment than the G2 treatment by 90-100% in the 0-20 and by 52-71% in the 20-40 cm soil layer. Soil S KCl40 is related to per cent of Al Act as SO 4 -Al (SO 4 -Al%) (r 2 = 0.59) (Figure 7a). In contrast, there was no relationship between SO 4 -S Soln and SO 4 -Al% (Figure 7b). In comparison, Toxic-Al was related to pH Soln more than pH CaCl 2 (Figure 8a,b), but not related to either Al CaCl 2 or Al Soln (data not presented).

Discussion
The Al species likely to be present when the soil pH Soln is below 5.0 are Org-Al, Toxic-Al, and SO 4 -Al [21,32]. The MINTEQ-predicted Toxic-Al activity in the soil profiles was up to 16 µmol L −1 in the sublayers (Figures 5c and 6c), which is above the critical range for harmful Al toxicity effects on Al sensitive plant growth [40]. In the LT, the application of L alone or in combination with G reduced Toxic-Al to a depth of 30 cm resulting in increased crop grain yield [5].
Changes in soil solution properties (SO 4 -S Soln , Ca Soln , Al Act , and SO 4 -Al) due to the application of G provide greater insight into the behaviour of soil Al and its effects on crop yield than the measured changes in soil chemical properties (S KCl40 , Ca Ex , Al CaCl 2 , Al Ex , and SO 4 -Al) [8]. In the ST, application of G2.5 increased the soil solution measurement SO 4 -S Soln (Figure 1a), Ca Soln (Figure 1c), IS Soln (Figure 1e), Al Act (Figure 2a), and SO 4 -Al (Figure 5d). The increase is Ca Soln results in the Ca 2+ displacing Al 3+ from the exchange site increasing Al Act , which in turn decreases pH Soln due to the associated release of H + from the cation exchange sites into the soil solution [15,17,18]. We did not identify this reaction using the soil measurements [8]. Instead, we observed G application increased S KCl40 , Ca Ex , and ECEC resulting in a small decrease in Al CaCl 2 and no change in Al Ex [8]. However, complexation of Al 3+ as SO 4 -Al also takes place, as predicted by MINTEQ. The G treatment's overall effect was to have the same Toxic-Al concentration as the control treatment ( Figure 5c) with pH Soln closely related to Toxic-Al (Figure 8a). The formation of SO 4 -Al appears to be sufficient to overcome the rise in Al toxicity associated with the greater Al Act when G is applied [55]. Therefore, the soil solution measurements confirm that the mechanism for the greater crop grain yield observed for the G treatments in the ST [5] is due to the greater SO 4 -S Soln and Ca Soln or greater S KCl40 and Ca Ex increasing IS Soln or EC resulting in the amelioration of the Al toxicity effect [15]. Simultaneously, the SO 4 -Al complexation plays an essential role in maintaining toxic-Al in the soil solution [2] at the same concentrations as the C treatment (Figures 5c and 8b). The ability of SO 4 -S to form SO 4 -Al is dependent on the SO 4 -S Soln concentration [31]. However, in this study, SO 4 -Al% was more related to S KCl40 (Figure 7a) than SO 4 -S Soln (Figure 7b) because of low SO 4 -Al% relative to SO 4 -S Soln for the C and L treatments in the LT experiment (Figure 7b).
Soil solution Al only accounts for less than 1% of Al Ex [56]. Similarly, in this study, soil solution, Al accounted on average for only 0.47% of the Al Ex . Hence, Al Ex acts as a reserve of Al to buffer Al Soln from changes when G is applied to the soil [56]. However, Al Act is poorly correlated with Al Ex [56] and Al CaCl 2 when examined across all treatments (Figure 4c,e). The highest regression correlation of r 2 = 0.41 across all treatments occurred for the relationship between pH Soln, and Al Act (Figure 4d), which is consistent with the observations of [56]. Mineral phase reactions can control Al Act [27]. For example, for a group of Queensland soils with G application, the equation for the relationship between pH Soln and log(Al Act ) is the same as the theoretical values for gibbsite dissolution (Log(Al Act ) = 8.04 − (3.00 × pH Soln )) [56]. However, for the LT experiment, log(Al Act ) the C and G treatments were greater than the theoretical values for gibbsite dissolution. Also, the lime treated soils of the ST experiment had high log(Al Act ) for the measured pH Soln because L application can result in the formation of trace quantities of more soluble Al minerals that subsequently controlled Al Act [57]. Hence, Al Act for Western Australian soils result from several mineral phase retention and release reactions [26].
The IS Soln values for the experimental sites were low, less than 2.5 mmol L −1 (Figure 1e,f), which is consistent with other measurements conducted on highly weathered soils from Queensland and the South Western Australia wheatbelt [42,58]. Hence, these soils have a high expression of Al toxicity due to the low IS Soln and EC. In the ST, G2.5 application increased IS Soln to 16.9 mmol L −1 in the 0-10 cm layer and 3.2-4.9 mmol L −1 in the subsoil (Figure 1e). The effect is to decrease the Al activity leading to a 14-19% increase in crop grain yield [5,15]. The immediate impact of G on these soil solution properties is illustrated by the leaching of SO 4 -S Soln (Figure 1a), and Ca Soln (Figure 1c), increasing IS Soln (Figure 1e) to the depth of 40 cm even after only 279 mm of rainfall over the period March 2017 to March 2018 [5].
In the LT, SO 4 -S and Ca applied as G is retained within the 50 cm layer as indicated by greater SO 4 -S Soln (Figure 1b), Ca Soln (Figure 1d), S KCl40 and Ca Ex [8]. However, SO 4 -S Soln was leached below the 50 cm [8], decreasing the effectiveness of the G2 treatment on crop yield increase overtime in the LT experiment [5]. There was an initial grain yield response of 16-23% in the LT experiment in the first year after application declining to 4-9% four years after the G2 application [5]. The Al toxicity limitation in the ST trial occurs below 10 cm while in the LT experiment, it exists in the 0-30 cm layer (Figures 5c and 6c). Because water flow from rainfall decreases with depth, the benefit of applying G to sites which have Al toxicity below 30 cm (ST experiment) should last longer compared to when Al toxicity exists in the 0-30 cm layer (LT experiment). In South Western Australia, for moderately acidic soils, Al toxicity in the 0-30 cm layer is widespread due to Tenosols' acidification process [59]. Indeed, for very acidic soils, longer-term acidification can result in Al toxicity appearing to a depth of 80 cm as illustrated by the soil profile presented by [60].
In both the ST and LT, G application increased the percentage of Al Act in the form of SO 4 -Al (Figures 5d and 6d). At the same time, there was no effect on pH Soln (Figure 2a,b) suggesting limited additional SO 4 -S sorption and release of OH − . There was no relationship between Al Act and Al CaCl 2 or Al Ex across all treatments (Figure 4c,e). In contrast, Al Act was related to pH Soln with the G treatments from the ST experiment having the highest concentration of Al Act (Figure 4d), but there was no impact of the G treatments on the relationship between Al Act and pH Soln of the ST experiment. Hence, G application increases Ca Soln and displaces Al 3+ and H + from the exchange sites [15,17]. However, there is no net effect on pH Soln because increased SO 4 -S Soln resulted in some SO 4 -S sorption and OH − displacement from the exchange sites [61].
In the LT, we observed the L treatments' effectiveness increased over time with grain yield response of 10-18% in the year of application rising to 20-36% in subsequent years [5]. Lime solubility is lower than G [11,12], resulting in no change in Al Act (Figure 2b) and pH Soln (Figure 2d) in the layers below 10 cm in the ST. In contrast, G application increases in SO 4 -S Soln (Figure 1a), Ca Soln (Figure 1c) and IS Soln (Figure 1e) to a depth of 50 cm in the first year after application. In the LT, the L4 treatment affected soil solution properties Al Act to a depth of 30 cm but pH Soln only in the 0-10 cm soil layer (Figure 2b,d). Similarly, L decreased Al CalCl2 to a depth of 30 cm, while pH CaCl 2 and Ca Ex only increased to a depth of 20 cm [8]. This observation is consistent with the previous publications which have shown the application of L sometimes has a small impact on the layers below 10 cm due to its low solubility and limited ability to be leached into the soil profile [1,8,11,62].
DOC Soln can form complexes with Al Soln that are not influenced by L application [63]. However, in both experiments, applied L increased DOC Soln (Figure 3b) because it increases organic matter solubility, microbial activity, production of soluble molecules and displacement of the previously adsorbed dissolved organic matter by other anions [64]. In agricultural soils, DOC Soln can range within 0 to 74 mg L −1 [64]. In soils with DOC Soln < 60 mg L −1 , such as those presented here (Figure 3), up to 80% of Al Soln can be bounded by DOC Soln for soil with pH < 5.0 [34,65]. In the ST experiment, which has a history of L application pH Soln was greater than 5.5 and Al Act was low with all of the Al Act calculated to be in the form of OH-Al (Figure 5b). In contrast, in the LT experiment, which has no L use history, pH Soln was 4.6, with only 45% of Al Act calculated to be associated with Org-Al in the 0-10 cm layer (Figure 6a). Hence, Toxic-Al was likely to be limiting wheat production, given the greater wheat yield in the first year where L had been applied [5]. The increase in Ca Soln can reduce DOC Soln due to increasing microbial consumption or adsorption by cation bridging [66]. However, at the LT, G application increased DOC Soln by 42% in the 0-10 cm layer, possibly due to SO 4 -S sorption desorbing DOC [66].

Conclusions
These soil solution measurements provide greater insight into the chemical reactions involved when lime and gypsum are applied to soils with an Al toxicity limitation for crop production than soil chemical measurements. In the short term, surface-applied gypsum increases soil solution sulfate and calcium to a depth of 50 cm despite low amounts of rainfall (279 mm). The increase in soil solution calcium resulted in the displacement of Al from the exchange sites leading to an increasing soil solution Al activity. However, there is no impact on Toxic-Al concentrations because this increase in Al activity is associated with Al sulfate formation, which is not harmful to root growth. The increase in soil solution sulfate, calcium and ionic strength together corrected the subsoil Al toxicity limitation.
In the long term, sulfate is leached from the soil profile resulting in a decline in the soil solution ionic strength and its effect on soil Al toxicity. Alkalinity from 4 t lime ha −1 leached into the soil profile reducing Al toxic forms to a depth of 30 cm. In the long term, gypsum gave no additional benefit over lime in terms of Toxic-Al but did maintain higher IS within the profile to 50 cm.