SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (23)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by de Wit, H. A.
Right arrow Articles by Mulder, J.
Right arrow Search for Related Content
PubMed
Right arrow Articles by de Wit, H. A.
Right arrow Articles by Mulder, J.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by de Wit, H. A.
Right arrow Articles by Mulder, J.
Soil Science Society of America Journal 63:1141-1148 (1999)
© 1999 Soil Science Society of America

DIVISION S-2-SOIL CHEMISTRY

Modeling Aluminum and Organic Matter Solubility in the Forest Floor Using WHAM

Helene A. de Wita, Marek Kotowskib and Jan Muldera

a Norwegian Forest Research Inst., 1432 Ås, Norway
b Dep. of Environmental Protection Engineering, Technical Univ. of Lublin, Lublin, Poland

heleen{at}nisk.no


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 Materials and methods
 Results and discussion
 Conclusions
 REFERENCES
 
Several studies suggest that solution concentrations of Al in organic surface soils are controlled by complexation with organic matter. We applied the mechanistic Windermere humic aqueous model (WHAM) to describe the solubility of Al and organic matter as observed in a batch equilibrium study with a forest floor Oe horizon. WHAM is unique in that it considers interactions of soil organic matter with protons and metals. We also compared WHAM with a previously proposed linear regression model that describes Al solubility. A range of soil Al contents was established by adding different amounts of Al in batch prior to titration with acid or base. The soil Al content was described by the bound Al ratio (BAR), defined as the equivalent ratio of organically bound Al and carboxyl groups. The bound Al ratio and pH ranged from 0.1 to 3, and from 1.7 to 6.3, respectively. Solutions were undersaturated with respect to Al(OH)3, except at BAR >= 2 and pH >= 4.5. Aluminum solubility increased with increasing BAR. Organic matter solubility was greatest at low BAR and high pH. WHAM reproduced the observed pH and Al concentrations, using parameters derived from experimental data. At BAR < 0.7, pH–pAl relationships were approximately linear. At BAR >= 0.7, there was a nonlinear increase in {Delta}pAl/{Delta}pH with pH. WHAM simulated the changing slope of the pH–pAl curves satisfactorily and reproduced observed trends in dissolved organic C (DOC) concentrations. This supports the hypotheses and assumptions concerning mechanisms for binding Al to soil organic matter in a forest floor, as embodied in WHAM.

Abbreviations: AAS, atomic absorption spectroscopy • calc, calculated • CFA, soil content of fulvic acids • CHA, soil content of humic acids • CHS, soil content of humic substances • DOC, dissolved organic carbon • FA, fulvic acids • HA, humic acids • HS, humic substances • I, ionic strength • meas, measured • optim, optimized • RMSD, root of the mean squared deviation • WHAM, Windermere humic aqueous model • {gamma}, parameter describing the distribution of CFA within 10 model-defined fractions of fulvic acids with specified hydrophobicities


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
Materials and methods
 Results and discussion
 Conclusions
 REFERENCES
 
MECHANISMS that control Al solubility in soils and surface waters are poorly understood. Most soil acidification models like ILWAS (Gherini et al., 1985), MAGIC (Cosby et al., 1985), and PROFILE (Warfvinge and Sverdrup, 1992) assume a simple equilibrium with gibbsite (Al(OH)3). However, several studies have shown that soil solutions in organic horizons and the upper mineral soil are undersaturated with respect to this mineral (David and Driscoll, 1984; Driscoll et al., 1985; Sullivan et al., 1986; Reuss et al., 1990; Mulder and Stein, 1994; Rustad and Cronan, 1995; Lawrence and David, 1997) with the largest undersaturation occurring in organic horizons. In addition, a cubic relationship between pH and pAl (the negative logarithm of the Al activity) as predicted by the gibbsite equilibrium is generally not observed. Evidence is accumulating that Al solubility in acid soils is controlled by complexation mechanisms with soil organic matter (Cronan et al., 1986; Mulder et al., 1989; Walker et al., 1990; Wesselink and Mulder, 1995; Berggren and Mulder, 1995).

The introduction of Al–organic matter interaction in a chemical equilibrium model could greatly improve the description of relationships between the concentrations of different cations in soil solutions. The recently developed Windermere humic aqueous model (WHAM) is a chemical equilibrium model considering specific and nonspecific ion binding by humic substances (HS) (Tipping, 1994). The Windermere humic aqueous model calculates the concentrations of cations and DOC in solutions in equilibrium with soils that contain HS, assuming that binding of cation to HS controls cation concentrations. Additionally, WHAM hypothesizes that DOC concentrations depend on the hydrophobicity of HS. Increased negative charge decreases hydrophobicity and, thus, results in increased DOC concentrations. The net charge on HS is calculated in consideration of the effect of ion binding. WHAM has been shown to be successful in the simultaneous description of pH and Al concentrations in organic soils in laboratory titration experiments (Tipping et al., 1995).

Previously, Cronan et al. (1986) and Walker et al. (1990) proposed an empirical linear regression model to describe pAl (the negative logarithm of Al3+ activity) in organic horizons as a function of pH and the Al content of the soil. A crucial parameter in this empirical model is the bound Al ratio (BAR). BAR is defined as the total equivalent of organically bound Al divided by the content of carboxyl groups in the soil: 3 x (moles of Al)/(moles of carboxyl groups). The central assumption in the model is that a linear relationship exists between pH and pAl for each BAR value (BAR range 0.1–1.0, pH range 3–5).

Our study was carried out with the following objectives: (i) to determine experimentally the solubility of Al and organic matter (DOC) in an Oe horizon of a Pinus sylvestris (L.) stand and to determine their dependency on the soil content of organically bound Al, expressed as BAR (the Al content of the Oe horizon was manipulated, using two experimental procedures to test the effect of sample treatment); (ii) to test WHAM by simulating the solubility of Al and DOC in a forest floor for a wide range of Al contents and pH values; and (iii) to compare WHAM with the previously mentioned linear regression model for describing Al solubility; i.e., experimentally obtained pH–pAl relationships.


   Materials and methods
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
Results and discussion
 Conclusions
 REFERENCES
 
Samples were taken from an Oe horizon of a sandy soil (Typic Udipsamment) in a Pinus sylvestris stand at Harderwijk in the Netherlands. The field moist soil was passed through a 0.5-mm sieve and homogenized. Half was stored at 4°C until later use in Experiment A, the remainder was freeze-dried for use in Experiment B. The field moisture content of the soil was 0.701 g g-1 and the ash content of the dried soil (after combustion at 650°C) was 0.189 g g-1.

We manipulated the Al content of the sieved Oe horizon following (i) the procedure described by Cronan et al. (1986) and Walker et al. (1990) (Experiment A), and (ii) a modification of the Cronan and Walker procedure to test the effects of sample pretreatment (Experiment B).

Experiment A: Procedure According to Cronan et al. (1986) and Walker et al. (1990)
Forty-gram subsamples were weighed into 1-L polycarbonate centrifuge bottles and 400 mL of 1 M HCl was added. After shaking on a reciprocal shaker for 5 h (100 strokes min-1), the suspension was centrifuged (40 min at 2750 g) and the supernatant was removed. The Al concentration in the supernatant was determined by atomic absorption spectroscopy (AAS). After repeating this procedure, the sample was washed five times with 100 mL of distilled, deionized water to remove the remaining HCl. The total amount of Al removed was 0.0227 mol kg-1. We assumed that the removed Al was part of the pool of organically bound Al and should be accounted for in Experiment B. Treated soil samples were considered to be H+ saturated, and they were freeze-dried and stored at 4°C. The content of carboxyl groups in the freeze-dried soil (104 cmolc kg-1) was determined by titration with 0.01 M NaOH to pH 7.00 under a N2 atmosphere. The range of the content of carboxyl groups in O horizons reported by Walker et al. (1990) was 79 to 449 cmolc kg-1. Soil samples with different BAR values were prepared by adding different amounts of Al(NO3)3, and subsequently, equivalent amounts of NaHCO3 to 200-mL suspensions containing 2.0 g of the H+-saturated soils. The mixtures were equilibrated while bubbling with N2 for 24 h under a N2 atmosphere, before excess salts were removed by multiple washing with distilled, deionized water. The suspensions were centrifuged (20 min at 13300 g), decanted, and the soil samples were freeze-dried. The BAR values obtained were 0.10, 0.35, 0.70, 1.00, 2.00, and 3.00.

Batch Titrations
For each BAR value in Experiment A, eight suspensions were made by adding 0.15 g of soil to 15 mL of a background electrolyte (0.005 M CaCl2) at room temperature under a N2 atmosphere. The background electrolyte used was the same as that used by Walker et al. (1990) and was added to keep the variation in ionic strength relatively small. The ionic strength of the solutions varied between 1.6 x 10-2 and 5 x 10-2. Titrations were done in batch mode by adding either 0.1 M HCl or 0.1 M NaOH to individual suspensions to obtain a pH range of 2.5 to 6.0. After 48-h equilibration on a reciprocal shaker (100 strokes min-1) the suspensions were centrifuged (20 min at 13300 g). After decantation and filtration (membrane filter, mixed [nitrate and acetate] cellulose esters, pore size 0.22 µm), the supernatants were analyzed for quickly reacting Al (Clarke et al., 1992), DOC, and pH. Quickly reacting Al, determined spectrophotometrically after a 2.3-s reaction with oxine (8-hydroxyquinoline) at pH 5, consists of all inorganic monomeric Al species except Al–F complexes. The detection limit for quickly reacting Al was 7 x 10-6 mol L-1. Total dissolved Al was determined spectrophotometrically after acidification of the samples using pyrocatechol violet (Velthorst, 1993). Total dissolved Al <7 x 10-6 mol L-1 was determined using graphite-furnace AAS. Dissolved organic C was determined using an Ionics Carbon Analyzer 555 (Thermo Instruments Systems, Breda, the Netherlands). The pH was determined using an Orion Research pH meter (model 701A) with an Orion combination pH electrode (type 91-56) (Fa. Ankersmit, Breda, the Netherlands).

Experiment B: Modification of Experimental Procedure Described by Cronan and Walker
Soil suspensions were made by adding 0.15 g freeze-dried soil to a 15-mL mixture of 0.005 M AlCl3 and 0.005 M CaCl2. The BAR value in Experiment B was calculated using the equivalent of organically bound Al present in suspension (Al added as AlCl3 + 0.0227 mol Al kg-1 soil, which was originally present in the soil, as determined in Experiment A), irrespective of Al being dissolved or sorbed. The suspensions were titrated with acid or base, and the ionic strength varied between 1.6 x 10-2 and 6 x 10-2. Ten suspensions were made for BAR 0.07 (the original soil sample, no AlCl3 added), and eight suspensions were made for all other BAR values. The BAR values obtained in Experiment B were slightly higher than those in Experiment A because of the presence of Al in the untreated soil sample and were 0.07, 0.17, 0.42, 0.77, 1.07, 2.07, and 3.07.

Computational Procedures
Inorganic speciation and the Al3+ activity were calculated using ALCHEMI (Schecher and Driscoll, 1987) correcting for temperature (22°C) and ionic strength. As input concentration for inorganic Al we used quickly reacting Al, unless concentrations were <7 x 10-6 mol L-1, in which case Al concentrations determined by graphite-furnace AAS were used for input data. We did not allow interaction between Al and DOC in the ALCHEMI calculations by setting DOC concentrations in the model input at zero. Titration curves were simulated using WHAM (Tipping, 1994).

WHAM is a chemical equilibrium model for soils, waters, and sediments and is based on a discrete site–electrostatic model of HS (Tipping, 1994). To predict the composition of the equilibrium soil solution, a model calibration is necessary. Model parameters that need to be optimized include the soil content of humic acids (CHA), the soil content of fulvic acids (CFA), and {gamma}, a distributed parameter. In the first optimization step, CHA and CFA are adjusted by minimizing the difference between observed and calculated pH and Al concentrations in an iterative, curve-fitting procedure. Using CHA and CFA, WHAM calculates cation relationships without considering organic matter solubility and subsequently computes the charges on HA and FA, which are functions of proton and metal sorption. WHAM assumes that HA is insoluble, and thus, DOC consists solely of FA. The solubility of FA is determined by its charge, which has been calculated previously, and its hydrophobicity, which is indicated by {gamma}. In the second optimization step, {gamma} is optimized using observed DOC concentrations. Using CHA, CFA, and {gamma}, the composition of the soil solution is calculated.

Essential in WHAM is submodel V (Tipping and Hurley, 1992; Tipping, 1993a and 1993b) developed for description of the competitive complexation and nonspecific binding of ions. Equilibrium constants for the binding of protons and metals to humic compounds used in WHAM have been derived from the literature and are reported in Tipping (1994). The proton and metal binding groups of HS are heterogeneous, having a range of pK values. For the sake of simplicity, the groups are distinguished into two types, A and B, each covering a different range of pK values. The median pK values are summarized in Table 1 (Tipping, 1994).


View this table:
[in this window]
[in a new window]
  		Table 1 Median equilibrium constants for binding of protons (pKA, pKB) and metals (pKMHA, pKMHB) to humic compounds used in the Windermere humic aqueous model (WHAM).{dagger}

 
Calibration of CHA and CFA is done by minimizing the root of the mean squared deviation for pH and Al, defined as

(1)


Note that the difference between [Alaq]meas, i and [Alaq]calc, i is divided by the mean observed Al concentration in order to give similar magnitudes to RMSD-pH and RMSD-Al. This implies that low Al concentrations carry relatively little weight in the determination of CHA and CFA, as they contribute little to RMSD-Al.

The parameter {gamma} describes the distribution of the only soluble humic compound in the model, FA, for 10 model-defined fractions with specified hydrophobicity characteristics. The distribution of the fractions of FA is expressed as follows:

where CFA is the soil content of FA, and i the number of the fraction (1 <= i <= 10). The most hydrophobic fraction of FA is assumed to adsorb appreciably, even though highly charged, while the most hydrophilic fraction is assumed to readily dissolve, even when carrying relatively little charge. The larger the value of {gamma}, the larger is the proportion of hydrophobic fractions. If {gamma} is 0, all fractions are of equal size. Thus, {gamma} > 0 implies that FA has relatively hydrophobic characteristics. Calibration of {gamma} involves minimization of RMSD-DOC:

(2)

For more details concerning the calibration of WHAM, reference is made to Tipping et al. (1995).


   Results and discussion
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
 Results and discussion
Conclusions
 REFERENCES
 
Batch Titrations
Figure 1 shows the increase in pH for all BAR curves upon addition of base. To reach a given pH, less base (or more acid) was added in Experiment A than in Experiment B. At zero acid or base addition (i.e., with a background electrolyte only), pH observations in Experiment A ranged from about 2.7 to 4.3, while pH observations in Experiment B clustered around 3. The titrations resulted in a reverse order of the BAR curves for Experiment A and B. For a given acid or base addition, the resulting solution pH increased in the order BAR 0.1 to BAR 3.0 in Experiment A, whereas this pH increase was in the order BAR 3.07 to BAR 0.07 in Experiment B.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 1 Titration curves showing the effect of base–acid additions on pH for different bound Al ratio (BAR) values. Lines show WHAM simulations

 
Figure 2 shows the increase in Al concentrations for all BAR curves upon addition of acid. At zero base addition, Al concentrations were close to zero in Experiment A, whereas Al concentrations in Experiment B ranged from zero to {approx}8 x 10-3 M. At equal acid additions and equal BAR values, more Al was dissolved in Experiment B than in Experiment A.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 2 Titration curves showing the effect of base–acid additions on concentrations of quickly reacting Al for different bound Al ratio (BAR) values. Quickly reacting Al includes all monomeric Al species except Al–F complexes. Lines show the WHAM simulations of inorganic Al species. Note the different scale of the y-axes

 
The apparent large differences in the titration curves of Experiments A and B are probably due to the different sample pretreatments in the experiments. In Experiment B, the BAR was established by addition of AlCl3 to the suspensions during titration; i.e., together with the addition of base or acid. Complexation of Al caused deprotonation of carboxylic groups on the organic material and, thus, a decrease in pH of the batch solution. Establishment of a higher BAR at a given base addition, by adding more Al, caused an enhanced proton liberation and, thus, a lower pH. By contrast, the samples used in Experiment A obtained their BAR values during pretreatment. All protons liberated through complexation of Al were removed by washing with distilled water prior to titration. Consequently, equal amounts of base addition and similar ionic strength at a given BAR led to a higher pH in Experiment A, than in Experiment B, especially at high BAR values. As Al solubility decreases with increasing pH, less Al is mobilized for a given BAR in Experiment A than in B.

Aluminum Solubility
Despite the differences between Experiment A and B shown in Fig. 1 and 2, the Al solubility curves for similar BAR values in Experiment A and B, calculated using ALCHEMI, are quite similar (Fig. 3) . This suggests that the acid washing of the samples in Experiment A, following Cronan et al. (1986), did not alter Al solubility curves for Experiment A and B significantly. As expected, the highest Al3+ activity (lowest pAl) was found at the highest BAR values and at low pH. With increasing pH, the observations may be described best by equilibrium with an Al(OH)3 phase . At increasing BAR values, the pH–pAl relationship becomes increasingly curvilinear (Fig. 3). Particularly for BAR >= 1, {Delta}pAl/{Delta}pH approaches the slope of the gibbsite solubility line at pH > 4.5, indicating an increasing pH dependency of Al3+ activity with increasing pH. Figure 3 confirms that the gibbsite equilibrium is a poor predictor for the Al activity in solution. Only at high soil Al content and at pH > 4.5 may equilibrium with Al(OH)3 be used to model Al3+ solubility.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 3 Aluminum solubility as a function of pH and bound Al ratio (BAR) compared with solubility of amorphous gibbsite. Symbols refer to measured pH values and to Al3+ activities calculated by ALCHEMI. Lines show WHAM simulations of pH and Al3+ activity

 
Organic Matter Solubility
In Fig. 4 , the largest DOC concentrations are observed at high pH and low soil Al content. The concentration of DOC increases rapidly with pH at BAR 0.1, whereas DOC mobilization is less pH dependent at higher BAR values. At BAR 3, the concentration of DOC decreases slightly when pH increases. Even at low pH and high BAR, when a relatively large portion of the reactive groups are most likely either protonated or complexed with Al, some DOC that is probably very hydrophilic remains present in solution.



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 4 Dissolved organic C (DOC) solubility as a function of pH and bound Al ratio (BAR) in Experiment A for a selection of BAR values. Insert shows DOC calculated by WHAM and observed pH. Note the change in scale of y-axis

 
Tipping and Woof (1990) argue for a major influence of charge on the release of humic compounds to the soil solution. At BAR 0.1, a decrease in pH will lead to protonation of carboxylic groups on HS, causing a decrease in net negative charge of HS. Thus, the solubility of HS is expected to decrease. At BAR 3 and high pH, most carboxylic groups will be occupied with Al and the net negative charge of HS can be expected to be relatively small. Here, a decrease in pH is associated with release of Al3+ (Fig. 3), which may lead to an increase in net negative charge of HS and to an increased solubility of organic matter.

Calibration of WHAM
We optimized CHA and CFA using all observations from Experiments A and B (Table 2) , excluding the data for BAR 3 and BAR 3.07 and observations with pH > 5.7 (the upper data points in the pH range for BAR 0.35 and 0.7 in Experiment A, and for BAR 0.07, 0.17, 0.42, and 0.77 for Experiment B). At pH > 5.7, DOC concentrations were relatively high and concentrations of quickly reacting Al were below the detection limit. We assumed that most Al as measured by AAS was organically complexed and that these values are an overestimation of the free Al present. We excluded the BAR 3 and 3.07 observations because many of the data points suggest equilibrium with Al(OH)3.


View this table:
[in this window]
[in a new window]
  		Table 2 Optimized values for the model parameters of soil content of humic acids (CHA) and soil content of fulvic acids (CFA) and their sum, the soil content of humic substances (CHS), in WHAM, including the error in simulation of pH (RMSD-pH) and Al concentrations (RMSD-Al).{dagger}

 
Optimized values for CHA were considerably larger than values for CFA (Table 2) for the dataset that included Experiments A and B and when Experiments A and B were considered separately. However, the optimized values for CHS were nearly the same in Experiment A and B. Optimization of subsets of the observations (results not shown) showed that CHA and CFA may vary considerably, even though their sum is fairly constant, indicating that one should be careful not to overinterpret these results. The optimized soil content of humic substances (CHSoptim) comprised {approx}0.25 g g-1 dry wt. of the Oe horizon. This is within the range of optimized CHS values reported for other O horizons by Tipping and Woof (1990); i.e., 0.04 to 0.48 g g-1 soil. Tipping et al. (1995) showed that CHSoptim was in reasonable agreement with HS extracted by 0.01 M NaOH (CHSmeas). In their study, the ratio CHSoptim/CHSmeas varied between 0.29 and 1.31, with a mean of 0.61.

We chose optimized values of CHA and CFA based on the lumped data set from Experiment A and B to calculate the composition of individual equilibrium solutions of both experiments. We considered that it was not reasonable to assume that CHA and CFA were different in Experiment A and B, as the same soil was used in both experiments. However, separate optimizations for Experiment A and B resulted in a considerably lower RMSD (pH + Al) for Experiment A than for the lumped data set (Table 2).

The Windermere humic aqueous model was calibrated to describe DOC concentrations by optimization of {gamma} . For organic surface soils Tipping et al. (1995) report {gamma} values ranging from 1.5 to 2.0 and RMSD-DOC ranging from 2.6 to 10.4 mg L-1.

WHAM Simulations
Figures 1 and 2 show that the calibrated WHAM model described the titration curves of both experiments well; the shape of the simulated curves closely follows the observations. The largest deviations between observations and simulations occurred at high BAR values and high acid additions in Experiment A, in which WHAM overestimated inorganic Al concentrations relative to the measured concentrations of quickly reacting Al.

The WHAM simulations of pH–pAl relationships in Fig. 3 were reasonably successful, especially considering that in contrast to Fig. 1 and 2, variables on both axes were simulated. However, at pH > 4 and BAR <= 0.42 there was a considerable discrepancy between model predictions and observations. The Al3+ activity calculated by WHAM was 5 to 1000 times lower than observed Al3+ activity (calculated by ALCHEMI); however, it should be noted that observed Al activities in this pH and BAR range are based on total Al concentrations as determined by graphite-furnace AAS. We assume that the major portion of total Al was organically complexed, because the DOC concentrations were relatively high (Fig. 4), and thus, the graphite-furnace AAS measurements represent an overestimation of inorganic Al in solution.

Concerning the DOC concentrations, the shape and the relative position of the curves for the separate BAR values were reproduced well, but the simulated concentrations deviated considerably from observed values (Fig. 4). At high pH and low BAR, the simulated DOC concentrations were about twice as high as the observed DOC concentrations. The minimum observed DOC concentration was 4 to 7 mg L-1 for all BAR values. By contrast, the minimum DOC concentrations predicted by WHAM decreased to zero. This implies that even the FA in the most hydrophilic of the 10 model-defined fractions of FA were carrying too little charge to overcome their model-assigned hydrophobicity and were adsorbed to the solid phase.

The WHAM simulation of DOC concentrations could be improved by assuming that a given minimum concentration of FA was dissolved independent of pH or BAR. When introducing a background concentration of 4 mg L-1 DOC in the model and repeating the optimization, we obtained and . The form of the curves remained essentially the same, and DOC concentrations at high pH and low BAR were still considerably overestimated.

Model Comparison
Cronan et al. (1986) and Walker et al. (1990) proposed an empirical linear regression model to describe pAl in organic horizons as a function of measured pH and BAR. We applied linear regression analysis to the pH–pAl relations obtained in Experiment A. The slopes ({Delta}pAl/{Delta}pH) of the regression lines increase with BAR, indicating an increasing pH-dependency of the Al activity (Table 3) . However, r2 decreases for increasing BAR, indicating that the linearity of pH–pAl relationship declines; this situation is also visible in Fig. 3. Thus, the use of linear regression analysis to describe the data at BAR >= 0.7 is questionable. By contrast, Cronan et al. (1986) and Walker et al. (1990) found a linear pH–pAl relationship for all BAR values considered; i.e., from BAR 0.1 to 1, in a pH range from 3 to 5. The pH range and BAR range covered by our experiments are wider than those reported by Walker et al. (1990) (Table 3). For BAR >= 0.7 and pH < 4, {Delta}pAl/{Delta}pH decreases with increasing BAR. This is in contrast with Walker et al. (1990) who reported a steeper slope of the pH–pAl curve with increasing BAR.


View this table:
[in this window]
[in a new window]
  		Table 3 Slope and r2 for the linear regression of pH against pAl for individual bound Al ratio (BAR) values in Experiment A. Mean slopes of regression lines for Oi and Oa horizons from seven locations in the USA are given for comparison

 
To quantify and compare the relative success of the model predictions, we considered deviations between model predictions and the data on logarithmic and linear scales. The linear regression model was calibrated using pAl with pH as an independent input parameter; thus, RMSD-pH was zero by definition. By contrast, calibration of WHAM involved minimization of a function that contains both pH and Al concentrations. For BAR 0.1, the data were described well by the linear model (Fig. 5) , indicated by a low RMSD-pAl (Table 4) . By contrast, WHAM predictions considerably overestimated pAl at higher pH; however, the difference between the RMSD-Al values (1 x 10-6 mol L-1 for WHAM vs. 5 x 10-6 mol L-1 for linear regression analysis) was insignificant compared with the average Al concentrations ({approx}1.65 x 10-4 mol L-1), and both models performed equally well.



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 5 Model descriptions (linear regression analysis and WHAM) simulations of Al solubility for bound Al ratio (BAR) 0.1 and 1. Symbols refer to measured pH values and to Al3+ activities calculated by ALCHEMI

 

View this table:
[in this window]
[in a new window]
  		Table 4 Root of mean squared deviation (RMSD) values for deviations between model-predicted and observed values for pH, pAl, and inorganic Al concentrations

 
For BAR > 0.1, linear regression explained more of the variation in pAl than WHAM did (except at BAR 1). On the other hand, the variation in Al concentrations was explained better by WHAM. For calculation of Al concentrations at low BAR (<= 0.35) both models were adequate. At higher BAR values, RMSD-Al increased for both models but WHAM performed best. Obviously, the model performance is closely linked to the model calibration procedure.

An advantage of linear regression is its ease of application, yet this modeling approach is totally empirical, and extrapolation to varying soil Al contents for the same soil is questionable. Each BAR value requires a new titration curve for calibrating the model. By contrast, the mechanistic approach in WHAM makes it possible to extrapolate to other soil Al contents. With one set of optimized values for CFA and CHA, WHAM was able to describe pH and Al concentrations in the BAR range 0 to 2 and in the pH range 2.5 to 6. WHAM successfully extrapolated to BAR 3 (Fig. 1 and 2), although the BAR 3 observations had not been used for calibration of the model. Nevertheless, use of WHAM requires a thorough calibration procedure that might be a barrier for the use of the model.

Extrapolation to other soils, using either WHAM or linear regression analysis, will probably be only partly successful in describing Al solubility in the forest floor. Both models need calibration with Al solubility data obtained with laboratory titration experiments that are similar to this study.


   Conclusions
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
 Results and discussion
 Conclusions
REFERENCES
 
The results from this study support the hypothesis that Al solubility in acid soils of the forest floor is controlled by complexation with soil organic matter. The data were obtained by two different procedures for sample pretreatment. No effect of sample treatment on Al solubility curves was found. The solubility of Al increased with an increasing soil content of organically bound Al, expressed as bound Al ratio (BAR). The solutions were undersaturated with respect to Al(OH)3, except at BAR >= 2 and pH >= 4.5. Organic matter solubility was largest at low BAR values and high pH. The pH dependency of organic matter solubility declined with increasing BAR.

Modeling results from WHAM closely followed the observed soil titration curves of pH and Al concentrations. pH–pAl relationships, which are calculated simultaneously by WHAM, were described adequately. However, pAl was systematically overestimated (Al3+ activities underestimated) at pH > 4 and BAR <= 0.42. This may be due to lack of adequate and analytical estimates of inorganic Al in solution that are necessary for a proper calibration of the model.

The success of WHAM in describing trends in DOC solubility for a wide range of experimental conditions suggests that the solubility of DOC is strongly influenced by the net charges associated with the humic compounds. On the other hand, the model did not allow for an observed minimum amount of DOC that is present irrespective of the calculated negative charge. Both WHAM and the empirical linear regression model proposed by Cronan et al. (1986) and Walker et al. (1990) successfully described Al solubility at BAR <= 0.35. The success of the model performance was linked to the calibration procedure of the model. At BAR >= 0.7, the pH–pAl relationship became curvilinear, which limited the use of linear regression analysis to model Al solubility in the O horizon.


   ACKNOWLEDGMENTS
 
The experiments were carried out at the Dep. of Soil Science and Geology at Wageningen Agricultural University in the Netherlands with financial support from the Additional Program on Acidification. We thank Eef Velthorst and Neel Nakken for help with the experimental work. Ed Tipping is gratefully acknowledged for helpful discussions. Marek Kotowski was supported by a NUFFIC grant. Helene A. de Wit is supported by a grant from the Norwegian Research Council (109475/720). Jan Mulder was supported by a postdoctoral fellowship from the Royal Netherlands Academy of Arts and Sciences (KNAW).

Received for publication January 28, 1998.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 Materials and methods
 Results and discussion
 Conclusions
 REFERENCES
 




This article has been cited by other articles:


Home page
 Soil Sci.Home page
P. Gruba and J. Mulder
Relationship between Aluminum in Soils and Soil Water in Mineral Horizons of a Range of Acid Forest Soils
Soil Sci. Soc. Am. J., June 18, 2008; 72(4): 1150 - 1157.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
H. A. de Wit, T. Groseth, and J. Mulder
Predicting Aluminum and Soil Organic Matter Solubility Using the Mechanistic Equilibrium Model WHAM
Soil Sci. Soc. Am. J., July 1, 2001; 65(4): 1089 - 1100.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (23)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by de Wit, H. A.
Right arrow Articles by Mulder, J.
Right arrow Search for Related Content
PubMed
Right arrow Articles by de Wit, H. A.
Right arrow Articles by Mulder, J.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by de Wit, H. A.
Right arrow Articles by Mulder, J.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Vadose Zone Journal Journal of Plant Registrations
Journal of Natural Resources
and Life Sciences Education		 Journal of
Environmental Quality