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 (11)
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.
Related Collections
Right arrow Soil Models
Right arrow Soil Analysis
Right arrow Soil Chemistry
Soil Science Society of America Journal 65:1089-1100 (2001)
© 2001 Soil Science Society of America

DIVISION S-2—SOIL CHEMISTRY

Predicting Aluminum and Soil Organic Matter Solubility Using the Mechanistic Equilibrium Model WHAM

Helene A. de Wit*,a, Tore Grosethb and Jan Mulderc

a Norwegian Forest Research Institute, Hoegskoleveien 12, N-1432 Aas, Norway
b Dep. of Chemistry, Univ. of Oslo, P.O. Box 1033 - Blindern, N-0315 Oslo, Norway
c Dep. of Soil and Water Sciences, Agric. Univ. Norway, Box 5028, N-1432 Aas, Norway

* Corresponding author (heleen{at}nisk.no)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION AND CONCLUSIONS
 REFERENCES
 
The mechanistic equilibrium model WHAM is used to describe solution–solid phase interactions in the forest floor with regard to Al and soil organic matter (SOM) solubility. WHAM takes into account specific and nonspecific ion-binding to humic compounds. Experimental data from a forest soil that was manipulated with respect to its Al content were obtained from batch studies and a field manipulation experiment. WHAM was parameterized using observations of pH, concentrations of inorganic Al, and DOC obtained in batch. Model fits of pH, concentrations of inorganic Al (>5 x 10-6 mol L-1), and DOC were good to tolerable (1, 9, and 15% deviation, respectively). Values of optimized model parameters agreed reasonably with analytically determined quantities. Using the optimized parameters, WHAM simulated addition of AlCl3 to the same soil. Comparison between model predictions and batch observations showed a deviation for pH, Al (>5 x 10-6 mol L-1), and DOC of 3, 60, and 15%, respectively. We regard this as a reasonable model performance and support for the assumption of an organic complexation control of Al solubility in organic soils. Application of WHAM to predict effects of AlCl3 addition in the field resulted in qualitative agreement between simulations and observations from tension lysimeters in the forest floor, but in a failure regarding the observed ranges of H, Al, and DOC. The discrepancy between model simulations and field observations may be explained qualitatively by a lack of equilibrium due to the diffusion-limited exchange of solutes between immobile water in micropores and mobile water in macropores.

Abbreviations: Al-qr, quickly reacting aluminum • BAR, bound Al ratio • CAL, soil content of reactive Al • CEC, cation-exchange capacity • CFA, soil contents of FA • CHA, soil contents of HA • CHS, soil content of HS • DOC, dissolved organic carbon • FA, fulvic acids • FA0, minimum concentration of dissolved FA • HA, humic acids • HS, humic substances • I, ionic strength • ICP, inductively coupled plasma optical emission spectroscopy • OM, organic matter • RDM, relative mean deviation • RMSD, root of mean squared deviation • SOM, soil organic matter • SSR, soil-to-solution ratio • TOC, total organic C • WHAM, Windermere Humic Aqueous Model • {gamma}, parameter describing the distribution of FA over 10 model fractions with varying degrees of hydrophobicity


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION AND CONCLUSIONS
 REFERENCES
 
DESCRIBING AND UNDERSTANDING Al solubility in forest soils has received much attention since the end of the 1970s when increased concentrations of Al were believed to deteriorate forest vitality. Nevertheless, controls of Al solubility are still poorly understood. In general terms, acid deposition on acid soils results in increased concentrations of Al in soil solutions. The acidification models MAGIC (Cosby et al., 1985), ILWAS (Gherini et al., 1985), and PROFILE (Warfvinge and Sverdrup, 1992) assume that Al solubility is controlled by equilibrium with an inorganic Al mineral, gibbsite. However, soil solutions collected in the field were often found to be undersaturated with respect to gibbsite, especially in the upper mineral soil horizons and in the forest floor (David and Driscoll, 1984; Reuss et al., 1990). Mechanisms that were proposed to explain the discrepancy between field observation and model predictions based on the gibbsite equilibrium include cation exchange (Reuss et al., 1990), chemical equilibrium with other mineral phases than gibbsite (Matzner and Prenzel, 1992), kinetic limitations of gibbsite dissolution (Alveteg et al., 1995), organic complexation (Wesselink et al., 1996), and diffusion-limited release of Al into rapidly percolating solution (Franken et al., 1995).

To date, strong evidence exists that Al solubility in organic horizons in acid forest soils is controlled by complexation with organic matter (Walker et al., 1990; Mulder and Stein, 1994; Berggren and Mulder, 1995). Walker et al. (1990) and de Wit et al. (1999) showed that Al solubility in organic soil horizons depends on the degree of Al saturation of soil organic matter (SOM). The mechanistic soil equilibrium model WHAM (Windermere Humic Aqueous Model), which incorporates proton- and metal-binding to humic compounds (Tipping, 1994), was developed to calculate equilibrium chemistry in water, sediments, and soils. So far, WHAM has been tested largely on data obtained from equilibrium studies in the laboratory and was shown to successfully describe solubility of Al in batch studies using both mineral and organic soils (Tipping et al., 1995; de Wit et al., 1999; Lofts et al., 2001). WHAM also described the solubility of SOM in a forest floor reasonably well over a range of pH and soil Al contents (de Wit et al., 1999).

The objectives of this study were (i) to use the mechanistic equilibrium model WHAM to improve the current quantitative understanding of solution–solid phase interactions regarding the solubility of Al and SOM in organic soils through batch experiments and (ii) to test whether WHAM can be extended from describing to predicting forest-floor chemistry (Al, DOC [dissolved organic carbon], pH) in batch and in a field manipulation experiment.

Two batch experiments were conducted using forest-floor soil from a field manipulation experiment in which tree responses to enhanced concentrations of Al in the soil solution were studied (de Wit et al., 2001). One batch study was used to parameterize the model WHAM, similar to previously reported tests of WHAM (e.g., de Wit et al., 1999). Once parameterized, the model was used to predict the effects of the addition of AlCl3 to forest-floor chemistry in batch and in the field.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION AND CONCLUSIONS
 REFERENCES
 
Location
A field manipulation experiment in a 45-yr-old Norway spruce [Picea abies (L.) H. Karst.] forest in Norway was set up in 1996 to test the Al toxicity hypothesis proposed by Ulrich (1983)(1984). According to the Al toxicity hypothesis, enhanced concentrations of Al in the soil solution limit root growth and nutrient uptake in trees and may result in a decrease in forest growth and vitality. The experiment was designed to establish potentially toxic Al concentrations in the soil solution by a weekly addition of various concentrations of AlCl3 in deionized water to 12 forest plots by means of an irrigation system (de Wit et al., 2001). Addition of AlCl3 was chosen, instead of addition of H2SO4 (which has been used in other field-manipulation experiments with a focus on the effects of acid deposition of trees) because target levels of Al in the soil solution were controlled better using AlCl3. Additionally, elevated concentrations of Al would be obtained throughout the soil profile including the forest floor. Geological deposits at the experimental location consist of fluvioglacial sandy sediments of {approx}60-m thickness. The homogeneous, sandy soil is classified as a Typic Udipsamment and has a thin (2–4 cm) O horizon consisting of Oi and Oe material. Twelve adjacent plots of 20 by 20 m2 were established of which nine plots were assigned to Al addition treatments and three plots to a reference treatment. The Al addition plots all received an initial high-dose AlCl3 addition (12 g Al m-2) to raise the soil Al content. Then various concentrations of AlCl3 were added weekly during the growing season in 10 mm of deionized water. Three of nine plots that had raised soil Al contents received no AlCl3 (A-0 Treatment) in the weekly irrigations and were a control of the initial Al addition. The remaining six plots received dilute concentrations of AlCl3 so that potentially toxic Al concentrations were established in the root zone. For the three plots assigned to Treatment A-1, the target Al concentration was 0.1 x 10-3 mol L-1. For Treatment A-2, the target Al concentration was 0.2 x 10-3 mol L-1. The Al concentrations that were added depended on expected dilution due to variations in precipitation and soil moisture content. The Al concentrations added to the plots in Treatment A-1 varied from 0.18 x 10-3 to 0.37 x 10-3 mol L-1. The Al concentrations added in Treatment A-2 varied from 0.37 x 10-3 to 0.74 x 10-3 mol L-1. The reference treatment consisted of a weekly addition of 10 mm of deionized water.

Soil Sampling and Analysis
The forest floor was sampled in September 1997, after the initial Al addition. Field-moist samples were stored at 4°C prior to sample treatment. Moss layer and needles were removed from each sample. The samples were air-dried for 1 wk at 28°C, sieved (5.4-mm diam. sieve), and homogenized. Next, subsamples from the reference plots (Treatment C) were pooled to one sample that will be referred to as Soil C, and subsamples from the Al-addition plots were pooled to another sample, Soil A. Cation-exchange capacity (CEC) was determined as the sum of cations extracted by 1 M NH4NO3 (base cations, Mn2+, and exchangeable acidity). Base cations, Al, and Mn were determined in the extract by using inductively coupled plasma optical emission spectroscopy (ICP). Exchangeable acidity was determined by titration of the extract to pH 7 (Ogner et al., 1999). Organically bound Al was determined by a 0.5 M CuCl2 extraction (Juo and Kamprath, 1979) and by a 0.1 M Na4P2O7 (pyrophosphate) extraction (van Reeuwijk, 1992; Porebska and Mulder, 1996). All extractions were done in triplicate. Al concentrations in the extracts were determined using FAAS PerkinElmer 5000 (Perkin Elmer Instruments, Norwalk, CT). The soil content of carboxyl groups was determined by titration with 0.01 M NaOH to pH 7.00 under N2 atmosphere.

Soil Solution Sampling and Analysis
Soil solution from the forest floor was sampled using zero-tension lysimeters and tension lysimeters. At each plot, two zero-tension lysimeters, each with an area of 0.090 m2, were installed under the forest floor. Soil solution drained freely in 2-L glass bottles. Eight rhizon tension lysimeters were installed per plot. Rhizon lysimeters (Eijkelkamp, Agrisearch Equipment, The Netherlands) are tubes (10 cm long, 2.5-mm diam.) that consist of porous polymer reinforced by a stainless-steel wire. Four lysimeters were connected to one evacuated (50 kPa) 250-mL bottle. The tension lysimeters were inserted in the Oe horizon, right above the mineral soil. Soil-solution samples were stored at 4°C prior to analysis. After filtration (pore size 0.45 µm, nitrocellulose filters, Millipore, Bedford, MA), soil-solution samples were analyzed for pH (Radiometer, Copenhagen, Denmark), total concentration of elements (Al, Ca, K, Mg, Na, Si) by ICP (Ogner et al., 1999), anions (nitrate, sulfate, phosphate) by IC (ion chromatography) (Ogner et al., 1999), quickly reacting Al (Clarke et al., 1992), and DOC (high-temperature catalytic oxidation; Dohrmann DC-190, Rosemount Analytical, Santa Clara, CA). Quickly reacting Al (Al-qr) is determined spectophotometrically using kinetic discrimination in a 2.3-s reaction with oxine at pH 5 and consists of monomeric inorganic Al species except F complexes (Clarke et al., 1992). The analysis had a detection limit of 0.8 x 10-6 mol L-1 Al-qr.

Batch-1: Titration with Acid or Base
Eight suspensions, replicated three times, were made for Soil A and for Soil C by adding 0.5 g of soil to 50 mL of a background electrolyte (0.5 x 10-3 mol L-1 KCl) at room temperature. The ionic strength of the background electrolyte was similar to the average ionic strength in the soil solution at the reference sites (as collected with zero-tension lysimeters) and was added to keep the variation in ionic strength, due to addition of acid or base, relatively small. Titrations were done in batch mode by adding either 0.01 M NaOH or 0.1 M HCl to individual suspensions to obtain a pH range of 2.5 to 5.0. The ionic strength varied from 5.0 x 10-4 (only background electrolyte) to 6.54 x 10-4 (the highest acid addition). After 48-h equilibration on a reciprocal shaker (165 strokes per min), the suspensions were centrifuged (15 min at 2800 x g). After decantation and filtration (pore size 0.45 µm, nitrocellulose filters, Millipore, Bedford, MA), the supernatants were analyzed following the same analytical procedure as described for the soil solution samples.

Batch-2: Addition of AlCl3
Seven suspensions, replicated three times, were made for Soils A and C by adding 0.5 g of soil to 50 mL of a mixture of background electrolyte (0.5 x 10-3 mol L-1 KCl) and 37 x 10-3 mol L-1 AlCl3. Al was added to obtain a range of initial Al concentrations varying from 0 to 0.93 x 10-3 mol L-1 Al, thus mimicking Al concentration in the irrigation water added in the field. Ionic strength (I) ranged from 5.4 x 10-4 (only background electrolyte) to 3.6 x 10-3. The suspensions were shaken, centrifuged, filtered, and analyzed according to the procedures described above.

Computational Procedures: Modeling Using WHAM
The equilibrium composition of the soil suspensions in batch titrations was described using WHAM (Tipping, 1994). The mechanistic model WHAM is a chemical equilibrium model for soils, waters, and sediments that includes specific and nonspecific ion-binding by humic substances. WHAM consists of submodels that calculate ion-binding to humic substances (HS) (model V), HS solubility, inorganic solution chemistry, Al-hydroxide and Fe-hydroxide precipitation, and clay cation exchange (not used here).

Model V, the central component of WHAM, assumes that HS consist of humic acids (HA) and fulvic acids (FA). Important characteristics of HA and FA are presented in Table 1. Humic acids and FA are hypothetical spherical molecules that carry proton-dissociating groups that bind metal ions by specific binding. Humic acids have a larger molecular weight and radius than FA, and a lower density of reactive groups (mol g-1). The reactive sites are distinguished in Type A sites (strong acids) and Type B sites (weak acids). Type A and Type B sites have different median pK values. Metal–humus interactions are described in terms of intrinsic equilibrium constants (Kintr) and electrostatic terms (Kelec) that take into account the effect of variable humic charge on binding:

(1)
where z is the charge of the metal (or proton), Z is the humic charge, and w is an electrostatic interaction factor depending on ionic strength. Thus the strength of the metal–humus bond depends on the net charge of the humic compound.


View this table:
[in this window]
[in a new window]
  		Table 1. Characteristics of model parameters humic acids (HA) and fulvic acids (FA) (default parameters reported in Tipping, 1994).

 
Nonspecific binding by accumulation of counter-ions in the diffuse double layer at the charged surface of HA and FA contributes to total binding of cations. Counter-ion accumulation is variable because of variability of humic charge. Counter-ion accumulation occurs in a layer of defined thickness around the molecules, and the diffuse layer volume is restricted to 25% of the total solution volume. Selectivity coefficients are used to calculate the contribution of major cations to counter-ion accumulation.

The WHAM describes the solubility of organic matter as the sorption/desorption of FA by soil solids. By definition, HA are not soluble. It is assumed that organic matter (OM) solubility is a function of charge and hydrophobicity of FA. In WHAM, FA are constructed as the sum of 10 fractions that differ only in their degree of hydrophobicity but that have the same proton- and metal-binding characteristics. The distribution of FA over the 10 fractions is expressed in terms of distribution parameter {gamma}:

(2)
where CFA is soil content of FA, and 1 <= i <= 10. At {gamma} = 1, all fractions are equal in magnitude. An increase in {gamma} implies an increase in the proportion of the hydrophobic fractions. Whether a given fraction FAi is dissolved or sorbed depends on the balance between hydrophobicity characteristics of the fraction and net charge of FA. For a given a net charge of FA, the most hydrophobic fraction of FA will tend to be adsorbed, whereas the least hydrophobic fraction of FA will tend to be dissolved. Theoretically, all fractions of FA may be adsorbed at a very low net charge of FA, thus resulting in a zero concentration of DOC. An increase in charge of FA will be associated with FA desorption.

Chemical equilibrium constants used in WHAM are derived from literature and are reported in Tipping (1994). Soil-specific input data for WHAM are the soil-to-solution ratio (SSR) (g L-1), the soil content of reactive Al (CAL) and the content of other relevant cations (mol g-1), the content of reactive humic substances (CHS) (g g-1), concentrations of dissolved anions (mol L-1), and the distribution parameter {gamma} (only for calculation of FA sorption). In our study, WHAM was parameterized by optimization of CHS, CAL, and {gamma}. The CHS term was defined as 0.84 x CHA + 0.16 x CFA (where CHA is the soil content of HA), as proposed by Tipping et al. (1995). A parameter FA0 was introduced to account for DOC that was dissolved independent of pH and Al content, and was determined by calculating the mean of the six lowest of all DOC concentrations measured in Batch-1.

Optimized values for CHS and CAL were obtained by a curve-fitting procedure that minimized the sum of the differences, F, between measured and calculated pH (pHmeas and pHcalc, respectively) and measured and calculated Al-qr (Al-qrmeas and Al-qrcalc, respectively) for n observations in Batch-1, according to F = FpH + FAl:

(3)

Note that FAl is defined by dividing the difference between measured and calculated Al-qr by the mean observed concentration of Al, in order to give FpH and FAl a similar magnitude. An optimum value for {gamma} was obtained by minimizing RMSD (root of mean squared deviation)-DOC (mg L-1):

(4)

Computational Procedures: Inorganic Speciation Using WHAM
Speciation of inorganic Al was done using the inorganic solution chemistry module of WHAM (Tipping, 1994), correcting for temperature (20°C) and ionic strength. Quickly reacting Al (Al-qr) was used as an estimate for dissolved inorganic Al. Thermodynamic constants are reported in Tipping (1994).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION AND CONCLUSIONS
 REFERENCES
 
Soil Chemical Characteristics of Soil C and Soil A
Some characteristics of Soil C (forest floor of nonmanipulated reference soil) and Soil A (forest floor with increased Al content) used in this study are shown in Table 2. The content of exchangeable and organically bound Al in Soil A was more than twice as high as in Soil C. Exchangeable acidity was higher, and exchangeable Ca was lower than in Soil C. The increase in exchangeable acidity in Soil A was almost entirely due to an increase in Al. In both soils, the dominant cation at the exchange complex was Ca. The degree of Al saturation of the complexation sites associated with organic matter was expressed as the bound Al ratio (BAR) (Cronan et al., 1986). Cronan et al. (1986) showed that Al solubility in forest floors could be described as a function of BAR and pH. Bound Al ratio is defined as the ratio of the soil content of reactive Al and the carboxyl content of the soil: BAR = 3 (moles of Al)/(moles of carboxyl groups).


View this table:
[in this window]
[in a new window]
  		Table 2. Characterization of Soil A (forest floor manipulated with respect to its Al content) and Soil C (reference soil).

 
The CuCl2–extractable Al was chosen as a measure for the content of "reactive" Al, that is, the fraction that is believed to have a rapid interaction with added solutions. Copper chloride has been proposed as an extractant of organically complexed Al (Juo and Kamprath, 1979). Thus we assumed that reactive Al was organically complexed, an assumption supported by Lofts et al. (2001). The BAR of Soil C was 0.12, which is within the range of BAR values in natural-forest floors reported by Walker et al. (1990) and De Wit et al. (1999). The BAR of Soil A was 0.27.

Batch-1
In batch experiment 1 (Batch-1), pH and concentrations of DOC increased, whereas concentrations of Al-qr decreased upon addition of base (expressed as added [OH-] – added [H+]; Fig. 1a,b). At given acid or base additions, more Al-qr was mobilized in Soil A than in Soil C. This reflects the higher Al content of Soil A. The concentration of Al-qr in Soil C decreased to values at the detection limit. For any given addition of base, pH in the solution of Soil A was lower than in the solution of Soil C, even though Soils A and C had equal contents of exchangeable H (calculated as the difference between exchangeable acidity and Al, Table 2). Probably more protons were mobilized in Soil A than in Soil C because of the preferential adsorption of the soil for polyvalent cations. Thus the mobilization of protons was possibly favored over the mobilization of Al. Concentrations of DOC increased with increasing base addition and were generally highest in Soil C (Fig. 1c). This suggests that the soil content of Al affects the solubility of SOM. Each observation was done in triplicate. Concentrations of DOC showed a relatively large variability (error bars in Fig. 1c), particularly at low additions of base where the ionic strength of the solution was low.



View larger version (32K):
[in this window]
[in a new window]
  		Fig. 1. Results of equilibrium experiments Batch-1 (a–c) and Batch-2 (d–f). In Batch-1, acid or base was added to Soil A and Soil C. In Batch-2, AlCl3 was added to Soil A and Soil C. Soil A has a higher Al content than Soil C. Open symbols represent observations for Soil C, and closed symbols for Soil A. Lines are WHAM simulations (dotted is Soil C; solid is Soil A). Note that the Y-axis for Al-qr is different for Batch-1 and Batch-2. Error bars show standard errors.

 
Batch-2
The Al that was added in Batch-2 was largely adsorbed. Increased concentrations of Al-qr (Fig. 1e) occurred, in particular, at additions >0.6 x 10-3 mol L-1 Al. At an Al addition of 0.74 x 10-3 mol L-1, the concentration of Al-qr was twice as high in Soil A as in Soil C, although BAR values for Soils A and C at this addition of Al were very close, that is, 0.70 and 0.59, respectively (Table 3). At this Al addition, pH was {approx}3.2 (Fig. 1d), where relatively small changes in pH are associated with large changes in Al activity. The pH declined upon Al addition because of decomplexation of H+ and exchange of added Al with cations, including protons, at the exchange complex (Fig. 1d). At given Al additions, pH in Soil A was lower than in Soil C, similar to Batch-1. Concentrations of DOC decreased with Al addition (Fig. 1f). Just as in Batch-1, the highest concentrations of DOC were found in Soil C, which is the soil with the lowest Al content.


View this table:
[in this window]
[in a new window]
  		Table 3. Bound Al ratio (BAR) in suspensions of equilibrium experiment Batch-1 (zero addition of Al) and Batch-2 for Soil C (reference soil) and Soil A (soil with increased Al content) as a function of added concentrations of Al.

 
Field Observations
The mean composition of the soil solution in the forest floor in 1998 sampled in the field using tension lysimeters is presented in Table 4. The observations showed considerable variability, demonstrated by the minimum and maximum values between brackets. Fluctuations in soil moisture content due to precipitation and evapotranspiration were an important source of variability. The added solution was on average diluted twice, but this varied considerably throughout the growing season. Concentrations of Cl in soil water were about twice as high in Treatment A-2 as in Treatment A-1. Cl concentrations in Treatment A-0 were slightly higher than in Treatment C, possibly due to mineralization of organic matter (Hjelm et al., 1995) into which Cl was incorporated during the initial additions of AlCl3 in 1996 and 1997. High concentrations of anions were associated with high concentrations of cations—as expected on the basis of the principle of electroneutrality. Thus the lowest pH was found in Treatment A-2. Total Al concentrations in Treatments C, A-0, and A-1 were of similar magnitude even though irrigation water in Treatment A-1 contained considerable concentrations of Al. Quickly reacting aluminum was close to the detection limit in Treatments A-0 and C, indicating that the largest part of total Al concentrations was organically complexed. Concentrations of DOC decreased in the order C {approx} A-0 > A-1 > A-2. Only in Treatment A-2 was a substantial part of total Al in an inorganic form.


View this table:
[in this window]
[in a new window]
  		Table 4. Average composition of field-collected soil solution of forest floor using tension lysimeters in Treatments C, A-0, A-1, and A-2 in 1998 (averaged over three plots per treatment). Treatment C is the reference treatment. Treatments A-0, A-1, and A-2 have a raised soil Al content and received weekly inputs of dilute AlCl3. Between brackets are minimum and maximum observations.

 
The observations from Treatments A-0, A-1, and A-2 suggest that addition of AlCl3 leads to enhanced cation concentrations, decreased DOC levels and, at relatively high additions, to a substantial increase in monomeric inorganic Al. Qualitatively, these effects agree with the effects of Al addition in Batch-2.

In soil solution collected using zero-tension lysimeters, higher concentrations of total Al, Al-qr, and Cl were observed than in soil solution collected by tension lysimeters (Table 5). Concentrations of total Al in zero-tension solution from Treatments A-1 and A-2 were close to the target Al concentrations, which were 0.1 x 10-3 and 0.2 x 10-3 mol L-1, respectively. Thus soil solution collected by zero-tension lysimeters resembled the input solution closer than soil solution collected by tension lysimeters. This suggests that zero-tension lysimeters collected solution that had interacted less with the solid phase than solution collected with tension lysimeters.


View this table:
[in this window]
[in a new window]
  		Table 5. Comparison of mean concentrations of Al-tot, Cl, and Al-qr in added solution to the forest floor, and forest-floor soil solution sampled by tension and zero-tension lysimeters in 1998 in Treatments C, A-0, A-1, and A-2 in 1998. Standard deviations are between brackets. Treatment C is the reference treatment. Treatments A-0, A-1, and A-2 have a raised soil Al content and received weekly inputs of dilute AlCl3.

 
Aluminum Solubility in Batch and in the Field
The pH–pAl plots for Batch-1 and Batch-2 are shown in Fig. 2. Observations where Al-qr was at the detection limit are not shown because of the limited accuracy of the speciation analysis at very low Al concentrations. This is illustrated in Fig. 1b, where increasing additions of base to Soil C did not result in a further decrease in Al-qr once the detection limit of Al-qr was reached. This would imply that concentrations of free Al are independent of pH at that part of the curve, which is contrary to our mechanistic understanding of Al solubility controls. All solutions were undersaturated with respect to amorphous gibbsite (pK = 8.0), but the degree of undersaturation decreased when pH increased. For pH < 3.7, Al activity in Batch-1 was lower than in Batch-2. Here, the soil content of Al, as indicated by BAR, was distinctly lower in Batch-1 than in Batch-2 (Table 3).



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 2. pH–pAl (negative logarithm of Al activity) for equilibrium experiments Batch-1 and Batch-2 for Soils A and C. Soil A has a higher Al content than Soil C. Symbols represent observations and lines are WHAM simulations.

 
The solubility of SOM increased with pH and decreased with soil Al content (Batch-1; Fig. 3a). The Batch-2 data points showed a similar trend for increase of DOC with pH, but because of only minor differences in BAR between Soils A and C at low pH (Table 3), only a weak trend was shown for the decrease in SOM solubility with increasing soil Al content.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 3. pH–DOC for equilibrium experiments Batch-1 (a) and Batch-2 (b) for Soils A and C. Soil A has a higher Al content than Soil C. Symbols represent observations and lines are WHAM simulations.

 
In Fig. 4, Al solubility in Soil A is compared for Batch-1 and field-collected solutions. According to Cronan et al. (1986), Al solubility depends on BAR value and pH. Thus the pH–pAl relationship found in Batch-1 for Soil A (shown by the regression line) could be expected to describe Al solubility in Soil A. However, pH–pAl data points from field samples showed considerable variability and deviated from the pH–pAl relationship derived from Batch-1, especially at lower pH. The slopes of pH–pAl regression lines based on field-collected samples were steeper, and minimum pAl values (= maximum Al activities) were lower than found in Batch-1, especially for zero-tension-collected solutions. The regression line for tension-lysimeter-collected solutions was between the regression lines for Batch-1 and zero-tension lysimeters. This suggests that soil solution collected with tension lysimeters was generally closer to equilibrium than soil solution collected using zero-tension lysimeters. This is supported by the comparison of Al and Cl levels in input and sampled solutions (Table 5), which shows that Al and Cl concentrations in the zero-tension solutions were more similar to the input solutions than the tension solutions. In conclusion, it is reasonable to assume that at least some of the solution samples collected in the field by tension lysimeters were close to equilibrium, which is a necessary condition for using an equilibrium model to describe field solution chemistry.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 4. pH–pAl relationships in Soil A for equilibrium experiment Batch-1 and for solutions collected in the field using tension and zero-tension lysimeters. Regression lines are also shown.

 
Calibration of WHAM
The success of a model calibration can be judged on the ability of the model to fit the data and on whether the optimized values for model parameters agree with analytically estimated values (Tipping et al., 1995).

The model fitted the observations of pH in Batch-1 as a function of base or acid addition very well (Fig. 1a). The relative mean deviation between fitted and measured (mean) pH (RDM-pH) was only 1% or 0.04 pH-unit (Table 6). For concentrations of inorganic Al (Al-qr) (Fig. 1b), the RDM-Al was about 25%, which is about 3 x 10-6 mol L-1. Judging from Fig. 1b, Soil C was apparently described better than Soil A, contrary to the RDM-Al value in Table 6. However, RDM-Al is expressed as a percentage of the mean Al-qr concentration. On average, lowest Al concentrations were found in Soil C. For inorganic Al concentrations >5 x 10-6 mol L-1, RDM-Al was 9%, and for concentrations <5 x 10-6 mol L-1, RDM-Al was 107% (measurements for Soils A and C combined). Thus WHAM fitted high concentrations of Al-qr considerably better than low concentrations of Al-qr. A consequence of this is that WHAM described Al solubility at high pH and low pAl rather poorly (Fig. 2).


View this table:
[in this window]
[in a new window]
  		Table 6. Deviation between WHAM simulations and observations for Soil A and Soil C.

 
One reason for the poor performance of WHAM at low concentrations of Al is the procedure that is followed for optimization of model parameters. In this procedure, the difference between observed and calculated concentrations of Al-qr are minimized. Concentrations of Al-qr ranged from about 10-4 to 10-6 mol L-1. The error in the model description of Al-qr (FAl) is thus mostly determined by the error in high concentration of Al-qr, which puts most weight on high concentrations of Al-qr in finding the optimum value for CHS and CAL. This optimization procedure is a sensible approach when a considerable number of observations are in the low concentration range, where analysis of Al-qr has a limited accuracy, as in this study. Lofts et al. (2001) use a different optimization procedure in which concentrations of Al are converted to a logarithmic scale (i.e., pAl), and FAl is defined in a similar way as FpH.

Another reason for the poor performance of WHAM in the high pH–pAl range is the discrepancy between the observed behavior of Al and its expected behavior on theoretical grounds. Observations of pH–pAl converge when pH increases, independent of soil Al content (Fig. 2), which was also shown by Walker et al. (1990). This suggests that at relatively high pH, Al activity is independent on the extent of Al binding of the soil, in contrast to conventional equilibrium expressions for chemical reactions, which are incorporated in WHAM.

WHAM fitted the observed increase in DOC concentrations with base addition, and the generally higher DOC level in Soil C than in Soil A, fairly well (Fig. 1c). However, the relatively sudden increase in DOC around a base addition of zero was not captured by WHAM. RDM-DOC (Table 6) was about 15% for both soils. WHAM described the correct range for pH and DOC in Fig. 3, but described a smooth increase in DOC with pH, contrary to the observed sharp increase around pH 4 in Soil A. Thus, the success of the model fit of the observations was greatest for pH and relatively high concentrations of Al.

The second measure for the success of the model fit is the comparison of optimized values for model parameters with analytically estimated quantities. Model parameters CHS and CAL were optimized using the data set of Batch-1 (Table 7). When comparing the optimum values of the model parameters with measured quantities, it is important to keep in mind the limitations of both the model parameters and the analytical methods. For example, the characteristics of HS are based on properties from isolated humic substances that may behave differently isolated than as part of a soil. Additionally, measuring the "reactive" soil pool of Al may be done using various extractants. Lofts et al. (2001) argue that extractions that aim to characterize various distinct soil pools of Al may not be so specific, since one is not sure which solid phases are being attacked. Given these uncertainties, Lofts et al. (2001) used analytically estimated values of reactive Al as a constraint for finding a value for CAl, rather than as an exact value, and allowed CAl to vary between 50 and 200% of CuCl2–extractable Al.


View this table:
[in this window]
[in a new window]
  		Table 7. Optimized model parameters for Soil C (reference soil) and Soil A (soil with increased Al content).

 
Here, the optimum value for CAL amounted to 50% of CuCl2–extractable Al in Soil C and to 35% of CuCl2–extractable Al in Soil A, which is for Soil A below the constraint used by Lofts et al. (2001). Values of CHS for Soils A and C were almost equal and amounted to {approx}11% of soil dry weight. No direct analytical determination of CHS was available and thus CHS was compared with the soil content of carboxylic groups by multiplying CHS (g g-1 soil) with the soil content of Type A reactive groups (pK value < 7, "strong" acid; Table 1). Thus the "optimized" COOH content was 75% of the measured COOH content for Soil A and 84% for Soil C.

The model parameter {gamma} is a distribution parameter and has no equivalent that can be measured. {gamma} was higher for Soil A than for Soil C, which indicates that according to the definition of {gamma}, FA in Soil C was more hydrophilic than FA in Soil A. The optimized value of {gamma} was low compared with other reported values (ranging from 1.5 to 2.3 [Tipping et al., 1995; de Wit et al., 1999]).

In conclusion, the optimized model parameters CHS and CAL were somewhat low, but within tolerable distance of analytically determined quantities, and allowed a good description of pH and of inorganic Al concentrations >5 x 10-6 mol L-1, that is, for those concentrations that are most relevant for evaluation of Al toxicity.

Simulation of Batch-2
The data set of Batch-1 was used for fitting the model parameters. Thus the data set of Batch-2 represented an independent data set for testing the ability of WHAM to describe solid–solution interactions in Soils A and C (Fig. 1d–f).

WHAM simulated the effects of AlCl3 addition on pH well (Fig. 1d) but tended to underestimate pH by 0.15 units at low Al additions. The RDM-pH in Batch-2 was as low as 3% (Table 6). WHAM predicted the distinct rise in Al-qr with Al addition fairly well for both soils, but tended to overestimate Al-qr at Al additions greater than 0.6 x 10-3 mol L-1. Not surprisingly, RDM-Al in Batch-2 was considerably larger than in Batch-1 for which WHAM was optimized. Again, high concentrations of Al were predicted better than low concentrations: RDM-Al was 61 and 109% for concentrations greater than and less than 5 x 10-6 mol L-1, respectively. Simulation of pH–pAl relationships for Batch-2 (Fig. 2) suffered from the same weakness as simulations for Batch-1. When Al activity was relatively high (pAl < 6), WHAM simulations were accurate, but at lower Al activity, the pH–pAl relationship predicted by WHAM deviated from measured relationships. Concentrations of DOC in Batch-2 were predicted somewhat better than in Batch-1, indicated by RDM-DOC (Table 6). The variability in measured DOC concentrations in Batch-2 was considerable, but suggested an increase of DOC with pH and with a decrease in soil Al content (Fig. 3) similar to Batch-1.

Given the complexity of the Al–OM interactions and the fact that the only changes with regard to the model input for Batch-1 were the total soil content of Al and inorganic anion level, we regard the model predictions of the effects of AlCl3 addition on pH and concentrations of inorganic Al and DOC as satisfactory. This strongly supports the description of the solution–solid phase interactions embodied in WHAM.

Simulation of AlCl3 Addition in the Field
WHAM was used to simulate field conditions by assuming that the forest floor in the field was a well-mixed box of soil and soil solution with higher SSRs than used in batch (SSR of 10 g L-1). This assumption implies that the mechanisms that control Al solubility in batch also apply in the field, that is, for a given soil Al content and pH, the same Al activity is expected in the field as in batch.

A realistic soil moisture content of a moist forest floor may be about 50%, which corresponds to an SSR (using a mean bulk density of 0.1 g cm-3) of 200 g L-1. WHAM calculated concentrations of H, Al-qr, and DOC for SSR values of 10, 50, 100, and 200 g L-1 for Soil A (Fig. 5). An increase in SSR implied a proportional increase in the pool of cations and humic compounds. The addition of Al as AlCl3 was negligible compared with the soil pool of Al that was present. Input Cl concentrations varied from 1 x 10-5 to 1.5 x 10-3 mol L-1. FA0 was set at 0.



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 5. WHAM simulations of concentrations of H+, inorganic Al, and DOC for soil to solution ratios (SSR) of 10, 50, 100, and 200 g L-1 against Cl concentrations in Soil A. Field measurements (tension lysimeters) are also shown. Symbols represent observations and lines are WHAM simulations.

 
Upon an increase in concentrations of Cl, WHAM calculated an increase of concentrations of H+ and inorganic Al and a decrease in DOC, especially at low Cl concentrations (<0.1 x 10-3 mol L-1). Only at SSR 10, concentrations of H and inorganic Al increased markedly with Cl concentration. The ranges that were simulated for concentrations of H and inorganic Al were very similar for SSR 50, 100, and 200 (separate lines hardly distinguishable in Fig. 5a,b), indicating that inorganic equilibrium solution chemistry at SSR>=50 was almost independent of SSR. Simulations of DOC levels were highest at SSR 200 and were approximately proportional to SSR. This is in agreement with Chapman et al. (1997) who measured DOC in water extracts of an Ap horizon (C content 4.5%) considering the effect of the SSR. In Chapman et al.'s experiments, SSR values ranged from 65 to 666 g L-1 and DOC levels increased with SSR from 8 x 10-3 to 17 x 10-3 g L-1.

Field observations of pH, Al-qr, and DOC showed considerable variability, which were very different from the predicted smooth relations between concentrations of Cl and the mentioned variables (Fig. 5). Generally speaking, WHAM predictions of an increase in H+ and inorganic Al concentrations and relatively constant DOC concentrations, with increasing Cl concentrations, were correct. However, the predicted ranges were far off observed ranges. In general, the closest fit to the measurements was at SSR 10, a value that is not representative for field conditions. The largest discrepancies between predictions and observations were for concentrations of inorganic Al and DOC. Thus, WHAM simulations of concentrations of H, inorganic Al, and DOC poorly agreed with measurements done in solutions collected with tension lysimeters in the field.


   DISCUSSION AND CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION AND CONCLUSIONS
REFERENCES
 
The mechanistic equilibrium model WHAM was used as a tool to evaluate solution–solid phase interactions regarding the solubility of Al and SOM in the forest floor. The model was tested against results of studies conducted in batch and in the field. WHAM was optimized using pH, Al-qr, and DOC measured in experiment Batch-1 and managed to describe pH and concentrations of inorganic Al and DOC successfully (Fig. 1a–c). Similar and equally successful tests of WHAM using both organic horizons and mineral soils were performed in Tipping et al. (1995), Berggren and Mulder (1995), De Wit et al. (1999), and Lofts et al. (2001). Simulation of an independent batch experiment in which AlCl3 was added to the same soils using the optimized model (Batch-2) resulted in adequate simulated values of pH and concentrations of inorganic Al and DOC (Fig. 1d–f). Optimized model parameters were within tolerable distance from analytically determined quantities. This strongly supports the assumptions about Al solubility being controlled by humic complexation, as embodied in WHAM. Additionally, it illustrates the ability of WHAM to be used as a tool to predict soil solution chemistry for other experimental conditions than it was calibrated for.

A prerequisite for using an equilibrium model to predict soil solution chemistry is that reactions have proceeded quickly enough so that equilibrium was attained. We tested whether Al3+ activity in the field-collected solutions was in equilibrium with the solid phase by using the pH–pAl relationship from the equilibrium study Batch-1 as a reference, similar to Berggren (1999). The pH–pAl relationship obtained in Batch-1 did not describe Al solubility in the field samples very well, but some overlap existed with Al solubility found in solutions collected by tension lysimeters, which suggested that some of these solutions were close to equilibrium (Fig. 4). However, WHAM predictions of concentrations of inorganic Al, H, and DOC corresponded poorly to measured concentrations in the solutions collected by tension lysimetry (Fig. 5).

Possible reasons for the discrepancy between WHAM simulations and observations may be an insufficient model description of soil chemistry in concentrated soil-solution systems and/or lack of equilibrium between collected solutions and the solid phase. In the latter case, WHAM could not be expected to reproduce the observed concentrations.

The equilibrium condition may be violated when contact between the soil and the incoming water is insufficient for reactions to proceed to near equilibrium, for example, when a substantial portion of the incoming water moves through the soil by so-called "macropore flow" (Reuss et al., 1986). Zero-tension lysimeters collect mobile solution that moves freely down the soil by force of gravity, whereas tension lysimeters collect a mixture of mobile and more slowly moving solution that is held by capillary forces. Centrifugation of soil expels solution that is even more strongly held (Giesler et al., 1996). This suggests that tension lysimeters collect soil solution that has interacted more closely with the solid phase than zero-tension lysimeters (Lawrence and David, 1996), which affects the chemistry of the solutions that were collected.

Giesler et al. (1996) reported significantly higher solute concentrations in forest-floor centrifugates than in leachate from the forest floor collected by zero-tension lysimeters. Total organic C (TOC) concentrations ranged from 0.14 to 0.61 g L-1 in centrifugates, which is the range of DOC simulated by WHAM for SSR 100 and 200 (Fig. 5). In zero-tension lysimeter leachate in the same soil, TOC ranged from 0.02 to 0.09 g L-1, similar to DOC levels in tension lysimeter solution (Fig. 5). The results of Giesler et al. (1996) agree with studies by Zabowski and Ugolini (1990) and Fernandez et al. (1995). Zabowski and Ugolini (1990) compared tension lysimeter and centrifuge soil solutions from an Oa horizon and found considerably higher DOC levels in centrifuge solutions than in solution collected by zero-tension lysimeters, that is, 0.24 g L-1 and 0.06 g L-1, respectively. Fernandez et al. (1995) compared solutions collected by low-tension lysimetry from a forest floor with zero-tension lysimeter solutions and found higher solute concentrations in solutions collected by tension lysimeters.

Results from the studies quoted suggest that differences in chemical composition of mobile macropore solution (collected by zero-tension lysimeters) and relatively immobile micropore solution (collected by tension lysimeters and centrifugation) may be related to the water residence time of the sampled solutions. The longer the residence time, the longer the reaction time and the higher the probability that the collected solution will be in equilibrium with the solid phase. This concept was approached quantitatively by Tipping (1996) who developed the hydrochemical model CHUM to describe streamwater chemistry at catchment scale. In CHUM, diffusion-limited interchange of solutes between relatively immobile micropore water and rapidly flowing macropore water is a key concept. Chemical soil–solution interactions were described using the principles of WHAM. The model CHUM was tested on a catchment in the UK and described streamwater chemistry reasonably well (Tipping, 1996).

The concept of diffusion-limited interchange of solutes between mobile macropore and immobile micropore water may supply a plausible qualitative explanation for the discrepancies between WHAM simulations and field observations in Fig. 5. Incoming water to the forest floor consisted of dilute AlCl3, which was partly rapidly transported through macropores in the forest floor and collected by zero-tension lysimeters. Tension lysimeters, however, collected a mixture of micropore solution, with a relatively longer residence time than macropore solution, and macropore solution. Thus zero-tension lysimeter solution resembled input solution more, whereas tension lysimeter solution resembled more the equilibrium solution, that is, micropore solution (see Table 5). Thus the apparent Al solubility found in solutions collected in the field may be dependent on the sampling method. Further studies comparing centrifuged solution with model predictions are warranted.


   ACKNOWLEDGMENTS
 
Thanks to Magne Huse and Egil Kortnes for assistance in the field, and to Gro Wollebæk and the analytical laboratory at NISK for doing part of the chemical analyses. Batch experiments and chemical analyses were carried out at NISK and at the university of Oslo. Ed Tipping is gratefully acknowledged for helpful discussions. H.A.W. was supported by a grant from the Norwegian Research Council (109475/720).

Received for publication June 27, 2000.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION AND CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Soil Sci.Home page
T. Scheel, C. Dorfler, and K. Kalbitz
Precipitation of Dissolved Organic Matter by Aluminum Stabilizes Carbon in Acidic Forest Soils
Soil Sci. Soc. Am. J., January 1, 2007; 71(1): 64 - 74.
[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 (11)
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.
Related Collections
Right arrow Soil Models
Right arrow Soil Analysis