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 (13)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Zhang, H.
Right arrow Articles by Payton, M. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Zhang, H.
Right arrow Articles by Payton, M. E.
Agricola
Right arrow Articles by Zhang, H.
Right arrow Articles by Payton, M. E.
Related Collections
Right arrow Nutrient Management
Right arrow Phosphorus
Right arrow Soil Chemistry
Published in Soil Sci. Soc. Am. J. 69:96-106 (2005).
© 2005 Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

Division S-2—Soil Chemistry

Path and Multiple Regression Analyses of Phosphorus Sorption Capacity

H. Zhanga,*, J. L. Schrodera, J. K. Fuhrmana, N. T. Bastad, D. E. Stormb and M. E. Paytonc

a Dep. of Plant and Soil Sciences
b Dep. of Biosystems and Agricultural Engineering
c Dep. of Statistics, Oklahoma State Univ., Stillwater, OK 74078
d School of Natural Resources, The Ohio State Univ., Columbus, OH 43210

* Corresponding author (hailin.zhang{at}okstate.edu)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Soil P saturation indices and P Langmuir adsorption maximum (Smax) are two environmental soil tests that provide valuable information for the proper management P in soils to avoid the overapplication of P. The objectives of this study were to determine Smax and develop P saturation indices for 28 Oklahoma benchmark soils and to use path analysis and multiple regression to examine the relationships between Smax and soil properties. Soil samples were analyzed for pH, clay content, oxalate extractable P (Pox), Al (Alox), Fe (Feox), and Mehlich-3 (M3) extractable P (PM3), Al (AlM3), Fe (FeM3), Ca (CaM3), and Mg (MgM3). The Smax value and saturation indices based on oxalate and M3 extractions were determined. The Smax value ranged from 34 to 500 mg kg–1 and was highly correlated with clay content (r = 0.79), organic C (r = 0.80), Alox (r = 0.88), and Feox (r = 0.83). Soil pH was not correlated (p > 0.05) with Smax. Path analysis showed significant direct effects (p < 0.01) between Alox and Smax and between Feox and Smax but these relationships were highly influenced by indirect effects of Alox and Feox. Multiple regression agreed well with path analysis and found that the combination of Alox and Feox were the two most important soil properties related to Smax of the soils studied. Significant relationships existed between AlM3 (r = 0.54) and Smax and between FeM3 (r = 0.54) and Smax. Three P saturation indices studied were highly correlated (p < 0.05) with each other. Our results show that Smax of Oklahoma soils may be predicted with oxalate extractable Al and Fe or M3 extractable Al, Fe, and Ca.

Abbreviations: AlM3, Mehlich-3 extractable Aluminum • Alox, oxalate extractable aluminum • DPS, degree of phosphorus saturation • FeM3, Mehlich-3 extractable iron • Feox, oxalate extractable iron • M3, Mehlich-3 • PSI, phosphorus saturation index • Pox, oxalate extractable phosphorus • PM3, Mehlich-3 extractable phosphorus, Psat, phosphorus saturation • PSIM3, phosphorus saturation index calculated with Mehlich-3 data • PSIox, phosphorus saturation index calculated with oxalate data • r, simple correlation coefficient • R2, coefficient of determination • Smax, phosphorus adsorption maximum • U, uncorrelated residue value


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
ALTHOUGH P is considered to be relatively immobile in the soil system (Johnson et al., 1997), there are mechanisms for P to leave the soil through plant uptake, loss in surface runoff, erosion of sediment, and leaching through the soil profile. If P is applied to the soil in excess of the crop requirement, P will generally build up in the soil, which increases the chances of P loss from the soil system (Sharpley et al., 1999). The N/P ratio of animal manures ranges from 1:1 to 4:1 (Zhang et al., 2004) and is generally less than the N/P ratio 8:1 taken up by most crops and pastures (USDA, 2001). The land application of animal manure to meet crop N needs often results in an overapplication of P, thus leading to an accumulation of P in the soil (Sharpley et al., 1999). Agricultural runoff is now considered a primary nonpoint source of P pollution because many point sources are largely under control (Daniel et al., 1994) and may lead to eutrophication of water bodies (Sharpley et al., 1999). To prevent P accumulations in soil, manure should be applied according to the P needs of the crop, and then supplemented with N fertilizer to accommodate the N needs of the crop (Daniel et al., 1994). Unfortunately, this is not always economically or practically feasible (Sharpley et al., 1996).

Many agricultural fields contain soil P levels that exceed crop requirements resulting from long-term manure application. In these situations, the environmental fate of P must be assessed. One tool that has been used in the evaluation of the environmental fate of P is soil P saturation. Phosphorus saturation is defined as the amount of P sorbed divided by the P sorption capacity of the soil. The concept of P saturation is meaningful as it estimates the degree to which P sorption sites have been filled and indicates the potential desorbability of soil P (Beauchemin and Simard, 1999). Phosphorus saturation has been highly correlated with P desorption such that P desorption increases with higher degrees of P saturation (Sibbesen and Sharpley, 1997). Phosphorus saturation is viewed as an environmental indicator of soil P because it has been found to be a good indicator of P availability to runoff and leachate (Kleinman and Sharpley, 2002).

Several approaches for the estimation of P saturation have been reported (Sharpley, 1995; Schoumans, 2000; Kleinman and Sharpley, 2002). One common approach is to determine acidified ammonium oxalate extractable P, Al, and Fe (Pox, Alox, Feox), and then calculate a degree of P saturation (DPS) (Schoumans, 2000). According to Schoumans (2000), DPS by the acid ammonium oxalate method is equal to the ratio of the amount of oxalate-extractable P divided by the P sorption capacity. It is assumed that the sum of the oxalate extractable Al and Fe equals the P sorption capacity. A critical DPS of 25% has been established for Dutch soils (Sharpley et al., 1996). Above this limit, the risk of P losses to leaching and surface runoff becomes unacceptable to the Dutch government and further applications of manure may be prohibited. However, this method may not be applicable to high pH soils, especially calcareous soils because the carbonates in calcareous soils tend to neutralize the acidic extracting solution (Loeppert and Inskeep, 1996; Kleinman and Sharpley, 2002). Furthermore, Al and Fe oxides are less significant for P sorption in high pH soils than acid soils (Lindsay, 1979).

A disadvantage of the definition of DPS is that its calculation depends on the P sorption capacity of the soil, which varies from horizon to horizon but is usually assessed by 0.5 (Alox + Feox) (Schoumans, 2000). However, it is possible to eliminate this assessment of P sorption capacity and to calculate an independent P saturation index (PSI) as shown in Eq. [1]:

[1]
where Pox, Alox, and Feox are expressed in mmol kg–1 soil.

Another approach proposed by Sharpley (1995) for the estimation of P saturation uses Mehlich-3 (M3) P (Mehlich, 1984) and the adsorption maximum (Smax) from P adsorption isotherms. It is referred to as Psat and is defined as:

[2]
where PM3 is M3 extractable P, Smax is P Langmuir adsorption maximum and are expressed in milligrams per kilogram (mg kg–1) of soil. Mehlich-3 extraction is widely used to assess the amount of P available to a plant during the growing season. It works well in soils with a wide range of pH (Mallarino, 1997).

Still another approach offered by Kleinman and Sharpley (2002) is similar to the acid ammonium oxalate approach but involves extraction of P, Al, and Fe with M3 as shown in Eq. [3].


[3]
where PM3, AlM3, and FeM3 are M3 extractable P, Al, and Fe, respectively, and are expressed in millimole per kilogram (mmol kg–1) of soil.

Phosphorus adsorption characteristics are influenced by one or a combination of chemical and mineralogical properties of soil such as clay type and content, Fe and Al oxides, organic C (OC), pH, and CaCO3 (Burt et al., 2002). Soil properties that have been correlated with P adsorption in soils include soil pH (Brennan et al., 1994; Dodor and Oya, 2000), OC and clay content (Singh and Tabatabai, 1977; Sanyal et al., 1993; Dodor and Oya, 2000), clay content, and Al and Fe oxide content (Sanyal et al., 1993; Dodor and Oya, 2000; Börling et al., 2001). Often, these properties are interrelated and autocorrelated (Basta et al., 1993), which makes it difficult to determine the components that contribute most to P adsorption in soils (Syers et al., 1973). Therefore, simple correlation analysis inadequately explains the relationships because correlation does not imply that a direct cause-and-effect relationship exists (Wright, 1921). Rather, the correlation analysis and the resulting coefficient may be influenced by indirect effects.

Path analysis is a statistical technique that partitions correlations into direct and indirect effects and distinguishes between correlation and causation (Wright, 1934; Afifi and Clark, 1984). Path analysis has been used extensively in agronomic studies (Gravois and Helms, 1992; Pantone et al., 1992; Cramer and Wehner, 2000; Zheng et al., 2002; Garcia del Moral et al., 2003) and to investigate relationships between soil properties and adsorption of heavy metals (Basta et al., 1993; Krishnasamy and Mathan, 2001). Several studies have used either simple correlation to examine relationships between individual soil properties and P sorption maxima or multiple regressions to evaluate the effect of different combinations of soil properties on P adsorption (Singh and Tabatabai, 1977; Sanyal et al., 1993; Brennan et al., 1994; Dodor and Oya, 2000; Börling et al., 2001). However, to the best of our knowledge, none have utilized path analysis to examine the contributions of the different soil properties to correlations established between soil properties and P adsorption. The objectives of this study were to (i) characterize P adsorption maximum (Smax) of major benchmark soils from Oklahoma; (ii) use path analysis to investigate the relationships between Smax and major soil properties; and (iii) to determine the relationships between different P saturation indices. The study will also provide useful information about Smax and P saturation indices in Oklahoma soils, which may be used by environmental managers to avoid overapplication of P and minimize eutrophication of water bodies.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Study Soils
Twenty-eight Oklahoma benchmark soils originally classified by Gray and Roozitalab (1976) and which had not received P additions as manure or commercial fertilizers within 3 yr of collection were chosen for P characterization to identify factors affecting their P sorption capacity. The soils were originally collected by Scott (1994). One sample per soil type was chosen for the study. Oklahoma has a diverse paleoclimate and geology with soils that represent many of the world soil orders (Table 1). Soils in this study represent the diversity of soils found in Oklahoma along with the major land resources areas (Fig. 1) . The soils were sampled from the surface horizon (A horizon or plow layer) then air-dried and ground to pass a 2.0-mm sieve.


View this table:
[in this window]
[in a new window]
  		Table 1. Classification and general properties of 28 Oklahoma benchmark soils.

 


View larger version (18K):
[in this window]
[in a new window]
  		Fig. 1. Locations of the 28 Oklahoma benchmark soils used in this study along with delineation of major land resource areas.

 
Soil Properties
Soil properties (clay content, OC content, and soil pH) analyzed by Scott (1994) were utilized in the study. Soil pH was measured in 1:2 soil/0.01 M CaCl2 suspension (McLean, 1982). Soil organic C content was determined by acid dichromate digestion according to Heanes (1984). Clay content was determined by the hydrometer method (Gee and Bauder, 1986).

Adsorption Isotherms
Phosphorus adsorption isotherms were determined according to the method of Graetz and Nair (2000). One gram of soil sample was equilibrated with 25 mL of varying concentrations of P in 0.01 M CaCl2 solution in 50-mL centrifuge tubes. The concentrations of the solutions were 0.0, 0.5, 1.0, 5.0, 10.0, 15.0, and 20.0 mg P L–1. The tubes were shaken for 24 h on an end-to-end shaker at 150 oscillations per min. The samples were then centrifuged for 10 min at 5211 x g and the supernatant decanted. The P in solution was then quantified colorimetrically using the ascorbic acid method (Kuo, 1996). The amount of P adsorbed was determined by the difference between the initial and final amounts of P in solution. Duplicate analyses were conducted on all study soils

Phosphorus adsorption isotherms were determined with the linearized form of the Langmuir equation Eq. [4].


[4]
where S equals the total amount of P retained, mg kg–1; C equals concentration of P after a 24-h equilibrium, mg L–1; Smax equals P sorption maximum, mg kg–1; k equals a constant related to the bonding energy, L mg–1 P. Smax was calculated by regressing C/S versus C, where C is the equilibrium solution P concentration and S is adsorbed P. The reciprocal of the slope of the linear regression is Smax (Olsen and Watanabe, 1957; Syers et al., 1973).

Soil Extractions
Acid ammonium oxalate extractable P, Al, and Fe were determined by shaking duplicate 1.5-g samples of soil with 30 mL of 0.5 M (COONH4)2 · H2O at pH 3.0 in 50-mL centrifuge tubes (Schoumans, 2000). Samples were shaken for 2 h, in the dark, on an end-to-end shaker at 150 oscillations per minute and centrifuged for 10 min at 5211 x g. Supernatants were analyzed for P, Al, and Fe using a TJA-9000 inductively coupled plasma-atomic emission spectroscopy (ICP–AES).

Mehlich-3 extractable P, Al, Fe, Ca, and Mg were determined by shaking duplicate 2.0-g samples of soil and 20 mL of M3 solution in 50-mL centrifuge tubes for 10 min on an end-to-end shaker (150 oscillations per minute). The samples were then centrifuged at 5211 x g for 10 min and supernatants were analyzed by ICP–AES. Duplicate analyses were conducted on the study soils.

Calculation of PSIox, PSIM3, and Psat
Phosphorus saturation indexes based on ammonium oxalate and M3 extractions were calculated using Eq. [1] and [3], respectively. P saturation with respect to Smax (Psat) was calculated using Eq. [2]. Phosphorus saturation indices were expressed as percentages.

Statistical Analyses
Statistical analyses were performed using PC SAS Version 8.2 (SAS institute, 2001). Two different statistical techniques (backward-stepwise regression analysis and path analysis) were utilized to evaluate the effect of soil properties on Smax. Backward-stepwise regression analysis was used to generate empirical models capable of predicting Smax based on soil properties. The backward-stepwise regression was used to identify crucial soil properties that explain most of the variation in Smax. Soil properties that did not explain a significant part of the variation (i.e., p > 0.05) were not used as independent variables in the multiple regression equation.

Path analysis differentiates between correlation and causation by partitioning simple correlation coefficients between independent variables (soil properties) and dependent variables (Smax) into direct and indirect effects (Afifi and Clark, 1984; Basta et al., 1993). Path analysis provides a numerical value for both direct and indirect effects, thus indicating the relative strength of causal relationships (Loehlin, 1987). Direct effects are referred to as path coefficients and are standardized partial regression coefficients (Basta et al., 1993).

Two path analysis models similar to those used by Basta et al. (1993) were also used to evaluate the relationships between Smax and soil properties (Fig. 2) . The direct effects of soil properties on Smax are represented by single-headed arrows while coefficients of intercorrelations between soil properties are shown by double-headed arrows. Indirect effects of soil properties on Smax are determined from the product of one double-headed arrow and one single-headed arrow. The independent variables of the first model were soil pH, clay content, OC content, and Alox and Feox. The second model examined the relationships between Smax and soil pH, clay content, OC content, AlM3, FeM3, CaM3, and MgM3 (Fig. 2). For both models, direct and indirect effects were obtained from multiple linear regression of soil properties on Smax and simple correlations between soil properties (SAS Institute, 2001). Additionally, an uncorrelated residue (U) was calculated for both models using the following equation.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 2. Path diagram for the relationship between soil properties and P adsorption by soil. The direct effects (Pij) of soil properties on P adsorption maxima (Smax) are represented by single-headed arrows while the indirect effects (rijPij) of soil properties are shown by double-headed arrows. Subscript designations for soil properties and P adsorption are identified numerically as follows: (1) pH = soil pH; (2) clay = clay content; (3) OC = organic carbon content; (4) Alox = acid ammonium oxalate extractable Al and AlM3 = Mehlich-3 extractable Al; (5) Feox = acid ammonium oxalate extractable Fe and FeM3 = Mehlich-3 extractable Fe; (6) CaM3 = Mehlich-3 extractable Ca; (7) MgM3 = Mehlich-3 extractable Mg; and (8) Smax = P adsorption maxima.

 

[5]
where R2 is the coefficient of determination. Path analysis results were determined from the following equations (Williams et al., 1990):

[6]

[7]

[8]

[9]

[10]

[11]

[12]
where rij is the simple correlation coefficient between soil property and P adsorption, Pij are path coefficients (direct effects), and rijPij are the indirect effects of soil property on P adsorption. Subscript designations are: (1) pH, (2) clay content, (3) OC content, (4) Alox or AlM3, (5) Feox or FeM3, (6) M3 extractable Ca (CaM3), (7) M3 extractable Mg (MgM3), and (8) Smax (Fig. 2).

The path analysis results can be summarized in a concise table, which consists of a matrix with the main diagonal representing direct effects and off-diagonal elements representing indirect effects (Williams et al., 1990). The position of each element in the matrix corresponds to its position in the normal equations presented above.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Soil Characteristics
The selected soil samples had a wide range of chemical and physical properties. Soil pH ranged from 4.3 to 8.1, clay content from 70 to 660 g kg–1, and OC from 4 to 30.0 g kg–1 (Table 1). Overall, the soils were moderately acidic (median pH = 5.9), low in OC (median OC = 12 g kg–1), and low in soil test P (median M3 P = 16 mg kg–1 and ranged from 2.5 to 140 mg P kg–1 soil; agronomic optimum M3 P = 30 to 50 mg kg–1; SERA-IEG-6, 2001).

Phosphorus extracted by the acid ammonium oxalate method varied by soil and ranged from 26 to 750 mg P kg–1 soil with a median value of 120 mg P kg–1 soil (Table 2). Ranges for Alox extracted by this method were from 140 to 2000 mg Al kg–1 soil while Feox ranged from 120 to 9600 mg Fe kg–1 soil. On average, M3 extracted approximately 22% of the P, 64% of the Al, and 8% of the Fe extracted by ammonium oxalate method. Our results are similar to those of Sims et al. (2002) who reported that M3 extracted approximately 84% of the Al and 19% of the Fe extracted by acid ammonium oxalate in soils typical of the Mid-Atlantic USA. The P sorption data were satisfactorily described by the linearized Langmuir equation with correlation coefficients ranging from 0.95 to 0.98. Phosphorus sorption maximum (Smax) estimated by Langmuir adsorption isotherms ranged from 34 to 500 mg P kg–1 soil with a median of 180 mg P kg–1 soil.


View this table:
[in this window]
[in a new window]
  		Table 2. Phosphorus sorption characteristics for 28 Oklahoma benchmark soils.

 
Path Analysis and Multiple Regression—Ammonium Oxalate
Results for the path analysis of P adsorption are shown in Table 3. Simple correlation coefficients (r) between pH, clay content, OC, Alox, Feox, and Smax are presented for comparison with path analysis results. The uncorrelated residual value (U) was low (0.28) while the coefficient of determination (R2) was high (0.92) indicating that the path analysis model explained the majority of variation in P adsorption by soil. Significant correlation coefficients (p < 0.01) were found between clay content (r = 0.79), OC (r = 0.80), Alox (r = 0.88), Feox (r = 0.83), and Smax (Table 3). A significant correlation (p > 0.05) was not found between Smax and soil pH. Our results are similar to those of other researchers who reported nonsignificant relationships existed between soil pH and Smax (Brennan et al., 1994; Dodor and Oya, 2000). Additionally, the results of our study are consistent with the findings of others (Singh and Tabatabai, 1977; Sanyal et al., 1993; Dodor and Oya, 2000) in that clay content and OC are both well correlated with P Smax. Furthermore, several researchers have found significant relationships between AlOX and Smax and between Feox and Smax (Sanyal et al., 1993; Dodor and Oya, 2000; Börling et al., 2001).


View this table:
[in this window]
[in a new window]
  		Table 3. Path analysis direct effects (diagonal, underlined) and indirect effects (off diagonal) of soil pH, clay (g kg–1), organic carbon (g kg–1), and Fe and Al extracted by acid ammonium oxalate or Mehlich-3 on P adsorption for 28 Oklahoma benchmark soils.

 
Path analysis partitions each r value into one direct effect and four indirect effects. Partitioning by path analysis showed significant direct effects by Alox (r = 0.47) and Feox (r = 0.32) on Smax. However, the direct effects of clay and OC content on P sorption were not significant (p > 0.05). Examination by path analysis revealed that the indirect effects of Alox (r = 0.38) and Feox (r = 0.20) were important contributors to the correlation between clay and Smax. Similarly, the indirect effects of both Alox (r = 0.36) and Feox (0.23) contributed greatly to the correlation between OC and Smax. Aluminum and Fe oxides exist in soil as discrete crystals, coatings on clay and humic substances and as mixed gels. They play an important role in adsorption in soil because of their high specific surface areas and reactivity (Sparks, 2003). The ammonium oxalate solution extracts amorphous Al and Fe oxides, which are the most reactive oxides in soil because of their size and consistently high surface areas (Loeppert and Inskeep, 1996). Therefore, it was not unexpected that there would be a significant intercorrelation between clay and Alox and also between clay and Feox. Our results suggested that the correlation between OC and P adsorption was indirect and represented P adsorbed by Al and Fe associated with organic matter. This concept is supported by a study by Borggaard et al. (1990) who found that P adsorption was not influenced by the removal of organic matter with H2O2.

Stepwise multiple regression identified a two-term model based on Alox and Feox that explained 91% of the variation in Smax (Table 4). The multiple stepwise regression agreed well with path analysis and identified the same two terms whose direct effects were identified as significant by path analysis. Our results agree with those of Börling et al. (2001) who used a multiple regression model and reported that the combination of Alox and Feox were the two most important soil properties related to P adsorption in soil.


View this table:
[in this window]
[in a new window]
  		Table 4. Multiple regression formulae describing the relationship between soil properties and P sorption maxima (Smax) for 28 Oklahoma benchmark soils.

 
Several researchers have suggested that acid ammonium oxalate extractions are not appropriate for calcareous soils where calcium dominates P sorption reactions because oxalate is precipitated as a calcium salt and oxalic acid reacts with calcium carbonate to change the pH of the buffer solution (Loeppert and Inskeep, 1996; Schoumans, 2000; Kleinman and Sharpley, 2002). Therefore, path analysis and multiple regression were conducted after the removal of six soils with pH >7.0 from the data set. Removal of the high-pH soils did not significantly change the results of the path analysis or the multiple regressions. After the removal of the high-pH soils, the simple correlations between Smax and soil properties were: clay content (r = 0.83), OC (r = 0.77), Alox (r = 0.87), and Feox (r = 0.86) (data not shown). Our results differ from those of Kleinman and Sharpley (2002) who evaluated 62 soils (37 acidic soils and 25 alkaline soils) and separated them into acidic and alkaline groups before doing correlation analysis. They reported highly significant relationships existed between Alox and P adsorption and between Feox and P adsorption for acidic soils. They did not report the correlations for the combined soils; but examination of their data set shows that significant relationships existed between Alox and P adsorption and between Feox and P adsorption which were greatly improved by the removal of the alkaline soils. Unlike their study, removal of the soils with pH >7 did not improve the relationships for our study. The differences between the two studies were most likely due to the number of alkaline soils. Approximately 21% of our soils were alkaline while approximately 40% the soils examined by Kleinman and Sharpley (2002) were alkaline. Additionally, only one of our study soils was pH >8 yet approximately 23% of their soils were pH >8 where Ca would dominate P sorption reactions (Lindsay, 1979). Although, a direct measurement of carbonates was not performed in either of the studies, CaM3 was measured in both studies. Their soils contained an average of approximately 9000 mg kg–1 CaM3 as compared with a mean of 1950 mg kg–1 CaM3 in our study soils (Table 2) suggesting their soils contained much greater concentrations of carbonates than our study soils. Additionally, M3 extractable Ca in our soils ranged from 94 to 7450 mg Ca kg–1 soil while CaM3 in their soils ranged from 452 to 33400 mg Ca kg–1 soil. Therefore, the difference between the two studies was most likely the result of the fact that their soils contained greater concentrations of carbonates when compared with our study soils.

Path Analysis and Multiple Regression—Mehlich-3
Results for the M3 path analysis model are summarized in Table 3. Simple correlation coefficients (r) between pH, clay content, OC, AlM3, FeM3, CaM3, MgM3, and Smax are presented for comparison with path analysis results. Similar to the ammonium oxalate model, U was low (0.32) and R2 was high (0.90), indicating the model constructed in this study explained most of the variation in P adsorption by soil. Significant correlations (p < 0.01) existed between Smax and clay and between Smax and OC as previously noted (Table 3). A significant correlation (p > 0.05) was not found between Smax and soil pH as previously noted. Significant relationships (p < 0.01) were also found between AlM3 (r = 0.54), FeM3 (r = 0.54), CaM3 (r = 0.54), MgM3 (r = 0.50), and Smax. Our results are similar to those found by Tran et al. (1990) who reported a significant relationship (r = 0.77) between AlM3 and P adsorption for 82 Quebec soils. However, they did not report the relationship between FeM3 or CaM3 and P adsorption in their study.

The M3 path analysis model found different direct and indirect effects from those of the acid ammonium oxalate path analysis model (Table 3). In contrast to the acid ammonium oxalate path analysis model, the M3 model showed significant direct effects by clay (r = 0.72) and OC (r = 0.27) on Smax. A significant direct effect of AlM3 on Smax also existed (r = 0.32), while a nonsignificant direct effects (p > 0.05) were found between FeM3, CaM3, MgM3, and Smax. Mehlich-3 extracted only about 8% of the Fe extracted by the oxalate method. This has also been shown by several other researchers (Maguire and Sims, 2002; Sims et al., 2002). Perhaps the low extraction efficiency of M3 affected the Fe inputs for the M3 path analysis model and changed the outputs for the direct effects of the model, thus allowing discrepancies to exist among the models. The existence of the significant direct effect between AlM3 and Smax suggested that the M3 extractant was similar to the acid ammonium oxalate extractant in that it extracted Al from the same non-crystalline pool in soil. Indeed other researchers have reported significant relationships between Alox and AlM3 but not between Feox and FeM3 (Tran et al., 1990; Fernandez Marcos et al., 1998). Path analysis revealed that the indirect effect of clay was an important contributor to the relationships between CaM3 and Smax and between MgM3 and Smax.

Stepwise multiple regression agreed well with path analysis. Stepwise multiple regression revealed the best model was a combination of the three same terms (clay, OC, and AlM3) whose direct effects were found significant by path analysis. The combination of these terms explained 89% of the variation in Smax.

Again, the soils with pH > 7.0 were removed and path analysis and multiple regression were conducted. Removal of the high-pH soils significantly changed the simple correlations found with path analysis by improving the relationships between AlM3 (r = 0.73), FeM3 (r = 0.67), CaM3 (r = 0.70), and Smax (data not shown). Our results agree somewhat with those of Kleinman and Sharpley (2002) who reported highly significant and improved relationships between AlM3 and Smax in acidic soils, when they separated soils into acidic and alkaline groups for correlation analysis. However, Kleinman and Sharpley (2002) reported a nonsignificant relationship between FeM3 and Smax.

Removal of the high-pH soils did not change the results of the path analysis or the multiple regression for the M3 model. Clay, OC, and AlM3 were still the only significant direct effects and were also identified by multiple regression as the three most important soil properties related to P adsorption in soil.

Correlations between Phosphorus Saturation Indices
The P saturation index (PSI) is the amount of extractable P (mmol kg–1) divided by the sum of the extractable Al and Fe (mmol kg–1). This saturation index was estimated using both acid ammonium oxalate (PSIox) and M3 (PSIM3) as extractants. The two different PSIs produced different values. The PSIox ranged from 2.8 to 38% and PSIM3 ranged from 1.6 to 26% (Table 2). Median values were 9.7% for PSIox and 3.9% for PSIM3. Additionally, a third saturation index (Psat) based on M3 P and Smax was calculated for the study soils. The Psat value ranged from 3 to 78% with a median of 15%.

Phosphorus saturation index using acid ammonium oxalate was highly correlated (p < 0.01, r = 0.87) with PSIM3 for all 28 study soils (Fig. 3) . Removal of soil samples with pH > 7.0 did not improve the correlation between PSIox and PSIM3 (data not shown). Our results are consistent with other researchers who have reported highly significant relationships between PSIox and PSIM3 (Kleinman and Sharpley, 2002; Maguire and Sims, 2002; Sims et al., 2002). Phosphorus saturation was not significantly related (p > 0.05) to PSIox when all 28 soils were included (Fig. 4A) . However, removal of soils with pH > 7.0, produced a highly significant relationship (p < 0.01, r = 0.79) between Psat and PSIox (Fig. 4B). A nonsignificant relationship (p > 0.05) was also observed between Psat and PSIM3 (Fig. 4C), but removal of soils with pH > 7.0 resulted in a highly significant correlation between Psat and PSIM3 (p < 0.01,. r = 0.85) (Fig. 4D). Close examination of Fig. 4A through 4D showed that two outliers (the Lebron and Mansic soils) which had a high pH (approximately 8), high PSIox, high PSIM3, and a low Psat greatly influenced the regression. These outliers appear in the lower right corners of Fig. 4A and Fig. 4C and their removal considerably improved the relationships between Psat and PSIox and between Psat and PSIM3. The Lebron and Mansic soils had much greater concentrations of CaM3 as compared with the other study soils (Table 2) suggesting that the soils contained greater concentrations of carbonates as compared with the other study soils which may have interfered with the M3 and oxalate extractions of Al and Fe and influenced the regression analysis.



View larger version (13K):
[in this window]
[in a new window]
  		Fig. 3. Relationship between P saturation indexes using acid ammonium oxalate (PSIox) and Mehlich-3 (PSIM3) as extractants for 28 Oklahoma benchmark soils. **p < 0.01.

 


View larger version (24K):
[in this window]
[in a new window]
  		Fig. 4. Relationships between a P saturation index calculated from Mehlich-3 extractable P and Smax (Psat), and (A) a P saturation index calculated from acid ammonium oxalate extractable data (PSIox) for all 28 soils; (B) PSIox for soils with pH < 7.0; (C) a P saturation index calculated from Mehlich-3 extractable data (PSIM3) for all 28 soils; or (D) PSIM3 for soils with pH < 7.0. **p < 0.01.

 
The degree of P saturation estimates how close the soil is to saturation and is viewed as an environmental indicator of soil P based on the fact that more P is released from soil to runoff or leaching as the degree of P saturation increases (Sharpley, 1995; Kleinman and Sharpley, 2002). Considerable research has been conducted on the use of PSIox as an environmental soil test (Hooda et al., 2000; Pautler and Sims, 2000; Maguire et al., 2001, Maguire and Sims, 2002, Sims et al., 2002). Our results indicate that other measurements of P saturation are well correlated with PSIox and may serve useful in identifying soils with increased risk for P loss.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Understanding the P sorption capacity of a soil can help to estimate the amount of P that a soil is capable of holding. Soil properties (clay content, OC content, Alox, and Feox) were highly correlated with Smax. Path analysis indicated that the direct effects for clay and OC were not significant and that these relationships were highly influenced by the indirect effects of Alox and Feox. Multiple regression also found that the combination of Alox and Feox were the two most important soil properties related to P sorption in soil. Therefore, Alox and Feox were more important soil properties for the direct estimation of P sorption than clay and OC. It also appeared that pH did not affect the relationship between Alox and Smax or between Feox and Smax. Thus, P sorption in Oklahoma soils may be estimated using the more economical and time saving oxalate extraction instead of time-consuming adsorption isotherms.

Aluminum, Fe, and Ca extracted by M3 were better correlated with Smax after the removal of the high-pH soils indicating pH affected the relationships. Thus, M3 extractions of Al, Fe, and Ca may be used to estimate P adsorption in Oklahoma soils. Mehlich-3 is a very common method used by many laboratories and represents an alternative to the ammonium oxalate method. Many testing facilities use ICP–AES for their analyses and including Al, Fe, and Ca analyses on M3 extracts would be a relatively easy extra step for many soil-testing laboratories.

Phosphorus saturation is increasingly viewed as an environmental indicator of soil P because it has been found to be a good indicator of P availability to runoff and leachate (Kleinman and Sharpley, 2002). Conversely, P saturation is usually estimated from data not readily available through testing laboratories. Our results show that PSIox was highly correlated with PSIM3 for all 28 soils. Furthermore, Psat was highly correlated with PSIox and with PSIM3 when two outliers containing large amounts of M3 extractable Ca were removed from the data set. Several researchers have shown that PSIox is related to P in surface runoff and to leachable P. The results of our study indicate that two other measurements of P saturation (PSIM3 and Psat) were well correlated with PSIox and may serve useful in identifying soils with increased risk for P loss. In particular, the correlation between PSIox and PSIM3 showed the potential use of the M3 extractant for estimating P saturation in soils, which provides useful information about the risk of runoff and leaching P in soils. Therefore, over-application of P can be limited.


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




This article has been cited by other articles:


Home page
 J. Environ. Qual.Home page
K. W. Goyne, H.-J. Jun, S. H. Anderson, and P. P. Motavalli
Phosphorus and Nitrogen Sorption to Soils in the Presence of Poultry Litter-Derived Dissolved Organic Matter
J. Environ. Qual., January 4, 2008; 37(1): 154 - 163.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
B. E. Haggard, D. R. Smith, and K. R. Brye
Variations in Stream Water and Sediment Phosphorus among Select Ozark Catchments
J. Environ. Qual., November 1, 2007; 36(6): 1725 - 1734.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
Y. Tang, H. Zhang, J. L. Schroder, M. E. Payton, and D. Zhou
Animal Manure Reduces Aluminum Toxicity in an Acid Soil
Soil Sci. Soc. Am. J., September 28, 2007; 71(6): 1699 - 1707.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
C. H. Bolster and G. M. Hornberger
On the Use of Linearized Langmuir Equations
Soil Sci. Soc. Am. J., September 28, 2007; 71(6): 1796 - 1806.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
C. J. Penn, G. L. Mullins, L. W. Zelazny, and A. N. Sharpley
Estimating Dissolved Phosphorus Concentrations in Runoff from Three Physiographic Regions of Virginia
Soil Sci. Soc. Am. J., September 20, 2006; 70(6): 1967 - 1974.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
R. L. Davis, H. Zhang, J. L. Schroder, J. J. Wang, M. E. Payton, and A. Zazulak
Soil Characteristics and Phosphorus Level Effect on Phosphorus Loss in Runoff
J. Environ. Qual., August 9, 2005; 34(5): 1640 - 1650.
[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 (13)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Zhang, H.
Right arrow Articles by Payton, M. E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Zhang, H.
Right arrow Articles by Payton, M. E.
Agricola
Right arrow Articles by Zhang, H.
Right arrow Articles by Payton, M. E.
Related Collections
Right arrow Nutrient Management
Right arrow Phosphorus
Right arrow Soil Chemistry

HOME HELP FEEDBACK