JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 1 March 2006
Published in J Environ Qual 35:599-610 (2006)
DOI: 10.2134/jeq2005.0135
© 2006 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
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 Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (5)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Allen, B. L.
Right arrow Articles by Camara, M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Allen, B. L.
Right arrow Articles by Camara, M.
Agricola
Right arrow Articles by Allen, B. L.
Right arrow Articles by Camara, M.
Related Collections
Right arrow Surface Water Quality
Right arrow Nutrients
Right arrow Nutrient Management
Right arrow Phosphorus

TECHNICAL REPORTS

Surface Water Quality

Soil and Surface Runoff Phosphorus Relationships for Five Typical USA Midwest Soils

B. L. Allen, A. P. Mallarino*, J. G. Klatt, J. L. Baker and M. Camara

Department of Agronomy, Iowa State University, Ames, IA 50011

* Corresponding author (apmallar{at}iastate.edu)

Received for publication April 25, 2005.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 REFERENCES
 
Excessively high soil P can increase P loss with surface runoff. This study used indoor rainfall simulations to characterize soil and runoff P relationships for five Midwest soils (Argiudoll, Calciaquaoll, Hapludalf, and two Hapludolls). Topsoil (15-cm depth, 241–289 g clay kg–1 and pH 6.0–8.0) was incubated with five NH4H2PO4 rates (0–600 mg P kg–1) for 30 d. Total soil P (TPS) and soil-test P (STP) measured with Bray-P1 (BP), Mehlich-3 (M3P), Olsen (OP), Fe-oxide-impregnated paper (FeP), and water (WP) tests were 370 to 1360, 3 to 530, 10 to 675, 4 to 640, 7 to 507, and 2 to 568 mg P kg–1, respectively. Degree of soil P saturation (DPS) was estimated by indices based on P sorption index (PSI) and STP (DPSSTP) and P, Fe, and Al extracted by ammonium oxalate (DPSox) or Mehlich-3 (DPSM3). Soil was packed to 1.1 g cm–3 bulk density in triplicate boxes set at 4% slope. Surface runoff was collected during 75 min of 6.5 cm h–1 rain. Runoff bioavailable P (BAP) and dissolved reactive P (DRP) increased linearly with increased P rate, STP, DPSox, and DPSM3 but curvilinearly with DPSSTP. Correlations between DRP or BAP and soil tests or saturation indices across soils were greatest (r ≥ 0.95) for FeP, OP, and WP and poorest for BP and TPS (r = 0.83–0.88). Excluding the calcareous soil (Calciaquoll) significantly improved correlations only for BP. Differences in relationships between runoff P and the soil tests were small or nonexistent among the noncalcareous soils. Routine soil P tests can estimate relationships between runoff P concentration and P application or soil P, although estimates would be improved by separate calibrations for calcareous and noncalcareous soils.

Abbreviations: BAP, bioavailable P • BP, Bray-P1 • DPSM3, degree of P saturation based on Mehlich-3 P, Fe, and Al • DPSox, degree of P saturation based on ammonium-oxalate P, Fe, and Al • DPSSTP, degree of P saturation based on soil-test P and a P sorption index • DRP, dissolved reactive P • FeP, P extracted with Fe-oxide-impregnated paper • M3P, Mehlich-3 P • OP, Olsen P • PSI, P sorption index • STP, soil-test P • TPR, total runoff P • TPS, total soil P • WP, water-extractable P


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 REFERENCES
 
PHOSPHORUS FERTILIZATION is often necessary to increase STP for optimal crop yield or to maintain STP at optimal levels. Excessive or inappropriate P application, however, has led to excessive P loss from agricultural fields and impairment of surface water resources through eutrophication. Above-optimum STP levels for crops in many fields of the USA and concentration of confined animal production operations may further increase manure application to soils in some areas (Whalen and Chang, 2001). The P delivery rate to water bodies is influenced by various source and transport factors including P application source, rate and method, soil P levels, field slope, soil erosion, surface runoff, subsurface drainage, and proximity to surface waters. The P index is an assessment tool developed in Iowa and other states to estimate the relative contribution of various source and transport factors to assign a risk rating to a field or management area (Lemunyon and Gilbert, 1993; Mallarino et al., 2002).

The soil P concentration is an important P source factor. Several routine P tests, such as BP, M3P, and OP are used to predict P sufficiency for crops. These tests are designed to extract a P fraction that correlates well with plant uptake, but not necessarily with P loss potential or algae growth in surface freshwater bodies. For example, acidic extractants such as those used by BP or M3P methods would likely dissolve Ca phosphates of low water solubility that may not be readily available for algal uptake (Self-Davis et al., 2000) and in many CaCO3-affected soils extract less P than other P tests because of reactions of the extracting solution with CaCO3 (Atia and Mallarino, 2002). Alternative tests have been proposed that may better estimate the soil P available to algae in aquatic environments. For example, WP has been extensively used in recent years to study soil P effects on DRP concentrations in runoff (Pote et al., 1996). Other methods use a sink approach to estimate soil and runoff P availability to algae, such as FeP that is often referred to as bioavailable P (Sharpley, 1993; Menon et al., 1997). Research results have shown, however, that routine agronomic tests and environmental tests usually are highly correlated (Atia and Mallarino, 2002; Kleinman and Sharpley, 2002; Maguire and Sims, 2002).

Soil P that could be desorbed from soil particles and lost from fields with surface runoff or water flow through the soil profile also can be estimated with P sorption–desorption isotherm determinations (Nair et al., 1984); however, these methods are time consuming and cannot be practically implemented as routine tests. Alternatively, a PSI derived from a single-point isotherm has been shown to be well correlated with P sorption isotherms (Bache and Williams, 1971). Researchers have used STP/PSI or STP/(PSI + STP) indices to estimate degree of P saturation in soil (Pote et al., 1999; Pautler and Sims, 2000; Westermann et al., 2001). Saturation indices that include PSI as a component could have a stronger theoretical basis for calcareous soils than other indices that are typically based on extractable Fe and Al. Because soil Fe and Al oxides play a major role in P sorption in many soils, indices of soil P saturation were developed based on P/(Fe + Al) molar ratios from an ammonium oxalate extraction (DPSox; Schoumans and Groenendijk, 2000) or a Mehlich-3 extraction (DPSM3; Khiari et al., 2000; Maguire and Sims, 2002). The DPSM3 is a particularly appealing index because the Mehlich-3 extractant is widely used for P and other elements and DPSM3 often correlates well with DPSox (Maguire and Sims, 2002; Sims et al., 2002). Morel et al. (2000) suggested that rapid soil P saturation indices can be more useful than STP across a range of soils with contrasting chemical and mineralogical properties. In their study, soil P saturation accounted for 94% of the variability in desorbed P; however, others have found that saturation indices and common routine soil P tests sometimes provide similar estimates of risk of dissolved P loss from fields (Kleinman et al., 1999; Pote et al., 1999; Pautler and Sims, 2000; Schoumans and Groenendijk, 2000). Furthermore, routine soil P tests provide useful estimates of total or bioavailable P concentrations in runoff from fields (Sims et al., 2002; Klatt et al., 2003). Therefore, use of locally calibrated, routine soil tests as environmental risk indicators has significant practical advantages because soil test data are widely available and their agronomic interpretations are well established (Sims et al., 2002).

Studies have shown that dissolved or bioavailable P in runoff increases with increased soil P measured by various methods, although relationships vary greatly across sites (Pote et al., 1996; Sharpley et al., 1996; Pote et al., 1999; Bundy et al., 2001). This variation occurs due to differences in management practices and variability in site-specific soil properties that affect P sorption–desorption and site hydrology. High water solubility of P sources and practices that lead to accumulation of P at or near the soil surface increase the risk of dissolved P loss. Also, the equilibrium between soluble, adsorbed, and chemically bound soil P varies greatly among soils with different mineralogical and chemical properties. Soil properties that result in low P adsorption (e.g., coarse texture, low concentration of Fe and Al oxides, and high P concentration) increase the potential for dissolved P loss (Sharpley et al., 1996; Kleinman et al., 1999; Pote et al., 1999; Morel et al., 2000; Pautler and Sims, 2000). These studies (most based on indoor or field simulated rainfall) showed that the DRP concentration usually increases linearly as STP increases but sometimes (especially at very high levels) DRP can increase curvilinearly or there is a change point after which the rate of increase is greater. Reported linear DRP increases range from 0.001 to 0.03 mg L–1 for every 1 mg kg–1 increase of BP or M3P, although DRP concentrations >2.0 mg L–1 have been shown for runoff events shortly after applying P fertilizer or manure.

Concerns related to surface water quality and excess P inputs to soils warrant further study of soil P levels and fractions and their relationships to the most bioavailable P fractions in surface runoff for typical soils of the U.S. Midwest. Of particular interest are relationships between TPS, P measured with routine or environmental soil P tests, and indices of soil P saturation, as well as how these soil P levels relate to runoff P concentrations shortly after fertilizers are applied and mixed with soil. A better understanding of these relationships also is necessary to optimize the effectiveness of comprehensive P assessment tools such as the P index, which use soil-test P to determine runoff P coefficients. Therefore, the objective of this study was to characterize soil P and surface runoff P relationships, for five typical soils of Iowa and several midwestern states having wide ranges of soil P levels, under similar conditions using an indoor rainfall simulation technique.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 REFERENCES
 
Soil Series
Five soils representative of large agricultural areas of Iowa and some neighboring states were selected for this indoor rainfall simulation study. Typical profiles were described in detail by Soil Survey Staff (USDA-NRCS, 2004). The Marshall series (fine-silty, mixed, superactive, mesic Typic Hapludolls) consists of well-drained, moderately permeable soils formed in loess on uplands and high stream benches, slope ranges from 0 to 20%, and is found in Iowa, Kansas, Missouri, and Nebraska. The Nicollet series (fine-loamy, mixed, superactive, mesic Aquic Hapludolls) consists of somewhat poorly drained, moderately permeable soils formed in calcareous loamy glacial till on plains and glacial moraines, slope ranges from 0 to 5%, and is found in Iowa and Minnesota. The Fayette series (fine-silty, mixed, superactive, mesic Typic Hapludalfs) consists of well-drained, moderately permeable soils formed in loess on uplands and high stream benches, slope ranges from 0 to 40%, and is found in Iowa, Illinois, Minnesota, and Wisconsin. The Tama series (fine-silty, mixed, superactive, mesic Typic Argiudolls) consists of well- to moderately well-drained, moderately permeable soils formed in loess on uplands and high stream benches, slope ranges from 0 to 20%, and is found in Iowa, Illinois, Minnesota, and Wisconsin. The Harps series (fine-loamy, mixed, superactive, mesic Typic Calciaquaolls) consists of poorly drained, moderately permeable calcareous soils formed in glacial till or alluvium on uplands, slope ranges from 0 to 3%, and is found in Iowa and Minnesota.

Soil Incubation, Sampling, and Analyses
Approximately 1000 kg of topsoil (15-cm depth) were collected from Iowa fields selected to have STP less than very high for crop production according to Iowa interpretations (Sawyer et al., 2002) and at least 10-yr histories of row crops and chisel-plow and disk tillage. Soil was collected from fields in Boone County (Nicollet soil), Cass County (Marshall soil), Dubuque County (Fayette soil), Marshalltown County (Tama soil), and Hancock County (Harps soil). Table 1 provides selected soil properties. Soil was sieved through a 1.3-cm screen and the moisture content was determined. Three batches of each soil (which would represent three treatment replications) were incubated separately with NH4H2PO4 fertilizer (monoammonium phosphate) rates of 0, 50, 125, 300, and 600 mg P kg–1 (on a 40°C oven-dry soil basis). Soils that received P were placed in a cement mixer and sprayed with an appropriate amount of a solution of fertilizer and water until the final moisture content approximated field moisture capacity for each soil. The soils were incubated in plastic containers indoors at ambient temperature of 22 to 26°C for 30 d. Although the 30-d incubation period may not be enough time for the fertilizer P to completely equilibrate with the soil, it represents well the actual time between fertilizer application and runoff for Iowa farms. The P applications and incubations were spaced over time so that the incubation period matched the pace of subsequent rainfall simulation work.


View this table:
[in this window]
[in a new window]
  		Table 1. Selected soil chemical and physical properties.

 
Approximately 1 kg of incubated soil was sampled for analyses immediately before each rainfall event, dried at 40°C, and crushed to pass through a 2-mm sieve. These samples were used for all soil tests except for TPS and organic C, for which subsamples were further crushed to pass a 0.5-mm screen. All tests were performed in duplicate. Soils were analyzed for BP, M3P, and OP following procedures recommended for the North-Central Region of the USA (Frank et al., 1998) using a colorimetric determination of extracted P based on the method of Murphy and Riley (1962). Total soil P was determined with an alkaline oxidation procedure (Dick and Tabatabai, 1977) adapted to an aluminum digestion block (Cihacek and Lizotte, 1990). Soil also was tested for WP with the procedure describe by Pote et al. (1996) by shaking 1 g of soil with 25 mL of deionized water for 1 h, centrifuging for 5 min at 266 m s–1 (27 100 x g), and filtering through Whatman no. 42 filter paper. Soil was analyzed for FeP following a procedure described by Chardon (2000). Briefly, Fe-oxide-impregnated filter paper disks (5.5-cm diameter, Whatman no. 50) were immersed in a solution of 10 g FeCl3·6H2O in 100 mL deionized water, removed from the solution, air dried, immersed in a 2.7 M NH4OH solution, and air dried again. Soil P was extracted by placing a paper disk into a bottle having 1 g soil and 30 mL of 0.01 M CaCl2 and shaking for 16 h. Phosphorus sorbed was determined by rinsing the disk with deionized water to detach any soil particles, air drying the disk, and removing P from the disk by shaking it for 1 h in 30 mL of 0.1 M H2SO4. The PSI (in mg kg–1) was determined using the single-point P sorption procedure suggested by Sims (2000). Briefly, 20 mL of a 75 mg P L–1 solution was added to 1 g of soil, shaken for 18 h, and filtered through a 0.45-µm pore-size filter. Phosphorus in all extracts was determined colorimetrically by the method by Murphy and Riley (1962).

Soil P saturation was estimated by DPSSTP, DPSM3, and DPSox indices. Extractable soil Fe and Al were measured by atomic absorption spectroscopy. The DPSox and DPSM3 indices were based on molar ratios (mmol kg–1) as described in Eq. [1] and [2], based on Khiari et al. (2000) and modified as suggested by Beck et al. (2004) by omitting a saturation factor ({alpha}m). Because of lack of consensus about how to calculate and use {alpha}m and because most studies in the past have used a value of 0.5 as a scaling factor, Beck et al. (2004) recommended omitting {alpha}m to facilitate interpretation and comparison of DPS across studies. The DPSSTP index was calculated (Eq. [3]) following a procedure used by Pautler and Sims (2000) using STP measured with BP, M3P, OP, WP, or FeP. Although the three saturation indices in this study are expressed as percentage P saturation, they are only indices and no actual P saturation is implied.

Formula 1[1]

Formula 2[2]

Formula 3[3]

Soil samples from the control treatment were also analyzed for soil characterization purposes. Total C was determined by a combustion method based on the procedure of Wang and Anderson (1998). Soil texture was determined by particle size distribution (Walter et al., 1978). Calcium carbonate was determined using the pressure-calcimeter method (Sherrod et al., 2002). Soil specific surface area was estimated with the method described by Cihacek and Bremner (1979). Soil pH was determined using a 1:1 soil/water ratio. Ammonium-acetate-extractable K, Ca, and Mg were determined as recommended for the north-central region (Warncke and Brown, 1998).

Rainfall Simulations and Surface Runoff Analysis
Plastic boxes measuring 82.5 cm in length, 42.3 cm in width, and 15 cm in depth were used for indoor rainfall simulations. The bottoms of the plastic boxes were first layered with 3.8 cm of silica sand, cheesecloth, and three successive layers of an equal mass of soil compacted to an approximate bulk density of 1.1 g cm–3 to a total soil thickness of 7.6 cm. A 0.95-cm-diameter hole and drain tube in the bottom of each box allowed subsurface drainage. Plastic tubes were attached to each box at the soil surface level (11.4 cm from bottom of the box) to collect runoff into bottles placed on a balance connected to a data logger. The boxes were set at a 4% slope and positioned 3.05 m below the rainfall simulator nozzles. The rainfall simulator design was described by Baker et al. (1997). The nozzles (Vee-Jet H1/2 U 316 SS, Spraying Systems, Wheaton, IL) were arranged in a rectangular grid (110-cm spacing along one direction and 77 cm along the other), and swept back and forth in a 90° arc at 1 oscillation s–1. The simulations were conducted at room temperature using tap water (DRP was <0.01 mg P L–1) at an intensity of 6.5 cm h–1, which represents a rainfall recurrence interval of 25 yr for Iowa (Huff and Angel, 1992). Two soil runoff boxes packed with soil from the same series received rain during each simulated rainfall event. For example, two replications of one treatment and soil received rain during the same event, and the third replication of that treatment and one replication of the next treatment received rain during the subsequent rainfall event. Runoff was collected during 75 min into a series of seven polyethylene bottles.

Bottles with runoff were shaken vigorously and a subsample was filtered (0.45-µm) for DRP analysis by the method of Murphy and Riley (1962). Subsamples of unfiltered runoff were analyzed for FeP and total P. Several authors (Sharpley, 1993; Menon et al., 1997) have referred to P extracted from unfiltered runoff with the FeP method as BAP. The BAP in runoff was determined as described for soil, except that an Fe-oxide-impregnated paper disk was shaken with 30 mL of runoff. Total runoff P (TPR) was determined with the persulfate–H2SO4 digestion method (American Public Health Association, 1998). Phosphorus in extracts was analyzed colorimetrically by the method of Murphy and Riley (1962). Total solids (TS) and total dissolved solids (TDS) also were determined following American Public Health Association (1998) methods and suspended solids were estimated as TS – TDS. All runoff analyses were performed in duplicate.

Data Management and Statistical Analyses
Flow-weighted mean P concentrations of the seven runoff samples for each soil series, treatment, and replication were used for this study. The mass of runoff P is not shown because an inherent limitation of boxed-soil studies is that some hydrologic characteristics associated with natural soil structure are lost and flow volumes are highly affected by the type and uniformity of soil packing in each tray, which make extrapolations of flow or P mass to field conditions quite unreliable. Average flow volumes for the Nicollet, Marshall, Fayette, Tama, and Harps soils (relative to a rainfall volume of 28.4 L) were 18.9, 21.2, 20.3, 19.6, and 17.6 L, respectively, with a standard deviation of 2.7 L across all soils and simulated rainfall events. Means of duplicate samples were used for correlation analysis and linear or quadratic regression analysis using SAS (SAS Institute, 2000). Quadratic equations are shown only when the quadratic term is significant (P ≤ 0.05) after the linear term. Linear coefficients (slopes) of relationships between measurements were compared across trials using the GLM procedure of SAS combined with a Bonferroni test (SAS Institute, 2000). Treatment means and regression lines for each soil series are shown in all figures.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
SUMMARY AND CONCLUSIONS
 REFERENCES
 
Characterization of Untreated Soil
Soil texture ranged from loam to silty clay loam across soils (Table 1). Specific surface area was lowest for the Nicollet and Fayette soils, intermediate for the Marshall and Tama soils, and considerably greater for the Harps soil. Total C was lowest for the Fayette soil; intermediate for the Marshall, Nicollet, and Tama soils; and greatest for the Harps soil. Soil pH was slightly acidic for the Nicollet and Fayette soils, near neutral for the Marshall and Tama soils, and alkaline for the Harps soil. Calcium carbonate was high for the Harps soil (67 g kg–1) and very low for the other soils (2–7 g kg–1). Concentration of oxalate-extractable Al (Alox) varied little across the soils, but oxalate-extractable Fe (Feox) was much greater for the Marshall, Tama, and Fayette soils than for the Nicollet and Harps soils. These differences may reflect variations in soil mineralogy or weathering. Concentrations of Mehlich-3-extractable Al and Fe (AlM3 and FeM3, respectively) were much lower than Feox and Alox (especially FeM3) for all the soils except the Harps soil, and relative AlM3 differences were similar to Alox. However, FeM3 for the Nicollet soil was similar to that for the Marshall, Tama, and Fayette soils and much less Fe was extracted by Mehlich-3 from the calcareous Harps soil. Phosphorus sorption capacity estimated by PSI was greatest for the Tama soil, intermediate for the Fayette, Marshall, and Harps soils, and lowest for the Nicollet soil. Total soil P of the untreated soils ranged from 374 to 689 mg P kg–1 and was lowest for the Fayette and Nicollet soils, intermediate for the Marshall and Tama soils, and greatest for the Harps soil. These TPS differences probably resulted from a combination of soil formation and management factors because all soils have been under production agriculture for at least six decades.

Relationships between Phosphorus Application and Soil Phosphorus
Soil analyses at the end of the 30-d incubation period showed that P application increased soil P measured by all methods in all soils (Fig. 1 and 2). Across all soils, soil P ranged from 3 to 530 mg kg–1 for BP, 4 to 640 mg kg–1 for OP, 10 to 675 mg kg–1 for M3P, 7 to 507 mg kg–1 for FeP, 2 to 568 mg kg–1 for WP, and 370 to 1360 mg kg–1 for TPS. Total soil P increased linearly (P < 0.05) as P application increased for all soils, which was an expected result because P rate increments were similar. Soil P extracted by other P tests increased linearly as P application increased with few exceptions. Very weak curvilinear increasing trends (greater rate of increase at high P application rates) were observed for all tests in the Fayette soil, for OP in the Harps soil, and for WP in the Harps, Nicollet, and Tama soils. However, R2 values for curvilinear trends were only 0.01 larger than for linear trends, except for M3P in the Fayette soil (0.03 larger) and, therefore, can be disregarded. Linear coefficients (mg kg–1 soil P increase per kg ha–1 P applied, assuming 2200 Mg soil ha–1) ranged from 0.9 to 1.1 for TPS, 0.3 to 0.9 for WP, 0.5 to 0.8 for FeP, 0.6 to 1.0 for M3P, 0.6 to 0.8 for BP, and 0.4 to 1.1 for OP. Rates of soil P increase differed only for the calcareous Harps soil compared with other soils and only for some tests. Linear coefficients of trends for FeP, WP, and OP were larger for the Harps soil. Linear coefficients for BP, M3P, and TPS did not differ across soils, or differences were small and inconsistent. For example, the BP increase with increasing P application was slightly less for the Tama and Harps soils than other soils while M3P increase was slightly less for the Marshall and Tama soils than other soils.


Figure 1
View larger version (30K):
[in this window]
[in a new window]
  		Fig. 1. Effect of P application rate on soil P measured by three routine soil tests. Curvilinear trends are shown only when the quadratic term was significant (P < 0.05) after a linear term. Different letters by the soil names indicate significantly different linear coefficients between soils for each soil test.

 

Figure 2
View larger version (31K):
[in this window]
[in a new window]
  		Fig. 2. Effect of P application rate on soil total P and P measured by two environmental soil tests. Curvilinear trends are shown only when the quadratic term was significant (P < 0.05) after a linear term. Different letters by the soil names indicate significantly different linear coefficients between soils for each soil test.

 
Correlation analyses across all soils between P rate and each soil test (Table 2) indicated high correlations for BP, M3P, and FeP (0.96–0.97), intermediate for OP and TPS (0.88 for both), and low for WP (0.80). Excluding the calcareous Harps soil resulted in higher and approximately similar correlations between P rate and all tests (0.95–0.99). Correlation coefficients among P tests across all soils were clearly lower for some relationships involving BP (mainly with OP, WP, and TPS; r = 0.72–0.83) and excluding the calcareous Harps soil improved correlations substantially (0.94–0.99). These results for BP, and also significantly lower rate of BP increase with increased P rate for the Harps soil, may be explained by the relatively lower P extraction by BP from this calcareous soil. A partial neutralization of the acidic and poorly buffered BP extractant by CaCO3 has been shown before for Iowa calcareous soils (Mallarino, 1997). In fact, only the OP and M3P routine tests are recommended in Iowa (Sawyer et al., 2002) and neighboring states for regions having slightly acid to CaCO3-affected soils. Other factors could have contributed to the relationship observed. For example, the Harps soil had the greatest organic matter concentration and the lowest extractable Fe and Al of all soils (Table 1). At greater concentrations, organic matter can decrease P sorption by coating calcite particles, leading to more P in available pools as measured by some tests (Huang and Schnitzer, 1986; Robbins et al., 1999).


View this table:
[in this window]
[in a new window]
  		Table 2. Correlation between P application rate and soil P measured by five tests for all soils (n = 75) and excluding the calcareous Harps soil (NonC).

 
Calculations from data in Fig. 1 showed that P application increased the fraction of TPS measured by the other tests for all soils. When no P was applied, the STP/TPS ratio across soils ranged from 0.6 to 4.6% for BP, 0.9 to 2.8% for OP, 2.7 to 5.2% for M3P, 1.5 to 3.7% for FeP, and 0.5 to 1.1% for WP. When P was applied at the highest rate, however, the STP/TPS ratio ranged from 28 to 50% for BP, 22 to 49% for OP, 36 to 63% for M3P, 29 to 38% for FeP, and 16 to 41% for WP. Sharpley et al. (1984) also reported that the proportion of TPS as BP increased from 4 to 29% in the surface 30 cm of a clay loam Texas soil that received various rates of cattle manure during a 5-yr period. Whalen and Chang (2001) reported that the proportion of TPS as OP increased from 13 to 27% in the surface 15 cm in a clay loam soil that received various rates of cattle manure during a 16-yr period in Alberta (Canada).

Increased application rates increased soil P saturation when evaluated by DPSSTP, DPSox, and DPSM3. Figure 3 shows that trends for DPSox and DPSM3 were linear (P < 0.05) for all soils except for the Fayette soil, in which a slight curvilinear trend was observed. Tests of differences of linear regression coefficients across soils for the two indices showed a greater rate of saturation increase for the Harps soil than other soils. This result is explained mainly by much lower extractable Fe in the calcareous Harps soil than in the noncalcareous soils (Table 1). Extractable Fe from the Harps soil was four (FeM3) to seven (Feox) times less than from other soils, while extracted P and Al differed by a factor of two or less. Among noncalcareous soils, DPSox linear coefficients for the Nicollet soil were slightly greater than for others, as would be expected from its lower PSI and extractable Fe and Al and greater organic matter. The DPSM3 linear coefficient for the Nicollet soil, however, did not differ from other noncalcareous soils, but that for the Fayette soil was significantly greater than for the Marshall and Tama soils. Moreover, differences for DPSM3 trends among noncalcareous soils were similar to those shown in Fig. 1 for M3P and P rate, which agrees with approximately similar FeM3 and AlM3 across all P application rates (not shown). Therefore, results for DPSox, and DPSM3 clearly indicated high rates of soil P saturation increase with increased P application rates for Harps but much smaller and inconsistent differences among the other soils.


Figure 3
View larger version (31K):
[in this window]
[in a new window]
  		Fig. 3. Effect of P application on three soil P saturation indices: degree of P saturation based on soil test P and a P sorption index (Degree of P SaturationSTP), based on Mehlich-3 P (Degree of P SaturationM3), and based on ammonium-oxalate P (Degree of P SaturationOX). Degree of P SaturationSTP is shown only for the Olsen P soil test. Curvilinear trends are shown only when the quadratic term was significant (P < 0.05) after a linear term. Different letters by the soil names indicate significantly different linear coefficients between soils for each index.

 
In contrast to results for DPSox and DPSM3, soil P saturation estimated by DPSSTP increased curvilinearly with increasing P rates for all soils when this index was calculated using BP, M3P, OP, FeP, or WP. Moreover, the ranking of soils for soil P saturation rate of increase sometimes differed from results for DPSox, and DPSM3. Only results for DPSSTP calculated using OP are shown in Fig. 3 because the ranking of soils and trends with increased P application rate were similar when other P tests were used. The rate of change of DPSSTP was contrastingly different for three groups of soils: Nicollet > Harps, Marshall, and Fayette > Tama. No single soil property in Table 1 fully explained DPSSTP differences among soils, although differences approximately follow results for PSI and Alox.

Correlations across all soils and treatments (Table 3) indicated that the DPSSTP index was better related to P application rate than DPSox or DPSM3 and were r = 0.96, 0.77, 0.71, respectively. All coefficients were high, however, when the Harps (calcareous) soil was excluded (r = 0.93–0.97). A contrasting difference between DPSSTP and DPSox or DPSM3 in our study may be explained by different sensitivity of PSI-based indices compared with indices based on extractable Al and Fe, because soil P measured by all extractants (including ammonium oxalate) was significantly correlated (not shown) across the four noncalcareous soils (r = 0.96–0.99) and across all soils (r = 0.72–0.98). Correlations across all soils (Table 3) indicated an approximately similar degree of correlation between DPSox or DPSM3 and soil P measured by the different tests but slightly higher correlations for DPSSTP. For example, correlation coefficients were 0.62 to 0.97 for DPSM3, 0.70 to 0.93 for DPSox, and 0.80 to 0.96 for DPSSTP. The ranking of soil P tests also were approximately similar for DPSox and DPSM3. For example, BP and M3P had the poorest correlation, whereas OP and WP had the best correlation. These results indicate that the soil P extraction mechanisms of OP and WP and the pools from which these tests extract P are better related (compared with other P tests) with saturation indices based on molar ratios of extracted P, Fe, and Al. The ranking of soil P tests for DPSSTP was the opposite from those for DPSox or DPSM3. For example, BP and M3P had the highest correlation, whereas OP and WP had the poorest, even though data shown for DPSSTP correspond to calculations with PSI and OP. Because the ranking of soil P methods for DPSSTP calculated with OP were similar to those for others (not shown) we conclude that the main factor explaining the differences across soils between DPSSTP and DPSox or DPSM3 is that DPSSTP is based on PSI and not on Fe and Al. The differences in correlations between the saturation indices and soil P tests across soils were due mainly to the calcareous Harps soil. Excluding the Harps soil from the calculation greatly improved all correlations and no substantial differences among indices were observed.


View this table:
[in this window]
[in a new window]
  		Table 3. Correlation between soil P saturation estimated by three indices (degree of P saturation based on soil test P and a P sorption index [DPSSTP], based on Mehlich-3 P [DPSM3], and based on ammonium-oxalate P [DPSOX]) and P application rate or soil P measured by five tests for all soils (n = 75) and excluding the calcareous Harps soil (NonC).

 
Relationships between Surface Runoff Phosphorus and Phosphorus Application Rates
Runoff P concentrations increased linearly with P application as measured by DRP and BAP for all soils (Fig. 4). Relationships also were linear across all soils (not shown), and correlation coefficients were 0.98 for DRP and 0.93 for BAP. The linear coefficients (mg L–1 runoff P increase per mg kg–1 applied P) ranged across soils from 0.0016 to 0.0037 for DRP and 0.0018 to 0.0049 for BAP. The calcareous Harps soil had a twofold greater rate of increase for DRP and BAP than other soils. These results indicate that runoff water removed proportionally more P from the Harps soil even though P application rates were the same across soils. The results agree with high rates of soil P increase with increased P application for the Harps soil as measured by OP, WP, FeP, DPSox, and DPSM3. Therefore, as soil P increases, these tests predict a greater risk of DRP and BAP loss in surface runoff from Harps soil than from the other soils if field slopes and management practices are similar.


Figure 4
View larger version (28K):
[in this window]
[in a new window]
  		Fig. 4. Effect of P application on runoff P concentration. Different letters by the soil names indicate significantly different (P < 0.05) linear coefficients between soils for each runoff P fraction.

 
For noncalcareous soils, runoff BAP increase with increased P application rates was, on average, 1.5 times greater than the DRP increase; however, runoff DRP and BAP linear coefficient differences among these soils were very small and usually not significant (P < 0.05). The only difference worth emphasizing was that DRP and BAP rates of increase with increased P rate were greatest for the Fayette soil, lowest for the Tama soil, and intermediate for the Nicollet and Marshall soils. These results agree with high rates of soil P increase with P application for Fayette and lower rates of increase for the Tama soil as measured by BP, M3P, OP, DPSox, and DPSM3, but not by FeP, WP, or DPSSTP. Therefore, the former tests seem better predictors of a slightly greater risk of runoff P loss for Fayette soil compared with other soils, especially Tama soil, with increased P rates if field slopes and management practices were similar. Greater rates of runoff DRP and BAP increase for the Fayette soil than for the Tama soil can be explained by differences in several soil properties (Table 1), such as lower specific surface, and clay, Feox, and Alox concentrations.

We also found that, as the P application rate increased, the ratios DRP/TPR and BAP/TPR increased linearly for all soils (not shown), although trends were steeper and better defined for the Marshall, Nicollet, and Fayette soils. For these soils, DRP/TPR without P application ranged from 0.03 to 0.07 but with the greatest P application rate ranged from 0.17 to 0.30. Similarly, BAP/TPR increased from a range of 0.07 to 0.10 without P application to a range of 0.29 to 0.48 with the high P rate. Andraski et al. (2003) found similar results on Wisconsin silt loam soils, where DRP/TPR increased from 0.04 to 0.15 and BAP/TPR increased from 0.10 to 0.23 as P application increased. Hence, the risk for P loss in the DRP and BAP runoff fraction increased at a proportionatey greater rate when the P application rate increased.

Relationships between Surface Runoff Phosphorus and Soil Phosphorus Concentration
Runoff DRP, BAP, and TPR concentrations always increased linearly (P < 0.05) with increased soil P measured by all tests; however, only in a few instances did linear regression coefficients differ across soils and these differences were small. Because of this result and the large number of possible comparisons, only data for selected relationships are shown. For DRP (Fig. 5), the most clear difference between soils or soil P tests was a steeper increase for the Harps soil than other soils with increased BP, M3P, and FeP but small or no soil differences with increased OP and WP. Only data for M3P, OP, and FeP are shown in Fig. 5 because relationships for BP and M3P were similar and those for WP and OP also were similar. For BAP (Fig. 6), the most clear result also was the different ranking of the Harps soil compared with other soils. The BAP increase for the Harps soil was less steep than for other soils for OP or WP but was intermediate or soil differences were not statistically different for BP, M3P, or FeP (only data for M3P, OP, and FeP are shown). Differences in DRP or BAP rates of increase among the noncalcareous soils were small, not significant for BP or M3P, and slightly steeper for the Fayette soil than for other soils when the other tests were used. Less steep runoff P increases with increased OP or WP for the Harps soil than other soils or tests seemed to disagree with data in Fig. 4 that showed greater DRP and BAP concentrations with increased P rate for the Harps soil. The results are explained by larger P application effects on soil P measured with OP and WP in the Harps soil than for any other soil or P test combination (Fig. 1 and 2). These results emphasize that use of some routine or environmental soil P methods to predict DRP or BAP in runoff must be based on separate calibrations for different soils.


Figure 5
View larger version (25K):
[in this window]
[in a new window]
  		Fig. 5. Relationship between soil-test P and dissolved reactive P in runoff. Different letters by the soil names indicate significantly different (P < 0.05) linear coefficients between soils for each soil test.

 

Figure 6
View larger version (26K):
[in this window]
[in a new window]
  		Fig. 6. Relationship between soil-test P and bioavailable P concentration in runoff. Different letters by the soil names indicate significantly different (P < 0.05) linear coefficients between soils for each soil test.

 
Our results showed linear increases of runoff DRP and BAP as soil P increased for all soils and trends; but in the literature, reports on data analyses to evaluate linear or nonlinear trends have shown variable results. Some reports indicate that runoff P concentration may increase at a greater rate once a threshold STP level is reached, which is also referred to as a change point. For instance, Sims et al. (2002) found DRP concentrations increased faster above 150 mg kg–1 M3P in silt loam and loamy sand Delaware soils that were generally more weathered, acidic, coarse-textured, and lower in organic matter than the soils in our study. McDowell et al. (2001) stated that it is often difficult to detect a change point, and that at least eight very different soil P levels encompassing an appropriate range would be required to detect a change point. Although our study included only five levels of P, there was no indication of a change point for any soil or test. Andraski and Bundy (2003) also reported linear increases for runoff DRP concentrations with increased soil P in a rainfall simulation study using a similar rainfall nozzle but with a 60-min runoff time conducted on Wisconsin soils that tested up to 130 mg kg–1 BP. The reported linear coefficient for noncalcareous silt loam soil was 0.0024 mg L–1 DRP for each mg kg–1 of BP, which was within the linear coefficient range in our study (0.0021–0.0028 mg DRP L–1 across soils). The reported linear coefficient for a calcareous (pH 7.5) silty clay loam soil of 0.012 mg DRP L–1 for each mg kg–1 of BP (they did not use OP or other tests we used) was greater than the DRP linear coefficient of 0.006 mg L–1 observed for the calcareous Harps soil in our study (not shown). These authors calculated that 1 mg DRP L–1 would correspond to 410 mg kg–1 BP (0–15-cm depth) for noncalcareous soils in their study. Similar calculations for BP in our study resulted in 383 mg kg–1 for Marshall, 453 mg kg–1 for Nicollet, 345 mg kg–1 for Tama, and 339 mg kg–1 for Fayette soils.

Table 4 summarizes linear relationships across soils between runoff P concentrations (DRP and BAP) and soil P measured by all methods by showing simple correlation coefficients. Correlation coefficients between DRP and the various soil tests across all soils ranged from 0.83 to 0.98. The poorest correlations were for BP and TPS (r = 0.83 and 0.91), while for all others (FeP, M3P, OP, and WP) were ≥0.94. Correlation coefficients between BAP and the soil tests across all soils ranged from 0.88 to 0.97. The poorest correlations were for TPS, BP, and WP (r = 0.88 for TPS and 0.91 for BP and WP) while for all others (FeP, OP, and M3P) were ≥0.96. Excluding the calcareous Harps soil greatly improved the correlation for BP (r increased to 0.97 and 0.98 for DRP and BAP, respectively), did not improve the correlation for FeP or OP, and marginally improved correlations for other tests. These results indicate that, except for BP and TPS, all other soil P extractants were highly (≥0.94) correlated with runoff DRP and BAP across all soils of the study.


View this table:
[in this window]
[in a new window]
  		Table 4. Correlation between runoff P concentration and soil P measured as dissolved reactive P (DRP) or bioavailable P (BAP, using Fe-oxide-impregnated filter paper) for all soils (n = 75) and excluding the calcareous Harps soil (NonC).

 
Relationships between Surface Runoff Phosphorus and Soil Phosphorus Saturation
Runoff DRP and BAP increased linearly (P < 0.05) with increased soil P saturation estimated by all indices. Relationships for DRP and the three saturation indices are shown in Fig. 7. Results for BAP are not shown because trends also were linear and differences among soils in rate of BAP increase were similar to those for DRP (although BAP was, on average, 1.5 times greater than DRP). Only relationships between DRP and DPSSTP calculated using OP are shown in Fig. 7 because results using other P tests were similar, although the magnitude of any difference between the calcareous Harps soil and other soils varied depending on the P test used. Data in Fig. 7 show two obvious results. One is that relationships for the Harps soil clearly deviated from those for other soils for the three saturation indices. The rate of DRP increase with increased DPSox for the Harps soil was similar to that for other soils (except the Nicollet soil) but a particular DRP value corresponded to a much greater DPSox value for the Harps soil than for other soils. The rate of DRP increase with increased DPSM3 was much smaller for the Harps soil than for all other soils and a particular DRP value also corresponded to a greater DPSM3 value for the Harps soil than for other soils. Therefore, these two saturation indices would overestimate DRP loss from calcareous Harps soil compared with other soils. In contrast, the rate of DRP increase with increased DPSSTP was steeper for the Harps soil than for other soils, especially for the three highest P application rates (a curvilinear increase). The DPSox and DPSM3 indices were developed in work mainly with acid or neutral soils and, therefore, interpretations of results for the Harps soil are uncertain; however, the DPSM3 index and an index based on the OP/PSI ratio have been used to study soil P saturation in calcareous soils (Westermann et al., 2001; Laboski and Lamb, 2004).


Figure 7
View larger version (26K):
[in this window]
[in a new window]
  		Fig. 7. Relationships between dissolved reactive P concentration in runoff and three soil P saturation indices: degree of P saturation based on soil test P and a P sorption index (Degree of P SaturationSTP), based on Mehlich-3 P (Degree of P SaturationM3), and based on ammonium-oxalate P (Degree of P SaturationOX). Degree of P Saturation STP is shown only for the Olsen P soil test. Curvilinear trends are shown only when the quadratic term was significant (P < 0.05) after a linear term. Different letters by the soil names indicate significantly different linear coefficients between soils for each index.

 
The second obvious result shown by Fig. 7 is that the rate of DRP increase for the Nicollet soil was less steep than for other noncalcareous soils for DPSox and DPSSTP, and also for DPSM3 although the difference did not achieve statistical significance at P < 0.05. These results indicate that all three saturation indices, but mainly DPSox and DPSSTP, would overestimate DRP loss from Nicollet soil compared with other soils. We cannot explain this result with certainty based on the measured soil properties (Table 1). For example, clay, specific surface, Feox (but not FeM3), and PSI were lowest for the Nicollet soil compared with the other noncalcareous soils, which suggests greater DRP loss for Nicollet soil under otherwise similar conditions; however, soil pH was lowest for Nicollet soil, which would suggest lower DRP loss because P sorption reportedly increases with decreased pH at these ranges (Mora and Canales, 1995).

Correlation analyses across all soils and treatments indicated approximately similar relationships between DRP or BAP and the three soil P saturation indices. Correlation coefficients between DRP and DPSox, DPSM3, or DPSSTP were 0.90, 0.93, and 0.89, respectively. Correlation coefficients between BAP and DPSox, DPSM3, or DPSSTP were 0.86, 0.87, and 0.92, respectively. Because of large differences already discussed for the Harps soil, the saturation indices correlated better with DRP and BAP when the calcareous Harps soil was excluded from calculations, although the rankings remained similar. Without the Harps soil, correlation coefficients ranged from 0.91 to 0.97 for DRP and from 0.93 to 0.98 for BAP. Because of inconsistencies discussed above for relationships among DRP (or BAP), DPSox, and DPSM3 across soils, we explored calculating DPS for both methods by including in Eq. [1] and [2] an {alpha}m coefficient of 0.5 and calculating it for each soil using PSI, as suggested by van der Zee and van Riemsdijk (1988). Calculated {alpha}m values with DPSox, for example, ranged from 0.21 to 0.36 for noncalcareous soils and from 0.44 to 0.72 for Harps. Relationships based on these calculations are not shown, however, because DPSox and DPSM3 calculated in these ways did not improve correlations with DRP and BAP and did not substantially reduce differences in estimates of DRP or BAP by these two indices. The DPSSTP index also correlated better with DRP and BAP when the Harps soil was excluded.

Theoretical reasons and some authors (Morel et al., 2000) suggest that soil P saturation indices would provide more similar relationships with runoff DRP or BAP across soils than routine soil P tests, mainly for DRP. However, comparisons of correlations between runoff P and the three saturation indices across all soils (r ≤ 0.93 for DRP and r ≤ 0.92 for BAP) with correlations for the soil tests in Table 4 indicate that this suggestion was not substantiated by our study, a result also found by others. In fact, correlations were greater for all soil tests, except BP.


   SUMMARY AND CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
REFERENCES
 
Phosphorus application increased soil P linearly when measured with all tests in all soils. A statistically significant greater soil P increase with increasing P rate (a curvilinear trend) was observed for some soils and tests, but the models' R2 values were only slightly greater (≤0.03) than for linear models. Linear coefficients of relationships between soil P and P rate usually did not differ or were very small and inconsistent across soils and tests. Exceptions were that FeP, WP, and OP increases per unit P applied were greater for the calcareous Harps soil than for the noncalcareous soils. Increased P rate increased DPSox and DPSM3 linearly, except for a small curvilinear increase for the Fayette soil, and increases for both methods were greatest for the Harps soil. In contrast, DPSSTP increased curvilinearly for all soils and tests and differences among soils were small. Correlations across all soils indicated that DPSSTP was better related to P application rate and soil P measured by various methods than DPSox or DPSM3 (r = 0.89–0.96, 0.70–0.93, and 0.62–0.97, respectively); however, correlation coefficients became greater and differences among indices became smaller when the Harps soil was excluded (r = 0.86–0.98).

Concentrations of DRP and BAP in runoff increased linearly with increased P application rates and soil P measured by all methods, although BAP was, on average, 1.5 times greater than DRP. The runoff P increase (mg P L–1) per unit P applied (mg P kg–1) across soils ranged from 0.0016 to 0.0037 for DRP and 0.0018 to 0.0049 for BAP. The greatest values were for the calcareous Harps soil (almost twofold greater) and differences among noncalcareous soils were small. Relationships between DRP or BAP and soil P measured by various extractants differed little among the noncalcareous soils but differed substantially for the Harps soil, mainly because of large differences in P extraction among methods for this calcareous soil compared with relatively small differences for the noncalcareous soils. Both DRP and BAP increased linearly with increased DPSox or DPSM3 but sometimes increased curvilinearly with DPSSTP. Correlations between runoff P and the three saturation indices across soils also were improved when the Harps soil was excluded. Environmental soil P tests (FeP and WP) and saturation indices (DPSox, DPSM3, and DSPSTP) were not better correlated with runoff P than routine soil tests recommended in the Midwest (OP and M3P) to assess crop P availability for soils ranging from slightly acid to alkaline pH due to CaCO3.

Overall, the results showed that differences in relationships between runoff P and P application or soil P were small or nonexistent among four noncalcareous soils but often differed for a calcareous soil. Routine soil P tests can be used to estimate runoff P concentration from these soils, although estimates would be improved by separate calibrations for calcareous and noncalcareous soils when the BP test is used. Care is needed when attempting to extrapolate results of this indoor simulation study with relatively short incubation times to field conditions, especially when soil P levels as high as those observed in this study are the result of many years of P application, and when the primary interest is in assessing P loads. Caution is also warranted when comparing results of this study to those of other studies that may use a different protocol for indoor rainfall simulations.


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




This article has been cited by other articles:


Home page
 J. Environ. Qual.Home page
B. L. Allen and A. P. Mallarino
Effect of Liquid Swine Manure Rate, Incorporation, and Timing of Rainfall on Phosphorus Loss with Surface Runoff
J. Environ. Qual., January 4, 2008; 37(1): 125 - 137.
[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]

This Article
Right arrow Abstract