«A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Crop, Soil, and Environmental Sciences ...»
Accepted fractionation techniques partition existing P into various pools to assess their availability to the environment. In these fractionation procedures, soils are subjected to increasingly aggressive serial extractions to dissolve secondary mineral compounds, thereby estimating P-bound forms and their resulting solubility (Kuo, 1996). Several studies have correlated the soluble and loosely bound soil P fractions (water or weak salt extractable P) to runoff DRP concentration in linear relationships (Pote et al., 1996; Pote et al, 1999; Torbert et al., 2002); however, other soil P fractions also play important roles in determination of environmentally available soil P. For instance, reductant soluble P formations (sodium citrate, sodium bicarbonate and sodium dithionate extractable P) dissolve when subjected to anaerobic conditions; therefore, this soil P fraction increases runoff P when in contact with surface water and increases algae available P when washed into waterways (Slomp et al., 1996; Ruiz et al., 1997).
Expected P fractions in soils may not be completely pH dependent as theory suggests and may vary depending on past soil management. Research by Beauchemin and coworkers (2003) suggested that appreciable amounts of Ca-P minerals exist in acidic soils while Fe-P minerals exist in basic soils. Similarly, Delgado and Torrent (2000) found considerable Ca-P minerals in both calcareous and acid limed soils under
were the predominant soil P fraction regardless of soil pH on heavily fertilized soils.
However, regardless of pH, soil secondary mineral formation and dissolution with P compounds controls P solubility and the resulting environmental availability responsible for increasing DRP concentrations in runoff water.
We encountered several unanticipated results when conducting runoff investigations on soils with purposely exaggerated wide ranges in STP levels (20 to 1,500 mg P kg" soil). Preliminary data suggested that as STP increased due to manure additions, soil pH and Ca concentrations also increased. Due to changes in inherent soil properties, the resulting runoff DRP concentrations did not linearly increase as anticipated, but plateaued or decreased after certain STP concentrations were reached.
Understanding P interactions in acidic soils receiving long term applications of manure are imperative to fully evaluate the STP threshold concept and determine the environmental fate of additional applications of P fertilizer or PL. The main objective of this study was to determine if changes in P secondary mineral fractions were responsible for buffering DRP loss in runoff as Mehlich-3 STP concentrations increased due to long term manure applications on traditionally acidic soils.
Rainfall simulations were conducted on a Captina SiL (Fine-silty, siliceous, active, mesic Typic Fragiudult) at the Arkansas Agricultural Research and Extension Center in Fayetteville; a Virginia Frederick SiL (Fine, mixed, semiactive, mesic Typic Paleudult) on production agricultural fields in Augusta and Rockingham Counties, VA;
the Southwest Missouri Center near Mt. Vernon, MO. All sites were certified as the perspective benchmark soil by NRCS personal and did not receive any manure applications for 1 yr prior to rainfall simulations. Exact fertilizer and management histories of each site are not available; however, historic PL applications resulted in soils with a wide range of Mehlich-3 (20 to 1554 mg P kg"1) and water extractable (2 to 122 mg P kg"1) STP concentrations (Table 3.1). The Captina and Hoberg SiL plots were established on a tall fescue (festuca arundinacea Schreb.) pasture with 100% ground cover while the Frederick SiL rainfall simulations were conducted on corn (Zea mays) fields immediately following silage harvest that resulted in 18% groundcover (Laflen et al., 1981). Captina and Hoberg SiL site locations had an average slope of 5% while Frederick SiL simulations were conducted on plots with an average slope of 7.5%.
Rainfall simulation plots were established according to the National Phosphorus Project Protocol (http://www.seral 7.ext.vt.edu/Documents/National_P_protocol.pdf) using a portable rainfall simulator (Humphry et al., 2002). Plots were 1.5 x 2.0 m and used aluminum edging to partition runoff into a flow collector on the downward slope end. Water used to simulate rainfall was passed through a series of exchange resins (cation-anion-cation) to reduce the inherent variability between source waters. The conditioned water had properties approximating natural rainfall, i.e., low buffering capacity and pH ~4. All water used for simulations was analyzed and determined to have P concentrations below instrumental detection limits. Rainfall was simulated at an intensity of 6.7 cm h"1 and the total amount of runoff was collected for 30 min. Each site
capacity between simulations as determined by measuring soil moisture content using a ThetaProbe soil moisture sensor (ML2, Delta-T Devices, Cambridge, UK) as described in the National P protocol.
Runoff water samples were collected for analysis from the total runoff volume for each simulation. A sub-sample was filtered through 0.45 um pore filter paper and acidified using concentrated HC1 (1 drop per 10 mL sample = pH ~ 2). Dissolved reactive P was analyzed colormetrically using a Sanplus System (SKALAR Inc., Norcross, GA 30071) (Murphy and Riley, 1962).
After the last simulation, representative soil samples from a 5-cm depth were randomly taken using 10 soil cores from each plot to form a composite sample. Soil samples were air dried and sieved through a 2-mm screen to remove non-soil fragments and to reduce aggregates. Mehlich-3 extractable P and Ca were determined by shaking 2 g soil with 20 mL Mehlich-3 extractant for 5 min on a reciprocating shaker at 200 evolution per minute (epm) and filtering through Whatman 42 filter paper (Whatman pic, Middlesex, UK) (Mehlich, 1984; Sims, 2000). Water-extractable P was determined by shaking 2 g soil with 20 g double deionized water for 1 h at 200 epm, filtering through
0.45um membrane filter, and acidifying to pH~2 by adding two drops concentrated HC1 (Self-Davis et al, 2000). Calcium and P were quantified using a SPECTRO CIROS inductively coupled argon plasma spectrometer (SPECTRO Analytical Institute, Kleve, Germany). Soil pH was determined on a 1:2 ratio (soihwater, w:w) using double deionized water.
Soil P was fractionated for samples of each soil series representing a wide range of Mehlich-3 extractable STP concentrations (low that received no PL fertilization to very high). Captina and Hoberg SiL soils were taken from undisturbed pasture to a depth of 5-cm while the Frederick SiL soil was sampled from fields that were tilled to a 10-cm depth. Phosphate-bound fractions were quantified using an inorganic fractionation scheme developed for non-calcareous soils (Kuo, 1996). This scheme solubilizes P in separate stages using increasingly aggressive serial extraction solutions. Due to nonstandardization of names for P extracted from each serial stage in the literature, we assigned names to each P fraction based on the expected cation primarily responsible for insolubility. Phosphorus was extracted into five fraction stages in the following order: 1) easily soluble and loosely bound P (SLB-P) extracted with 1 MNH4CI; 2) Al phosphates (Al-P) extracted with 0.5 MNH 4 F; 3) Fe phosphates (Fe-P) extracted with 0.1 MNaOH;
4) reductant soluble P (RS-P) extracted with 0.3 MNasCsHeOy, 1 MNaHC0 3, and Na2S2C4; and the least soluble fraction; 5) Ca phosphates (Ca-P) extracted with 0.25 M H2SO4. Decanted extracts were immediately colorimetrically analyzed for reactive P concentration using a Cary 50 Bio UV-Visible spectrophotometer (Varian Inc., Palo Alto, CA 94304) (Murphy and Riley, 1962). Total Pi was calculated by adding together the successive fraction extracts. Total P analysis was conducted using digestion with perchloric acid as described by Kuo (1996) and the organic P (P0) fraction of soil P was found by difference (Pt - Pi = P0) (Tiessen and Moir, 1993).
Rainfall simulations were established as completely randomized designs with Mehlich-3 STP as a fixed effect on 11, 8 and 11 locations for Captina, Frederick and Hoberg SiL, respectively. Captina and Hoberg SiL plots were replicated 3 times while Frederick SiL plots were replicated 4 times and replication was analyzed as a random effect. Correlation between STP and DRP in runoff was established using simple linear regression (PROC REG) with SAS software (SAS Institute, 2003). Relationships of water extractable P as a percentage of Mehlich-3 P versus Mehlich-3 STP and exchangeable Ca were established by taking natural logarithms of the x and y axis and analyzed as a linear model using regression (PROC REG). For illustration purposes, these functions are presented as a power relationship with the associated linear regression r and ^-values.
Only the highest order statistically significant correlation was reported for all regression models. A significance level of p0.10 was established a priori.
The soil P fraction study was established as a completely randomized design replicated three times. Mehlich-3 STP served as a fixed effect for each soil series while replication was a random effect. Analysis of variance (PROC GLM) was conducted using SAS software (SAS Institute, 2003) and means separated using Fisher's protected least significant difference tests (LSD). A significance level of p0.10 was established a priori. In order to discuss locations within each soil series that had various background concentrations of STP and cations, data will be discussed as a percentage of Pt.
Wide STP ranges in the same soil series resulted from varying intensities of previous PL applications. Mehlich-3 P concentrations ranged from 37 to 646 mg P ha"1 for Captina SiL, 82 to 1162 mg P kg"1 for Frederick SiL and 20 to 1554 mg P kg"1 for Hoberg SiL soils (Table 3.1). Water-extractable P soil concentrations ranged from 9 to 122, 5 to 92 and 6 to 71 mg P kg"1 for Captina, Frederick, and Hoberg SiL soils, respectively (Table 3.1).
Runoff DRP correlated with Mehlich-3 STP in quadratic relationships for all three soils (Fig. 3.1). Research by Pote and coworkers (1996) established linear relationships between Mehlich-3 extractable P and runoff DRP on a Captina SiL, but STP concentrations did not exceed 500 mg P kg"1. Peak DRP concentrations in runoff were released from soils with STP concentrations of 495, 458, and 1084 mg P kg _1 for Captina, Frederick, and Hoberg SiL, respectively (Fig. 3.1). Above these STP concentrations, DRP in runoff plateaued or decreased. Quadratic relationships suggest that linear relationships may not be appropriate for modeling DRP release from soil series with high Mehlich-3 STP concentrations (Sharpley et al., 2004). Sequential soil P fraction extracts and relationships between exchangeable Ca and Mehlich-3 STP discussed later may provide reasons for quadratic relationships between DRP and Mehlich-3 extractable STP at high soil P concentrations.
Statistically, water-extractable STP correlated with runoff DRP in a quadratic relationship for Frederick SiL but in linear relationships with Captina and Hoberg SiL (Fig. 3.1). For a Captina SiL, Pote and coworkers (1996) found a higher correlation with
values, Mehlich-3 provided the best correlation with DRP in runoff with Captina and Frederick SiL while water extractable P provided the best fit for Hoberg SiL soils in our study. For these three soils, it appears that the acidic Mehlich-3 extractant is acceptable for prediction of STP and DRP relationships until a soil specific STP threshold is reached. Below this threshold, Mehlich-3 provides a similar or better model for prediction of DRP than water extractable P according to r2 and/?-values; however, above this threshold Mehlich-3 may dissolve water insoluble compounds and over-predict environmentally available P (Sharpley et al., 2004).
An analysis comparing water extractable P as a percentage of Mehlich-3 P (water extractable P + Mehlich-3 P x 100) versus Mehlich-3 STP concluded that a power relationship exists (r2 = 0.60) (Fig. 3.2). A power function is indicative of a threshold value limiting the usefulness of Mehlich-3 as an environmental predictor of DRP at high Mehlich-3 STP concentrations (Sharpley et al., 2004). From the power relationship, water extractable P per unit Mehlich-3 STP rapidly decreased between 150 and 400 mg Mehlich-3 P kg"1. A decreasing ratio of water extractable P to Mehlich-3 P indicates that the acidic Mehlich-3 extracts P-formations not readily extracted by rainfall; thereby, giving an abnormally high prediction of DRP loss during rainfall on high STP soils. Both the Captina and Frederick SiL soils' Mehlich-3 STP and runoff DRP asymptote (Fig. 3.1) occurred at the beginning of the plateau of the power relationship (Fig. 3.2) while the Hoberg SiL's was situated on the plateau. Research by Sharpley and coworkers (2004) found a Mehlich-3 STP breakpoint of 412 mg P kg"1 using a split-line model. In their model, Mehlich-3 STP values above 412 mg P kg"1 had a differing relationship with
prediction of DRP runoff loss cannot be assumed for low and high Mehlich-3 STP concentrations as an overestimation of runoff P occurs. The three soils used in our study had a similar relationship to Sharpley and coworkers' (2004) breakpoint since water insoluble P compounds were extracted by Mehlich-3 after this value was exceeded.
Long term PL additions had a positive influence on soil pH (Fig. 3.3), similar to results found by Hue (1992) and Kingery and coworkers (1994). According to Lindsay (1979), available soil Ca allows for formation of Ca-P secondary minerals as pH increases. As P; is shifted from environmentally available SLB-P fractions into water insoluble secondary minerals such as Ca-P, P is not readily available to plants nor as susceptible to loss in runoff as DRP (Torbert et al., 2002; Sharpley et al., 2004).
Maximum pH was reached in the Captina SiL with STP concentrations of 425 mg P kg"1.