sorption of escherichia coli in agricultural soils ... - Soil Chemistry

SORPTION OF ESCHERICHIA COLI IN AGRICULTURAL SOILS INFLUENCED BY SWINE MANURE CONSTITUENTS J. A. Guzman, G. A. Fox, C. J. Penn

ABSTRACT. Sorption of fecal bacteria in soils treated with swine effluents as a fertilizer involves complex mechanisms that are functions of the effluent constituents, soil colloid properties, and fecal bacteria population. Swine effluents contribute significant solutes and organic compounds during application and potentially increase the soil solution pH and ionic strength when in contact with soils. The objective of this research was to investigate sorption of fecal bacteria on soils treated with swine effluents and to derive a fecal bacteria sorption model based on soil properties. Sorption of Escherichia coli was investigated using a series of artificial and natural soils treated with swine effluent at varying dilution ratios (i.e., manure effluent concentrations). Note that this research conducted sorption tests based on multiple strains of the E. coli population in the manure effluent, as opposed to a specific E. coli strain. Fecal bacteria in swine effluents mainly consisted of attached bacteria as opposed to free cells in suspension (i.e., planktonic form). Sorption of surface‐bonded E. coli appeared to be controlled by processes occurring in the substrate (i.e., surfaces to which the bacteria attached) as a function of the soil solution pH and ion exchange. For soils up to 30% clay content and 3.0% total carbon content, nonlinear equations characterized the sorption of E. coli in the artificial and natural soils. For soils with percent clay less than or equal to 11%, total carbon played a primary role in the estimated sorption of fecal bacteria. For soils with clay percentages larger than 11% and lower than 30%, sorption of fecal bacteria was directly correlated to amorphous aluminum and iron concentrations, percent clay, and total carbon. In addition, dispersion resulting from alkalinity buildup due to adding effluent to the soils in the presence of soil organic matter decreased observed sorption of E. coli in some cases, especially at higher effluent ratios. Experimental data indicated that predictions based exclusively on the percent clay content overestimated sorption of E. coli on soils treated with swine effluent in most soils. Manure effluent concentration, the presence of attached bacteria, and soil dispersion under high effluent concentrations should be considered when modeling fecal bacteria transport in soils. Future environmental research should be conducted with actual manure sources and multiple strains of an E. coli population. Keywords. Attached bacteria, Bacteria transport, Escherichia coli, Fecal bacteria, Sorption, Swine effluents.

F

ollowing manure application in agricultural lands, soil sorption of fecal bacteria is influenced by the soil's physical and chemical properties, animal waste effluent characteristics, and bacteria proper‐ ties (Foppen et al., 2005; Pachepsky et al., 2006; Torkzaban et al., 2008; Kim and Walker, 2009; Guzman et al., 2009, 2010). Under nutrient‐depleted environments (e.g., aged ma‐ nures commonly used as fertilizers), fecal bacteria are mainly found surface‐bonded (e.g., reversibly or irreversibly at‐ tached) rather than as free cells in suspension (i.e., planktonic form) (Dunne, 2002; Winfield and Groisman, 2003). Soil sorption of fecal bacteria following manure application re‐ sults from complex mechanisms and interactions occurring

Submitted for review in November 2010 as manuscript number SW8910; approved for publication by the Soil & Water Division of ASABE in January 2012. Presented at the 2009 ASABE Annual Meeting as Paper No. 096239. The authors are Jorge A. Guzman, ASABE Member, Post‐Doctoral Research Hydrologist, USDA‐ARS Grazinglands Research Laboratory, El Reno, Oklahoma; Garey A. Fox, ASABE Member, Associate Professor, Department of Biosystems and Agricultural Engineering, Oklahoma State University, Stillwater, Oklahoma; and Chad J. Penn, Assistant Professor, Department of Plant and Soil Sciences, Oklahoma State University, Stillwater, Oklahoma. Corresponding author: Garey A. Fox, Department of Biosystems and Agricultural Engineering, 120 Ag Hall, Oklahoma State University, Stillwater, OK 74078; phone: 405‐744‐8423; fax: 405‐744‐ 6059; e‐mail: [email protected].

between a soil particle and the bacterium, as well as soil par‐ ticles and organic compounds in which fecal bacteria are at‐ tached. These sorption mechanisms are mediated by interactions occurring at the surface/bulk solution interface of the soil colloids and organic compounds (Brown et al., 2000; ter Laak, 2005) as changes in the soil solution pH and ion exchange occur. Parallel to this, animal wastes contribute significant loads of organic compounds and solutes to soils when applied as fertilizers, along with a diverse microbial load (Choudhary et al., 1996; Leung and Topp, 2001). Apply‐ ing animal wastes to soils promotes changes in the soil prop‐ erties and microbial community (Gerzabek et al., 1997; Haynes and Naidu, 1998; Eghball et al., 2004; Turner et al., 2010). Microbial communities found in aged manures stored over long periods in pits prior to application significantly deviate from the microbial communities directly excreted from pigs in most cases (Leung and Topp, 2001). Soil sorption of fecal bacteria has been typically investi‐ gated using planktonic monostrain cultured bacteria in batch soil column or microfluid devices. Sorption is typically de‐ scribed by linear or nonlinear relationships (Ling et al., 2002; Mankin et al., 2007; Bolster et al., 2009) or by the Derjaguin‐ Landau‐Verwey‐Overbeek (DLVO) theory (Redman et al., 2004; Torkzaban et al., 2008). However, these studies be‐ come limited in applicability with complex soil/bulk solution interactions driven by changes in the soil solution pH and ion‐ ic strength commonly found when soils are treated with ani‐

Transactions of the ASABE Vol. 55(1): 61-71

E 2012 American Society of Agricultural and Biological Engineers ISSN 2151-0032

61

mal waste. In addition, previous investigations are typically limited to quartz particles or glass beds (Walker et al., 2005; Yang et al., 2008; Kim and Walker, 2009), which are not rep‐ resentative of the highly reactive soil minerals (i.e., mostly variably charged) and organic matter (i.e., variably charged) common in agricultural fields (Mankin et al., 2007). A few studies have been conducted to investigate the fate and sorption of fecal bacteria on soils treated with manure fertilizers. Batch experiments using loam and clay loam top‐ soils were conducted to evaluate the effect of bovine manure colloids on E. coli attachment (Guber et al., 2005) and to compare the attachment of fecal coliforms to the same two soil types and its soil fractions (Guber et al., 2007). In both series of experiments, cultured bacteria (e.g., fecal coliforms or E. coli) isolated from bovine strains were suspended in dis‐ tilled water. Other studies on sorption of fecal bacteria have been conducted on loess clay loam soil (clay, silt, sand, and organic matter contents of 29%, 45%, 26%, and 0.8%, re‐ spectively) treated with waste effluents (Gantzer et al., 2001; Kouznetsov et al., 2004). Nonlinear relationships were re‐ ported describing sorption of fecal coliform in soils. Ling et al. (2002) proposed a linear relationship between the E. coli distribution coefficient and the natural logarithm of the clay content based on two different soils (14% and 35% clay con‐ tent with 0.84% and 0.54% organic matter content). Mankin et al. (2007) conducted similar experiments and reported Freundlich isotherms describing E. coli sorption to sand and silt loam soils but concluded that linear isotherms better fit low initial E. coli concentrations. Further research is needed on the sorption of particle‐associated fecal bacteria and the influence of the swine effluent constituents. This research hypothesizes that manure constituents and the predominance of attached fecal bacteria found in aged animal wastes result in unique sorption observations when compared to previous sorption studies utilizing suspended monostrain free cells (e.g., planktonic). The objectives of this

Soil ID

[a] [b] [c]

62

Name

study were to: (1) investigate sorption of E. coli in a range of soils after liquid swine manure application, (2) determine the potential for using nonlinear equations to describe this sorp‐ tion, and (3) utilize soil properties for estimating the nonlin‐ ear coefficients. The development of such predictive equations will be useful in estimating the detectable E. coli sorption simultaneously from multiple strains in liquid swine manure‐amended soils for improved modeling of E. coli fate and transport. Note that soil sorption of fecal bacteria is not synonymous with fecal bacteria immobilization, as fecal bacteria can be attached to soil particles that can become mo‐ bile when suspended in solution.

MATERIALS AND METHODOLOGY SOIL SAMPLES Natural soils (i.e., Oklahoma and Iowa soils; table 1) and artificial soils (i.e., artificially constituted and treated natural soils; table 2) were used in E. coli sorption experiments per‐ formed at room temperature (23°C ±0.5°C). Benchmark Oklahoma soils (natural soils) were provided by the Soil, Wa‐ ter, and Forage Analytical Laboratory (SWFAL) at Oklaho‐ ma State University, and the Iowa soil (natural soil) was collected from the Iowa State Research Farm in Nashua, Iowa. Soils were air‐dried, ground, and sieved to pass a 2 mm opening. Soil particle size distribution was estimated by the wet sieve and hydrometer method (Gee and Or, 2002). Soil pH was measured in the natural and artificial soils using 1:1soil to distilled water dilution (Thomas, 1996). Amor‐ phous Al and Fe mineral content from the Oklahoma bench‐ mark soils (Fuhrman, 2000) were extracted using the acid ammonium oxalate reagent (McKeague and Day, 1966). Artificial soils were used to assess and quantify the effect of soil minerals and total carbon on attachment of E. coli to soils from manure effluents. Artificially constituted soils were prepared by mixing Silurian sand (U.S. Silica Co.,

Table 1. Soil properties of the natural soils used in the sorption experiments. Soil Clay TN[a] TC[a] (%) (%) pH (%) Classification

Al[b] Fe[b] (mmol kg‐1) (mmol kg‐1)

PR

Pratt

Sandy, mixed mesic Lamellic Haplustalfs

6.3

7

0.04

0.36

5

2

DO

Dougherty

Loamy, mixed, active, thermic Arenic Haplustalfs

5.2

8

0.06

0.48

7

5

DA

Darnell

Loamy, siliceous, active, thermic, shallow Udic Haplustepts

5.4

11

0.06

0.57

10

3

BE

Bernow

Fine‐loamy, siliceous, active, thermic Glossic Paleudalfs

4.3

11

0.13

1.3

11

12

CO

Cobb

Fine‐loamy, mixed, active, thermic Typic Haplustalfs

5.5

16

0.06

0.4

19

6

CA

Camasaw

Typic Hapludults

5.7

21

0.12

2.83

44

29

EA

Easpur

Fine‐loamy, mixed, superactive, thermic Fluventic Haplustolls

5.9

22

0.09

0.96

16

10

SA

Sallisaw

Fine‐loamy, siliceous, superactive, thermic Typic Paleudalfs

5.5

22

0.15

1.26

44

23

PA

Parsons

Fine, mixed, active, thermic Mollic Albaqualfs

6.5

30

0.12

1.41

29

48

SL[c]

Floyd sandy loam

Fine‐loamy, mixed, superactive, mesic Aquic Pachic Hapludolls

6

4.1

0.21

2.07

n.a.

n.a.

LS

Loamy sand

Arenic Haplustalfs

7.4

2.1

0.01

0.15

n.a.

n.a.

TN = total nitrogen, and TC = total carbon. Amorphous soil mineral after McKeague and Day (1966). Soil from Iowa; all other soils are from Oklahoma.

TRANSACTIONS OF THE ASABE

Table 2. Soil properties of the artificial soils used in the sorption experiments. Soil Clay TN[a] (%) pH (%) Constituents

Soil ID

TC[a] (%)

C5P2

Silurian sand, kaolinite, and peat moss

4.4

5

0.03

0.75

C10P2


4.5

10

0.03

0.75

C20P2


4.3

20

0.03

0.75

C5P4


‐‐

5

0.06

1.51

C10P4


‐‐

10

0.06

1.51

C20P4


‐‐

20

0.06

1.51

C5P8


‐‐

5

0.11

3.02

C10P8


‐‐

10

0.11

3.02

C20P8


‐‐

20

0.11

3.02

C5

Silurian sand and kaolinite

5.4

5

0.02

0.00

C10


5.3

10

0.01

0.00

C20


5

20

0.01

0.00

SS

Silurian sand

‐‐

0

0.00

0.00

SSPt

Silurian sand and peat moss

4.5

0

0.03

0.75

SL‐TCR[b]

Natural soil; TC reduced

9.1

4.1

0.00

0.64

LS‐TCR[b]

Natural soil; TC reduced

8.4

2.1

0.08

0.15

[a] [b]

TN = total nitrogen, and TC = total carbon. TCR = natural soil treated to reduce total carbon.

Berkeley Springs, W.V.) and well crystalline kaolinite clay (KGa‐1b, The Source Clay Repository, Purdue University, West Lafayette, Ind.) at three different kaolinite clay and to‐ tal carbon contents (table 2). In cases in which total carbon was considered, sphagnum peat moss (Premier Horticulture, Inc., Quakertown, Pa.) was used. Before mixing, the peat moss was oven‐dried at 65°C for 24 h and sieved to pass a #200 sieve (74 mm openings) in order to mimic a fine soil par‐ ticle size composition. For all soils and peat (e.g., sieved frac‐ tion), total carbon and total nitrogen content were estimated using a TruSpec carbon and nitrogen analyzer (Leco Corp., St. Joseph, Mich.). The two treated natural soils (i.e., artificial soils) were the result of reducing the total carbon content in the loamy sand (LS) and sandy loam (SL) soils using H2O2 reagent, follow‐ ing the procedures described by Kunze (1965). A 120 g sam‐ ple of soil was placed in a glass container, and sodium acetate (NaOAc) was added to achieve a 1:1 soil to NaOAc volume ratio. An initial 30 mL of hydrogen peroxide reagent (30% concentration) was added, and the mixture was allowed to stand overnight. The following day, the suspension was stirred and heated in order to remove excess H2O2 while

[a] [b]

small H2O2 aliquots were added until the soil lost its black color. With the completion of the organic matter decomposi‐ tion, the pH was increased to 8 using Na2CO3, and then the solution was boiled for 10 min. After the solution cooled, a 1 N NaCl wash followed at an equivalent volume. The solu‐ tion was allowed to stand for one day or until a clear superna‐ tant was observed. The supernatant was then discharged, and the remaining soil was washed with distilled water twice and then allowed to stand. Once a clear supernatant was observed, the supernatant was poured out and the soil was oven‐dried at 65°C. ESCHERICHIA COLI SOURCE Liquid swine manure from the Swine Research and Edu‐ cational Center at Oklahoma State University was collected in a 5 d composite sample and stored at 4°C in a closed plastic container for one year. Storage of the swine effluent mim‐ icked anaerobic conditions, but not necessarily the tempera‐ ture, typically found in manure pits before manure application. Before the experiments, the stored manure was mixed with fresh liquid swine manure aliquots in order to in‐ crease the E. coli concentration to a desired concentration (e.g., approximately 20,000 most probable number (MPN) per mL) and mimicked the mixing cycle found in manure pits (e.g., aged manure mixed with fresh manure). Major ions were measured in the final effluent mixture us‐ ing inductively coupled plasma atomic emission spectrosco‐ py (ICP‐AES) and elemental digestion analysis (EPA 3051 method) for the liquid and solid components, respectively (table 3). Solids were separated after 90 min of centrifugation at 400g using 40 mL of liquid swine manure. The percentage of solids and liquid was estimated by weight after pouring the supernatant and evaporating the remained liquid at 65°C. SWINE MANURE PARTICLE SIZE AND ASSOCIATED E. COLI POPULATION Independent of the sorption tests, the particle size distribu‐ tion of the swine manure and its associated E. coli population were determined by serial filtration. The serial filtration tech‐ nique was used to determine the percentage of the fecal bacteria population associated to a specific particle size in the swine effluent. Note that swine effluents were not in contact with soils during the filtration experiments. Two filtration media (stainless steel and nylon mesh) were used due to par‐ ticular sorption effects that may occur as a function of the filtration material, pore geometry, and spatial distribution: (1) 1.27 cm diameter and 0.16 cm thickness stainless steel disks (Applied Porous Technologies, Inc., Tariffville, Conn.) with 53, 40, 20, and 10 mm openings, and (2) nylon mesh (BioDesign, Inc., Carmel, N.Y.) with 50, 35, 20, 15 and 8 mm openings. Liquid samples (1 mL) were acquired before and after each successive filtration stage for E. coli enumeration in triplicate experiments. Pressurized air or vacuum systems were used due to clogging in pore openings below 20 mm

Source[a]

%

Na

Table 3. Composition of the swine effluent utilized in the study. Ca Mg K S B P Fe Zn

Liquid AS (mg L‐1) Liquid NS (mg L‐1) Solid (mg kg‐1)

99 99 1

401 394 0.6

159 159 1.5

44.6 15.2 0.7

923 1029 0.9

818 346 0.6

1.2 1.2 n.a.

137 55.8 1.6

2.0 1.2 1580

0.74 0.20 1808

Cu

Mn

0.41 0.42 ‐0.05[b] ‐0.07[b] 367 305

Al

Ni

pH

0.11 0.10 n.a.

n.a. n.a. 9.0

7.5 7.5 n.a.

Liquid AS = swine effluent used in the artificial soils (AS), and Liquid NS = swine effluent used in the natural soils (NS). Measurements over the detection limit.

Vol. 55(1): 61-71

63

when using the porous metal disks or nylon mesh media. Al‐ though sorption and straining mechanisms are difficult to dif‐ ferentiate as the pore size opening of the filtration media decrease, the use of serial filtration served to estimate the bacteria population associated with specific particle sizes. In addition, note that the serial filtration conducted on swine ef‐ fluents was not used for comparison with sorption of fecal bacteria when in contact with soils (e.g., mixture of soil and swine effluent) but rather to assess the attached or free cell fecal bacteria population. SOIL SORPTION EXPERIMENTS Soil samples for E. coli quantification were prepared in triplicate by placing 6.0 g of air‐dried soil (i.e., at room tem‐ perature) in a 50 mL plastic centrifuge tubes. The soil was then mixed with 6.0 mL effluent (approximately 10°C at the time of mixing with the soil) prepared by diluting liquid swine manure in distilled water at five different dilution ra‐ tios (1:5, 2:4, 3:3, 5:1, and 6:0 mL of liquid swine manure to mL of distilled water). A total of 15 samples per soil type were prepared (tables 1 and 2). These sorption experiments were not conducted using the typical approach in which the solute or sorbent concentrations were changed while keeping other properties constant. In this case, other properties such as solution pH, ionic strength, and available exchangeable ions also changed with the bacteria loads. Even though soil samples were prepared from disturbed soils, the experiments were intended to mimic the top 1 to 2 cm of soil in contact with swine effluent following manure application. Different sorption conditions may occur deeper in the soil profile as ef‐ fluents are transported through the soil matrix, such as reduc‐ tion in solute concentration and organic compounds, ion exchange, and filtering, which fall outside the scope of this article. After soil sample preparation, E. coli sorption was quanti‐ fied using the following procedure. First, treated soils were resuspended using a vortex (Genie 2, G‐560, Scientific In‐ dustries, Inc., Bohemia, N.Y.) for 2 to 3 s and then allowed to stand for a 5 min equilibration time. This equilibration time was equivalent to that used by Ling et al. (2002) and sim‐ ilar to other reported sorption studies with fecal bacteria (Huysman and Verstraete, 1993). The equilibration time was short enough to minimize microbial activity (e.g., inactiva‐ tion, bacteria growth, competition, and predation) that may affect the initial fecal bacteria population, but long enough for the main ion exchange processes to occur following the mixing process. Note that effluents were used with their en‐ tire biological and chemical loads proportional to the dilution ratio. Second, treated soils were centrifuged for 3 min at 48g to decant clay particles from E. coli (i.e., differential centrifu‐ gation) as described by Ling et al. (2002). Note that the com‐ plete methodology proposed by Ling et al. (2002) was not followed as it implied a second treatment with a saline solu‐ tion quantifying “strong adsorption.” Tests conducted to quantify E. coli removal by centrifugation in swine effluents (no soil was added) rather than sorption indicated that less than 6% of the E. coli population was removed by gravita‐ tional forces during the 3 min centrifugation time. As an ex‐ ample, initial E. coli concentrations from six samples of swine effluent averaged 4.4 × 106 MPN per 100 mL with a standard deviation of 1.2 × 106 MPN per 100 mL; following a 3 min centrifugation at 48g, E. coli concentrations de‐ creased to 4.2 × 106 MPN per 100 mL with a standard devi‐

64

ation of 1.4 × 106 MPN per 100 mL. This removal (i.e.,5.2%) was insignificant in the computations. Subsamples from the supernatant were quantified for E.coli concentrations (MPN mL‐1) using IDEXX Colilert re‐ agent and Quanti‐Tray 2000 (IDEXX, Westbrook, Maine), an EPA‐approved method for E. coli quantification. Note that E.coli enumeration using Quanti‐Tray 2000 is based on fluorescence under UV light (Garbrecht et al., 2009; Guzman et al., 2009, 2010), and color in the sample should not impact E. coli quantification. Electrical conductivity and soil pH were measured after centrifugation (i.e., triplicate samples) for each dilution ratio and refrigerated for additional solution analysis of major ions (e.g., Na, K, Ca, Mg, S, P, B, Fe, Zn, Cu, Mn, and Al) using ICP‐AES. This research assumed that centrifugation did not lead to cell lysis due to collision be‐ tween E. coli (i.e., attached or planktonic) and soil particles. Sutera and Mehrjardi (1975) concluded from experiments with blood cells (6 to 8 mm size) under turbulent flow that shear stresses above 2,500 to 3,000 dynes cm‐2 may result in cell lysis. These shear stresses were equivalent to cells centri‐ fuged at approximately 400g to 480g. SAR, E. COLI QUANTIFICATION, AND NONLINEAR E. COLI SORPTION The sodium absorption ratio (SAR) was computed based in the ion concentration from each effluent dilution: [ Na + ]

SAR = [Ca

2+

] + [ Mg 2 + ]

(1)

where [Na+], [Ca2+], and [Mg2+] are the soil solution ion con‐ centration in mmol L‐1. Electrical conductivity (EC) and soil pH were measured from effluent dilutions after mixing with soils. Sorbed E. coli was computed and expressed per gram of dry soil as the difference in E. coli population in the initial ef‐ fluent volume (e.g., 6 mL) and the E. coli population in the supernatant after centrifugation: qSP =

V (Co − C LP ) M soil

(2)

where qSP is the sorbed E. coli concentration (MPN g‐1 soil), V is the effluent volume (mL), Co and CLP are the initial and supernatant E. coli concentrations (MPN per 100 mL), and Msoil is the mass of soil (g). Data from the triplicate experi‐ ments were averaged and then fit to the following nonlinear equation: B qSP = A ⋅ C LP

(3)

where A and B are empirical coefficients. This equation is mathematically equivalent to the Freundlich isotherm equa‐ tion. Note that qSP represents sorption of E. coli in the solid phase in MPN g‐1, and CLP is the concentration of E. coli in solution after centrifugation in MPN mL‐1. The coefficient of determination (R2) between the observed and predicted data was used to quantify the strength of the correlation. A set of equations was derived to estimate A based on soil properties. The percent clay content served as a key and/or explicit variable in the corresponding equation due to the fact that increased clay content resulted in increased E. coli sorp‐ tion. During the model derivation, a threshold in the percent clay content was identified to influence the estimation of A,


and therefore the soils were divided into two groups: (1) soils with percent clay content less than or equal to 11%, and (2)soils with percent clay content between 11% and 30%. Note that this threshold may change as more soils are incor‐ porated into the analysis. Finally, relationships between A and B were derived based on E. coli sorption from the natural soils. Derivation of the predictive A equations (i.e., two group of soils) was conducted using the natural soils dataset (e.g.,percent clay content and TC, and amorphous mineral and ion concentrations). The unknown coefficients of the nonlinear equations were solved by optimization techniques using the Solver add‐in in Microsoft Excel. The R2 value be‐ tween the observed and estimated A and the average absolute error (SSerr,avg ) between the observed and estimated sorbed fecal bacteria concentrations were used as the objective func‐ tions to assess the solution: SS err ,avg =

1 n

∑[(q

sp ) obs

− ( q sp ) est ]2

(4)

where n is number of observations from triplicate samples for each dilution ratio and natural soil, and (qsp )obs and (qsp )est are the observed and estimated sorbed E. coli concentrations in MPN g‐1, respectively.

RESULTS AND DISCUSSION

100

80

Average E. coli Size

Percent of E. coli Population Finer (%)

ATTACHED E. COLI IN SWINE EFFLUENTS The majority of the fecal bacteria population in the swine effluent was found to be particle‐associated rather than free cells (fig. 1). Assuming an E. coli size in the range of 0.8 to 3.0 mm (vertical dashed lines in fig. 1), the planktonic E. coli population (i.e., free cells) estimated from the linear trend line (fig. 1) was less than 10% of the initial effluent detectable E. coli population. Note that data from both the metal disks and the nylon mesh were combined to generate the single lin‐ ear trend line between opening size and percent of the E. coli population. The serial filtration indicated that at least 60% of the total swine effluent detectable E. coli population was particle‐associated with 8 to 50 mm particles sizes (i.e., likely organic compounds). Although these populations (i.e., planktonic and particle‐ associated) may change as a function of the particle size sepa‐ ration and E. coli enumeration techniques, attached bacteria populations commonly dominate planktonic populations in

60

40

2

R = 0.88

20

0 0.1

1

10

100

Opening Size (μm)

Figure 1. Particle size distribution of organic compounds in swine efflu‐ ents associated with the percentage of E. coli population finer than a cer‐ tain mesh size from microsieving experiments. Filled circles represent data from nylon mesh, and hollow circles represent data from metal disks.

Vol. 55(1): 61-71

oxygen‐depleted swine effluents (Leung and Topp, 2001). This may not be the case in aerated swine effluents in which aerobic planktonic populations rapidly overcome more stable anaerobic bacteria populations (Leung and Topp, 2001). Therefore, in aged manures, removal of attached E.coli from solution by sorption and/or flocculation/precipi‐ tation was potentially controlled by processes occurring in the substrate (i.e., organic compounds in which the bacteria were attached) or bacteria aggregates/clumping when in con‐ tact with soils. On the other hand, plaktonic E. coli were the potential bacteria population subjected to surface bonding (e.g., reversible or irreversible) during mixing. EFFLUENT SOLUTE CONCENTRATIONS The liquid swine manure was rich in organic compounds (e.g., aggregates and colloids) and solutes (table 3) that in‐ creased the pH and ionic strength of the soils' bulk solution (fig. 2a). Note that EC (e.g., associated to specific swine ef‐ fluent dilution ratios in fig. 2a) and ionic strength are com‐ monly linearly correlated. Following centrifugation, the pH in the bulk solution increased to greater than the initial efflu‐ ent pH (e.g., 7.5) in most natural soils (fig. 2a). Swine effluent constituents played an important role in sharply increasing the bulk solution pH as a function of the effluent mixing ratio and soil pH buffer capacity. In addition, note in figure 2a the pH variability for dilution ratios below or equal to 2:4 (e.g.,swine effluent to distilled water) compared to dilution ratios above 2:4. This may help explain further differences in sorption for some specific soils at low swine effluent dilution ratios. On the other hand, the bulk solution pH in the artificial soils did not exceed the initial pH in the swine effluent (fig.2a), perhaps due to the high pH buffer capacity of peat. Changes (e.g., increases and decreases) occurring in the soil solution pH following manure application have been re‐ ported from laboratory and field experiments as a temporary condition (e.g., weeks), which would be important in imme‐ diate sorption of bacteria to soils but is dependent on manure source (e.g., cattle or swine) and condition (e.g., fresh or aged). Changes in the bulk solution pH were explained by CO2 degasification (Husted et al., 1991; Safley et al., 1992) and aeration (Leung and Topp, 2001), as the swine effluent was initially shaken and increased in temperature during cen‐ trifugation: −

CO 2 + H 2 O ⇔ HCO 3 + H +

(5)

Essentially, a loss of CO2 from the system (i.e., degassing) forced the equilibrium in equation 5 to shift toward the react‐ ants (left side), which resulted in a decrease of solution pro‐ tons, thereby increasing the pH. As an example, Lovanh et al. (2010) documented greenhouse gas fluxes (e.g., methane, carbon dioxide, nitrous oxide, and ammonia) in fields follow‐ ing manure application. Note that CO2 and HCO3‐ are microbial byproducts in swine effluents; HCO3‐ is also found in the pig mucosal duodenum secretions (Odes et al, 1995). Moreover, ion exchange of Mg2+ associated with carbonates on clay minerals (fig. 2f) and Na+, K+, and Ca2+ in organic compounds from the effluent solution (figs. 2c, 2d, and 2e) may have contributed to the increase in pH. Other variables may play a role in pH changes, such as oxidation of volatile fatty acids and ammonia concentrations, which have been reported to buffer swine effluent pH (e.g., 6.3 to 7.5) under anaerobic conditions but favor increases in pH as

65

Figure 2. Soil solution pH, sodium absorption ratio (SAR), electrical conductivity (EC), and equilibrated E. coli concentrations (dashed lines in fig. 2a are the average EC per dilution ratio, i.e., mL of liquid swine manure to mL of distilled water), major ions sorbed (qSP ) by the natural soils and in the swine effluent, and concentrations (Cman ) of ions in the liquid manure effluent before contact with the soil (dashed lines in figs. 2c to 2h are the solute concentrations in the swine effluent before mixing with soil, and dots are the sorbed solute concentrations after mixing).

ammonia volatilization occurs (Georgacakis et al., 1982; Paul and Beauchamp, 1989). INFLUENCE OF CLAY, TOTAL CARBON, AND DISPERSION ON E. COLI SORPTION As expected, for the artificial and natural soils, higher clay content increased E. coli sorption directly but not proportionally (fig. 3). Figures 3a, 3b, and 3c present sorption data for artificial soils with clay contents of 5%, 10%, and

66

20%, respectively. Figures 3d, 3e, and 3f present sorption data for natural soils with 0% to 11%, 11% to 20%, and 20% to 30% clay contents, respectively. In general, the minimum and the range of sorbed E. coli concentrations increased for soils with higher clay content. These differences become less apparent when the clay content exceeded approximately 10%. In other words, the magnitude of the additional sorption decreased as the percent clay content increased to 10%.


Figure 3. Sorption of E. coli from swine effluents for the artificial soils with clay contents of (a) 5%, (b) 10%, and (c) 20%; and sorbed E. coli concentrations for the natural soils with clay contents in the range of (d) 0% to 10%, (e) 10% to 20%, and (f) 20% to 30%. See tables 1 and 2 for soil identification and soil properties. Dashed lines and regression equations correspond to the artificial soils with 5%, 10%, and 20% kaolinite clay contents and not total carbon added. Arrows indicate samples for which a sharp decrease in sorption was observed.

Increases in total carbon contents also increased sorption of E. coli for the 5% clay in the artificial soils (fig. 3a). The regression lines in figures 3a, 3b, and 3c were derived based on the artificial soil containing only Silurian sand and kaolinite. With peat (total carbon) added to the soils, sorbed E. coli concentrations increased for the same solution E. coli concentration (fig. 3a). Similar increases in sorbed E. coli concentrations for artificial soils with 10% and 20% clay content were not observed when adding total carbon (figs. 3b and 3c). Therefore, increased total carbon did not necessarily correlate to increased E. coli sorption depending on the clay content.

Vol. 55(1): 61-71

In some soils (e.g., C5P8, CA, EA, PA, and SA in fig. 3), sharp decreases in sorption (i.e., lower sorbed E. coli concentrations) were observed at the maximum effluent concentration (figs. 3a and 3f). The arrows in figures 3a and 3f indicate data points at which lower sorbed E. coli concentrations were observed at higher solution E. coli concentrations. Note that EC was an indicator of effluent application rates (fig. 2a). Soil dispersion (Gupta et al., 1984) and the presence of particle‐associated E. coli were hypothesized to explain these results. When soil dispersion occurs, particle‐associated E. coli move into soil solution and therefore are quantified as solution phase E. coli as opposed to sorbed E. coli. Although the bacteria are particle‐

67

associated, these bacteria are mobile and susceptible to transport in the liquid phase. Note in the natural soils (figs.3d to 3f) that this phenomenon was present on soils with high clay content, while in the artificial soils (figs. 3a to 3c) it was observed at the lowest clay content but higher total carbon content. These observations indicate the importance of soil mineralogy and total carbon (i.e., as an indicator of soil organic matter) on sorption of fecal bacteria from swine effluents. Humic substances are recognized to promote and stabilize soil aggregates, acting as binding elements favoring sorption. However, under sharp increases in the soil solution pH, humic substances can favor dispersion, which results in a decrease in sorption (Tarchitzky and Chen, 2002). In addition, humic substances coating soil clay minerals can suppress sorption due to ion competition (Tarchitzky and Chen, 2002; Yamaguchi et al., 2004; Huang et al., 2005). The sodium adsorption ratio (SAR) in the natural soils (e.g., approximately 12 after the 1:5 dilution ratio; fig. 2a) indicated that exchangeable Na+ and K+ (i.e., dispersing ions; figs. 2c and 2d) might be contributing to soil dispersion. Note that most of the Na+ and K+ added concentrations remain in solution after mixing with soils. On the other hand, concentrations of exchangeable ions Ca2+ and Mg2+ (i.e.,flocculating ions; figs. 2e and 2f) in solution decrease and increase proportional to EC, respectively. These ion exchange processes occurred as the solution ionic strength increased (e.g., as the dilution ratio decreased) and the pH increased (e.g., average of 8.2 after the 2:4 dilution ratio; initial pH in swine effluent was 7.5), further favoring dispersion. Dispersion of the clay minerals was visibly present in all soils to some degree as the swine effluent dilution ratio decreased. It was also hypothesized that dispersion of the effluent organic compounds or bacteria aggregates/clumping increased detection of E. coli in solution as particles carrying attached E. coli were physically split apart. Note that bacteria aggregates/clumping has been reported to cause an underestimation of bacteria enumeration when applied to actual manure conditions (Fries et al., 2006). However, further investigation is needed on this hypothesis.

A = 8.9(Al) + 5.8(Fe) + 0.8(clay) − 153.5(TC)

(7)

The correlation between the observed A derived from experimental data and the predicted A derived from equations 6 and 7 had a coefficient of determination of R2 = 0.92 (fig. 4a). Surprisingly, B was related to A based on the following equation (fig. 4): B = −0.1ln( A) + 0.9

(8)

This relationship had a coefficient of determination of R2= 0.94 (fig. 4b). The larger values for B (e.g., close to 1.0) were found for soils with percent clay contents less than or equal to 11%, except for the SL soil. Note that the SL soil (from Iowa) had been under long‐term treatment with swine effluent as a fertilizer, and changes in its soil properties may have occurred. Figure 5 compares the observed and predicted sorbed E.coli concentrations using equation 3 with the derived coefficients for the natural soils. The predicted sorbed E. coli concentrations were not biased toward soils with a specific clay content or total carbon content. Like any other non‐ physically based equation, this relationship applies for the soils and conditions investigated, and use outside of this range should be cautioned. Note that the artificial soils were not included in the regression, as their compositions were limited to assess the effect of total carbon and soil minerals

PREDICTIVE EQUATIONS FOR SORPTION OF E. COLI In both artificial and natural soils, the nonlinear equations proposed in equation 3 described sorption of multiple E. coli strains at different percent clay and total carbon contents. However, the nonlinear equations could not model the sharp decrease in sorption observed in some soils (e.g., CA, EA, PA, and SA; arrows in fig. 3f). Note that these equations and their coefficients are unique in that they characterize different soil and manure application suspensions. In addition, note that these coefficients are not physically based. Changes in A were related to soil properties such as percent clay content and TC, and amorphous Al and Fe minerals concentration. For soils with percent clay contents less than or equal to 11%, A was proportional to TC: A = 39.0(TC) 2.4

(6)

For soils with percent clay contents larger than 11% but lower than 30%, A depended on amorphous Al and Fe mineral concentration (mmol kg‐1), percent clay content (clay), and TC:

68

Figure 4. Estimated nonlinear coefficients (A and B) from the predictive equations developed in this research: (a) predicted versus observed nonlinear A for the natural soils in mL g‐1, and (b) relationship between the nonlinear equation coefficients (A versus B) for the natural soils. Soils after reduction in total carbon and artificial soils are included in the figure for comparison. See tables 1 and 2 for soil properties.


Figure 5. Comparison of estimated versus observed (average of triplicate experiments) sorbed E. coli concentrations (qSP in MPN g‐1) from the proposed predictive equation, and the theoretical perfect fit (1:1 line shown as dashed line). See table 1 for soil properties. Table 4. Statistical model comparison for the predictive model and the Ling et al. (2002) equation. Shaded rows indicate soils with percent clay lower or equal to 11%.[a] Observed Predictive Model Ling et al. (2002)

[a]

Soil BE CA CO DA DO EA PA PR SA SL LS

(qsp )avg

SD

(qsp )avg

SD

(SSerr )avg

(qsp )avg

SD

(SSerr )avg

4,365 3,954 4,382 3,116 2,328 4,953 4,061 1,920 2,741 1,816 1,691

2,356 1,613 1,837 1,575 1,333 2,500 1,587 858 1,346 578 811

3,063 3,571 3,445 2,814 2,961 2,697 4,031 2,988 4,653 3,064 840

1,305 1,498 1,501 1,337 1,446 1,279 1,513 1,676 1,403 1,288 629

1,771 682 1,065 497 720 2,645 1,095 1,376 2,393 1,435 909

8,891 36,263 19,058 12,815 7,873 28,351 71,226 6,839 53,630 781 221

6,714 32,343 17,592 8,475 4,965 26,548 73,616 4,676 47,762 721 165

6,218 42,584 20,473 11,541 6,443 42,584 93,537 6,050 66,366 1,090 1,585

Mean

3,212

1,490

3,102

1,352

1,326

22,359

20,325

27,134

See table 1 for soil identification: (qsp )avg is the average sorbed E. coli concentration, SD is the standard deviation, and (SSerr )avg is the average absolute error between observed and estimated sorbed E. coli (eq. 4). All values are in units of MPN g‐1.

on sorption of E. coli and were unrealistic when compared to field conditions. The dependence of the predictive equations on the percent clay content was expected. However, in soils with percent clay contents less than or equal to 11%, A was nonlinearly related to the percent TC. For other soils, A was directly correlated to the amorphous Al and Fe mineral concentration and the percent clay and TC. These results were not surprising due to the high variability found in cation exchange capacity (CEC) in soils with soil organic matter and clay contents lower than 5% and 30%, respectively. Organic

Vol. 55(1): 61-71

matter and amorphous Al and Fe coating clay mineral colloids usually modify the electrochemical surface properties of clays minerals (Zhuang and Yu, 2002). In natural conditions, increases in the soil solution pH increase the surface negativity of humic substances (e.g., organic matter) and significantly change the clay surface potential as a function of the soil mineral composition and coating. The average absolute error (eq. 4) between the estimated and the observed sorbed E. coli concentrations (qsp ) using equations 6, 7, and 8 was 1,326 MPN g‐1 (table 4). Note that

69

fitting the experimental data with individual isotherms for each individual experiment resulted in an average absolute error of 740 MPN g‐1. Across all soils, averaged sorbed E. coli concentrations (qsp ) were estimated by the predictive model to be 3,102 MPN g‐1 compared to the observed value of 3,212MPN g‐1. The predicted qsp exhibited a lower standard deviation than the observed qsp (1,352 MPN g‐1 versus 1,490MPN g‐1) (table 4). The proposed equations also decreased the error over the Ling et al. (2002) equation, which predicted an average qsp of 22,359 MPN g‐1 with a standard deviation of 20,325 MPN g‐1 (table 4). Residuals from the estimated E. coli sorption were normally distributed and proportional to the increase in the E. coli concentration in solution. This increase in error deviation was at least partially the result of the most probable number method used to quantify E. coli concentrations (Gronewold and Wolpert, 2008).

SUMMARY AND CONCLUSIONS Swine effluents are rich in solutes and organic colloids, and upon contact with soils result in varying sorption mechanisms for E. coli. Experiments conducted with free cells suspended in inert solutions may not be appropriate in this situation. Research conducted on permanently charged surfaces (e.g., sandy soils or glass beds) are not representative of actual agricultural fields. Generally speaking, soil clay minerals (i.e., variably and permanently charged) and organic matter (i.e., variably charged) are fundamental soil properties that may result in varied attractive or repulsive forces when in contact with swine effluents (i.e., solutions rich in exchangeable ions and other constituents that may sharply increase the soil solution pH). Sorption of attached E.coli was primarily controlled by sorption of the bacteria substrate (likely organic matter) or bacteria aggregates/ clumping rather than the bacterium itself. In addition, buildup in alkalinity of the bulk solution and ion exchange of solutes such Na+, K+, Ca2+, and Mg2+ may decrease sorption of E. coli in the presence of organic matter as a result of soil dispersion. Degassing of CO2 from the swine effluent when aerated via shaking or with increased temperature and ion exchange explained the increase in pH observed during the sorption experiments. Even though complex sorption mechanisms occurred when multiple‐constituent liquid swine manure was applied to the soils, sorption of multiple E. coli strains was represented by nonlinear equations, which were valid except in cases of soil dispersion, with a 35% average absolute error in sorbed E. coli concentration. Using equations that directly correlate E. coli sorption to percent clay content overestimated sorption of E. coli when applied to soils treated with swine effluents (average absolute error of 563% in sorbed E. coli concentration). Note that the proposed equations and their coefficients are unique in that they characterize different soil and manure application suspensions, rather than typical isotherms characterizing sorption for a constant bulk solution condition. The proposed equations will be useful when modeling sorption of fecal bacteria in soils (Guzman et al., 2012). The intent of this study was to decrease the uncertainty associated with predicting sorption of fecal bacteria when swine effluents come into contact with soils. Soil bacteria transport

70

is the result of multiple processes and mechanisms in which soil sorption of fecal bacteria is not a synonym of fecal bacteria immobilization, as transport of particle‐associated bacteria may occur following rainfall or irrigation events in agricultural lands. While not as reproducible as deriving equations for sorption from numerous experiments conducted using a single free‐cell E. coli strain, the proposed equations are intended to better represent the sorption of the detectable E. coli population from multiple strains. Care should be taken in using these equations when the manure sources and properties of the manure vary considerably. These equations should not be considered transferable to other environmental situations. As demonstrated, changes in environmental conditions (soil solution pH, exchangeable ions, etc.) control sorption mechanisms. Future research should be conducted that simultaneously investigate multiple strains of the E. coli population from manure sources. ACKNOWLEDGEMENTS The authors acknowledge the financial assistance of a 2007‐2011 USDA Cooperative State Research, Extension, and Education Service (CSREES) National Research Initiative Grant (Award No. 2007‐35102‐18242).

REFERENCES Bolster, C. H., B. Z. Haznedaroglu, and S. L. Walker. 2009. Diversity in cell properties and transport behavior among 12 different environmental Escherichia coli isolates. J. Environ. Qual. 38(2): 465‐472. Brown, P. A., S. A. Gill, and S. J. Allen. 2000. Metal removal from wastewater using peat. Water Res. 34(16): 3907‐3916. Choudhary, M., L. D. Bailey, and C. A. Grant. 1996. Review of the use of swine manure in crop production: Effects on yield and composition and on soil and water quality. Waste Mgmt. Res. 14(6): 581‐595. Dunne, W. M. 2002. Bacterial adhesion: Seen any good biofilms lately? Clin. Microbiol. Rev. 15(2): 155‐166. Eghball, B., D. Ginting, and J. E. Gilley. 2004. Residual effects of manure and compost applications on corn production and soil properties. Agron. J. 96(2): 442‐447. Foppen, J. W. A., A. Mporokoso, and J. F. Schijven. 2005. Determining straining of Escherichia coli from breakthrough curves. J. Contam. Hydrol. 76(3‐4): 191‐210. Fries, J. S., G. W. Characklis, and R. T. Noble. 2006. Attachment of fecal indicator bacteria to particles in the Neuse River estuary, NC. J. Environ. Eng. 132(10): 1338‐1345. Fuhrman, J. K. 2000. Phosphorus sorption and desorption characteristics of selected Oklahoma soils. MSc thesis. Stillwater, Okla.: Oklahoma State University. Gantzer, C., L. Gillerman, M. Kuznetsov, and G. Oron. 2001. Adsorption and survival of faecal coliforms, somatic coliphages, and F‐specific RNA phages in soil irrigated with wastewater. Water Sci. Tech. 43(12): 117‐124. Garbrecht, K., G. A. Fox, J. A. Guzman, and D. Alexander. 2009. E. coli transport through soil columns: Implications for bioretention cell removal efficiency. Trans. ASABE 52(2): 481‐486. Gee, G. W., and D. Or. 2002. The solid phase. In Methods of Soil Analysis: Part 4. Physical Methods, 278‐283. Madison, Wisc.: SSSA. Georgacakis, D., D. M. Sievers, and E. L. Iannotti. 1982. Buffer stability in manure digesters. Agric. Wastes 4(6): 427‐441. Gerzabek, M. H., F. Pichlmayer, H. Kirchmann, and G. Haberhauer. 1997. The response of soil organic matter to manure


amendments in a long‐term experiment at Ultuna, Sweden. European J. Soil Sci. 48(2): 273‐282. Gronewold, A. D., and R. L. Wolpert. 2008. Modeling the relationship between most probable number (MPN) and colony‐forming unit (CFU) estimates of fecal coliform concentration. Water Res. 42(13): 3327‐3334. Guber, A. K., D. R. Shelton, and Y. A. Pachepsky. 2005. Effect of manure on Escherichia coli attachment to soil. J. Environ. Qual. 34(6): 2086‐2090. Guber, A. K., Y. A. Pachepsky, D. R. Shelton, and O. Yu. 2007. Effect of bovine manure on fecal coliform attachment to soil and soil particles of different sizes. Appl. Environ. Microbiol. 73(10): 3363‐3370. Gupta, R. K., D. K. Bhumbla, and I. P. Abrol. 1984. Effect of sodicity, pH, organic matter, and calcium carbonate on the dispersion behavior of soils. Soil Sci. 137(4): 245‐251. Guzman, J. A., and G. A. Fox. 2012. Implementation of biopore and fecal bacteria fate and transport routines in the Root Zone Water Quality Model (RZWQM). Trans. ASABE 55(1): 73-84. Guzman, J. A., G. A. Fox, and J. Payne. 2010. Surface runoff transport of Escherichia coli after poultry litter application on pastureland. Trans. ASABE 53(3): 779‐786. Guzman, J., G. A. Fox, R. Malone, and R. Kanwar. 2009. Escherichia coli transport from surface‐applied manure to subsurface drains through artificial biopores. J. Environ. Qual. 38(6): 2412‐2421. Haynes, R. J., and R. Naidu. 1998. Influence of lime, fertilizer, and manure applications on soil organic matter content and soil physical conditions: A review. Nutr. Cycl. Agroecosyst. 51(2): 123‐137. Huang, P. M., M. K. Wang, and C. Y. Chiu. 2005. Soil mineral‐ organic matter‐microbe interactions: Impacts on biogeochemical processes and biodiversity in soils. Pedobiologia 49(6): 609‐ 635. Husted, S., L. S. Jensen, and S. S. Jørgensen. 1991. Reducing ammonia loss from cattle slurry by the use of acidifying additives: The role of the buffer system. J. Sci. Food Agric. 57(3): 335‐349. Huysman, F., and W. Verstraete. 1993. Water‐facilitated transport of bacteria in unsaturated soil columns: Influence of inoculation and irrigation methods. Soil Biol. Biochem. 25(1): 91‐97. Kim, H. N., and S. L. Walker. 2009. Escherichia coli transport in porous media: Influence of cell strain, solution chemistry, and temperature. Colloid Surfaces B 71(1): 160‐167. Kouznetsov, M. Y., Y. A. Pachepsky, L. Gillerman, C. J. Gantzer, and G. Oron. 2004. Microbial transport in soil caused by surface and subsurface drip irrigation with treated wastewater. Intl. Agrophysics 18(3): 239‐247. Kunze, G. W. 1965. Pretreatment for mineralogical analysis. In Methods of Soil Analysis: Part 1. Physical and Mineralogical Properties, 568‐577. C. A. Black et al., eds. Madison, Wisc.: ASA. Leung, K., and E. Topp. 2001. Bacterial community dynamics in liquid swine manure during storage: Molecular analysis using DGGE/PCR of 16S rDNA. FEMS Microbiol. Ecol. 38(2‐3): 169‐177. Ling, T. Y., E. C. Achberger, C. M. Drapcho, and R. L. Bengtson. 2002. Quantifying adsorption of indicator bacteria in a soil‐water system. Trans. ASAE 45(3): 669‐674. Lovanh, N., J. Warren, and K. Sistani. 2010. Determination of ammonia and greenhouse gas emissions from land application of swine slurry: A comparison of three application methods. Bioresource Tech. 101(6): 1662‐1667.

Vol. 55(1): 61-71

Mankin, K. R., L. Wang, S. L. Hutchinson, and G. L. Marchin. 2007. Escherichia coli sorption to sand and silt loam soil. Trans. ASABE 50(4): 1159‐1165. McKeague, J. A., and J. H. Day. 1966. Dithionite and oxalate extractable Fe and Al as aids in differentiating various classes of soils. Canadian Soil Sci. 46(1): 13‐22. Odes, H. S., R. Muallem, R. Reimer, S. Ioffe, W. Beil, M. Schwenk, and K.‐F. Sewing. 1995. Effect of somatostatin‐14 on duodenal mucosal bicarbonate secretion in guinea pigs. Digest Dis. Sci. 40(3): 678‐684. Pachepsky, Y. A., A. M. Sadeghi, S. A. Bradford, D. R. Shelton, A. K. Guber, and T. H. Dao. 2006. Transport and fate of manure‐ borne pathogens: Modeling perspective. Agric. Water Mgmt. 86(1‐2): 81‐92. Paul, J. W., and E. G. Beauchamp. 1989. Relationship between volatile fatty acid, total ammonia, and pH in manure slurries. Biological Wastes 29(4): 313‐318. Redman, J. A., S. L. Walker, and M. Elimelech. 2004. Bacteria adhesion and transport in porous media: Role of the secondary energy minimum. Environ. Sci. Tech. 38(6): 1777‐1785. Safley, L. M., J. C. Barker, and P. W. Westerman. 1992. Loss of nitrogen during sprinkler irrigation of swine lagoon liquid. Bioresource Tech. 40(1): 7‐15. Sutera, S. P., and M. H. Mehrjardi. 1975. Deformation and fragmentation of human red blood cells in turbulent shear flow. Biophys J. 15(1): 1‐10. Tarchitzky, J., and Y. Chen. 2002. Rheology of sodium‐ montmorillonite suspensions: Effects of humic subatances and pH. SSSA J. 66(2): 406‐412. ter Laak, T. L. 2005. Sorption to soil of hydrophobic and ionic organic compounds: Measurement and modeling. PhD diss. Utrecht, Netherlands: Utrecht University. Thomas, G. W. 1996. Soil pH and soil acidity. In Methods of Soil Analysis: Part 3. Chemical Methods, 475‐490. D. L. Sparks, ed. Madison, Wisc.: ASA and SSSA. Torkzaban, S., S. S. Tazehkand, S. L. Walker, and S. A. Bradford. 2008. Transport and fate of bacteria in porous media: Coupled effects of chemical conditions and pore space geometry. Water Resour. Res. 44: W04403, doi:10.1029/2007WR006541. Turner, J. C., J. A. Hattey, J. G. Warren, and C. J. Penn. 2010. Electrical conductivity and sodium adsorption ratio changes following annual applications of animal manure amendments. Comm. Soil Sci. Plant Analysis 41(9): 1043‐1060. Walker, S. L., J. E. Hill, J. A. Redman, and M. Elimelech. 2005. Influence of growth phase on adhesion kinetics of Escherichia coli D21g. Appl. Environ. Microbiol. 71(6): 3093‐3099. Winfield, M. D., and E. A. Groisman. 2003. Role of nonhost environments in the lifestyles of Salmonella and Escherichia coli: Minireview. Appl. Environ. Microbiol. 69(7): 3687‐3694. Yamaguchi, T., T. Takei, Y. Yazawa, M. T. F. Wong, R. J. Gilkes, and R. S. Swift. 2004. Effect of humic acid, sodium, and calcium addition on the formation of water‐stable aggregates in Western Australia wheatbelt soils. Australian J. Soil Res. 42(4): 435‐439. Yang, H. H., J. B. Morrow, D. Grasso, R. T. Vinopal, A. Dechesne, and B. F. Smets. 2008. Antecedent growth conditions alter retention of environmental Escherichia coli isolates in transiently wetted porous media. Environ. Sci. Tech. 42(24): 9310‐9316 Zhuang, J., and G. R. Yu. 2002. Effects of surface coatings on electrochemical properties and contaminant sorption of clay minerals. Chemosphere 49(6): 619‐628.

71

72
