APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Dec. 1994, p. 4500-45080099-2240/94/$04.00+0Copyright C 1994, American Society for Microbiology

Effect of Mineral and Organic Soil Constituents on MicrobialMineralization of Organic Compounds in a Natural Soilt

DAVID B. KNAEBEL,lt THOMAS W. FEDERLE,2* DREW C. McAVOY,2 AND J. ROBIE VESTAL'§

Department of Biological Sciences, University of Cincinnati, Cincinnati, Ohio 45221-0006,' andEnvironmental Science Department, The Procter and Gamble Co., Cincinnati, Ohio 452172

Received 16 May 1994/Accepted 11 September 1994

This research addressed the effect of mineral and organic soil constituents on the fate of organic compoundsin soils. Specifically, it sought to determine how the associations between organic chemicals and different soilconstituents affect their subsequent biodegradation in soil. Four "C-labeled surfactants were asepticallyadsorbed to montmorillonite, kaolinite, illite, sand, and humic acids. These complexes were mixed with a

woodlot soil, and "4CO2 production was measured over time. The mineralization data were fitted to variousproduction models by nonlinear regression, and a mixed (3/2)-order model was found to most accuratelydescribe the mineralization patterns. Different mineralization patterns were observed as a function of thechemical and soil constituents. Surfactants that had been preadsorbed to sand or kaolinite usually showedsimilar mineralization kinetics to the control treatments, in which the surfactants were added to the soil as an

aqueous solution. Surfactants that had been bound to illite or montmorillonite were typically degraded tolesser extents than the other forms, while surfactant-humic acid complexes were degraded more slowly than theother forms. The desorption coeflicients (Kd) of the soil constituent-bound surfactants were negativelycorrelated with the initial rates of degradation (k,) and estimates of "4CO2 yield (PO) as well as actual totalyields of "4CO2. However, there was no relationship between Kd and second-stage zero-order rates ofmineralization (ko). Microbial community characteristics (biomass and activity) were not correlated with any

of the mineralization kinetic parameters. Overall, this study showed that environmental form had a profoundeffect on the ultimate fate of biodegradable chemicals in soil. This form is defined by the physicochemicalcharacteristics of the chemical, the composition and mineralogy of the soil, and the mode of entry of thechemical into the soil environment.

Synthetic organic chemicals (xenobiotics) enter terrestrialenvironments as a result of intentional application, wastedisposal practices, accidental spills, land use patterns, andatmospheric fallout. Their fate in soils is a function of bothabiotic and biotic processes, which include transport andsorption, chemical catalysis, photodegradation, and biodegra-dation. Biodegradation of a chemical in soil is determined bythe presence and size of degrader populations, as well as

environmental conditions that affect the activity of thesepopulations and the bioavailability of the chemical.

Several studies have shown that various soil constituentsmay alter the biodegradation of chemicals in solution. Ben-zylamine degradation was substantially inhibited when incu-bated in the presence of montmorillonite (31), whereas diquatdegradation was severely inhibited in the presence of mont-morillonite clays but not in the presence of kaolinite clays (48).In contrast, Scow and Alexander (41) found substantial de-creases in the rates and extents of phenol and glutamatedegradation in the presence of kaolinite. Other workers haveshown that naphthalene or ot-hexachlorocyclohexane must

* Corresponding author. Mailing address: Environmental ScienceDepartment, The Procter and Gamble Co., Ivorydale Technical Cen-ter, Cincinnati, OH 45217. Phone: (513) 627-7296. Fax: (513) 627-8198.

t Dr. Vestal died of brain cancer in August 1992 at the age of 49. Wemiss him greatly and valued him as a colleague, mentor, and friend. Wehope that publication of his unfinished work will be a living memorialto his dedication to science.

t Present address: Microbiology, Molecular Biology and Biochem-istry Department, University of Idaho, Moscow, ID 83843.

§ Deceased.

desorb from the soil matrix before biodegradation can occur

(1, 36). Ogram et al. (33) likewise demonstrated that 2,4-dichlorophenoxyacetic acid had to be available in the soilsolution before biodegradation could begin but found thatboth sorbed and free bacteria could degrade the 2,4-dichloro-phenoxyacetic acid. In another study, the effect of the soilmatrix on chemical biodegradation was microorganism spe-

cific, such that one bacterial species was capable of metaboliz-ing sorbed naphthalene but another species required naphtha-lene to be in solution (13). In contrast to the aforementionedstudies, other studies have shown that biodegradation ofpentachlorophenol and toluene was not retarded by interac-tions with the soil matrix (3, 38).Many of these previously mentioned studies were performed

with dilute suspensions of the soil constituents, which are notrepresentative of typical soil conditions. Only a few studieshave assessed the effects of added soil constituents underenvironmentally realistic conditions. Bellin et al. (3) found thatsorption of pentachlorophenol to alkaline and acidic soilsincreased with increasing additions of sludge but that biodeg-radation was not affected. O'Connor et al. (32) reported thatsludge additions did not significantly affect either the adsorp-tion or biodegradation of a related chemical, 2,4-dichlorophe-nol, in calcareous soils. In other studies, addition of variouslevels of different clays to soils generally resulted in decreasedmineralization of a chemical to CO2 (25, 44, 45). These studiesdid not differentiate the effects of the added clay on overallmicrobial activity and on the bioavailability of the test chemi-cal. Furthermore, none of these studies of addition of sludgeand clays to soils have specifically addressed the interactionbetween the chemicals and the added constituents.The present study is unique in that organic chemicals were

4500

Vol. 60, No. 12

EFFECT OF SOIL CONSTITUENTS ON BIODEGRADATION

adsorbed to specific soil constituents before being introducedinto a natural soil and the incubation conditions reflected insitu conditions of temperature and soil water activity (aj).Experiments were designed to minimize changes in the soilchemistry and mineralogy while presenting the chemical to thenatural microbiota in a defined environmental form. Four"C-labeled surfactants with different physicochemical proper-ties were individually associated with purified soil constituents(sand, montmorillonite, kaolinite, illite, and humics), and smallamounts of these complexes were introduced into a naturalwoodlot soil. The mineralization kinetics of the complexedsurfactants were compared with one another and with thoseof controls, which were conventionally dosed with aqueoussolutions or dispersions of the chemicals. In addition, theparameters describing mineralization were compared with thedesorption coefficients (Kd) of the chemical from the soilconstituent.The four surfactants differed in their charge and hydropho-

bicity and therefore modeled many of the physicochemicalinteractions which can occur between organic chemicals anddifferent soil constituents. In addition to serving as excellentmodel compounds, surfactants have a wide range of consumerand industrial uses (35) and are increasingly important in theremediation technology of in situ soil washing (10, 28). Mostmodern surfactants are usually quickly biodegraded in surfacesoils, with half-lives of 3 weeks or less (19, 47), but somenonionic surfactants may have longer residence times in somesoils (46) and may affect microbial processes therein (22). Inaddition, the extent of mineralization of surfactants added tosoils has been shown to vary greatly as a function of soil type(19).

MATERLILS AND METHODS

Soil. Bonnell soil was obtained from a woodlot in HamiltonCounty, Ohio. The upper 5 to 8 cm was aseptically collected,partially dried under a stream of sterile air, and sieved. The6-2-mm fraction was used for all experiments. The soil was asilty loam and was mapped as a fine mixed, Mesic, TypicHapludalf (24). It had a pH of 7.13, an organic carbon contentof 4.7%, a cation-exchange capacity of 27.5 meq 100 g-' (2), asurface area of 49.6 m2 g-1 (14), and a clay content of 6.3%.The clay fraction was predominantly mica, with a moderatelevel of vermiculite and a trace of smectite.

Soil constituents. Montmorillonite (SWy-1), kaolinite, andillite were purchased from the Clay Mineral Society, Universityof Missouri, Columbia. The illite was pulverized prior to use.The sand was collected near Daytona Beach, Fla., and com-busted at 550°C for 4 h before use. Humic acid was purchasedfrom Aldrich Chemical Co., Milwaukee, Wis.

Chemicals. [U-ring-'4C]dodecyl linear alkylbenzene sulfo-nate (LAS) with a specific activity of 67.8 ,uCi mg-1 waspurchased from New England Nuclear, Wilmington, Del.[U-ethoxy-14C]dodecyl linear alcohol ethoxylate (LAE) (aver-age of 8.5 ethoxy residues per molecule) and [U-14C]sodiumstearate were purchased from Amersham, Arlington Heights,Ill. LAE had a specific activity of 5.4 ,uCi mg-', and stearatehad a specific activity of 196.8 ,uCi mg-'. [1-`4C]stearyl tri-methylammonium chloride (STAC) was synthesized at Procterand Gamble and had a specific activity of 12.5 ,uCi mg-'.[U-14C]acetate was purchased from Amersham; its specificactivity was 57 mCi mmol-1. The radiochemical purity of thesurfactants as determined by thin-layer chromatography wasgreater than 98%, except for sodium stearate, which had apurity greater than 96%. All other chemicals used were

LAS

LAE

XO.H8.5

STAC

cI

Na-Stearate* * * * * * . * *_CON4,~~~~~~~COO- Na+

* * * * * * * * *

FIG. 1. Chemical structures of the compounds used in this study.The location of the 14C is indicated by *.

reagent grade. Structural formulas of the surfactants are shownin Fig. 1.

Microbiological analysis. Subsamples of the soil were ana-lyzed for overall microbial biomass and activity at both theinitiation and termination of the experiments. Microbial activ-ity was determined by measuring the rate of ["4C]acetateincorporation into microbial lipids, using the method of Mc-Kinley et al. (29) with one modification. Instead of using a soilslurry, activity measurements were done at 100% water-hold-ing capacity to represent more realistic a, ranges in the soil.The concentration of ["4C]acetate was 42.6 jiM. Soil microbialbiomass was determined by measuring CHCl3-extractable lipidphosphate as described by Findlay et al. (11).

Preparation of chemical-soil constituent complexes. All soilconstituents were sterilized by autoclaving for 40 min at 121°C.A 4-g portion of each soil constituent was placed in a round-bottom (250 ml) flask with 4 to 12 ml of absolute methanolcontaining 14.5 ,ug of the 14C-labeled chemicals. This ratio ofsurfactant and soil constituent resulted in a final concentrationof approximately 50 ng of surfactant per g of soil in themineralization experiments amended with 100 mg of thecomplex. The flasks were placed on a rotary evaporator, andthe materials were mixed at 40°C. A vacuum was applied, andthe solvent slowly evaporated as mixing continued. The com-plexes were further dried aseptically in a vacuum oven andstored over Drierite in sterile containers. Alcohols were usedinstead of water to avoid high temperatures during the dryingprocess as well as to ensure sterility during the adsorptionprocess.

Specific activity was determined by placing a known quantityof each complex in 0.5 ml of methanol or ethanol in a 20-mlscintillation vial (Kimble), vortexing for 5 min, and mixing with15 ml of Scintiverse II (Fisher) and sufficient water to form agel. Radioactivity determinations were made on a Packard

VOL. 60, 1994 4501

4502 KNAEBEL ET AL.

Tri-Carb 2200CA scintillation counter, using an efficiencytracing protocol for quench correction. Specific activities of thebound surfactants closely agreed with theoretical specific ac-tivities, based on the total "4C activity added to each soilconstituent.

Mineralization assays. Mineralization of the bound surfac-tants was measured in serum bottle radiorespirometers asdescribed by Knaebel and Vestal (20). The surfactant-soilconstituent complexes were added at a low level (1.25%, wt/wt)to avoid altering the biology and chemical and physical char-acteristics of the soil. A small amount (100 mg) of thesurfactant-soil constituent complex was mixed with 8.0 (dryweight) of soil. Sterile deionized distilled water was then addedto bring the mixture to 70% gravimetric water-holding capac-ity. Controls were treated similarly, except that the surfactantwas dissolved or dispersed in the water used to adjust thewater-holding capacity. Four replicates of each treatment weretested. Abiotic controls were autoclaved three times andamended with 4.7% formalin and 30 mg of thimerosol kg-1 aspreviously described (19). The respirometers were incubated inthe dark at room temperature (ca. 22°C). Mineralization wasmeasured over a period of 2 months.

Desorption assays. Desorption coefficients (log Kd) weredetermined by standard techniques (la) adapted for desorp-tion applications. An aliquot of each complex was placed intosterile 10 mM NaN3 in acid-washed polypropylene or glasscentrifuge tubes. The tubes were vortexed briefly and placedhorizontally on an orbital shaker (120 rpm) at room tempera-ture (22 to 24°C). At predetermined intervals, the tubes werecentrifuged and a portion of the supernatant was removed forradioactivity determination. The complexes were resuspended,and the process was repeated until the level of radioactivity inthe supernatant remained constant. Equilibria were assumedwhen the radioactivity in the supernatant was constant. Thisusually occurred within 2 to 4 h, but experiments were con-ducted for 24 to 48 h. The Kd was calculated from the ratio ofradioactivity remaining bound to the soil constituents (dpm permilligram) to radioactivity in the supernatant (dpm per milli-liter). The former was estimated by subtraction of the super-natant value from the total added radioactivity.Data analysis. Mineralization results were fitted to first-

order (23) and 3/2-order mineralization models (7). The twomodels are similar, but the 3/2-order model has additionalterms for growth and a zero-order mineralization rate.The first-order model has the following form: P = PO [1 -

e(-klt)], where P is the percentage of compound mineralized attime t, PO is the asymptotic percentage of compound convertedto 14C02, and k1 is the first-order rate constant (day-1). The3/2 model (with linear growth term) has the following form: P= Po {1 - [-kt - (k2t2)/2]} + (kot), where P is the percentageof compound mineralized at time t, PO is the percentage ofcompound converted to 14Co2 during first-order metabolism,k, is a proportionality rate constant (day-'), k2 is a lineargrowth rate term, and ko is a zero-order rate constant (percentday- ). On the basis of the small amounts of labile carbon (i.e.,the surfactants) which were added to the soil (50 ng g-1), nogrowth was assumed and k2 was removed from the equation(7), to give the following: P = PO [1 - e( klt)] + (kot).The mineralization data were fitted to the models by the

Quasi-Newton minimization technique in the NONLIN mod-ule of Systat, Evanston, Ill. The model that best fit themineralization data was determined by the F-test procedureoutlined by Robinson (37). The GLM module of SAS Institute,Cary, N.C., and the MGLH module of Systat were used forstatistical analyses. Comparisons within treatments (i.e., sur-

400

-o

0

-j

Q)0Ec

(a)300 1

200 F

100 [

0

16

-o-0

IN

tLC

0q0

12

8

4

A

HY

Initial Mont Kaol Hllit Sand Hum Control

FIG. 2. Microbial biomass and activity in the Bonnell soil beforeand after the mineralization experiments. Data are shown as mean +

1 standard deviation. (a) Microbial biomass, measured as nanomolesof lipid phosphate per gram (dry weight) (gdw) of soil (n = 3). (b)Microbial activity of the soil before and after the mineralizationexperiments, expressed in terms of soil mass (n = 4). Initial, initialbiomass of the soil prior to mineralization experiments; Mont, mont-morillonite; Kaol, kaolinite; Illit, illite; hum, humics.

factants or soil constituents) were made by the Scheffe multi-ple-comparison test (9, 40, 43).

RESULTS

The Bonnell soil had an initial microbial biomass of 141nmol of lipid phosphate g (dry weight)-', which representsapproximately 2.8 mg of biomass carbon g (dry weight)-1 (49),and an initial activity of 2.97% [14C]acetate incorporated intomicrobial lipids h-Vl g (dry weight)-' (Fig. 2). The biomass andphysiological activity of this soil were similar to those of othersoils in this geographical region (19). After the mineralizationexperiments were terminated, mean microbial biomass andactivity were elevated compared with the initial values in mosttreatments, including the control (Fig. 2). This increase inmean value was also accompanied by increased variabilitywithin each treatment. Hence, the biomass in treatmentsamended with surfactant-soil constituent complexes was notsignificantly different from that in the control treatments (P 30.05) (Fig. 2a). Similarly, microbial activity, whether expressedin terms of soil mass (Fig. 2b) or microbial biomass (data notshown), did not exhibit any significant treatment effects (P >0.05).

In 92% of the cases, the 3/2-order model showed signifi-cantly better fit to the mineralization data than the first-ordermodel on the basis of F-tests that control for the extra terms inthe model (40). Two examples of this better fit are shown inFig. 3. This model estimates an initial exponential rate con-stant (k1), a yield parameter (PO), a linear (or exponential)

APPL. ENVIRON. MICROBIOL.

77///////////// fi//

T

.,K

EFFECT OF SOIL CONSTITUENTS ON BIODEGRADATION 4503

30

20

Q)N 10

0L.U)

0

C: 50Q00

U) 400L

30

20

1 0

0

v Raw DataO First order model fit0 3/2 order model fit

0 2 4 6

0 1 0 20 30 40 50 f

- 0

&' &

VIC

1+

_ S .s Dn

CD

(0 0

CD

CD

_ _ .0D

_. _

CD

= a c

5' O

CD

(-'_

CD3

CD00'

0' --w CDCA

CD

CD3 2

CD5C&

.

CD CD

CDOCA

D

O z

60

Time in Days

FIG. 3. Comparison of the fit of 3/2-order and first-order models tomineralization data. The mineralization data were from the stearatecomplexed with humics (a) and sand (b). The inset shows the details ofthe first 8 days of mineralization.

growth parameter (k2), and a linear mineralization rate (ko).The k2 estimates were usually zero in this study (data notshown), so no growth was assumed. Therefore, the growthterm was removed from the model (7). This was consistent withthe minimal changes in biomass during the mineralizationexperiments (Fig. 2a) and the low level of added surfactants(50 ng g').

Because of the error minimization calculations used by thenonlinear regression algorithm, the estimates of equivalentparameters of the 3/2-order model differed in a consistentfashion from those of the first-order model. The general trendswere that the initial rate estimates (kl) calculated by the3/2-order model were slightly higher than the estimates of thefirst-order rate constant and that the yield estimates (PO)tended to be lower than the asymptotic yield estimates of thefirst-order model. However, the PO values for the 3/2-ordermodel are unique in that they indicate the point at which themineralization rate changes from a pseudo-first-order reactionto a zero-order reaction. They do not indicate the asymptoticyield of CO2. In most circumstances, PO represented 70% (forstearate and STAC) to 80% (LAE and LAS) of the total finalpercentage recovered as 4'CO2 (Table 1). For this reason, thetotal amounts of 14Co2 evolved after 60 days are included inTable 1.

CD 0 e- = 03

F'

p

I+C

I+cC

-J

I+C

C

I+

C

I+

I+C

p

I+C

I+C

I+

l+

C

I+C

I+

l+

00 ----

1± 1± 1+ 1±1+ 1+

O kOO .O

00 j C

1+ 1+ 1+ 1+1+ 1+oc =nc

I+ I+ I+ I+ It l+C C oCo C o

o~- ~cro o

00

1+ 1+ 1+ 1+1+ 1+

00- 0 "

It l+ I+ I+ I+ I+

00 C o o"C

1+ 1+ 1+ 1+ 1+1+

p p pppp

C,0 C-a

1+ 1+ 1+ 1+1+ 1+C C CCCC

o0 -oo oCo

I+ I+ I+ I+ I+ I+

1+ 1+ 1+ 1+ 1+ 1+

C C oCCoC

1+ 1+ 1+ 1+ 1+ 1+

0 o owoC

I+ I+ I+1I+I+1I+

o Co

I+ I+ I+1I+I+ I+

pn pn ppppo

I+ I+ I+ I+I+1I+

kO - o£e

I+ I+ I+ I+I+1I+C o C-C-o

I+ I+ I+1I+I+1I+C o Jo-o oo

I+I+ I+ I+1±++I+C C CCC

C .~ .C.

i+ I+ I+1I+I+1I+C C CCCCn

l+ I+ I+ I+I+ I+oJ C CCCCoo. 00 oJo o

I+ I+ I+ I+1I+±I+- o C Co o-*00 _ . _

VOL. 60, 1994

vr-_L.

(D0'

0

CD(0

0'

(D

t'

0s.0

(0

0'0:0D(0

(0

00

ci,H

F.'<3CL P

HNr. 3-E E

c-

;rlCD

4504 KNAEBEL ET AL.

50

40

30

20

Q) ~~~~~~~~~~~~2010z 1 00

j J1o00 1 2 3 4

c 80

FIG 4.)MnrlztoofbudofreLE()adLAS()i

70-16050~~~~~~~~~~630 T

60 -20 2

10/ ~~~~~~~01 2 34 50

0 10 20 30 40 50 60

Time in DaysFIG. 4. Mineralization of bound or free LAB (a) and LAS (b) in

the Bonnell soil, shown as the cumulative production of 14Co2 overtime. Each line represents four replicates. Error bars indicate 1standard deviation. The initial 4 to 5 days of mineralization are shownin the insets. The soil constituents to which the surfactants were boundare shown in Fig. 3a. Symbols: 0, montmorillonite; *, illite; V,kaolinite; V, sand; 1, humics; *, control.

The environmental form of the surfactants had variable butsignificant effects on the rate and extent of mineralization ofthe radiolabeled chemicals. In some instances, different chem-icals also exhibited distinct mineralization kinetics when asso-ciated with the same soil constituent. Figure 4 shows evolutionof 14C02 from the various chemical-soil constituent complexesas a function of time. For LAE, the total yield of 14C02 duringthe experiment was similar (-55%) when LAE was associatedwith kaolinite, sand, or humic acid or added to the soil inaqueous solution (Fig. 4a; Table 1). The yield was significantlylower when LAE was bound to illite (43%) or montmorillonite(34%). During the initial 4 days, LAE added as an aqueoussolution was mineralized more rapidly (k1 = 0.58 day-') thanwas LAE added in the bound form (Fig. 4a inset). The k, forLAE bound to montmorillonite, kaolinite, illite, or sand wasapproximately 0.45 day- 1, while that for LAE bound to humicacid was 0.23 day-'. Notably, mineralization of the LAE-humic acid complex exhibited a sigmoidal mineralization pat-tern. The ko ranged from 0.12% day-' for montmorillonite-bound LAE up to 0.21% day-' for kaolinite-bound LAE.The microbial mineralization of the various environmental

forms of LAS was different from that observed for the analo-gous forms of LAE (Fig. 4b). The total yield of 14C02 for LASafter 60 days was greatest when LAS was associated with

a)

. _40

6L)c

a)U

6)

10

0

20

15

1 0

50 5

0 A

-5 01 3 5

0 10 20 30 40 50 60Time in Days

FIG. 5. Mineralization of bound or free stearate (a) and STAC (b)in the Bonnell soil, shown as the cumulative production of '4Co2 overtime. See the legend to Fig. 4 for explanation of symbols.

kaolinite (75%), sand (70%), or illite (68%). The yield waslower when LAS was added to the soil as an aqueous solution(56%) or when bound to montmorillonite (59%) or humicacids (48%). Notably, mineralization of the LAS-humic acidcomplex was preceded by a 3-day lag. The initial rate of LASmineralization (k1) was similar for most of the environmentalforms (k1 0.55 day-'), except for the LAS-humic acidcomplex, which was mineralized much more slowly than theother forms (k1 0.19 day-') (Fig. 4b inset; Table 1). The koof all forms of LAS was relatively constant and averaged 0.23%dayf1

Stearate mineralization in many ways mimicked that of LAS;however, the total yield of 14C02 from LAS was on average1.8-fold higher than that for stearate (Fig. 5a; Table 1). Theyield of 14C02 was highest when stearate was associated withkaolinite (43%), sand (47%), or illite (43%) or added as anaqueous solution (42%). The yield was lower when stearatewas bound to montmorillonite (39%) or humic acid (29%). Inall cases, mineralization of stearate was not preceded by a lagand did not exhibit a sigmoidal pattern. With the exception ofthe humic acid complex, the initial rates of mineralization of allforms of stearate were similar (k1 = 0.20 to 0.25 day-'). The k1of the humic acid complex was 0.15 day-'. These initial rateswere lower than those observed for LAE or LAS. The ko wasagain relatively constant across all treatments (0.24 to 0.29day-'), except for the montmorillonite (0.08 day-') and humicacid (0.16 day-1) complexes.

APPL. ENVIRON. MICROBIOL.

EFFECT OF SOIL CONSTITUENTS ON BIODEGRADATION 4505

TABLE 2. Desorption coefficients (Kd) for the surfactant-soil constituent complexesa

Kd for:Soil constituent

LAE LAS Stearate STAC

Sand 98 41 479 794Kaolinite 162 155 1,413 2,399Illite 141 66 676 1,258Montmorillonite 98 22 33 275Humic acid 135 1,698 2,884 2,239

Mean 123 110 832 1,071

aKd is calculated as: (GJ/B)1C5, where G5 is the total quantity of materialdesorbed (micrograms), B is the dry weight of soil (or soil constituent), and C. isthe concentration of chemical in solution at equilibrium (micrograms permilliliter).

The yield of 14CO2 from STAC was the lowest for all thematerials tested, regardless of environmental form (Fig. 5b).Like stearate, yields were greatest and comparable whenSTAC was added to the soil as an aqueous solution (20%) oras a kaolinite (21%), sand (19%), or illite (19%) complex(Table 1). They were substantially lower for humic acid (9%)or montmorillonite (7%) complexes, which also had the lowestinitial mineralization rates (k1 0.12 day-') compared with

the other forms (- 0.20 day-1). The ko values for STACmineralization were also the lowest observed with any of thechemicals. Montmorillonite and humic complexes had ko val-ues of approximately 0.05% day-', while those of the otherforms averaged 0.11% day-'.

Table 2 shows the desorption coefficients (Kd) for thevarious chemical-soil constituent complexes. The Kds for LAEwere similar for all soil constituents (98 to 162 liters/kg). Incontrast, those for LAS were highly variable, ranging from 22liters/kg for montmorillonite to 1,698 liters/kg for humic acidcomplexes. On average, the desorption coefficients for thestearate and STAC complexes were considerably greater thanthose of other two surfactants. The Kds for the humic acid-bound surfactants were usually the largest for any of the soilconstituent-surfactant complexes, and the Kds for montmoril-lonite- and sand-bound surfactants were usually the lowest.The low values for montmorillonite were unexpected, given itslarge surface area. However, because of the use of alcoholrather than water during the adsorption process, this 2:1 claywas not fully expanded and many interstitial sites were notavailable.

Figure 6 shows the relationships between the desorptioncoefficient (log Kd) and the various mineralization param-eters. For all of the surfactant-soil constituent complexes, logKd was negatively correlated with k1 (r = -0.862; P S 0.01),PO (r = -0.663; P - 0.01), and the total recovery of 14CO2(r = -0.620; P s 0.01) but was not significantly related to ko(r = -0.182; P - 0.05). For LAS and STAC, k, was inverselyrelated to the desorption coefficient of the various soil constit-uent complexes. No such relationships, however, existed forthe LAE and stearate complexes (Fig. 6a). More importantly,the average k1 and log Kd for each compound was inverselycorrelated. Thus, the differences among chemicals as well as

within the STAC and LAS treatments accounted for theobserved correlation between k1 and log Kd. In contrast, thenegative correlation between PO and log Kd appears to belargely dependent on the differences between the surfactantsand not to be due to the soil constituents to which they were

bound.The microbial characteristics of the soil (activity and bio-

0.8

0.6 F

0-

>0IxaE

0.4 -

0.2 1

0.0

80

60 F

a

40

20 [

0

0

b.

a

c:a

2 3 4 1 2 3 4

log(Kd)FIG. 6. Relationship between the mineralization kinetic parame-

ters (kl, ko, PO, and final yields of 14CO2) and log Kd. (a) Log Kd versus

the initial mineralization rate, k1 (day-'); (b) log Kd versus detach-ment-dependent rate, ko (% day-'); (c) log Kd versus yield component,P0 (%); (d) log Kd versus final yields of "4Co2 (%). Confidenceintervals (99% level) are shown. The different surfactants are identi-fied in panel b.

mass) were not correlated with any of the mineralizationkinetic parameters or total yields of 14CO2 (P : 0.05 in allcases).

DISCUSSION

These experiments demonstrated that under realistic soilconditions, the microbial mineralization of organic chemicalsin soil is controlled in part by interactions between thechemicals and the soil matrix constituents. These interactionsaffected both the yields of 14CO2 from radiolabeled chemicalsintroduced into soils and the kinetics of mineralization. Whileother studies have demonstrated that purified soil constituentscan adsorb and potentially affect the subsequent degradationof chemicals, most of these studies have not been performedunder typical soil solution conditions with intact soil commu-nities. Commonly, the adsorption and desorption of chemicalsis examined in dilute solutions of the soil constituent, which donot represent the typical soil solutions or soil water activity(a,) (6, 12, 31, 38, 39, 41). The unique value of this study is thatmineralization of chemicals in known environmental forms wasmeasured under realistic aw conditions, at which the biological,chemical, and physical characteristics of a natural soil were

largely preserved.Mineralization in this study exhibited multiphasic kinetics,

consistent with multiple controlling factors. Typically, themineralization data were better described by the 3/2-ordermodel than the first-order model. The improved fits resulted

(a)

*--.. '-.I

..I...5a

O LAE * LAS (b)o STAC * Stearate

O-...* .s

o ......

o o3

(c)

Io....0.ua

oL 0..I

(d)

0

**Sv---.... , .. .o . s ...*-l- ....v1.,0 D

0 b

VOL. 60, 1994

4506 KNAEBEL ET AL.

from the inclusion of the ko term in the 3/2-order model, whichdescribes a zero-order process that follows the initial first-order phase of mineralization (7). Brunner and Focht (7)found that ko was constant for a variety of substrates in any onesoil and suggested that it describes the indigenous rate ofcarbon release in a soil. Others have concluded that biphasicbiodegradation kinetics are related to desorption, since modelsthat have a desorption term typically provide better fits to soilbiodegradation data (17, 21, 30, 42). Furthermore, biphasicdesorption kinetics have been frequently observed for organicchemicals in soil and are characterized by a relatively quickinitial release of the chemical from the soil or soil constituent,followed by a slower desorption phase (4, 18, 30, 34, 38). In thepresent study, it is likely that ko was indicative primarily of slowdesorption of the "4C substrates from the soil matrices. Thisconclusion is based on observations that the ko values varied asa function of the surfactant and the soil constituent to which itwas bound. The ko was greatest when the surfactants had beenpreadsorbed to sand or kaolinite and lowest when they hadbeen bound to humics or montmorillonite.

Association of the surfactants with soil minerals (sand andclays) led to variable effects on the mineralization kineticparameters. Sand and kaolinite complexes exhibited some ofthe highest rates (k1 and ko) and extents (PO and total yields of14C02) of mineralization, while illite and montmorillonitecomplexes generally exhibited lower PO and total yields of14GO2 but similar initial mineralization rates (k1). PO values forthe illite complexes were 9 to 34% lower than those for thecorresponding kaolinite complexes, with LAE and STACexhibiting the greatest treatment effect (Table 1). Montmoril-lonite had an even more dramatic effect on PO, with 6 to 68%lower values for the montmorillonite complexes than for thecorresponding kaolinite complexes. Notably, k1 for all claycomplexes were similar to each other and to those for sandcomplexes. In contrast, ko values for the montmorillonitecomplexes of LAE, stearate, and STAC were approximatelyhalf of what was observed for the corresponding illite, kaolin-ite, and sand complexes.These observations are consistent with different degrees of

bioavailability related to different types of association betweenthe chemicals and the soil minerals. One fraction of thechemical appeared to be readily bioavailable and rapidlymineralized, and the rate at which this fraction was mineralizedwas described by the initial mineralization rate (k1). Signifi-cantly, k1 and the desorption coefficients (Kd) were significantlynegatively correlated with each other. The relative size of thisreadily mineralized fraction was reflected by the magnitude ofPO, which also was inversely correlated with Kd. A secondfraction was more slowly mineralized, and its mineralizationappeared to be equilibrium dependent. This equilibrium de-pendence was reflected in the similarity of the ko values of thekaolinite, illite, and sand complexes to those of the treatmentsinvolving aqueous solutions. A third fraction, exclusive tomontmorillonite, was much more slowly mineralized and notentirely equilibrium dependent. The predominance of thisform was reflected in the much lower ko rates for the mont-morillonite complexes of LAE, stearate, and STAC.

Attributing the observed mineralization kinetics to the de-sorption behavior of the chemicals is consistent with thedifferent nature of the sand and clays involved. Sand hasexclusively external binding sites. Kaolonite and illite arenonswelling 1:1 clays, whose binding surfaces are primarilyexternal. Illite has 3 to 5 times the ion-exchange capacity and5 to 10 times the surface area of kaolinite (5). In contrast,montmorillonite is an expandable 2:1 clay, with 2 to 7 times theion-exchange capacity and 3 to 8 times the surface area of illite,

but approximately 80% of the binding surfaces are interstitial(5). Differences in PO among the clay complexes probablyreflect the relative abundance of binding sites, whereas thelower ko values exhibited by montmorillonite complexes ofLAE, stearate, and STAC were probably due to much slowerdesorption and decreased bioavailability of chemicals whichmigrated into interstitial surfaces following their hydration inthe soil. Notably, the ko of the LAS-montmorillonite complexwas similar to that of all other LAS complexes. This uniquebehavior for LAS is probably associated with its size-relatedexclusion from the interstitial spaces of montmorillonite.While the linear surfactants (LAB, stearate, and STAC) havesmall enough diameters (<10 A [<1 nm]) to migrate into theinterlayers (10 to 20 A [1 to 2 nm]), the benzene ring(diameter, 35 A [3.5 nm]) makes this migration nearly impos-sible for LAS. A similar observation (exclusion of a chemicalfrom within the interlayer spaces of montmorillonite) wasmade by Ogram et al. for 2,4-dichlorophenoxyacetic acid (33).

These results expand on and are generally consistent withexperiments that demonstrate the ability of clays to inhibitbiodegradation of organic compounds. They are consistentwith the observations that benzylamine degradation by aPseudomonas sp. was slower when montmorillonite (1 g li-ter-1) was present (31) and that nucleic acids bound tomontmorillonite were generally oxidized to lesser extents thanthose bound to kaolinite or illite (16). They were not, however,consistent with the observations of Scow and Alexander (41)that the degradation of phenol, glutamate, and p-nitrophenolwas slower when they were incubated in aqueous solutionscontaining kaolinite. This incongruity demonstrates the differ-ent results that can be observed when testing biodegradationprocesses at realistic soil a, in contrast to testing them in dilutesolutions.The mineralization of the surfactants was consistently inhib-

ited when they had been preadsorbed to humics. Whereasmost of interactions between the chemicals and clays (or sand)were probably ionic or involved hydrogen bonding, those withthe humics were probably hydrophobic and in some instancesmay have involved covalent bonding (8, 27). The initial min-eralization rates of all chemical-humic acid complexes weremuch lower than those of the other complexes. In addition, theonset of humic acid-LAS complex mineralization was precededby a 3-day lag period, and mineralization of LAE exhibited asigmoidal pattern. Furthermore, the desorption coefficients forhumic complexes of LAS, stearate, and STAC were muchgreater than those of the other complexes. These complexesexhibited much lower P0 values than the other complexes,which suggests a smaller pool of readily bioavailable chemicalin the humic complexes. These results agree with other reportsthat showed that nonbiological binding events or enzymaticincorporation between chemicals and soil organic matterwould lead to much longer persistence in the environment (8,15, 26, 27).The log Kd values of the compounds and soil constituents

were negatively correlated with the kinetic parameters k1, PO,and total measured yield (Fig. 6). These correlations suggestthat the greater the affinity of a chemical for the environmentalmatrix, the less it will be available to the indigenous degradingpopulations. These relationships, however, should be viewedwith caution, because a short-term equilibration constant(determined over 24 to 48 h) is being compared with alonger-term (60 day) process. Furthermore, in the short-termKd determinations, the surfactants partition between a largeamount of solution and a small amount of soil constituentsolids. In the biodegradation experiments, the surfactantspartition into a much smaller volume of soil solution and are

APPL. ENvIRON. MICROBIOL.

EFFECT OF SOIL CONSTITUENTS ON BIODEGRADATION 4507

available not only to the microorganisms but also to other soilprocesses. Since these Kd measurements did not take intoconsideration the full complexity of the interactions of achemical with the various components of soil, it should not betaken for granted that sorptive chemicals with high measuredKd will not degrade as extensively or rapidly as less sorptivechemicals.-Laboratory studies that examine the biodegradation of a

chemical in soil usually are conducted by adding the chemicalto a soil as an aqueous solution or dispersion or by addinginsoluble chemicals to the soil in an organic solvent. Theresults of this study indicate that the soil constituents withwhich the chemical becomes associated can have a profoundeffect on the observed results. As a consequence, to mostaccurately assess the biodegradation of a chemical in a soil, itis important that the chemical be added in the same mannerthat it normally enters the soil environment. Thus, chemicalsthat enter soil in association with sewage sludge, such assurfactants, are best studied preassociated with sludge. Simi-larly, the fate of other chemicals that enter soils (e.g., pesti-cides) should be studied in the same form in which they arecustomarily applied.

In summary, a chemical in a soil environment can be insolution, sorbed weakly or tightly to inorganic and organic soilconstituents, or covalently incorporated into soil organic mat-ter, such as humics. The bioavailability of the chemical forbiodegradation will depend on which of these environmentalforms is predominant. How the chemical is distributed amongthese forms is determined by its physical and chemical char-acteristics, its mode of entry, and the composition and miner-alogy of the soil. The ultimate fate of a chemical in soil is theresult of several competing processes: biodegradation, irre-versible sorption, humification, and diffusion into interstitialspaces not accessible to microorganisms. The outcome of thiscompetition is dependent on the kinetics of these variousprocesses and the points from which they begin. These startingpoints are defined by the mode of entry of the chemical intothe soil matrix and its form at the time of this entry.

ACKNOWLEDGMENTS

We thank Carl Johnston, Jack Pretty, and Joseph Caruso forassistance with the inductively coupled plasma emission spectroscopyanalysis; B. Grosser, C. Johnston, and D. Moll for helpful discussions;and C. Bollinger for editorial assistance.

This work was supported by a grant from the Environmental ScienceDepartment of the Procter and Gamble Company.

REFERENCES1. Al Bashir, B., T. Cseh, R. Leduc, and R. Samson. 1990. Effect of

soil/contaminant interactions on the biodegradation of naphtha-lene in flooded soils under denitrifying conditions. Appl. Micro-biol. Biotechnol. 34:414-419.

la.American Society for Testing and Materials. 1987. AmericanSociety for Testing and Materials method E1195-87, standard testmethod for determining a soption constant (K.) for an organicchemical in soil and sediments. American Society for Testing andMaterials, Philadelphia.

2. Bascomb, C. L. 1964. Rapid method for the determination ofcation exchange capacity of calcareous and non-calcareous soils. J.Sci. Food Agric. 15:821-823.

3. Bellin, C. A., G. A. O'Connor, and Y. Jin. 1990. Sorption anddegradation of pentachlorophenol in sludge amended soils. J.Environ. Qual. 19:603-608.

4. Boesten, J. J. T. I., and L J. T. van der Pas. 1988. Modelingadsorption/desorption kinetics of pesticides in a soil suspension.Soil Sci. 146:221-231.

5. Bohn, H. L., B. L. McNeal, and G. A. O'Connor. 1979. Soilchemistry. John Wiley & Sons, Inc., New York.

6. Bosetto, M., P. Arfaioli, and P. Fusi. 1993. Interactions of alachlorwith homoionic montmorillonites. Soil Sci. 155:105-113.

7. Brunner, W., and D. D. Focht. 1984. Deterministic three half-orderkinetic model for degradation of added carbon substrates in soil.Appl. Environ. Microbiol. 47:167-172.

8. Dec, J., and J.-M. Bollag. 1988. Microbial release and degradationof catechol and chlorophenols bound to synthetic humic acid. SoilSci. Soc. Am. J. 52:1366-1371.

9. Einot, I., and K. R. Gabriel. 1975. A study of the powers of severalmethods of multiple comparisons. J. Am. Stat. Assoc. 70:574-583.

10. Ellis, W. D., J. R. Payne, and G. D. McNabb. 1985. Treatment ofcontaminated soils with aqueous surfactants. United States EPAproject summary. EPA/600/S2-85/129. U.S. Environmental Protec-tion Agency, Washington, D.C.

11. Findlay, R. H., G. M. King, and L Watling. 1989. Efficacy ofphospholipid analysis in determining microbial biomass in sedi-ments. Appl. Environ. Microbiol. 55:2888-2893.

12. Gu, B., and H. E. Doner. 1992. The interaction of polysaccharideswith silver hill illite. Clays Clay Miner. 40:151-156.

13. Guerin, W. F., and S. A. Boyd. 1992. Differential bioavailability ofsoil-sorbed naphthalene to two bacterial species. Appl. Environ.Microbiol. 58:1142-1152.

14. Heilman, M. D., D. L. Carter, and C. L. Gonzalez. 1965. Theethylene glycol monoethyl ether (EGME) technique for determin-ing soil surface area. Soil Sci. 100:409-413.

15. Hsu, T.-S., and R. Bartha. 1976. Hydrolyzable and nonhydrolyz-able 3,4-dichloroaniline-humus complexes and their respectiverates of biodegradation. J. Agric. Food Chem. 24:118-122.

16. Ivarson, K. C., M. Schnitzer, and J. Cortez. 1982. The biodegrad-ability of nucleic acid bases adsorbed on inorganic an organic soilcomponents. Plant Soil 64:343-353.

17. Jacobsen, C. S., and J. C. Pedersen. 1992. Mineralization of 2,4-dichlorophenoxyacetic acid (2,4-D) in soil inoculated with Pseudo-monas cepacia DBO1 (pRO101), Alcaligenes eutrophus AEO106(pRO101) and Alcaligenes eutrophus JMP134 (pJP4): effects ofinoculation level and substrate concentration. Biodegradation 2:253-263.

18. Karickhof, S. W. 1980. Sorption kinetics of hydrophobic pollut-ants in natural sediments, p. 193-205. In R. H. Baker (ed.),Contaminants and sediments: analysis, chemistry, biology, vol. 2.Ann Arbor Science Publishing, Ann Arbor, Mich.

19. Knaebel, D. B., T. W. Federle, and J. R. Vestal. 1990. Mineraliza-tion of linear alkylbenzene sulfonate (LAS) and linear alcoholethoxylate (LAE) in 11 contrasting soils. Environ. Toxicol. Chem.9:981-988.

20. Knaebel, D. B., and J. R. Vestal. 1988. Comparison of double-vialto serum bottle radiorespirometry to measure microbial mineral-ization in soils. J. Microbiol. Methods 7:309-317.

21. Knaebel, D. B., and J. R. Vestal. 1992. Effects of intact rhizospheremicrobial communities on the mineralization of surfactants insurface soils. Can. J. Microbiol. 38:643-653.

22. Laha, S., and R. G. Luthy. 1991. Inhibition of phenanthrenemineralization by nonionic surfactants in soil-water systems. En-viron. Sci. Technol. 25:1920-1930.

23. Larson, R. J. 1984. Kinetic and ecological approaches for predict-ing biodegradation rates of xenobiotic organic chemicals in naturalecosystems, p. 677-686. In M. J. Klug and C. A. Reddy (ed.),Current perspectives in microbial ecology. American Society forMicrobiology, Washington, D.C.

24. Lerch, N. K., W. F. Hale, and D. D. Lemaster. 1982. Soil survey ofHamilton County, Ohio. U.S. Department of Agriculture, SoilConservation Service, Washington, D.C.

25. Martin, J. P., and K. Haider. 1986. Influence of mineral colloidson turnover rates of soil organic carbon, p. 283-304. In P. M.Huang and M. Schnitzer (ed.), Interactions of soil minerals withnatural organics and microbes. Soil Science Society of America,Inc., Madison, Wis.

26. Martin, J. P., K. Haider, and L. F. Linhares. 1979. Decompositionand stabilization of ring-14C-labeled catechol in soil. Soil Sci. Soc.Am. J. 43:100-104.

27. Martin, J. P., A. A. Parsa, andK Haider. 1978. Influence of intimateassociation with humic polymers on biodegradation of ['4C]labeledorganic substrates in soil. Soil Biol. Biochem. 10:483-486.

VOL. 60, 1994

4508 KNAEBEL ET AL.

28. McDermott, J. B., R. Unterman, M. J. Brennan, R. E. Brooks,D. P. Mobley, C. C. Schwartz, and D. K. Dietrich. 1989. Twostrategies for PCB soil remediation: biodegradation and surfactantextraction. Environ. Prog. 8:46-51.

29. McKinley, V. L., T. W. Federle, and J. R. Vestal. 1982. Effects ofpetroleum hydrocarbons on plant litter microbiota in an arcticlake. Appl. Environ. Microbiol. 43:129-135.

30. Mihelcic, J. R., and R. G. Luthy. 1991. Sorption and microbialdegradation of naphthalene in soil-water suspensions under deni-trifying conditions. Environ. Sci. Technol. 25:169-177.

31. Miller, M. E., and M. Alexander. 1991. Kinetics of bacterialdegradation of benzylamine in a montmorillonite suspension.Environ. Sci. Technol. 25:240-245.

32. O'Connor, G. A., J. R. Lujan, and Y. Jin. 1990. Adsorption,degradation and plant availability of 2,4-dichlorophenol in sludge-amended calcareous soils. J. Environ. Qual. 19:587-593.

33. Ogram, A. V., R. E. Jessup, L. T. Ou, and P. S. C. Rao. 1985.Effects of sorption on biological biodegradation rates of (2,4-dichlorophenoxy)acetic acid in soils. Appl. Environ. Microbiol. 49:582-587.

34. Pavlostathis, S. G., and K. Jaglal. 1991. Desorptive behavior oftrichloroethylene in contaminated soil. Environ. Sci. Technol. 25:274-279.

35. Richtler, H. J., and J. Knaut. 1988. World prospects for surfac-tants, p. 3-58. In Proceedings of the 2nd World SurfactantCongress. European Committee for Surfactants and their OrganicIntermediates (CESIO), Brussels.

36. Rijnaarts, H. H. M., A. Bachmann, J. C. Jumelet, and A. J. B.Zehnder. 1990. Effect of desorption and intraparticle mass transferon the aerobic biomineralization of a-hexachlorocyclohexane in acontaminated calcareous soil. Environ. Sci. Technol. 24:1349-1354.

37. Robinson, J. A. 1985. Determining microbial kinetic parametersusing nonlinear regression analysis: advantages and limitations inmicrobial ecology. Adv. Microb. Ecol. 8:61-114.

38. Robinson, K. G., W. S. Farmer, and J. T. Novalk 1990. Availability

of sorbed toluene in soils for biodegradation of acclimated bacte-ria. Water Res. 24:345-350.

39. Rowlands, W. N., and R. J. Hunter. 1992. Electroacoustic study ofadsorption of cetylpyridiunium chloride on kaolinite. Clays ClayMiner. 40:287-291.

40. SAS Institute, Inc. 1985. SAS user's guide: statistics, version 5edition. SAS Institute, Inc., Cary, N.C.

41. Scow, K. M., and M. Alexander. 1992. Effect of diffusion on thekinetics of biodegradation: experimental results with syntheticaggregates. Soil Sci. Soc. Am. J. 56:128-134.

42. Scow, K. M., and J. Hutson. 1992. Effect of diffusion and sorptionon the kinetics of biodegradation: theoretical considerations. SoilSci. Soc. Am. J. 56:119-127.

43. Sokal, R. R., and F. J. Rohlf. 1981. Biometry, 2nd ed. W. H.Freeman & Co., New York.

44. Sorensen, L. H. 1972. Stabilization of newly formed amino acidmetabolites in soil by clay minerals. Soil Sci. 114:5-11.

45. Sorensen, L. H. 1975. The influence of clay on the rate of decay ofamino acid metabolites synthesized in soil during the decomposi-tion of cellulose. Soil Biol. Biochem. 7:171-177.

46. Valoras, N., J. Letely, J. P. Martin, and J. Osborn. 1976. Degra-dation of a nonionic surfactant in soils and peat. Soil Sci. Soc. Am.J. 40:60-63.

47. Ward, T. E., and R. J. Larson. 1989. Biodegradation kinetics oflinear alkylbenzene sulfonate in sludge-amended agricultural soils.Ecotoxicol. Environ. Saf. 17:119-130.

48. Weber, J. B., and H. D. Coble. 1968. Microbial decomposition ofdiquat adsorbed on montmorillonite and kaolinite clays. J. Agric.Food Chem. 16:475-478.

49. White, D. C., R. J. Bobbie, J. S. Herron, J. D. King, and S. J.Morrison. 1979. Biochemical measurements of microbial mass andactivity from environmental samples, p. 69-81. In J. W. Costertonand R. R. Colwell (ed.), Native aquatic bacteria: enumeration,activity and ecology. ASTM STP 695. American Society forTesting and Materials, Philadelphia.

APPL. ENVIRON. MICROBIOL.