Abiological loss of endosulfan and related chlorinated organiccompounds from aqueous systems in the presence and absence of

oxygen

T.F. Guerin *

3/32 Wolli Creek Road, Banksia, Sydney, New South Wales 2216 Australia

Received 24 July 2000; accepted 25 January 2001

‘‘Capsule’’: Endosulfan, related OC pesticides, and major degradation products are studied in aquatic systems and factorsinfluencing persistence, and implications to biodegradation studies are explored.

Abstract

Endosulfan is a cyclodiene organochlorine currently widely used as an insecticide throughout the world. This study reports thatthe endosulfan isomers can be readily dissipated from aqueous systems at neutral pH in the absence of biological material orchemical catalysts, in the presence or absence of oxygen. The study showed that aldrin, dieldrin, and endosulfan exhibit bi-phasic

loss from water in unsealed and butyl rubber sealed vessels. Half-lives are substantially increased for endosulfan I when oxygen isremoved from the incubation vessel. The study conditions, where PTFE was used, were such that loss due to volatilization andalkaline chemical hydrolysis was eliminated. Half-lives determined from these data indicate that the parent isomers are much less

persistent than the related cyclodienes, aldrin and dieldrin, confirming the findings of previous studies. The major oxidation productof endosulfans I and II, endosulfan sulfate, is less volatile and can persist longer than either of the parent isomers. Endosulfansulfate was not formed in any of the treatments suggesting that it would not be formed in aerated waters in the absence of microbialactivity or strong chemical oxidants. Since endosulfan sulfate is formed in many environments through biological oxidation, and is

only slowly degraded (both chemically in sterile media and biologically), it represents a predominant residue of technical gradeendosulfan, which finds its way into aerobic and anaerobic aquatic environments. The data obtained contributes to and confirmsthe existing body of half-life data on endosulfan I and II and its major oxidation product, endosulfan sulfate. The half-life data

generated from the current study can be used in models for predicting the loss of chlorinated cyclodiene compounds from aqueoussystems. The findings also highlight the importance of critically reviewing half-life data, to determine what the predominant pro-cesses are that are acting on the compounds under study. # 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Endosulfan; Degradation; Pesticide; Half-lives; Hydrolysis; Non-biological degradation; Bi-phasic loss; Persistence; Abiotic loss; Biode-

gradation; Endosulfan sulfate; Toxicity; Physico-chemical; Risk

1. Introduction

Endosulfan is a cyclodiene organochlorine possessinga labile, cyclic sulfite diester group. Endosulfan is awidely used agricultural chemical and it has been detect-ed in an increasing number of environmental samplesin recent years (Guerin and Kennedy, 1991; Guerin,1993; Mansingh and Wilson, 1995; Mansingh et al.,1997; Miles and Pfeuffer, 1997; Guerin, 1999b). Of key

concern regarding its widespread distribution, particu-larly in water environments, is its high acute toxicity tofish (Table 1). There are, however, relatively few studiesdescribing the fate of endosulfan in aquatic systems(Greve, 1971; Walker et al., 1988; Peterson and Batley,1991; Singh et al., 1991; Guerin and Kennedy, 1992;Guerin, 1993; Peterson and Batley, 1993; Mansingh andWilson, 1995; Mansingh et al., 1997; Miles and Pfeuffer,1997). In addition, it has not been extensively deter-mined to what extent losses of the endosulfan isomersresult from chemical degradation as opposed to dis-sipation by other means, such as volatilization andadsorption, in aqueous systems (Guerin and Kennedy,1992; Guerin, 1993). Although there are a number ofreports describing biological oxidation of endosulfan to

0269-7491/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.

PI I : S0269-7491(01 )00112-9

Environmental Pollution 115 (2001) 219–230

www.elsevier.com/locate/envpol

* Present address: Shell Engineering Ltd, NSW State Office, PO

Box 26, Granville 2142 NSW, Australia. E-mail address: turlough.

guerin@shell.com.au.

E-mail addresses: turloughg@hotmail.com, turlough.guerin@

bigpond.com (T.F. Guerin).

the sulfate, there is no clear evidence for its formation insterile soils or water (Guerin, 1993). This indicates thatliving organisms may be necessary to bring about theoxidation of endosulfan to form endosulfan sulfate inthe environment. Previous studies have also reportedthat isomerization can occur between the parent isomersin aqueous systems, with the reaction favouring forma-tion of endosulfan I (a-endosulfan; Schmidt et al.,1997).

The primary aim of the research described in thispaper was to determine the half-life of the parent endo-sulfan compounds in sterile aqueous solutions with andwithout the presence of oxygen. A further aim was todetermine whether endosulfan sulfate could be formedunder these conditions.

2. Materials and methods

2.1. Chemicals

Endosulfan and its degradation products were a giftfrom Hoechst, Germany. Aldrin and dieldrin were pro-vided by Shell Chemicals, Australia. cis- and trans-Aldrin diol were kindly provided by Shell ChemicalsResearch, UK. Hexane (Nanograde) and methanol(ChromAR) were purchased from Mallinckrodt Chemi-cals. The key physico-chemical properties of the endo-sulfan compounds are summarized in Table 2.

2.2. Incubation conditions in aerobic study

The pesticides, endosulfan I, II, aldrin and dieldrin andthe endosulfan degradation product, endosulfan sulfate(100–500 ppm in 10 ml methanol) were added to eitherNanopure-filtered, distilled and deionized water, steri-lized by autoclaving at 121�C for 20 min, or sterilized100 mM potassium phosphate-buffered yeast mannitolmedium with filtered (45 m) soil extract (10% v/v of a 50g l�1 soil in water extract; (Guerin, 1993) in 4 ml Whea-ton vials, to give final amounts of 1–5 mg l�1. The head-space volume was 2 ml in the vials. The quantities of 1–5

mg l�1 pesticide used in this study reflects those com-monly used in studies of the microbial degradation ofpesticides, and in aquatic toxicology, where bioassaysare performed. The medium was also sterilized byautoclaving for 20 min at 121�C. It is also noted that thequantities of added pesticide meant that these com-pounds were close to the limits of the solubility for thesecompounds in the aqueous phase (Table 2). Duplicatevials were made of unsilanized borosilicate glass andprepared according to the protocol described (Guerin,1993). The medium contents have previously beendescribed (Guerin, 1993). Vessels were sealed with eitherPolytetrafluoroethylene (PTFE)-lined butyl rubber seals(Wheaton, Millville NJ, USA supplied by EdwardsInstrument, Narellan, Australia) or non-PTFE-linedbutyl rubber stoppers. Duplicate vessels were kept at30�0.5�C, in an incubator for 30 days. No furtherattempt was made to artificially aerate the incubationflasks during the course of the experiment, and sterileair at atmospheric pressure from a laminar flow cabinetwas used as the gas atmosphere in the headspace ofthe flasks. The media was kept sterile throughout theexperiment, and this was checked by heterotrophic platecounts and microscopic examination (Guerin, 1993).The pH of the media was adjusted to 7�0.05 with KOHor HCl prior to dispensing into the incubation vessels.Incubation vessels were kept in the dark.

Surface microlayer subsamples were also taken, intriplicate, after 4 h (30�C) from vessels set aside espe-cially to determine whether there was any difference inpesticide concentration at the liquid–air interface. Thiswas performed by withdrawing a volume of 100 ml fromthe surface, an amount equivalent to the top 0.9 mmof the medium, using a pipette. The pipette was rinsedwith solvent and the medium extracted as previouslydescribed (Guerin and Kennedy, 1992). An equivalentvolume removed from the medium bulk was also ana-lyzed for pesticides.

Table 1

Overview of acute toxicity of key endosulfan compoundsa

Compound Toxicity LD50 (mg kg�1)

Insects Fish Birds Mammals

Endosulfan I 5.5 0.001–0.01b 26–1000 9.4–40

Endosulfan II 9.0 0.001–0.01 26–1000 177

Endosulfan sulfate 9.5 0.001–0.01 –c 8–76

Endosulfan diol >500 1–10 – >1500

a Summarized from the literature (Guerin, 1993; Anonymous,

1998).b The lower the lethal dose, i.e. LD50 value, the higher the toxicity.c –, indicates that there was no data available.

Table 2

Liquid-phase physico-chemical properties of major endosulfan com-

pounds

Compound Solubility in

water (S) ppma

Log Kow v.p. (Pa)a,b Hc

Endosulfan I 0.51 3.6 0.0004 0.72

Endosulfan II 0.45 3.83 8.0�10�5 0.04

Endosulfan sulfate 0.48 3.66 3.7�10�5 0.03

Endosulfan diol 300.0 3.68 2.3�10�6 1.3�10�4

a Values reported are from the PhysProp and DATALOG Data-

bases from Syracuse Research Corporation where available (Meylan

and Howard, 2000). Values for endosulfan diol are from elsewhere

(Guerin and Kennedy, 1992; Guerin, 1993).b Vapor pressure in units of Pa.c Henry’s constant (H)=v.p./S in units of Pa m3 mol�1, calculated

from the v.p. and S data reported in this table.

220 T.F. Guerin / Environmental Pollution 115 (2001) 219–230

2.3. Incubation conditions in oxygen-limited study

The incubations conditions were as before exceptthe incubation vessels were evacuated with N2 gasat the commencement of the incubation period. Resa-zurin was added to the medium and this dye remainedin its reduced form throughout the entire incubationperiod (i.e. clear) in all the incubation vessels.

2.4. Extraction, recovery and analysis of parentpesticides and degradation products

Duplicate incubation vessels containing the aqueousmedia (2 ml) had their contents quantitatively trans-ferred (i.e. sacrificed) into the reservoir of a 10 mlliquid–liquid partitioning device (Mixxor1 by Genex)and were extracted, recoveries determined, and analysesconducted according to the method previously de-scribed (Guerin and Kennedy, 1992). Samples weretaken on days 0, 2, 4, 6, 8, 15 and 30. A total solventvolume of 10 ml hexane/acetone/methanol/medium(15:5:2:2) was also added to the Mixxor1 reservoirs.The piston of the Mixxor1 was moved 60 times in itsreservoir to partition the pesticides into the solventphase. After the phases were allowed to separate (�1min), the solvent layer was decanted off the aqueousphase directly from the Mixxors1 into volumetric flasksand the total volumes were made to either 10 or 25 mlwith hexane. Subsamples were dried with anhydroussodium sulfate prior to analysis by GC–ECD.

2.5. Analysis of data

Half-lives for all the compounds were determined.The data analysis was conducted in three stages. Dur-ing the initial data analysis (incorporating the datacollected at all sampling times, i.e. 0–30 days), thesquare root of the correlation coefficient (r2) was deter-mined from the log exponential decay plots i.e. log100�C/Co vs t plots using the trend line function inMicrosoft Excel 2000 (Microsoft). In the second stageof the analyses, data were analyzed, excluding the dataat day 15 and 30, to determine whether the pesticideloss was bi-phasic, that is, exponential during the initialphases of the experiment (0–8 days), followed by a laterlinear phase (8–30 days). Where there was bi-phasic (i.e.non-first-order or non-linear) loss occurring, evidentfrom the 100�logC/Co vs t plots, 8–30 day data wasplotted and half-lives were also calculated using thisdata. Multivariate analysis were conducted to comparesets of data using the regression function in MicrosoftExcel 2000 (Microsoft). This generated P-values (5%level of significance) for these comparisons. Univariateanalyses were conducted on individual treatments (e.g.aldrin, unsealed, 0–30 days) to determine P values (5%level of significance).

3. Results and discussion

3.1. Analysis using 0–30 day data

For each of the treatments, a line was plotted throughthe data to determine whether the data collected atall the sampling times fitted an exponential decay curve.Since the concentration data was plotted as log values,then an exponential decay plot would be a straight lineon such graphs.

The dissipation of aldrin and dieldrin from PTFEsealed vessels observed the first order decay model. Thiswas also the case with the data from dissipation ofendosulfan I and II from the unsealed and PTFE sealedflasks, and the dissipation of endosulfan sulfate fromunsealed, BRS and PTFE sealed flasks. This is illu-strated in Fig. 1A–C.

These data indicated, however, that the first orderexponential decay model did not adequately describe thelosses of pesticide over the entire period of the experi-ment (0–30 days), particularly for losses of the morevolatile pesticides from the unsealed and butyl rubbersealed vessels.

The dissipation of endosulfan I and II from water inPTFE-sealed vessels (r2=0.93) and endosulfan sulfatefrom water in unsealed vessels (r2=0.95) fitted the firstorder decay model well, and therefore a single half-lifevalue was adequate to describe the loss of these com-pounds within these treatments. There was a significantdifference (P<0.05) between the pesticide half-lives inthe butyl rubber sealed and PTFE-sealed flasks, and theunsealed and PTFE-sealed vessels, but no differenceswere measured between the butyl rubber sealed and theunsealed vessels. The half-lives of the compounds ineach of the aerobic treatments are recorded in Table 3.

3.2. Analysis using 0–8 day data

The data that were obtained earlier in the trial (i.e. upto the 8th day of incubation, only) for the unsealed andBRS sealed treatments, generally showed larger r2

values and an improved fit to the first order exponentialdecay model (compare columns 2, 3, 5, 6 in Table 3).

In the unsealed and butyl rubber sealed treatments, itwas found that data from the first 8 days fitted themodel of first order exponential decay better than datafrom over the entire experiment (i.e. 30 days). Thesefindings indicated that the loss of pesticides from theaqueous media, particularly in the butyl rubber-sealedand the unsealed vessels, did display a first order expo-nential decay, but only over the initial phase of theexperiment. Volatilization from the unsealed vessels, andabsorption by butyl rubber in the butyl rubber sealedvessels, could explain these losses early in the experiment.As with the 0–30 day data, there was a significant differ-ence (P<0.05) between the pesticide half-lives in the

T.F. Guerin / Environmental Pollution 115 (2001) 219–230 221

butyl rubber sealed and PTFE-sealed flasks, and theunsealed and PTFE-sealed vessels, but no differenceswere measured between the butyl rubber sealed and theunsealed vessels. The 0–8 day data for aldrin, dieldrinand endosulfan are plotted in Fig. 2A and B.

3.3. Analysis of 8–30 day data

When the 8–30 day data from the unsealed treatmentswas analyzed, increased half-lives were obtained. Thesehalf-lives increased from 0.5, 2, and 14 (for the 0–8 daydata), to 78, 46, and 29 days for aldrin, dieldrin, andendosulfan I, respectively. These increased half-livesreflected the predominant dissipation processes whichwere acting on the compounds after the initial fast rateof loss due to volatilization. These longer term rates ofloss reflecting the chemical degradation and slow deso-rption processes, showed that there was an 18–75�increase in the half-life length when data from 8–30 dayswas considered. These slower processes were chemicaldegradation and slow desorption of the compoundsfrom the glass and through the aqueous media, andsubsequent volatilization (Fig. 3).

3.4. Losses of aldrin and dieldrin

There were high rates of loss of aldrin and dieldrinfrom both unsealed incubation vessels and flasks sealedwith butyl rubber when the 0–30 day data was con-sidered. In the butyl rubber sealed vessels, aldrin anddieldrin had largely disappeared at the 30th day, withhalf-lives of 2.9 and 3.7 days, respectively. These resultswere similar to the values of 0.5 and 2 days for half-livesin the open flasks. The only difference in the pattern ofloss between these two treatments was that the initialrate of dieldrin disappearance in the unsealed vesselswas slightly higher than in the butyl rubber sealed ves-sels. In contrast, when aldrin and dieldrin were incu-bated in similar sterile media or in water alone withPTFE-lined butyl rubber seals, there was a considerablyslower rate of loss. The half-lives under these conditionswere >58 and >22 days, respectively.

Losses of aldrin and dieldrin were very low underconditions of limited oxygen with half-life values of 134and 46 days, respectively (Fig. 4A and Table 7).

The bi-phasic pattern of loss was particularly pro-nounced for aldrin, dieldrin and endosulfan I loss fromunsealed flasks (Table 4). A possible reason for thisnon-first order loss could have been that a large prop-ortion of pesticide was concentrated at the liquid–airinterface (surface microlayer) as previously reported(Guerin and Kennedy, 1992). As a consequence, thesepesticides would be expected to volatilize early in theexperiment because of their close proximity to the liquid–air interface. Peterson and Batley (1991) have alsosuggested this mechanism as a reason for the loss of

endosulfan from aqueous media in laboratory experi-ments. After this surface quantity had entirely volati-lized, further losses must have come from the solutionbulk through the process of diffusion. Thereforeanother possible reason for this bi-phasic loss may bedue to a rim effect, where the more rapid diffusionoccurs at high adsorptions (i.e. at concentrations higherthan the compound’s solubility in the aqueous medium

Fig. 1. Dissipation of chlorinated compounds under oxygenated con-

ditions (0–30 day data) (A) aldrin and dieldrin, (B) endosulfan I and

II, and (C) endosulfan sulfate.

222 T.F. Guerin / Environmental Pollution 115 (2001) 219–230

early in the experiment) across the glass to the surface.As the pesticide levels decrease in the vessels, theamounts remaining after the initial high losses will moreclosely approximate the upper limits of the compound’ssolubility. This may explain, at least in part, why thepesticide loss is slower after 8–30 days incubation asthe forces of solubilization would tend to override thoseexerted by volatilization (Guerin and Kennedy, 1992;Guerin, 1993).

In the current study, approximately 40–50% of theoriginally applied pesticides recovered from the vesselscontaining aqueous medium (1–5 mg ml�1 originallyadded), was found to be concentrated at the interfacesof the system. Also, an attempt was made to subsamplethe liquid–air interface and measure the pesticide con-centration. However, there was no detectable differencebetween the surface sample and that in the bulk of themedium. Therefore, the pesticides must have accumu-lated at the liquid–glass interface only under theseconditions. Although it is possible that a slower rateof hydrolysis could occur at this interface, insulatedfrom the effect of the hydroxyl ion, this is unlikely toexplain the increased half-lives for the endosulfan iso-mers reported in this study with effective sealing. Asimilar increase in half-life was also observed for themore stable compounds such as dieldrin, when the flaskswere sealed with PTFE.

This distribution to the glass-medium interface wasobserved when the compounds under study were addedto either water or microbial growth media. An homo-genous distribution of pesticide throughout the entiresystem was achieved by adding 0.1% Tween 80. All ofthe compounds studied were distributed throughout thesystem in a similar fashion, and all responded similarlyto the detergent treatment in both the sterile distilledwater and growth medium. There was a greater dis-tribution of all the pesticides to the liquid–glass inter-face in the vessels containing pure water. A similar effectwas observed by Peterson and Batley (1991) when theysubsampled from a larger total volume of water con-taining both endosulfan isomers at lower concentrations(<0.1 ppm) in polycarbonate vessels. These findingsand the results from the current study therefore illus-trate the importance of avoiding subsampling whenanalyzing aqueous extracts containing relatively highconcentrations of endosulfan and related cyclodienes.

The distribution of pesticides in aqueous systems is ofparticular importance in microbial degradation studieswhere the availability of the compound is likely toaffect its degradation. Thus, when the pesticide is addedin small amounts of solvent to the aqueous phase (ashas generally been reported in microbial pesticide

Table 3

Half-lives (r2 values) in days for pesticide loss using the first order exponential decay model on data from various timesa

Compound Unsealed (days) BRS-sealed (days) PTFE-sealed (days)

0–30 0–8 8–30 0–30 0–8 8–30 0–30 0–8

Aldrin 13.6 (0.37) 0.52 (0.73) 77.9 (0.55) 1.1 (0.79) 2.9 (0.98) 21.6 (0.71) >200b (0.22) 57.8 (0.34)

Dieldrin 2.8 (0.65) 2.01 (0.98) 46.4 (0.90) 3.4 (0.94) 3.7 (0.98) 2.3 (0.99) 75 (0.54) 21.8 (0.82)

Endosulfan I 25.8 (0.83) 13.8 (0.89) 29.0 (0.93) 7.9 (0.83) 6.6 (0.96) 3.0 (0.66) 21.7 (0.93) 13.8 (0.89)

Endosulfan II 18.0 (0.90) �200b (0.001) 18.2 (0.91) 12.7 (0.83) 9.6 (0.96) 5.7 (0.63) 24.2 (0.93) 36.2 (0.36)

Endosulfan sulfate 37.3 (0.95) 51.8 (0.48) 35.3 (1.0) 34.0 (0.5) 12.4 (0.78) 68.9 (0.18) 102.5 (0.56) 65.1 (0.37)

a Unsealed=cotton wool was used to stopper the flasks, BRS=butyl rubber sealed vessels, and PTFE=Teflon-lined butyl rubber seals.b Values of >200 and �200 days were given because of the relatively short time of the trials (30 days).

Fig. 2. Dissipation of chlorinated compounds under oxygenated con-

ditions (0–8 day data) (A) aldrin and dieldrin, and (B) endosulfan I

and II.

T.F. Guerin / Environmental Pollution 115 (2001) 219–230 223

degradation studies, often followed by evaporation ofsolvents with N2 gas), its distribution in the incubationvessel will tend to be associated with the interfaces. Thegeometry of the incubation vessel as well as the con-stituents of the medium will effect the pesticide dis-tribution. In microbial degradation experiments, whereinsoluble compounds are added in methanol or a similarsolvent, an apparent increase in pesticide concentrationwith time will be observed in the bulk of the medium,once exponential growth commences and lipids increasein quantity. This effect may be overcome by completelysacrificing the entire treatment incubation flasks at eachsampling time.

The inclusion of the very chemically stable cyclodienes,aldrin and dieldrin, in PTFE-sealed vessels in this studyprovided internal controls that indicated disappearancepredominantly from physical losses, thereby providingthe maximum limits of these processes in the systemstudied. The experimental conditions were too mild andthe incubation period was too short to allow substantialchemical degradation of these compounds. The persis-tence of aldrin and dieldrin in these incubations thereforerepresents the maximum limits for either slow volatiliza-tion or other possible processes such as irreversiblebinding to container surfaces. Thus, any differences

between the persistence of these internal controls andthat of the endosulfan compounds represents the actualdisappearance by chemical reaction. The very slow rateof disappearance of aldrin and dieldrin in the PTFE-sealed vessels in both water and growth medium con-firmed that the system was well sealed.

3.5. Dissipation of endosulfan isomers

In both the butyl rubber sealed and the unsealed ves-sels, the endosulfan isomers were lost at fast rates. After8 days of incubation, the calculated half-lives of endo-sulfans I and II in the butyl rubber sealed flasks were 7.9and 12.7 days, respectively. When endosulfans I and IIwere incubated in unsealed vessels, the half-lives variedfrom 13.8 to 29 and 18 to �200 days, respectively. Inboth the butyl rubber sealed and the unsealed flasks,comparing 0–8 day data, endosulfan I loss was higherthan that of endosulfan II. This result reinforces pre-vious findings that endosulfan I is more volatile thanendosulfan II (Goebel et al., 1982; Worthing andWalker, 1987; Singh et al., 1991; Guerin and Kennedy,1992; Guerin, 1993).

When the vessels were sealed with PTFE, rates ofdisappearance for both isomers in both water and

Fig. 3. Experimental vessel and graphical presentation of bi-phasic loss.

224 T.F. Guerin / Environmental Pollution 115 (2001) 219–230

microbial growth medium were considerably lowercomparing 0–30 day data. The half-lives were 21.7 and24.2 days for endosulfans I and II, respectively. Underoxygen-limited conditions, the half-lives of endosulfanI and II, were �200 and 58 days, respectively (Fig. 4A,B and Table 7). The effect of PTFE sealing was to sub-stantially reduce the volatilization of the parent com-pounds from the flasks.

It was clear from the butyl rubber sealed and unsealedtreatments that endosulfan I is more volatile thanendosulfan II. Given the relative chemical inertness ofthe PTFE-sealed systems and that traces of endosulfandiol were detected in the same system, it is reasonable toconclude that both endosulfan isomers were chemicallydegraded in the aqueous incubations. Based on thisdata, endosulfan II may be more chemically labile thanendosulfan I.

In previous studies, it has been observed that endo-sulfan II also disappeared at a faster rate than endosulfanI. Under aerobic conditions at a lower temperature of22�C, the half-lives of endosulfans I and II in a potas-sium phosphate buffered, minimal salts medium (pH6.5), were 88 and 40 days, respectively (Miles and Moy,1979). In their paper no mention was made on how thevessels were sealed. The half-lives of endosulfans I andII in non-sterile seawater (pH 8.0) were 4.9 and 2.2days, respectively (Cotham and Bidleman, 1989). Theseincubations were carried out aerobically and at 20oCunder laboratory lighting. In another study, incuba-tions in lake water showed that the half-life of endo-sulfan I was 35 days at pH 7 and 105 days at pH 5.5(Greve, 1971). It was shown in the same study thatwhen iron hydroxide gel is mixed with water, the rate ofhydrolysis is considerably accelerated. Other research-ers have reported half-lives of 10–43 days under con-trolled laboratory conditions, at pH values of 8.5(Southan and Kennedy, 1995), and values of <3 daysfor both isomers in laboratory water columns of un-reported pH (Logan and Barry, 1996). Guerin (1999a)has reported losses of endosulfan I under sterile anae-robic conditions, as part of a 30-day anaerobic bio-degradation study, with losses of endosulfan I at 20, 10and 2% (of that originally applied) when this com-pound was added at 1, 2, and 10 ppm, respectively,indicating rates of loss are dependent on the mass ofadded pesticide. The latter findings are consistent withwater insoluble pesticides desorbing from the glass sur-face into the medium. Also, in the Guerin (1999a) study,when microorganisms were present, half-lives of endo-sulfan I varied between 5 and 15 days, substantiallyincreasing its loss. It has also been shown that endo-sulfan losses can be significantly minimized from watersolutions if the incubation vessels are sealed to preventvolatilization (Guerin and Kennedy, 1992; Guerin,1993). Several biodegradation studies in liquid culturehave demonstrated the importance of sealing incubationvessels with Teflon or PTFE as previously discussed(Guerin, 1995, and references cited therein). However,not all biodegradation studies have employed PTFEand this should be considered as a critical criterionwhen reviewing and evaluating degradation datareported in the literature. It was noted that there was nointerconversion between isomers under the conditionsdescribed in the current study.

Fig. 4. Dissipation of chlorinated compounds under oxygen-limited

conditions (0–30 day data) (A) aldrin, dieldrin, endosulfan I and II

(1 ppm), and (B) endosulfan I and II (10 ppm), and (C) endosulfan

sulfate (5 and 50 ppm).

T.F. Guerin / Environmental Pollution 115 (2001) 219–230 225

The role of pH is important, particularly when therates of endosulfan loss are compared across differentstudies. This is because the endosulfan isomers are sus-ceptible to alkaline hydrolysis (Goebel et al., 1982).Thus, rates of hydrolysis at pH 8 will be a �10 timesfaster than the rates at pH 7. Some differences in thehalf-lives previously reported may be due to differencesin temperature, which may also affect the hydrolysisrates of pesticides. Since endosulfans I and II are vola-tile, the temperature at which the experiments are car-ried out is also very important.

In the current study there was no significant difference(P>0.05) in the degradation rates of either endosulfanisomer comparing incubation in water and in themicrobial growth medium. This indicates that the soil,peptone, or yeast extract and inorganic minerals had nomeasurable effect on the persistence of the isomers.

3.6. Losses of endosulfan sulfate

In all of the experiments conducted, endosulfan sul-fate was relatively stable and considerably more persis-tent than the parent isomers. The half-life of endosulfansulfate in the sterile water was calculated at 103 dayswhen sealed with PTFE compared with 30 days in theunsealed vessels. Its persistence in the vessels sealed withbutyl rubber (10 day half life), compared with that inthe unsealed vessels (30 day half-life), was not sig-nificant, and this was likely to be due to the wide varia-tion in the analysis of endosulfan sulfate data aspreviously reported by (Guerin et al., 1992).

The data on the dissipation of endosulfan sulfatefrom the PTFE-sealed vessels fitted the model of firstorder exponential decay poorly, when all the samplingtimes were analyzed. The very low r2 values obtainedwith endosulfan sulfate in the PTFE-sealed vessels,suggests little or no relationship between endosulfansulfate concentrations and time. However, from theextraction and analysis of endosulfan sulfate previouslyreported (Guerin and Kennedy, 1992), it is likely thatanalytical error was also important and contributed tothe very low r2 values. Endosulfan sulfate was evenmore stable under conditions of limited oxygen, withhalf-lives typically �200 days (Fig. 4C and Table 5).

The results of the dissipation of endosulfan sulfatetherefore indicate a limitation of calculating the half-lifeof this compound using the approach described here.

This approach is more appropriate for determining thehalf-lives of the parent isomers, where the differencesbetween the half-lives are not as great, and where ana-lytical variation is low.

Miles and Moy (1979) have also reported on the per-sistence of endosulfan sulfate in aqueous media andhave given a value for its half-life, under the previouslydescribed conditions, as >140 days. The reported per-sistence of endosulfan sulfate in the aqueous systemsstudied in the current work, and from this report in theliterature, indicates that this endosulfan transformationproduct is likely to remain in water environments muchlonger than the parent isomers. It has previously beenshown not to be readily biodegradable. However, in realenvironments, there may be other processes of endo-sulfan sulfate removal such as strong adsorption to soiland sediment particles. It should be recognized thatbecause of the relatively short time frame of the trials,the r2 values and corresponding half-life data for endo-sulfan sulfate, has been included in the data set for thesake of completeness and these do not represent defini-tive values. Further research would be needed to deter-mine definitive values for the persistence of thiscompound in aqueous systems.

3.7. The role of volatilization in pesticide disappearance

From the increased losses in the open vessel incuba-tions, it is evident that the endosulfan dissipation inthese treatments was primarily owing to volatilization.

The rates of volatilization of the endosulfan isomersin the open-vessel experiments were similar to thosefrom the butyl rubber sealed experiments. These resultsindicated that volatilization of these compounds fromthe aqueous media has a similar effect as that of thebutyl rubber seals. Aldrin and dieldrin were also lost atfast rates in similar incubations, confirming thatabsorption into the butyl rubber seals was the majorcause of loss (Guerin and Kennedy, 1992). Extractionand analysis of the butyl rubber seals after the incuba-tion indicated that all the cyclodienes had becomeabsorbed into this sealing material.

No hydrolysis products of endosulfan, endosulfansulfate, aldrin or dieldrin were detected in hexane–acetone extracts from the open or butyl rubber sealedtreatments. Thus, it is evident that the major cause ofdissipation of all compounds in the unsealed flasks was

Table 4

Summary of bi-phasic loss data from trials with unsealed vessels

Phase of trial (days) Main loss mechanisms Relative rate of loss Calculated half-lives (days)

Aldrin Dieldrin Endosulfan I

Initial (0–8) Volatilization Fast 0.52 2.01 13.8

Latter (8–30) Desorption, chemical degradation Slow 77.9 46.4 29

226 T.F. Guerin / Environmental Pollution 115 (2001) 219–230

volatilization, and absorption in the butyl rubber sealedflasks, rather than chemical degradation.

Volatilization from uninoculated controls in aerobicmicrobial degradation studies is likely to be a significantfactor in overall pesticide disappearance in unsealedsystems, particularly with organochlorine pesticidessuch as endosulfan. In one study, 30% of appliedendosulfan I was reported to have volatilized from aseawater/sediment microcosm (sealed with poly-urethane) during the first 4 days of the experiment(Cotham and Bidleman, 1989). Others demonstratedthat polystyrene absorbed both endosulfan isomersstrongly, compared with glass (Peterson and Batley,1991). The current findings therefore confirm thesefindings, and illustrate the importance of sealing aque-ous systems containing these compounds, with an inertmaterial such as PTFE.

The high volatilization rate of endosulfan I, is due toits low water solubility and relatively high vapor pres-sure, or its high Henry’s constant. The ratio of liquid-phase vapor pressure and solubility, or solid-phasevapor pressure and solubility, provides a value for theHenry’s constant. This relationship may be used toshow the difference in the relative rates of volatilizationof the parent endosulfan isomers and of the recalcitrantcyclodienes, aldrin and dieldrin. In illustrating theimportance of the Henry’s constant of a compound, it isconvenient to introduce the concept of fugacity. Thefugacity is the escaping tendency of a compound from aparticular phase. This can be expressed mathematicallyas f=C/Z. In this expression, f is the fugacity (units ofpressure Pa), C is the concentration (units of mol m�3)and Z is the fugacity capacity (units of mol m�3 Pa�1).Each compound has its own fugacity and at equili-brium, compounds will accumulate in phases withthe lowest fugacity, or highest Z values. So in water, thefugacity capacity is the inverse of the compound’s Hen-ry’s constant (H) (Guerin and Kennedy, 1992). This isdescribed by the equation, Z=Zwater=1/H. The calcu-lated fugacities of the cyclodiene compounds understudy are equivalent to their vapor pressure values in

the same phase because the concentration (C) is equal totheir water solubilities for the solid compounds.

In calculating the Henry’s constant, values for watersolubility and vapour pressure must be for the samephase, that is, both for the liquid-phase or both for thesolid-phase. The values presented in Table 2 are forthe solid-phase for each of the pesticides.

Some of the behavior observed in the butyl rubber-sealed and unsealed vessels can be accounted for bydifferences in their calculated fugacities. The fastestrates of disappearance from both of these treatmentswere that of aldrin, which also had the lowest Z value,or greatest fugacity. From the vapor pressure and solu-bility data obtained from the literature, endosulfan I hasa Henry’s constant of 0.72, approximately 18 timesthat of endosulfan II (H=0.04), which correlates wellwith the greater rate of disappearance from the non-PTFE-sealed vessels (consistent with volatilization andabsorption mechanisms of loss). Aldrin had the highestHenry constant of 4.95, while dieldrin was lower at 0.53.

3.8. Detection and analysis of potential hydrolysisproducts of endosulfan

Trace levels of the hydrolysis product, endosulfandiol, were detected after 30 days incubation in flaskscontaining either parent isomer of endosulfan at thebeginning of the experiment. The recovery treatmentsshowed that this degradation product was extractedwhen spiked into zero time vessels. The identity ofendosulfan diol was confirmed using two different gaschromatographic columns (Guerin and Kennedy, 1992).The highest concentrations of endosulfan diol weredetected in the PTFE-sealed incubations. With endo-sulfan I, these concentrations were 0.08–0.1 ppm ofendosulfan diol after 30 days. This rate of endosulfandiol formation correlates well for the calculated half-lifeof endosulfan I of approximately 22 days in the sterilemedia. This rate of formation, however, was not stoi-chiometric, as approximately 0.5 ppm of endosulfandiol would have been expected to form over this period

Table 5

Half-lives for pesticide loss under oxygen-limited conditions using the first order exponential decay model (0–30 days)a

Compound 1 ppmb 10 ppmc

Half-life (days) r2 Half-life (days) r2

Aldrin 134 0.003 – –

Dieldrin 46 0.14 – –

Endosulfan I �200 <0.001 >200 0.001

Endosulfan II 58 0.24 97 0.71

Endosulfan sulfate �200 0.43 �200 0.01

a PTFE-lined butyl rubber sealed vessel, evacuated 7� prior to incubation (see methods); values >200 days were given because of the relatively

short time of the trials (30 days). Values of �200 indicated that the calculated half-lives were greater than 1000 days.b 5 ppm of endosulfan sulfate was used because of its higher analytical detection limits.c 50 ppm of endosulfan sulfate was used because of its higher analytical detection limits.

T.F. Guerin / Environmental Pollution 115 (2001) 219–230 227

if there was complete transformation. Some of the dif-ference between the amount of endosulfan diol expectedand that which was observed, may have been due to thereduced extraction efficiency of the endosulfan diol.Higher concentrations of 0.1–0.15 ppm endosulfan diolwere detected in the endosulfan II incubations under thesame conditions, consistent with its lower chemical sta-bility. Much lower amounts of endosulfan diol (<0.01ppm) were detected in the endosulfan sulfate incuba-tions after 30 days, and then only in PTFE-sealed incu-bations. These findings correlate well with the observedstability of this compound under these conditions.

The potential hydrolysis products of dieldrin, cis- andtrans-aldrin diol, were not detected in any of the treat-ment incubations containing dieldrin, although theunderivatized standard compounds were chromato-graphed successfully under the conditions described foranalyzing the parent compounds (Guerin et al., 1992).Given the highly recalcitrant nature of dieldrin, and themild incubation conditions of water and growth medi-um, no hydrolysis products were expected to form.

Polytetrafluoroethylene (PTFE) exhibits chemical andphysical properties which, when coated onto butyl orsilicone rubber, make it suitable for sealing aqueousmedia that is in contact with semi-volatile or hydro-phobic compounds such a the cyclodiene pesticides.These characteristics of PTFE are its high resistance toheat, inertness to chemical attack over a wide range oftemperatures, low moisture absorption and permeabil-ity (<0.01% in 24 h), high physical strength, very

high thermal stability, and flexibility. Because ofits high resistance to temperature, it is also autoclav-able (Schlanger and Baumgartner, 1980; Guerin, 1993).However, dry heat and radiation may also sterilize it.These properties are listed in Table 6. Since this materialis impermeable to gases, including N2, O2, H2, and CO2,it can also be used to maintain anaerobic conditions inflasks containing media for the growth of microorgan-isms. Of these properties, its high chemical resistance isof greatest importance in biodegradation studies as itprevents absorption of the compounds under study, intothe sealing material.

4. Conclusions

This study reports that the endosulfan isomers can bedissipated from simple aqueous systems at neutral pH inthe absence of biological material or chemical catalysts.When the incubation vessels are sealed with PTFE, thenendosulfan II is more readily degraded than endosulfanI, a phenomenon already observed in various aqueoussystems. The study also showed that under PTFE-sealedconditions, but in oxygen-limited conditions, the half-lives are more than doubled indicating that the parentisomers of endosulfan are more stable under these con-ditions. This result is in contrast to that obtained inunsealed systems, where the loss of endosulfan I isgreater than that of endosulfan II. Half-lives determinedfrom the data indicate that the parent isomers are much

Table 6

Physico-chemical properties of polytetrafluoroethylene (PTFE)

Inertness to chemical attack

Low moisture absorption and permeability

Impermeable to N2, O2, H2 and CO2

High physical strength and high thermal stability

Low coefficient of friction

Very low dielectric constant and excellent electrical insulator

Non-stick (anti-adhesion) surfaces

Flexible, making suitable for sealing vessels in biodegradation or dissipation studies

Table 7

Implications and recommendations from the current study

1. The endosulfan isomers, like aldrin and dieldrin, volatilize readily from incubation vessels containing microbial growth media, which are

unsealed, or sealed with butyl rubber.

2. Published values for half-lives of volatile and semi-volatile values should be critically reviewed prior to use in modelling their degradation to

see whether the study has taken bi-phasic effects into account and use the data accordingly.

3. Biodegradation assays should include sterilized cells as a control to minimize glass surface binding effectsa.

4. Half-life values generated in the current study can be used in modelling the dissipation of the endosulfan and related compounds from

aqueous systems.

5. When endosulfan (or aldrin and dieldrin) are added to aqueous media at levels higher than their solubility in water, adsorption effects are

likely to retain the pesticide at the glass-media interfaces until microbial growth becomes significant, and causes it to desorb. Therefore in

sterile treatments, entire vessels should be extracted without prior subsampling, particularly with compounds that have low water solubility.

6. Anaerobic biodegradation assays should be sealed with Teflon-lined butyl rubber to avoid pesticide absorption, while maintaining an oxygen

impermeable seal.

a Particularly in aerobic studies where the medium surface has direct contact with the atmosphere through a cotton wool plug.

228 T.F. Guerin / Environmental Pollution 115 (2001) 219–230

less persistent than the related cyclodienes, aldrin anddieldrin. However, the major oxidation product ofendosulfans I and II, endosulfan sulfate, is less volatileand can persist longer than either of the parent isomers.

Endosulfan sulfate was not formed in any of thetreatments in the current study. This suggests thatendosulfan sulfate would not be formed in aeratedwaters in the absence of microbial activity or strongchemical oxidants. Since endosulfan sulfate is formed inmany natural environments through biological oxida-tion, and is only slowly degraded (both chemically insterile media and biologically), it represents a pre-dominant residue of endosulfan in aerobic aquaticenvironments.

Both the endosulfan isomers dissipated from theincubation vessels at faster rates when the vessel wassealed with butyl rubber, than when they were sealedwith PTFE. Conversely, the relatively inert PTFE sealsgreatly reduced losses from volatilization and absorp-tion, thus providing the necessary conditions for study-ing the chemical degradation of the cyclodienes.

Analysis of the data on the dissipation of the cyclo-dienes has indicated that the loss of the more volatilecyclodienes, aldrin, dieldrin and endosulfan I, fromunsealed and butyl rubber sealed treatments, is bi-phasic.

The fact that there was very little difference betweenthe rates of dissipation of aldrin and dieldrin frommedia sealed with butyl rubber and that which wereunsealed, showed that butyl rubber sealing was ineffec-tive. As known from previous biodegradation studies,such a rubber seal is therefore unsuitable for microbialdegradation studies when endosulfan or other volatile/semi-volatile compounds are studied. Although butylrubber has a very low permeability towards oxygen, ithas a high affinity for organic compounds (e.g. hexaneand volatile organochlorine pesticides). Conversely,PTFE, due to its very low coefficient of friction andresistance to chemical reaction, has an extremely lowporosity to volatile/semi-volatile organic compounds.PTFE-lined rubber therefore provides an ideal seal foranaerobic degradation studies with compounds of highvolatility, for example see Guerin (1999a).

A significant finding was the complete absence of theformation of endosulfan sulfate. This is a toxic degra-dation product (Table 1), and is the major oxidativeproduct of endosulfan in the environment. This was truefor both sterile incubations and incubations containingsoil extracts in the well defined liquid media used forcultivating anaerobic and aerobic bacteria. From thecurrent study, it is unlikely that endosulfan sulfateforms in naturally occurring waters under anaerobicconditions, either with or without microorganisms pres-ent. However, endosulfan diol was formed in the sterileincubations, indicating that this degradation productmay be formed in the absence of any microbial activity.Furthermore, under the conditions described, there was

no interconversion between the parent isomers of endo-sulfan during the study period.

Further implications for studying the behaviour ofchlorinated organic compounds in aqueous systems arealso given (Table 7). It is imperative that in any aqueousincubation containing volatile/semi-volatile organiccompounds, such as endosulfan, aldrin or dieldrin, spe-cial precautions must be taken to reduce volatilization.An important demonstration in this study is that of thenecessity to seal aqueous incubation vessels with Teflon-lined butyl rubber seals to prevent volatilization whichwould have otherwise reduced the apparent half-lives ofthe compounds under study. For these volatile organo-chlorines, unlined butyl rubber was shown to be in-effective as a vessel stopper, and may even enhance theloss of these compounds from sterile aqueous media.These findings are fundamental to the design of futurebiodegradation experiments as losses of these com-pounds due to volatilization, as well as from chemicalhydrolysis, are also likely to occur. These losses canconfound the results of biodegradation experiments,making it difficult to determine which losses are actuallya result of biological activity. Therefore, these findingswere applied to the design of experiments aimed atdetermining the role of indigenous soil microorganismsin the biodegradation of endosulfan under anaerobicconditions and have allowed the biodegradation poten-tial of indigenous populations of anaerobic micro-organisms to be determined (Guerin, 1999a).

References

Anonymous, 1998. Review of endosulfan. Australia, National Regis-

tration Authority for Agricultural and Veterinary Chemicals, Can-

berra.

Cotham, W.E., Bidleman, T.F., 1989. Degradation of malathion,

endosulfan, and fenvalerate in seawater and seawater/sediment

microcosms. Journal of Agricultural and Food Chemistry 37,

824–828.

Goebel, H., Gorbach, S.G., Knauf, W., Rimpau, R.H., Huttenbach,

H., 1982. Properties, effects, residues and analytics of the insecticide

endosulfan. Residue Reviews 83, 1–122.

Greve, P.A., 1971. De persistentie van endosulfan in oppervlakte-

water. Ghent Rijksinstituut Mededlingen 36, 439–447.

Guerin, T. F. 1993. The relative significance of biodegradation and

physico-chemical dissipation of endosulfan from water and soil.

Department of Agricultural Chemistry and Soil Science University

of Sydney, Sydney, New South Wales, Australia, pp. 240.

Guerin, T.F., 1995. Anaerobic biodegradation of the chlorinated

hydrocarbon endosulfan. In: Hinchee, R.E., Hoeppel, R.E., Ander-

son, D.B. (Eds.), Bioremediation of Recalcitrant Organics, Vol. 7.

Battelle Press, Ohio, pp. 157–164.

Guerin, T.F., 1999a. The anaerobic degradation of endosulfan by

indigenous microorganisms from low-oxygen soils and sediments.

Environmental Pollution 106, 13–21.

Guerin, T.F., 1999b. Natural attenuation of a low mobility chlori-

nated insecticide in low-level and high-level contaminated soil: a

feasibility study. Remediation 9, 51–63.

Guerin, T.F., Kennedy, I.R., 1991. The biodegradation of endosulfan

in cotton growing soils. The Australian Cotton Grower 12, 13–15.

T.F. Guerin / Environmental Pollution 115 (2001) 219–230 229

Guerin, T.F., Kennedy, I.R., 1992. Distribution and dissipation of

endosulfan and related cyclodienes in sterile aqueous systems:

implications for studies on biodegradation. Journal of Agricultural

and Food Chemistry 40, 2315–2323.

Guerin, T.F., Kimber, S.W.L., Kennedy, I.R., 1992. Efficient one-step

method for the extraction of cyclodiene pesticides from aqueous

media and the analysis of their metabolites. Journal of Agricultural

Food and Chemistry 40, 2309–2314.

Logan, D.C., Barry, M.J., 1996. The fate of endosulfan in aquatic

microcosms and effects on benthic communities. Intersect ’96, Syd-

ney, Australia, RACI, P155.

Mansingh, A., Robinson, D.E., Dalip, K.M., 1997. Insecticide con-

tamination of the Jamaican environment. Trends in Analytical

Chemistry 16, 115–123.

Mansingh, A., Wilson, A., 1995. Insecticide contamination of Jamai-

can environment III. Baseline studies on the status of insecticidal

pollution of Kingston Harbour. Marine Pollution Bulletin 30,

640–645.

Meylan, B., Howard, P., 2000. PhysProp and Datalog Databases.

Syracuse Research Corporation. (http://esc.syrres.com).

Miles, C.J., Pfeuffer, R.J., 1997. Pesticides in canals of south Florida.

Archives of Environmental Contamination and Toxicology 32,

337–345.

Miles, J.R.W., Moy, P., 1979. Degradation of endosulfan and its

metabolites by a mixed culture of soil microorganisms. Bulletin of

Environmental Contamination and Toxicology 23, 13–19.

Peterson, S.M., Batley, G.E., 1991. Fate and Transport of Endosulfan

and Diuron in Aquatic Ecosystems. CSIRO Division of Coal and

Energy Technology, Sydney, New South Wales.

Peterson, S.M., Batley, G.E., 1993. The fate of endosulfan in aquatic

ecosystems. Environmental Pollution 82, 143–152.

Schlanger, L.M., Baumgartner, E.R., 1980. Fluoroplastics. Modern

Plastics Encyclopedia 57, 31–34.

Schmidt, W.F., Hapeman, C.J., Fettinger, J.C., Rice, C.P., Bilboulian,

S., 1997. Structure and asymmetry in the isomeric conversion of

beta- to alpha-endosulfan. Journal of Agricultural and Food

Chemistry 45, 1023–1026.

Singh, N.C., Dasgupta, T.P., Roberts, E.V., Mansingh, A., 1991.

Dynamics of pesticides in tropical conditions. Kinetic studies of

volatilization, hydrolysis and photolysis of dieldrin and a- and b-endosulfan. Journal of Agricultural Food Chemistry 39, 575–579.

Southan, S.K., Kennedy, I.R., 1995. Dissipation of the pesticide

endosulfan from cotton field runoff water using an improved appa-

ratus for the simulation of field conditions. Fourth Environmental

Chemistry Symposium, Darwin, RACI, EO-22.

Walker, W.W., Cripe, C.R., Pritchard, P.H., Bourquin, A.W., 1988.

Biological and abiotic degradation of xenobiotic compounds in in

vitro estuarine water and sediment/water systems. Chemosphere 17,

2255–2270.

Worthing, C.R., Walker, S.B., 1987. The Pesticide Manual: a World

Compendium. The British Crop Protection Council, 1081, Thorn-

ton Heath, UK.

230 T.F. Guerin / Environmental Pollution 115 (2001) 219–230