Spatial and temporal regulation of the pesticide dieldrin within industrial catchments

The Science of the Total Environment 251r252 Ž2000. 255᎐263

Spatial and temporal regulation of the pesticide dieldrin within industrial catchments A.A. Meharg a,U,1, J. Wright a , G.J.L. Leeks b, P. Wass b, D. Osborn a a

Institute of Terrestrial Ecology 2 , Monks Wood, Abbots Ripton, Huntingdon, Cambridgeshire, PE17 2LS, UK b Institute of Hydrology 2 , Maclean Building, Crowmarsh Gifford, Wallingford, Oxfordshire, UK Accepted 5 January 2000

Abstract The river catchments of south Yorkshire support a very high density of wool processing industries. Dieldrin was once used as a moth proofing agent, as a sheep dip, and as a pesticide to protect wool fleeces during storage and transport, all of which caused pollution of these catchments due to textile processing. Weekly sampling of four of these rivers revealed two classes of dieldrin contamination: the Aire and Calder Žthe rivers which support very high concentrations of wool processing industries. had higher concentrations Žaveraging ; 3 ngrl. than the Don and Trent Ž; 1 ngrl.. The average flux of dieldrin from these rivers into the Humber estuary was 9.8 grday, with the Aire Žof which the Calder is a tributary . and the Trent contributing almost equally, with a smaller contribution from the Don. The Trent has the highest average flow, explaining its large contribution to dieldrin flux. Less detailed sampling of rivers from the north Humber catchment which drain predominantly rural areas had dieldrin concentrations similar to the heavily industrialized southern catchment rivers. This suggests that dieldrin from agronomic and domestic usage may be more persistent than the pollution caused by textile processing industries. Evidence is presented to suggest that the principle dieldrin sources to the Humber catchments are sewage treatment plants, and that the dieldrin sources are in rapid equilibrium with the water column. 䊚 2000 Elsevier Science B.V. All rights reserved. Keywords: Aire; Calder; Dieldrin; Don; Humber; Trent

U

Corresponding author. Present address: Department of Plant & Soil Science, University of Aberdeen, Cruickshank Building, St. Machar Drive, Aberdeen AB24 3UU, UK. 2 Component institute of the Natural Environment Research Council’s Centre for Ecology and Hydrology. 1

0048-9697r00r$ - see front matter 䊚 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 0 . 0 0 3 8 8 - 0

256

A.A. Meharg et al. r The Science of the Total En¨ ironment 251 r 252 (2000) 255᎐263

1. Introduction The organochlorine insecticide dieldrin was widely used as a moth proofing agent in the textile industries ŽBrown et al., 1979; Velduisen, 1991. which historically has led to considerable pollution of rivers in the southern Humber catchment ŽLowden et al., 1969; Brown et al., 1979; Boryslawskyi et al., 1985; Edwards et al., 1997; Meharg et al., 1998.. It is a red-listed substance under EU legislation due to its high toxicity and long environmental persistence ŽMeharg et al., 1998.. Rivers feeding this catchment ŽAire, Calder and their tributaries . support high densities of textile processing industries ŽLowden et al., 1969; Brown et al., 1979; Boryslawskyi et al., 1985; Edwards et al., 1997.. Levels of over 6000 ngrl dieldrin have been recorded in sewage treatment effluent and 5000 ngrl in the water column of rivers on this catchment during 1976 ŽBrown et al., 1979.. Since this period, levels have dropped dramatically, generally falling below their Environmental Quality Standard of 10 ngrl, although occasional samples still exceed this value ŽEdwards et al., 1997; Meharg et al., 1998.. Observed decreases were concurrent with phasing out of the use of dieldrin as a moth proofing agent: it ceased to be used by 1986 ŽVelduisen, 1991.. However, there still may be current sources of dieldrin in this catchment, particularly effluent from wool scouring ŽShaw, 1994.. To investigate the spatial and temporal dynamics of dieldrin and its fluxes into the North Sea Žvia the Humber Estuary., its concentration was monitored routinely Žat least weekly. in the southern Yorkshire Rivers ŽAire, Calder and Don. and the contrasting River Trent, draining in the east midlands. All these rivers feed the Humber estuary. Fine scale regulation of dieldrin within these catchments was investigated.

2. Materials and methods 2.1. Sampling The rivers Aire, Calder, Don and Trent were sampled at weekly intervals from February 1995

to February 1997. Additional samples were collected during spate events at 2᎐3-h sampling frequencies. Sampling strategy and locations are outlined in Leeks et al. Ž1997.. The Aire, Don and Trent were sampled from fixed points on their lower reaches, upstream of the tidal limit, but as close to this limit as practically possible given the constraints imposed by access and by sampling protocols. The Calder is a tributary of the Aire. Mid-river water samples, in well mixed reaches of the rivers, and downstream of all sewage treatment plants were obtained. Samples were collected in 500-ml glass bottles with airtight PTFE lined tops which were pre-rinsed with HPLC grade acetone and hexane and dried with a stream of electron capture grade ŽECD. grade nitrogen ŽDistillers MG, UK. prior to use. The samples were stored at 4⬚C in the dark before extraction. Further samples were collected from the rivers draining agricultural catchments in N Yorkshire ŽSwale, Nidd, Wharfe, Ouse, Derwent and Ure. to contrast with the southern industrialized rivers using the sampling locations outlined in Leeks et al. Ž1997.. An integrated event monitoring system was designed, using commercially available components and integrating them to provide a system capable of monitoring in situ turbidity as well as collecting samples throughout spate events Leeks et al. Ž1997.. The main components were a logging device ŽCampbell CR10., nephelometric and absorptiometric turbidity sensors ŽPartech IR12 and Partech IR40, respectively., a depth measuring device ŽDruck 208 pressure transducer . and river water automatic bulk sampler ŽEpic 1011 wastewater sampler., described in detail by Wass et al. Ž1997.. 2.2. Sample preparation The contents of the sample bottle were mixed by shaking and inversion and 500 ml transferred to a 500-ml volumetric flask without filtering. The contents of the flask were then sequentially extracted with 4 = 10 ml of hexane ŽHPLC grade, Rathburns, Scotland. and the hexane extracts were transferred to a Kuderna-Danish flask via a funnel containing anhydrous sodium sulphate.

A.A. Meharg et al. r The Science of the Total En¨ ironment 251 r 252 (2000) 255᎐263

The sodium sulphate was rinsed with 10 ml of hexane which was also collected in the KudernaDanish flask. The extract was reduced to approximately 1.5 ml at 80⬚C and then further reduced to exactly 1 ml using a stream of nitrogen ŽECD grade.. The extract was purified using alumina column chromatography. The alumina column consisted of a 0.8-mm i.d. glass column plugged with silanized glass wool and packed with 0.8 g of alumina. The alumina Žaluminium oxide, neutral, Brockman grade 1, Merck, UK. was prepared by heating in a furnace at 700⬚C for 4 h, followed by deactivation with 5% water. The whole extract was placed on the alumina column and eluted with hexane. The first 5 ml of the eluant was collected and reduced to exactly 0.5 ml under a stream of ECD grade nitrogen. An internal standard Ž2,6-dichlorobenzonitrile. was added to the cleaned up extract. Dieldrin was determined by gas chromatography with electron capture detector ŽGC-ECD.. A blank sample bottle was included with the test sample bottles during the sampling procedure to take account of contamination during transport and from exposure to the atmosphere whilst sampling. In the laboratory, the blank bottle was rinsed with 5 = 10-ml aliquots of hexane. This hexane extract was then treated in the same way as the samples. A recovery test was performed for every third batch of samples. Sieved air dried soil Ž0.5 g. was added to a 500-ml volumetric flask. The soil was then spiked with 1 ml of hexane containing dieldrin to simulate extraction from the sediment phases. The hexane was gently evaporated with a stream of ECD grade nitrogen: the volumetric flask was then made up to the mark with hexane washed distilled water. The contents of the flask were shaken and left overnight before being analysed. A portion of the spiking solution was kept for analysis with the recovery sample. The data presented in this paper have not been corrected for recoveries. Limits of detection Žl.o.d.. were determined for each batch, and were calculated as four times the standard deviation of the blank noise at the point in the chromatograph where the compound eluted. The limits of detection in

257

water samples was calculated as 0.02 ngrl., and the recovery was 84%. 2.3. GC-ECD analysis The GC-ECD model was a Varian 3400 gas chromatograph interfaced with a Varian 8200CX autosampler, splitrsplitless injector and electron capture detector. The column was a 50 m= 0.22 mm i.d. column ŽSGE ŽUK. Ltd.. connected with glass press-fit connector to a 5 m= 0.22 mm non-polar deactivated retention gap ŽSGE ŽUK. Ltd... A single 5-␮l injection was used per run. The injector was held at a temperature of 200⬚C and operated in splitrsplitless mode with the split valve opening after 2 min. The detector temperature was 250⬚C, and the detector make-up gas was ECD grade nitrogen at 25 ml miny1 . The carrier gas was hydrogen with an average linear velocity of 450 mm sy1 . The column oven temperature program was: an initial temperature of 60⬚C held for 2 min, 4.5⬚C miny1 to 170⬚C, held for 2.5 min, 2.5⬚C miny1 to 200⬚C held for 5 min, 2⬚C miny1 to 280⬚C held for 0 min and then 40⬚C min to 320⬚C held for 2 min. The identity of a chromatographic peak was found by comparison of its relative retention time with the relative retention times of the peak in the dieldrin standard. This operation was performed automatically by the chromatography data system. Using graphical in-house software a visual check was then made of the assigned peak identities, by overlaying sample, blank and standard chromatograms. Quantification was by the internal standard method. Samples were run in batches of eight. Standards were run before and after the samples and the response factor used was an average of the two. Blanks were run before and after standards to check for carryover. 2.4. Measurement of instantaneous flow The measurement of water discharge and bulk sampling for analyses were carried out at the same monitoring sites. Instantaneous discharge was recorded every 15 min from the ultrasonic flow gauges on the Trent and Calder, a broadcrested weir velocity area gauging station on the

A.A. Meharg et al. r The Science of the Total En¨ ironment 251 r 252 (2000) 255᎐263

258

Aire and a velocity area gauging station on the Don. Suspended solids was determined as outlined in Leeks et al. Ž1997.. 2.5. Statistical modelling General Linear Modelling ŽGLM. was used to analyse the spatial and temporal patterns in the data ŽMinitab v.11, Minitab Inc., PA.. To enable investigation of the temporal patterns, data were classed by month and sampling year, with on average 3.5 measurements per determinand for each month.

3. Results and discussion Mean concentrations of dieldrin on the Aire and Calder were ; 3 ngrl, contrasting with the Don and Trent which had concentrations one third of this value ŽTable 1.. Maximum values approached the Environmental Quality standard of 10 ngrl on the Aire and exceeded it on the Calder. Both the Don and Trent maximum values were well below their EQS levels ŽTable 1.. The

levels on the Aire are consistent with a long-term monitoring program at the tidal limit Ža similar sampling location was also used in this study. for this river ŽEdwards et al., 1997.. The study of Edwards et al. Ž1997. showed that dieldrin had dropped on this sampling location from values reaching 300 ngrl in 1976 to their current levels. Other historical data also shows the extent to which this catchment was once polluted with dieldrin by the moth proofing industries. In 1976 dieldrin levels were recorded at 4900 ngrl on tributaries of the Calder ŽBrown et al., 1979. and values above 200 ngrl were regularly recorded on a number of sampling points on both the Aire and Calder ŽLowden et al., 1969.. In a study conducted in 1981 dieldrin levels in a sewage treatment outfall situated on a tributary of the Calder were reported as being as high as 10 110 ngrl, with river concentrations regularly exceeding 1000 ngrl. The Don in contrast, in the study of Lowden et al. Ž1969., had much lower concentrations, with a maximum value of 34 ngrl. The decline of dieldrin in the Aire and Calder can be attributed to the phasing out of the use of dieldrin in moth proofing ŽVelduisen, 1991.. How-

Table 1 Summary statistics for instantaneous flow, dieldrin concentrations and fluxes for the Aire, Calder, Don and Trent a River

Instantaneous flow Mean Žm3 sy1 .

Min. Žm3 sy1 .

Aire Calder Don Trent

30.3 16.7 11.6 57.3

River Aire Calder Don Trent

Dieldrin concentration Mean Žngrl. 2.9 3.0 1.2 1.0

Min. Žngrl. 0.3 - 0.02 - 0.02 - 0.02

Max. Žngrl. 9.9 16.0 4.0 3.9

River Aire Calder Don Trent

Dieldrin flux Mean Žgrday. 5.5 2.6 1.0 3.3

Min. Žgrday. 0.6 - 0.02 - 0.02 - 0.02

Max. Žgrday. 21.8 11.6 5.9 9.9

a

Data presented are the mean of monthly averages.

7.0 2.0 2.5 20.0

Max. Žm3 sy1 . 201.0 118.0 83.1 260.0

A.A. Meharg et al. r The Science of the Total En¨ ironment 251 r 252 (2000) 255᎐263

ever, there is evidence that other sources may be currently contributing to dieldrin levels on these rivers. Wool fleeces imported from abroad often contain pesticides which are currently banned in the UK such as HCH formulations and dieldrin ŽShaw, 1994.. These chemicals are used abroad in sheep dips and used to preserve fleeces during storage and transportation. When fleeces are scoured Žthe first stage in wool processing. to remove grease, fat soluble chemicals such as HCH isomers and dieldrin are released into sewage treatment plants in the scouring effluent ŽShaw, 1994.. Shaw Ž1994. calculated that wool scouring on the Aire and Calder would give rise to river concentrations of 0.3᎐0.5 ngrl and 0.1᎐0.6 ngrl, respectively, based on an estimate that 90% removal of dieldrin by sewage treatment works ŽSTWs.. However, this 90% removal was a general figure for organochlorine pesticides. Croll Ž1969. found that conventional STWs did not significantly reduce dieldrin levels and if this is the case wool scouring would contribute considerably to the concentrations observed in this study ŽTable 1.. Dieldrin could have agronomic sources as it was once highly used for seed dressing and sheep dipping. However, even at the height of dieldrin usage, monitoring showed that in catchments which supported agronomic and industrial activities that industrial sources dominated as illustrated by levels reported in the Severn basin during 1966 ŽCroll, 1969.. Agronomic use of dieldrin was progressively limited from 1976 onwards. These historic uses may explain some of the dieldrin contamination observed in the Humber catchment. Some of the rivers ŽSwale and Nidd. draining agronomic regions of the Humber catchment had average dieldrin levels approaching those observed on the Aire and Calder ŽTables 1 and 2.. All the other North Yorkshire rivers ŽWharfe, Derwent and Ure. had levels of dieldrin similar to those reported on the Don and Trent ŽTables 1 and 2.. Although the Don and Trent support major urban conurbations ŽSheffield, and Leicester and Nottingham, respectively., they also have a large agronomic contribution to their catchments as they drain the Staffordshire and Derbyshire uplands and lowlands, which tend

259

Table 2 Concentrations of dieldrin in rivers draining the predominantly agricultural catchments of North Yorkshire collected during 1994r1995 River

Number of samples

Conc. Žngrl.

Swale, Catterick Bridge Swale, Thornton Manor Nidd, Skip Bridge Wharfe, Tadcaster Ouse, Clifton Bridge Ouse, Acaster Malbis Derwent, Budwith Ure, Boroughbridge

4 9 7 15 13 18 12 7

0 2.6 2.7 1.5 1.6 0.8 1.0 0.8

to be fertile pastures ŽJarvie et al., 1997.. Mixed farming also occurs in Midlands catchments drained by the Trent. That fact that the once highly polluted Aire and Calder rivers now have concentrations typical of rivers draining rural catchments may indicate that the persistence of dieldrin in these river systems is greater in rural compared with industrial catchments. The Aire has the highest dieldrin flux followed by the Trent ŽTable 1.. These fluxes are low, ranging from 1 to 5.5 grday for the four rivers. In terms of longer term changes in flux it is important to compare these fluxes with historic measurements Ž1968. at similar sampling locations on the Aire and Calder ŽLowden et al., 1969.. Current flux on the Aire Ž5.5 grday. is more than double that for the Calder Ž2.6 ngrday., primarily due to the larger instantaneous flow on the Aire ŽTable 1.. In 1968 they showed that the Calder dominated dieldrin flux with 630 grday compared with 200 grday on the Aire, reflecting the high density of moth proofing activities on rivers feeding the Calder ŽBrown et al., 1979; Boryslawskyi et al., 1985.. The decline of dieldrin flux on the Calder again emphasizes that the historic pollution of this catchment with dieldrin has greatly dissipated. There are a number of papers reporting temporal trends on the Aire and Calder ŽLowden et al., 1969; Brown et al., 1979; Boryslawskyi et al., 1985; Edwards et al., 1997.. However, these do not correlate concentrations with other parameters. Edwards et al. Ž1997. reports long-term trend

260

A.A. Meharg et al. r The Science of the Total En¨ ironment 251 r 252 (2000) 255᎐263

data for the Aire showing very large decreases in concentrations while Boryslawskyi et al. Ž1985. showed that there were very clear within week

cycles in dieldrin concentrations within rivers receiving effluent from factories which were moth proofing textiles. Croll Ž1969. reported temporal

Fig. 1. Monthly averages of dieldrin concentrations Žfilled histograms., dieldrin flux Žtriangles., temperature Žblack cross with grey line., instantaneous flow Žsolid line. and chloride concentration Ždashed line. for the rivers Aire, Calder, Don and Trent.

A.A. Meharg et al. r The Science of the Total En¨ ironment 251 r 252 (2000) 255᎐263

patterns in dieldrin levels for a number of rivers in Essex and Kent with high resolution over a complete year. No clear seasonal related patterns emerged. In the study reported here, there do appear to be some seasonal patterns of dieldrin concentrations and fluxes within all four rivers ŽFig. 1.. Concentrations appear to rise during late summerrearly autumn over both years of sampling, although concentrations stay elevated for longer during the second year. The rise in both the concentration and the flux of dieldrin occurs after sustained periods of low flow and high river temperatures. The inter-year comparisons may be difficult to interpret due to the unusually low instantaneous flows observed on the Aire and Calder during the winter of the second year ŽFig. 1..

261

Table 3 General Linear Model analysis of factors Žriver, chloride concentration and temperature. correlating with monthly averaged dieldrin concentrations on the Aire, Calder, Don and Trent Source

d.f.

Adj. MS

F

P

River Cl Temperature River= Cl River= temp. Cl = temp. River= Cl = temp. Error

3 1 1 3 3 1 3 78

12.2 11.9 7.5 9.1 15.3 3.2 12.0 3.4

3.64 3.53 2.22 2.70 4.62 0.96 3.58

0.016 0.064 0.140 0.051 0.005 0.330 0.018

Individual dieldrin concentrations were correlated with a wide range of factors Žparticulate,

Fig. 2. Plots of dieldrin concentrations Žcircles with fitted dashed line., instantaneous flow Žcrosses with dotted lines. and suspended sediment Ždot-dash line. for a spate event on the Calder and two events on the Trent.

262

A.A. Meharg et al. r The Science of the Total En¨ ironment 251 r 252 (2000) 255᎐263

organic carbon, nitrate, instantaneous flow, particulate iron and aluminium, suspended sediment, chloride, temperature. for each river only chloride and temperature proved significant. A GLM was devised to study the interaction between river, chloride and temperature ŽTable 3.. Other GLM models were performed but no other parameters were significant Ždata not shown.. In any case, many of the parameters were inter-related, i.e. high chloride occurs at low flows, high temperatures are associated with low flows, suspended sediment is highly positively correlated with instantaneous flow etc. The GLM model ŽTable 3. shows that there were significant differences in dieldrin concentrations between rivers, a river = chloride and river = temperature interaction, and a significant river times chloride times temperature interaction. As high chloride concentrations are associated with low flows and high temperatures, and chloride sources are dominated by sewage treatment plants, the temporal modelling suggests that dieldrin is predominantly arising from STWs which dominate flux under low flow conditions. This in line with other studies on dieldrin sources to rivers ŽCroll, 1969; Lowden et al., 1969; Brown et al., 1979; Boryslawskyi et al., 1985.. Chloride concentrations did not drop during the first full winter over which sampling was conducted, associated with low river flows measured over that period ŽFig. 1.. Evidence from sampling on the Trent and Calder during winter spate events indicate that dieldrin sources are in rapid equilibrium with the water column as little or no dilution occurs on doubling of instantaneous flow ŽFig. 2.. Similarly, sediment loadings clearly do not regulate dieldrin concentrations as sediment loadings increase two to fivefold during spates ŽFig. 2., confirming the observation from weekly sampling that sediment concentrations are not closely correlated with dieldrin levels in the water column. However, data from Brown et al. Ž1979. indicate that 20᎐30% of dieldrin is associated with particles larger than 0.7 ␮m for Mag brook Ža minor tributary of a river feeding the Calder., which at the time of sampling was a highly polluted Žtotal dieldrin concentration of ; 4000 ngrl.. For a

much less polluted river feeding Chesapeake Bay in the US, 43% of dieldrin was associated with particles greater than 0.7 ␮m ŽGodfrey et al., 1995.. To unravel the regulation of dieldrin within the Humber catchment more detailed studies looking at dieldrin association with particulates at point sources Žsuch as STWs., rather than just total water column measurements at the confluence of rivers are required.

4. Conclusion The Aire and Calder appear to have recovered from massive dieldrin contamination from the textile industries sustained during the 1960s᎐ 1980s, now having concentrations typical of rivers draining rural catchments. This is encouraging on a national and international level, as dieldrin was widely used in moth proofing industries worldwide. It appears that its residence time in industrial catchments is relatively short in the context of other organochlorine contamination scenarios. STWs appear to be the principle source of dieldrin to the four rivers ŽAire, Calder, Don and Trent. investigated in detail, although the need for further study has been highlighted if this is to be verified, and the mechanisms underpinning dieldrin source regulation understood. There is strong seasonal cycling in dieldrin concentrations, indicating release from historic stores during summer months of high temperature and low river flows. This study concentrated on catchments where industrial sources were the dominant dieldrin source. Surveys of agricultural rivers in the N Yorkshire catchment however, had similar current levels of dieldrin as the industrialized rivers. Understanding the release of dieldrin from historic agricultural sources is highlighted as an area of future study.

Acknowledgements We wish to thank the research staff at the LOIS RACS-R laboratory in York for conducting the sampling and Isabella Tindall at IH for

A.A. Meharg et al. r The Science of the Total En¨ ironment 251 r 252 (2000) 255᎐263

managing the LOIS database. Janice Horne assisted with sample preparation at Monks Wood and we are grateful for her help. References Boryslawskyi, M., Garrood, A., Morphy, M.J., 1985. Spatial and temporal patterns of dieldrin pollution in the Holme Catchment, West Yorkshire, England. Environ. Pollut. Ser., B 10: 129᎐139. Brown, L.G., Bellinger, E.G., Day, J.P., 1979. Dieldrin pollution in Holme catchment, Yorkshire. Environ. Pollut., 18: 203᎐211. Croll, B.T., 1969. Organo-chlorine insecticides in water ᎏ Part I. J. Water Treat. Exam., 18: 255᎐274. Godfrey, J.T., Foster, G.D., Lippa, K.A., 1995. Estimated annual loads of selected organic contaminants to Chesapeake Bay via a major tributary. Environ. Sci. Technol., 29: 2059᎐2064. Jarvie, H.P., Neal, C., Robson, A.J., 1997. The geography of the Humber catchment. Sci. Total Environ., 194r195: 87᎐99.

263

Leeks, G.J.L., Neal, C., Jarvie, H.P., Casey, H., Leach, D.V., 1997. The LOIS river monitoring network: strategy and implementation. Sci. Total Environ., 194r195: 101᎐109. Lowden, G.F., Saunders, C.L., Edwards, R.W., 1969. Organochlorine insecticides in water ᎏ Part II. J. Water Treat. Exam., 18: 275᎐287. Meharg, A.A., Wright, J., Osborn, D., 1998. The frequency of environmental quality standard ŽEQS. exceedence for chlorinated organic pollutants in rivers of the Humber catchments. Sci. Total Environ., 210r211: 219᎐228. Shaw, T., 1994. Agricultural chemicals in raw wool and the wool textile industry. J. Inst. Water Environ., 8: 287᎐290. Velduisen, D.R., 1991. Technical and economic aspects of measures to reduce water pollution from the textile finishing industry. Report to the Commission des Cummunautes Europeennes. ´ Wass, P.D., Marks, S.D., Finch, J.W., Leeks, G.J.L., Ingram, J., 1997. Monitoring and preliminary interpretation of in-river turbidity and remote sensed imagery for suspended sediment transport studies in the Humber catchment. Sci. Total Environ., 194r195: 263.