FEMS Microbiology Ecology 49 (2004) 71–82 www.fems-microbiology.org
The role of microbial community composition and groundwater chemistry in determining isoproturon degradation potential in UK aquifers Andrew Johnson a
a,*
, Neville Llewellyn a, Jennifer Smith a, Christopher van der Gast b, Andrew Lilley b, Andrew Singer b, Ian Thompson b
Centre for Ecology and Hydrology Wallingford, Benson Lane, Wallingford, Oxfordshire OX10 8BB, UK b Centre for Ecology and Hydrology Oxford, Mansfield Road, Oxfordshire OX1 3SR, UK Received 28 November 2002; received in revised form 15 September 2003; accepted 23 March 2004 First published online 17 April 2004
Abstract The community response of indigenous sandstone, chalk and limestone groundwater microorganisms to the addition of the commonly used herbicide isoproturon was examined. The addition of 100 lg l1 isoproturon generally caused an increase in species diversity determined by chemotaxonomic analysis (fatty methyl ester analysis) of isolates resulting from incubation of cultures at 18 C for 4 days. Amongst the groundwater samples to which isoproturon was added, isoproturon degradation rates were correlated with increasing dominance of a few species. However, the changes in community profile associated with isoproturon degradation varied from site to site. Repeated sub-culturing with 100 lg l1 isoproturon and sterile groundwater was carried out to examine whether this level of pesticide could exert a selection pressure, and hence stimulate more rapid degradation. Significantly increased degradation was observed in a groundwater sample from the chalk, but not in sandstone, or limestone samples. The addition of filter-sterilised sandstone groundwater to bacteria on filter paper from slow degrading limestone sites significantly improved their degrading performance. The addition of filter-sterilised limestone groundwater to the sandstone bacteria reduced their degradation rate only slightly. The data suggested that the nature of the indigenous community does influence pesticide degradation in groundwater, but that the groundwater chemistry may also play a role. 2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved. Keywords: Groundwater; Isoproturon; Biodegradation; Microbial community response
1. Introduction The Chalk, Permo-Triassic sandstone and Jurassic limestone aquifers supply approximately a third of the drinking water requirements of England and Wales in the UK [1]. Agriculture over these unconfined aquifers includes the production of winter cereal with which herbicide use is associated. Herbicides have penetrated to groundwater, albeit at low concentrations (sub lg l1 ), in the UK [2,3]. Despite its importance as a UK *
Corresponding author. Tel.: +44-1491-838800; fax: +44-1491692424. E-mail address:
[email protected] (A. Johnson).
drinking water resource we know very little about the microbial ecology of the groundwater in the fractured rock aquifer environment. In particular, we do not know what factors influence the response of the indigenous microorganisms of the groundwater environment to low levels of pesticide contamination. Isoproturon, 3-(4-isopropylphenyl)-1,1-dimethylurea, is used frequently in Europe to suppress Blackgrass growth in winter cereals. It is the most widely used organic pesticide used in the UK with approximately 3300 tonnes applied in 1997 [4]. It has been demonstrated that bacteria with the competence to degrade isoproturon can be found in chalk, limestone, sandstone and alluvial aquifers in the UK [5–8]. However, these studies have
0168-6496/$22.00 2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.femsec.2004.03.015
72
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
shown that this potential is inconsistent, with wide variations in degradation rate occurring within the same field, and even boreholes ‘losing’ and regaining the ability to degrade isoproturon [6]. Also the degradation rates vary from negligible to fast between different aquifer types [7]. What are the limiting factors which cause this wide variation in degradation rates between different groundwater samples? A number of possibilities present themselves. Might the concentration of the pesticide, or previous exposure to high concentrations of pesticides in the groundwater influence subsequent degradation rates? Pahm and Alexander [9] observed reduced mineralisation of p-nitrophenol at 1–5 lg l1 compared to 100 lg l1 . Similarly Tor€ ang et al. [10], observed differences in the 2,4-D degradation rate depending on the spiking concentration in sandy aquifer samples from Denmark. Biodegradation rates achieved in microcosms from chalk groundwater samples in the UK were not correlated with previous exposure to isoproturon in the boreholes [6]. Thus, in groundwater subject to contamination by only very low and transient concentrations of pesticides, insufficient selection pressure probably exists to stimulate or sustain pesticide degrading communities. Might degradation be linked to the local groundwater chemistry? Unlike the UK, an isoproturon degradation potential has been absent in most Danish shallow sandy aquifers [11,12]. A possible explanation for its absence may be related to pH, since, many Danish groundwaters, appear to be more weakly acid (pH 5.3–6.4) than those mostly found in the fractured rock aquifers used for drinking water supply in the UK [13,14]. Studies into variations in isoproturon degradation potential in a British soil found shorter half-lives were positively correlated with increasing pH [15]. However, as the majority of the major UK aquifers have a neutral to weakly alkaline pH, so pH alone cannot explain the inter-site differences observed in the UK [7]. If the degradation is cometabolic, due to a fortuitous similarity with a natural substrate, then concentration would be less important than the size of the existing degrading population and the presence of the cometabolite. For instance, secondary plant metabolites have been demonstrated to stimulate microbial degradation of xenobiotics a phenomenon referred to as analogue enrichment [16,17]. Another explanation is that differences in degradation rates are a function of the presence, or abundance of competent microbial consortia. Recent in situ studies in aquifers have revealed that continuous exposure to a mixture of herbicides (each at 40 lg l1 ) caused acclimation of exposed microbial communities, and the development of similar community structures [18]. Similarly, soil studies also suggest that isoproturon degradation may be dependent on subtle interactions
between different species in a microbial consortium [19]. Investigations into the impact of isoproturon on soil communities found that the composition and metabolic potential of the community can change with exposure [20,21]. For instance Variovorax, a genus that has previously been implicated in the degradation of phenlyurea and other herbicides [22], has been reported to become more abundant. To better understand the microbial ecology and possible limiting factors associated with isoproturon degradation in groundwater the following hypotheses were tested: (1) That the addition of isoproturon will influence the species diversity and community structure. (2) That rapid isoproturon degradation rates are associated with particular changes in the microbial community. (3) That rapid isoproturon degradation is related to and can be predicted by the presence of certain key species. (4) That competent degrading bacteria are universally present in groundwater samples, but that degradation is stimulated by factors associated with the local groundwater chemistry. The principle method used was the incubation of fresh groundwater samples with 100 lg l1 isoproturon. This technique has been used previously to test whether indigenous microorganisms have the competence to degrade isoproturon [5], and has revealed surprisingly variable responses in the microbial community to isoproturon within the same field site [6]. Thus, whilst using isoproturon at the 100 lg l1 level cannot tell us that degradation would actually occur in the natural situation (usually no more than 0.3 lg l1 [23]), it can reveal real differences in the microbial ecology between different groundwater sites.
2. Materials and methods 2.1. Sample collection from the field In order to study the Triassic sandstone aquifer environment, sites were selected on the outcrop near Mansfield in Nottinghamshire. The locations were at the Gleadthorpe farm (GT2), Welbeck, and Clumber Park, which are no more than 10 km from one another (Table 1). These sites are within the Eastern portion of the Sherwood Sandstone, where the groundwater chemistry is dominated by calcium-bicarbonate [24]. At present, around 10% of the drinking water requirement for England and Wales comes from this type of aquifer. At Welbeck and Clumber Park groundwater samples were taken from a sample tap. To reduce the potential for contamination, the water samples were only collected after the sample taps had been opened for 30 min and over a 100 l discarded.
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
73
Table 1 Description of borehole sites used for groundwater sampling Site
Site landuse
Soil type
Installation year
Borehole description
Isoproturon exposure/ presence?
GT2 (sandstone)
Experimental farm. Winter cereals used
Cuckney series, brown sand of slightly stony loamy sand to 0.9 m [35]
1995
Drilled to 17.6 mbs [7] with plastic casing extending to 12 mbs
Isoproturon applied in this, or adjoining field in 1996, 1998 and 2000
Clumber Park (sandstone)
In woodland with the nearest agriculture over 2 km away
As above
Pre-1909
7 mbs, in use to supply the estate with 67–135 m3 drinking water per day
No measurements
Welbeck (sandstone)
In a 0.25-ha wood surrounded by permanent pasture
As above
1988
90 mbs, cased out in mild steel, a sample tap in the pump house
No measurements
Site WON (chalk)
In fields which undergo arable crop rotation
Andover series, slightly stony silty clay loam over fragmented chalk with brown soil to 0.5 mbs [36]
1991
10 mbs cased out with plastic tube
0.2–0.05 lg l1 (1995–1998) [23]
Western Court (chalk)
In fields which undergo arable crop rotation
As above
1992
15 mbs cased out with plastic tube
0.06–0.3 lg l1 at the latter (D. Gooddy pers comm.)
Bridgets (chalk)
Permanent grassland
As above
1994
30 mbs cased out with plastic tube
No measurements
Coleby (limestone)
Beside road with surrounding fields being both arable and pasture
Elmton 1 series, brown, calcareous, slightly stony clay loam to sandy clay loam to 0.3 mbs [35]
1974
Monitoring borehole drilled to 15.5 m with a solid lining to 2 mbs
One measurement only carried out in 2000 revealed 0.04 lg l1
Welbourn (limestone)
Beside road, surrounded by arable farmland
As above
1976
Monitoring borehole drilled to 22.5 m with a solid lining to 3.5 mbs
No measurements
Chalk aquifers provide around 18% of the drinking water requirement for England and Wales [1]. For the chalk area, samples were taken from observation boreholes previously installed by BGS at Site WON, Bridgets farm and Western Court which are 3 km distant from one another in Hampshire (Table 1). Limestone aquifers provide around 15% of the drinking water requirement for England and Wales [1]. For this area, samples were taken from Coleby and Welbourn (Table 1). These boreholes were within 14 km of each other. Groundwater was collected from the observation boreholes using a small submersible electric pump. Five borehole volumes were pumped out and discarded before collecting samples in sterile bottles. Groundwater was stored (24 h maximum) at 4 C prior to use. 2.2. Site measurements Determination of the unstable parameters pH, redox potential (Eh), dissolved oxygen (DO2 ), alkalinity and temperature was carried out on site. Conductivity was also measured in the field using a portable conductivity bridge and meter (Mettler-Toledo, Schwerzenbach,
Switzerland). A flow through cell connected directly to the pumped supply was used for the measurement of pH, Eh (platinum electrode, Mettler-Toledo) and dissolved oxygen (DO2 membrane electrode and meter, Mettler-Toledo) in order to represent as closely as possible the in situ conditions. Bicarbonate was measured in the field using a digital titrator (Hach, Loveland, USA). Following 0.45-lm filtration, dissolved organic carbon (DOC) was measured using a TOCsin II Aqueous Carbon Analyser (Phase Separations Ltd, Queensferry, UK). The detection limit is 0.2 mg l1 (0.1 mg l1 ). 2.3. Degradation studies with microcosms The microcosms were prepared by placing 12 g of matrix material previously extracted from the unsaturated zones 4–6 mbs of either site WON for incubation of chalk groundwater, Gleadthorpe for incubation of sandstone groundwater or Longwood (near Lincoln) for the limestone groundwater into 120 ml sterile screw-top plastic containers (Bibby-Sterilin, Stone, UK). These matrix samples had been recovered one year previously and stored in their original sealed 10 cm diameter, 50 cm
74
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
long core liners [6,7] using either dry percussion or rotary air flush techniques to avoid contamination with drilling fluids. The matrix samples were dried and ground to 1–4 mm diameter before use. The containers with their solid matrix components were then autoclaved at 121 C and 15 psi for 30 min. Triplicate 70 ml aliquots of the groundwater samples were added to the appropriate containers. Groundwater samples from site WON and Western Court were used to represent chalk groundwater, and GT2, Welbeck and Clumber Park to represent sandstone groundwater. Previous studies have demonstrated that in the absence of matrix material, groundwater degradation rates are minimal [5,25]. From a 150 mg l1 isoproturon methanol stock, 3.2 ml were taken and made up to 60 ml in pure water to give an 8 mg l1 working stock which was 0.45 lm filtered (PTFE Acrodisc, Gelman, Ann Arbor, USA). From this working stock 0.875 ml was used to spike the groundwater samples to give a final concentration of 100 lg l1 . The controls including the relevant matrix material were autoclaved (30 min) as described above, but after the addition of groundwater and prior to spiking with the pesticide. In order to address concerns caused by trace quantities of methanol (0.06%) carried over into the microcosms and influencing the microbial community, 0.04 ml methanol was added to some 70 ml groundwater samples without isoproturon. To study the impact of isoproturon on the communities a third non-sterile treatment was left un-spiked by either isoproturon or methanol. Incubation was at 20 C and samples taken at approximately 50-day intervals over 250 days. The containers were sampled by drawing off 1.5 ml by sterile disposable pipette (polyethylene, Bibby-Sterilin, Stone, UK). Samples were then introduced to a 2-ml syringe and passed through a 0.45-lm PTFE filter (Gelman, Ann Arbor, USA) into a glass vial and stored at 4 C, prior to chemical analysis. 2.4. Microbial counts On receipt at the laboratory and before use in the microcosms a simple viable count of the groundwater samples was carried out. This involved taking triplicate 1 ml sub-samples of groundwater which were added to 9 ml of quarter strength Ringer’s solution [26] and vortexed for 1 min. Suspensions were serially diluted and 100 ll spread plated onto R2A (Difco, UK) amended with 50 mg l1 cyclohexamide (Sigma, UK) to prevent fungal growth, to determine bacterial counts. The plates were incubated at 18 C for 4 days and those containing between 20 and 200 colonies counted. Counts were expressed as colony forming units (cfu) per ml of groundwater. In addition samples were taken aseptically from microcosms; for the chalk groundwater samples 40 and 146 days and the sandstone samples 15 and 132 days after establishment, respectively. The microcosms were
vigorously shaken immediately before sampling to resuspend the sediment, to assist counting of organisms both in the water and the sediment. Total cell counts were determined by microscopy using membrane filtration and staining with 40 ,6-diamidino-2-phenylindole (DAPI) [27]. 2.5. Bacterial isolate characterisation In conjunction with the microbial count tests on the microcosms, 75 colonies were randomly selected from three replicate R2A plates (25 per plate) for each of the microcosm (3 aquifer types, 3 replicates, with and without the presence of isoproturon (850 in total). Plates with between 20 and 200 colonies after 10 days incubation and used to determine cfu, were selected for the isolation. Collected isolates were sub-cultured on TSBA to ensure purity. Isolates from the second sample (146 days for chalk and 132 days for sandstone) were characterised by analyses of the fatty acid methyl ester content (FAME) using gas chromatography [28]. In brief, cells were harvested after 24 h incubation on TSBA at 28 C and whole cell fatty acids were saponified, methylated and extracted. FAME analysis was performed using a Hewlett–Packard HP6890 series gas chromatograph (Hewlett–Packard, Berkshire, UK) and using Microbial Identification System (MIS) software (Microbial ID, Newark, DE). All isolates were identified using ‘‘Aerobe Library’’ version 4 (1999). In total 719 isolates were characterized by FAME, since a number of isolates were lost prior to FAME due to poor growth on TSBA. 2.6. Assessment of bacterial community structure The diversity and structure of communities identified by FAME analysis was analysed further using three complementary indices. To do this, bacterial identifications from FAME analysis were used to the species level [28]. The indices used were: (1) Species richness ðS Þ – this is simply the number of species identified in a sample. (2) Simpson’s index (L0 ) – formally L0 , is the probability that two isolates taken at random from a sample, will be the same species. Of the three measures this is the most sensitive to changes in the frequency of the more abundant species. (3) Shannon–Wiener index (H 0 ) – this is a combined figure that reflects the extent of diversity and the evenness of isolate distribution between taxa. It is sensitive to changes in the frequency of common and less common (though not the rarer) species. These indices were calculated by: s X ni ðni 1Þ N ðN 1Þ i¼1 s X pi2 ;
Simpson’s index of concentration ðL0 Þ ¼
i¼1
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
Shanon–Wiener index ðH 0 Þ ¼
s X ðpi Þðloge pi Þ i¼1
(ni is the number of isolates in the ith species, s is the numbers of species in sample, pi is the proportion of species i in the sample, N is the number of individuals in the sample). The indices (species richness, Simpson’s index and Shannon–Wiener index) were adjusted using a resampling method [29] for a standard sample size of n ¼ 20. 2.7. Repeated enrichments in groundwater spiked with isoproturon An enrichment experiment with 100 lg l1 isoproturon and sterile groundwater was set up to examine whether this level of pesticide could exert a selection pressure on the microbial community, and hence stimulate more rapid degradation. Established microcosms used to study isoproturon degradation and between 200 and 280 days of age were used as the inoculum. For the chalk groundwater, 120-ml screw top containers were filled with 60 ml Western Court groundwater (stored at 4 C) and 12 g chalk and then autoclaved for 30 min. Isoproturon was then added to the autoclaved water to give a final concentration of 100 lg l1 as described above, together with 10 ml from each of the replicates of the established microcosm degradation experiment (Bridgets). For the sandstone samples (GT2 and Clumber Park), this was repeated using GT2 groundwater, and for the Limestone samples Welbourn groundwater for Welbourn and Coleby were used (all groundwater previously stored at 4 C). When transferred into the fresh sterile groundwater the chalk cultures were 170-day old, the sandstone cultures 153-day old, and limestone cultures 303-day old, respectively. Incubation and sampling followed the protocol described above. After 316 days of this new incubation, following vigorous mixing of the microcosms, 10 ml from each container was transferred to fresh sterile groundwater and sterile solid matrix as a second subculture. Sub-samples from the second sub-culture were transferred again to a third sub-culture after 240 days incubation. Degradation was monitored by sampling every 50 days for changes in isoproturon concentration and looking for the formation of metabolites. 2.8. Influence of groundwater type on isoproturon degradation To test the role of the groundwater chemistry in stimulating or retarding the potential of the indigenous microorganisms to degrade isoproturon, a filter switch technique was used. On the basis of previous contrasting performance the sandstone and limestone groundwaters
75
were compared. The sandstone samples were represented by GT2 and Clumber Park, which had demonstrated previous rapid isoproturon degradation (Table 5). The limestone samples were represented by Welbourn and Coleby. The inoculum and groundwater used came from the second sub-culture experiment, after 240 days incubation with isoproturon. Each of these microcosms were divided into two 20 ml aliquots. To separate the groundwater and indigenous microorganisms from one another, the samples were filtered through a sterile 0.2 lm PVDF filter (47 mm Durapore membrane filter, Millipore, Bedford, USA) held in a sterilized (autoclaved) magnetic 250 ml filter holder (Gelman, Ann Arbor, USA) with a Buchner flask. From previous tests with 14 C-labelled isoproturon, PVDF filter discs had been selected as the best compromise between practical filtration (not too hydrophobic) and not too sorptive of isoproturon (unlike cellulose-based filters). Thus, for example a 20 ml sub-sample of one of the GT2 microcosms was 0.2 lm filtered. The moist filter paper with its retained bacteria was immediately placed aseptically into a 100 ml Sterilin pot with 7 g of sterile sandstone aquifer matrix (previously drilled from the GT borehole as described in 2.3) together with 40 ml of filter-sterilised GT2 groundwater, to act as a positive control (Fig. 1). The other 20 ml sub-sample of GT2 was similarly filtered, but in this case the filter paper containing the sandstone groundwater bacteria was transferred to a 120 ml Sterilin pot containing 7 g sterile limestone aquifer matrix, and 40 ml filter-sterilised limestone groundwater from Welbourn. Thus, the sandstone bacteria were bathed in limestone groundwater in the presence of limestone aquifer material (from Longwood as described previously). The microcosms were then spiked as before to give a 100 lg l1 final concentration. Similarly, the three replicates of the sandstone sample from Clumber Park second sub-culture were divided and half incubated with limestone Welbourn groundwater and half placed back with its native groundwater. For the limestone bacteria of Welbourn and Coleby, these were divided and incubated either with their own groundwater or that of the sandstone GT2. All the microcosms were then incubated at 20 C for 300 days and sampled approximately every 50 days to determine isoproturon degradation. To check the efficiency of the 0.2 lm PVDF filtration technique, 10 ml from GT2 and Clumber Park (sandstone groundwater), and Welbourn (limestone groundwater) were passed through individual filter discs into sterile Buchner flasks. One millitres was taken from the filtrate and spread on 3% TSBA plates and incubated at 20 C for 3 days. Growth on the plates was compared with 1 ml spread on TSBA plates taken from the original groundwater samples. In addition the effectiveness of the filtration was further examined by microscopy of DAPIstained samples.
76
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
bacteria separated from native groundwater on filter paper Returned to native filtered groundwater
Incubated with foreign filtered groundwater Incubated with foreign bacteria (on filter paper) 0.2 µm filtration of groundwater sample
filtered groundwater Returned to native bacteria (on filter paper)
Fig. 1. Schematic representation of experimental method associated with examining the influence of groundwater type on isoproturon degradation.
2.9. Analysis of isoproturon and metabolites After filtration (0.2 lm) samples were stored in PTFE capped glass vials at 4 C prior to analysis. The samples were stored no longer than one week prior to analysis. All analytical standards were obtained from Sigma–Aldrich (Poole, UK). Isoproturon, isopropyl aniline, mono and didesmethyl-isoproturon were determined by high performance liquid chromatography (HPLC) with UV detection. A 150-ll sample was separated on a C18 column (Columbus from Phenomenex Ltd, UK; 250 21 mm, 5 lm) using an isocratic 35:65 acetonitrile:water mobile phase at a flow rate of 200 ll/min. Detection was made at 240 and 220 nm, which also enabled the determination of peak identity and purity. Stock solutions were prepared in methanol with calibration standards and samples prepared in 50:50 methanol:water. The limit of detection and quantitation for these determinants was 1 and 5 lg l1 , respectively. The appearance of any of the metabolites mentioned above was taken as confirmatory evidence of biodegradation having taken place [6,30,31].
3. Results 3.1. Groundwater sample characteristics The chemical and microbiological characteristics of the groundwater samples used are shown in Table 2. All the groundwaters had dissolved oxygen at levels close to saturation, given a temperature of 9–13 C, whilst pH values were neutral to weakly alkaline (7.1–7.7) and DOC values between 0.7 and 2.8 mg l1 . The lowest numbers of culturable bacteria were associated with the limestone groundwater samples and the highest with the sandstone. The water sampling method for all sites was
essentially the same, with borehole purging prior to sample collection. Clumber Park and Welbeck were different from the other sites in that these boreholes were equipped with their own pumps and the samples were taken from sample taps. The total amount of DOC in the groundwater and numbers of culturable organisms were not directly related to one another. Unfortunately when it came to measure the parameters at Welbourn the water level had fallen below the borehole depth, so further samples could not be taken. 3.2. The response of the indigenous microbial community to 100 lg l1 isoproturon Analysis of the microbial communities associated with the different treatments showed that the introduction of isoproturon had no consistent impact on microbial total or cultured counts either at the beginning or towards the end of the incubations. Total microbial counts decreased significantly (P < 0:05) in microcosms with chalk and sandstone from a mean 1.0 108 – 3.0 107 ml1 after 117 d. incubation, in both spiked and un-spiked samples. The addition of 0.6% methanol alone (used as an isoproturon carrier solvent) had no detectable effect on the microbial communities (data not shown). All the groundwater samples collected from the chalk and sandstone sites degraded isoproturon and caused the formation of monodesmethyl-isoproturon (Figs. 2 and 3) that was not detected in sterile controls. The majority of the parent isoproturon was not converted to monodesmethyl-isoproturon, as has been previously observed in groundwater [6,7]. In fact with soil biodegradation studies the major metabolites are considered to be hydroxylated forms [30,31] which are not detectable with this eluent. To help compare the degradation rates for all the groundwater samples (Table 3), the data were
Isoproturon (µg/L)
0.7 (0.3) 0.7 (0.4)
13 (0.1) 3.3 (0.2) 27 (7.8)
3.8 (0.09) 2.6 (0.5) 16 (1)
100 80 60 40 20 0
328 273
50
100
150
200
250
300
Time (d)
Fig. 2. Degradation of isoproturon in chalk groundwater collected from Western Court (N), Western Court sterile treatment (M), Site WON (d) and Site WON sterile treatment (s) (mean of three observations with standard deviations). Data are given for the metabolite monodesmethyl-isoproturon (––) using the same site symbols (open) to represent the different treatments. Arrows indicate when microcosms sampled for viable counts, total counts and FAME characterization.
Isoproturon (µg/L)
NA, information witheld; ND, carbonate prevented accurate measurement here and NC, data not collected.
120
*
1.1 2.8 NC 7.3 0153 1260 0079 6143 Limestone Welbourn Coleby
15–19.5 mbs 7.5 mbs
Arable Arable
NC 9.8
NC 147
NC 951
176 114 187 0.7 1.2 2.3 502 517 1218 7.7 7.7 7.5 4594 3722 4622 3739 4604 3706 Sandstone Welbeck Clumber Park Gleadthorpe (GT 2)
5.5–7 mbs 8–12.0 mbs 8–10.0 mbs
Park Park Exptal
7.0 8.8 9.4
415 418 444
ND ND 1.2 7.2 7.1 7.2 4522 1332 4518 1322 NA Chalk Bridgets farm no. 2 Western Court Site WON
24–26 mbs 6.8–9 mbs 4.5–6 mbs
Pasture Arable Arable
7.6 12.7 9.1
309 364 ND
618 590 716
301 298 281
Viable count (R2A) and standard deviation from three replicates (cfu ml1 103 ) DOC (mg l1 ) Eh (mV)
120
0
SEC (lS/cm)
HCO3 (mg l1 )
140
pH
DO (mg l1 )
77
160
Location Watertable (1999– 2001) Grid ref. Ordinance Survey 1:50,000 Borehole
Table 2 Chemical and microbial characterisation of groundwater samples investigated in this study (DO, dissolved oxygen; Eh, redox potential; SEC, electrical conductivity and DOC, dissolved organic carbon)
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
100 80 60 40 20 0 0
30
60
90
120
150
180
210
240
270
Time (d)
Fig. 3. Degradation of isoproturon in sandstone groundwater sampled from GT2 (N), Welbeck (d), Clumber Park (r) with sterile treatments (dashed lines, open symbols) (mean of three observations with standard deviations). Data are given for the metabolite monodesmethylisoproturon (––) using the same site symbols (open) to represent the different treatments. Arrows indicate when microcosms sampled for viable counts, total counts and FAME characterization.
entered into the Model Manager program (Cherwell Scientific, UK) to calculate the half-lives and give confidence levels for isoproturon in a simple first order model for the parent molecule. Whilst first order decay may not be entirely appropriate for these situations, it does provide a useful means of comparison: Mp ðtÞ ¼ M0 expðkp tÞ; where Mp is the concentration of isoproturon, M0 is the initial concentration of isoproturon and kp is the first order rate constant of the parent. As observed previously with studies on GT2 groundwater [7], its degradation of isoproturon was very consistent with good replication occurring (Table 3) and a half-life of around 100 d. Welbeck and Clumber Park showed greater variance between replicates with aggregate mean half-lives of 124 and 166 d. It should be noted that both these groundwater samples came from
78
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
Table 3 Isoproturon half-lives calculated as simple first order reactions for each individual replicate sample in microcosms used to determine community impact 95% confidence range
R2
Sandstone 24.11.99 GT2 SG1 106 SG2 96 SG3 97
69–143 89–103 88–106
0.95 0.99 0.99
Welbeck SW1 SW2 SW3
181 91 102
72–287 64–118 67–138
0.87 0.97 0.96
Clumber Park SC1 SC2 SC3
123 210 167
35–212 127–292 57–277
0.83 0.93 0.84
Sample and code
Half-life (days)
Chalk 10.11.99 WON4 CWO1 211 CWO2 239 CWO3 223 Western Court CWC1 CWC2 CWC3
206 89 68
160–262 199–277 170–276 149–261 25–154 11–125
Bre Bac 1% 1%
1%
SW–
Pse 89%
Pse 93% Ace 6%
Agr 7%
SC–
SC+ Bac 12%
Pse 63% Sph NM 4% 1%
Aci 1% Pse 24%
Agr 1%
Fla 1%
Chr 1%
CWC+
Pse 43%
Pse 49%
Sta 1%
Bre 36%
Phe 1% Och 1%
En 1%
Cel Com 4% 1%
Sph 32%
NM 19%
CWC–
Str 1%
Azo Bac 3% 3% Noc 1%
Chy 1% Com 1%
Azo 16%
NM 7%
boreholes within parks which were a few km distant from agricultural farms. For the chalk, site WON had good replication with an aggregate half-life of 224 days (Table 3), but Western Court had some replicates showing rapid degradation of isoproturon (CWC2 and CWC3) and another slow (CWC1). Assessment by culture based methods detected no consistent impact of the presence of 100 lg l1 isoproturon on the taxa composition of microbial communities isolated from chalk or sandstone groundwaters, after 146 and 132 days exposure, respectively. However, cluster analysis of the data grouped together all the isoproturon treated samples at 2 Euclids and likewise untreated at 2 Euclids, but separated the treatments at 4.8 Euclids (data not shown), reflecting the distinctiveness of exposed communities. In most samples pseudomonad species, the most abundant group (Fig. 4), similarly showed no consistent difference between treatments in terms of the presence of specific species. However, when the data were analysed mathematically (Table 4) in terms of species richness (number of species) and community structure (diversity indices) differences were detected, irrespective of groundwater type. Overall, the presence of isoproturon increased species richness and diversity with SC (sandstone, Clumber Park) groundwater showing the most marked effect (H 0 increased from 1.75 to 2.07) followed by CWC (chalk, Western Court) and a more muted effect in SW (sandstone, Welbeck) groundwater (Table 4). In groundwater microcosms the isoproturon degradation
Cur Hyd Chr 1% 1% Stv 1% 1% NM 1%
Sta 1%
SW+
0.97 0.99 0.97 0.97 0.88 0.87
Met Pae Sph 1% 3% 1%
Aci 13% Azo 1% Bre 3% Cel 1% Chr 18% Esc 1% Jan 6% Pho NM Yer 3% 1% 1%
Fig. 4. Isolation frequencies of the culturable bacterial community detected in groundwaters in sandstone, Welbourn (SW) and Coleby (SC) microcosms sampled after 140 days and chalk Western Court (CWC) sampled after 130 days, with (+) or without ()) isoproturon present (mean of all three replicates). Ace, Acetobacter; Aci, Acidovorax; Agr, Agrobacterium; Azo, Azospirllium; Bac, Bacillus; Bre, Brevundimonas; Cel, Cellulomonas; Chr, Chromobacterium; Chy, Chryseomonas; Com, Comamonas; Cur, Curtobacterium; En, Enterococcus; Esc, Escherichia; Fla, Flavobacterium; Hyd, Hydrogenophaga; Jan, Janthinobacterium; Met, Methylobacterium; Noc, Nocardia; Och, Ochrobatrum; Pae, Paenibacillus; Phe, Phenylobacterium; Pho, Photobacterium; Pse, Pseudomonas; Sph, Sphingomonas; Sta, Staphylococcus; Str, Streptococcus; Stv, Streptoverticillium; Yer, Yersinia; and NM, No Match.
rates correlated (R2 ¼ 0:56, P ¼ 0:009) with increased proportions of the three most numerous genera (Fig. 4). That is the presence of isoproturon increased the relative abundance of a few species that presumably were associated with increased rates of degradation. However, no consistent pattern could be detected in terms of the presence of key taxa and high rates of degradation. Thus, for example, the community profile associated with high degradation in the sandstone Welbourn site was different from the sandstone Coleby site (Fig. 5).
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
79
Table 4 Diversity measures – species richness (S ), Simpson’s index (L0 ), and Shannon–Wiener index (H 0 ) – estimated for sandstone groundwater Welbourn (SW), and Coleby (SC) with chalk from Western Court (CWC) with (+) and without ()) isoproturon Name
SW+ SW) SC+ SC) CWC+ CWC)
Ranking the estimated values of S , L and H 0a
Total number of isolates
Estimated values S
L0
H0
S
L0
H0
73 73 67 61 65 66
7 7 9 8 8 7
0.163 0.179 0.100 0.184 0.174 0.211
1.77 1.65 2.07 1.75 1.79 1.65
1–3 1–3 6 4–5 4–5 1–3
5 3 6 2 4 1
4 2 6 3 5 1
These values were standardised to a uniform sample size of 20 isolates by taking the median values from 1000 re-samplings with all replicates combined. The S and H 0 values rise with increasing diversity, while the L0 values fall with increasing diversity. Also given are the rankings of these values. a Ranking of the estimated values of S , L0 and H 0 is from least diverse (1) to most diverse (6).
with fresh 100 lg l1 isoproturon (Table 5), a variety of responses were seen. In some cases, degradation rates did not change, in others a transient improvement reverted to the original performance by the third subculture. In only one case, Bridgets, did the repeated subculturing lead to a putative increase in degradation rates. In this case, an original half-life began at 312 days, then dropped to 269 days on first sub-culture, 104 days on second sub-culture and finally 130 days on the final sub-culture (Table 5).
3.3. Impact of enrichment on isoproturon degradation rates To examine whether a more vigorous isoproturon degrading population would develop with repeated additions of the pesticide, an enrichment experiment was established. By comparing the degradation rates for the first microcosms and that after repeated sub-culturing
300 Half-Life (d)
3.4. The influence of groundwater type on isoproturon degradation
R2 = 0.5507
250 200
In a test of the 0.2 lm filtration process, no colony growth whatsoever could be found when 1 ml of the filtrate was plated out on TSBA plates. However, tests with direct counting indicating some cell-like could pass through the filter. Using the filter technique the greatest change was detected when the normally slow degradation of the limestone bacteria, from Welbourn and Coleby (predicted half-lives 6.3 and 4.7 years, respectively), were incubated with groundwater (filter-sterilised) from the fast-degrading sandstone site GT2 (Fig. 6). In the
150 100 50 0 30
40
50
60
70
80
90
100
% Dominant Genera Fig. 5. Plot of the three most numerous genera against isoproturon degradation (half-life).
Table 5 Isoproturon half-lives (DT50) and standard deviation in parentheses calculated as simple first order reactions for microcosms established to determine the influence of repeated enrichment with 100 lg/L and sterile groundwater at 20 C Material
Sandstone aquifer
Site
GT2
Original 1st sub-culture (316 days) 2nd sub-culture (240 days) 3rd sub-culture (377 days)
Limestone aquifer Clumber Park
Welbourn DT50 (days)
R
DT50 (days)
R
DT50 (days)
R2
0.98 0.99
240 (54) 112 (33)
0.75 0.96
759 (38) 765 (181)
0.89 0.73
247 (51) 247 (172)
0.95 0.87
312 (30) 269 (44)
0.99 0.94
95 (25)
0.98
135 (77)
0.97
11,550 (350)
0.4
385 (86)
0.73
104 (10)
0.96
73 (48)
0.97
266 (12)
0.92
577 (142)
0.77
552 (155)
0.70
130 (28)
0.97
71 (12) 78 (11)
The duration of the incubation period is given in the first column.
2
Bridgets
R
R
2
Coleby
DT50 (days)
DT50 (days)
2
Chalk aquifer
2
80
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
% Isoproturon remaining
120 100 80 60 40 20 0 0
50
100
150
200
250
300
Time (d)
Fig. 6. Degradation of 100 lg l1 isoproturon by limestone Welbourn bacteria with its native groundwater (j), or filter sterilized sandstone GT groundwater (), similarly limestone Coleby bacteria incubated with its native groundwater (r), or filter sterilized sandstone GT groundwater (}) (mean of three observations with standard deviations). Data are given for the metabolite monodesmethyl-isoproturon (––) using the same site symbols (open) to represent the different treatments.
% Isoproturon remaining
120 100 80 60 40 20 0 0
50
100
150
200
250
300
Time (d)
Fig. 7. Degradation of 100 lg l1 isoproturon by sandstone GT bacteria with its native groundwater (N), or filter sterilized limestone Welbourn groundwater (M), similarly sandstone Clumber park bacteria incubated with its native groundwater (d), or filter sterilized limestone Welbourn groundwater (s) (mean of three observations with standard deviations). Data are given for the metabolite monodesmethyl-isoproturon (––) using the same site symbols (open) to represent the different treatments.
presence of the filter-sterilised sandstone groundwater the isoproturon degradation half-lives decreased to 161 days for Welbourn, and 187 days for Coleby, respectively. In contrast, when the fast degrading sandstone bacteria from GT2 and Clumber Park were incubated with filter-sterilised limestone groundwater from Welbourn, degradation was retarded only slightly (Fig. 7).
4. Discussion In previous studies of the fate of isoproturon in groundwaters we demonstrated that its rate of degradation varied considerably, both spatially and temporally [6,7]. An aim of this study was to determine the microbial basis of this observed variation and in par-
ticularly the possible correlation with microbial community composition, abundance and groundwater chemistry. The results of this study demonstrated that the presence of 100 lg l1 isoproturon increased the diversity of the bacterial community. However, the extent to which isoproturon increased diversity at 100 lg l1 , which is a much higher concentration than the trace levels (<1 lg l1 ) detected in groundwaters in the UK, was small. The relatively weak selective effect of 100 lg l1 of isoproturon suggests an even smaller likelihood that trace levels detected in groundwater in the field would select for specialist populations. The minor selective effect exerted by isoproturon was reflected in the response of the individual communities examined in this study, which were inconsistent in their response, revealing no trends in terms of preferential response of specific populations. In contrast to this general picture, repetitive subcultures of microbial communities from sandstone (Clumber Park) and chalk (Bridgets) led to decreased isoproturon half-lives, which in the latter case was maintained. The accelerated rates of degradation on repeated sub-culturing, observed in Bridget samples was the only exception. There is some evidence that accelerated degradation can occur in soil [32,33]. Some adaptation with another herbicide has been observed in groundwater microcosm studies containing sediment and groundwater exposed for 40 days to 30 lg l1 phenoxy acid, the herbicide degraded more rapidly than in sediments not previously exposed [34]. This would suggest that some pesticides represent more valuable nutrient and energy sources than isoproturon, so preferentially selecting specific populations. Also it is probable that isoproturon concentrations of 0.1 lg l1 , which more typically occur in the environment, are too low to have a significant lasting impact on exposed microbial communities, indeed in a previous study on chalk, the ability of the groundwater samples to degrade isoproturon was not correlated to previous field exposure to isoproturon in their boreholes [6,23]. The moderate and inconsistent effects of isoproturon, demonstrated in this study, suggest that the variable nature of isoproturon degradation in groundwaters are unlikely to be due solely to microbial processes, but may also be influenced by physico-chemical factors in the groundwater. The presence, or absence of key substrates might be playing a role, for instance, it has recently been demonstrated that the production of cometabolites by a bacterial strain in a consortia stimulated isoproturon breakdown by another member of the group [19]. The results of this and previous studies suggest that these controlling factors may be more complex than previously suspected and certainly more subtle. This study contributes in several ways to pin-pointing the key controlling factors. We have demonstrated that:
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
• The increased degradation rates, across the different groundwater microcosms, correlated with increasing dominance of the three most numerous genera. It is possible that these differences in community structure either correlated with, or generated, the classes of environmental change that were important contributors to degradation rates. Despite this no consistent patterns in the terms of the presence or dominance of specific taxa could be correlated with degradation rates. This could be due to the fact that the populations responsible for isoproturon degradation represented only a small proportion of the total community, as well as the limits of the detection methods used. • Isoproturon exposure increased the extent of community diversity, but had no detectable effect on microbial counts. • Isoproturon degradation was detected in microbial communities that were very different in composition suggesting that the genes for degradation are widely distributed throughout the community. • Only in one case could isoproturon degradation rates be stimulated by enrichment using 100 lg l1 isoproturon. Given the absence of correlation of isoproturon degrading potential related to previous exposure in groundwater noted in previous work on chalk [6], these data support the theory that fast degrading microbial communities in groundwater are unlikely to occur as a result of continued exposure to isoproturon in the field. • Isoproturon degradation rates by the indigenous microbial community could be improved or retarded according to the bathing groundwater, indicating a possible role of the groundwater chemistry in mediating degradation rates.
Acknowledgements This study was funded by the NERC CEH Integrating fund. The authors thank Sarah Harman, Andy Dixon and Janice Trafford for technical support and Daren Gooddy for helpful information.
References [1] Downing R.A. (1998) Groundwater, Our Hidden Asset. Earthwise Publications, British Geological Survey. [2] Anon. (1995) Pesticides in the Aquatic Environment, Water Quality Series no. 26, Report of the National Rivers Authority. HMSO, London. [3] Anon. (1999) Pesticides in the Aquatic Environment 1997. Environment Agency, Howbery Park, Wallingford, UK. [4] Environment Agency (2001) Pesticides 2000: A summary of monitoring of the aquatic environment in England and Wales. Environment Agency. www.environment-agency.gov.uk.
81
[5] Johnson, A.C., Hughes, C.D., Williams, R.J. and Chilton, P.J. (1998) Potential for aerobic isoproturon biodegradation and sorption in the unsaturated and saturated zones of a chalk aquifer. J. Contam. Hydrol. 30, 281–297. [6] Johnson, A.C., White, C. and Bhardwaj, C.L. (2000) Potential for isoproturon, atrazine and mecoprop to be degraded within a chalk aquifer. J. Contam. Hydrol. 44, 1–18. [7] Johnson, A.C., White, C., Bhardwaj, C.L. and Dixon, A. (2003) The ability of indigenous microorganisms to degrade isoproturon, atrazine and mecoprop within aerobic UK aquifer systems. Pest. Man. Sci. 59, 1291–1302. [8] Issa, S. and Wood, M. (1999) Degradation of atrazine and isoproturon in the unsaturated zone: a study from Southern England. Pest. Sci. 55, 539–545. [9] Pahm, M.A. and Alexander, M. (1993) Selecting inocula for the biodegradation of organic compounds at low concentrations. Microbial Ecol. 25, 275–286. [10] Tor€ang, L., Albrechtsen H.-J. and Nyholm, N. (2000) Biodegradation kinetics at low concentrations (<1 lg l1 ) for aquifer pesticide contaminants. In: Proceedings of the International Conference on Groundwater Research, Copenhagen, Denmark, 6–8 June 2000 (Bjerg, P.L., Engesgaard, P. and Krom T.D., Eds.). AA Balkema, Netherlands. [11] Larsen, L., Sørensen, S.R. and Aamand, J. (2000) Mecoprop, isoproturon and atrazine in and above a sandy aquifer: Vertical distribution of mineralisation potential. Environ. Sci. Technol. 34, 2426–2430. [12] Albrechtsen, H.-J., Mills, M.S., Aamand, J. and Bjerg, P.L. (2001) Degradation of herbicides in shallow Danish aquifers: an integrated laboratory and field study. Pest. Man. Sci. 57, 341–350. [13] Kristensen, G.B., Sørensen, R.S. and Aamand, J. (2001) Mineralization of 2,4-D, mecoprop, isoproturon and terbuthylazine in a chalk aquifer. Pest. Manag. Sci. 57, 531–536. [14] Tuxen, N., Tuchsen, P.L., Rugge, K., Albrechtsen, H.-J. and Bjerg, P.L. (2000) Fate of seven pesticides in an aerobic aquifer studied in column experiments. Chemosphere 41, 1485–1494. [15] Walker, A., Jurado-Esposito, M., Bending, G.D. and Smith, V.J.R. (2001) Spatial variability in the degradation rate of isoproturon in soil. Environ. Pollut. 111, 407–415. [16] Yen, K.-M. and Gunsalus, I.C. (1982) Plasmid gene organization: naphthalene/salicylate oxidation. Proc. Natl. Acad. Sci. USA 79, 874–878. [17] Chen, S.-H. and Aitken, M.D. (1999) Salicylate stimulates the degradation of high-molecular weight polycyclic aromatic hydrocarbons by Pseudomonas saccharophilia P15. Environ. Sci. Technol. 33, 435–439. [18] De Lipthay, J.R., Johnsen, K., Albrechtsen, H.-J., Rosenberg, P. and Aamand J. (2004) Bacterial diversity and community structure of a subsurface aquifer exposed to realistic low herbicide concentrations. FEMS Microbiol. Ecol. doi:10.1016/j.femsec. 2004.02.007. [19] Sørensen, S.R., Ronen, Z. and Aamand, J. (2002) Growth in coculture stimulates metabolism of the phenylurea herbicide isoproturon by Sphingomonas sp. Strain SRS2. Appl. Environ. Microbiol. 68, 3478–3485. [20] Roberts, S.J., Walker, A., Rarekh, N.R., Welch, S.J. and Waddington, M.J. (1993) Studies on a mixed bacterial culture from soil which degardes the herbicide linuron. Pest. Sci. 39, 71–78. [21] El Fantrossi, S., Verschuere, L., Verstraete, W. and Top, E.M. (1999) Effect of phenylurea herbicides on soil microbial communities estimated by analysis of 16S rRNA gene fingerprints and community level physiological profiles. Appl. Environ. Microbiol. 65, 982–988. [22] Kamagata, Y., Fulthorpe, R.R., Tamura, K., Takami, H. and Tiedje, J.M. (1997) Pristine environments harbour a new group of oligotrophic 2,4-dichlorophenoxyacetic acid-degrading bacteria. Appl. Environ. Microbiol. 63, 2266–2272.
82
A. Johnson et al. / FEMS Microbiology Ecology 49 (2004) 71–82
[23] Johnson, A.C., Besien, T.J., Bhardwaj, C.L., Dixon, A., Gooddy, D.C., Haria, A.H. and White, C. (2001) Penetration of herbicides to groundwater in an unconfined chalk aquifer following normal soil applications. J. Contam. Hydrol. 53, 101–117. [24] Edmunds, W.M., Cook, J.M., Kinniburgh, D.G., Miles, D.L. and Trafford J.M. (1989) Trace element occurrence in British groundwaters. British Geological Survey Report SD/89/3. [25] Albrechtsen, H.-J., Smith, P.M., Nielsen, P. and Christensen, T.H. (1997) Importance of sediment fines in laboratory studies on the degradation of organic chemicals in aquifers. Water Res. 31, 2287–2299. [26] Orcutt, F.S. (1940) Methods for more accurately comparing plate counts for soil bacteria. J. Bacteriol. 39, 100. [27] Wagner, M., Amann, R., Lemmer, H. and Schleifer, K.-H. (1993) Probing activated sludge with oligonucleotides specific for proteobacteria: inadequacy of culture-dependent methods for describing microbial community structure. Appl. Environ. Microbiol. 59, 1520–1525. [28] Thompson, I.P., Bailey, M.J., Ellis, R.J. and Purdy, K.J. (1993) Subgrouping of bacterial populations by cellular fatty acid composition. FEMS Microbiol. Ecol. 102, 75–84. [29] Edwards, M.L., Lilley, A.K., Timms-Wilson, T.H., Thompson, I.P. and Cooper, I. (2001) Characterisation of the culturable heterotrophic bacterial community in a small eutrophic lake (Priest Pot). FEMS Microbiol. Ecol. 35, 295–304.
[30] Mudd, P.J., Hance, R.J. and Wright, S.J.L. (1983) The persistence and metabolism of isoproturon in soil. Weed Res. 23, 239–246. [31] Lehr, S., Gl’’agen, W.E., Sanderman, H., Beese, J.F. and Scheunert, I. (1996) Metabolism of isoproturon in soils originating from different agricultural management systems and in cultures of isolated soil bacteria. Int. J. Anal. Chem. 65, 231–243. [32] Bending, G.D., Shaw, E. and Walker, A. (2001) Spatial heterogeneity in the metabolism and dynamics of isoproturon degrading microbial communities in soil. Biol. Fertil. Soils 33, 484–489. [33] Sørensen, S.R. and Aamand, J. (2003) Rapid mineralisation of the herbicide isoproturon in soil from a previously treated Danish agricultural field. Pest Manag. Sci. 59, 1118–1124. [34] Tuxen, N., De Lipthay, J.R., Albrechtsen, H.-J., Aamand, J. and Bjerg, P.L. (2002) Effect of exposure history on microbial herbicide degradation in an aerobic aquifer affected by a point source. Environ. Sci. Technol. 36, 2205–2212. [35] Hodge, C.A.H., Burton, R.G.O., Corbett, W.M., Evans R. and Seale, R.S. (1984) Soils and their use in Eastern England. Soil Survey of England and Wales, Bulletin No. 13, Harpenden. [36] Jarvis, M.G., Allen, R.H., Fordham, S.J., Hazelden, J., Moffat, A.J. and Sturdy, R.G. (1984) Soils and their use in South East England. Soil Survey of England and Wales, Bulletin No. 15, (1984) Harpenden.
