Abstract

Introduction

Water contaminants, such as excessive nutrients, sediments and organic matter in streams and their receiving waters are a worldwide problem (Cooke et al. 1993; European Environment Agency 1998; USEPA 1996). Increased nutrients and organic matter (eutrophication) can lead to algal blooms (CSIRO Australia 1996) that reduce the amenity of water. Identifying the source of pollutants is an important step in the development of effective remedial action.

Organic marker compounds (biomarkers), such as lipids, have useful attributes for tracing water pollutants to their source in agricultural environments (Nash and Halliwell 2000). Lipids are a heterogeneous class of organic compounds that contain variable length, branched, hydrocarbon chains, including cyclic hydrocarbons, and a range of functional groups, and as a result are often species specific (Jones, Nichols et al. 1994).

Coprostanol is an example of a sterol used to study faecal pollution (Leeming and Nichols 1996; Nichols and Leeming 1991; Nichols, Leeming et al. 1996). It is formed by the hydrogenation of the double bond between C₅ and C₆ in the second hexane ring of cholesterol. While only trace amounts of cholesterol are found in plant tissues, it is an important membrane component of animal cells (Christie 1989). Consequently, cholesterol consumed by meat eating animals is converted to coprostanol on passage through the gut and is found in the faeces (Martin et al. 1973; Rosenfeld and Gallagher 1971). Similarly 24-ethylcoprostanol and 24-ethylepicoprostanol can be used to distinguish between faeces from herbivores and carnivores (Leeming, Ball et al. 1994). In both cases the sterol profiles of the faeces reflect the diet of the source animal and conversions in the digestive tract (Leeming et al. 1995).

This study investigates the concentrations of neutral lipids of the sterol and other alcohol groups in overland flow exported from pasture-based grazing systems and source materials. Lipids were extracted directly from source materials, the particulate (>0.45 μm) and filtrate fractions (<0.45 μm) of the overland flow, and from the particulate and filtrate fractions of water extracts of source materials.

Materials and Methods

Sample collection and storage

Details of the source materials and overland flow used in these studies are presented in Table 1. Source materials were collected from farms in the Gippsland region of southern Australia.

All equipment (i.e. spatulas, scissors, soil corer, glassware, bottles, sieves) was thoroughly cleaned prior to use. All source materials were collected from more than one location or animal within a farm in order to minimise bias. Homogenised bulk samples were coned and divided into three unequal sub-samples. One of the sub-samples was frozen prior to its direct analysis while another was extracted by gently oscillating (Ratek, Australia, 32 mm orbit diameter, 30 cycles/min.) ca. 200 g of the source material for 6 h in 2 L of deionised water to simulate overland flow (Nash et al. 2003). After oscillation the sample was allowed to settle for 1 min. and the supernatant was decanted through glass fibre (GF/F) filters (particulate fraction). Filtrates were collected and passed through conditioned SPE disks (filtrate fraction). The GF/F filters and solid phase extraction (SPE) disks were wrapped individually in aluminium foil and frozen prior to analysis.

Overland flow samples were collected during October and December 2000, and in January 2001 for comparison with the source materials. Water samples were manually collected from established monitoring sites at Arawata (38°29’S, 140°57’E), Darnum (38°10’S, 146°03’E), Trafalgar (38°09’S, 146°08’E) and the Macalister Research Farm (MRF) near Maffra in the Macalister Irrigation District (38°00’S, 146°54’E). Overland flow samples were taken at the start of the storm/irrigation event (start), and during (Mid or Composite) or end of the event (End).

Table 1. Selected detail of sample types, locations and timing for the potential source materials

Sample	Details	Replicates	Date sampled
Cow faeces	Grass fed (fresh)	2	20 Nov. 2000
	Grass fed (air-dry)	1	22 Nov. 2000
	Grain and grass fed (fresh)	1	21 Nov. 2000
	Grain and grass fed (air-dry)	2	21 Nov. 2000
	Grass fed (fresh)	1	11 Jan. 2001
	Grain and grass fed (fresh)	1	11 Jan. 2001
Lolium perenne (Ryegrass)	1 leaf stage	2	20 Nov. 2000
	3 leaf stage	2	20 Nov. 2000
	1 leaf stage	1	9 Jan. 2001
	3 leaf stage	1	10 Jan. 2001
Trifolium repens (White clover)		1	21 Nov. 2000
		1	11 Jan. 2001
Rumex spp. (Dock)		1	22 Nov. 2000
Arctotheca calendula (Capeweed)		1	22 Nov. 2000
Plantago lanceolata (Plantain)		1	22 Nov. 2000
Hypochoeris radicata (Flatweed)		1	22 Nov. 2000
Decaying vegetation	Ellinbank	3	9 Nov. 2000
	MRF paddock A	1	10 Jan. 2001
	MRF paddock B	1	10 Jan. 2001
Surface soil classification	Dermosol	2	9 Nov. 2000
Fertiliser	Di-ammonium phosphate	1	8 Nov. 2000
	Single superphosphate	1	8 Nov. 2000

Analytical Procedures

The source materials and particulate fractions (GF/F) were quantitatively extracted using a modified single-phase dichloromethane/methanol/water method (Bligh and Dyer 1959). Gas chromatographic (GC) determinations and quantitation of selected compounds were performed on the solvent containing the O-TMSi ethers using a Hewlett Packard (HP) 5890 GC. Peak identification was based on comparison of retention time data with data obtained for laboratory standards. Peak areas were quantified using Waters Millenium (Waters Australia Pty Ltd, Australia) chromatography software. Individual compound concentrations less than 10 ng/g were considered trace amounts. The detection limits for individual sterols were ca. 1-2 ng/g. Verification of the identity of individual compounds by GC-MS analyses was performed using a HP5890 GC and a Thermoquest/Finnigan GCQ-Plus (Thermo-Finnigan, USA) mass spectrometer.

Statistics

The statistics employed were for the most part descriptive, to present the mean and range (or potential range) of the data. Bar graphs are used to present the means, with error bars for the range. A logarithmic scale is used to visually accommodate values that may differ by orders of magnitude, and the range, which typically was in proportion to the mean. A ratio of abundance between two compounds is represented by the distance between the means on the log scale.

Results and Discussion

The sterols and other alcohols investigated in this study are presented in elution order in Table 2 together with their common name and class. The chemical profiles of these compounds in some potential source materials and the particulate and filtrate fractions of their water extracts are presented in Figure 1. These profiles were compared to the concentrations in the particulate and filtrate fractions of overland flow samples collected from several farms (Figs. 2, 3). The concentration (μg/g dry weight or μg/L for overland flow) of individual compounds varied between sources and within sources between sampling locations.

Cow Faeces

In the majority of direct cow faeces extracts, the most abundant neutral lipids were phytol, hexacosanol, 24-ethylcoprostanol, 24-ethyl-5α-cholestanol, 24-ethylepicoprostanol, 24-ethylcholesterol, coprostanol and cholesterol (Fig. 1). One faeces sample contained a large concentration of octacosanol, causing the mean concentration of this lipid to be greater than that for cholesterol. The four 24-ethyl sterols/stanols have been used to differentiate sources of faecal pollution (Nichols, Leeming et al. 1996). The sterol profile is consistent with those previously published for cow and sheep faeces (Nichols, Leeming et al. 1996). A lipid that has yet to be identified, labelled m/z 291, was found in all samples.

Air-dry cow faeces generally contained higher concentrations of the neutral lipids than the fresh faeces although air-drying appeared to change the relative proportions of lipids in individual samples (data not shown). Interestingly, the ratios of related compounds were consistent across both air-dry and fresh faecal samples. For example, the ratio of cholesterol to coprostanol generally ranged between 0.7 and 1.1, coprostanol to epicoprostanol generally 6.1 to 7.1, 24-ethylcholesterol to 24-ethylcoprostanol 0.2 to 0.4 and 24-ethylcoprostanol to 24-ethylepicoprostanol 1.4 to 2.8.

The particulate fraction of the water extracts from cow faeces contained significantly lower concentrations of most neutral lipids than the direct extracts. The most abundant lipids were 24-ethyl-5α-cholestanol, phytol and hexacosanol. In contrast to the direct extracts of faeces, the concentrations of lipids present in the water extracted particulate fraction of fresh faeces were higher than the air-dry faeces. This probably indicates a smaller particle size in the fresh samples. Although the changes in concentrations were quite large between the direct extracts and the particulate fraction of the water extract, the ratios of the related compounds were similar, with the exception that the ratio of coprostanol to epicoprostanol changed from 6.1 to 7.1 to between 2.8 and 6.6. These data support the hypothesis that these compounds have useful properties for use in tracing cow faeces.

The lipids that were present in the filtrate fraction of the water extracts of faeces had generally similar proportions of individual lipids to the particulate fraction but lower concentrations overall. Phytol, which had been prevalent in the direct extracts and particulate fractions, was not detected at all in the filtrate fraction. Given that the lipids studied are surface active (i.e. would be expected to readily attach to soil colloid), it is not surprising that their concentrations in the filtrate decreased. Their presence in this fraction may be indicative of colloidal material. The trends observed in the ratios of related lipids present in the direct extracts and particulate phase of cow faeces were not observed in the filtrate fraction.

Table 2. Lipid names and peak identification numbers in elution order.

Peak number	Lipid	Common name	Class
1	2-hexadecen-1-ol	phytol	branched fatty alcohol
2	tetracosanol	C24 long chain alcohol	fatty alcohol
3	hexacosanol	C26 long chain alcohol	fatty alcohol
4	5β-cholestan-3β-ol	coprostanol	sterol
5	5β-cholestan-3α-ol	epicoprostanol	sterol
6	cholesta-5,22E-dien-3β-ol	22-dehydrocholesterol	sterol
7	cholest-5-en-3β-ol	cholesterol	sterol
8	5α-cholestan-3β-ol	5α-cholestanol	sterol
9	octacosanol	C28 long chain alcohol	fatty alcohol
10	24-methylcholesta-5,22E-dien-3β-ol	brassicasterol	sterol
11	24-methylcholesta-5,7,22-trien-3β-ol	ergosterol	sterol
12	24-ethyl-5β-cholestan-3β-ol	24-ethylcoprostanol	sterol
13	24α-methylcholest-5-en-3β-ol	campesterol	sterol
14	24-ethyl-5β-cholestan-3α-ol	24-ethylepicoprostanol	sterol
15	unknown	Arctotheca m/z 163	unknown
16	24-ethylcholesta-5,22E-dien-3β-ol	stigmasterol	sterol
17	24-methyl-5α-cholest-7-en-3β-ol	Δ7-ergostenol	sterol
18	3β-hydroxyolean-12-ene	β-amyrin	triterpene
19	24-ethylcholest-5-en-3β-ol	24-ethylcholesterol	sterol
20	24-ethyl-5α-cholestan-3β-ol	24-ethyl-5α-cholestanol	sterol
21	24-ethylcholesta-5,24(28)Z-dien-3β-ol	isofucosterol	sterol
22	3β-hydroxyurs-12-ene	α-amyrin	triterpene
23	triacontanol	C30 long chain alcohol	fatty alcohol
24	unknown	hopanol #1	hopanol
25	4,23,24-trimethyl-5α-cholest-22-en-3β-ol	dinosterol	sterol
26		fernol	hopanol
27	24-ethyl-5α-cholest-7-en-3β-ol	Δ7-stigmastenol	sterol
28	unknown	hopanol #2	hopanol
29	unknown	m/z 291	unknown
30	dotriacontanol	C32 long chain alcohol	fatty alcohol

Grasses and Weeds

The most abundant neutral lipids in most vegetation were phytol and 24-ethylcholesterol (Fig. 1). This is not surprising as 24-ethylcholesterol is one of the main plant sterols and phytol is present in plants in similar quantities to chlorophyll (Fruton et al. 1963). The abundance of phytol in vegetation is probably responsible for the high concentrations of phytol compared to other lipids in cow faeces.

As with the cow faeces, there was a general decrease in lipid concentrations in the water extractable particulate fractions of ryegrass compared to the direct extracts. However, there was also a change in the relative concentrations of individual lipids. The ratio of 24-ethylcholesterol to 24-ethyl-5α-cholestanol in the particulate fraction of the water extract typically doubled as ryegrass progressed from the 1 leaf to the 3 leaf stage, emphasising the effect of stage of growth and/or grazing. For all vegetation, the ratio of 24-ethylcholesterol to 24-ethylcoprostanol in the particulate fraction was generally >4.

Of the other vegetation, each species appeared to have a unique chemical profile. For example, the m/z 163 biomarker was unique to capeweed (Arctotheca m/z 163, Fig. 1) and triacontanol dominated all extracts and fractions of white clover. The unidentified Arctotheca m/z 163 lipid would appear to have particular value in tracing studies. It was not detected in any other source materials apart from cattle faeces. Capeweed was most likely a small component (<10%) of the pastures grazed by the cattle in late spring.

Figure 1. Average chemical profiles for some possible source materials (Direct) and the particulate (>0.45 μm, Particulate) and filtrate (<0.45 μm, Filtrate) fractions of their water extracts. Error bars indicate the variation between samples where multiple samples were analysed, and likely variation where only single samples were analysed.

Other Sources (data not shown)

In the direct extracts, the fertilisers contained low concentrations of lipids compared to all other source materials. The most abundant lipids were 24-ethylcholesterol, cholesterol, coprostanol and hexacosanol. All concentrations were higher for DAP than SSP and that probably reflects the oil coating applied to DAP during manufacture and some contamination of the SSP. No neutral lipids were detected in either the particulate or filtrate fractions of the water extracts of either fertiliser.

The lipid analyses of the direct extracts and particulate fraction of the decaying grasses and soil contained similar neutral lipids to the grass/weed and faecal samples, except for the Arctotheca m/z 163 lipid. Being a composite of faeces and vegetation, this is not surprising. Again, the filtrate fraction contained few lipids detected in the direct extracts and particulate fraction.

Overland flow

In the overland flow the three major neutral lipids present in the particulate material were hexacosanol, phytol and 24-ethylcoprostanol (Fig. 2). The order of abundance for other lipids depended on the site.

The generally lower concentration of lipids from the Arawata site presumably reflects soil hydrology. While overland flow predominates at the other sites, at Arawata infiltrated water is expressed as overland flow near the monitoring point and its passage through the soil would be expected to decrease lipid concentrations.

Samples were collected from the Macalister Research Farm (MRF) for an overland flow event on section B (8.9 ha) and two events on paddocks A (2.3 ha) and B (2.6 ha) and section A (14 ha). This provided an opportunity to compare lipid concentrations at different scales and time intervals. Many of the differences in concentrations of lipids at the different scales and time intervals were statistically significant (P<0.05). Paddocks A and B contained higher concentrations of most neutral lipids in the particulate material during the first overland flow event than in the second. In paddock B, the higher concentrations may be due to the paddock being grazed 2 d prior to the first event, whereas grazing occurred 24 d prior to the second overland flow event

Figure 2. Chemical profiles of the particulate (>0.45 μm, Particulate) and filtrate (<0.45 μm, Filtrate) fractions of overland flow collected from Arawata and Darnum. Error bars indicate the expected variation in concentrations if multiple samples were analysed.

The particulate phase of the first overland flow event from section A contained low concentrations (many significantly lower, P<0.05) of most neutral lipids compared to subsequent sampling. The major lipids were 24-ethylcholesterol, stigmasterol, phytol, cholesterol and hexacosanol; suggesting material of predominantly plant origin. The lower concentrations of compounds that may be of animal origin may reflect the grazing pattern. When the paddock adjacent to the monitoring station was grazed only 6 d prior to the second overland flow event most lipid concentrations were higher than the sample from the first irrigation. The sample taken at the beginning of the overland flow event had a lipid profile in order of abundance 24-ethyl-5α-cholestanol, phytol, hexacosanol, cholesterol and octacosanol. Mid-way through the overland flow event, the profile had changed to phytol, hexacosanol, 24-ethyl-5α-cholestanol, octacosanol and dotriacontanol. The major changes evident between the beginning and mid-point of the flow were the decrease in the concentration of compounds that might be attributed to animals and increase in compounds that might be attributed to plants. It is possible that compounds derived from animals, such as cows, are mobilised more rapidly than compounds from plants

Figure 3. Chemical profiles of the particulate (>0.45 μm, Particulate) and filtrate (<0.45 μm, Filtrate) fractions of some overland flow collected from the Macalister Research Farm. Error bars indicate the expected variation in concentrations if multiple samples were analysed.

Grazing occurred 15 d prior to both overland flow events in paddock A. The higher concentrations of neutral lipids (P<0.05 for most lipids) in the particulate fraction of overland flow from the first irrigation was attributed to the use of irrigation water from a re-use pond (i.e. drainage water) for the first irrigation and channel water for the second irrigation.

Section B was sampled during one overland flow event. The chemical profile for the particulate fraction in order of abundance was phytol, Δ7-ergostenol, hexacosanol, cholesterol and 24-ethylcholesterol. The last time a paddock in this section had been grazed was 8 d prior to overland flow, explaining the similar values overall for the lipids compared with those observed in Section A during the second overland flow event. The reason for the elevated Δ7-ergostenol compared to other samples is unclear.

The related lipid ratios (i.e. precursor/product) are surprisingly similar for the particulate fractions of the overland flow samples from the MRF, dryland sites and the faeces samples. For example the ratio of coprostanol to epicoprostanol was 4.5 to 6.3 for the dryland sites, 4.5 to 8.5 for the MRF and 2.8 to 6.6 for the water extractable particulate fraction of faeces, and 24-ethylcoprostanol to 24-ethylepicoprostanol 1.4 to 2.8 for dryland sites, 1.3 to 2.3 for the MRF and 1.4 to 2.8 for faeces. In the filtrate fraction of the overland flow samples a relationship appeared to exist between 24-ethylcoprostanol and 24-ethylepicoprostanol, 0.6 to 1.2 for the dryland sites and 0.5 to 1.2 for the MRF.

The ratio of 24-ethylcholesterol to 24-ethylcoprostanol varied widely between sites (0.3 to 1.8 for the dryland sites, and 1.1 to 15 for the MRF). For the sites at the MRF, paddock B that had the greatest period between grazing and the overland flow event and a ratio of 15, while section A had only 6 d between grazing and overland flow with a ratio of 1.1. This suggests that this ratio could possibly be used to attribute pollutants to grazing.

Concluding Discussion

This study has investigated the concentrations of neutral lipids of the sterol and alcohol groups in overland flow exported from pasture based grazing systems, likely source materials and water extracts of those source materials. The results suggest that the pattern of these lipids may be used to identify source materials. Although many of the observations discussed in this paper are not significant (P<0.05), the conservative nature of the statistics used does not preclude the existence of any relationships. Where statistical significance has been shown, the results are unambiguous.

Acknowledgements

The authors would like to acknowledge financial support from the Dairy Australia Limited, GippsDairy and the Victorian Department of Primary Industries. The authors also thank the Macalister Research Farm Cooperative, Stuart Tweddle, Ian Bayley and Mike Malone for the use of their farms.

References

Bligh EG, Dyer WJ (1959) A rapid method of total lipid extraction and purification. Canadian Journal of Biochemistry and Physiology 37, 911-917.

Christie WW (1989) 'Gas chromatography and lipids - A practical guide.' (The Oily Press: Scotland)

Cooke GD, Welch EB, Peterson SA, Newroth PR (1993) 'Restoration and management of lakes and reservoirs.' (Lewis Publishers: Boca Raton USA)

CSIRO Australia (1996) 'Port Phillip Bay Environmental Study.' Commonwealth Scientific and Industrial Research Organisation Australia, Dickson, ACT.

European Environment Agency (1998) 'Europe's environment: The second assessment.' European Environment Agency, Copenhagen, Denmark.

Fruton JS, Simmonds S (1963) 'General Biochemistry.' (John Wiley and Sons, Inc.: London, England)

Jones GJ, Nichols PD, Shaw PM (1994) Analysis of microbial sterols and hopanoids. In 'Chemical methods in Prokaryotic Systematics'. (Eds M Goodfellow and AG O'Donnell) pp. 164-195. (John Wiley and Sons Ltd.: New York, USA)

Leeming R, Ball A, Ashbolt N, Jones G, Nichols P (1994) Distinguishing between human and animal sources of faecal pollution. Chemistry in Australia 61, 434-435.

Leeming R, Ball A, Ashbolt N, Nichols P (1995) Using faecal sterols from humans and animals to distinguishing faecal pollution in receiving waters. Water Research 30, 2893-2900.

Leeming R, Nichols PD (1996) Concentrations of coprostanol that correspond to existing bacterial indicator guideline limits. Water Research 30, 2997-3006.

Martin WJ, Ravi Sabbiah MT, Kittle BA, Birk CC, Naylor MC (1973) Nature of faecal sterols and intestinal bacterial flora. Lipids 8, 208-215.

Nash DM, Halliwell DJ (2000) Tracing phosphorus transferred from grazing land to water. Water Research 34, 1975-1985.

Nash DM, Hannah M, Clemow L, Halliwell D, Webb BJ, Chapman D (2003) A laboratory study of phosphorus mobilisation from commercial fertilisers. Australian Journal of Soil Research 41, 1201-1212.

Nichols PD, Leeming R (1991) Tracing sewage in the marine environment. Chemistry in Australia 58, 274-275.

Nichols PD, Leeming R, Rayner MS, Latham V (1996) Use of capillary gas chromatography for measuring faecal-derived sterols. Application to stormwater, the sea-surface microlayer, beach greases. Regional studies and distinguishing algal blooms and human and non-human sources of sewage pollution. Journal of Chromatography 733, 497-509.

Rosenfeld RS, Gallagher TF (1971) Reduction of esterification of cholesterol and sitosterol by homogenates of faeces. Journal of Lipid Research 12, 192-197.

USEPA (1996) 'Environmental indicators of water quality in the United States.' USEPA, Office of Water, EPA 841-R-96-002, Washington DC USA.