Construction of Pathogen Budgets for Sydney Drinking Water Catchments

Home , Kedumba River, Mulwaree River, Mummel, River Lett

Deterministic model of microbial sources, fate and transport: a quantitative tool for pathogen catchment budgeting

Presented by

Christobel Ferguson

Submitted in total fulfillment of the requirements of the degree of

Doctor of Philosophy

Department of Biotechnology and Biomolecular Science

Faculty of Science

The University of New South Wales

June 2005

The object of the bacteriological examination of any water-supply is to ascertain whether that particular water is, or is not, one the consumption of which may give rise to disease or be prejudicial to its users, either at the time of examination or subsequently. Hygienists are unanimous in recognising that sewage and the excreta of human beings, diseased or healthy, must be looked upon as potential vehicles for disease production. The presence of the excreta of animals must also be looked upon as prejudicial, since it may contain harmful bacteria and other parasites. It is clear, therefore, that the detection of the presence of sewage and of human excreta, and to a lesser extent of animal excreta, must be the aim of the water bacteriologist.

William G. Savage, B.Sc. M.D. from The Bacteriological Examination of Water-Supplies. London 1906.

ABSTRACT

The most important priority for the management of Australian drinking water catchments is the control of pathogen loads delivered to raw water reservoirs and treatment plant intakes. A process-based mathematical model was developed to estimate pathogen catchment budgets (PCB) for Cryptosporidium, Giardia and E. coli loads generated within and exported from catchments. The model quantified key processes affecting the generation and transport of microorganisms from humans and animal excreta using land use and hydrologic data, and catchment specific information including point sources such as sewage treatment plants and on-site systems. The PCB model was applied in the Wingecarribee catchment, Sydney and used to predict and rank pathogen and indicator loads in dry weather, intermediate (<30 mm in 24 h) and large wet weather events (100mm in 24 h). Sensitivity analysis identified that pathogen excretion rates from animals and humans, and manure mobilisation rates were the most significant factors determining the output of the model. Comparison with water quality data indicated that predicted dry weather loads were generally within 1-2 log10 of the measured loads for Cryptosporidium and E. coli and within 1 log10 for Giardia. The model was subsequently used to predict and rank pathogen and indicator loads for the entire (16 000 km2) Sydney drinking water catchment.

iii

ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project’s design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ………………………………………………….

ACKNOWLEDGMENTS

I would like to thank my supervisors for their support and guidance; Professor Brett Neilan (Biotechnology and Biomolecular Science, University of New South Wales), Professor Nicholas Ashbolt (Civil and Environmental Engineering, University of New South Wales), Dr. Daniel Deere (Water Futures Pty Ltd and CRC for Water Quality and Treatment) and Dr. Barry Croke (Integrated Catchment Assessment and Management Centre, Australian National University). I particularly thank Barry for coding the model in ICMS and FORTRAN and helping me to test the model by sensitivity analysis.

I would like to thank the Cooperative Research Centre for Water Quality and Treatment, American Water Works Association Research Foundation, Water Services Association of Australia, Melbourne Water and the Sydney Catchment Authority. I would also like to thank the following people for their generous assistance. Martin Krogh, Chris Chafer, Bala Vigneswaran, Peter Paterson, Danielle Camenzulli, Julian Long, Stuart Naylor, Bruce Whitehill, Gary Bownds, Penny Knights, Tony Paull and Ben Shallis, Sydney Catchment Authority. Katrina Charles, Christine Kaucner, Dr Cheryl Davies, Nanda Altavilla, Dr Peter White, Paul Gwynne, Andrew Feitz, Paul Beavis, Dr David Roser, Robby Smith and Lynnette Menzies, University of New South Wales. Dr Peter Cox, Malcolm Warnecke, Dr Mark Angles, Merran Griffith, Richard Teffer, Myly Truong, Monica Logan, Mike Mannile, Hamish Manzi, Terry Adams and Lyn Tamsitt, Sydney Water Corporation. Professor Tony Jakeman and the research group at the Integrated Catchment Assessment and Management Centre, Australian National University. Dr. Ana Maria de Roda Husman, Dr. Jack Schijven and Dr Peter Teunis, National Institute of Public Health and the Environment (RIVM), the Netherlands. Dr. Melita Stevens, Melbourne Water. Professor Bill Cooper, University of North Carolina, USA.

Joerg Rodehutskors, University of Lippe and Hoexter, Germany. Stephen Burgun and Professor Richard Whittington, Sydney University. Peter Jackson and Dr. Paul Hackney, University of Western Sydney. Dr. Gertjan Medema and Dr. Wim Heijnen, KIWA Water Research, the Netherlands. Dr. Dennis Mulcahy, Dr. Dennis Steffensen, Rachael Miller and Professor Don Bursill, Cooperative Research Centre for Water Quality and Treatment. Professor Bob Wasson, Centre for Resource and Environmental Studies, Australian National University. Dr. Therese Flapper and Grant Leslie, Ecowise Environmental. Dr. Annette Davison, Water Futures. Special thank you to Elena Cotto for formatting the manuscript.

Finally, I would like to thank my parents for their unqualified support of my academic pursuits, and my husband, Dr. Peter Beatson, for his encouragement and for the numerous discussions regarding various aspects of this work, without which, it might not have been completed.

PUBLICATIONS

Ferguson, C.M., Altavilla, N., Ashbolt, N.J. & Deere, D.A. (2003a) Prioritising

Watershed Pathogen Research. Journal of American Water Works Association, 95(2),

92-102.

Ferguson, C.M., Ashbolt, N.J. & Deere, D.A. (2004) Prioritisation of catchment management in the Sydney catchment - construction of a pathogen budget. Water

Science and Technology: Water Supply, 4(2), 35-38.

Ferguson, C.M., Croke, B.F.W., Beatson, P.J., Ashbolt, N.J. & Deere, D.A. (submitted- a) Development of a process-based model to predict pathogen budgets for the Sydney drinking water catchment. Journal of Water and Health.

Ferguson, C.M., Davies, C.M., Kaucner, C., Krogh, M., Rodehutskors, J., Deere, D.A.

& Ashbolt, N.J. (in press) Field scale transport of Cryptosporidium parvum, E. coli and

PRD1 bacteriophage in surface runoff from bovine faecal pats under simulated rainfall.

Journal of Water and Health.

Ferguson, C.M., de Roda Husman, A.M., Altavilla, N., Deere, D. & Ashbolt, N.J.

(2003b) Fate and transport of surface water pathogens in watersheds. Critical Reviews in Environmental Science and Technology, 33(3), 299-361.

Ferguson, C.M., Kaucner, C., Krogh, M., Deere, D. & Warnecke, M. (2004)

Comparison of methods for the concentration of Cryptosporidium oocysts and Giardia cysts from raw waters. Canadian Journal of Microbiology, 50, 675-82.

Chalmers, R., Ferguson, C.M., Caccio, S., Gasser, R., Abs EL-Osta, Y., Heijnen, L.,

Xiao, L., Elwin, K., Hadfield, S., Sinclair, M. & Stevens, M. (2005) Direct comparison of selected methods for genetic categorisation of Cryptosporidium parvum and

vii

Cryptosporidium hominis species. International Journal for Parasitology, 35(4), 397-

410.

Charles, K.J., Ashbolt, N.J., Ferguson, C., Roser, D.J., McGuinness, R. & Deere, D.A.

(2003c) Impacts of centralised versus decentralised sewage systems on water quality in

Sydney's drinking water catchments. Water Science & Technology, 48(11-12), 53-60.

Cox, P., Griffith, M., Angles, M., Deere, D.A. & Ferguson, C.M. (2005) Concentrations of pathogens and indicators in animal feces in the Sydney watershed. Applied and

Environmental Microbiology, 71(10), 5929-5934.

Davies, C., Kaucner, C., Altavilla, N., Ashbolt, N., Hijnen, W., Medema, G., Deere, D.,

Krogh, M. & Ferguson, C. (2004a) Pathogen fate and transport in surface water flow.

Water (Australia), 31(3), 57-62.

Davies, C.M., Altavilla, N., Krogh, M., Ferguson, C.M., Deere, D.A. & Ashbolt, N.J.

(2005a) Environmental inactivation of Cryptosporidium oocysts in catchment soils.

Journal of Applied Microbiology, 98(2), 308-17.

Davies, C.M., Logan, M.R., Rothwell, V.L., Krogh, M., Ferguson, C.M., Charles, K.,

Deere, D.A. & Ashbolt, N.J. (in press) Soil inactivation of viruses in septic seepage.

Letters in Applied Microbiology.

Miller, K., Ferguson, C.M., Gillings, M.R., Mitchell, H., Pappayut, S., Angles, M., Cox,

P., Brusentiev, S. & Neilan, B. (submitted) Comparison of tracing and tracking tools for identifying bacterial contamination in drinking water catchments. Environmental

Science and Technology.

viii

EXECUTIVE SUMMARY

The World Health Organisation (WHO) has identified that pathogen contamination continues to be the greatest risk to the quality of drinking water supplies. Many water utilities are thus embracing the concept of risk management and the implementation of a multi-barrier approach utilizing hazard analysis of critical control points (HACCP) to reduce the risk of pathogen contamination of drinking water supplies. Traditional management of the water supply chain has focused on water treatment plants and disinfection of the distribution networks as the principal barriers to contamination. However, many water utilities are re-discovering the value of judicious management of the drinking water catchment as a significant barrier to potential contamination of the water supply. The management of raw water quality in drinking water catchments requires:

• quantification of point and diffuse pollutant sources; • evaluation of alternate pathogen management practices; • implementation of appropriate pathogen (faecal) control measures; and • prediction and verification of water quality improvements at the catchment scale.

This study attempted to address the above objectives by describing the development of a deterministic model or pathogen catchment budget (PCB), to quantify and predict pathogen and bacterial indicator loads and their dispersion at catchment- scales. The framework for the PCB model was based on a conceptual model of catchment processes that control the generation, fate and transport of pathogens in drinking water catchments (Chapter 1). This chapter also summarises previously described mathematical models for microbial fate and transport in surface water runoff. 0 reviews the literature relating to the quantification of pathogen sources in drinking water catchments. The factors controlling the transport of microorganisms in catchments were also reviewed and published (Ferguson et al., 2003b).

New data quantifying the concentrations of Cryptosporidium, Giardia and E. coli extracted from animal faeces and sewage effluent in the Sydney drinking water

catchments were collated (Chapter 3). Quantification of pathogen and indicator concentrations in a wide range of animal faeces showed that wildlife species excreted low concentrations of indicators and pathogens compared to domestic livestock. Analysis of sewage effluent showed that pathogen and indicator concentrations were highly variable between plants. This variability was probably related to the variety of treatment processes and their efficacy, and possibly changing levels of infection in the contributing communities. Investigation of surface transport mechanisms at field-scale was used to quantify manure mobilisation rates of microorganisms from bovine faecal material (Chapter 4). The results indicated that from least to most, the order of transportation efficiency was Cryptosporidium oocysts, E. coli then PRD1 phage. Typical rainfall events mobilised up to 1% of the Cryptosporidium oocysts from fresh faecal pats over a distance of 10 m downslope (18˚ slope) of bare soil sub-plots. Subsequent rainfall events only mobilised 0.01 - 0.1% of the original total oocyst load from the same faecal pats one-week later.

The PCB model was comprised of five discrete modules describing the hydrology, a land/animal budget, on-site sewage systems, inputs from sewage treatment plants (STPs) and an in-stream processes module (Chapter 5). The model predicted total daily input loads of Cryptosporidium, Giardia and E. coli generated within a catchment for a typical dry weather day, intermediate wet weather (30 mm rainfall in <24 h) or wet weather day (100 mm rainfall in <24 h). For each of these scenarios the model also predicted daily loads exported from each catchment accounting for microorganisms lost via settling and decay, as the loads were routed through downstream catchments. At the base of the Wingecarribee catchment (site 50) the PCB model predicted the daily dry weather loads exported from the catchment were 6.2 log10 oocysts of Cryptosporidium, 6.0 log10 oocysts of Giardia and 8.4 log10 mpn of E. coli.

In large wet weather events this increased significantly to 10.5 log10 (oo)cysts of

Cryptosporidium and Giardia and 14.9 log10 mpn of E. coli. These export loads can be used to predict potential delivery to downstream storage reservoirs and may be used as input variables to the reservoir fate and dispersion model developed by Hipsey et al. (2005). Additional data outputs from the model include spatial maps of catchment loads and sub-catchment rankings of the predicted input loads by both raw daily load and the load per unit area.

Sensitivity analysis was used to assess the importance of thirteen key parameters in the PCB model in determining the export loads predicted by the model (Chapter 6). Single parameter perturbations of the model indicated that the load of pathogens excreted per person per day, pathogen concentrations in animal faeces, animal density and rate of direct deposition were the major parameters determining export loads in dry weather. In wet weather, manure mobilisation and animal density were the most significant parameters. Sensitivity analysis of the sub-catchment input rankings per unit area showed that pathogen excretion rates from animals and humans, and manure mobilisation rates were the most significant factors determining the sub-catchment rankings. Comparison to water quality data collected from the Wingecarribee catchment showed that model predictions for dry weather were within 1-2 log10 of the measured load for Cryptosporidium and E. coli and within 1 log10 for Giardia. The lack of wet weather events limited the amount of data that was able to be collected leaving the wet weather predictions largely untested. Pathogen and indicator loads for the two wet weather events that did occur were similar to the intermediate wet weather load predicted for site 3a, but 1-3 log10 lower than the predicted loads at sites 8 and 49. These results were reasonable given that the rainfall events were also lower (24 and 8 mm) than the size of the rainfall event (<30 mm in 24 h) used in the model calculations.

The PCB model was subsequently used to predict input and export loads for the entire SCA catchment for dry, intermediate and large wet weather events (Chapter 7). The sub-catchments were also ranked by raw predicted load and the load per unit area. In dry weather both the raw and unit area rankings for Cryptosporidium and Giardia were dominated by sub-catchments receiving effluent from STPs, particularly downstream of Kellys Ck (sub-catchment 2503, Moss Vale STP), Mittagong Ck (2504, Bowral STP) and Berrima (2506, Berrima STP). In contrast, predicted E. coli loads were highest in sub-catchments impacted by diffuse pollution from native forest and urban areas including Bundanoon Ck and Lower Mulwaree sub-catchments respectively. When ranked by unit area the highest E. coli loads were predicted for Lower Mulwaree, Leura Falls Ck and Upper Kedumba River sub-catchments. In intermediate and large wet weather events the load from STPs was augmented by the input of diffuse pollution generated by the washoff of faecal material from the land surface and contaminants were quickly routed to downstream sub-catchments. Thus sub-catchments such as 1001 (Warragamba reservoir) which are downstream of

numerous sub-catchments receive large inputs of Cryptosporidium and E. coli. For example, the predicted load of Cryptosporidium oocysts, Giardia cysts and E. coli exported to Warragamba reservoir in a large wet weather event were 11.6, 11.3 and 15.7 log10 microorganisms respectively. Despite this the Warragamba reservoir sub- catchment does not occur in the top ten ranked sub-catchments for loads generated per unit area as the load received was mitigated by its size. The highest loads of Giardia were predicted for sub-catchments dominated by improved pasture with cattle (2502 Kellys Ck, 304 Jembaicumbene Ck and 2627 Woolshed Ck). However, the highest loads per unit area were predicted for sub-catchments that were also impacted by STPs, 2503 downstream of Kellys Ck (Bowral STP), and 303 Gillamatong Ck (Braidwood STP).

The PCB model can be used to predict daily loads of Cryptosporidium, Giardia and E. coli both generated in, and exported from drinking water sub-catchments. The model can also rank sub-catchments by total daily loads or by the load per unit area to identify those sub-catchments that represent the greatest risk of pathogen contamination. This information enables water utility managers to identify and prioritise sub- catchments that pose the greatest potential risk for the delivery of pathogens to downstream reservoirs and to rank those sub-catchments that may require the implementation of control measures to meet local water quality objectives. Future modifications of the model could be undertaken to calculate “infectious pathogen unit” budgets by calculating the proportion of the total pathogen catchment budget that represent potentially viable and infectious pathogenic microorganisms. This would be particularly relevant for the robust protozoan pathogens Cryptosporidium and Giardia. Incorporation of the PCB model into an existing catchment modelling product also using ICMS software (CatchMODS) would enable the model to be used as a scenario analysis tool. This would enable water utility managers to perform cost-benefit analysis of catchment management scenarios for the simultaneous assessment of their impact on pathogen, nutrient and sediment loads. A major outstanding issue is the need for the collection of water quality and flow data during rainfall events to test the wet weather predictions of the model. This constraint is partly related to the persistent drought conditions experienced during this study but is also more generally related to the difficulty and cost of installing and maintaining infrastructure in remote catchment locations.

xii

TABLE OF CONTENTS

Executive Summary ...... ix Table of Contents...... xiii Index of Tables...... xviii Index of Figures ...... xx Abbreviations ...... xxiii Nomenclature ...... xxvii

Chapter 1 Introduction...... 1 Development of a conceptual model...... 1 Review of existing information...... 2 Identification of knowledge gaps...... 11 Selection of index microorganisms...... 15 Review of catchment scale factors...... 16 Review of currently available fate and transport models for pathogens...... 17 Aims of this study ...... 32 Structure of dissertation ...... 33

Chapter 2 Pathogen sources in drinking water catchments ...... 35 Introduction...... 35 Natural processes – inputs from wildlife ...... 35 Population density...... 36 Volume of manure ...... 38 Pathogen prevalence and shedding intensity...... 39 Animal age and behaviour ...... 46 Zoonotic transfer ...... 47 Catchment characteristics...... 48 Farm management – inputs from domestic livestock ...... 48 Population density...... 49 Volume of manure ...... 49 Pathogen prevalence and shedding intensity...... 51 Animal age and behaviour ...... 62 Zoonotic transfer ...... 62 Catchment characteristics...... 63 Manure storage and treatment...... 64 Urban development ...... 65 Sewage treatment plants (STPs) ...... 66 On-site systems (septic leachate) ...... 72 Stormwater / urban runoff...... 77

xiii

Faecal dispersion from recreational use of water bodies ...... 78 Construction of a pathogen source budget...... 78 Chapter 3 Pathogen and indicator concentrations in animal faeces and sewage effluent in the Sydney catchment ...... 81 Introduction...... 81 Materials and Methods...... 83 Sample collection...... 83 Animal faeces...... 83 Sewage effluent...... 84 Sample preparation ...... 85 Animal faeces...... 85 Sewage effluent...... 85 Comparison of bacterial enumeration of faeces with and without sonication ....86 Bacterial analysis ...... 86 Faecal coliforms...... 87 C. perfringens spores...... 87 Protozoan analysis ...... 88 Animal faeces...... 88 Sewage effluent...... 90 Viral analysis...... 91 Animal faeces...... 91 Sewage effluent...... 91 Results...... 92 Faecal coliforms and C. perfringens recovery with and without sonication ...... 92 Faecal coliforms...... 93 Animal faeces...... 93 Sewage effluent...... 94 C. perfringens spores ...... 95 Animal faeces...... 95 Sewage effluent...... 96 Pathogenic protozoa...... 97 Animal faeces...... 97 Sewage effluent...... 100 Viruses ...... 100 Animal faeces...... 100 Sewage effluent...... 101 Moisture content of animal faeces ...... 102 Discussion...... 102 Animal faeces...... 102 Sewage effluent...... 106 Conclusions...... 107 Synthesis of data for use in the pathogen model...... 108

Chapter 4 Field-scale simulation of microbial transport from bovine faecal pats in surface waters...... 117 Introduction...... 117

xiv

Materials and Methods...... 119 Location ...... 119 Preparation of field plots...... 119 The rainfall simulator...... 120 Preparation of inocula ...... 121 Preparation of artificial faecal pats ...... 123 Simulated rainfall experiments ...... 123 Sample analysis...... 124 Data analysis ...... 125 Runoff ...... 126 Plot and flume samples ...... 126 Results...... 127 Rainfall simulation...... 127 Characterisation of artificial faecal pats...... 128 Cryptosporidium oocysts ...... 129 E. coli ...... 133 PRD1 bacteriophage ...... 135 Discussion ...... 136 Conclusions...... 140

Chapter 5 Construction of a Pathogen Budget: Case study in the Wingecarribee catchment ...... 141 Introduction...... 141 Development of a pathogen catchment budget (PCB)...... 144 Model description ...... 146 Hydrologic module ...... 149 Assumptions...... 150 Land module ...... 150 Assumptions...... 153 Assumptions...... 155 Sewage treatment plant module ...... 156 Assumptions...... 156 Assumptions...... 158 On-site systems module ...... 158 Assumptions...... 159 In-stream module ...... 159 Assumptions...... 160 Assumptions...... 160 Application in the Wingecarribee ...... 161 Hydrologic and in-stream modules ...... 164 Land module ...... 166 Sewage treatment plant module ...... 169 On-site systems module ...... 169 Outputs from the model for the Wingecarribee catchment...... 170 Predicted loads ...... 170 Sub-catchment rankings...... 175

Raster diagrams...... 176 Discussion...... 183 Conclusions...... 184

Chapter 6 Testing the model...... 187 Sensitivity analysis...... 187 Analysis of the export loads...... 188 Analysis of sub-catchment inputs by area rankings...... 191 Comparison of model outputs to water quality data ...... 201 Sample site locations...... 201 Site 3a (Maugers)...... 205 Site 8 (Wingecarribee River, Sproules Lane) ...... 206 Site 13 (Wingecarribee River at Burradoo) ...... 207 Site 22a (Medway Rivulet at Oldbury Farm)...... 208 Site 48 (Joadja Creek)...... 209 Site 49 (Wingecaribee River, Greensteads) ...... 210 Sampling Program...... 210 Materials and methods ...... 211 Physicochemical analyses...... 211 Bacterial analyses...... 212 Protozoan analyses...... 212 Statistical analyses ...... 213 Results and Discussion ...... 214 Physico-Chemical analysis ...... 214 Bacterial and Protozoan analysis ...... 214 Molecular source tracing tools...... 224 Introduction...... 224 Materials and Methods...... 226 Bacterial Strains ...... 226 DNA Extraction ...... 227 BOX-PCR ...... 228 IGS-PCR/Rsa 1 ...... 228 Random Amplified Polymorphic DNA (RAPD)...... 229 Automated ribotyping ...... 229 Results and Discussion ...... 230 Analysis of DNA Fingerprints ...... 230 Appropriateness of E. coli as an MST indicator ...... 239 Strain Diversity within a Single Water Sample ...... 239 Conclusions...... 242

Chapter 7 Application of the model to the entire Sydney drinking water catchment245 Description of SCA catchments...... 245 Input data requirements for all SCA catchments ...... 251 Hydrologic module ...... 251 Land module ...... 251 Sewage treatment plant module ...... 251

xvi

On-site sewage systems module ...... 254 In-stream module ...... 254 Model Outputs...... 255 Predicted loads ...... 255 Sub-catchment rankings...... 264 Raw rankings...... 264 Per unit area rankings...... 264 Discussion ...... 272 Conclusions...... 274

Chapter 8 General discussion ...... 275

References...... 287

Appendix 1...... 322 Subcatchment data file for all of the Sydney drinking water catchments......

xvii

INDEX OF TABLES

Table 1.1 Scoring criteria for review of existing information ...... 3 Table 1.2 Example ranking of references for pathogen fate and transport ...... 6 Table 1.3 Analysis of knowledge gaps in a conceptual catchment...... 12 Table 1.4 Identification of factors that influence each catchment process (shading) with the major knowledge gaps highlighted (dark shading)...... 14 Table 1.5 Summary of catchment models...... 28 Table 2.1 Estimates of wildlife animal density for native vegetation land use areas ...... 38 Table 2.2 Manure production rates for wildlife animals...... 39 Table 2.3 Pathogen hazards arising from animals in surface water catchments.....40 Table 2.4 Concentrations of Cryptosporidium spp. in faeces of wildlife animals...... 42 Table 2.5 Concentrations of Giardia spp. in faeces of wildlife animals...... 45 Table 2.6 Mean concentrations of faecal coliforms and E. coli in faeces of wildlife animals ...... 46 Table 2.7 Estimates of domestic animal density...... 49 Table 2.8 Manure production rates for domestic and companion animals ...... 50 Table 2.9 Prevalence and concentrations of Cryptosporidium spp. in faeces of domestic livestock and companion animals...... 52 Table 2.10 Prevalence and concentrations of Giardia spp. in faeces of domestic livestock and companion animals ...... 58 Table 2.11 Concentrations of faecal coliforms and E. coli in faeces of domestic livestock and companion animals ...... 60 Table 2.12 Concentrations of E. coli O157:H7 in faeces of domestic livestock and companion animals ...... 61 Table 2.13 Mean pathogen concentrations in raw and treated STP effluent ...... 67 Table 2.14 Summary of SCA sewage treatment plant characteristics modified from Paterson and Krogh (2003) ...... 70 Table 2.15 Average dry weather flow and mean microbial concentrations in treated STP effluent in SCA catchments from Paterson and Krogh (2003) and Krogh and Paterson (2002) ...... 71 Table 2.16 Protozoan parasites in raw sewage effluent at Braemar STP (Australian Water Technologies, 2002a) and (Olley & Deere, 2003) ...... 72 Table 2.17 Cryptosporidium and Giardia (oo)cyst concentrations in sewage effluent from on-site systems, Robertson and Oakland, NSW...... 76 Table 2.18 Protozoan parasites in unsewered urban runoff (Australian Water Technologies, 2002a)...... 77 Table 3.1 Effect of sonication on recovery of bacteria from animal faecal samples.92 Table 3.2 Faecal coliform concentrations in animal faeces ...... 93 Table 3.3 Faecal coliform concentrations in sewage effluent from STPs in the SCA area of operations...... 94 Table 3.4 C. perfringens spore concentrations in animal faeces...... 96 Table 3.5 C. perfringens spore concentrations in sewage effluent from STPs in the SCA area of operations ...... 97 Table 3.6 Cryptosporidium and Giardia concentrations in animal faecal samples ....99 Table 3.7 Cryptosporidium and Giardia concentrations in sewage effluent from STPs in the SCA area of operations ...... 100

xviii

Table 3.8 Summary of virus data for animal faecal samples ...... 101 Table 3.9 Moisture content for animal faecal samples...... 102 Table 3.10 Volume of manure produced per animal per day...... 110 Table 3.11 Cryptosporidium oocyst concentrations in animal faeces...... 112 Table 3.12 Giardia cyst concentrations in animal faeces ...... 113 Table 3.13 E. coli concentrations in animal faeces...... 114 Table 3.14 Animal access to streams and the likelihood of their deposition in streams and riparian zones ...... 116 Table 4.1 Microbial concentrations in fresh artificial bovine faecal pats at time zero ...... 129 Table 4.2 Mean concentrations and SNK groupings for Cryptosporidium, E. coli and PRD1 bacteriophage for plot samples (surface runoff from bare and vegetated sub-plots after simulated rainfall of 55 mm.h-1 for 30 min) ...... 131 Table 4.3 Mean concentrations and SNK groupings for Cryptosporidium, E. coli and PRD1 bacteriophage for flume samples (surface runoff at 10 m distance from bare and vegetated sub-plots after simulated rainfall of 55 mm.h-1 for 30 min) ...... 132 Table 5.1 Summary of stocks and flow processes for total pathogen unit budget (TPU) ...... 148 Table 5.2 Land use categories for the model derived from the Sydney Catchment Authority’s geographic information system layers...... 151 Table 5.3 Animal density by land use categories...... 152 Table 5.4 Microbial concentrations in manure and manure characteristics...... 155 Table 5.5 Microorganism characteristics ...... 156 Table 5.6 Estimated mean daily flow for sampling sites in the Wingecarribee...... 165 Table 5.7 Subcatchment data file for the Wingecarribee catchment...... 167 Table 5.8 Post-treatment concentrations of microorganisms in Wingecarribee STP effluent ...... 169 Table 5.9 Ranking of Wingecarribee sub-catchments with the highest predicted pathogen and E. coli loads (raw load and per unit area) ...... 176 Table 6.1 Perturbation factors for the sensitivity analysis ...... 188 Table 6.2 Sensitivity analysis of sub-catchment rankings for input per unit area budgets ...... 194 Table 6.3 Site locations for water quality monitoring in the Wingecarribee catchment...... 202 Table 6.4 Physico-chemical results for water collected from the 6 routine sites...... 214 Table 6.5 Bacteriological and Protozoan results from the 6 routine sites...... 215 Table 6.6 Bacteriological and Protozoan results from randomly sampled sites ...... 216 Table 6.7 Location and strain numbers for E. coli isolates collected from raw waters in the Sydney and Shoalhaven catchments ...... 227 Table 6.8 Comparison of the groups produced by different typing methods for E. coli isolates ...... 231 Table 7.1 Sub-catchments within the Sydney drinking water catchment ...... 246 Table 7.2 Arithmetic mean post treatment concentrations of Cryptosporidium, Giardia and E. coli in sewage effluent from STPs in the SCA area of operations (Australian Water Technologies, 2002b; Krogh & Paterson, 2002; Paterson & Krogh, 2003) ...... 253 Table 7.3 Buffer capacity and overflow volumes for STPs in the SCA area of operations ...... 254 Table 7.4 Ranking of Sydney sub-catchments generating the highest predicted pathogen and E. coli loads (raw load and per unit area) ...... 265

xix

INDEX OF FIGURES

Figure 1.1 Conceptual model of a catchment...... 4 Figure 1.2 Link between catchment research and management...... 10 Figure 4.1 Layout of field plots situated on loam soil with a slope of 18º...... 122 Figure 4.2 Cumulative runoff volume (L) for bare and vegetated sub-plots for simulated rainfall events (Runs 1 to 6) ...... 128 Figure 4.3 Cumulative loads of Cryptosporidium oocysts exported from 10 m bare soil sub-plots for experiments 1 and 2. Samples collected at one minute intervals; control run = samples 1-35, fresh run = samples 36-70, aged run = samples 71-105 ..133 Figure 4.4 Cumulative loads of E. coli exported from 10 m bare soil sub-plots for experiments 1 and 2. Samples collected at one minute intervals; control run = samples 1-35, fresh run = samples 36-70, aged run = samples 71-105 ...... 135 Figure 4.5 Cumulative loads of PRD1 phage exported from 10 m bare soil sub-plots for experiments 1 and 2. Samples collected at one minute intervals; control run = samples 1-35, fresh run = samples 36-70, aged run = samples 71-105 ...... 136 Figure 5.1 System view of the model (PCB) showing the linkages between the input data and the 5 separate subroutines or modules to produce the model outputs ...... 161 Figure 5.2 Map of the Wingecarribee catchment showing the 52 sub-catchments used in this study. The top panel shows the distribution of land use across the catchment, while the bottom panel shows the stream network and the sampling sites 163 Figure 5.3 Cryptosporidium loads log10 oocysts a) generated in each sub-catchment and b) exported from each sub-catchment and routed downstream (not in sequential downstream order) ...... 172 Figure 5.4 Giardia loads log10 cysts a) generated in each sub-catchment and b) exported from each sub-catchment and routed downstream (not in sequential downstream order) ...... 173 Figure 5.5 E. coli loads log10 mpn a) generated in each sub-catchment and b) exported from each sub-catchment and routed downstream (not in sequential downstream order) ...... 174 Figure 5.6 Cryptosporidium input loads log10 oocysts generated within each sub- catchment per km2 in a) dry weather, b) intermediate wet weather event and c) large wet weather event ...... 177 Figure 5.7 Cryptosporidium loads log10 oocysts exported from each sub-catchment in a) dry weather, b) intermediate wet weather event and c) large wet weather event.178 Figure 5.8 Giardia input loads log10 cysts generated within each sub-catchment per km2 in a) dry weather, b) intermediate wet weather event and c) large wet weather event ...... 179 Figure 5.9 Giardia loads log10 cysts exported from each sub-catchment in a) dry weather, b) intermediate wet weather event and c) large wet weather event...... 180 Figure 5.10 E. coli input loads log10 mpn generated within each sub-catchment per km2 in a) dry weather, b) intermediate wet weather event and c) large wet weather event ...... 181 Figure 5.11 E. coli loads log10 mpn exported from each sub-catchment in a) dry weather, b) intermediate wet weather event and c) large wet weather event...... 182 Figure 6.1 Relative change in total Cryptosporidium loads exported during wet weather events from four representative sub-catchments (52, 2, 28, 40). The dominant microorganism source for each sub-catchment is indicated in the legend...... 189 Figure 6.2 Calaang Creek at Maugers (site 3a) ...... 205

Figure 6.3 Wingecarribee River at Sproules Lane (site 8) ...... 206 Figure 6.4 Wingecarribee River at Burradoo (site 13) ...... 207 Figure 6.5 Medway Rivulet at Oldbury Road (site 22a) ...... 208 Figure 6.6 Joadja Creek (site 48)...... 209 Figure 6.7 Wingecarribee River at Greensteads (site 49)...... 210 Figure 6.8 Predicted versus measured daily dry weather Cryptosporidium loads exported from sub-catchments in the Wingecarribee...... 218 Figure 6.9 Predicted versus measured daily intermediate wet weather Cryptosporidium loads exported from sub-catchments in the Wingecarribee...... 218 Figure 6.10 Predicted versus measured daily dry weather Giardia loads exported from sub-catchments in the Wingecarribee...... 220 Figure 6.11 Predicted versus measured daily intermediate wet weather Giardia loads exported from sub-catchments in the Wingecarribee...... 220 Figure 6.12 Predicted versus measured daily dry weather E. coli loads exported from sub-catchments in the Wingecarribee ...... 222 Figure 6.13 Predicted versus measured daily intermediate wet weather E. coli loads exported from sub-catchments in the Wingecarribee...... 222 Figure 6.14 BOX –PCR was performed on the 101 E. coli isolates from water samples and run on 2% agarose gel. Isolates were sorted into clonal lines and a representative fingerprint from each is shown above ...... 235 Figure 6.15 IGS-PCR followed by restriction digestion with Rsa1. Digests were separated on 2% agarose. Isolates were sorted into clonal lines and a representative fingerprint from each line is shown above...... 235 Figure 6.16 Representative patterns of each E. coli ribogroups. DNA extracts from the isolates was treated with EcoRI and automated ribotyping was performed with the RiboPrinter® system (Qualicon, Inc., Wilmington, USA)...... 237 Figure 6.17 RAPD patterns produced by the environmental isolates of E. coli. RAPD PCR products were separated on 3% agarose gels with a 100bp marker (M) included in each ...... 238 Figure 6.18 Accumulation Curve for the isolates of the 26-03-2002 Collection. The number of new observations is plotted against the number of total observations. Isolates were randomized for entry into the curve to ensure independence of observation...... 240 Figure 7.1 Map of the Sydney drinking water catchments...... 250 Figure 7.2 Cryptosporidium loads log10 oocysts generated in each sub-catchment (not in sequential downstream order)...... 258 Figure 7.3 Cryptosporidium loads log10 oocysts exported from each sub-catchment and routed downstream (not in sequential downstream order) ...... 259 Figure 7.4 Giardia loads log10 cysts generated in each sub-catchment (not in sequential downstream order) ...... 260 Figure 7.5 Giardia loads log10 cysts exported from each sub-catchment and routed downstream (not in sequential downstream order) ...... 261 Figure 7.6 E. coli loads log10 mpn generated in each sub-catchment (not in sequential downstream order) ...... 262 Figure 7.7 E. coli loads log10 mpn exported from each sub-catchment and routed downstream (not in sequential downstream order) ...... 263 Figure 7.8 Cryptosporidium input loads log10 oocysts generated within each sub- catchment per km2 in a) dry weather, b) intermediate wet weather event and c) large wet weather event ...... 266

xxi

Figure 7.9 Cryptosporidium export loads log10 oocysts exported from each sub- catchment in a) dry weather, b) intermediate wet weather event and c) large wet weather event ...... 267 Figure 7.10 Giardia input loads log10 cysts generated within each sub-catchment per km2 in a) dry weather, b) intermediate wet weather event and c) large wet weather event ...... 268 Figure 7.11 Giardia export loads log10 cysts exported from each sub-catchment in a) dry weather, b) intermediate wet weather event and c) large wet weather event ...... 269 Figure 7.12 E. coli input loads log10 mpn generated within each sub-catchment per km2 in a) dry weather, b) intermediate wet weather event and c) large wet weather event ...... 270 Figure 7.13 E. coli export loads log10 cfu exported from each sub-catchment in a) dry weather, b) intermediate wet weather event and c) large wet weather event...... 271

xxii

ABBREVIATIONS

α Degree of statistical significance θ Soil moisture characteristic φX174 (virion) Small spherical DNA bacteriophage to E. coli AWTS Aerated wastewater treatment systems AFP Artificial faecal pat (bovine faeces) ANCOVA Analysis of covariance ANOVA Analysis of variance ARI Average recurrence interval BASINS Better assessment science integrating point and non-point sources (model) BEV Bovine enterovirus B. fragilis Bacteroides fragilis BMP Best management practice BOM Bulk organic matter bp Base pair(s) CC-PCR Cell culture polymerase chain reaction CDC Centre for Diseases Control and Prevention CI Confidence interval Ck Creek cm Centimetre C. parvum Cryptosporidium parvum CPE Cytopathic effect C. perfringens Clostridium perfringens Crusted Artificially-prepared bovine faecal pat dried in an incubator at 20°C for 24 h d Day DAPI 4’,6-diamidino-2-phenylindole stain DAPI/PI 4’,6-diamidino-2-phenylindole/propidium iodide stain df Degrees of freedom DIC Direct interference contrast DLVO theory Derjaguin-Landau-Verwey-Overbeek theory

xxiii

dNTP Deoxyribonucleoside triphosphate d/s Downstream dw Dry weight E. coli Escherichia coli FISH Fluorescent in situ hybridization g Gram g Gravity (centrifugation) GI Gamma-irradiated GLM Generalized linear model GM Geometric mean GWLF Generalized watershed loading function (model) h Hour HACCP Hazard analysis and critical control point(s) HSPF Hydrologic simulation program FORTRAN (model) Hwy Highway IDEA Intermittently decanted extended aeration IFA Immunofluorescent antibody staining IHACRES Identification of unit hydrographs and component flows from rainfall, evaporation and streamflow (hydrologic model) IMS Immunomagnetic separation Indicator(s) Faecal indicator microorganisms IPU Infectious pathogen units (proportion of TPU budget that is pathogenic) k Inactivation rate(s) kL Kilolitre km Kilometre L Litre LA Load allocation(s) log Logarithm m Metre M Molar MEM Minimal essential medium min Minute

xxiv

mL Millilitre ML Megalitre mm Millimetre mo Month mol. Molecular MPN Most probable number MS2 (virion) F-specific RNA coliphage to E. coli n Number of observations N North

N0 Concentration of viable oocysts at time 0 ND Not determined NDC Natural disinfection criteria Neth The Netherlands NI Non-irradiated nm Nanometre NOM Natural organic matter

Nt Concentration of viable organisms at time t NTU Nephelometric turbidity units (oo)cysts Cryptosporidium oocysts and Giardia cysts p Probability p Number of positive samples Pathogen(s) Faecal-oral pathogen(s) PBS Phosphate-buffered saline PCR Polymerase chain reaction Pers. comm. Personal communication pfu Plaque-forming unit(s) pI Isoelectric point PRD1 (virion) Somatic phage to Salmonella typhimurium LT2 PROMISE Pathogen emission model QMRA Quantitative microbial risk assessment R River R2 Coefficient(s) of determination RBC flume Replogle Bos Clemmens flume

xxv

rpm Revolutions per minute RT-PCR Reverse transcriptase polymerase chain reaction SCA Sydney Catchment Authority SD Standard deviation SEDMOD Sediment explicit delivery model SNK Student-Newman-Keuls Test STARS Sediment/chemical transport with advection, resuspension and settling (model) STP Sewage treatment plant t Time TMDL Total maximum daily load(s) TPU Total pathogen unit (budget) μg Microgram UK United Kingdom μL Microlitre μm Micrometre μS Microsiemens u/s Upstream Uncrusted Artificially-prepared bovine faecal pat less than 2 h old UNSW University of New South Wales USA United States of America UV Ultraviolet VBNC Viable but non culturable WAM Watershed assessment model (hydrologic model) WATNOT Pathogen dispersion model WLA Waste load allocation(s) Wt. Weight yr Year

xxvi

NOMENCLATURE

δj,i Survival rate of microorganism (j) in material (i) (soil or water) λ Land use category a Area of sub-catchment (l) (km2)

As,l Number of animals of species (s) in sub-catchment (l)

βj Number of microorganisms (j) excreted per person per day (accounts for excretion rate from infected person multiplied by the prevalence rate within the population bl Buffer capacity for sewage treatment plant (STP) in sub-catchment (l) (ML) cj,l Post treatment concentration of microorganism (j) in sub-catchment (l)

Cl Proportion of the population in sub-catchment (l) connected to an STP -1 ds The amount of manure produced (kg.day ) for animal species (s)

Ds Probability of species (s) defecating directly into stream e Event duration (days)

Ej,l Exported load of microorganism (j) for sub-catchment (l)

m E ,lj Measured export load of microorganism (j) from sub-catchment (l) fl Relative mean annual rainfall for sub-catchment (l) (rainfall with respect to mean annual rainfall for total catchment area)

Fj Probability of microorganism (j) (both bound and not bound to suspended sediment particles) being deposited or bound to bed sediment over a 1 km reach (0-1)

Gl Identifies the downstream sub-catchment into which an upstream sub-catchment (l) drains

Hl Identifies the sub-catchment location of the STP to which an upstream sub- catchment (l) was connected

Ij, k, l Input to stream of microorganism (j) from source (k) for sub-catchment (l)

Lj Land budget for microorganism (j) m Catchment moisture deficit (mm)

Ms Fraction of faeces on land that would be transported to stream in a large rainfall/runoff event for each animal species (s) nl Proportion of population connected to a sewage treatment plant in sub-catchment (l) (0-1)

Od Proportion of on-site systems connected to streams in dry weather

xxvii

Ow Proportion of on-site systems connected to streams in wet weather -1 Pj,s The concentration (microorganism.kg manure) of microorganism (j) in animal species (s) r Rainfall depth (mm) rl Event rainfall depth for subcatchment (l) L R l Local reach length from node (l), assumed to be the square root of the sub-catchment area (km)

Rl Reach length between node (l) and next node downstream (km) s Stream order

Sl Population of sub-catchment (l) t Threshold of moisture deficit (m) for producing flow

Tl Travel time from node (l) input to receiving node (days) U Effective rainfall (mm)

Ul Effective rainfall (mm) for sub-catchment (l)

U0 Scale factor for event size (event size for mobilisation to equal [1-exp(-1)] of the maximum value) (mm) v Flow velocity (m.s-1) V Volume of effluent produced per person per day (160 L) W Volume of effluent transported in a wet weather event (ML)

Xs Access to streams for species (s) (whether stream is fenced or animals are housed) (0-1)

Yj,l Measured concentration of microorganism (j) per litre at sub-catchment (l)

Zl Mean daily flow measured in ML at the same location as Yj,l in sub-catchment (l)

xxviii

Chapter 1 Introduction

A large proportion of this chapter has been published as: Ferguson, C. M., Altavilla, N., Ashbolt, N. J. and Deere, D. A. (2003) Prioritizing Watershed Pathogen Research. Journal of American Water Works Association 95(2): 92-102.

Previous studies have quantified faecal indicator microorganisms (“indicators”) and faecal-oral pathogens (“pathogens”) for a variety of diffuse and point sources within catchments. Others have quantified concentrations of pathogens and indicators in receiving waters. However, relatively few have quantified the effect of the processes that link sources to the hydrological catchment. Quantification of these processes would enable pathogen data to be incorporated into existing hydrologic models with the potential to predict the dispersion of pathogen loads in drinking water catchments.

This study describes the development of a mathematical model to predict the origin, fate and transport of the pathogens Cryptosporidium and Giardia and the indicator bacteria Escherichia coli in drinking water catchments. The model has been applied to the Sydney drinking water catchment. This extensive and moderately developed catchment of over 16 000 square kilometres is managed by the Sydney Catchment Authority (SCA). There are approximately 100 000 people and 1 million head of livestock within the catchment. Pathogen sources include eleven sewage treatment plants (STPs), many unsewered townships and a number of dairies. Extensive sheep and cattle grazing country comprise the bulk of the catchment.

Development of a conceptual model

The first step was to identify the processes that govern pathogen sources, and their subsequent fate and transport within catchments, and to develop these into a conceptual model (flow chart) (Figure 1.1). The conceptual model could then be used as a framework for the development of a mathematical model. The conceptual model represents the human impacts of urban development and farm management, as well as

the effects of riparian vegetation and natural processes. The natural processes represent the baseline of catchment activities that influence pathogen fate and transport in all catchments. Other processes such as the effect of riparian vegetation, farm management techniques and levels of urban development that occur within a catchment are defined by local conditions and the extent of their impact will be specific for each catchment.

Water utility managers can use the conceptual model to illustrate the key processes that take place within a drinking water catchment. This enables managers to identify pathogen sources and implement control measures that protect drinking-water consumers. However, in the absence of at least semi-quantitative data, the model is of limited use. Quantitative data is necessary to enable accurate prediction of exposure, which is the first part of quantitative microbial risk assessment (QMRA), and to reveal the extent to which any particular source may contribute pathogens to drinking water. Furthermore, the model cannot quantify how control measures would reduce exposure for a given design specification, without sound fate and transport data on pathogens.

In summary, quantitative information is required for the implementation of systematic catchment management approaches. In the US, for example, this might be used for developing pathogen Total Maximum Daily Loads (TMDL) (USEPA, 2001). This would provide a basis for setting Waste Load Allocations (WLA) and Load Allocations (LA) and supporting Best Management Practice (BMP) programs to ensure that QMRA derived exposure targets are met.

Review of existing information

Following development of the conceptual model, the next step was the collation and interpretation of existing information that could be used in the development of the mathematical model. Scoring criteria were used to rank more than 200 papers such that the information derived from them could be evaluated for quality (Table 1.1). The scoring criteria were weighted so that higher scores reflected more comprehensive methods and more complete study design and statistical analysis. This provided semi-

quantitative and quantitative data on pathogen sources and processes identified within the conceptual model.

Table 1.1 Scoring criteria for review of existing information

Criterion (Total maximum value = 14) Value Laboratory Quality System Yes or No Sampling Details 3 Sample volume tested 1 Description of Experiment - Brief 1 Complete 2 Method Used 3 Bacteria - Plate count 1 MPN1 tube method 2 Membrane Filtration or Colilert® 3 Viruses - Phage without appropriate controls 1 Phage including appropriate controls 2 Enteric viruses by cell culture or PCR 3 Protozoa - Immunofluorescence (IFA) 1 IFA+DIC2 or DAPI3 or excystation 2 IMS4+IFA + infectivity & internal controls 3 Use of adequate experimental controls 2 Percentage recovery calculated 1 Adjustment for % recovery 1 Study Design / periodicity 4 Temporal variation 1 Seasonal variation 1 Physico-chemical variation 1 Spatial variation 1 Statistical Analysis 2 Brief 1 Complete analysis 2

1 MPN – Most probable number 2 DIC – Differential interference contrast microscopy 3 DAPI – 4’,6-diamidino-2-phenylindole stain 4 IMS – Immunomagnetic separation

Figure 1.1 Conceptual model of a catchment

There was considerable variation in the quality of the reviewed papers, reflected in the range of scores for the reviewed articles (3 to 13). The application of the scoring criteria assisted the interpretation of data and results, allowing comparisons to be made between papers utilising different methods. It was acknowledged that data from earlier publications will have been derived using methods that may have been superseded and that this is a function of progress. This is not to say that such data is not useful, but that it must be interpreted with consideration to the remaining scoring criteria. For example, if data were obtained using a method rated as 1, the publication could still score as high as 12 if the sampling details, study design and statistical analysis were comprehensive. A summary of the major contributions, from a selection of the most highly rated scientific articles reviewed by this process are shown in Table 1.2. This table also highlights the attributes of the paper that contributed to the final score.

The review of available knowledge has been used to direct and support current catchment management activities. Meanwhile, the processes that were identified as having the most significant knowledge gaps were targeted for new research. Research priorities can be set considering the magnitude of the knowledge gap in combination with the relative benefit and the resources required to reduce the gap. This iterative approach is summarized in the flow diagram given in Figure 1.2. As new data becomes available it can be incorporated into both the conceptual and mathematical models.

Table 1.2 Example ranking of references for pathogen fate and transport

Reference Score Subject Major Contribution/s Advantages Constraints 1 (Atherholt et 12 Effect of rainfall on Statistically significant Robust detection Data not corrected al., 1998) Giardia and associations between methods, equivalent for recovery Cryptosporidium parasite concentrations and sample volumes efficiency rainfall and associated examined for parasite increases in particulate concentrations matter (Brenner, 11 Uses phages to The presence of host- Showed significant Reliance on phages Brenner & Salmonella typhimurium specific phages were correlations between as suitable models Schwartz, WG 49 and B. fragilis indicative of septic microbiological for enteric virus 1999) HSP 40 as indicators of contamination while faecal parameters and stream survival and the presence of enteric coliforms and enteric temperature, transport viruses in surface waters viruses could persist in suspended solids and contaminated by septic catchments, particularly rainfall discharge during summer (Conboy & 10 Identifies factors that Demonstrates a link Detailed descriptions Risk assessments Goss, 2000) increase the risk of between manure spreading of the effect of soil were not confirmed microbial contamination and risk of contamination of type, geology, well using quantified of groundwater groundwater wells construction and farm microbial analysis management practices

Table 1.2 (cont’d.) Example ranking of references for pathogen fate and transport

Reference Score Subject Major Contribution/s Advantages Constraints 1 (Deborde et al., 13 Male-specific and somatic Examined virus occurrence Detailed description of Reliance on 1998) coliphages were in septic systems, sites and sampling coliphages as detectable in septic attachment and movement design. Compares the models of enteric effluent and the impacted through groundwater transport of microbial virus survival and groundwater. Proposed tracers to the transport transport setback distances of 30.5 of the hydrological m (100 ft) may be tracer, bromide, insufficient for some including hydrogeological settings breakthrough curves (Freire-Santos 9 Examines the combined Factorial design to assess Viability assessed by Recovery efficiency et al., 2000) influence of combination of temperature, two different methods. not specified, environmental factors on salinity and storage time on Robust statistical questionable the viability of the viability of C. parvum analysis validity of Cryptosporidium oocysts ‘viability’ methods in water

(Hansen & 13 Evaluation of the effect of Demonstrates link between Detailed description of Oocyst viability was Ongerth, 1991) time and catchment catchment use and oocyst site and sampling not assessed characteristics on concentration design. Samples Cryptosporidium oocyst adjusted for recovery concentrations in river efficiency water

Table 1.2 (cont’d.) Example ranking of references for pathogen fate and transport

Reference Score Subject Major Contribution/s Advantages Constraints 1

(Jewett et al., 10 Examines the effect of Bacterial transport was Identifies ionic Data limited to P. 1995) ionic strength and pH on unaffected by changes in strength as an fluorescens, no the transport of P. pH. Decreasing the ionic important factor for information on fluorescens P17 using strength decreased the bacterial transport and pathogens with laboratory columns and bacterial collision efficiency demonstrates that different isoelectric large pore glass fibre glass filters can be points or filters used as a supplement hydrophobicity to columns (Joergensen et 9 Effects of land use, Significant numbers of Detailed Indicators were not al., 1998) manure spreading and faecal indicators were measurements of soil detected at adjacent earthworms on the transferred into the soil via characteristics related groundwater wells movement of faecal earthworm burrows to microbial biomass indicators into the soil following manure and faecal indicator application. Particularly on concentrations grassland soil compared to arable soil (Mawdsley et 10 Vertical and horizontal Oocysts not adsorbed to Detailed description of Data not corrected al., 1996) transport of soils but retained in run off experimental design for recovery Cryptosporidium in soils and soil extraction efficiency method

Table 1.2 (cont’d.) Example ranking of references for pathogen fate and transport

Reference Score Subject Major Contribution/s Advantages Constraints 1

(Medema & 13 Models the discharge and Uses two models to Detailed description of Protozoan loads Schijven, 2001) dispersion of Giardia and integrate water quality and experimental design from agricultural Cryptosporidium in pathogen survival data to and model runoff not included. surface water predict parasitic protozoan development. Model Discharge from concentrations in surface predicted protozoan combined sewer water concentrations to overflows was not within an order of modified for rainfall magnitude at 5/6 sites events

(Medema, 10 Influence of temperature Calculation of die-off rates Robust detection Size of laboratory Bahar & and autochthonous in river water methods used for all microcosms and Schets, 1997b) microorganisms on the pathogens and recovery efficiency survival of pathogens and indicators not specified indicators in river water (Tate et al., 9 Hydrological transport of Quantification of oocyst Detailed description of Method of 2000) Cryptosporidium oocysts transport after rainfall site and sampling estimating recovery from faecal deposits design efficiency not specified

1 Maximum possible score was 14 (see Table 1.1)

Select and Review using collate relevant quality scoring information criteria

Identify knowledge gaps

Develop and refine Conceptual Enhanced Undertake Model knowledge research to meet 7 utility needs

Prioritize watershed management for drinking-water

Hydrologic quality outcomes Assessment of watershed exposure for model QMRA TMDL

WLA and

BMP

The main body text defines abbreviations.Performance Monitoring and Compliance

Abbreviations are defined in the main text.

Figure 1.2 Link between catchment research and management

Identification of knowledge gaps

The conceptual model (Figure 1.1) identified the dominant processes in drinking water catchments and highlighted the knowledge gaps in the literature regarding these processes. The relative magnitude of the knowledge gap was estimated using expert judgment. The most engineered processes for example, sewage treatment, are relatively well understood compared to processes such as buffer strip entrapment and host prevalence. A summary of the knowledge gaps, the management questions they relate to, the research required and proposed control measures, is given in Table 1.3. The magnitude of gaps in our collective knowledge of fundamental processes is generic to all catchments. However, the relative importance of gaps is catchment-specific. For example, the removal of pathogens by sewage treatment plants (STP) is only relevant where there is one or more STP in the catchment.

Table 1.3 highlights the significant knowledge gaps relating to the conceptual model and potential control measures. There was a number of knowledge gaps that were identified as large or medium, representing a wide variety of research project needs. However, this complexity was reduced because each of the catchment processes could be defined by a set of common factors that principally influence the fate and transport of pathogens. For example, the effectiveness of wetland retention basins is primarily influenced by visible and UV light, rainfall / hydrology, vegetation type, surface properties and biological activity. Each of the catchment processes identified in the conceptual model can be defined in terms of these factors, as outlined in Table 1.4 In conclusion it was found that it was not necessary to undertake research into numerous, specific, catchment processes, such as “buffer strip entrapment”, “wetland retention” and so on. Instead, it was apparent that the need was to quantify the relationships between the underlying factors and pathogen fate and transport behaviour. This information can then be used as inputs to a mathematical model to simulate the fate and transport processes relating to pathogens. This will enable catchment managers to predict pathogen transport and attenuation across diverse catchments as well as to investigate the influence of a broad range of control measures.

Table 1.3 Analysis of knowledge gaps in a conceptual catchment

Process Management Question Knowledge Research need Control Measure gap size Sewage How to reduce viable Small Ongoing evaluation of all new STP and on-site treatment Treatment pathogen density processes and performance of on-site processes (various) systems Faecal Does direct access of Medium It is evident that access will lead to Preventing direct access of Disintegration humans and animals lead pathogen presence in water. humans and animals to and Dispersion to pathogen presence in Quantification of the health risk is not waterways water ways and, if so, to established and relates to host what extent prevalence, defecation behaviour and faecal dispersion processes in water. Manure How to reduce viable Medium Evaluation of treatment options and Separation of solids and Treatment pathogens in animal waste determination of best management liquids. Composting and practices (BMPs) bunding prior to manure spreading Wetland What quantity of pathogens Medium How efficient are wetlands at trapping Well designed wetlands at Retention are retained by wetlands and inactivating pathogens (compared appropriate locations in the to indicator microorganisms) catchment Retention Pond Can retention pond Medium How efficient are retention ponds at Well-designed retention Entrapment entrapment decrease trapping pathogens? Is sediment ponds at appropriate locations pathogen loads to resuspension and pond overflow a risk, in the catchment. Guide for waterways and if so, to what extent BMP related to retention ponds

Table 1.3 (cont’d.) Analysis of knowledge gaps in a conceptual catchment

Process Management Question Knowledge Research need Control Measure gap size Buffer Strip Trap efficiency of each Large Build on extensive knowledge relating Riparian buffer strips. BMP Entrapment type of pathogen related to to sediment particles by establishing for the installation (including slope, hydrological factors, distribution of pathogens according to sizing) and maintenance of vegetation type and particle size and association with buffer strips drainage faecal/soil matter Environmental Reduction in viable Medium Identify the combined stressors acting Encouraging inactivation by Inactivation pathogen density in synergy as occurs in natural systems. natural processes Including both physico-chemical and ecological stressors Host Prevalence What animals to target and Large What are the prevalence patterns of Feral animal control. what are the target stock pathogens in feral and native wild Minimize opportunities for densities animal populations related to zoonotic and interspecies population density? What is the transfer of pathogens potential for transmission between different animal and human populations and how does this relate to relative population densities of hosts and infectivity of pathogens to humans Soil Retention Trap efficiency of each Medium Build on extensive knowledge relating Drainage interruptions type of pathogen related to to sediment particles by establishing (stormwater and on-farm retention time, vegetative distribution of pathogens according to practices, wetlands, retention factors, surface properties particle size in faecal matter and basins). Control of sub- and hydrological factors inactivation during retention surface transport. BMP for considering combinations of synergistic surface cover and surface physico-chemical and ecological treatments (vegetation cover, stressors ploughing)

Table 1.4 Identification of factors that influence each catchment process (shading) with the major knowledge gaps highlighted (dark shading)

Factor Catchment

+ Process pH ºC F/T V/UV M R/H NOM NH4 P ST SP Biota

Environmental Inactivation

Soil Retention

Faecal Disintegration & Dispersion Wetland Retention

Buffer Strip Entrapment

Retention Pond Entrapment Manure Treatment

Sewage Treatment

Host Prevalence

ºC – Temperature F/T – Freezing and thawing V/UV – Visible and UV light M – Moisture content R/H – Rainfall and hydrology NH4+ - Ammonia content NOM – Vegetation and Natural organic matter P – Physical processes ST – Soil type SP – Surface properties

Selection of index microorganisms

Although the aetiological agent responsible for waterborne disease remains unidentified in many cases, in recent decades the protozoan pathogens Cryptosporidium and Giardia have been identified as the cause of numerous outbreaks of waterborne disease (Hrudey & Hrudey, 2004; Hunter, Waite & Ronchi, 2003). In addition to these protozoan pathogens, the bacterial pathogen E. coli O157:H7 and pathogenic viruses such as Rotavirus and Norwalk have also been implicated in waterborne disease outbreaks with severe consequences for consumers (Hrudey & Hrudey, 2004). Although viruses are a significant source of waterborne disease outbreaks, data on the extent and cause of these outbreaks is frequently limited due to the difficulty and cost of analysing samples for these organisms. This study will therefore quantify the occurrence, fate and transport of the protozoan pathogens Cryptosporidium and Giardia and the indicator bacteria E. coli as an index for bacterial pathogens.

Cryptosporidium and Giardia are waterborne parasites that inhabit the intestinal tract of humans and animals, and can cause gastrointestinal illness. They are released into the environment (via faeces) in inactive forms that are resistant to many natural and artificial stressors (including disinfection). These are called oocysts for Cryptosporidium and cysts for Giardia. New hosts are infected by ingestion of (oo)cysts. As the number (dose) of (oo)cysts required to cause infection is small (about 10 organisms), and infected individuals shed (oo)cysts in very high numbers (typically 1010 over the course of the disease), and the (oo)cysts are highly persistent outside the host, their disease causing potential is quite high. Cryptosporidium oocysts are spherical, diameter 4 to 6 µm. Giardia cysts are larger and elliptical, size 8-12 by 7-10 µm.

Faecal coliform bacteria have frequently been used as an indicator of microbiological contamination of water because they are common inhabitants of the intestinal tract of both humans and warm-blooded animals, and are generally excreted in high numbers. The faecal coliform group is defined as Gram-negative bacteria that can ferment lactose with the production of acid and gas at 44ºC within 24 hours; they primarily include species from the genera Escherichia and Klebsiella. It is estimated

that between 91-95% of detected faecal coliforms are E. coli (P. Cox pers. comm.). The specific enumeration of E. coli, rather than faecal coliforms is now recommended practice for water quality analysis in many countries, including Australia. This is due to the more specific relationship between E. coli, faecal contamination, and pathogen presence. However, much of the existing data relates to faecal coliform counts and to make use of this data the Virginia Department of Environmental Quality have derived a formula to transform faecal coliform data to E. coli data (Virginia Department of Environmental Quality, 2003). The formula is:-

E. coli concentration = 2 -0.0172*(Faecal coliform concentration 0.91905) (1)

where the bacterial concentrations are in cfu.100 mL-1.

The majority of E. coli serotypes are non-pathogenic. However, some serotypes possess enterotoxigenic, enteropathogenic, enteroinvasive or enterohaemorrhagic virulence factors. The occurrence of enterohaemorrhagic E. coli O157:H7 in drinking water supplies has caused a number of serious waterborne outbreaks of disease resulting in acute and chronic illness and fatalities. The most widely known outbreak occurred in Walkerton, Ontario in 2000. However, a number of waterborne outbreaks caused by E. coli O157:H7 occurred prior to Walkerton including, Cabool, Missouri (USA) and Saitama (Japan) in 1990, and Washington County, New York (USA) in 1999 (Hrudey & Hrudey, 2004).

Review of catchment scale factors

Although the specific knowledge gaps can be clearly identified (Table 1.4), the application of research findings to catchments is complicated by the necessity to address these issues at a variety of scales. Water utilities have identified that quantitative information regarding the factors influencing pathogen fate and transport is required at all levels of scale:

• the total catchment scale is important for determining the relative significance of sources that are distant from abstraction points and how ameliorative steps may be effective; • the sub-catchment and land unit scale is important for prioritizing BMP programs according to relative pathogen exposure reductions; • the single site or field plot scale is necessary for comparing the effectiveness of on-site treatment and containment practices and comparing one development with another; and • the laboratory scale enables pathogen survival and transport to be examined under controlled conditions enabling a wide range of parameters to be tested for their single and combined effects.

Examination of the literature reveals that the majority of research studies have examined factors at either the laboratory-scale or the field plot-scale, for example, (Camesano & Logan, 1998; Jewett et al., 1995; Mawdsley et al., 1996; Stoddard, Coyne & Grove, 1998). Fewer studies have been conducted at the sub-catchment or total catchment scale, probably reflecting the increased cost and logistical difficulties (Brenner, Brenner & Schwartz, 1999; Medema & Schijven, 2001). Great care needs to be taken in moving between scales, extrapolation is possible however, the extent of extrapolation is limited and should be verified by appropriate calibration and ground- truthing. Therefore, a key priority in catchment pathogen research should be to examine the pivotal factors driving fate and transport of pathogens and to quantify their effect at all relevant scales. Quantitative data needs to be generated in a format suitable for use in existing hydrologic models. This should facilitate integration of pathogen fate and transport factors with other catchment characteristics, as well as permitting some extrapolation of experimental observations across time and scales.

Review of currently available fate and transport models for pathogens

Mathematical models are at the core of systematic catchment characterization and management programs. Models have both predictive and hypothesis testing functions. They enable knowledge to be applied generically in simulations to any

sufficiently attributed catchment. However, their output is only as good as the combination of the calibration input and the validity of the assumptions used. Model calibration needs to be undertaken locally for specific catchments.

There are numerous catchment modelling platforms available and all focus on the hydrological cycle as the basis for simulating processes within catchments relating to water yield and, in some cases, water quality. Such models commonly simulate sediment or nutrient transport. Pathogens are particles in themselves, and may become particle-associated (particularly with the fine/clay fractions). Therefore, the platforms potentially exist for modelling pathogens in catchments, but only to the extent that the behaviour of pathogens in catchments is understood.

A variety of modelling approaches can be used including; conceptual, stochastic, black box, theoretical, and deterministic or process based models. Conceptual models are designed according to a conceptual understanding of the hydraulic cycle with empirically determined functions to describe the various subprocesses (Pitt & Voorhees, 2003). Pitt and Voorhees (2003) define stochastic models as those based on the assumption that the flow at any time is a function of the antecedent component and a random component. They describe black box models as using mathematical functions that are fitted to the data without regard to the processes that they represent. This approach can be useful when little information is known about the system to be modelled. Theoretical models are written as a series of mathematical functions describing a theoretical concept of the hydrologic cycle (Pitt & Voorhees, 2003). Deterministic or process based models assume that the process being described can be defined in physical terms without any random component. The advantage of using a process based approach to model ecological systems is that the model builds on existing information that is known about the system. This information could consist of conceptual models, published literature and historic water quality and hydrological information.

Most of the existing catchments modelling platforms fall into two groups, simplified or complex hydrological models. Simple models, such as IHACRES and STARS generally require a small number of parameters, resulting in modest data requirements, and are easy to calibrate and test. While these models give a modest set

of outputs, the confidence in the estimated values can be high providing there is sufficient data available for testing, or even better, validation. Rainfall-runoff models such as IHACRES (Jakeman, Littlewood & Whitehead, 1990) can be used to fill in gaps in datasets, as well as extending the period of available stream flow. The low number of parameters required by these models allows parameter values to be used on a regional basis (Post & Jakeman, 1996). This allows the models to be applied to ungauged catchments and catchments with little data available.

The more complex hydrologic-based models have been developed to simulate and predict the volume and movement of sediments and/or nutrients and often incorporate a Geographical Information System (GIS) framework. For example, the Spatially Explicit Delivery MODel (SEDMOD) (Fraser, Barten & Pinney, 1998; Fraser, Barten & Tomlin, 1996) and the USEPA Better Assessment Science Integrating Point and Non-point Sources (BASINS) model both require input of parameters that include flow-path hydraulic roughness, gradient, slope shape, stream proximity, and a normalized soil moisture index. Similarly, the BASINS Non-point Source Model (NPSM) is a planning-level catchment model that integrates both point and non-point sources. It is capable of simulating non-point source runoff and associated pollutant loadings, accounting for point source discharges, and performing flow and water quality routing through stream reaches and well-mixed reservoirs. Both models are GIS-based systems developed by USEPA to assess water quality in catchments. BASINS addresses three objectives: (1) to facilitate examination of environmental information, (2) to provide an integrated catchment and modelling framework, and (3) to support analysis of point and nonpoint source management alternatives. It was originally released in September 1996, but regularly updated and provides easy-to-use planning- level catchment modelling. BASINS uses most of the simulation capabilities of the Hydrologic Simulation Program – FORTRAN (HSPF) which was also released by USEPA, yet use of default settings may provide relatively rough estimates of pathogen behaviour.

HSPF comprises a set of computer codes that can simulate the hydrologic, and associated water quality processes on pervious and impervious land surfaces, and in streams and well-mixed impoundments. Generally HSPF is a very powerful and comprehensive modelling environment dealing with non-point source. In order to

benefit from these models, however, appropriate parameters must be input. To simulate the removal of solids associated pathogens, such as oocysts, the wash off potency factor of oocysts, which is an intrinsic transport characteristic, must be known prior to running the model. For example, full understanding of oocyst transport characteristics would be the basis to applying HSPF to Cryptosporidium. Furthermore, the attenuation rates in response to environmental variables need to be understood.

Pioneering studies of pathogen and indicator modelling have been undertaken, with some success. For example, Jenkins et al. (1984) developed a process-based model to predict bacterial levels in upland catchments in the United Kingdom. The model utilizes a “mass balance” approach to describe bacterial sources and transport in catchments as a multiple storage and release system (Jenkins et al., 1984). Inputs of faecal bacteria from animals are transport by rainfall and runoff processes to the stream network. The model also incorporates bacterial loss through sedimentation and decay functions. The authors identified two deficiencies in the model requiring further work. These were the lack of detailed information on bacterial characteristics associated with the transfer between the land and “channel” stores, and the need to incorporate larger and more complex catchments (to account for the effects of routing contaminants downstream).

A process-based manure source and runoff model (MWASTE) was developed by Moore et al. (1989). The model calculates faecal coliform and faecal streptococci concentrations in runoff from land-applied waste for a variety of animal species and utilising a range of management techniques. The hydrologic data file is calculated using a field-scale model developed to evaluate Chemicals, Runoff, and Erosion from Agricultural Management Systems (CREAMS). The model inputs include animal species, type, and numbers and manure management factors such as collection frequency, spread duration, field size, buffer strip width, spread volume, and soil pH. MWASTE must be run individually for each animal species, but results can be superimposed to calculate runoff for fields that have manure spread from more than one species. The model incorporates bacterial loss from inactivation and retention by buffer strip entrapment. However, only part of the model was tested using a sensitivity analysis.

A probabilistic model developed by Walker et al. (1990) uses Monte Carlo simulation to combine selected deterministic relationships with statistical information on rainfall and temperature, to predict faecal coliform bacteria concentrations in surface runoff. The model (COLI) predicts minimum and maximum bacteria concentrations in runoff resulting from a storm presumed to occur immediately after manure is applied to the land. The model includes parameters to describe the characteristics of the catchment, manure management practices, bacteria characteristics, soil properties, seasonal effects, and waste management practices. The runoff component is calculated using the modified universal soil loss equation (MUSLE). Subroutines are used to calculate hourly temperature to predict bacterial decay and if the period of decay is greater than one day, the manure is assumed to be stored and the decay calculation is modified to include air temperature. The model includes separate manure management classes (surface application, subsurface application, pasture deposition, and no manure) and the yield for a single storm event is the sum of the yields from each of the four area classes. The model was designed as an evaluation tool for the assessment of various manure management strategies and therefore does not include inputs from other sources or predict the impact on receiving water quality.

Fraser et al. (Fraser, Barten & Pinney, 1998) used a GIS-based hydrologic model (SEDMOD) to estimate the load of faecal coliforms in streams for several sub- catchments of the Hudson River in the state of New York. The model characterised five key transport parameters; flow-path hydraulic roughness, slope gradient, slope shape, stream proximity, and a normalized soil moisture index. Using these parameters it was possible to predict the relative contribution of varied livestock operations in a heterogeneous landscape. A combination of temperature, turbidity and SEDMOD- predicted faecal coliform delivery, accounted for 80% of the variation in faecal coliform discharge between sub-catchments (Fraser, Barten & Pinney, 1998). Other models, such as HSPF, also now include faecal coliform modelling components. However, there is a limited correlation between pathogens and indicators such as faecal coliforms (Ashbolt, Grabow & Snozzi, 2001). Furthermore, Fraser et al. (Fraser, Barten & Pinney, 1998) noted that the application of their model in the field was limited by a number of poorly understood factors including:

• the assumption of steady state conditions which did not allow for variation in environmental conditions, e.g. rainfall (which in many catchments dominates pathogen loadings); • transport variables were linearly combined to estimate delivery potential whereas in the field these variables can interact in complex ways; • stream processes such as settling, mortality and entrainment were not included; • background contamination by native and feral animals was not evaluated; and • the model assumed that animal livestock deposited faecal material uniformly over the pastures and did not account for either spatial or temporal variation in faecal deposition.

Walker and Stedinger (1999) developed a model that accounted for pathogen loading from diffuse pollution providing a reasonably comprehensive oocyst attenuation model to predict Cryptosporidium concentrations in the raw water supplied to New York City from the Catskill-Delaware catchment. They based their model on a generalized watershed loading function (GWLF) model (Haith & Shoemaker, 1987) utilising first-order decay functions to estimate oocyst decay in manure and in water. However, there were a number of assumptions made in the calculation of this model for which data was not available. For example, the assumption that desiccation does not effect new or intermediate manure and that oocyst numbers will decrease by 25% per day in old manure, are not validated. The model also assumed that oocysts travel with manure, but the authors noted that if oocysts were easily freed from manure due to runoff processes, then the estimates of oocyst concentrations in the surface waters would be too low (Walker & Stedinger, 1999). The authors also noted that the model did not account for surface entrapment and filtration processes that can significantly reduce pathogen concentrations in farm runoff.

A study in the Netherlands modelled the discharge of parasitic protozoa into surface water and the dispersion into rivers and streams using an emission model (PROMISE) and a dispersion model (WATNAT) (Medema & Schijven, 2001). The use of these models combined observational occurrence data and experimental data from

laboratory studies into a single integrated description. The input data required for the PROMISE model included the average excretion of (oo)cysts per inhabitant per year, the total number of inhabitants and the removal of Cryptosporidium and Giardia by biological wastewater treatment. In the dispersion model, WATNAT, it was assumed that parasites were transported in surface water by advection, like other small particles. The input data for the WATNAT model required input from discharges of treated and untreated domestic wastewater and the concentration of Cryptosporidium and Giardia in raw and treated wastewater. Survival data for Cryptosporidium and Giardia derived from the literature enabled the authors to incorporate inactivation rates into the model. However, they noted that although sedimentation and re-suspension of attached (oo)cysts were also likely to be important factors in (oo)cyst dispersion and transport, they could not be included in the model because too few quantitative data were available. The concentration of Cryptosporidium and Giardia in surface water that was calculated by the combination of these models was close to the observed concentration at five of the six surface water sites. The site where the model predicted lower concentrations than were observed was influenced by agricultural runoff, which was not incorporated into the model. The authors note the importance of incorporating non- point pollution into the model and highlight that the data required to do this include:-

• the occurrence of Cryptosporidium and Giardia in manure from livestock in the Netherlands; and • quantitative information regarding transmission routes of (oo)cysts from manure to surface water (runoff and soil infiltration).

Steets and Holden (2003) developed a mass balance-based, mechanistic model to describe the fate and transport of faecal coliforms in an estuarine coastal lagoon. The model predicts concentrations in the water column, sediments and receiving waters for summer and winter conditions. The approach of the model is mechanistic and generic but the model itself contains parameters that are site and season specific, facilitating the ability of the model to be modified to apply to different geographical areas. The model includes decay functions and sedimentation and resuspension parameters. The model indicated that in summer dry flow conditions allowed the lagoon to function as a retention basin for faecal coliforms entering from the creek. However during winter the lagoon amplified faecal coliform concentrations due to resuspension of faecal coliforms 23

from the sediments and this caused increased concentrations in the waters downstream of the mouth of the lagoon. The model was tested using field data and results showed the predicted concentrations internal to the lagoon were within an order of magnitude. Some differences in the predicted and observed concentrations within the lagoon were possibly due to overestimation of faecal coliform inactivation (k values) or underestimation of faecal coliform advective dispersion. The model was also evaluated using sensitivity analysis indicating that the under-predictions of summer faecal coliform concentrations were likely due to errors in k estimates, as k was the most sensitive parameter for this seasonal condition. The winter model simulation was highly sensitive to resuspension values, and was much less affected by k values.

Tian et al. (2002) developed a transport model for the potential contamination of waterways with E. coli from animal-grazed farmlands. The model is GIS- and process- based, with spatial features that include; grazing location, land topography, distance to nearby streams and distance through the stream network to the catchment outlet. The temporal features of the model include population dynamics on the land surface, in flow and on streambeds. The model uses a mass balance approach to examine grid cells on the land surface and in networked stream segments. The model consists of six components: non-point source dynamics and distribution, attenuation processes, diffuse runoff, point sources, transport via runoff to nearby streams, and in-stream mobilisation. The hydrologic basis of the model is the GIS- based Watershed Assessment Model (WAMView) (Bottcher & Hiscock, 2001). The WAMView model provides estimates of surface runoff from individual land units as well as the stream flow within the catchment. The model includes many of the processes that determine the origin, fate and transport of microorganisms from diffuse pollution, particularly the attenuation and diffusion processes. However, the sources of E. coli were derived using a standardized stocking unit (SU) that is not fully explained. The authors comment that the model does not account for inputs by direct animal faecal deposition, inputs from wildlife or seasonal effects, and that these components could be included in further refinements of the model.

Similarly, Collins and Rutherford (2004) also used WAM as the hydrologic basis for a model that estimates E. coli concentrations in streams draining grazing country in New Zealand. The model uses a conceptual model and mass balance

approach to incorporate inputs from seepage zones and direct faecal deposition from domestic livestock (but not wildlife). The model also includes decay rates to estimate die-off of E. coli and transport within surface runoff is modelled in two stages. The first stage estimates washout or detachment from faecal deposits and the second stage uses a delivery index to estimate transport to the stream network. The components of the delivery index were; proximity to stream, slope and flow accumulation, and the index was used in conjunction with the volume of surface runoff generated. Subsurface transport to the stream network was included using a percolation index, ranging from 0 to 0.1. Only percolation generated within 50 m of a stream is capable of delivering E. coli to that stream. The model was tested using both sensitivity analysis and with water quality data. The delivery index, particularly the proximity to streams was identified as an important parameter in the model, as was the proportion of bacteria that were excreted directly onto seepage zones. Simulations indicated that when cattle have access to streams, varying the rate of direct deposition between 1 and 15% had a significant impact on the predicted concentrations of E. coli. The study also included scenario analyses to estimate the potential value of riparian buffer zones.

A multiple regression approach was used by Crowther et al. (2003) to model spatial variation in faecal indicator concentrations in waterways using land use and topographic data. The model predicted presumptive coliform, E. coli and enterococci concentrations for water samples collected in a large rural catchment in Wales for both base flow and high flow conditions. The model utilised GIS land use, slope and sub- catchment morphometry data, as well as absolute and relative surface-flow distances to the sub-catchment outlet. At base flow the strongest correlations were with land use in close proximity (< 2 km) to the sub-catchment outlet. During base flow, the proportion of improved pasture within 1 or 2 km of the outlet accounted for 76.1% and 67.3% of the variance in E. coli and enterococci concentrations, respectively. At high flow, concentrations increased by an order of magnitude, and the strongest correlations were with land use in the < 5 km or < 60% relative surface-flow distance to the sub- catchment outlet. The authors noted that no improvement in the model was achieved by including slope and that loss of indicator organisms during high flow was minimal, presumably due to increased depth, velocity and turbidity of the water. They concluded that flow distance was not a key factor affecting microbial water quality during high

flow and that for catchment management to be effective, control measures need to be implemented catchment wide, and not just in the immediate vicinity of the outlet point.

Recently, Dorner et al. (2004) developed a probabilistic model for estimating the production of Cryptosporidium spp. and Campylobacter spp. from livestock in the Grand River catchment, Ontario, Canada, using parameters drawn from published studies. They used β-distributions to estimate pathogen prevalence and Г-distributions to estimate animal pathogen shedding intensity. The pathogen loads were then calculated using daily manure production rates by summing the number of pathogens shed by all positive animals. As different animal types and age classes may show varying pathogen prevalence or shedding intensities, the simulations were performed according to the animal categories defined in the 2001 Canadian Census of Agriculture. The results indicated that although cattle were responsible for the production of the largest volume of manure, other domestic livestock also contributed large numbers of both Cryptosporidium spp. and Campylobacter spp. They noted that the predicted loading rates were highly sensitive to the concentration of pathogens in animal faeces (shedding intensity). The model calculates the “input” loads from the diffuse agricultural pollution sources but does not include inputs from wildlife or the delivery mechanisms for transport to the stream network. Thus the model calculates a “maximum potential load” for the production of pathogens from domestic livestock animal sources. Further refinements of the model will include the inputs from sewage treatment plants (Dorner, Huck & Slawson, 2003) and will involve linking the model to the established hydrologic model WATFLOOD.

Three alternative approaches were assessed by Vinten et al. (2004) to predict the delivery of E. coli from livestock sources to bathing waters in Scotland. They included a lumped soil transport model, a regression model using water quality data and a distributed catchment model (PAMIMO-C). The input to all three models was the soil E. coli pool which was supplemented by grazing animals and slurry, and depleted by inactivation (calculated using two pool first-order inactivation kinetics). The results indicated that all of the models tended to under predict the delivery of E. coli to the receiving waters, especially during base flow conditions. The regression model showed the best fit to the data, followed by PAMIMO-C. The lumped soil transport model

severely under predicted E. coli concentrations, as it did not account for surface flow transport.

Most recently Haydon and Deletic (2005) have linked a simple conceptual mass- balance pathogen model to a stormflow-baseflow model to predict pathogen concentrations in stream flows from catchments. The model has been used to predict E. coli concentrations for baseflow and stormflow conditions in two Australian catchments. The model utilizes a pollutant buildup and washoff module to estimate pathogen deposition, storage, movement and decay within a pathogen store. Deposits into the store are a function of land use, and evapotranspiration is used to calculate loss from the store (accounting for the effects of temperature and moisture). Movement out of the store is a function of respective flows taken from the hydrologic model, and the pathogen level of the store. The model can be run on a daily time step, but gives a better estimate of E. coli concentrations during stormflow if run with an hourly time step. Further refinements of the model will include improvements to the load estimations and the use of Bayesian techniques to reduce the levels of uncertainty in the model.

Table 1.5 summarises each of the described catchment models, highlighting their strengths and weaknesses and their potential usefulness for predicting pathogen fate and transport.

Table 1.5 Summary of catchment models

Model Advantages Constraints Applicability IHACRES Few input parameters required Model output is dependent on good Rainfall-runoff model that can be Easy to calibrate and test quality rainfall and stream data. Several applied locally to catchments with Can be used on a regional basis catchments with stream data are needed good datasets, or for for regionalisation ungauged/poorly gauged catchments through regionalisation STARS Few input parameters required Requires calibration – needs sufficient Estimating sediment and chemical Easy to calibrate and test water quality data (eg salt) exports BASINS Can calculate runoff and pollutant loadings A simple approach that uses USA data A planning level catchment model from point and non-point sources through where quality assurance is suspect in that can integrate both point and stream reaches and reservoirs; uses GIS some cases. User-friendly tools may give non-point sources of pollutants approach and focused on TMDL estimations rise to inappropriate use of output data HSPF Existing versions of this model include faecal Requires large amounts of data for a wide Only applicable to catchments with coliforms range of input parameters comprehensive existing GIS and Assumes that pathogens are transported hydrological information that can be with particles but the relationship is not integrated with water quality data quantified (Jenkins et al., Simple process-based model for predicting Not able to use GIS data Predicts faecal coliform 1984) indicator concentrations in water and No routing mechanism concentrations in water and sediment from diffuse sources Model not validated sediment in small upland catchments MWASTE Predicts indicator concentrations in surface Not able to use GIS data Predicts faecal coliform and faecal runoff from manure / agricultural areas No routing mechanism streptococci concentrations in Has the ability to run various management Only includes indicator bacteria surface runoff from agricultural scenarios Only includes manure sources areas

Table 1.5 (cont’d.) Summary of catchment models

Model Advantages Constraints Applicability COLI Predicts faecal coliform concentrations in Not able to use GIS data Predicts faecal coliform surface runoff from manure / agricultural No routing mechanism concentrations in surface runoff areas Only includes indicator bacteria from agricultural areas for single Has the ability to run detailed management Only includes manure sources storm events scenarios and includes catchment / land use features SEDMOD Interface with a geographical information Requires large amounts of data for a wide Useful for predicting the relative system (GIS) platform facilitates data input range of input parameters contribution of diverse livestock Can prioritise relative contributions of non- operations within a variety of land point source pollutants use types GWLF Few input parameters required Some assumptions may not be valid e.g. Can be utilised to predict pathogen decay and transport coefficients for loads utilising first order decay pathogens kinetics Estimates of pathogen loads may be too low PROMISE and Simple emission and dispersion models that Currently calibrated for regional data Currently only applicable for use WATNAT can predict the concentration of pathogens in specific to the Netherlands within the Netherlands receiving waters receiving point source Data not available to account for non- pollution point source pollution from agricultural livestock (Steets & Holden, A mechanistic model to describe the fate and Only includes indicator bacteria Able to predict water and sediment 2003) transport of faecal coliforms in a coastal Only inputs are diffuse pollution concentrations of faecal coliforms lagoon from diffuse sources dispersed into Accounts for advective flow, dispersion, estuarine lagoons decay and sedimentation and resuspension

Table 1.5 (cont’d.) Summary of catchment models

Model Advantages Constraints Applicability Tian et al. (2002) Interface with a geographical information Only models E. coli bacteria Only applicable to catchments with system (GIS) platform facilitates data input Does not include direct deposition of comprehensive existing GIS Calculates runoff and streamflow using WAM faecal material to streams, inputs from information Sensitivity analysis of various scenarios used wildlife or groundwater transport Able to estimate E. coli loads, to test variables in the model mechanisms surface runoff and streamflow (Collins & Calculates runoff and streamflow using WAM Does not include direct deposition from Estimates E. coli loads, surface Rutherford, 2004) Includes direct deposition from livestock wildlife runoff and streamflow for diffuse Has a delivery index for pathogen transport to Data not yet available for some of the pollution of rural catchments the stream network and includes subsurface parameters (e.g. rates of direct deposition) transport Includes in-stream dynamics of deposition and resuspension Sensitivity analysis and water quality validation testing of the model and scenario analyses to evaluate effectiveness of riparian buffer zones (Crowther et al., Uses catchment land use GIS data to estimate Does not include data on stocking rates or Predicts faecal indicator 2003) faecal indicator concentrations in streams point sources concentrations for streams in base Not linked to a hydrologic model flow or high flow conditions Need to have water quality data Dorner et al. Uses β-distributions to calculate pathogen Does not yet include inputs from wildlife, Probabilistic model to estimate (2004) prevalence in domestic animals and Г- STPs, or on-site systems. Does not yet maximum pathogen loads generated distributions to estimate pathogen shedding include pathogen inactivation or stream per day from domestic livestock in intensity routing. catchments

Table 1.5 (cont’d.) Summary of catchment models

Model Advantages Constraints Applicability (Vinten et al., Compares a soil transport model, regression Does not include point sources Likely to under predict microbial 2004) analysis and a distributed catchment model Does not account for salinity of receiving impacts on bathing water quality (PAMIMO-C) to predict E. coli transport to waters or sunlight intensity receiving waters Soil transport model does not allow for surface water delivery Haydon and Estimates E. coli concentrations in stream Does not give good estimates of total Simple model for the estimation of Deletic (2005) water during baseflow and stormflow loads, probably due to the simplified microorganism concentrations at conditions representation of the pathogen store peak flows from rainfall and flow component of the model data

IHACRES – Identification of unit hydrographs and component flows from rainfall, evaporation and streamflow STARS – Sediment/chemical Transport with Advection, Resuspension and Settling BASINS – Better assessment science integrating point and non-point sources model HSPF – Hydrologic simulation program FORTRAN MWASTE – model developed by Moore et al. (1989) COLI – model developed by Walker et al. (1990) SEDMOD – Spatially explicit Delivery Model GWLF – Generalized watershed loading function PROMISE and WATNOT – models used by Medema and Schjiven (2001) WAM – Watershed assessment model used by Tian et al. (2002) and Collins and Rutherford (2004)

Aims of this study

The aim of this work is to produce a mathematical model that predicts pathogen (Cryptosporidium and Giardia) and indicator bacteria (E. coli) loads at different locations in Sydney’s drinking water catchments. The model will integrate existing information with new quantitative data addressing the knowledge gaps identified in the review of the literature. The approach will be to use the conceptual model of the processes that influence pathogen sources, fate and transport within drinking water catchments as a framework for the development of the mathematical model. The model will incorporate existing data with GIS data for the Sydney drinking water catchments and with new data collected in the course of this study.

New data will be collected to quantify Cryptosporidium, Giardia and E. coli concentrations and prevalence in a range of domestic and wildlife animal faeces within the Sydney drinking water catchment. Surface water transport of Cryptosporidium, E. coli and a bacteriophage (PRD1) will be quantified using artificial rainfall simulation at field-scale. The mathematical model will be constructed using data for the Wingecarribee sub-catchment in the SCA’s area of operations. The model will be refined using sensitivity analysis, and tested using grab-sample water quality data collected from the Wingecarribee sub-catchment. The refined model will then be applied to all of the SCA’s drinking water catchments.

The model will be designed as a management tool to identify those sub- catchments that represent the highest risk to water quality and human health i.e. those sub-catchments that export the highest total pathogen loads in the closest proximity to the raw water storages. The development of such a model will enable local water utilities to identify and then prioritise sources of pathogen risks enabling the prioritized implementation of control measures. Ideally, the model will be used to predict the potential improvements in water quality that can be derived from the implementation of control measures. This can subsequently be verified by collecting local quantitative data, which can also be used to calibrate and refine the model as an iterative process. This tool will enable water utilities to assess the relative importance of the many sources of pathogens in their catchments and to proactively manage microbial risks to drinking water quality.

The model will predict the total pathogen units (TPU budget) arising and exported from each drainage unit per day in dry and wet weather conditions. The ultimate aim will be to further refine the model to predict the proportion of the TPU budget that represents an actual risk to human health. That is the proportion of the TPU that is comprised of viable, infectious and virulent strains of each microorganism (infectious pathogen unit, or IPU budget). This information would be a necessary input to a quantitative microbial risk assessment (QMRA) of the drinking water supply.

Structure of dissertation

Following the development of the conceptual model a systematic review of the literature was undertaken to determine existing data related to the quantification of pathogen sources (Chapter 2) and the mechanisms that govern their transport in surface waters (Ferguson et al., 2003b). New data quantifying the concentrations of Cryptosporidium, Giardia and E. coli found in animal faeces and sewage effluent in the Sydney drinking water catchments are described in Chapter 3. Quantification of field- scale transport of Cryptosporidium, E. coli and PRD1 bacteriophage in surface water using artificial rainfall simulations are described in Chapter 4. The construction of the mathematical model using the data from the previous chapters and from the Wingecarribee sub-catchment is shown in Chapter 5. Chapter 6 describes the refinement of the model using sensitivity analysis, comparison of model outputs with water quality data and the evaluation of molecular source tracing tools for use in discriminating pollutant sources in the Wingecarribee. Application of the refined model to all of the SCA catchments is described in Chapter 7, followed by a general discussion of the model (Chapter 8) and a reference section. Appendix 1 summarises the subcatchment data required to run the model for all of the SCA catchments.

Chapter 2 Pathogen sources in drinking water catchments

Introduction

The conceptual model (Figure 1.1) identified the processes responsible for the origin, fate and transport of pathogens in drinking water catchments. Most faecal indicators are generic to warm-blooded animals. As a result, humans, wildlife populations and domestic livestock all contribute indicator organisms to surface waters. Some studies have shown that there is a high level of diversity of faecal indicator strains isolated from environmental samples (McLellan, Daniels & Salmore, 2003; Renter et al., 2003). Therefore, an important information need for catchment managers is the likely source of indicators. If the indicators have arisen from human faecal waste, the risk for any particular indicator concentration is higher than from an animal source. Indicators from domestic animal waste present less risk for the same indicator load whilst indicators from wildlife are likely to indicate less risk still. False alarms can arise where indicators are found from wildlife sources and are thought to be arising from human sources. For example, enteroviruses were found in the Sydney catchment but were typed and found to be bovine enteroviruses (BEV) rather than human enterovirus strains, indicating cattle-derived rather than human-derived pollution (Rothwell et al., 2004). The difficulty with distinguishing the source of faecal indicator organisms can be alleviated by examining samples for selected index pathogens in combination with indicators or by utilizing tracing and tracking tools that are able to identify the origin of environmental isolates.

Natural processes – inputs from wildlife

The primary natural process that contributes pathogens to catchments is the excretion of faecal material from host animals. The natural processes that mitigate the survival and dispersion of these pathogens include environmental inactivation, soil retention, and riparian vegetation (Figure 1.1).

The term wildlife is used to describe both native and feral free-living species of animals. Native animal species are defined as those free-living fauna that are indigenous to a specific geographical area, usually a continent or island. Frequently these species are protected and cannot be kept by humans as companion animals. Feral species can be defined as those free-living fauna species that are not native to a specific geographical area and which are not confined or managed as domestic animals. These species are frequently regarded as pest species that damage native flora and fauna and can be the subject of hunting and culling activities. Within Australia the legal culling of native animals is also allowed for some species, such as kangaroos, due to their abundance in certain regions.

Both domestic livestock and wildlife can carry pathogens that infect humans and have been identified as the cause of waterborne disease outbreaks (Hrudey & Hrudey, 2004). Indeed, the majority of recently emerging pathogens of human significance are zoonotic, that is, they are capable of infecting multiple host species (Woolhouse, 2002). In general the virulence of pathogens for humans and the prevalence in the host animals are both lower in wildlife than domestic livestock animals. Understanding what pathogens are likely to arise from particular animal sources, and what the likely infectivity would be, helps in identifying risks and control strategies. Our collective understanding of the relationship between pathogen species, strains, serotypes and genotypes, and the importance of specific virulence determinants on the one hand, and the host range and infectivity on the other, is rapidly evolving. Therefore, water utilities and catchment managers need to maintain a watching brief for emerging knowledge relating to pathogen risks to water supplies. In all catchments wildlife animals will host and excrete a range of pathogens and indicator organisms in their faeces. The potential impact of this material on the quality of surface waters is determined by the following factors.

Population density

Animal density by area is an important determinant of pathogen loadings. Higher animal density results in a larger volume of manure excreted per unit area. Thus there is an increase in pathogen source material that may be transported in runoff to

surface waters and/or deposited directly to streams. From an epizoological perspective, as animal density and population size increases, a threshold is reached where the circulation of a particular pathogen within that animal reservoir can become sustainable over time.

It is difficult to quantify animal densities for wildlife because animal movement is uncontrolled and because animal populations vary with season and environmental conditions, with many species being migratory. As well as being difficult to estimate animal density it is also difficult to estimate host pathogen prevalence since signs of infection and disease can be difficult to assess (Artois et al., 2001). Human population expansion has encouraged the emergence of infectious diseases due to increasing population density especially in urban areas, combined with human encroachment into wildlife habitat. Human destruction of habitat contributes to increased density of wildlife in the remaining areas. Ecological techniques such as count estimation and scat analyses are the approaches most commonly used to estimate wildlife populations, although in the absence of such locally acquired data, published estimates of wildlife populations can be used for similar ecosystems.

Although it is difficult to estimate animal abundance it is known that feral cats are distributed across the continent, occupying all habitats (Ramsay, 1994). Densities in favourable habitats are estimated at one per square kilometre (Ramsay, 1994). In farmland regions in New Zealand they can be as high as 3.5 cats per square kilometre (Langham & Charleston, 1990). Kangaroo populations in native vegetation areas can range from 200 to 500 animals per square kilometre (M. Roberts pers. comm.). For this study a mean density of 200 will be used for land use areas designated as native vegetation. Pigs are probably the most significant contributors among feral animals due to their large body size, significant populations, and habit of foraging and wallowing in and around water bodies. The population has been estimated at around 1 pig.km-2 in the protected catchment zone (SCA Pest Control Group pers. comm. and (Australian Water Technologies, 2002a)). Other feral animals of (economic) consequence in the catchment include rabbits, foxes, goats, dogs and deer (Australian Water Technologies, 2002a). Estimates of wildlife animal population densities for native vegetation land use areas in the SCA’s area of operations are given in Table 2.1.

Table 2.1 Estimates of wildlife animal density for native vegetation land use areas

Animal Country Density (per km2) Reference Kangaroos Australia 200 M. Roberts pers. comm. Marsupials Australia 20 D. Ashton pers. comm. Pigs (feral) Australia 1 D. Ashton pers. comm. Deer Australia 0.5 D. Ashton pers. comm. Rodents UK Similar to human (O'Keefe et al., 2003) population Rabbits Australia 50 D. Ashton pers. comm. Foxes Australia 1 D. Ashton pers. comm. Dogs (feral) Australia 0.25 D. Ashton pers. comm. Cats (feral) Australia 1 (Ramsay, 1994) Cats (feral) UK 0.1 - 0.33 (O'Keefe et al., 2003)

Volume of manure

Manure excretion rates and volumes for wildlife are much less well documented than for domestic animals. It is likely that even within the same geographical area wildlife will excrete less than the equivalent domestic species due to the lack of supplemental feeding and the inferior quality of feed. To assess the impact of wildlife manure it is necessary to know how often wildlife species defecate and the volume of manure produced per day. Many wild animals use defecation as a means of demarcating territory (Clough et al., 2003) and this affects the distribution of manure and the likelihood that it will reach surface waters. Estimates of manure production rates for wildlife are shown in Table 2.2.

Table 2.2 Manure production rates for wildlife animals

Animal kg manure Reference .animal-1.d-1 Kangaroos 0.2 M. Roberts pers. comm. Pigs (feral) 6.2 (Australian Water Technologies, 2002a) Deer 1 Similar to sheep (Australian Water Technologies, 2002a) Rabbits 0.078 (Medema, 1999) Dogs (feral) 0.5 (Australian Water Technologies, 2002a)

Pathogen prevalence and shedding intensity

A wide variety of pathogens can be carried and excreted by domestic animals and wildlife, however not all of them are a risk to surface water quality. The major bacterial, protozoan and viral pathogens of risk to surface water quality are listed in Table 2.3.

There is little data on native and feral animal concentrations and/or prevalence that are suitable for giving accurate estimates of mean pathogen concentrations and there is usually a high level of variance. It is therefore recommended that both published data and locally acquired data be reviewed together in deriving estimates of pathogen concentrations in animal faeces. Geographical variation in types of animal species and environments (catchment characteristics) will influence the risk that animal species present to surface water quality in different parts of the world. Also, the interaction of domestic animals with wildlife may significantly impact on the concentration, type and strain of pathogens carried and excreted by wildlife populations. In the United States studies have shown that up to 45% of beaver and 48% of muskrats are shedding Giardia cysts (Hibler & Hancock, 1990). Beaver have frequently been implicated as a source of Giardia contamination for surface waters (Cole, 1990). However, it is not known whether increases in the number of reported cases of giardiasis in backcountry visitors are due to better diagnosis or reflect increased levels of surface water contamination (Cilimburg, Monz & Kehoe, 2000).

Table 2.3 Pathogen hazards arising from animals in surface water catchments

Hazards Common sources Occasionally Not known to be sources sources Bacteria Salmonella spp, Birds and mammals Reptiles Campylobacter spp Yersinia Pigs Reptiles enterocolitica E. coli O:157 Cattle ? ? M. bovis Cattle ? ? M. avium Birds Parasites Cryptosporidium Placental mammals, Birds (indirectly Marsupial mammals, parvum genotype II rodents via eating monotreme mammal faeces) mammals, reptiles, Giardia Most placental Birds (Upton) Reptiles, marsupial mammals, mammals, including beavers monotreme mammals Toxoplasma gondii Most mammals Birds, reptiles

Viruses Hepatitis E Pigs Other mammals, birds, reptiles Indicators E. coli All mammals and Reptiles? ? birds Enterococci, All mammals Reptiles? ? Streptococci Clostridia Dogs All mammals ? and birds

A Canadian study of Giardia spp. and Cryptosporidium spp. prevalence in humans, wildlife and agricultural sources found that wildlife had the lowest prevalence of Giardia spp. (3.28%) and Cryptosporidium spp. (0.94%) (Heitman et al., 2002). Sewage influent had the highest prevalence of Giardia cysts (48.8%) and Cryptosporidium oocysts (5.42%) however, the highest concentrations of both parasites occurred in domestic livestock manure (calf and dairy cattle faeces respectively). Although the overall prevalence of each parasite was extremely low, the prevalence of Giardia spp. in aquatic mammals (beaver and muskrat) was quite high. This suggests a potential for aquatic mammals to become infected by human and agricultural contamination (Heitman et al., 2002). As animal reservoirs, infected beaver and

muskrat could amplify background levels of Giardia and Cryptosporidium spp. in catchment waterways. Interestingly, the Cryptosporidium strain isolated from the beavers closely resembled the pattern displayed by the “cattle” genotype (Heitman et al., 2002). They concluded that although wildlife as a whole did not appear to be an important source of river contamination the impact of aquatic mammals was in need of separate assessment.

An Australian study of pathogen concentrations in domestic and wildlife animal faeces found that concentrations were generally higher in domestic animal faeces than in native and feral species, probably due to higher animal density and domestic animal confinement. For example, the mean concentration of Cryptosporidium in wild swine faeces was 0 oocysts.g-1 while the mean concentration in domestic pigs was 36 oocysts.g-1 (Australian Water Technologies, 2002b). Supporting this theory, a study of feral pigs in California by Atwill et al. (1997) showed that once a threshold or density was exceeded, the risk of shedding Cryptosporidium by individual feral pigs was significantly higher (p=0.003); no pigs from low density populations were shedding C. parvum (<1.9 pigs.km2) compared to 9 to 10% of pigs from high density populations (>2 pigs.km2). Giardia cysts were shed by 6 to 8% of feral pigs regardless of age, indicating that the risk of surface water contamination with cysts will be a function of climatic variables and animal behaviour within the riparian zone.

A summary of published Cryptosporidium, Giardia and E. coli concentrations in wildlife faeces are shown in Table 2.4, Table 2.5 and Table 2.6 (respectively). All values are per gram wet weight of faeces.

Table 2.4 Concentrations of Cryptosporidium spp. in faeces of wildlife animals

Animal Country n Prevalence Mean Reference % oocysts .g-1 (range) (range) Eastern Grey Australia 2340 5 – 31.5 (10 – 2 x 107) (Power et al., 2004) Kangaroo Eastern Grey Australia 3557 6.7 - Power pers. comm. Kangaroo Eastern Grey Australia 25 32 204 (Davies et al., Kangaroo (0 – 3263) 2005b) Bison (ranch) Canada 41 4.9 2369† (Heitman et al., 2002) Pigs (feral) USA 221 5.4 - (Atwill et al., 1997) Pigs (feral, <8 USA 62 11 - (Atwill et al., 1997) mo) Pigs (feral, >9 USA 159 3 - (Atwill et al., 1997) mo) Horses Canada 1 0 0 (Heitman et al., 2002) Deer Canada 649 0.15 12† (Heitman et al., 2002) Deer (fallow) UK 16 6 3000 (Sturdee, Chalmers & Bull, 1999) Deer (muntjac) UK 42 10 3000 (Sturdee, Chalmers & Bull, 1999) Deer (white- USA 34 8.8 - (Rickard et al., tailed, <6 mo) 1999) Deer (white- USA 360 5.0 - (Rickard et al., tailed, >6 mo) 1999) Elk (wild) Canada 34 0 0 (Heitman et al., 2002) Elk (ranch) Canada 38 4.9 3742† (Heitman et al., 2002) Moose Canada 177 0 0 (Heitman et al., 2002) Hedgehog UK 4 25 3000 (Sturdee, Chalmers & Bull, 1999) Porcupine Canada 8 0 0 (Heitman et al., 2002) Beaver Canada 334 2.4 509† (Heitman et al., 2002) Beaver USA 87 0 - (Zhou et al., 2004a) Otter USA 20 0 - (Zhou et al., 2004a)

Table 2.4 (cont’d) Concentrations of Cryptosporidium spp. in faeces of wildlife animals

Animal Country n Prevalence Mean Reference % oocysts .g-1 (range) (range) Rodents (mice) UK 300 24 - (Bodley-Tickell, Kitchen & Sturdee, 2002) Rodents (mice) UK 242 22 - (Chalmers et al., 1997) Rodents (wood UK 230 21 - (Chalmers et al., mouse) 1997) Rodents (mice UK 485 32.8 4.4 x 104 (Sturdee et al., and wood mice) SD 145.5 2003) Rodents (mice) Canada 2 0 0 (Heitman et al., 2002) Squirrel UK 8 0 - (Sturdee, Chalmers & Bull, 1999) Squirrel Canada 15 0 0 (Heitman et al., 2002) Squirrel USA 309 16 5.4 x 104 (Atwill et al., 2001) Muskrat Canada 23 0 0 (Heitman et al., 2002) Muskrat USA 237 11.8 - (Zhou et al., 2004a) Shrew UK 20 35 (3000 – (Sturdee, Chalmers 2.5 x 104) & Bull, 1999) Pygmy shrew UK 10 10 1.0 x 104 (Sturdee, Chalmers & Bull, 1999) Bank vole UK 123 13 - (Chalmers et al., 1997) Barn animals UK 940 29.9 4.7 x 104 (Sturdee et al., (mice, voles and SD 126.6 2003) shrews) Racoon USA 51 3.9 - (Zhou et al., 2004a) Racoon USA 100 13 - (Snyder, 1988) Rabbits Neth 31 6.5 1200 (Medema, 1999) Rabbits UK 28 7 3000 (Sturdee, Chalmers & Bull, 1999) Hare Canada 453 0 0 (Heitman et al., 2002) Hare (Brown) UK 2 0 - (Sturdee, Chalmers & Bull, 1999) Badger UK 26 15 3000 (Sturdee, Chalmers & Bull, 1999) Weasel UK 1 0 - (Sturdee, Chalmers & Bull, 1999)

Table 2.4 (cont’d) Concentrations of Cryptosporidium spp. in faeces of wildlife animals

Animal Country n Prevalence Mean Reference % oocysts .g-1 (range) (range) Stoat UK 2 0 - (Sturdee, Chalmers & Bull, 1999) Fox Neth 2 0 0 (Medema, 1999) Fox USA 76 7.9 - (Zhou et al., 2004a) Fox UK 23 9 3000 (Sturdee, Chalmers & Bull, 1999) Coyote Canada 99 0 0 (Heitman et al., 2002) Wolf Canada 2 0 0 (Heitman et al., 2002) Polecat UK 2 0 - (Sturdee, Chalmers & Bull, 1999) Dogs (feral) USA 59 3.8* - (Hackett & Lappin, 2003) Mallard duck Neth 168 0 0 (Medema, 1999) Geese (Canada) Canada 57 0 0 (Heitman et al., 2002) Geese (Canada) Canada 209 23.4 - (Zhou et al., 2004b) Geese (Canada) Canada 209 2.4‡ - (Zhou et al., 2004b) * Study was conducted on domestic dogs, no available estimates for feral dogs † Mean of positive samples ‡ Proportion of all samples that contained either C. parvum or C. hominis

Table 2.5 Concentrations of Giardia spp. in faeces of wildlife animals

Animal Country n Prevalence Mean Reference 1 (%) cysts .g- Bison (ranch) Canada 41 14.6 2649† (Heitman et al., 2002) Pigs (feral) USA 221 7.6 - (Atwill et al., 1997) Pigs (feral, <8 mo) USA 62 6 - (Atwill et al., 1997) Pigs (feral, >9 mo) USA 159 8 - (Atwill et al., 1997) Horse Canada 1 0 0 (Heitman et al., 2002) Deer Canada 649 0.15 1168† (Heitman et al., 2002) Deer (white-tailed, USA 34 2.9 - (Rickard et al., 1999) <6 mo) Deer (white-tailed, USA 360 1.1 - (Rickard et al., 1999) >6 mo) Elk (wild) Canada 34 0 0 (Heitman et al., 2002) Elk (ranch) Canada 38 15.8 1665† (Heitman et al., 2002) Moose Canada 177 0.6 168† (Heitman et al., 2002) Porcupine Canada 8 0 0 (Heitman et al., 2002) Beaver - 45 (Hibler & Hancock, 1990) Beaver Canada 334 8.7 1654† (Heitman et al., 2002) Rodents (mice) Canada 2 0 0 (Heitman et al., 2002) Squirrel Canada 15 0 0 (Heitman et al., 2002) Muskrat - - 48 - (Hibler & Hancock, 1990) Muskrat Canada 23 78.3 9574† (Heitman et al., 2002) Rabbits Neth 31 0 0 (Medema, 1999) Hare Canada 453 0 0 (Heitman et al., 2002) Fox Neth 2 0 0 (Medema, 1999) Coyote Canada 99 5 1577† (Heitman et al., 2002) Wolf Canada 2 0 0 (Heitman et al., 2002) Dogs (feral) USA 59 5.4* - (Hackett & Lappin, 2003) Mallard duck Neth 168 24 4.4 x 105 (Medema, 1999) Geese (Canada) Canada 57 0 0 (Heitman et al., 2002) * Study was conducted on domestic dogs, no available estimates for feral dogs † Mean of positive samples

Table 2.6 Mean concentrations of faecal coliforms and E. coli in faeces of wildlife animals

Animal n Faecal coliforms E. coli Reference 1 1 cfu .g- cfu .g- Eastern Grey 25 - 5.8 x 105 (Davies et al., Kangaroo SD 5.3 x 105 2005b) Rodents - 3.3 x 105 1.2 x 105* (Geldreich, 1978) (mice) Chipmunk - 1.48 x 105 5.6 x 104* (Geldreich, 1978) Rabbits - 20 16* (Geldreich, 1978) Duck - 3.3 x 107 8.0 x 106 (Geldreich et al., 1962) Duck - - 3.3 x 107 (Jones & White, 1984) Geese - 3.2 x 104 -1 x 106 - (Weiskel, Howes & Heufelder, 1996) Seagulls - - 1.3 x 108 (Jones & White, 1984) * Converted from faecal coliform count using equation 1

Animal age and behaviour

Animal age will affect the concentration and shedding rate of many pathogens. The ratio of adult to juvenile animals may significantly affect the total load of pathogens generated, with juvenile animals frequently shedding much greater pathogen loads. The frequency of animal defecation is also dependant on the type of pathogen infecting the host. Some pathogens such as Campylobacter or E. coli O157 may not cause clinical symptoms in the host, such that there are probably no differences between infected and uninfected hosts. However other pathogens such as Salmonella spp. can cause diarrheal resulting in increased frequency of defecation in infected animals. Native animals may host less virulent strains of pathogens, may be less likely to defecate in and around streams and be less likely to damage riparian areas. For example, cattle are far more hazardous than kangaroos in terms of all of these factors (Australian Water Technologies, 2002b; Power et al., 2001).

Animal behaviour significantly affects the impact of wildlife activities on surface water quality. The availability and location of drinking water and types of vegetation can determine the amount of time spent, and thus the amount of faecal material deposited in the riparian zone. The ease of access to waterways affects the extent of direct deposition of faecal material in waterways. For example, wild swine frequently inhabit the riparian zone, defecating directly into the water and damaging stream banks and channels, while kangaroo species spend little time in the riparian zone. The presence of competing animals and/or predator animal species will influence animal range and territory and in particular may limit the time that some species spend at watering holes.

Zoonotic transfer

Wildlife species are capable of becoming infected with human-infectious pathogens originating from domestic animals. In gathering information on catchments, sites where wildlife can interact with humans and domestic animals, or their waste, are potential sites for pathogen transfer. These can include sewage disposal areas, rubbish tips, manure spreading areas and stock grazing areas near to nature reserves.

A study by Bull et al. (1998) in the UK showed a significant difference in the species and prevalence of Cryptosporidium in bank voles. On Skomer Island, where there were no domestic livestock, 51% of live-trapped animals were positive for Cryptosporidium species and 85% were shedding Cryptosporidium muris. However, on the mainland the prevalence of Cryptosporidium was one-quarter of that on Skomer Island and the species normally found was Cryptosporidium parvum. The authors suggested that the prolonged absence of livestock on Skomer Island had allowed the host parasite interaction to reach its natural balance, which supports the view that small rodents are the main and preferred host of C. muris but not for C. parvum (Bull et al., 1998). A key inference from this work is that where wildlife interact with domestic animals or humans, pathogens carried by wildlife are more likely to be infectious to humans.

Catchment characteristics

The characteristics of catchments can be an important determinant of pathogen loading to waterways (Tiedemann et al., 1987). The distribution of different land-use types within the catchment will determine the types of animals present at various locations and will directly affect their access to and potential impact on surface waters. The location, aspect, slope, soil type, and vegetation cover of catchments, particularly the riparian areas, may mitigate the impact of wildlife activities. In general, land use designated as native vegetation; nature reserves or watercourses will attract the highest level of wildlife abundance. Management practices for these areas will influence the distribution, fate and transport of pathogens from wildlife faeces. For example, planting of vegetation buffer zones will reduce the impact of surface runoff entering waterways. The hydrological connectivity of wildlife habitation areas as determined by slope, impervious surfaces and the presence or absence of riparian buffer zones can mitigate the potential impact of wildlife activities in the catchment e.g. retention ponds and contour banks.

Despite the contamination that can arise from wildlife, it is important to consider the role of wildlife within the broader catchment context. For many surface waters the contribution of contamination from wildlife may be relatively insignificant compared to that from domestic animal or human sources. Therefore, the significance of wildlife for any particular source water depends on local factors.

Farm management – inputs from domestic livestock

The farm management processes that contribute pathogens to catchments are the excretion of faecal material from host animals and the subsequent handling, storage and disposal of animal manure and slurry. The host animals include both domestic livestock for agricultural production and domestic companion animals such as cats and dogs. The processes that mitigate the survival and dispersion of pathogens from these sources include manure treatment and storage, retention pond entrapment, wetland retention and farm management activities such as stock rotation and restricting access via fencing.

The potential impact of this material on the quality of surface waters is influenced by the following factors.

Population density

Domestic animals are often kept at stocking ratios much higher than can be maintained for native and feral species that do not receive supplemental feeding. This means that pathogen prevalence is often much higher in domestic species than in wildlife. High domestic animal populations are more likely to transmit pathogens to the corresponding feral species in the same geographical area and vice versa. Estimates of domestic animal population densities are given in Table 2.7.

Table 2.7 Estimates of domestic animal density

Animal Density Reference (per km2) Cattle (grazing) 500 Estimated from discussions with the RLPB† Cattle (intensive) 2000 Estimated from discussions with the RLPB† Sheep 500 Estimated from discussions with the RLPB† Pigs 5000 Estimated from discussions with the RLPB† Horses 3 D. Ashton pers. comm. Goats 2 D. Ashton pers. comm. Dogs (domestic) 260* (O'Keefe et al., 2003) Dogs (domestic) 400* (Olley & Deere, 2003) Cats (domestic) 160* (O'Keefe et al., 2003) Cats (domestic) 400* (Olley & Deere, 2003) Poultry 5000 Author’s estimate * Density in urban land use areas † RLPB – Rural Lands Pasture Protection Board, Moss Vale Office

Volume of manure

Animal excretion rates and volumes for domestic animals are reasonably well documented compared to those for wildlife. Regional differences occur due to different feeding regimes and the quality and quantity of feed available as well as management practices. Estimates of manure production rates for domestic livestock and companion animals are shown in Table 2.8. The average weight of cattle faecal deposits in a study

carried out in New Zealand by Davies-Colley et al. (2002) was 920 g (n=5) per deposit. However, Larsen et al. (1994) reported that cattle defecated several deposits per event and that these may total 2 to 3 kg.

Table 2.8 Manure production rates for domestic and companion animals

Animal kg manure Reference .animal-1.d-1 Cattle (dairy) 55 (American Society of Agricultural Engineers, 1999) Cattle (dairy) 63 (van Eerdt, 1998) Cattle (dairy) 6.4 (Medema, 1999) Cattle (dairy) 50 (Dorner, Huck & Slawson, 2004) Cattle (beef) 21 (American Society of Agricultural Engineers, 1999) Cattle (beef) 27.251 (Australian Water Technologies, 2002a) Cattle (beef) 23 (Dorner, Huck & Slawson, 2004) Calves 5.6 (American Society of Agricultural Engineers, 1999) Calves 9.62 (van Eerdt, 1998) Calves (<1 yr) 5 (Dorner, Huck & Slawson, 2004) Sheep 1.1 (American Society of Agricultural Engineers, 1999) Sheep 1.0 (Ortega-Mora et al., 1999) Sheep 1.0 (Australian Water Technologies, 2002a) Sheep (ewes) 4 (Dorner, Huck & Slawson, 2004) Sheep (lambs) 1 (Dorner, Huck & Slawson, 2004) Pigs 5.1 (American Society of Agricultural Engineers, 1999) Pigs 6.2 (Australian Water Technologies, 2002a) Pigs (sows) 15 (Dorner, Huck & Slawson, 2004) Pigs (growers) 5 (Dorner, Huck & Slawson, 2004) Horses 23 (American Society of Agricultural Engineers, 1999) Horses 28 (Dorner, Huck & Slawson, 2004) Goats 2.6 (American Society of Agricultural Engineers, 1999) Goats 1.0 (Australian Water Technologies, 2002a) Deer 1.0 Similar to sheep (Australian Water Technologies, 2002a) Dogs (domestic) 0.5 (Australian Water Technologies, 2002a) Dogs (domestic) 0.45 (Weiskel, Howes & Heufelder, 1996) Dogs (domestic) 0.1 – 0.2 (O'Keefe et al., 2003) Poultry (layers) 0.12 (American Society of Agricultural Engineers, 1999) Poultry (layers) 0.088 (van Eerdt, 1998) Poultry (layers) 0.11 (Dorner, Huck & Slawson, 2004)

1 mean for dairy and beef cattle and calves 2 combined mass of urine and faeces (calf slurry), animals <24 wks

Pathogen prevalence and shedding intensity

There is substantially more data available on the concentration of pathogens in domestic animal faeces compared to wildlife faeces. This is because cattle, and to a lesser extent sheep have been repeatedly implicated as sources of waterborne outbreaks of cryptosporidiosis (Becher et al., 2005) and E. coli O157:H7 infections in humans. In many outbreaks the source of the causal organism has not been definitively identified. Thus it is now recognized that novel molecular techniques are required to genotype and subgenotype contaminating isolate(s) to conclusively discriminate and identify the sources of waterborne outbreaks (Chalmers et al., 2005). Although, it is well known that cattle may carry strains of Cryptosporidium that are infectious to humans (primarily C. parvum bovine genotype), there is no convincing evidence to date, to support the role of cattle as reservoirs of human infectious strains of Giardia (Becher et al., 2005). Initial estimates from mathematical models will need to use prevalence data and total (oo)cyst or bacterial counts to predict the potential load of microorganisms to waterways. However, genotyping and infectivity assessments will be required to determine the proportion of these total loads that represent an actual risk to human health (IPU budgets).

A summary of published Cryptosporidium, Giardia, E. coli and E. coli O157 concentrations in domestic livestock and companion animal faeces are shown in Table 2.9, Table 2.10, Table 2.11 and Table 2.12 (respectively).

Table 2.9 Prevalence and concentrations of Cryptosporidium spp. in faeces of domestic livestock and companion animals

Animal Country n Prevalence % Mean Reference -1 (95% CI) oocysts.g ((range)) (range) Cattle (dairy) UK 768 2 - (Bodley-Tickell, Kitchen & Sturdee, 2002) Cattle (dairy) Germany 8064 21.5 - (Joachim et al., ((19 –36)) 2003) Cattle (dairy) USA 19 10.5 - (Fayer et al., 2000) Cattle (dairy) Canada 54 28 490† (Heitman et al., 2002) Cattle (dairy) Canada 38 8 18.8† (Heitman et al., 2002) Cattle (dairy) UK 516 3.5 1778 (Sturdee et al., (SD 1.2) 2003) Cattle (dairy, USA 998 0 2.1 x 104 * (Wade, Mohammed >24 mo) (1- 7.9 x & Schaaf, 2000) 104) Cattle (dairy, USA 810 0 2.1 x 104 (Wade, Mohammed 6-24 mo) (1- 7.9 x & Schaaf, 2000) 104) Cattle (beef) UK 553 62.4 – 92.0‡ - (Scott et al., 1995) Cattle (beef) UK 220 3.6 1371 (Sturdee et al., (SD 0.5) 2003) Cattle (beef, USA 240 7.1 3.38† (Atwill et al., 2003) >24 mo) Cattle (beef, USA 120 8.3 2.63† (Atwill et al., 2003) >24 mo, pre- calving) Cattle (beef, USA 120 5.8 4.46† (Atwill et al., 2003) >24 mo, post calving) Cattle (beef, Denmark 55 3.6 - (Enemark et al., >24 mo) 2002) Cattle (beef, Denmark 62 29.0 - (Enemark et al., >24 mo) 2002) Cattle (beef, USA 484 0.6 - (Atwill et al., 1999) >12 mo) Cattle (beef, 7 USA 118 28.8 - (Fayer et al., 2000) – 9 mo) Cattle (beef, Canada 54 9 - (Olson et al., >6 mo) 1997b)

Table 2.9 (cont’d.) Prevalence and concentrations of Cryptosporidium spp. in faeces of domestic livestock and companion animals

Animal Country n Prevalence % Mean Reference -1 (95% CI) oocysts.g ((range)) (range) Cattle UK 810 5.4 19# (Hutchison et al., 2004) Cattle Spain 131 72 - (Lorenzo-Lorenzo, Ares-Mazas & Villacorta Martinez de Maturana, 1993) Cattle (dairy Australia 20 30 55 (Davies et al., & beef) (SD 232) 2005b) Cattle (dairy Spain 225 17.8 - (Quilez et al., & beef) 1996b) Calves (dairy) UK 727 21 - (Bodley-Tickell, Kitchen & Sturdee, 2002) Calves (dairy) USA 500 5.6 - (Sobieh et al., 1987) Calves (dairy) Mexico 512 25 - (Maldonado- Camargo et al., 1998) Calves (dairy) USA 7369 22.4 - (Garber et al., 1994) Calves (dairy) UK 367 52$ 1.0 x 105 (Sturdee et al., (SD 349.8) 2003) Calves (dairy) UK 272 23.2$$ 2.4 x 104 (Sturdee et al., (SD 21.2) 2003) Calves (dairy, USA 468 19.7 - (Santin et al., 2004) <11 mo) Calves (dairy, USA 1135 2.4 2.1 x 104 (Wade, Mohammed <6 mo) (1- 7.9 x & Schaaf, 2000) 104) Calves (dairy, Australia 54 48 - (Becher et al., <2 mo) (34.8, 61.5) 2005) Calves (dairy, USA 503 50.3 - (Santin et al., 2004) <2 mo) Calves (dairy, France 440 43.4 - (Lefay et al., 2000) <3 wk) (38.8– 48.0)^ Calves (dairy, USA 342 51 - (Ongerth & Stibbs, 1- 3 wk) 1989) Calves (dairy, France 1628 17.9 - (Lefay et al., 2000) <2 wk) (16.1 – 19.8) Calves (beef) USA 1053 11 - (USDA, 1994)

Table 2.9 (cont’d.) Prevalence and concentrations of Cryptosporidium spp. in faeces of domestic livestock and companion animals

Animal Country n Prevalence % Mean Reference -1 (95% CI) oocysts.g ((range)) (range) Calves (beef, USA 915 5.6 - (Atwill et al., 1999) <11 mo) Calves (beef, Canada 20 5 - (Ralston, Allister & <6 mo) Olson, 2003) Calves (beef, Canada 25 15 - (Olson et al., <6 mo) 1997b) Calves (beef, Denmark 18 11.1 - (Enemark et al., < 2 mo) 2002) Calves (beef, Denmark 24 20.8 - (Enemark et al., < 2 mo) 2002) Calves (dairy Australia 25 84 7460 (Davies et al., & beef, <12 (SD 2.5 x 2005b) mo) 104) Calves (dairy Spain 121 14 - (Quilez et al., & beef, <4 1996b) mo) Calves (dairy Spain 78 53.8 - (Quilez et al., & beef, < 6 1996b) wk) Calves (dairy Spain 844 47.9 - (Castro-Hermida, & beef, 1 - 3 Gonzales-Losada & wk) Ares-Mazas, 2002) Sheep Italy 156 12 - (de Graaf et al., 1999) Sheep Spain 14 93 67.5 (Ortega-Mora et al., 1999) Sheep UK 24 29.2 10~ (Hutchison et al., 2004) Sheep UK 90 75 - (Chalmers et al., 2002) Sheep UK 250 6.4 2800 (Sturdee et al., (SD 4.3) 2003) Sheep USA 23 17.4 - (Xiao, Herd & Rings, 1993) Sheep Australia 20 75 957 (Davies et al., (SD 3445) 2005b) Sheep (>6 mo) Canada 22 27 - (Olson et al., 1997b) Lambs UK 255 12.9 1.8 x 104 (Sturdee et al., (SD 23.8) 2003)

Table 2.9 (cont’d.) Prevalence and concentrations of Cryptosporidium spp. in faeces of domestic livestock and companion animals

Animal Country n Prevalence % Mean Reference -1 (95% CI) oocysts.g ((range)) (range) Lambs Australia 13 77 270 (Davies et al., (SD 497) 2005b) Lambs (<6 Canada 40 23 - (Olson et al., mo) 1997b) Lambs (2-3 USA 23 78.3 - (Xiao, Herd & wk) Rings, 1993) Lambs (0-2 Spain - 71 - (Ortega-Mora et wk) al., 1999) Pigs UK 126 13.5 58## (Hutchison et al., 2004) Pigs Canada 40 0 0 (Heitman et al., 2002) Pigs Australia 18 33 14.3 (Davies et al., 2005b) Pigs (sows) USA 19 0 - (Xiao, Herd & Bowman, 1994) Pigs (>6 mo) Canada 4 100 - (Olson et al., 1997b) Pigs (<6 mo) Canada 78 5 - (Olson et al., 1997b) Pigs (1-6 mo) Spain 620 21.9 - (Quilez et al., 1996a) Pigs (6-8 wk) Australia 18 66 472 (Davies et al., 2005b) Pigs (6-8 wk) USA 30 26.6 - (Xiao, Herd & Bowman, 1994) Horses UK 416 8.9 2067 (Sturdee et al., (SD 2.4) 2003) Horses (>6 Canada 24 21 - (Olson et al., mo) 1997b) Horses (<6 Canada 10 10 - (Olson et al., mo) 1997b) Goats Italy 42 19 - (de Graaf et al., 1999) Goats Spain 367 11 - (de Graaf et al., 1999) Goats USA 19 26 - (de Graaf et al., 1999) Deer (white- USA 11 9 (Fayer et al., 1996) tailed, farmed)

Table 2.9 (cont’d.) Prevalence and concentrations of Cryptosporidium spp. in faeces of domestic livestock and companion animals

Animal Country n Prevalence % Mean Reference -1 (95% CI) oocysts.g ((range)) (range) Dogs Spain 81 7.4 - (Causape et al., (domestic) 1996) Dogs USA 59 3.8 - (Hackett & Lappin, (domestic) 2003) Cats Australia 40 10 - (McGlade et al., (domestic) 2003) Cats USA 600 8.3+ - (McReynolds et al., (domestic) 1999) Cats Australia 162 1.2 - (Sargent et al., (domestic) 1998) Chicken USA 33 27 - (Ley et al., 1988) (broiler) Chicken USA 17 6 - (Ley et al., 1988) (layer) Chicken Netherlands 16 27 2100 (Medema et al., (layer) 2001)

* n = 27, overall mean regardless of animal age ‡ = 62.4% positive by faecal smear which increased to 92% positive with sucrose flotation # Geometric mean of positive samples, n=44 ~ Geometric mean of positive samples, n=7 ## Geometric mean of positive samples, n=17 † Mean concentration in positive animals ^ 90.5% of animals tested had symptoms of diarrhoea $ Home bred calves $$ Bought in calves @ Lower result for farm with good management, higher result for farm with poor management + Based on seroprevalence using C. parvum specific IgG

In the study by Wade et al. (2000) no association was found between the season of the year when animals were sampled and the prevalence rates of infection with C. parvum. An important difference between dairy and beef calves is the age when the maximum prevalence of Cryptosporidium shedding develops. In beef calves the peak occurs at about 2 months whereas in dairy calves the peak shedding intensity is reached at about 2 weeks of age, suggesting that they are infected within a few days of birth (Atwill et al., 1999). The recent study by Santin et al. (2004) included genotyping of the Cryptosporidium isolates, this revealed that approximately 85% of the infections in

pre-weaned calves (<2 mo old) were with C. parvum, the zoonotic species, but only 1% of the infections in post-weaned calves (<11 mo) were with C. parvum. The most prevalent genotype in the post-weaned calves was the Cryptosporidium Bovine B genotype (55%) followed by C. andersoni (13%). A survey by Fleming et al. (1999) indicated that Cryptosporidium was at least as, if not more prevalent in young pigs as in lambs.

Table 2.10 Prevalence and concentrations of Giardia spp. in faeces of domestic livestock and companion animals

Animal Country n Prevalence Mean Reference (%) cysts.g-1 (range) Cattle (dairy, USA 998 0.2 3000 * (Wade, >24 mo) (1- 8.5 x 104) Mohammed & Schaaf, 2000) Cattle (dairy, 6- USA 810 3.5 3000 (Wade, 24 mo) (1- 8.5 x 104) Mohammed & Schaaf, 2000) Cattle (dairy) Canada 54 25 29.9† (Heitman et al., 2002) Cattle (dairy) Canada 38 10 1.5† (Heitman et al., 2002) Cattle (beef) Canada 20 15 38.5 (Ralston, Allister & Olson, 2003) Cattle (beef, 7 – USA 118 37.3 - (Fayer et al., 9 mo) 2000) Cattle (beef, >6 Canada 54 11 - (Olson et al., mo) 1997b) Cattle UK 810 3.6 10‡ (Hutchison et al., 2004) Calves (dairy, USA 1135 20.1 3000 (Wade, <6 mo) (1- 8.5 x 104) Mohammed & Schaaf, 2000) Calves (dairy, < Australia 54 89 (80.5, 37.6# (Becher et al., 11 wk old) 97.7) 2005) Calves (beef, <6 Canada 25 31 - (Olson et al., mo) 1997b) Calves (beef, Canada 20 21 2 (Ralston, Allister 25-27 wk) & Olson, 2003) Calves (beef, 5 Canada 20 85 2230 (Ralston, Allister wk) & Olson, 2003) Sheep UK 24 20.8 20~ (Hutchison et al., 2004) Sheep (>6 mo) Canada 22 9 - (Olson et al., 1997b) Lambs (<6 mo) Canada 40 57 - (Olson et al., 1997b) Pigs UK 126 2.4 68## (Hutchison et al., 2004) Pigs Canada 40 18 16.1† (Heitman et al., 2002)

Table 2.10 (cont’d.) Prevalence and concentrations of Giardia spp. in faeces of domestic livestock and companion animals

Animal Country n Prevalence Mean Reference (%) cysts.g-1 (range) Pigs (sows) USA 19 0 - (Xiao, Herd & Bowman, 1994) Pigs (>6 mo) Canada 71 18 - (Olson et al., 1997b) Pigs (<6 mo) Canada 147 3 - (Olson et al., 1997b) Pigs (6-8 wk) USA 30 3.3 - (Xiao, Herd & Bowman, 1994) Horses (>6 mo) Canada 24 25 - (Olson et al., 1997b) Horses (<6 mo) Canada 10 0 - (Olson et al., 1997b) Dogs USA 59 5.4 - (Hackett & (domestic) Lappin, 2003) Dogs Italy 17 15 - (Berrilli et al., (domestic) 2004) Cats (domestic) Australia 40 80 - (McGlade et al., 2003)

* Overall mean regardless of animal age, n=263 ‡ Geometric mean of positive samples, n=29 # Arithmetic mean of average cyst concentrations for 3 age groups (<3 wk, 4-7 wk and > 8 wk calves) ~ Geometric mean of positive samples, n=5 ## Geometric mean of positive samples, n=3 † Mean concentration in positive animals

A recent study by Berrilli et al. (2004) indicated that not all Giardia cysts carried by dogs were likely to be infectious for humans. The study showed that 76.5% of isolates were genotyped as Assemblages C, D or mixed C/D and only 23.5% exhibited potential zoonotic genotypes (Assemblage A and A/C mixed Assemblages).

Table 2.11 Concentrations of faecal coliforms and E. coli in faeces of domestic livestock and companion animals

Animal Country n Faecal E. coli Reference coliforms Mean -1 -1 cfu.g cfu.g Cattle (dairy) NZ 5 - 1.2 x 107 (Davies-Colley et al., 2002) Cattle (dairy) USA - 2.3 x 105 8.4 x 104* (Geldreich et al., 1962) Cattle (dairy & Australia 20 - 2.1 x 106 (Davies et al., 2005b) beef) (SD 2.1 x 106) Cattle UK - - 2.3 x 105 (Jones & White, 1984) Calves (dairy & Australia 25 - 4.2 x 109 (Davies et al., 2005b) beef, <12 mo) (SD 1.3 x 1010) Sheep Australia 20 - 5.9 x 106 (Davies et al., 2005b) (SD 8.1 x 106) Sheep USA - 1.6 x 107 4.1 x 106* (Geldreich et al., 1962) Sheep UK - - 1.6 x 107 (Jones & White, 1984) Lambs Australia 13 - 6.9 x 106 (Davies et al., 2005b) (SD 1.5 x 107) Pigs USA 3.3 x 106 9.7 x 105* (Geldreich et al., 1962) Pigs Australia 20 - 1.2 x 107 (Davies et al., 2005b) (SD 2.4 x 108) Pigs UK - - 3.3 x 106 (Jones & White, 1984) Pigs (6–8 wk) Australia 20 - 1.2 x 108 (Davies et al., 2005b) (SD 1.5 x 107) Horses USA - 1.26 x 5.8 x 103* (Geldreich, 1978) 104 Dogs (domestic) USA - 2.3 x 107 5.8 x 106* (Geldreich, 1978) Cats (domestic) USA - 7.9 x 106 2.2 x 106* (Geldreich, 1978) Chicken USA - 1.3 x 106 4.1 x 105* (Geldreich et al., 1962) Turkey UK - - 3.0 x 105 (Jones & White, 1984)

* Converted from faecal coliform count using equation 1

Table 2.12 Concentrations of E. coli O157:H7 in faeces of domestic livestock and companion animals

Animal Country n Prevalence Mean Reference (95% CI) cfu.g-1 (range) Cattle (dairy) Australia - - 43 (Fegan & Desmarchelier, 2003) Cattle (beef) Canada 4790 13.3 - (LeJeune et al., 2004) Cattle UK 810 13.2 1200* (Hutchison et al., 2004) Calves (dairy, >2 USA 263 4.9‡ - (Zhao et al., mo & <4 mo) 1995) Calves (dairy, <2 USA 399 1.5‡ - (Zhao et al., mo) 1995) Calves (dairy, >2 USA 132 5.3^ - (Zhao et al., mo & <4 mo) 1995) Calves (dairy, <2 USA 171 2.9^ - (Zhao et al., mo) 1995) Sheep UK 24 20.8 780# (Hutchison et al., 2004) Sheep UK 677 6.5 (<100 - >106) (Ogden, MacRae (4.6, 8.4) & Strachan, 2004) Pigs UK 126 11.9 3900† (Hutchison et al., 2004)

* Geometric mean of positive samples, n=107 # Geometric mean of positive samples, n=5 † Geometric mean of positive samples, n=15 ‡ Control herds, not previously tested for E. coli O157 and randomly selected ^ Case herds, previously tested by USDA and positive for E. coli O157

Animal age and behaviour

Direct deposition of faeces in watercourses is probably the chief means by which domestic animal waste enters surface waters, and may serve as a first approximation to the total load. Surface transport will also be significant in areas where a continuous liquid layer connection to the watercourse can exist during wet weather events.

Hafez et al. (1969) determined that faecal deposits from cattle were nonuniformly distributed throughout pasture areas. This nonuniform distribution could result in approximately 0.4 to 2% of the pasture being covered by manure annually. However, in certain areas such as fence lines, gates, bedding and watering areas the density could be much higher (Johnson, Gary & Ponce, 1978). In one study the location of randomly sampled pats was recorded, 45.4% were on grazing areas, 23.2% at feed/watering stations, 11.4% on riparian areas, 10.4% along trails, and 9.8% under trees (Hoar et al., 1999). Free-ranging cattle will defecate 12 times daily on average (Larsen et al., 1994; Thelin & Gifford, 1983) and with unrestricted access to waterways, free ranging cattle will defecate in streams 3.4% of the time in August (late summer) and 1.7% in November (late autumn) (Larsen et al., 1988). Where streams are the only water source the rate may be higher, 6.7 to 10.5% (Gary, Johnson & Ponce, 1983). Cattle are known to spend 5-30 times as much time in riparian zones than would be predicted from surface area alone (Belsky, Matzke & Uselman, 1999). A study by Davies-Colley et al. (2002) showed that when crossing a creek the rate of direct faecal deposition from cattle was approximately 10% (25 deposits from a herd of 256 animals) with an additional 11 fresh deposits found within 200 m of the crossing area. Although no published studies could be found, it is thought that the rate of direct faecal deposition by sheep in riparian zones is much lower than for cattle.

Zoonotic transfer

As domestic animal populations increase, the likelihood of pathogens being carried by that population also increases. This increases the opportunity for transmission to susceptible wildlife species located in the same geographical area and

vice versa. Pathogens also spread between domestic animals (Shere et al., 2002) as well as between domestic and wild animals. Therefore, domestic and human pathogen sources can pollute water directly or, indirectly via wildlife.

Catchment characteristics

Catchment characteristics can be significantly altered by variations in farm management techniques. For example, grazing pasture sites that lead to congregation of animals around waterways will lead to increased faecal deposition in areas that will readily contribute pathogens to streams in the event of storms. Tate et al. (2000) simulated the release of Cryptosporidium from fresh calf pats by storms. Model pats (4 x 0.2 kg: 1.5x108 oocysts.kg-1) were placed on a 0.5 m2 soil plot (equivalent pat density to areas of cattle concentration) and subjected to intense artificial rainfall of 7.62 cm.h-1 (a 1 in 100 yr 30 min storm) for 90 min. About 1.2% of total oocysts appeared in overland flow (most in the first 30 minutes). Runoff however accounted for only 4% of the volume added by precipitation, and oocysts were not enumerated in the remainder which appeared as leachate (the plot had been pre-wetted so that precipitation and runoff-plus-subsurface flow were in equilibrium).

It is important to recognise that high-risk sources are not only those that promote direct faecal deposition to water. Pathogens can persist for some considerable time in faeces, particularly in cold-climate winters, where pathogens can persist throughout the winter to be released from land along with snow melt or during infrequent storms (Kistemann et al., 2002; Tiedemann et al., 1987). Climatic information needs for catchment managers, therefore, are relevant to wildlife pollution risk assessment. The seasonality of floodwater and snowmelt inflows to source waters, and the temperature ranges found in water and land systems, relate to risk. Climate change trends also impact the migration and abundance of wildlife populations, which in turn determine manure loadings to stream networks. For example, warmer winter temperatures and decreased snow depth could allow animals to extend their distributions into higher latitudes than may presently occur (Ayres & Lombardero, 2000).

Manure storage and treatment

In recent years management practices regarding handling, treatment and storage of farm animal faeces and manures have come under increasing scrutiny as a potential source of pathogens (Natural Resource Agriculture and Engineering Service, 2000). Farming activities vary depending on the products being produced and thus environmental concerns and the resources available to deal with them vary. Manure management strategies need to address the following concerns:

• Potential environmental damage; • Maximising nutrient value; • Impact on neighbours; • Minimising damage to the land; • Reducing costs and inconvenience.

Some of the management alternatives either in current use or being proposed by the USDA include; daily manure spreading, storage, odour control, solid separation, composting, biodrying, high solids anaerobic digestion, anaerobic digestion, lagoon treatment, sequencing batch reactors and total resource recovery (Natural Resource Agriculture and Engineering Service, 2000). Management strategies that reduce the risk of pathogen contamination by either increasing the rate of pathogen inactivation or decreasing the risk of transport and dispersion include:

• Storage of manure, high ammonia concentrations and temperature favour pathogen inactivation (particularly when the manure temperature exceeds 55ºC) • Increased storage time; • Drying • Anaerobic digestion • Incorporation of manure into the soil instead of surface spreading.

Urban development

Catchments that contain urban development introduce additional sources of pathogens primarily from human sources, but also from companion animals such as cats and dogs. Human sources include sewage effluent, septic leachate from on-site sewage disposal systems, stormwater and direct deposition from human recreation activities in rivers and storages. Although the level of efficacy varies widely, most human faeces will undergo some form of treatment before being discharged to the environment. The exceptions include direct deposition during recreation and illegal connections of sewage pipes to stormwater drains. The worst-case scenario for human faecal pathogen contamination is thus the discharge of raw faeces or sewage.

The pathogenic content of human faeces is determined by a number of factors including the rate of pathogen infection in the human population, the type of pathogen and the age of the individual (Anderson et al., 1998). The prevalence of Cryptosporidium shedding has been shown to be highly dependant on country and population and can range from 0-60%, with the higher proportion of shedding among diarrhoeic individuals (Ungar, 1990). Anderson et al. (1998) estimated that Cryptosporidium infection in the general population of the USA ranged from 0-5% and that Giardia infection ranged from 0-10%. These figures are similar to Cryptosporidium shedding estimates from the Centre for Diseases Control and Prevention (CDC) of 0.5-1% of the general population. Recently, Horman et al. (2004) estimated the prevalence of Cryptosporidium infection in the general Nordic population was 0.99% (95% CI, 0.81; 1.19, n=955) and that the rate of Giardia infection was 2.97% (95% CI, 2.64; 3.31, n=4505). The prevalence of C. parvum in the southern region of the Czech republic has been estimated at 5.8% (Chmelik et al., 1998).

E. coli bacteria are ubiquitous to the human gut and thus it would be estimated that the infection rate approximates 100% of the population. Shedding intensity or pathogen concentrations excreted by humans for Cryptosporidium and Giardia are assumed to vary log normally about a median value of 1 x 106 (oo)cysts.g-1 (Jakubowski, 1984; Robertson, Smith & Paton, 1995) with contents < 1 x 105 and > 1 x 107 (oo)cysts.g-1 occurring less than 1% of the time (Anderson et al., 1998). The

concentration of E. coli in human faeces is approximately 1 x 1010 cfu.g-1 (Feachem et al., 1983).

Sewage treatment plants (STPs)

The numbers and types of pathogens found in wastewater will vary both spatially and temporally, depending on the disease incidence in the population, season, water use, economic status of the population and the quality of the potable water (Rose & Carnahan, 1992). Wastewater treatment processes are designed to reduce nutrient and pathogen loads that are discharged to waterways. There are a wide variety of treatment processes available and various combinations of processes can be used. Common components of conventional treatment processes include primary sedimentation, a biological step using either activated sludge or trickling filters followed by secondary treatments such as retention or oxidation ponds. Tertiary treatment processes involve reducing pathogens and enhancing disinfection by the removal of soluble and particulate organic matter. Filtration is probably the most common tertiary process, although coagulation, particularly with lime can result in significant reductions in pathogens (Yates & Gerba, 1998). While the ability of conventional treatment processes to remove pathogens has been demonstrated in pilot and full-scale systems, the reliability of the performance of these processes can be highly variable, given the dynamic quality of raw wastewater. The issue of reliability is important since fluctuations in treatment efficacy can result in significant pathogen loads being discharged to wastewaters, and/or significant risk to public health if the wastewater is recycled. The variability in treatment processes and efficacy is evident in the wide variation in concentrations of pathogens present in wastewaters (Table 2.13).

Table 2.13 Mean pathogen concentrations in raw and treated STP effluent

Type of n Country Crypto- Giardia E. coli E. coli Reference effluent sporidium cysts.L-1 cfu.L1 O157 oocysts.L-1 cfu.L-1 Raw 8 Spain - - 1.6 x 1.6 x (Garcia- sewage 108 103 Aljaro, influent Bonjoch & Blanch, 2005) Raw 3 Italy 4.5 53.6 - - (Carraro et sewage SD 0.8 SD 6.7 al., 2000) influent Raw - France 10 - 100 1000 – - - (Rouquet et sewage 1 x 104 al., 2000) influent Raw - USA 15 490 (Rose et al., sewage 2001) influent Primary 11 Canada 0 200 - - (Heitman et al., 2002) Primary 11 Canada 30 160 - - (Heitman et al., 2002) Primary 10 Canada 0 420 - - (Heitman et al., 2002) Primary 10 Canada 130 700 - - (Heitman et al., 2002) Primary 11 Canada 70 390 - - (Heitman et al., 2002) Primary 10 Canada 178 170 - - (Heitman et al., 2002) Primary 14 Canada 40 3980 - - (Heitman et al., 2002) Primary 10 Canada 0 1470 - - (Heitman et al., 2002) Primary 12 Canada 0 5270 - - (Heitman et al., 2002) Primary 10 Canada 0 0 - - (Heitman et al., 2002) Primary 9 Canada 0 2210 - - (Heitman et al., 2002) Primary 9 Canada 0 880 - - (Heitman et al., 2002) Primary 11 Canada 220 930 - - (Heitman et al., 2002)

Table 2.13 (cont’d.) Mean pathogen concentrations in raw and treated STP effluent

Type of n Country Crypto- Giardia E. coli E. coli Reference effluent sporidium cysts.L-1 cfu.L1 O157 oocysts.L-1 cfu.L-1 Primary 9 Canada 0 700 - - (Heitman et al., 2002) Primary 11 Canada 0 400 - - (Heitman et al., 2002) Primary 8 Canada 0 1260 - - (Heitman et al., 2002) Primary - Scotland 43 8390 - - (Robertson et al., 2000)

Secondary 10 Italy 37 - 1.7 x - (Bonadonna SD 9 106 et al., 2002) SD 0.7 x 106 Secondary - Scotland 343 2030 - - (Robertson et al., 2000)

Secondary - Scotland ND 7 - - (Robertson et al., 2000)

Secondary - Scotland 12 1050 - - (Robertson et al., 2000)

Tertiary - Scotland 22 50 - - (Robertson (filtration) et al., 2000)

Tertiary - USA 0.0003 0.011 - - (Rose et al., 2001) Tertiary 11 Italy 0.21 1.39 - - (Carraro et SD 0.06 SD 0.51 al., 2000)

Since the microbial quality of effluent from wastewater systems varies widely and is dependent on many local factors, catchment specific wastewater data is needed for an accurate assessment of localised water quality impacts. There are 11 major sewage treatment plants within the SCA area of operations. Paterson and Krogh (2003) described the size, treatment type and efficacy of each of these plants, and the key characteristics are shown in Table 2.14. Major treatment plant upgrades are currently underway for several plants including Bowral, Goulburn, Lithgow, Bundanoon,

Wallerawang and Braemar. Historic water quality and flow data for treated STP effluent from these plants is shown in Table 2.15.

Data on the quality of raw sewage (primary effluent) was available for Braemar STP. Cryptosporidium (oo)cysts were enumerated in one mL subsamples, and the results were used to calculate contribution of (oo)cysts from untreated human waste (Table 2.16). For comparison, using combined data from two STPs (KRAL and WEST) the average annual protozoan loads shed by humans in the Netherlands were estimated to be Cryptosporidium 4.4 x 105 and Giardia 1.4 x 106 (oo)cysts.person-1.d-1 (Medema et al., 2001) which are similar to the values for Braemar STP.

Table 2.14 Summary of SCA sewage treatment plant characteristics modified from Paterson and Krogh (2003)

STP Year Design EP# Treatment Treatment type Disinfection ADWF ‡ WW+ Discharge area Commissioned capacity serviced level (KL.d-1) storage (ML) Braemar 2001 14000 6000 Secondary IDEA* UV 2250 2.6 Iron Mines Ck Berrima 1990 2000 1450 Secondary IDEA TEP† 370 11 Oldbury Ck Bowral 1936, -76, -82 10000 9975 Secondary Trickling filter TEP 2500 0 Mittagong & 2 Pasveer rivulet Moss Vale 1995 9000 7000 Secondary IDEA UV 1600 19.7 Whytes Ck Bundanoon 1982 2000 1900 Secondary IDEA TEP 510 0 Reedy Ck Goulburn - 31500 28000 Secondary Trickling filter MP^ 6800 45 Gorman Rd farm Marulan <1994 - 450 - Aeration & MP 110 44 Land reuse MP Braidwood >1985 3000 960 Secondary Trickling filter MP 240 0 Local Ck Mt Victoria 1970, 1997 990 1000 Tertiary IDEA UV & tertiary 300 0 Land treatment sand filtration & Fairy dell Ck Wallerawang <1994 2225 2100 Secondary Trickling filter TEP 530 0 Pipers Flat Ck Lithgow - 21000 18000 Secondary Trickling filter TEP 4400 Lagoons Farmers Ck

# EP - equivalent population ‡ ADWF – Average dry weather flow * IDEA – Intermittently decanted extended aeration † TEP – Tertiary effluent pond ^ MP – Maturation pond + WW – Wet weather storage capacity

Table 2.15 Average dry weather flow and mean microbial concentrations in treated STP effluent in SCA catchments from Paterson and Krogh (2003) and Krogh and Paterson (2002)

STP Years n ADWF‡ n Faecal coliforms E. coli* n Cryptosporidium† n Giardia†

Mean Median Mean Median Units KL.d-1 cfu.L-1 SD cfu.L-1 oocysts.L-1 cysts.L-1 Braemar 1996-2000 1461 1535 71 1.5 x 105 1.9 x 105 5.6 x 104 32 9.16 6.31 33 49.9 17.9 Berrima 1996-2000 1551 321 48 1160 1784 650 33 0.05 0 33 2.38 0 Bowral 1994-2000 2312 2332 84 8778 2.9x 104 4200 34 16.6 15.0 34 43.8 32.1 Moss Vale 1997-2000 1095 2062 84 509 2733 300 32 1.15 0.66 32 7.77 4.82 Bundanoon 1997-2001 1096 648 43 2328 3711 1200 33 0.09 0 33 1.44 0.33 Goulburn May 2002 - ND^ 33 6.5 x 104 7.4 x 104 2.6 x 104 33 18.4 11.0 33 3.20 1.66 Marulan - - ND - ND ND ND - ND ND - ND ND Braidwood 1998-1999 602 277 33 110 100 74 33 7.38 2.91 33 0.69 0 Mt Victoria 1997-2000 1096 135 182 2.7 x 105 1.3 x 106 9.7 x 104 - ND ND - ND ND Wallerawang - - ND - ND ND ND - ND ND - ND ND Lithgow May 2002 - ND 35 2250 3540 1200 32 0.12 0 32 9.25 2.76

‡ ADWF – Average dry weather flow * Converted from faecal coliform count using equation 1 † (Oo)cyst counts are adjusted for recovery efficiency, all data collected during May 2002 ^ ND not determined

Table 2.16 Protozoan parasites in raw sewage effluent at Braemar STP (Australian Water Technologies, 2002a) and (Olley & Deere, 2003)

-1 Pathogen Prevalence (oo)cysts.L Per person Per person p/n* in +ve volume of raw contribution of -1 -1 samples sewage (L.d ) ((oo)cysts.d ) 4 5 Cryptosporidium 2/7 (29%) 1.08 x 10 240 7.5 x 10 4 6 Giardia 6/7 (86%) 6.07 x 10 240 1.3 x 10

* p = number positive, n = number tested, sampling weekly for single eight week summer period - seasonal effects unknown

On-site systems (septic leachate)

The performance and sustainability of on-site sewage treatment and disposal systems have frequently been assessed as poor compared to centralised sewage systems. This evaluation has been reinforced by incidents such as the Hepatitis A outbreak at Wallis Lakes in Australia, and by the many reported incidents of contaminated groundwater drinking water supplies overseas (Hrudey & Hrudey, 2004). The primary threat to public health posed by on-site systems is from viruses, mainly due to their small size, which facilitates their transport, and also due to their robust survival in the environment. The recognition of this threat has led increasingly to research being focused on the occurrence and removal of pathogens, particularly viruses, from water and wastewater treatment systems, disinfection systems and transport through soil in land application systems, sand filters and vegetated buffer strips.

The Sydney Catchment Authority (SCA) manages a catchment area of approximately 16000 km2 that includes a number of unsewered urban areas such as Robertson, Oakdale and Oakland. For these towns and villages it is necessary to determine the cost and potential benefits of installing centralized sewer systems compared to alternative treatment strategies. Such assessments need to predict the potential environmental impacts of the various treatment alternatives compared to the existing status of water quality. Three main types of on-site sewage treatment systems are used in the Sydney catchment. These are:

• Septic tank plus soil absorption; • Pump out septic tanks; and • Aerated water treatment systems (AWTS).

The standard septic system consists of a primary sedimentation tank where supernatant overflows to a gravel-filled absorption trench, where it infiltrates to the underlying soil layer. AS 1546.1:1998 suggests minimum tank volumes for these systems of 3000 litres for 1 to 5 persons and 4000 L for 6 to 10. However, most extant systems are 2050 litres. Capacity is reduced over time by accumulation of sedimented solids. The tank is supposed to be pumped out when capacity has been reduced such that detention time is less than 24 hours, this may be required every 3 to 5 years - AS 1547:2000 assumes a sludge build up rate of 80 L.yr-1. Protozoan removal rates in the septic tank by either death or sedimentation would be negligible considering the short detention time (days). The minimum trench area required is determined by the effluent volume and the loading rate (mm.d-1) which can be absorbed under local soil conditions, guidelines for sizing absorption systems are provided in AS 1547:2000. Tank size upgrades to accommodate higher loads are common, but often unaccompanied by extra absorption capacity. Protozoan pathogens would not be physically retained by the coarse media in the absorption trench, but might be effectively trapped in the surrounding soil. Absorption trenches do however vary in effectiveness, and effluent may surface if a system's fluid capacity is exceeded. This can be caused by heavy precipitation, inadequate trench area, or if the permeability of the surrounding soil is or becomes insufficient (for example by growth of 'biomats'). Protozoa in surfacing effluent may enter nearby watercourses via overland flow.

A time-step model described by Jelliffe (1997) used water balance in the absorption trench to predict effluent surfacing. Inputs were flow from the tank and infiltration of precipitation, versus losses by percolation into the subsoil and evapotranspiration (the latter was effectively negligible). Failing systems were considered to have half the fluid absorption capacity required under AS 1547, and failure rates were estimated to be 5 to 10% - this might be reasonable for recently installed septics designed to AS 1547, however an earlier NSW survey reported visible surfacing effluent in 45% of septic systems (O'Neill, Roads & Wiese, 1993) which

might be more representative of older areas. The model predicted that effluent surfacing should occur rarely from properly functioning STSA systems, whereas contaminant release from failing systems was frequent, substantial and dominated the overall export budget.

Pump out septic systems should not release effluent to the surrounding environment, tank contents are periodically removed and disposed via municipal sewerage at the owner's cost. A high level of delinquency is however to be expected, with overflows or effluent illegally pumped out to local stormwater or watercourses.

AWTS are multi-stage biological treatment systems where effluent is degraded and clarified, and usually chlorinated before discharge, via sub-surface or (more typically) surface irrigation. Protozoan removal rates by AWTS are unquantified to date, but might be anticipated to be somewhat less effective than the equivalent municipal-scale secondary treatment process (activated sludge with post-settling), especially considering that in practice reliability of Watts is low, a very high percentage (70-95%) fail to meet regulatory discharge standards (Charles et al., 2001) based on all quality requirements. Surface application may also promote protozoan transport in runoff.

It is estimated that there are more than 18 000 on-site sewage treatment and disposal systems within the SCA’s area of operation. Of these approximately 79% are septic tanks, 1% are pump out systems and 20% are aerated wastewater treatment systems (AWTS) (Charles et al., 2001). The poor performance of these systems has been widely reported and, thus has been the focus of management reforms. However, the impact of these failures has not been quantified, nor is the location of these systems documented, making it difficult to assess the potential impact on waterways. Although the causes of on-site system failure are varied, appropriate design and regular maintenance may address many of the risks associated with on-site sewage disposal.

Limited information is available on pathogen concentrations in septic tanks or AWTS. However, calculations indicate that maximum concentrations of human enteric viruses could exceed 1010 viruses per litre for short periods when there is an infected resident in the household. The rate of virus inactivation and removal from wastewater

in septic tanks, primarily due to sedimentation with attached particles, is very low. Hence, land application systems and subsequent effluent transport in buffer distances still need to achieve large reductions of viruses to protect water quality. Deborde et al. (1998) reported ranges of 0.26 - 4.4 viruses.L-1 in effluent from a school septic tank system serving 350 people. Lewis and Stark (1993) reported ranges of 0.07 viruses per litre to greater than 59 per litre for household septic tanks, with viruses detectable in effluent for up to 137 days following an infection. In centralised sewage systems, virus concentrations of 104 - 105 viruses.L-1 of raw sewage have been reported (Gerba, 2000a).

It is possible that the concentrations of viruses in septic tanks may be much higher than reported due to variability in rates of excretion and duration of infection. For example, an infected individual can shed between 102 viruses per gram of stool during Coxsackie and echovirus infections that may continue for 2 weeks to 4 months; but up to 1012 viruses per gram for rotavirus infections, which may continue for 2 to 3 months (Gerba, 2000b). Virus inactivation or removal from wastewater within the septic tank is primarily due to sedimentation with attached particles, and is estimated at less than one log10 (USEPA, 2002). Removal in secondary treatment systems is expected to be slightly higher than septic tanks with an additional 1-2 log10 by disinfection (Charles et al., 2003b). Weiskel et al. (1996) found a geometric mean of 1.17 x 105 cfu.100 mL-1 faecal coliforms in septic effluent in the USA (n=4). Little data is available regarding the concentrations of parasitic protozoans in on-site system effluent. Data on the microbial quality of sewage effluent from septic and AWTS systems within the SCA area of operations are shown in Table 2.17.

Table 2.17 Cryptosporidium and Giardia (oo)cyst concentrations in sewage effluent from on-site systems, Robertson and Oakland, NSW

On-site n E. coli n Cryptosporidium (Oocysts.L-1) Giardia (Cysts.L-1) System (cfu.L-1)

Prevalence Mean SD Average Prevalence Mean SD Average (%) † content (%) † content

All septics 45 7.0 x 24 21 8.7 x 4.4 x 1.8 x 104 33 486 1140 162 107 104 105 Community 9 4.7 x 5 40 4.2 x 9.4 x 1.7 x 105 60 1050 2100 630 septics 106 105 105 Single 35 8.9 x 18 16 3.5 1.5 0.56 21 356 772 74 residences 107 AWTS 1 3.6 x 1 - <4 - - - <4 - - 104

* Negative results (not detected) not included † Estimate: average content = (prevalence x mean of positives)

Stormwater / urban runoff

Runoff from unsewered urban areas is believed to be the fourth largest source of Cryptosporidium oocysts in the SCA catchment areas (Australian Water Technologies, 2001). Weiskel et al. (1996) found a median faecal coliform density of 8.5 x 103 cfu.100 mL-1 (inter-quartile range of 1.6 x 103 – 4.0 x 104 cfu.100 mL-1, n=48) for stormwater draining residential and commercial areas. No difference was found between first-flush samples and those collected after 0.6 cm of rainfall (p=0.194, n=12). The results for dry weather drainage from an unsewered township in the Sydney catchment area are shown in Table 2.18.

Table 2.18 Protozoan parasites in unsewered urban runoff (Australian Water Technologies, 2002a)

n Cryptosporidium (oocysts.L-1) Giardia (cysts.L-1) Prevalence Median Range Prevalence Median Range 6 33 0 0 - 1 100 43 12 – 392

These results show that there is the potential for significant input of pathogens from stormwater runoff from urban areas. This indicates that the model will need to incorporate potential loads from urban animals/stormwater sources. An important component of estimating the quality of stormwater is to be able to estimate the inputs from companion animals. Protozoan loading on urban stormwater catchments might be estimated from density of urban animals (dogs, cats, rodents). This requires estimates of the density of companion animals and knowledge about how the faeces will be disposed of in the urban environment. Hackett and Lappin (2003) indicated that almost half of the homes in the USA (47.5%) have at least one dog.

Faecal dispersion from recreational use of water bodies

Direct input of faecal material can be expected from bathers and other recreational users of waterways. The incidence and impact of actual defecation are difficult to evaluate. Anderson et al. (1998) assigned a probability of 0.001 and faecal mass of 50 to 200 g per bather. Within the SCA area of operations recreational use of waterways is limited to passive secondary (boating) activities on one reservoir only. Primary contact recreation is prohibited in all reservoirs and storages. Only limited bathing occurs in streams and waterways, primarily adjacent to rural townships, remote from off-take points and generally limited to the summer months.

Construction of a pathogen source budget

The first step in the construction of a pathogen budget for drinking water catchments is to construct a source generation budget that describes the origin of pathogens to the system. The mathematical model will need to include an animal/land use module that incorporates inputs from both wildlife and domestic animal livestock sources. It will need to include estimates of population density for each species (wildlife and domestic livestock). Estimates of wildlife animal density are difficult to obtain because their movement is uncontrolled. However, estimates can be made from published data combined with locally acquired biological surveys, scat surveys, trapping and observations from catchment management staff. When estimating animal densities it will be useful to relate abundance to a particular land use as this will enable them to be extrapolated across the catchment area.

For each animal species it will be necessary to estimate the amount of manure excreted per day, the rate of host prevalence and the average concentration of representative or index pathogens. Manure excretion rates can be derived from published information although pathogen concentrations are likely to be more variable and should be crosschecked against locally collected data. The choice of representative pathogens should be made based on estimates of the likely pathogen risks depending on the geographical location and the downstream water treatment available. Collection of pathogen concentration data should focus on animal species that are most abundant,

those that excrete the largest volumes of manure, and those that inhabit the aquatic or riparian zones. It may also be necessary to assess whether or not the ratio of adult to juvenile animals is important by comparing pathogen concentration data for adults and juveniles to animal abundance by age. Animal behaviour will affect the rate of direct faecal deposition to streams, and this in turn is affected by factors such as climate, vegetation and the availability of shade and alternate sources of water. Although only limited data is available on these factors, they are beyond the scope of this study. A future SCA project is planned to assess the impact of animal behaviour patterns utilising radio satellite tracking collars. The potential for zoonotic transfer of pathogens should also be evaluated as part of a risk strategy for the management of catchments, however it will not be considered in the mathematical model.

The mathematical model will also need input modules to describe the origin of pathogens from human sources via STPs and on-site systems. These modules will quantify the inputs from human activities in those catchments that contain farming activities with rural residential and/or urban areas. The inputs from stormwater runoff will be captured indirectly in the animal/land use module by including estimates of animal population inputs that occur in urban areas, such as companion animals and rodent populations. The potential input of pathogens via direct deposition and faecal disintegration from human sources during recreational activities is considered to be small in comparison to the other potential sources in the Sydney drinking water catchments. This process will therefore not be included in the mathematical model. Application of the pathogen model to different catchments that allow recreational activities in water storages would warrant inclusion of this process, and would require further refinement of the model.

Chapter 3 Pathogen and indicator concentrations in animal faeces and sewage effluent in the Sydney catchment

Sections of this chapter have been published as: Cox, P., Griffith, M., Angles, M., Deere, D. A. and Ferguson, C. M. (2005) Concentrations of pathogens and indicators in animal faeces in the Sydney watershed. Appl. Environ. Microbiol. 71(10), 5929-5934.

Introduction

The transmission of waterborne pathogens continues to present a significant disease burden (Craun, 1979; Hunter, Waite & Ronchi, 2002) with protozoan, viral and bacterial pathogens having been associated with numerous waterborne outbreaks. One potential source of these pathogens in drinking water catchments is the faeces of domestic animal and wildlife populations. Pathogens from animal faeces may enter waterways by direct deposition or as a result of overland runoff containing faecal material deposited in the catchment. To quantify the impact from animal sources it is necessary to estimate the concentration of potential pathogens present in animal faeces (shedding intensity) and to estimate the prevalence of the potential pathogen in the animal population. The amount of pathogens generated will also be affected by the range of animal ages, with juvenile animals usually carrying higher burdens than adult animals. Animal behaviour will determine the sites where faeces are deposited in relation to the stream network while animal population densities will affect the pathogen load and the rate and range of zoonotic transmission. The pathogen loads carried and excreted by animal populations will also be subject to seasonal and climatic variation.

Chapter 2 identified that there is limited published information on the concentration and prevalence of potential waterborne pathogens and faecal indicator bacteria in wildlife populations (Aramini et al., 1999; Atwill et al., 2001; Chalmers et al., 1997; Graczyk et al., 1998) and even less for Australian species (Power et al., 2004). By comparison, data for potential pathogen concentrations in the faeces of domestic animals are more abundant, particularly for the enumeration and prevalence of

the protozoan parasites Cryptosporidium and Giardia (Atwill et al., 1998; Atwill et al., 2003; Atwill, Johnson & Pereira, 1999; Atwill, McDougald & Perea, 2000; Graczyk et al., 2000; Olson et al., 1999; Olson et al., 1997a). However, there were large differences in pathogen prevalence and shedding intensity across geographic locations, suggesting that published data need to be supplemented with local, catchment specific estimates.

Another significant source of pathogens in catchments identified in Chapters 1 and 2 are inputs from sewage treatment plants (Figure 1.1, Table 2.15). The wide variation in the microbial quality of sewage effluent is partly due to the variety of treatment methods used, but is also related to the wide variability in the quality of wastewater itself. The variability of wastewater matrices can cause significant variation in the efficacy of the treatment processes. Wide variation in pathogen removal could result in significant numbers of pathogens passing through a treatment plant for various periods of time (Yates & Gerba, 1998). Thus to estimate sewage effluent inputs within the Sydney catchment it will be necessary to supplement the existing data on pathogen concentrations for each of the STPs.

Quantification of total pathogen loads from various animal hosts and from sewage point sources is the essential first step in the calculation of a source material budget. Although there have been no reported waterborne disease outbreaks in the population served by drinking water from this catchment, further analysis of isolates by genotyping and virulence testing will facilitate the estimation of the IPU budget. Once total pathogen unit TPU budgets have been calculated for a catchment subsequent analysis can then be performed to estimate the proportion of the source material that are infectious pathogens and thus represents a risk of human infection (IPU) budget.

This study was undertaken to provide a cross-sectional estimate of the prevalence and intensity of shedding of pathogenic and indicator microorganisms in domestic and wildlife animals present in a large semi-protected drinking water catchment. The prevalence and shedding intensity data for animal faeces and the sewage effluent quality data collected during this study will be used to supplement the data presented in Chapter 2. The combined data will be used to construct an animal/land source budget for the pathogen model (Chapter 5).

C. Ferguson (SCA) and P. Cox (Sydney Water) were responsible for the scope, design and management of the study. M. Mannile, M. Krogh and C. Ferguson performed sample collection and faecal samples were processed at the Sydney Water Laboratories at West Ryde under the supervision of M. Angles, M. Logan, M. Warnecke and G. Ault. Data collation, analysis and report writing were undertaken by M. Griffiths, M. Angles and P. Cox. C. Ferguson carried out additional data interpretation and statistical analysis. C. Ferguson, M. Angles and P. Cox modified parts of the report for publication.

Materials and Methods

Sample collection

Animal faeces

Domestic animal faecal samples were collected from four catchments within the Sydney Catchment Authority (SCA) area of operations. They were the Wollondilly, Braidwood, Upper Cox’s and Wingecarribee catchments. These catchments are mixed land use with significant agricultural activities interspersed with urban and rural residential areas. Sheep and cattle grazing are the dominant agricultural land uses in the catchment. High sheep stocking rates are present in the Wollondilly catchment near Goulburn (approximately 550 000 sheep) and in the Shoalhaven catchment near Braidwood (approximately 109 000 sheep). Beef cattle grazing numbers are consistent throughout the catchments at a stocking density of about 25% of the sheep population. Dairy farms are also present, predominantly within the Wingecarribee catchment (approximately 10 500 adult cattle). There are at least 8 piggeries in the catchment, but no piggery effluent is discharged to waterways. The distribution of animal populations across different age groups in these catchments is unknown. We therefore aimed to estimate point prevalence for the overall population by collecting samples from a cross- section of animal ages. The exception was adult cattle (> 12 months of age) and calves (<12 months of age). Samples were collected twice from each catchment during April and May 2002 (autumn), except for the Wingecarribee, which was sampled three times.

Each domestic animal faecal sample consisted of a composite (minimum of 3 scats) from adult cattle, calves, sheep, pigs, dogs, horses, poultry, and cats. Samples were aseptically collected in sterilized wide mouth containers and stored at 4°C during transport and until analysis, with the exception of samples for viral analysis that were stored at -80°C. Following Animal Ethics approval, appropriately trained staff collected native and feral animal faecal samples. The majority of samples were collected from the Sydney University research facility, Arthursleigh farm, in the Wollondilly catchment, and the SCA lands at Braidwood. Both of these areas are mixed land use agricultural areas, the later interspersed with fragmented areas of native vegetation. These samples were collected in May and June 2002 (autumn to early winter). The native animals targeted included: the eastern grey kangaroo (Macropus giganteus), wombats (Vombatus ursinus), pademelon (Thylogale billardierii), swamp wallaby (Wallabia bicolor), brush-tail possum (Trichosurus vulpecula), platypus (Ornithorhynchus anatinus), common brown antechinus (Antechinus stuartii), wood duck (Chenonetta jubata) and a native rodent (Rattus fuscipes). Feral animals targeted included: pigs (Sus scrofa), foxes (Vulpes vulpes), rabbits (Oryctolagus cuniculus), goats (Capra aegagrus hircus), deer (Odocoileus virginianus), carp (Cyprinus carpio) and cats (Felis catus). The majority of samples were processed individually, however where the scat volume was too small for analysis, samples from the same species and sub-catchment were occasionally combined to make a composite. Faecal samples from feral animals were collected direct from the rectum post-mortem. Native animal samples were collected aseptically from the interior surface of Elliott traps or from the ground surface soon after defecation. Freshness of surface samples was ascertained by observation of defecation and presence of a high-moisture sheen.

Sewage effluent

Three treated sewage effluent samples were collected from ten of the eleven STPs in the catchment. They included; Wallerawang, Bowral, Lithgow, Mt Victoria, Moss Vale, Braemar, Bundanoon, Berrima, Goulburn and Braidwood STPs. The treatment processes for each of these plants is described in Table 2.14. Effluent flow volumes for each these STPs are described in Table 7.3. 84

The grab samples were collected from the EPA license sampling point. Samples were collected between the 22nd of April and the 12th of June 2002. Field measurements were made for pH, temperature, turbidity and electrical conductivity. Most samples were transported on ice for the duration of the sampling trip, with the exception of protozoan pathogen samples, which were stored at ambient temperature due to the large sample volume required for analysis.

Sample preparation

Animal faeces

Samples were thoroughly mixed using a sterile tongue depressor and a 1 g aliquot of faecal material was weighed into 99 mL of sterile Ringer’s solution. To assist in releasing the bacteria from the sediment and faecal material, each sample was sonicated for 30 s using a High Intensity Ultrasonic Processor (Vibra-Cell VCX 400, Sonics and Materials Inc., Newtown, CT, USA) as described by Davies et al. (1995). Following sonication, each sample was thoroughly shaken before subsampling into 50 mL volumes. When small volumes of faecal material were provided, a 0.5 g or 0.1 g aliquot was weighed into 99.5 mL and 99.9 mL of sterile Ringer’s solution, respectively. The sonication probe was disinfected between samples by immersion in 12.5 % sodium hypochlorite solution for 1 minute, followed by a rinse with sterile deionized water. The probe was then immersed in sterile 26% sodium thiosulphate for 10 s followed by a final rinse with sterile deionized water. To ensure that the cleaning process was effective, blank (sterile Ringer’s solution) control samples were processed in each batch for any evidence of carry over during the sonication process.

Sewage effluent

No specific pre-treatment was required for bacterial and protozoan analysis. For viral analysis sewage effluent samples were assessed on receipt to determine their suitability for ultrafiltration. If considered unsuitable for ultrafiltration, virus particles

were concentrated with 8% PEG-6000 and the concentrate resuspended in 40 mL Minimal Essential Medium (MEM). If considered clean enough for ultrafiltration, based on the absence of observable turbidity, samples were concentrated by an Amicon ultrafiltration cartridge (100,000 mol. wt. cutoff), followed by a further concentration in 8% PEG-6000 and resuspension in 40 mL MEM.

Comparison of bacterial enumeration of faeces with and without sonication

Throughout the project, a sample from a batch was randomly selected and processed in parallel with the other samples but without sonication. Instead, samples were thoroughly mixed and shaken before serial dilutions were prepared and processed. This was done to assess whether the use of sonication damaged either the vegetative cells or the C. perfringens spores sufficiently to reduce recoveries. Laboratory Quality Control samples including faecal material spiked with E. coli and C. perfringens and sterile dilution water were processed for every batch.

The difference in recovery data of faecal coliforms and C. perfringens treated by sonication and not treated by sonication was analysed by the Wilcoxon signed-ranks test (Helsel & Hirsch, 1992). Probability values (p) less than 0.05 were considered significant.

Bacterial analysis

Isolation and enumeration of faecal coliforms was carried out using standard membrane filtration methods as described in the Standard Methods for the Examination of Water and Wastewater (APHA, 1998). Isolation and enumeration of C. perfringens spores was carried out using a standard membrane filtration method (Standards Australia, 2000). In each case, serial dilutions were prepared from the sonicate (after settling for 10 min) or sewage sample and 1 mL of the dilutions, placed into 25 mL of sterile Ringer’s, was filtered through a 0.45 μm pore size cellulose ester membrane (Millipore, Sydney, Australia).

Faecal coliforms

Faecal coliforms were isolated and enumerated by placing the membrane onto an absorbent pad pre-soaked in mFC broth (without the addition of 1% Rosolic Acid in 0.2 N sodium hydroxide; Difco, Basingstoke, UK). Bacteria were resuscitated at 35°C for 2-4 h, followed by incubation for 18-22 h at 44.5°C. All blue colonies greater than 1 mm in diameter were counted and recorded as presumptive faecal coliforms. A maximum of 6 typical colonies from each sample was confirmed by inoculation into Lauryl Tryptose broth containing a Durham tube and Tryptone water (Oxoid, Sydney). After incubation at 44.5°C for 24 h, the tubes were examined for growth, gas and indole production. Colonies giving a positive Lauryl Tryptose broth and a positive indole result were confirmed as faecal coliforms.

C. perfringens spores

For the isolation and enumeration of C. perfringens spores the sample was heat shocked to 75ºC for 10 min to kill vegetative cells and immediately cooled on ice. One mL of each serial dilution was filtered and the membrane placed on Perfringens Agar (Oxoid) supplemented with 4-methylumbelliferyl phosphate (Sigma, Sydney). Perfringens Agar plates were incubated in an anaerobic jar with an anaerobic gas generating kit (Anaerogen - Oxoid) at 35ºC for 18-24 h. Colonies were examined for fluorescence under long wave UV light (366 nm) and all black or grey colonies that exhibited some degree of fluorescence or no fluorescence (atypical C. perfringens) were subcultured onto Brain Heart Infusion Agar (Oxoid). The Brain Heart Infusion Agar (BHIA) was incubated anaerobically at 35ºC for 18-24 h then colonies from the BHIA were inoculated into tubes of Lactose Gelatin and Nitrate Motility Medium that had been pre-boiled for 10 min to remove oxygen. The tubes were incubated aerobically at 35ºC for up to 48 h for Lactose Gelatin tubes and 24 h for Nitrate Motility Medium tubes. Colonies that produced gas and hydrolysed gelatin in Lactose Gelatin media, were non-motile and reduced nitrate in Nitrate Motility media were confirmed as C. perfringens.

Protozoan analysis

Animal faeces

A 0.5 g aliquot of the faecal sample was weighed into a sterile 50 mL centrifuge tube. To quantify recovery efficiency each individual sample was seeded with ColorSeed™ (Biotechnology Frontiers Ltd, BTF, Sydney, Australia) by vortex mixing a ColorSeed™ vial for 30 s, and adding the contents to the faecal sample. The ColorSeed™ vial was rinsed twice with approximately 4 mL elution buffer (Warnecke, Weir & Vesey, 2003), each time vortex mixing and adding to the faecal aliquot. The seeded faecal suspension was refrigerated overnight at 4°C. The dispersal of the faecal material was initiated by the addition of 20 mL 0.002 M sodium pyrophosphate and vortex mixing for 2 min. After incubation at room temperature for 30 min the faecal suspension was centrifuged at 3000 g for 10 min at 4°C. Using vacuum aspiration, all but approximately 7 mL of the supernatant was removed and the pellet resuspended by vortex mixing. The pellet was re-dispersed by sieving through a metal mesh (approximately 2.5 cm square; pore size, approximately 1.5 mm) into a sterile 50 mL centrifuge tube. The original sample tube was rinsed with sterile deionized water and these rinsings were passed through the sieve. The combined sieved material was stirred with a sterile wooden stick and the stick rinsed into the tube. The volume was made up to 50 mL with sterile deionized water and centrifuged at 3000 g for 10 min at 4°C. The supernatant was vacuum aspirated to leave approximately 5 mL in the tube. The tube was vortex mixed to resuspend the pellet, and the material transferred to an L10 tube (Dynal™, Oslo). The 50 mL tube was rinsed with 3 mL sterile deionized water, vortex mixed and the rinse transferred to the L10 tube. Rinsing was repeated with a further 2 mL of sterile deionized water.

To facilitate the detection of low concentrations of (oo)cysts and to maximize specificity we used immunomagnetic separation (IMS) utilising the Dynal™ GC Combo kit™ as per the manufacturer’s instructions, with the following modifications: Following the transfer of material from the L10 tube to a 1.5 mL tube, the supernatant from the 1.5

mL tube was used to rinse the L10 tube, and then transferred back to the 1.5 mL tube. Diethyl ether was used in dispersal and de-fatting on calf faecal samples as described by Davies et al. (2003). In addition, the liquid containing the dissociated cysts and oocysts was not neutralized with sodium hydroxide if samples were stained on the same day.

Dissociation was performed as per the Dynal™ GC Combo kit instructions, with the following modifications: with the microcentrifuge tubes in the Dynal™ magnetic apparatus (MPC-M) and the magnetic strip in place, the tubes were rested horizontally for 10 s to allow the magnetic beads to attach to the MPC-M. The supernatant was discarded, careful not to disturb the bead matrix. The centrifuge tubes were uncapped and 300 µL of 0.1 M HCl was added to each. After recapping each tube was vortex mixed for 10 s and incubated at room temperature for 10 min in a vertical position. The tubes were again vortex mixed for 10 s and each tube was tapped at the base to dislodge any remaining pellet. The magnetic strip was replaced in the MPC-M and the tubes were repositioned in the MPC-M. The tubes were allowed to rest horizontally for 10 s. The acid wash was transferred to a 13 mm membrane (Millipore, Sydney) with the vacuum on, with care taken to not disturb the bead matrix.

To stain the Cryptosporidium oocysts and Giardia cysts, three drops of cold methanol was applied to the surface of each membrane and allowed to stand for one minute. The excess fluid was aspirated under vacuum, and the membrane surface washed with 1 mL of methanol. The membrane was allowed to dry for 5 s and overlaid with 80 μL of 4’, 6’-diamidino-2-phenylindole (DAPI, 0.8 µg.mL-1) and allowed to stand for 2 min. The entire membrane surface was washed with 200 μL of wash buffer (BTF) while vacuum aspirating. Eighty µL of Easystain™ (BTF) was applied to the membrane and allowed to incubate for 15 min. The fluid was then aspirated under vacuum. The membrane surface was washed with 200 µL wash buffer (BTF) while vacuum aspirating. The membrane was transferred to a microscope slide with forceps and overlaid with mounting medium (BTF) and a cover slip applied.

The slide-mounted membranes were examined using an Axioskop epifluorescence microscope (Zeiss, Germany) employing filter set 09 (blue light excitation) for the Easystain™ (BTF), filter set 02 (UV light excitation) for DAPI

staining and filter set 15 (green light excitation) for the ColorSeed™ (BTF). The membrane was initially scanned at 200 x magnification for appropriate objects stained with EasyStain™, which were further examined at higher magnification where necessary to ensure correct identification. The identification criteria used for EasyStain™-labelled and DAPI stained objects were as described in USEPA method 1623 (USEPA, 1999).

Each raw result was adjusted for the recovery efficiency of the individual sample matrix using the following calculation:

(NI/NR)*no of oocysts counted (2)

Where NI = number of ColorSeed™ inoculated into the sample, and NR = number of ColorSeed™ recovered from the sample. Sample results with a recovery efficiency of <1% were discarded prior to statistical analysis (n = 12 for Cryptosporidium and n = 3 for Giardia).

Sewage effluent

Sewage samples were spiked with an internal control using ColorSeed™ (BTF) as described in the previous section. The minimum volume of sewage effluent tested was 1 L. Samples were concentrated through a 293 mm diameter flat-bed filter unit (Millipore) containing a 2 µm pore size track-etched polycarbonate membrane (Osmonics, USA) using a peristaltic pump (Watson-Marlow unit 604S, UK). The membrane was then transferred to a perspex sheet, and the concentrate eluted to a 50 mL centrifuge tube (Iwaki) using a 0.1% Tween 80 solution (ICN Biochemicals, Inc.) and a rubber squeegee. The concentrate was then centrifuged at 1620 g for 10 min and the supernatant aspirated to leave a 5 mL pellet. This method was based on the protocol originally described by Ongerth and Stibbs (1987). Sewage effluent samples were then processed using IMS and detected using IFA and DAPI as described in the previous section.

Virus analysis

Animal faeces

Faecal samples were diluted by suspending 2 g of homogenized faeces in 20 mL of phosphate buffered saline (PBS). A 10 mL aliquot of the faecal suspension was extracted with an equal volume of carbon tetrachloride, to remove lipids. Six mL of extracted faecal sample was inoculated onto cell culture after passing through a sterile

0.45 μm syringe filter to remove bacteria and fungi. Confluent cell cultures (LLC-MK2, MRC-5, and HEp-2) were prepared in cell culture tubes (Corning, Crown Scientific, Sydney, Australia). These cell lines are sensitive for the detection of reoviruses, many adenoviruses and enteroviruses including polioviruses, Coxsackie viruses and echoviruses. One mL was inoculated into 2 tubes of each cell culture, which equated to 0.6 g of the original sample. Due to the cellular toxicity of samples from herbivorous animals, a 1:10 dilution of sample was also inoculated, equating to 0.06 g of the original sample. Inoculated cell cultures were incubated at 37°C for 90 min to allow viral adsorption to the cells. Following this incubation, the inoculum was removed from the cell culture and replaced with fresh media to minimize sample toxicity and incubated at 37°C. The cultures were examined microscopically at least twice a week for a period of 28 days, or until viral cytopathic effect (CPE) was evident. Enterovirus, reovirus and adenovirus identification was achieved by observing the specific appearance of CPE on the various cell lines. All viral isolates were confirmed by sub-passaging.

Sewage effluent

A 6 mL aliquot of the concentrated sewage effluent sample was inoculated into two flasks or tubes of each cell culture (LLC-MK2, MRC-5, and HEp-2) and then processed as described in the previous section.

Results

Faecal coliforms and C. perfringens recovery with and without sonication

The recovery of faecal coliforms from animal faeces was significantly greater (p = 0.02, n = 12) in the sonicated samples. There were only 2 occasions involving one dog and one calf faecal sample, when recoveries using sonication were less than those obtained with no sonication (Table 3.1).

Table 3.1 Effect of sonication on recovery of bacteria from animal faecal samples

Faecal Concentration recovered (x 106 cfu.g-1 wet wt.) sample Faecal coliforms C. perfringens Sonication No Sonication Sonication No Sonication Cat 32 5.1 25 26 3 2.6 2.8 2.7 4.5 3.2 4.5 3.2 41 36 69 51 Dog 78 74 78 74 31 9 9.1 5.8 9 8 5.1 5.8 26 29 0.031 0.037 Poultry 250 10 0.0051 0.003 1.6 1.4 0.001 0.001 Calf 1.5 2.8 <0.0001 <0.0001 Possum 0.18 0.17 <0.0001 <0.0001

There was no significant difference in the recovery of C. perfringens spores from animal faeces either with or without sonication (p = 0.08, n = 12).

Faecal coliforms

Animal faeces

All pooled animal faecal samples from domestic animals contained high levels of faecal coliforms (Table 3.2).

Table 3.2 Faecal coliform concentrations in animal faeces

Faecal coliforms (cfu.g-1 wet wt.) Animal p/n Median Range Domestic animals Calves 9/9 3.0 x 106 5.8 x 105 – 8.3 x 108 Cattle 9/9 1.8 x 105 1.3 x 103 – 8.5 x 106 Sheep 9/9 6.6 x 105 1.0 x 105 – 1.9 x 108 Horse 9/9 3.8 x 104 4.0 x 103 – 3.3 x 106 Pig 9/9 7.1 x 106 5.8 x 105 – 4.1 x 108 Poultry 9/9 1.1 x 108 1.6 x 106 – 9.5 x 108 Cat 8/8 2.3 x 106 3.3 x 104 – 4.1 x 107 Dog 9/9 3.1 x 107 8.4 x 106 – 1.2 x 108 Native wildlife Antechinus 3/12 <100 <100 – 5.0 x 103 Brushtail possum 2/2 1.6 x 105 1.4 x 105 – 1.8 x 108 Wood duck 8/9 8.1 x 103 <100 – 6.9 x 104 Kangaroo 11/11 2.1 x 106 7.5 x 103 – 1.1 x 109 Platypus 4/11 <100 <100 – 4.6 x 104 Wallaby* 10/10 6.5 x 105 600 – 2.9 x 107 Rat 5/5 2.1 x 103 1.2 x 103 – 8.0 x 104 Wombat 7/7 4.0 x 103 100 – 5.7 x 104 Feral wildlife Carp 0/1 <100 - Deer 1/1 2.2 x 106 - Fox 1/1 9.6 x 106 - Goat 5/5 1.4 x 106 4.6 x 104– 3.4 x 109 Rabbit 3/4 5.0 x 105 <200 – 1.0 x 106 Feral cat 2/2 6.9 x 106 2.8 x 106 – 1.1 x 107 Feral pig 5/5 4.1 x 104 1.0 x 103 – 4.9 x 109

* includes Swamp wallaby and Wallaroo sample p = number of positive samples n = total number of samples

The highest median concentration from the domestic animal samples was 1.1 x 108 cfu.g-1 from poultry faecal samples. The poultry faecal samples also had the highest maximum faecal coliform concentration in domestic animals of 9.5 x 108 cfu.g-1, followed by the calf samples (8.3 x 108 cfu.g-1). Wildlife animals also had high median concentrations with the majority ranging between 8.1 x 103 and 9.6 x 106 cfu.g-1. The highest concentrations were recorded in samples from a feral pig (4.9 x 109 cfu.g-1), feral goat (3.4 x 109 cfu.g-1) and kangaroo (1.1 x 109 cfu.g-1). The exceptions were antechinus, platypus and carp, which had median concentrations of <100 cfu.g-1.

Sewage effluent

Faecal coliform concentrations in treated effluent from Mt. Victoria, Braidwood and Moss Vale STPs were low (<150 cfu.L-1) (Table 3.3).

Table 3.3 Faecal coliform concentrations in sewage effluent from STPs in the SCA area of operations

Faecal coliforms (cfu.L-1) STP Mean Median Range Braemar 1850 550 <10 - 5000 Berrima 2530 2900 1100 – 3600 Bowral 1.1 x 104 3000 1000 – 3.0 x 104 Moss Vale 140 60 <10 – 360 Bundanoon 5170 2500 1000 – 1.2 x 104 Goulburn 7.3 x 104 7.4 x 104 4000 – 1.4 x 105 Braidwood 110 100 100 – 140 Mt Victoria <10 <10 <10 - <10 Wallerawang 1.0 x 104 9100 6300 – 1.6 x 104 Lithgow 1930 1400 1000 – 3400 n = 3

However, high mean concentrations (>1.0 x 104 cfu.L-1) were recorded for effluent from Goulburn, Bowral and Wallerawang STPs. Mean concentrations at Bundanoon, Berrima, Lithgow and Braemar were also reasonably high (greater than 1000 cfu.L-1) and the most variation in effluent quality was recorded at Braemar STP.

C. perfringens spores

Animal faeces

Concentrations of C. perfringens spores in pooled animal faecal samples from domestic animals were highly variable (Table 3.4) with the majority of horse, calf, sheep and adult cattle samples being equal to, or less than the detection limit of 100 cfu.g-1. The domestic and feral cat samples had the highest median concentrations of C. perfringens (3.3 x 106 cfu.g-1 and 3.4 x 106 cfu.g-1, respectively), followed by the domestic pig (2.9 x 106 cfu.g-1). Concentrations of C. perfringens spores in wildlife faeces were low, frequently below the detection limit of 100 cfu.g-1. The exceptions were the feral cat, wood duck and fox samples and one sample of kangaroo faeces.

Table 3.4 C. perfringens spore concentrations in animal faeces

C. perfringens spores (cfu.g-1 wet wt.) Animal p/n Median Range Domestic animals Calves 2/9 <100 <100 – 6.6 x 105 Cattle 2/9 <100 <100 – 500 Sheep 2/9 <100 <100 – 7.2 x 106 Horse 3/9 <100 <100 – 100 Pig 9/9 2.9 x 106 200 – 1.8 x 107 Poultry 8/9 4.6 x 103 <100 – 4.3 x 106 Cat 8/8 3.3 x 106 7.9 x 105 – 6.9 x 107 Dog 8/9 3.6 x 105 <100 – 5.7 x 107 Native wildlife Antechinus 0/12 <100 <100 – <1.0 x 103 Brushtail possum 0/2 <100 <100 – <100 Wood duck 4/9 <100 <100 – 8.0 x 104 Kangaroo 1/11 <100 <100 – 600 Platypus 0/11 <100 <100 – <100 Wallaby* 0/10 <100 <100 – <100 Rat 0/5 <100 <100 – <100 Wombat 2/7 <100 <100 – 100 Feral wildlife Carp 0/1 <100 - Deer 0/1 <100 - Fox 1/1 4.3 x 103 - Goat 5/5 <100 <100 – <100 Rabbit 0/4 <100 <100 – <100 Feral cat 2/2 3.4 x 106 2.3 x 106 – 4.4 x 106 Feral pig 0/5 <100 <100 – <100

* includes Swamp wallaby and Wallaroo sample p = number of positive samples

Sewage effluent

There were few C. perfringens spores present in the treated effluent samples from Mt Victoria, Braidwood, Bundanoon and Berrima STPs (<400 cfu.L-1). The highest concentrations recorded were 8400 cfu.100 mL-1 from Moss Vale STP, 5200 cfu.100 mL-1 from Goulburn STP, 4700 cfu.100 mL-1 from Wallerawang STP and 3800 cfu.100 mL-1 from Bowral STP. There was a large amount of variability in the C. perfringens data from Bowral, Goulburn and Moss Vale STPs (Table 3.5).

Table 3.5 C. perfringens spore concentrations in sewage effluent from STPs in the SCA area of operations

C. perfringens spores (cfu.L-1) STP Mean Median Range Braemar 2630 1500 1000 - 5400 Berrima 350 360 160 – 520 Bowral 1.4 x 104 2700 <10 – 3.8 x 104 Moss Vale 2.8 x 104 1100 320 – 8.4 x 104 Bundanoon 320 350 160 – 460 Goulburn 1.9 x 104 4800 < 1.0 x 105– 5.2 x 104 Braidwood 280 270 <20 – 560 Mt Victoria 30 10 <10 - 70 Wallerawang 2.1 x 104 1.0 x 104 4500 – 4.7 x 104 Lithgow 990 880 280 – 1800

n = 3

Pathogenic protozoa

Animal faeces

Cryptosporidium oocysts were detected in at least one composite sample of the faeces from each of the domestic animal species with the exception of poultry where no oocysts were detected (Table 3.5). Median concentrations, however, were zero for most domestic species, except pigs (367 oocysts.g-1) and sheep (17 oocysts.g-1) (Table 3.6). In the case of native animals, kangaroos (4/11 samples), and possum (2/2 samples) were positive for Cryptosporidium oocysts. Median values for the native and feral animals were also zero with the exception of the possum faeces (54 oocysts.g-1), deer faeces (6 oocysts.g-1), and rabbit faeces (38 oocysts.g-1). Fewer samples were collected from feral animals compared to the domestic and native animals due to the difficulty in obtaining samples.

Giardia cysts were present in at least one sample of the faeces from each of the domestic animal species (Table 3.6). The highest maximum concentrations were recorded in samples from the domestic cat (>7143 cysts.g-1), dog (>6061cysts.g-1), pigs (>16,667 cysts.g-1) and sheep (504 cysts.g-1). In contrast to the Cryptosporidium oocyst

isolation, median values in the domestic animals were greater than zero for the majority of animal species (Table 3.6). Giardia cysts were less common in native and feral animal faeces than Cryptosporidium oocysts, with Giardia cysts only detected from the wood duck (4/9) and the fox (1/1) samples.

Table 3.6 Cryptosporidium and Giardia concentrations in animal faecal samples

Cryptosporidium (oocysts.g-1 wet wt.) Giardia (cysts.g-1 wet wt.) Animal ColorSeed™ ColorSeed™ p/n Median Range p/n Median Range recovery (%) recovery (%) Domestic animals Calves 4/7 9 0 - 183 44 - 85 7/9 133 0 – 533 3 - 40 Adult cattle 2/9 0 0 – 10 42 - 87 7/9 68 0 – 293 37 - 80 Sheep 6/9 17 0 – >6897 22 - 86 6/9 26 0 – 504 13 - 73 Horse 1/9 0 0 – 5 20 - 82 2/9 0 0 – 8 45 - 81 Pig 7/9 367 0 – 6000 1 - 49 5/9 11 0 – >1.6 x 104 2 - 61 Poultry 0/7 0 0 – 0 1 - 32 2/9 0 0 – 67 2 - 70 Cat 3/7 0 0 – 17 59 - 93 1/7 0 0 – >7143 8 - 58 Dog 2/8 0 0 – >5000 5 - 91 5/8 835 0 – >6061 6 - 77 Native wildlife Antechinus 0/3 0 0 – 0 8 - 11 0/8 0 0 – 0 3 - 51 Brushtail possum 2/2 54 20 – 89 41-65 0/2 0 0 – 0 40 - 41 Wood duck 0/9 0 0 – 0 27 - 68 4/9 0 0 – 339 38 - 77 Kangaroo 4/11 0 0 – 1257 14 - 81 0/11 0 0 – 0 29 - 78 Platypus 0/6 0 0 – 0 49 - 85 0/6 0 0 – 0 54 - 100 Wallaby* 0/10 0 0 – 0 32 - 87 0/10 0 0 –0 17 - 81 Rat 0/4 0 0 – 0 29 - 69 0/4 0 0 – 0 30 - 70 Wombat 0/5 0 0 – 0 16 - 61 0/5 0 0 – 0 25 - 73 Feral wildlife Carp 0/1 0 - 56 0/1 0 - 33 Deer 1/1 6 - 31 0/1 0 - 47 Fox - - - - 1/1 1.5 x 104 - 4 Goat 0/3 0 0– 0 16 - 27 0/3 0 0 – 0 34 - 58 Rabbit 1/2 38 0 – 77 13 - 33 0/2 0 0 – 0 19 – 63 Feral cat 0/1 0 - 51 0/1 0 - 61 Feral pig 0/5 0 0 – 0 1 - 79 0/5 0 0 – 0 1 - 47 * includes Swamp wallaby and Wallaroo sample, p = number of positive samples, n = total number of samples

Sewage effluent

Cryptosporidium and Giardia were detected in the final effluent from all STPs on at least one sampling occasion (Table 3.7). Samples collected from Bowral, Goulburn, Moss Vale and Mt Victoria STPs were positive for both Cryptosporidium and Giardia on all sampling occasions. The highest Cryptosporidium concentration recorded was >292 oocysts.L-1 (0% DAPI positive) from Mt Victoria STP. The highest Giardia concentration was >138 cysts.L-1 (19% DAPI positive) from the same Mt Victoria STP effluent sample. Low concentrations of both Cryptosporidium and Giardia were found in effluent from Bundanoon and Braidwood STPs.

Table 3.7 Cryptosporidium and Giardia concentrations in sewage effluent from STPs in the SCA area of operations

Cryptosporidium (oocysts.L-1)† Giardia (cysts.L-1)† STP p/n Mean SD p/n Mean SD Braemar 2/3 1.0 1.1 3/3 42.5 54.2 Berrima 2/3 10.1 17.5 3/3 5.3 8.8 Bowral 3/3 17.0 15.9 3/3 88 108.9 Moss Vale 3/3 1.0 0.8 3/3 46.8 56.1 Bundanoon 2/3 0.1 0.1 2/3 0.5 0.4 Goulburn 3/3 38.2 33.4 3/3 8.7 6.8 Braidwood 1/2 5.0 7.0 1/2 0.3 0.5 Mt Victoria 3/3 100.7 165.7 3/3 115.1 20.4 Wallerawang 1/3 0.3 0.5 3/3 45.5 45.6 Lithgow 1/2 0.8 1.1 2/2 47.4 64.1

† adjusted for recovery efficiency using ColorSeed™ as an internal control n = 3

Viruses

Animal faeces

Infectious enteric viruses were detected predominantly in the domestic animal faecal samples (Table 3.8). Viruses were detected in 4/5 calf faeces tested, with the majority of samples being positive for reovirus. One calf faeces sample tested positive for enterovirus, and was subsequently confirmed as a strain of bovine enterovirus

100

(BEV). Similarly, 4/6 adult cattle faecal samples were positive for virus, however in this case only reovirus was isolated. Only 1 faecal sample from pigs was positive for virus (reovirus). No viruses were isolated from domestic cats, dogs, horses, poultry or sheep. Reovirus was detected in 1 antechinus faecal specimen (Table 3.8). No viable virus was detected in any of the other native or feral animal faecal samples; carp (n = 1), deer (n = 1), possum (n = 2), platypus (n = 4), rat (n = 4), wombat (n = 7), duck (n = 9), wallaby/wallaroo (n = 10), kangaroo (n = 11), feral cat (n = 1), fox (n = 1), feral rabbit (n = 2), feral goat (n = 5) and feral pig (n = 5).

Table 3.8 Summary of virus data for animal faecal samples

Species n Adenovirus Reovirus Enterovirus p p p Domestic animals Pig 4 0 1 0 Calves 5 0 3 1 Adult cattle 6 0 4 0 Native wildlife Antechinus 4 0 1 0 n = total number of samples p = number of positive samples 0 = no positive samples

Sewage effluent

Sewage treatment plant effluent was sampled three times from each STP. Two samples from Wallerawang STP and one sample from Bowral STP were positive for Adenovirus. Reoviruses were detected in one sample each from Bowral, Braidwood, Goulburn and Lithgow STPs. Enterovirus was not detected in any of the STP effluent samples.

101

Moisture content of animal faeces

Table 3.9 shows the moisture content of the faecal samples for which there was sufficient material available for analysis. The moisture content for the calf and adult cattle faecal samples were higher than from any other species and showed little variation between samples (median 85.3% and 86.1% wet wt., respectively).

Table 3.9 Moisture content for animal faecal samples

Moisture content (% wet wt.) Animal n Median Range Domestic animals Calves 9 85.3 79.8 – 87.6 Cattle 9 86.1 85.0 – 88.1 Sheep 9 69.0 57.1 – 77.3 Horse 9 76.4 72.5 – 81.1 Pig 9 71.2 68.9 – 74.2 Poultry 9 70.3 57.8 - 78.4 Cat 8 58.7 46.9 – 95.7 Dog 9 68.3 64.3 – 72.4 Native wildlife Brushtail possum 1 56.8 - Wood duck 3 73.0 71.8 – 79.5 Kangaroo 5 73.1 46.0 – 74.9 Wallaby* 8 74.8 61.5 – 82.7 Wombat 5 72.9 64.0 – 75.4 Feral wildlife Deer 1 66.2 - Goat 1 53.7 - Feral pig 2 74.2 73.8 – 74.6

Discussion

Animal faeces

Point prevalence studies that examine pathogen prevalence in a given population at a particular point in time are relatively common (Enemark et al., 2002; Fayer et al., 2000; Olson et al., 1997b; Ralston, Allister & Olson, 2003; Xiao, Herd & Bowman, 1994) while, fewer studies have quantified pathogen excretion rates (Ortega-Mora et al.,

102

1999; Sturdee et al., 2003). Although prevalence studies are useful for disease management, these types of studies are insufficient for risk assessment. Both the prevalence and concentration of pathogens in faeces are required to establish pathogen source loads in catchments and hence assess the risk from faecal contamination of source waters. Estimation of source loads is further complicated because it is difficult to compare the data from different studies due to variations in methodologies, pooling of faecal samples, seasonality, animal age, herd immunity and temporal variation. For example, Power et al. (2004) showed that the proportion of Cryptosporidium infections in Eastern Grey kangaroos that could not be detected by normal clinical methods contributed a substantial load of oocysts in the catchment. For this reason we used a sensitive detection method capable of quantifying low concentrations of Cryptosporidium oocysts in animal faeces and included both wildlife and domestic animals, focusing on those species that were most abundant in the catchment and which excreted the greatest volume of manure.

Concentrations of C. perfringens spores, Cryptosporidium and Giardia were higher in the faeces of domestic animals compared to those from the native and feral animals. These results are similar to those of a Canadian study by Heitman et al. (2002) that found concentrations of Cryptosporidium oocysts and Giardia cysts in wildlife faeces were significantly lower than in domestic livestock. Faecal coliforms were isolated from all of the faecal samples examined in this study, whether from domestic, feral or native animals and at high concentrations in most cases. This indicated that animal sources represent a significant potential load of faecal coliforms in these drinking water catchments. Mathematical models that rely on faecal coliforms as surrogates for the estimation of pathogen loads (Fraser, Barten & Pinney, 1998; Tian et al., 2002) could therefore result in an overestimation of risk. In contrast, C. perfringens spores were mostly isolated from the domestic animal faeces and were rarely detected in the native or feral animal populations.

A comparison of sonication treatment and no sonication treatment on the recoveries of faecal coliforms and C. perfringens showed that the sonication method gave equivalent or higher counts for the majority of samples tested and would therefore be recommended as the method of choice. These findings are in agreement with a

103

previous study on the effectiveness of sonication for enumerating bacteria from sediments (Emerson & Cabelli, 1982).

The prevalence of viruses and the protozoan parasites Cryptosporidium and Giardia in a faecal survey will depend on the current levels of infection in the animal population. Cryptosporidium and Giardia were identified in at least one sample from each domestic animal species sampled (except poultry). The highest prevalence of Cryptosporidium from pooled samples from domestic animals was identified in pigs (78%). This is higher than the prevalence rates recorded by other studies, for example a study of 27 pig farms in Spain recorded a prevalence of 21.9% (Morgan et al., 1999a), while in Canada Olson et al. (1997b) reported a prevalence of 11% for domestic pigs. Pooled faecal samples from sheep also had higher prevalence rates of Cryptosporidium (67%) than previous studies; 10.1% recorded by Majewska et al. (2000) and 24% reported by Olson et al. (1997b). The higher prevalence rates in this study may have been caused by; the pooling of faecal samples, the inclusion of faeces from a cross- section of animal ages, and the use of IMS which is a more sensitive method for the detection of Cryptosporidium oocysts in animal faeces (Atwill et al., 2003). A recent study by Davies et al. (2005b) showed improved detection of Cryptosporidium oocysts using IMS, and they also detected a high prevalence of Cryptosporidium in adult and juvenile sheep (75% and 86%, respectively) in the Sydney catchment.

Over half the samples collected from domestic calves, adult cattle, dogs, pigs and sheep were positive for Giardia. The prevalence of Giardia was highest in pooled samples from adult cattle and calves (78%); this was much higher than the 40.6% for calves reported in a survey in New Zealand (Hunt, Ionas & Brown, 2000) but consistent with the results of Olson et al. (1997b). Very few wildlife faecal specimens were positive for Cryptosporidium or Giardia (oo)cysts. Cryptosporidium oocysts were identified in samples from possums, kangaroos, deer and rabbits. No positives were detected in the feral pig samples, although this result is probably biased due to the low number of animals sampled, since a study by Atwill et al. (1997) found that the prevalence of Cryptosporidium in feral pigs in California was 5.4%. The overall prevalence of Cryptosporidium in pooled kangaroo faecal samples in this study (36%) is similar to the rate recorded by Power et al. (2001) of 31.5% for samples collected during autumn. The study by Power et al. (2001) identified a strong seasonal effect on

104

the prevalence of Cryptosporidium in kangaroos. Future work could investigate the seasonal effect of prevalence rates for other wildlife species included in this study facilitating the evaluation of seasonal impact on the overall animal/land source generation budget.

While the trends for Cryptosporidium and Giardia are similar to faecal coliforms and C. perfringens, a reliance on the use of faecal indicator organisms would lead to an overestimation of the concentration of pathogens in faeces from native and feral animals. The application of molecular genotyping techniques (Morgan et al., 1999a; Morgan et al., 1999b) to faecal samples may facilitate the differentiation of sources, at least for the Cryptosporidium and Giardia isolates. The determination of the different genotypes of Cryptosporidium and Giardia will enable a more robust assessment of the risk that they pose to the drinking water supply. Further work needs to be carried out on the isolates from this study to determine their genotypes and hence assess the possible risks from their presence in the catchment.

Similar to the bacterial indicators and protozoa, viruses were isolated more frequently from the domestic animal population, compared to either the native or feral populations, albeit only from calves, cattle and pigs. Of the enteric viruses tested, reoviruses were isolated most often. The isolation of reovirus from the animal populations was not surprising given that these viruses are ubiquitous; the only three known serotypes (Ginsberg, 1980) infect a range of vertebrates including cattle, sheep, swine and other mammals.

The results for moisture content were consistent with those of a previous study by Davies et al. (2004b). It is not yet known whether the higher moisture content of bovine faeces enhances the survival of Cryptosporidium oocysts compared to other faecal matrices. However, the median moisture content of most other faecal samples was much lower ranging from 53 - 77 %. The cat faecal samples were highly variable, probably because they were the only samples collected from litter trays that are designed to absorb moisture.

105

Sewage effluent

The treatment plants sampled in this study all use different treatment processes prior to the release of effluent to the environment (Table 2.14). All are classed as secondary treatment plants, with the exception of Mt Victoria STP (tertiary). Concentrations of faecal coliforms, C. perfringens, Cryptosporidium and Giardia were detected in relatively high concentrations from all the STP treated effluents. The exception was the Mt Victoria STP where concentrations of faecal indicators were very low, to non-detectable but concentrations of Cryptosporidium and Giardia were high. Goulburn and Wallerawang STPs had the highest median concentrations of both faecal coliforms and C. perfringens spores. The highest maximum concentration for faecal coliforms was 1.4 x 105 cfu.L-1 from Goulburn STP. The arithmetic mean concentrations of faecal coliforms were >1000 cfu.L-1 for all of the STPs except Mt. Victoria, Braidwood and Moss Vale. Similarly, the mean concentrations of C. perfringens spores were generally >500 cfu.L-1 with the exception of Mt. Victoria, Braidwood, Bundanoon, and Berrima. The highest maximum concentration of C. perfringens was 8.4 x 104 cfu.L-1 from Moss Vale STP.

Cryptosporidium and Giardia were detected in the final effluent from all STPs on at least one sampling occasion. Cryptosporidium oocysts were found in 75% of samples, and Giardia cysts were found in 93% of samples analysed. All samples collected from Bowral, Goulburn, Moss Vale and Mt Victoria STPs were positive for both Cryptosporidium and Giardia. Mean densities ranged from 0.1 to 101 oocysts.L-1 for Cryptosporidium, and 0.3 to 115 cysts.L-1 for Giardia. The highest mean concentrations of both Cryptosporidium and Giardia were for effluent from Mt. Victoria STP. This contrasts with the faecal indicator bacteria results that showed very low concentrations of both faecal coliforms and C. perfringens spores for effluent from Mt. Victoria STP. Mt Victoria STP uses pasveer intermittent aeration ditches, a tertiary filter and ultraviolet disinfection to treat the effluent prior to its release into the environment. These results indicate that the use of C. perfringens as a surrogate for Cryptosporidium removal by treatment processes needs to be supplemented with specific information about the treatment processes and their impact on pathogen viability and infectivity.

106

Adenoviruses were present at two sites (Bowral and Wallerawang), reoviruses at four sites (Bowral, Braidwood, Goulburn and Lithgow). On the 23/04/02 treated sewage effluent from the Bowral STP contained both adenovirus and reoviruses. This was the only occasion that more than one type of enteric virus was isolated from a treated sewage sample. Whether Bowral STP is contributing a substantial viral load is difficult to determine from the limited data available. Unfortunately the presence/absence test used to detect viable enteric viruses gives no indication of the actual numbers being contributed by the STPs. Future work would ideally include some quantitative assessment of viral loads discharged in sewage effluent from these plants. However, the presence of enteric viruses in 33% of the treated effluents indicates that the STPs are contributing human infectious viruses to the catchment.

Overall, Goulburn STP typically provided the poorest effluent quality with consistently high faecal coliform concentrations and very variable C. perfringens spore concentrations. Goulburn STP uses a trickling filter and tertiary maturation pond to treat the effluent. Final effluent from this treatment plant is irrigated onto local land, and is only discharged into the Wollondilly River following periods of peak flow that exceed the effluent storage capacity. Cryptosporidium and Giardia were identified in all three samples collected. The water quality of the final effluent from Wallerawang STP was also relatively poor, containing high faecal coliforms and C. perfringens concentrations. Cryptosporidium was detected in one of the three samples collected, while Giardia was present in all samples. There were two positive identifications for adenovirus. Both Goulburn and Wallerawang STPs rely on trickling filters and settling ponds as the main process for effluent treatment.

Conclusions

This study carried out in late autumn and early winter of 2002 was a dry weather, cross-sectional survey of the prevalence and intensity of microbial shedding from animals within the Sydney drinking water catchment. The results show that faecal indicators and pathogens were more prevalent, and shedding intensity was generally higher for domestic animals than wildlife. This has implications for the calculation of an animal/land source budget since domestic animals also excrete significantly larger

107

volumes of manure per day compared to wildlife animals. The result is that domestic animal species, such as cattle, are likely to excrete the most pathogens per animal per day because they have high prevalence rates, shedding intensity and excrete large volumes of manure.

For these reasons the pathogen model will need to quantify the inputs from cattle both by direct deposition to the stream network, and via surface water runoff from faecal pats deposited on land. The next chapter will investigate the quantification of microbial transport rates for Cryptosporidium, E. coli and PRD1 bacteriophages from bovine faecal pats using artificial rainfall simulation at field-scale.

Although this study only included a small number of additional STP effluent samples, it does supplement the existing information on pathogen concentrations in effluent discharged from the STPs operated in the Sydney drinking water catchments. Further characterization of effluent quality over a longer period of time would improve the estimates of pathogen concentrations and variability of effluent quality from these plants.

Synthesis of data for use in the pathogen model

To identify the most suitable values for use in the mathematical model the locally collected field data reported in this chapter were compared with the published literature summarized in Chapter 2. Synthesis of these data was directed at identifying the most appropriate values for use in a pathogen model for Australian drinking water catchments. Other factors considered in the selection of values included:

• the sample size used in each study; • the methodology used; • the availability of data for various animal types.

108

The following sections outline the values selected for use in the construction of the model. The animal density data provided by SCA catchment officers, published and local researchers (Table 2.1 and Table 2.7) were used as the best available estimates for animal density in the SCA catchment.

Data on the volume of manure produced per animal per day were selected from the literature and local studies, as outlined in Table 3.10.

109

Table 3.10 Volume of manure produced per animal per day

Animal Species Manure Reference Comments (kg.d-1) Cattle, grazing 27.25 (Australian Water Mean for dairy and beef cattle, similar and intensive Technologies, to the value of 23 kg reported by 2002a) Dorner et al. (2004) and the range 27- 40 kg reported by Larsen et al. (1994) Sheep 1 (Australian Water Good agreement on this value, and Technologies, similar to the 1.1 kg value reported by 2002a; Ortega- the ASAE (1999) Mora et al., 1999) Pigs, domestic 6.2 (Australian Water Similar to the 5.1 kg value reported by and feral Technologies, the ASAE (1999) 2002a) Kangaroos 0.2 M. Roberts pers. Estimate from researcher working on comm. kangaroo abundance and distribution Dogs, domestic 0.5 (Australian Water Similar to the estimate (0.45 kg) of and feral Technologies, Weiskel et al. (1996) 2002a) Cats, domestic 0.2 - Estimated as slightly less than half the and feral amount excreted by dogs per day Horses 23 (American Society Similar to value of 28 kg reported by of Agricultural Dorner et al. (2004) Engineers, 1999) Goats 1 (Australian Water Less than the estimated value of 2.6 kg Technologies, from the ASAE (1999) possibly due to 2002a) differences in diet Deer 1 (Australian Water Would be expected to be similar to Technologies, sheep and goats based on similar size 2002a) and diet Other 0.2 - Similar or less than the amount marsupials excreted by kangaroos Rodents 0.01 - Estimated as ten times less than the volume from poultry Rabbits 0.2 - Estimated as similar to cats, this value is higher than the 0.078 kg estimated by Medema (1999) Foxes 0.3 - Estimated as slightly higher than the volume excreted by cats and rabbits Poultry 0.12 (American Society Similar to values of 0.11 (Dorner, of Agricultural Huck & Slawson, 2004) and 0.088 Engineers, 1999) (van Eerdt, 1998)

110

The rate of manure mobilisation was estimated by assessing the likely rate of manure movement towards the riparian zone in wet weather, based on faecal material composition, weight and form and the likely location of the deposited material in the catchment. The most highly mobile faeces was deemed to be faeces from domestic cats and dogs, since this is frequently deposited on hard, impervious surfaces in urban areas, and is often washed into stormwater. These faeces were assigned a value of 0.05 indicating that 5% of the total volume of material in the catchment would be moved into the stream in a wet weather event. Rodents in the catchment have unlimited access to waterways and stormwater drains, for these reasons, rodents were also assigned a manure mobilisation rate of 0.05. The most mobile faecal material from catchment livestock are the faeces from sheep, kangaroos, goats, deer, marsupials, and rabbits, these faeces were assigned a mobilisation rate of 0.03. Less mobile faeces from cattle, pigs, feral dogs and cats, and foxes were assigned a mobilisation rate of 0.02. The least mobile faecal material deposited by animals included in the model was horse and poultry faeces which were assigned a value of 0.005.

Table 3.11 to Table 3.13 summarise pathogen and indicator concentrations in animal faeces. Each table shows the values selected for use in the model, the source of the data and the reasons for selecting these values. Where available, data from local studies performed in the Sydney drinking water catchment were selected in preference to data from studies carried out in geographically distant locations.

111

Table 3.11 Cryptosporidium oocyst concentrations in animal faeces

Animal Species Cryptosporidium Reference Comments -1 (log10 oocysts.kg.d ) Cattle, grazing and 3.7 (Australian Water Arithmetic mean of intensive Technologies, 2002b; combined data (n=29), Davies et al., 2005b) similar to Atwill et al. (2003) Sheep 4.9 (Australian Water Arithmetic mean of Technologies, 2002b; combined data (n=29), Davies et al., 2005b) similar to Ortega-Mora et al. (1999) Pigs, domestic 5.5 (Australian Water Arithmetic mean of Technologies, 2002b; combined data (n=29), Davies et al., 2005b) slightly higher than Hutchison et al. (2004) Pigs, feral 0 (Australian Water Small sample size Technologies, 2002b) (n=5), no comparative data Kangaroos 6.5 (Australian Water Arithmetic mean of Technologies, 2002b; combined data (n=36) Davies et al., 2005b) Dogs, domestic and 5.5 (Australian Water Arithmetic mean (n=9) feral Technologies, 2002b) Cats, domestic and 3.4 (Australian Water Arithmetic mean (n=8) feral Technologies, 2002b) Horses 2.9 (Olley & Deere, 2003) Small sample size (n=5), 3 log10 lower than Sturdee et al. (2003) Goats 0 (Australian Water Small sample size Technologies, 2002b) (n=3), no comparative data Deer 3.3 (Australian Water Very small sample size Technologies, 2002b) (n=1), 3 log10 lower than Sturdee et al. (1999) Other marsupials 4.5 (Australian Water Based on results for Technologies, 2002b) possums, very small sample size (n=2) Rodents 5.4 (Australian Water Based on results for Technologies, 2002b) Antechinus (n=8) Rabbits 3.7 (Australian Water Very small sample size Technologies, 2002b) (n=2), lower estimate than (Medema, 1999) and (Sturdee, Chalmers & Bull, 1999) Foxes 0 (Australian Water Only one sample tested, Technologies, 2002b) same result as Medema et al. (1999) but 3 log10 lower than Sturdee et al. (1999) Poultry 6.3 (Medema et al., 2001) Most likely not human infectious Cryptosporidium (n=16)

112

Table 3.12 Giardia cyst concentrations in animal faeces

Animal Species Giardia Reference Comments -1 (log10 cysts.kg.d ) Cattle, grazing and 5.2 (Olley & Deere, 2003) (n=10), slightly less intensive than Wade et al. (2000) Sheep 5.6 (Olley & Deere, 2003) (n=10), no comparative data Pigs, domestic 5.7 (Olley & Deere, 2003) (n=8), slightly higher than Hutchison et al. (2004) Pigs, feral 0 (Australian Water (n=5), no comparative Technologies, 2002b) data Kangaroos 0 (Australian Water No positive samples Technologies, 2002b) (n=11), no comparative data Dogs, domestic and 5.4 (Australian Water Median value (n=9), no feral Technologies, 2002b) comparative data Cats, domestic and 5.9 (Australian Water (n=8), no comparative feral Technologies, 2002b) data Horses 3.4 (Olley & Deere, 2003) (n=4), no comparative data Goats 0 (Australian Water Small sample size (n=3) Technologies, 2002b) Deer 0 (Australian Water Very small sample size Technologies, 2002b) (n=1), much lower than the 6 log10 mean of Heitman et al. (2002) Other marsupials 0 (Australian Water Based on results for Technologies, 2002b) possums, very small sample size (n=2) Rodents 0 (Australian Water Based on results for Technologies, 2002b) Antechinus (n=8), same result as Heitman et al. (2002) 0 (Australian Water Very small sample size Technologies, 2002b) (n=2), however same result as Medema (1999) Foxes 5.7 (Australian Water Only one sample tested, Technologies, 2002b) similar result to the mean of 6.1 log10 for Coyote (Heitman et al., 2002) Poultry 2.8 (Olley & Deere, 2003) (n=4), no comparative data

113

Table 3.13 E. coli concentrations in animal faeces

Animal Species E. coli Reference Comments -1 (log10 cfu.kg.d ) Cattle, grazing 9.3 (Australian Water (n=9), identical to the mean Technologies, 2002b) reported by Davies et al. (2005b) (n=20) Cattle, intensive 10.1 (Davies-Colley et al., (n=5), possible that higher 2002) density leads to slightly higher concentrations Sheep 10.4 (Australian Water (n=9), similar to the mean of Technologies, 2002b) 9.8 log10 cfu.kg reported by Davies et al. (2005b) (n=20) Pigs, domestic 10.8 (Australian Water (n=9), similar to the mean of Technologies, 2002b) 10.1 log10 cfu.kg reported by Davies et al. (2005b) (n=20) Pigs, feral 12.1 (Australian Water Small sample size (n=5), no Technologies, 2002b) comparative data Kangaroos 11.2 (Australian Water Sample size (n=11), higher Technologies, 2002b) than the mean of 8.7 log10 cfu.kg reported by Davies et al. (2005b) (n=25) Dogs, domestic 10.6 (Australian Water (n=9), similar to Geldreich and feral Technologies, 2002b) (1978) Cats, domestic 10.0 (Australian Water (n=8), similar to Geldreich Technologies, 2002b) (1978) Cats, feral 9.8 (Australian Water Very small sample size Technologies, 2002b) (n=2), very similar result to domestic cats Horses 8.7 (Australian Water (n=9), 2 log10 higher than Technologies, 2002b) reported by Geldreich (1978) Goats 11.9 (Australian Water (n=5), no comparative data Technologies, 2002b) Deer 9.3 (Australian Water Only one sample tested, no Technologies, 2002b) comparative data Other marsupials 7.2 (Australian Water Arithmetic mean of platypus, Technologies, 2002b) wombat and possum data (n=20) Rodents 7.4 (Australian Water Based on results for rats Technologies, 2002b) (n=5), similar to Geldreich (1978) Rabbits 8.7 (Australian Water Small sample size (n=4), 4 Technologies, 2002b) log10 higher than Geldreich (1978) Foxes 9.9 (Australian Water Only one sample tested, no Technologies, 2002b) comparative data Poultry 11.4 (Australian Water (n=9), 3 log10 higher than Technologies, 2002b) Geldreich (1978)

114

Due to the lack of local data on the distribution of livestock by age within the Sydney drinking water catchment this parameter was not included in the model. Table 3.14 summarises the values selected to represent the most important aspects of animal behaviour. Animal access to streams was based on wildlife having full access to waterways (access = 1) and domestic livestock having more or less access dependent on farm management practices. Domestic pigs must be housed in sheds and no discharge of effluent to waterways is allowed therefore access was 0. Animal behaviour was accounted for as the likelihood of animal species entering and defecating in the stream assuming they had unlimited access. Likelihood was assumed to vary between 0 and 1 with pigs (0.025), cattle and marsupials attributed with the highest likelihood of defecating in streams (0.01). This data will be incorporated into the mathematical model as outlined in Chapter 5.

115

Table 3.14 Animal access to streams and the likelihood of their deposition in streams and riparian zones

Animal Species Access‡ Likelihood* Comments

Cattle, grazing 0.1 0.01 Limited access (Gary, Johnson & Ponce, 1983), relatively high likelihood Cattle, 0.05 0.01 Less access than grazing animals, intensive relatively high likelihood Sheep 0.1 0.005 Same access as cattle, likelihood half that of cattle Pigs, domestic 0 0.025 No access, high likelihood if they have access Pigs, feral 1 0.025 Full access, high likelihood if they have access Kangaroos 1 0.001 Full access, low likelihood Dogs, domestic 0.2 0.001 More access than grazing animals, low likelihood Dogs, feral 1 0.001 Full access, low likelihood Cats, domestic 0.2 0.001 More access than grazing animals, low likelihood Cats, feral 1 0.001 Full access, low likelihood Horses 0.01 0.005 Usually housed away from waterways, thus ten times less access than grazing cattle, likelihood half that of cattle Goats 0.01 0.005 Usually housed away from waterways, thus ten times less access than grazing cattle, likelihood half that of cattle Deer 1 0.005 Full access, likelihood half that of cattle Other 1 0.01 Full access, relatively high likelihood marsupials Rodents 1 0.005 Full access, likelihood half that of cattle Rabbits 1 0.001 Full access, low likelihood Foxes 1 0.001 Full access, low likelihood

‡ Animal access to streams based on wildlife having full access and domestic livestock having more or less access dependent on farm management practices * Estimates of animal behaviour, increasing or decreasing the likelihood of animal species entering the stream assuming they have unlimited access

116

Chapter 4 Field-scale simulation of microbial transport from bovine faecal pats in surface waters

This chapter has been accepted as: Ferguson, C. M., Davies, C. M., Kaucner, C., Krogh, M., Deere, D. and Ashbolt, N. J. Field-scale transport of Cryptosporidium parvum, E. coli and PRD1 bacteriophage in surface water runoff from bovine faecal pats under simulated rainfall. The Journal of Water and Health (in press).

Introduction

Although previous studies have quantified catchment input concentrations of pathogens (Ashendorff et al., 1997; Chauret et al., 1995; Jellison, Hemond & Schauer, 2002; Ong et al., 1996; Xiao et al., 2001) few have quantified the relationship between pathogen inputs and the catchment processes that influence their fate and transport (Davies et al., 2004b; Tate et al., 2000; Trask et al., 2004). Animal faecal deposits to land remain a significant source of pathogenic protozoa in catchments (Atwill et al., 2003; Olson et al., 1997a) and have been linked to rainfall event-related increases in pathogen concentrations in surface waters (Ashbolt & Roser, 2003; Atherholt et al., 1998). In particular, the potential pathogen inputs from dairy and beef cattle warrant further investigation since these animals are frequently the largest and most abundant domestic livestock in drinking water catchments. They also excrete a much larger volume of manure than other domestic livestock and they can host human infectious strains of both Cryptosporidium oocysts and E. coli. Although cattle are known to excrete Giardia cysts their potential for causing human infections is still being investigated (Becher et al., 2005).

A review of catchment processes that control pathogen fate and transport identified the need to quantify pathogen transport processes at a range of scales, noting that most studies to date have relied on observations made from laboratory experiments (Ferguson et al., 2003b). These previous studies examined microbial transport using either repacked soils (Atwill et al., 2002) or intact soil columns or blocks (Davies et al., 2004b; Mawdsley, Brooks & Merry, 1996; Trask et al., 2004). However, there is a need to verify if observations made at laboratory-scale can be extrapolated to field-scale and

117

to determine mobility for a range of pathogens (bacterial, viral and protozoan). Tate et al. (2000) studied Cryptosporidium oocyst transport at field scale and determined that slope was an important factor, with oocyst transport increasing significantly with slope. The majority of microbial transport studies have examined either indicator bacteria or Cryptosporidium oocysts. Although virus transport through subsurface soil and into groundwater has been studied extensively (Jin, Pratt & Yates, 2000; Schijven et al., 2003; Schijven & Hassanizadeh, 2002), few studies have investigated the potential transport of enteric viruses in surface water runoff from failing on-site sewage systems (Charles & Ashbolt, 2004).

The quantification of pathogen and indicator fate and transport at field-scale will facilitate the development of more comprehensive hydrologic models with an increased capacity to estimate microbiological parameters at a range of scales. These experiments will examine the transport of Cryptosporidium oocysts, E. coli bacteria and PRD1 bacteriophage in surface waters over bare and vegetated soil plots 10 m in length. The inclusion of PRD1 in these experiments was based on the need to evaluate the potential for surface water contamination arising from the transport of viral pathogens from failing on-site systems. The quantification of virus transport at field-scale will facilitate the future inclusion of enteric viruses into the pathogen model.

This study examined the surface transport of Cryptosporidium parvum oocysts, E. coli and PRD1 bacteriophage released from bovine faecal pats at field-scale, using simulated rainfall events similar to those employed in previous laboratory-scale experiments (Davies et al., 2004b). The objective was to quantify pathogen transport using a range of bacterial, protozoan and viral index organisms and to examine the cumulative release of pathogens in a subsequent rainfall event after one-week field- storage at ambient temperature. The aim was to collect field-scale data on pathogen transport to facilitate the development of mathematical models to better predict pathogen fate and transport at catchment scales.

C. Ferguson and C. Davies (UNSW) were responsible for the scope, design and management of the study. C. Kaucner co-ordinated the field work and processing of samples at the UNSW laboratory, sample analysis was performed by C. Davies, C. Kaucner, P. Beatson and J. Rodehutskors. Sydney Water Laboratories at West Ryde

118

under the supervision of P. Cox and M. Warnecke performed some of the analyses for Cryptosporidium. J. Rodehutskors calibrated the rainfall simulation equipment. M. Krogh performed statistical analysis while data collation and report writing were undertaken by C. Ferguson.

Materials and Methods

Location

Field experiments were conducted in situ, adjacent to (10 m) the location of collection of the intact soil blocks used in the study by Davies et al. (2004b) at Arthursleigh Farm, Marulan. The farm is located in the Sydney drinking water supply catchment approximately 200 km south west of Sydney, New South Wales (NSW), Australia (34°33’06.2’’N, 150°03’22.6’’E). Livestock had not grazed the location for at least twelve months prior to preparation of the soil plots. However, the area was accessible to wildlife and care was taken therefore, to remove any faecal deposits on the surface of the soil prior to the experiments. The soil was characterised as loam in texture (49% sand, 27% silt and 24%clay) with soil properties as described previously (Davies et al., 2004b). The soil plots were predominantly (>70%) vegetated with the introduced grass Phalaris aquatica L. Two other grasses Avena barabata Pott. Ex Link. and Carex inversa R. Br. were also present, but at much lower densities (<10%). An unidentified daisy species also made up approximately 10% of the vegetative cover. Less abundant species were not identified.

Preparation of field plots

Two experiments were performed using separate plots prepared adjacent to each other, with each experiment consisting of three rain events (Runs) one week apart: a control run without any faecal pats in place (Runs 1 & 4), a fresh faecal pat run (Runs 2 & 5), and an ‘aged’ faecal pat run (Runs 3 & 6). The design and layout of a plot is illustrated in Figure 4.1. Each test plot was constructed by inserting metal borders directly into the soil to a depth of 10 cm, creating two separate sub-plots each 10 m by 1 119

m on land with a slope of 18 degrees. One week before the start of each experiment the vegetation was removed from one side of the plot using a turf cutter (Ryan® Jr. Sod Cutter, Jacobsen, North Carolina USA). The loose soil on the surface of the bare sub- plot was then lightly packed down using a turf roller. To simulate grazing by livestock, the vegetated sub-plot was mown to a height of 2 cm and all loose clippings removed from the plot by hand. Any missed grass stalks were trimmed using scissors and removed. The extent of vegetation cover on the vegetated sub-plots was estimated at 67% ± 8% (n=16) Davies et al. (2004b). Surface runoff was collected at 2.5 m, 5 m and 7.5 m distances (plot samples). At the base of each 10 m sub-plot surface runoff was collected by way of Replogle Bos Clemmens (RBC) flumes (flume samples). These calibrated flumes were used because they are suitable for measuring flow for open channel systems (Bos, Replogle & Clemmens, 1993).

The rainfall simulator

The field-scale rainfall simulator was designed and constructed by Dr Paul Hackney and co-workers, University of Western Sydney, NSW, Australia. The rainfall simulator was erected above the surface of the prepared soil plots to produce a rainfall intensity of 55 mm.h-1. The simulator was made up of four separate stands each capable of producing rainfall over an area of 19 m2. Each stand consisted of rotating spray nozzles mounted on a tripod. Raw source water was pumped from the Wollondilly River and mixed with rainwater in a storage tank. Water from the storage tank was pumped into a 2000 litre water tank through an in line filter that removed particles greater than 10 μm in size (CUNO Pacific Pty Ltd., Blacktown, Australia). The tank was connected via hoses to the rainfall simulator, and the rainfall intensity controlled by adjusting the pressure gauges for each stand. All raw water used for the experiments was disinfected with sodium hypochlorite (>1 ppm) for a minimum of 18 h and then neutralized with an excess of sodium thiosulfate until no chlorine residual remained, prior to each experiment. Six rainfall calibration runs were performed to test the repeatability of the 55 mm.h-1 artificial rainfall simulation events. Rainfall calibration over the entire plot area was measured for each run using a total of ten rain gauges, eight of which were directly under the nozzles and two located 1 m either side of the plot. 120

Preparation of inocula

The fresh faecal pat runs (Runs 2 and 5) required the preparation of fresh artificial faecal pats inoculated with E. coli, PRD1 bacteriophages and Cryptosporidium oocysts. An overnight broth culture of E. coli was prepared in 50 mL of Tryptone Yeast Glucose broth (Oxoid, Australia) and centrifuged at 2500 g for 10 min. The supernatant was discarded and the pellet resuspended in 50 mL of sterile ultra pure water. One millilitre of this suspension was used to inoculate each 1 kg faecal pat with approximately 109 - 1010 organisms. The Cryptosporidium parvum oocysts were purified from defatted fresh calf faeces by density gradient flotation in sucrose solution as described by Upton (1997). The stock oocyst suspension was gamma-irradiated at 90 kGy using a 60Co source (Steritech Pty Ltd., Wetherill Park, NSW, Australia). The stock oocyst suspension containing approximately 8 x 107 oocysts was used to inoculate each 1 kg faecal pat. Oocyst stock suspensions were enumerated using IFA and well slides. Phage stock suspension containing approximately 3.6 x 109 PRD1 virions, collected from the supernatant after centrifugation (2500 g) of an infected overnight Salmonella typhimurium LT2 host culture was used to inoculate each 1 kg faecal pat.

121

1 m 1 m

Top of slope

2.5 m P4 P1

1 2 3 1 2 3

P2 2.5 m P5

4 5 6 4 5 6

2.5 m

7 8 9 7 8 9

2.5 m P6 P3 Down slope

Figure 4.1 Layout of field plots situated on loam soil with a slope of 18º

Vegetated ground Bare ground Numbered moisture probes (P1 – P6) Rainfall simulator stand Flumes Numbered sample collection wells (1-9)

122

Preparation of artificial faecal pats

Cow faeces was collected in bulk (approximately 11 kg) from a dairy farm (Corstorphine Dairy, Camden, NSW, Australia) and sterilized by gamma-irradiation at 90 kGy using the 60Co source as described previously by Davies et al. (2004b). Five artificial faecal pats were prepared for each experiment by inoculating and mixing the microorganisms into 1 kg portions of sterilized cow faeces (200 mm diameter and 30 mm deep) prepared in spring-sided baking trays. The artificial faecal pats were placed in an incubator at 20°C for 6 h followed by overnight storage at ambient temperature. All faecal pats were transported to the field and two pats were placed 10 cm from the top edge of both the bare and vegetated sub-plots. Initial (time zero) concentrations of each microorganism were determined by analysing three replicate samples from the fifth artificial faecal pat within 6 hours of performing each Control Run.

Ambient and faecal pat temperatures were recorded at 15 min intervals for the duration of the experiments using iButtons™ (Maxim/Dallas Semiconductor Corp., Dallas, Texas, USA). To quantify changes in aged faecal pats during ambient field- storage, two replicate samples were removed from each aged faecal pat immediately prior to the final rainfall simulation for both experiments (Runs 3 and 6). Each replicate was analysed for moisture content, E. coli, Cryptosporidium and PRDI phage concentrations.

Simulated rainfall experiments

The rainfall intensity and duration of the simulated events were chosen from event average recurrence interval (ARI) curves provided by Sydney Catchment Authority, for a wide geographical spread of locations within the Sydney catchment. The rainfall simulator was calibrated to produce a rainfall event with an intensity of 55 mm.h-1 and duration of 30 min. These rainfall event characteristics are identical to those used by Davies et al. (2004b) in laboratory experiments and are also relevant to North American conditions based on previous rainfall simulation studies (Atwill et al., 2002; Thelin & Gifford, 1983; Trask et al., 2001). This rainfall event represented a 1 in 1 year ARI for the majority of locations selected from the Sydney catchment. 123

During each simulated rainfall event surface runoff samples were collected at three distances along the plot (plot samples) and from the base of the 10 m plot (flume samples) (Figure 4.1). Plot samples were collected from the 9 sample collection wells using buried (top level with ground) sterile 500 mL containers. Sample containers at each plot site were replaced at six min intervals. Flume samples were collected as runoff reached the 10 m distance and exited the flume, samples were collected in 10 L plastic bladders (EB407; Entapak Pty. Ltd., Dandenong, Victoria, Australia). The bladders were changed every minute and the flow rate measured. All surface runoff was collected.

The nearest rainfall gauge (HillTop-Glendusk, approx. 5 km distant) was used to monitor rainfall throughout the six-week experimental period. There was negligible natural rainfall in this time with the maximum rainfall on an experimental day of 3.5 mm recorded after the experiment on the day of run 1. The maximum rainfall recorded on any day in the six-week period was 7.5 mm, 2 days before run 5. Most days recorded rainfall of 0 mm natural rainfall. Soil moisture probes (Theta ML2x-UM-1.21 probes, Measurement Engineering Australia (MSE), Adelaide, SA, Australia) were installed on the plots as illustrated in Figure 4.1.

Sample analysis

Due to the large volumes of surface runoff collected, flume samples were stored at ambient temperature while plot samples (500 mL) were stored in ice-chests during transit to the laboratories (<3 h). All samples were then refrigerated at 4ºC until analysed. Where sample volume was sufficient, samples were analysed individually. However, due to the small volumes collected for some vegetated plot samples, a number were pooled prior to analysis. Plot and flume samples were analysed for concentrations of Cryptosporidium spp., E. coli and PRD1 phage.

Cryptosporidium oocysts in surface runoff samples were concentrated by sequential centrifugation at 2500 g for 10 min to a maximum packed pellet of 1 mL, with the exception of two samples from the vegetated sub-plot that were concentrated

124

using Hemoflow hollow fibre ultrafiltration cartridges (HF80S; Fresenius Medical Care AG, Bad Homburg, Germany) (Simmons et al., 2001). The packed pellets were then processed using IMS and IFA as described by Davies et al. (2004b). Faeces and soil samples were processed and analysed for Cryptosporidium using the method reported by Davies et al. (2003). The recovery efficiency for soil matrices was determined by adding 100 ColorSeed™ Cryptosporidium parvum oocysts (BTF Decisive Microbiology, North Ryde, NSW, Australia) to 10% of randomly selected samples. For water matrices, samples of similar turbidity were grouped together and then ColorSeed™ oocysts (BTF) spiked into 10% of samples, ensuring that each group of similar turbidity samples contained at least one spiked sample. Recovery efficiency was calculated as (NC/NI * 100) where NC was the number of ColorSeed™ oocysts counted in the sample and NI was the number of ColorSeed™ inoculated into the sample.

For enumeration of E. coli and PRD1 phage in faeces and soil, a 1 g sample was diluted in 20 mL of sterile 0.002 M sodium pyrophosphate, vortexed for 2 min then allowed to stand for 30 min at room temperature. After further vortexing for 15 s, the suspensions were allowed to settle for 10 min at room temperature and 2 mL of the supernatant withdrawn from 2 mm below the surface. This aliquot was diluted accordingly and assayed for PRD1 and E. coli, using the double agar layer technique (ISO, 1996) with the Salmonella typhimurium LT2 host and the Quantitray™ format of Colilert®-18 (IDEXX, USA) respectively. Water matrices were serially diluted if required and then assayed as described above. Bacteria and phage analyses were completed within 24 h of sample collection.

Data analysis

Statistical analysis was performed separately for plot samples and flume samples. To examine the effect of distance, the three plot samples collected at each distance were averaged and the average value used in all subsequent analyses. Due to large differences in runoff volume, data from the bare and vegetated sub-plots were analysed separately. Microbial concentrations that were less than the detection limit were assigned the value of half of the detection limit. Data generated from the field experiments were analysed using Analysis of Variance (ANOVA) and Analysis of 125

Covariance (ANCOVA) using the SAS Generalized Linear Model (GLM) procedure (Version 8.1, SAS Institute Inc., Cary, NC, USA). The Student-Newman-Keuls Test (SNK) was used to test for significant differences between means at the α = 0.05 level. In all analyses factors were assumed to be fixed and significance determined using a Type 1 error rate of α = 0.05. Residuals were examined for all analyses to assess the homogeneity of variances.

Runoff

Runoff volumes for plot samples were analysed using a three factor ANOVA with factors time (0-6, 6-12, 12-18, 18-24 & 24-30), faecal pat age (control, fresh, aged) and distance (2.5, 5 & 7.5 m). Flume samples were analysed using a two factor ANOVA with factors time (0 to 1, 1 to 2 etc.) and faecal pat age (control, fresh, aged).

Plot and flume samples

Cryptosporidium oocysts, E. coli and PRD1 bacteriophage concentrations were first log-transformed (Log10(X+1)) and then analysed using either a two or three factor ANOVA with factors time, faecal pat age and distance (plot samples only). ANCOVA was also used with the same factors, but including runoff volumes within the sub-plot as the covariate to see if differences in runoff volumes affected the decision on significance of factors. For plot samples only the fresh (2 and 5) and aged (3 and 6) runs could be compared statistically because of differences in the way in which samples were chosen and pooled together for analysis during the control runs (1 and 4).

126

Results

Rainfall simulation

At the end of each experimental run the rainfall gauges were checked to ensure that the rainfall event reached the required intensity. The average rainfall intensity for the six runs was 55.2 ± 0.91 mm.h-1. Since all surface water runoff was collected from the flumes the volume of the flume samples were used to calculate the cumulative surface runoff from each sub-plot for each rainfall event (Figure 4.2). Figure 4.2 shows that for all rainfall events there was considerably more surface runoff from the bare sub- plots compared to the vegetated sub-plots, presumably due to greater infiltration on the vegetated sub-plots. The reduced volume of surface runoff from the vegetated sub-plots limited the volume of sample available for analysis and consequently limited the number of results making statistical comparison of bare and vegetated sub-plots difficult for most parameters in the study.

On the bare sub-plots average flume runoff volumes were significantly higher (p<0.0001) from the aged faecal pat runs (8995 mL.min-1, n=60) compared to the control (4408 mL.min-1, n=57) and fresh faecal pat runs (5165 mL.min-1, n=60). This was not surprising since rainfall events were performed sequentially and thus soil moisture contents increased over the course of each experiment (Runs 1 to 3 and Runs 4 to 6). All three probes on the vegetated sub-plots started with slightly higher soil moisture contents before the experiments and retained higher soil moisture contents after the rainfall simulations than the probes on the corresponding bare sub-plots. This was expected since the presence of vegetation increases water infiltration and would improve retention of higher soil moisture content. Vegetation would also increase the size of the boundary layer thus reducing moisture loss due to evaporation. Throughout most runs, probe 6 located at the base of the vegetated sub-plots showed higher soil moisture content than the soil in the upper slopes of either sub-plot. This effect was also noted for Probe 3 on the bare sub-plot during experiment 2 suggesting that the effect was partially due to slope, and was further enhanced by the presence of vegetation. On bare soil sub-plots there was no significant variation in runoff volumes over time during rainfall events (p=0.568).

127

1000 Run 1 Bare

e ( ) 100 Run 2 Bare Run 3 Bare 10 Run 4 Bare f vo Run 5 Bare 1 Run 6 Bare 0.1 Run 1 Veg

l e runof lum L Run 2 Veg 0.01 Run 3 Veg

Cumu ativ Run 4 Veg 0.001 Run 5 Veg 0 10203040 Run 6 Veg Time

Figure 4.2 Cumulative runoff volume (L) for bare and vegetated sub-plots for simulated rainfall events (Runs 1 to 6)

Characterisation of artificial faecal pats

Arithmetic mean concentrations of microorganisms spiked into the artificial faecal pats as measured at time zero and after one-week field-storage at ambient temperature are shown in Table 5.1. The results indicate homogeneity of the organisms within the pats, and therefore the suitability of the pat preparation method. Both experiments were carried out during early winter and iButton™ measurements indicated that although ambient temperatures ranged from -2°C to 25°C, faecal pat temperatures only varied from 1°C to 14°C. The moisture content of the faecal pats decreased from 90% to between 65 and 85%. Table 5.1 shows that Cryptosporidium oocyst concentrations remained unchanged after one-week field-storage at ambient temperature. PRD1 bacteriophage concentrations all decreased during the one-week of -1 field-storage by 2 to 3 log10 pfu.kg dwt. Six of the eight faecal samples analysed for E. -1 coli concentrations showed decreases of 1 to 2 log10 cfu.kg dwt. However, two samples from Experiment 2 (one each from the bare and vegetated sub-plots) showed an increase in E. coli concentrations of approximately 1 log10. It is possible that these two replicates were taken from within the centre of the faecal pat where higher moisture retention may have contributed to enhanced survival and possible replication. The

128

majority of samples were taken from near the edge of the faecal pat to minimize pat disturbance prior to the final rainfall simulations, however the exact locations were not recorded.

Table 4.1 Microbial concentrations in fresh artificial bovine faecal pats at time zero

E. coli Cryptosporidium PRD1 phage Sample -1 -1 -1 (log10 cfu.kg ) (log10 oocysts.kg ) (log10 pfu.kg ) Experiment 1 Mean Time 0 wet wt.(n=3) 10.09 6.40 8.87 SD wet wt. 9.23 5.67 8.29 Total load Time 0 wet wt. 10.39 6.70 9.17 Mean Time 0 dry wt.(n=3) 11.09 7.40 9.87 Mean 1-week dry wt. (n=4) 10.1 6.98 7.36 Experiment 2 Mean Time 0 wet wt. (n=3) 9.73 6.53 9.20 SD wet wt. 9.47 5.97 8.64 Total load Time 0 wet wt. 10.03 6.84 9.51 Mean Time 0 dry wt.(n=3) 10.73 7.53 10.20 Mean 1-week dry wt. (n=4) 11.8 7.54 7.96 SD Standard deviation

Cryptosporidium oocysts

The recovery efficiency of Cryptosporidium oocysts was consistent for each of the matrices tested. The arithmetic mean percent recovery from water was 55.3% ± 21.3 (n=54) and for soil 63.0% ± 4.4 (n=3). A previous study determined the arithmetic mean recovery efficiency for Cryptosporidium oocysts from adult cattle faeces was 40.0% ± 16 (n=10) (Davies et al., 2003). All results were adjusted for the percent recovery prior to statistical analysis, noting that recovery efficiencies were approximate since samples were processed in batches, and ColorSeed™ was not spiked into each individual sample.

129

Table 4.2 shows the mean Cryptosporidium concentrations in surface runoff collected from the plot samples. Mean concentrations in surface runoff from fresh faecal pats were significantly higher than from the aged faecal pats on the bare sub-plots (p=<0.0001) but not on the vegetated sub-plots. On both bare and vegetated sub-plots Cryptosporidium concentrations in surface runoff increased over time during the rainfall simulations (p=0.009, p=0.018 respectively). The effect of distance was also significant on both bare and vegetated sub-plots (p=0.0002, p=0.035 respectively). Mean Cryptosporidium concentrations in the surface runoff from the bare sub-plots decreased from 3.0 log10 at the 2.5 m distance to 2.0 log10 at the 7.5 m distance. On the vegetated sub-plots the mean concentrations decreased from 1.3 log10 to 0.5 log10 over the same distance. Thus once mobilised the decrease in concentration was similar on both sub- plots however on the vegetated sub-plots the mean concentrations mobilised to the 2.5 m distance were less than half the mean concentrations on the bare sub-plot indicating that initial release and transport on the vegetated sub-plots was much lower than on the bare sub-plots.

Table 4.3 shows the mean log10 Cryptosporidium oocyst concentrations in surface runoff from the bare and vegetated sub-plots collected at the flumes. Due to fewer results being available for analysis the only significant effect for the flume samples was the higher mean concentrations in runoff from the bare sub-plots from fresh faecal pat runs compared to the control and aged faecal pat runs (p=0.0001). Flume samples from the vegetated sub-plots showed no significant differences in Cryptosporidium oocyst concentrations with either time or run type.

130

Table 4.2 Mean concentrations and SNK groupings for Cryptosporidium, E. coli and PRD1 bacteriophage for plot samples (surface runoff from bare and vegetated sub-plots after simulated rainfall of 55 mm.h-1 for 30 min)

n n C. parvum E. coli PRD1 -1 -1 -1 Bare Vegetate (log10 oocysts.L ) (log10 mpn.mL ) (log10 pfu.mL ) sub-plots d Bare Vegetated Bare Vegetated Bare Vegetated sub-plots sub-plots sub-plots sub-plots sub-plots sub-plots sub-plots Mean SNK Mean SNK Mean SNK Mean SNK Mean SNK Mean SNK Age Fresh 30 18 3.20 A 1.09 A 3.84 A 0.98 A 1.85 A 0.23 A Aged 30 18 1.80 B 0.88 A 3.09 B 0.71 A 1.15 B 0.08 A Time 0 – 6 12 - 1.81 A - - 2.30 A - - 0.49 A - - 6 – 12 12 - 2.69 B - - 3.81 B - - 1.57 B - - 12 – 18 12 12 2.80 B 0.83 A 3.80 B 0.54 A 1.96 B 0.08 A 18 – 24 12 12 2.73 B 0.60 A 3.71 B 0.90 A 1.60 B 0.11 A 24 – 30 12 12 2.48 B 1.52 B 3.72 B 1.08 A 2.01 B 0.27 A Distance 2.5 m 20 12 3.04 A 1.34 A 3.85 A 1.12 A 1.69 A 0.35 A 5.0 m 20 12 2.46 B 1.11 AB 3.56 A 0.85 A 1.45 A 0.08 A 7.5 m 20 12 2.00 C 0.51 B 2.99 A 0.55 A 1.37 A 0.06 A

- = not tested n = number of samples included in the analysis mpn = most probable number pfu = plaque forming units SNK = Student Newman Keuls test, values with different letters are significantly different at the p<0.05 level.

131

Table 4.3 Mean concentrations and SNK groupings for Cryptosporidium, E. coli and PRD1 bacteriophage for flume samples (surface runoff at 10 m distance from bare and vegetated sub-plots after simulated rainfall of 55 mm.h-1 for 30 min)

n n Cryptosporidium parvum n E. coli n PRD1 -1 -1 -1 Bare Vegetated (log10 oocysts.L ) Bare (log10 mpn.mL ) Bare (log10 pfu.mL ) Sub-plots Sub-plots Bare Vegetated Sub-plots Bare Sub-plots Bare sub-plots sub-plots sub-plots sub-plots Mean SNK Mean SNK Mean SNK Mean SNK Age Control 10 - 0.20 B - - 14 1.13 A - - Fresh 12 6 1.53 A 0.39 A 21 3.26 B 12 1.77 A Aged 11 2 0.36 B 0.72 A 22 2.95 B 9 0.86 B Time 4 - 5 5 - 0.62 A - - 6 1.75 A 3 1.34 A 7 -8 ------5 2.77 A - - - 9 - 10 6 - 1.00 A - - 6 2.80 A 4 1.54 A 12 - 13 ------5 2.82 A - - - 12 - 18 - 2 - 0.00 A ------14 - 15 6 - 0.65 A - - 6 2.60 A 3 1.57 A 17 - 18 ------5 2.59 A - - - 18 - 24 - 3 - 0.77 A ------19 - 20 6 - 0.69 A - - 6 2.66 A 4 1.42 A 22 - 23 ------4 2.48 A - - - 24 - 25 5 - 0.85 A - - 5 2.92 A 3 1.31 A 24 - 30 - 3 - - 0.50 A ------27 - 28 ------4 2.97 A - - - 29 - 30 5 - 0.61 A - - 5 2.65 A 4 1.11 A - = not tested n = number of samples included in the analysis mpn = most probable number pfu = plaque forming units SNK = Student Newman Keuls test, values with different letters are significantly different at the p<0.05 level

132

Figure 4.3 shows the cumulative total load of Cryptosporidium oocysts exported from the bare sub-plots for both experiments. The total export from the control runs (time 1 to 35) were 108 and <1 oocysts, respectively. The cumulative loads for both fresh runs (time 36 to 70) were approximately 4.5 log10 oocysts while the loads from the aged runs (time 71 to 105) were one to two orders of magnitude lower (2.8 – 3.6 log10 oocysts, respectively). The net transport of Cryptosporidium from the two experiments were reasonably consistent, the fraction of material being exported in the fresh runs was about 1% of the total load of oocysts deposited onto the plots in the faecal pats. The subsequent aged runs released approximately 0.01 to 0.1% of the initial oocyst load.

4 oocysts log10 log10 oocysts

1 Cryptosporidium 0 0 102030405060708090100110 Time

Cryptosporidium Experiment 1 Cryptosporidium Experiment 2 Fresh faecal pats

Figure 4.3 Cumulative loads of Cryptosporidium oocysts exported from 10 m bare soil sub-plots for experiments 1 and 2. Samples collected at one minute intervals; control run = samples 1-35, fresh run = samples 36-70, aged run = samples 71-105

E. coli

Table 5.2 shows the mean log10 E. coli concentrations for the surface runoff from the bare and vegetated sub-plots. The mean E. coli concentrations from the bare sub-plots was significantly higher for the fresh compared to the aged runs (p=0.029). On the bare sub-plots there was also significant variation in E. coli concentrations over

133

time during the rainfall event, with samples collected during the first time period (0 to 6 min) having significantly lower concentrations of E. coli than any of the other time intervals in the rainfall simulation (p=0.026). Although mean E. coli concentrations decreased slightly with increasing distance from the faecal pat, the differences were not significant, indicating that E. coli were easily mobilised across the surface of the bare sub-plot.

As observed for Cryptosporidium the mean E. coli concentrations in the surface runoff from the vegetated sub-plots were not significantly different between the fresh and aged runs (p=0.321). However, mean E. coli concentrations from the vegetated -1 sub-plots (0.98 and 0.71 log10 mpn.mL for fresh and aged pats) were considerably -1 lower than from the bare sub-plots (3.84 and 3.09 log10 mpn.mL respectively). On the vegetated sub-plots there was also no significant variation in E. coli concentrations over time or with increasing distance from the faecal pats.

Table 4.3 shows that the mean E. coli concentrations in flume samples from the bare sub-plots was significantly higher for both the fresh and aged runs compared to the control run (p=0.0002). There was also no significant variation in E. coli concentrations over time during the rainfall event. The limited volume of runoff collected at the flume from the vegetated sub-plots precluded the ability to undertake statistical analysis for E. coli or PRD1 phage.

Figure 4.4 shows the cumulative total loads of E. coli exported from the bare sub-plots for both experiments. The totals exported from the control runs (time 1 to 35) were 6.7 and 5 log10, respectively. The cumulative loads for both fresh runs (time 36 to

70) were approximately 8.5 log10, thus two to 3 orders of magnitude higher than the control runs. Approximately 1% of the initial E. coli load deposited in the fresh faecal pats was exported 10 m across the bare soil sub-plots to reach the flumes. The loads exported from the aged runs (time 71 to 105) varied. In experiment 1 the load exported was less than from the fresh faecal pat run (7 log10). However in experiment 2 it was slightly higher than from the fresh run (9.2 log10). Thus the loads exported from the aged faecal pats were between 0.1 and 10% of the initial E. coli load placed on the sub- plot.

134

11 10 9 8 7 6

log10 cfu cfu log10 5 4

E. coli E. 3 2 1 0 0 102030405060708090100110 Time

E. coli Experiment 1 E. coli Experiment 2 Fresh faecal pats

Figure 4.4 Cumulative loads of E. coli exported from 10 m bare soil sub-plots for experiments 1 and 2. Samples collected at one minute intervals; control run = samples 1-35, fresh run = samples 36-70, aged run = samples 71-105

PRD1 bacteriophage

The mean PRD1 phage concentrations in the surface runoff from the bare sub- plots were significantly higher for the fresh compared to the aged runs (p=0.001). There was also significant variation in PRD1 phage concentrations over time during the rainfall event, with samples collected during the first time period (0 to 6 min) having significantly lower concentrations of PRD1 phage than any of the other time intervals in the rainfall event (p=0.009). However, in contrast to Cryptosporidium, the mean PRD1 phage concentrations were very similar at all distances down the bare sub-plots, indicating that PRD1 was highly mobile in the surface runoff (Table 5.2). On the vegetated sub-plots the mean PRD1 phage concentrations in the surface runoff were not significantly different between the fresh and aged runs (p=0.262) nor were there any significant differences over time or with distance. However, the mean concentrations from the vegetated sub-plots were considerably lower than from the bare soil sub-plots.

Table 5.3 shows the mean log10 PRD1 phage concentrations for flume runoff collected from the bare sub-plots indicating that concentrations from the fresh runs were significantly higher than from the aged runs (p=0.049). As with Cryptosporidium and

135

E. coli there was no significant variation in PRD1 phage concentrations over time during a rainfall event.

Figure 4.5 shows the cumulative total loads of PRD1 phage exported from the bare sub-plots for experiments 1 and 2. The total exports from the control runs (time 1 to 35) were approximately 3 log10 and <1, respectively. The cumulative loads for both fresh runs (time 36 to 70) were 7 and 6 log10, respectively. The loads from the aged faecal pat runs (time 71 to 105) were 6 and 4 log10 respectively. The fraction of material exported from the fresh faecal pats ranged from 0.1 to 1% of the total PRD1 phage deposited onto the plots. However the load exported one-week later from the aged faecal pats was more variable ranging from 0.005 to 0.1% of the initial total load of PRD1 phage.

10 9 8 7 6 5 4 3

PRD1 phage log10 pfu pfu log10 phage PRD1 2 1 0 0 102030405060708090100110 Time

PRD1 phage Experiment 1 PRD1 phage Experiment 2 Fresh faecal pats

Figure 4.5 Cumulative loads of PRD1 phage exported from 10 m bare soil sub- plots for experiments 1 and 2. Samples collected at one minute intervals; control run = samples 1-35, fresh run = samples 36-70, aged run = samples 71-105

Discussion

Large-scale field experiments to examine the transport of microorganisms in water, soil and groundwater are logistically difficult as well as time consuming and labour intensive. They also become more difficult when screening for a range of

136

microbial species to elucidate relationships between surface characteristics and microbial transport (Ferguson et al., 2003b; Jewett et al., 1995). The majority of studies conducted have examined transport of microorganisms through groundwater and comparatively few studies have examined pathogen transport in surface waters (Atwill et al., 2002; Mawdsley et al., 1996; Tate et al., 2000; Trask et al., 2004). Davies et al. (2004b) examined the release of Cryptosporidium oocysts from faecal pats and their transport through and over intact soil blocks under different simulated catchment characteristics of slope, vegetation status and rainfall event intensity/duration. In laboratory (Davies et al., 2004b) and field experiments (current study), the volumes of surface runoff from the vegetated plots were much lower than on the bare plots. However, the extent of runoff reduction due to the presence of vegetation in this study was much higher than anticipated making statistical comparisons difficult due to the reduced volumes of sample available for analysis. This does however highlight the role of increased infiltration and vegetation cover as a mechanism for retarding pathogen transport.

The volume of surface runoff increased with distance down the plot, which is not surprising given the slope (18º). The significantly higher volume of runoff generated from the aged faecal pat runs, compared to the control and fresh runs, was expected due to sequential use of the plots. Soil moisture content increased rapidly by (5 – 15%) with each artificial rainfall simulation and would plateau at a level usually 5 - 10 % higher than the pre-rainfall event level with the increased soil moisture contributing to rapid soil saturation and hence surface runoff. It should be noted that plot surface runoff samples were collected in 500 mL containers that were capped once filled, thus the runoff volumes for plot samples were not accurate reflections of total surface runoff volumes. However, since all surface runoff was collected at the flumes statistical analysis of runoff volumes was performed using the flume sample data collected from the 10 m distance. Capping of plot sample containers may also have biased the estimation of microbial concentration data for these samples. However, significant differences over sample time was only evident for samples collected in the 0- 6 minute time interval which was always significantly lower than all of the subsequent time intervals (Table 4.2). The exception was the estimate of Cryptosporidium oocyst concentrations in surface runoff from the vegetated sub-plots at the 24-30 min. The mean concentration in these samples was significantly higher than for the previous time

137

intervals (p=0.018) suggesting that Cryptosporidium oocyst transport on the vegetated sub-plots was just starting to increase as the event was concluding. Since several of the plot samples for this time interval were recorded as 500 mL volumes it is possible that -1 the mean concentration 1.52 log10 oocysts.L was underestimated.

Microorganism concentrations in the faecal pats were subject to change during the one-week exposure in the field between the fresh and aged runs. However, temperature variation within the faecal pats was buffered by up to 10°C compared to the ambient temperature and only PRD1 phage showed significant decay between fresh and aged runs. Two samples showed an increase in the concentration of E. coli between the fresh and aged runs, a phenomenon similar to that reported by Robinson et al. (2004). On rangelands full-sized pats (2 to 3 kg) are crusted within 48 hours and dry throughout in 15 days (Thelin & Gifford, 1983) suggesting that desiccation of undisturbed pats may lead to significant decreases in Cryptosporidium oocyst concentrations. However, Kemp et al. (1995) found steady levels of oocysts throughout the year in drainage from pasture improved by manure spreading, suggesting enhanced survival in manure slurry compared to intact faecal pats.

Concentrations of Cryptosporidium oocysts were higher in the surface runoff from the bare sub-plots than from the vegetated sub-plots, where little runoff was generated even after 30 min of rainfall simulation. On both bare and vegetated sub- plots Cryptosporidium was the only organism that showed significant reductions in mean concentrations with increasing distance from the faecal pats. Despite this, the net transport of Cryptosporidium oocysts was still 4.5 log10 over 10 m of bare soil with a slope of 18°. Interestingly, using the same rainfall intensity and duration (55 mm.h-1 for

30 min) the previous study by Davies et al. (2004b) observed a 4.5 log10 export of Cryptosporidium over 1m. However, the slope of the soil in the previous study was only 10°, suggesting that a doubling of the slope increased the distance of Cryptosporidium oocyst transport on bare soil by an order of magnitude. Tate et al. (2000) demonstrated that Cryptosporidium was released from small model calf pats for at least six weeks in a Californian rangeland with a Mediterranean climate (springtime, intermittent storms), though in steadily lesser quantities.

138

While Cryptosporidium is considered an important human waterborne pathogen and E. coli provides a link to previous studies, the relevance of PRD1 bacteriophage in the scenario of release and transport of pathogens from animal faeces is not immediately obvious. The main risks/sources of human enteric viruses in catchments are sewage discharges and septic seepage (Charles et al., 2003a) rather than animal faecal deposits, since most enteric viruses tend to be very host-specific; with the possible exceptions of hepatitis E virus (Pina et al., 2000) and Norovirus group II (Bull et al., in preparation). The decision to include the bacteriophage PRD1 in this study was made based on the following rationale: PRD1 is similar in size (60-80 nm) and characteristics to many animal enteric viruses, particularly adenoviruses. It may therefore model the transport of animal viruses such as adenoviruses (Maluquer de Motes et al., 2004) or bovine enteroviruses (Rothwell et al., 2004) from faecal deposits in a catchment. The use of PRD1 in the present study is also relevant as a model for surface transport of human viruses from septic seepage (Nicosia, Rose & Stark, 2001), given its UV resistance and general persistence (Gerba et al., 2003).

Both E. coli and PRD1 phage were easily mobilised in the surface runoff and showed no decrease in concentration with increasing distance from the faecal pats on either bare or vegetated sub-plots. These results are consistent with field experiments by Bales et al. (1995), which showed that the bacteriophage PRD1 could travel considerable distances (12 m) over prolonged periods of time (25 d) without losing infectivity. Although PRD1 phage was highly mobile, the concentration of PRD1 phages in runoff from the vegetated plots was at least one log lower than the concentrations on the bare sub-plots. This is probably due to the reduction in actual runoff volume, rather than a decrease in the mobilisation rate, since it is clear that the viruses move easily with the water fraction. These results have major implications for the effective use of vegetated riparian buffer zones as a means of reducing pathogen transport in surface waters. Since many studies only examine bacterial and protozoan transport, it is likely that these studies will underestimate the potential for the transport of enteric viruses to surface waters.

For each of the three microorganisms used in this study, successive rainfall events were able to remobilize and transport microorganisms from the faecal pats into surface runoff. Only PRD1 phage showed significant decay between the fresh and aged

139

runs and substantial microbial loads remained in the faecal pats (Table 5.1). Although, the concentrations present in the runoff represent a substantial source of contamination to receiving waters, the proportion transported in each rainfall event was only a small fraction of the pathogens deposited in the faecal material (~ 1%).

Conclusions

Both E. coli and PRD1 phage were easily transported with surface runoff over 10 m distances on the bare soil sub-plot. The transport of Cryptosporidium oocysts was significantly reduced with increasing distance and mean concentrations were much lower on vegetated than bare soil sub-plots. The data generated in these experiments will be used to estimate animal manure mobilisation rates for the mathematical model described in the next chapter. Future work could include a comparison of different manure matrices to determine whether initial release and subsequent transport is affected by the composition and structure of the manure matrix.

140

Chapter 5 Construction of a Pathogen Budget: Case study in the Wingecarribee catchment

Parts of this chapter have been published as: Ferguson, C. M., Ashbolt, N. J. and Deere, D. A. (2004) Prioritisation of catchment management in the Sydney catchment - construction of a pathogen budget. Wat. Sci. Tech.: Wat. Suppl. 4(2): 35-38. Parts of this chapter have been submitted as: Ferguson, C. M., Croke, B. F. W., Beatson, P. J., Ashbolt, N. J. and Deere, D. A. (2004) Development of a process-based model to predict pathogen budgets for the Sydney drinking water catchment. J. Wat. Health .(submitted-a).

Introduction

The Sydney Catchment Authority (SCA) was formed in 1999 to manage the catchment providing the bulk water supply to Sydney, Australia, a city of four million people. Its formation was a direct result of the detection of Cryptosporidium and Giardia at levels of concern in Sydney’s water supply in August 1998. The incident was precipitated by a series of very large rainfall-runoff events within the catchment following a prolonged period of drought. The pathogens Cryptosporidium and Giardia are parasitic protozoa that cause gastrointestinal illness. Illness is usually mild and of short duration, but infection in immunocompromised hosts can become persistent with cryptosporidiosis potentially causing mortality. Community disease surveillance data indicated that there was no detectable increase in the reported level of illness during the period of the incident, suggesting that the genotypes of both Cryptosporidium and Giardia present in the supply were non-infectious for humans. Neither the sources nor the genotype of the pathogens were identified. However, it was confirmed that poorer water quality arising from catchment runoff had short-circuited the main reservoir and reached the treatment plant in a much shorter time than previously predicted by the retention time for the storage (Hawkins et al., 2000).

The SCA developed a strategy to identify and prioritise the research and management needs within the catchment. An important component of this strategy was to reduce the risk of pathogen contamination by implementing a multi-barrier approach to the protection of raw drinking water quality. The development of a conceptual model and a review of existing information identified some significant knowledge gaps

141

regarding the fate and origin of pathogens in drinking water catchments (Ferguson et al., 2003a). Preliminary work focused on the identification of the key processes that govern pathogen fate and transport in surface waters (Ferguson et al., 2003b) enabling the SCA to prioritise its pathogen research program. The knowledge needs were identified as follows (1) Characterisation of sources within the catchment, including domestic, native and feral animals, and septic seepage (2) Quantifying the processes that determine pathogen fate and transport (3) Development of tracing and tracking tools to identify the origin of faecal contamination and (4) development of a model for the estimation of catchment pathogen loadings from various sources over time and space. The aim of this last objective is to integrate all of the research information into a single modelling tool that will enable catchment managers to prioritise pathogen control measures within catchments.

The development of a model requires detailed knowledge and understanding of the sources of microorganisms, the processes that influence their mobilisation and transport, and the factors that cause their inactivation and/or loss. However, few studies have attempted to quantify the source, fate and transport of pathogens entering the surface waters of catchments, with most assessments to date relying on estimates for faecal indicator bacteria (Fraser, Barten & Pinney, 1998; Jenkins et al., 1984; Tian et al., 2002). Therefore, it is not surprising that a high priority for water utilities is to develop a methodology for quantifying pathogen hazards in catchments that enables them to spatially identify pathogen “hotspots” and target the implementation of effective barriers and control measures. Similarly, Walker et al. (1990) developed a probabilistic model that used Monte Carlo simulation to combine selected deterministic relationships with statistical information on rainfall and temperature. The model (COLI) predicted faecal coliform bacteria concentrations in surface runoff resulting from a storm presumed to occur immediately after animal manure was applied to land surfaces. In a later study Fraser et al. (1998) used a geographical information system (GIS) based hydrologic model (SEDMOD) to estimate the load of faecal coliforms in streams for several sub-catchments of the Hudson River in the state of New York. However, all of these models were limited to predicting transport of faecal indicator bacteria.

142

This reliance on faecal indicator bacteria may lead to an underestimation of pathogen risks given the frequent lack of correlation between the presence of indicator bacteria and of pathogens, and the differences in their fate and transport characteristics. Walker and Stedinger (1999) developed a model that accounted for pathogen loading from diffuse pollution to predict Cryptosporidium concentrations in the raw water supplied to New York City from the Catskill-Delaware catchment. They based their model on a generalized watershed loading function (GWLF) model (Haith & Shoemaker, 1987), utilising first-order decay functions to estimate oocyst decay in manure and in water. However, the model did not include estimates of surface washoff or the release of pathogens from the faecal matrix. A more recent study in the Netherlands modelled the discharge of Cryptosporidium and Giardia into surface water and the dispersion into rivers and streams using an emission model (PROMISE) and a dispersion model (WATNAT) (Medema & Schijven, 2001). The authors noted however that this model was unable to account for the impact of diffuse agricultural pollution and was thus primarily a point source and dispersion model. Several other faecal indicator models have also been developed recently (Collins & Rutherford, 2004; Crowther et al., 2003; Tian et al., 2002) and at least one other pathogen model is currently under development (Dorner, Huck & Slawson, 2004). None are yet commercially available.

This study describes the development of a process-based mathematical model or pathogen catchment budget (PCB) to quantify pathogen and faecal indicator loads within catchments. The model is based on a conceptual model that identified key processes for microbial sources and transport within drinking water catchments (Ferguson et al., 2003a). The model uses a mass-balance approach and will be used to construct pathogen and faecal indicator budgets for all of the SCA sub-catchments. The model will predict the total loads generated and the total loads exported from each sub- catchment for the pathogens Cryptosporidium and Giardia and the faecal indicator E. coli. Inputs to the model include land use data and catchment specific information to predict pathogen loads.

C. Ferguson (SCA), P. Beatson (Ecowise) and B. Croke (Australian National University) were responsible for the conceptual design of the model. C. Ferguson reviewed and collated all of the published information and analysed the new data for use

143

in the model. C. Ferguson and B. Croke designed the structure and components of the model into 5 modules. B. Croke coded the model in ICMS and FORTRAN program languages. C. Ferguson prepared the publications with input from the co-authors.

Development of a pathogen catchment budget (PCB)

A pathogen budget can be constructed by quantifying the primary stocks and flows and their subsequent loss as secondary stocks and flows as outlined in the equation below:

Yield = Primary Stock - Primary flows - Secondary Stock - Secondary flows (3)

Stocks are sources and accumulations of contaminants (pathogens or faecal indicators). The state of a catchment is best described in terms of stocks because they modulate the behaviour of the catchment by accumulating the difference between inflows and outflows to a particular part of a catchment or a particular process. By this means stocks create delays because they provide inertia to the system. Stocks therefore are the cause of disequilibria in a catchment by decoupling rates of flow. Flows are the rates of increase or decrease in stocks. The net flow into a stock is the rate of change of the stock. Inflow(s) are the quantity of the inflow at any time between the initial time

(to) and the current time (t). For example, a reservoir accumulates a stock of water (or pathogens) at a rate given by the difference between its inflows and outflows, beginning with an initial stock (to). If the inflow and outflow are constant, so too is the stock. If the outflow is greater than the inflow, the stock will be reduced, until eventually the initial stock (to) is also reduced to a new value which is in equilibrium with the inflow and outflow; assuming that the inflows and outflows are constant. If, however, the stock alone is known, there is a very large range of variation in inflows and outflows that can produce a particular behaviour of the stock. Information about inflows and outflows cannot therefore be derived from stocks alone.

A pathogen budget can be an iterative tool with the first phase addressing the stocks of total pathogen units (TPU) and their flows within the catchment per unit of time (such as proportion of yearly stock per rainfall event). The second phase assesses

144

the proportion of the total stock that represents infectious pathogen units (IPU) capable of causing infection and/or illness in humans. The primary sources of pathogen stock within catchments are (1) animals, including domestic, native and feral species and (2) human faecal effluent, including STPs and on-site treatment systems. Using the conceptual model to identify the dominant stocks and flows within a catchment determines which factors need to be quantified in a mathematical model to calculate the TPU budget.

Budgets need to also address the issue of scale since a catchment can be decomposed into any size, as required for a particular problem or issue. For example, a stock and flow analysis could be performed for an individual development application in the Warragamba Catchment, or for the entire Warragamba Catchment. In the case of the whole catchment, the sink may be Warragamba reservoir, or stream sediment upstream of Warragamba Dam. For a development application for a townhouse, for example, on a hillside, the source may be an on-site system, while the sink may be the creek at the bottom of the hill. Temporal scale is also important since many budgets are mean annual accounts, averaged over decades to smooth out variability in individual flow events. However, for acute-acting hazards such as pathogens time series should be constructed from flows and therefore for changes in stocks. Such time series can be derived from monitoring known relationships with events, for example rainfall. Although stocks change only by inflows and outflows, stocks also determine flows. For example, the stock of pathogens can become so depleted by the absence of domestic animals on a paddock that the flow (transport of pathogens to the stream) decreases the concentration of pathogens in the water. The stock controls the flow, with the flow rate in individual events being a function of both the stock size and the magnitude of the transporting event. Feedbacks and the resulting dynamics are only clearly seen where flow (and stock change) time series data are available. Feedbacks in highly averaged budgets can be inferred but not readily analysed.

145

By decomposing a river network into links or into mega-links between large river junctions (so-called nodes), the stock and flows of the river and adjacent landscape and point sources of contaminants can be analysed as a spatially distributed system. By mapping, and storing in a GIS, the river network point sources and land uses, spatially distributed stock and flow analysis will provide managers with a significant input to decision rules. But even at a lower spatial resolution, such as ranking major sub- catchments in terms of their input of contaminants to downstream channels, is of use. The calibration of a pathogen budget with sub-catchment spatial resolution can be achieved by collecting water quality monitoring data during dry and wet weather events and by investigating sources of contamination using molecular tracing and tracking tools.

Model description

The model was based on the earlier conceptual model that represented the primary sources of microorganisms within catchments as animals, including domestic and wildlife species, and human wastewater, including sewage treatment plants (STPs) and on-site septic systems (Ferguson et al., 2003a). The importance of these sources is mediated by many factors including host prevalence and excretion rates, inter-host transfer rates, the number of animals and the amount of faecal material generated. Transport mechanisms include natural processes. Rainfall intensity and duration, for example, affect surface runoff. Connectivity or proximity to waterways affects delivery of pathogens that have been mobilised. Some of these processes also contribute to microbial inactivation and further complexity arises from the variable effect of some factors that may either increase or decrease microbial loads, for example soil type. Secondary processes that generate microorganisms within a catchment are often human- generated, such as farm management and urban development. But natural processes such as sediment re-suspension may also be significant. Physical factors such as soil type, slope and distance from sources to waterways determine the spatial location of microbial stores within catchments. Secondary sources of microorganisms in catchments include sediment in watercourses, dams and ponds, ephemeral streams and wetlands. Secondary processes that transport microorganisms include sediment resuspension, overflow of ponds, dams and wetlands, interflow, surface runoff, manure

146

spreading and the introduction of new animal species to the catchment. The sources and processes affecting microbial survival and transport within catchments are shown in Table 5.1.

147

Table 5.1 Summary of stocks and flow processes for total pathogen unit budget (TPU)

Primary Stock Primary flow processes Effect Stock Secondary flow processes Effect STP discharges Temperature ↓ Channel sediment Overflow ↑ On-site systems Surface runoff ↑ Retention ponds Animal access ↑ Domestic animals Rainfall ↑ Dams Stormwater runoff ↑ Feral animals Distance / Time ↓ Soil Leakage of On-site systems ↑ Native animals Ultraviolet radiation ↓ Wetlands Manure spreading ↑ Visible light ↓ Ephemeral ponds Riparian buffer zones ↓ Settling rate ↓ Food chain Manure/effluent treatment ↓ Competition & predation ↓ Human access ↑ Soil type ↑ or ↓ Tillage practices ↑ or ↓ Ammonia ↓ Interflow ↑ Preferential flow paths ↑ or ↓ Sediment resuspension ↑ Vegetative matter ↑ New host arrives ↑ Freeze/thaw ↑ or ↓ Imported animal feed ↑ Natural organic matter ↑ or ↓ pH ↑ or ↓ Moisture content ↑ Aggregation ↑ or ↓ Nutrients ↑ Bold text indicates those factors that were considered most significant and thus were included in the mathematical model

148

Not all of these processes need to be quantified to derive an initial pathogen budget. The approach taken was to assess the relative importance of the various processes and to assign mathematical representations to the main sources and processes. The assessment was based on prior knowledge, but the precise structure of the model was also dependent on the availability of the input data. The prior knowledge includes the literature review (Chapter 2) and the new data (Chapters 3 and 4). The Sydney drinking water incident of 1998 and a number of previous studies highlighted the relationship between rainfall events and increased concentrations of pathogens in surface waters (Atherholt et al., 1998; Ferguson et al., 1996; Roser et al., 2003), sometimes, but not always, associated with the occurrence of waterborne disease outbreaks (Curriero et al., 2001). Additionally, a study by Davies et al. (2004b) demonstrated that the mobilisation and transport of pathogens in the Sydney catchment was significantly influenced by the occurrence of intense, short duration wet weather events. Hence the model was constructed to predict pathogen and indicator outputs for the predominant conditions in the catchment defined as “dry” (<5 mm in 24 h), “intermediate wet weather” (30 mm in 24 h) and “wet weather flood events” (100 mm in 24 h). It should be noted that these definitions are adjustable within the PCB model.

The PCB model consists of 5 components: a hydrologic module, a land budget module, an on-site systems module, a sewage treatment plant (STP) module and an in- stream transport module. The model is coded using the Interactive Component Modelling System (ICMS) software freely available from the Commonwealth Scientific Information and Resource Organisation (CSIRO). The software can be requested from the website (www.clw.csiro.gov.au/products/icms).

Hydrologic module

The hydrologic module uses the non-linear loss module of the IHACRES rainfall-runoff model described by Croke and Jakeman (2004). Briefly, this model assumes an initial catchment moisture deficit and using the distribution of surface rainfall (GIS layer), an amount of rainfall is converted into a depth of effective rainfall (rainfall that ends up as streamflow) for each sub-catchment. The effective rainfall is

149

used to estimate the wet weather mobilisation of faeces that have been deposited on the land (as described in the land module).

The effective rainfall for sub-catchment l (Ul ) is then given by:

− l tr ll (1−+= emrU ) for < tm

[]l ()−−− ttmr l −+= temr for t +<≤ rtm l (4)

= 0 for +> rtm l

where m is the initial catchment moisture deficit, t is the flow threshold and rl is the even t rainfall depth for sub-catchment (l), which is given by the mean event rainfall depth r, and the spatial variation in rainfall fl.

Assumptions

1. The catchment moisture deficit (m) describes the moisture deficit prior to a wet weather event (an input value). 2. All sub-catchments used the same initial moisture deficit and model parameters. However, rainfall surface modifies the event rainfall for each

sub-catchment by the factor (fl). 3. The drainage threshold (t) is the moisture deficit (saturate water content – actual water content) of a sub-catchment before runoff and is generated for a wet weather flood event. 4. The depth of effective rainfall (U) depends only on the amount of rainfall and the soil moisture. 5. The antecedent dry period is adjustable (30 days used in this study). 6. The amount of rainfall is adjustable (30 mm and 100 mm for intermediate and large events respectively, in the current simulations).

Land module

The number of microorganisms leaving the sub-catchment is summed over all animal species present in the sub-catchment. Animal species are assigned as present or absent for a particular land use at a defined density. Animal density per sub-catchment

150

is calculated from the GIS layers using the land use categories described in Table 5.2 and the animal density data shown in Table 5.3. Faecal material deposited on the land surface decays at the rate for microbial inactivation in soil. Faecal material, mobilised to the stream in wet weather or deposited in the stream, decays at the inactivation rate for each microorganism in water. Decay is calculated based on the estimated travel time to reach the sub-catchment outlet.

Table 5.2 Land use categories for the model derived from the Sydney Catchment Authority’s geographic information system layers

Land use Land use type SCA GIS Land use layer category (λ)

1 water water

2 improved pasture cattle Agriculture – Unimproved Pasture, Intensive Pasture & 50 % of Improved Pasture

3 improved pasture sheep 50 % of Agriculture Improved Pasture

4 commercial & industrial Urban (Industrial, Commercial, Builtup Areas), Transport (airfield, highway, roads & railway), mining (Mines, Quarries, Tailings), Lands – Degraded

5 intensive animals cattle

6 intensive animals dogs

7 intensive animals pigs

8 intensive animals poultry

9 intensive plants Horticulture – Orchard, Pollution (Biosolids disposal)

10 urban residential Urban – Residential, Pollution – Waste disposal, Water filtration & STPs

11 rural residential Urban (Rural Residential, Open Space Recreation, Environmental Protection), Recreation – Golf course, Heritage - cemetery

12 forestry with native fauna

13 forestry with native and Plantation (Native vegetation, New and Old feral fauna softwood growth, Cleared), Vegetation (Forest or Woodland, Heath, Rainforest, Sparse, Wetland)

151

Table 5.3 Animal density by land use categories

Animal Species Type† Access‡ Likelihood* Density Land use category (λ)§ (km2) 1 2 3 4 5 6 7 8 9 10 11 12 13 Cattle grazing 0 0.1 0.01 500 0 1 0 0 0 0 0 0 0 0 0 0 0 Cattle intensive 0 0.05 0.01 2000 0 0 0 0 1 0 0 0 0 0 0 0 0 Sheep 0 0.1 0.005 500 0 0 1 0 0 0 0 0 0 0 0 0 0 Pigs domestic 0 0 0.025 5000 0 0 0 0 0 0 1 0 0 0 0 0 0 Pigs feral 1 1 0.025 1 0 0 0 0 0 0 0 0 0 0 0 0 1 Kangaroos 1 1 0.001 200 0 0 0 0 0 0 0 0 1 0 0 1 1 Dogs domestic 0 0.2 0.001 400 0 0 0 0 0 1 0 0 0 1 1 0 0 Dogs feral 1 1 0.001 0.25 0 0 0 0 0 0 0 0 0 0 0 0 1 Cats domestic 0 0.2 0.001 400 0 0 0 0 0 0 0 0 0 1 1 0 0 Cats feral 1 1 0.001 1 0 0 0 0 0 0 0 0 0 0 0 0 1 Horses 0 0.01 0.005 3 0 0 0 0 0 0 0 0 0 0 1 0 0 Goats 0 0.01 0.005 2 0 0 0 0 0 0 0 0 0 0 1 0 0 Deer 1 1 0.005 2 0 0 0 0 0 0 0 0 0 0 0 0 1 Marsupials 1 1 0.01 20 1 0 0 0 0 0 0 0 1 0 0 1 1 Rodents 1 1 0.005 50 0 0 0 1 0 0 0 0 1 1 1 1 1 Rabbits 1 1 0.001 50 0 0 0 0 0 0 0 0 1 0 0 0 1 Foxes 1 1 0.001 5 0 0 0 0 0 0 0 0 1 0 0 0 1 Poultry 0 0 0.01 5000 0 0 0 0 0 0 0 1 0 0 0 0 0

† = wildlife species are represented as 1, domestic animals as 0 ‡ = animal access to streams based on wildlife having full access and domestic livestock having more or less access dependent on farm management practices * = estimate of animal behaviour increasing or decreasing the likelihood of animal species entering the stream assuming they have unlimited access § = signifies whether the animal species is present (1) or absent (0) for this land use category

152

In dry weather, the only linkage between the land budget module and the in- stream transport module was through direct input into the stream (i.e. animals defecating directly into the stream). This is calculated based on an estimate of the access to streams (wild animals have unrestricted access, domesticated animals may be prevented from accessing streams). In addition to access, an estimate of the likelihood of a particular species defecating into the stream is included. The number of microorganisms entering the stream is given by:

a ,1, lj = ∑ ,, XdDPAI ssssjls (5) s=1

where Ij,1,l is the input to stream of microorganism j from animal sources for sub- catchment l, As,l is the number of animals of species s in sub-catchment l, Pj,s is the concentration (in microorganisms/kg) of microorganism j in the faecal material of animal species s, Ds is the probability of species s defecating directly into a stream, ds is the amount of manure produced (kg/day/animal) by animal species s, Xs is the access to streams for species s.

Assumptions

1. Any faeces deposited directly into the stream become available for transport; i.e. the faeces disperse relatively quickly. 2. The concentrations of microorganisms in manure, and the manure production rates were estimated from previous studies and fieldwork (Cox et al., 2005; Davies et al., 2005b) (Table 5.4, and chapter 3, section 3.6). 3. Access to streams was assumed to be 100% for native and feral animals. Access of domestic animals was based on local knowledge and field observations. e.g. domestic pigs within the catchment must be housed in barns thus access to streams was 0. However, cattle were free range so access was set at 0.1 based on the observations of Gary et al. (1983). 4. The likelihood of direct faecal deposition to streams was estimated at 1% for cattle and 2.5% for wild pigs.

153

5. Material deposited to land is assumed to start to decay after one day and decay rates were based on microorganism inactivation rates for soil (Table 5.5).

The model uses decay rates expressed as the proportion surviving per day (δj,i) for microorganism j in material i. The instream decay module uses the decay rates for water (δj,1), while the manure deposited to land decays according to the δ values for survival in soil (δj,2). Inactivation rates can also be reported as the rate of inactivation per day (k values). Davies et al. (2005b) reported inactivation rates for Cryptosporidium in soil ranged from 0.0135 to 0.0151 at 20ºC while in manure they ranged from 0.0107 to 0.0234 (at 20ºC). As these values were not substantially different and because manure deposited to land will often be trampled and mixed into the soil, we decided to use an averaged k value of 0.02 based on these and other studies. The k values for E. coli in water and soil are calculated using data from a number of previous studies (Crane & Moore, 1986; Crane, Westerman & Overcash, 1980; Khatiwada & Polprasert, 1999; Lau & Ingham, 2001; Medema, Bahar & Schets, 1997a; Sherer et al., 1992; Stoddard, Coyne & Grove, 1998; Trevisan, Vansteelant & Dorioz, 2002). The k values for Giardia in water and soil are also calculated using data from a previous study (Anderson et al., 1998).

The wet weather budget includes the build up of material on the land, and the likelihood of mobilisation to the stream. The build up of the store of microorganisms on the land depends on the length of the antecedent dry period, the assumed storage at the start of the antecedent dry period, and the decay rate for each microorganism in soil.

a 2 1− M δ js 2, ,1, lj ∑ MI s ()()−−= []l /exp1 0 PAUU ,, sjls (6) s=1 1− δ j 2,

where Ms is the fraction of faeces for animal species s on land that would be transported to stream in a large rainfall-runoff event, Ul is the effective rainfall generated in sub-catchment l, and U0 is the scale factor for event impact.

154

Assumptions

1. Mobilisation rate of manure assigned to each species is a considered estimate based on the size, shape and consistency of faecal material (Table 5.4, see also chapter 3, section 3.6).

2. Mobilisation varied with effective rainfall Ul as outlined in equation (6). 3. Microorganisms remaining after the preceding event correspond to a full

mobilisation (Ul >>U0) of a store initially at the equilibrium value.

4. U0 is assumed to be 50 mm (adjustable within the model).

Table 5.4 Microbial concentrations in manure and manure characteristics

Animal species Manure Manure Cryptosporidium Giardia E. coli -1 -1 (s) (kg.d ) mobilisation (log10 (log10 (log10 cfu.kg.dP )P rate# oocysts.kg.d-1) cysts.kg.d-1) Cattle grazing 27.25 0.02 3.7 5.2 9.3 Cattle intensive 27.25 0.02 3.7 5.2 10.1 Sheep 1 0.03 4.9 5.6 10.4 Pigs domestic 6.2 0.02 5.5 5.7 10.8 Pigs feral 6.2 0.02 0 0 12.1 Kangaroos 0.2 0.03 6.5 0 11.2 Dogs domestic 0.5 0.05 5.5 5.4 10.6 Dogs feral 0.5 0.02 5.5 5.4 10.6 Cats domestic 0.2 0.05 3.4 5.9 10.0 Cats feral 0.2 0.02 3.4 5.9 9.8 Horses 23 0.005 2.9 3.4 8.7 Goats 1 0.03 0 0 11.9 Deer 1 0.03 3.3 0 9.3 Marsupials 0.2 0.03 4.5 0 7.2 Rodents 0.01 0.05 5.4 0 7.5 Rabbits 0.2 0.03 3.8 0 8.7 Foxes 0.3 0.02 0 5.7 9.9 Poultry 0.12 0.005 6.3 2.8 11.4

# These values are estimates for which no empirical data is yet available

155

Table 5.5 Microorganism characteristics

Microorganism δ : Proportion of the βj : Fj : Fraction bound (j) initial population Microorganisms instream to surviving.d-1 (%) excreted.person-1.d- sediment (%) 1

(log10) Soil Water On-site effluent Water Cryptosporidium 0.9550 0.9772 3.875 5 Giardia 0.6310 0.7943 4.114 5 E. coli 0.7079 0.4266 10.000 50

Sewage treatment plant module

Selection of sub-catchments connected to STPs was based on proximity to a STP, and spatial connection of urban areas. STP connectivity was calculated based on the proportion of the total population located in urban land use areas (λ=10) compared to the total sub-catchment population. In urban areas 98% of the population was assumed to be connected to the STP. In dry and intermediate wet weather conditions the number of microorganisms entering the stream is given by:

,3, = VcnI ,ljllj (7)

where Ij,3,l is the input to stream of microorganism j from the STP in sub- catchment l, nl is the population connected to a sewage treatment plant in sub-catchment l, V is the volume of water used per person per day and cj,l is the post treatment concentration of microorganism j from the STP in sub-catchment l.

Assumptions

1. The dry weather budget was simply the product of the population connected to the STP, the volume of water used per person per day and the post treatment microorganism concentration measured in the water released by the STP.

156

2. The volume of effluent produced per person per day is adjustable (160 L in this study).

In wet weather the volume of effluent that may be released during an event can be allocated based on the buffer capacity for each STP and available data on overflow volumes. The microbial load excreted per person per day was calculated by multiplying the percent prevalence of infection in the community by the concentration of microorganisms excreted per infected person per day. The percent prevalence of microbial infection in the population was estimated at 100% for E. coli. Jones and

White (1984) report the daily load of E. coli excreted per person as 9.3 log10 cfu similar to the value reported by Feachem et al. (1983) of 10 log10 cfu. The PCB model uses an excretion rate of 10 log10 cfu per person per day.

The prevalence of both Giardia and Cryptosporidium was estimated at 1% which is within the ranges used by Anderson et al. (1998) (0-10% for Giardia and 0-5% for Cryptosporidium) and similar to the results of Hellard et al. (2000). Hellard et al. (2000) examined asymptomatic individuals in Melbourne, Australia and reported the percent prevalence of infection of Giardia at 1.6% and Cryptosporidium at 0.4%. The values used in the model are also similar to the prevalence reported in asymptomatic individuals in Scandinavian countries of 2.97% for Giardia and 0.99% for Cryptosporidium (Horman et al., 2004). The concentration of Cryptosporidium and Giardia excreted by infected persons was estimated using results from raw effluent analysis on samples collected from Braemar STP (Table 2.16). The load per infected 5 6 person per day was estimated at 7.5 x 10 Cryptosporidium oocysts and 1.3 x 10 P Giardia cysts (Australian Water Technologies, 2002a). These values are similar to the estimates of 4.4 x 105 Cryptosporidium oocysts and 1.4 x 106 Giardia cysts for protozoan loads shed per person per day in the Netherlands (Medema et al., 2001).

157

The number of microorganisms entering the stream from STPs is given by:

,3, = β jllj + bWWenI l )/()( (8)

where βj is the number of microorganisms j excreted per person per day (accounts for excretion rate from infected person multiplied by the prevalence rate within the population), W was the volume of effluent transported in a wet weather event

(ML), e is the event duration and bl is the buffer capacity for the STP in sub-catchment l (ML).

Assumptions

1. The wet weather budget was the load of microorganisms entering the STP (population connected · microorganisms.person-1.day-1) buffered by the available storage at the STP. 2. Any water entering in excess of the buffer was assumed to leave the STP without treatment.

On-site systems module

The input of microorganisms to the stream from on-site systems is assumed to depend on the population using on-site systems; an estimate of the number of microorganisms excreted per person per day; and the fraction of on-site systems connected to the stream. The only difference between wet and dry conditions for the on-site systems module is the level of connectivity to streams. Inputs from on-site systems are given by:

2 I ,2, lj (Sl β ))1( ( ()OOOC dwdlj [ exp1 {−−−+−= ( l UU 0 ) }]) (9)

where Ij,2,l was the input to stream of microorganism j from on-site systems in sub-catchment l, Sl was the population of sub-catchment l, Cl was the proportion of the

158

population in sub-catchment l connected to an STP, and Od and Ow were the proportion of on-site systems connected to streams in dry and wet weather, respectively.

Assumptions

1. 1% of on-site systems were connected to the stream under dry conditions

(Od), in wet conditions (Ow) this increased to 20%. 2. There was no decay of microorganisms between on-site systems and the stream network. 3. The variation of the fraction of on-site systems connected to streams with size of event is described by equation (9).

In-stream module

Little data was available regarding the rate of attachment of microorganisms to sediment particles and their subsequent rate of deposition in streams, lakes and reservoirs. Recent studies by Davies et al. (2005b), Kaucner et al. (2004) and Brookes et al. (submitted) suggest that Cryptosporidium and Giardia (oo)cysts remain largely unattached to particles and may remain suspended in the water column for considerable lengths of time. However, E. coli was more likely to attach to particles and was correlated with particles in the size range 74.7 – 157 μm (Brookes et al., submitted). The in-stream export from the sub-catchments during dry weather is given by:

2 z ⎡ L L ⎤ ⎡ ⎤ ()l vR l Rl ()vR ii Ri ,lj = ⎢∑ IE lkj δ j,,, 1 ()1 j +− IF 3,, lj ⎥ + ⎢∑ E δ jij ,, 1 ()1− F j ⎥ (10) ⎣k=1 ⎦ ⎣ i=1 ⎦

L where Ej,l was the exported load of microorganism j from sub-catchment l, R l B was the local reach length for node l (km – assumed to be a where a is the sub- 2 catchment area in km ), vl was the flow velocity over the reach, Fj was the probability of microorganism j being deposited or bound to bed sediment over a 1km reach and Rl was the reach length between node l and the next node downstream (km).

159

Assumptions

1. In dry weather, all microorganisms bound to sediment settle out, and there is no resuspension of settled material in either dry or wet weather. 2. A fixed rate of 50% of E. coli were assumed to be bound to sediment and thus lost through settling. 3. Cryptosporidium and Giardia primarily remain in the water column with only 5% becoming bound and lost through settling. The stream reach (km) was divided by the flow velocity to estimate the loss due to settling per km for each sub-catchment. 4. Inactivation was calculated using the microorganism specific decay rate for water (Table 5.5) and an estimated travel time. The decay rate was

expressed as a percentage of the initial population surviving per day (δj,i). 5. There was no decay of microorganisms entering the river network from the STPs before reaching the outlet of each sub-catchment due to the STP being located near the sub-catchment outlet.

In wet weather the in-stream export from the sub-catchments was given by:

2 z ⎡ L ⎤ ⎡ ⎤ ()l vR l ()vR ii ,lj = ⎢∑ IE ,, jlk δ j 1, + I ,,3 jl ⎥ + ⎢∑ E δ jij 1,, ⎥ (11) ⎣k =1 ⎦ ⎣ i=1 ⎦

Assumptions

1. During dry weather (low flow conditions), the flow velocity was assumed to be 0.1 m.s-1. During intermediate wet weather events flow velocity was assumed to be 1 m.s-1 and for the larger wet weather event flow velocity was assumed to be 3 m.s-1. All flow velocity values are adjustable for each sub-catchment.

The PCB model consists of five modules: hydrologic, land, STP, on-site systems and in-stream modules. The hydrology module has 4 parameters, though one of these

(flB )B is derived from a rainfall surface. The land module has 8 parameters. The STP

160

module has 7 parameters, 3 of which are obtained from recorded values (cj,l, W and bl). The on-site systems module has 6 parameters, though two of these are included in other modules (U0 and βj). The in-stream module has 5 parameters, though Rl is determined L from the DEM (and R l estimated from the sub-catchment area). Thus the PCB model has 27 parameters, though it should be noted that some of these parameters are matrices. The ICMS program links these modules as outlined in Figure 5.1.

Figure 5.1 System view of the model (PCB) showing the linkages between the input data and the 5 separate subroutines or modules to produce the model outputs

Application in the Wingecarribee

The PCB model was developed and tested using data for the Wingecarribee catchment located approximately 200 km south west of Sydney in the Sydney drinking water catchment. The Wingecarribee is a mixed land use catchment comprised of unsewered and sewered urban areas, rural residential, improved pastures, horticulture, native vegetation and grazing areas as well as a protected upland swamp. The

161

catchment was divided into 52 sub-catchments (Figure 5.2) based on a spot sampling of the water quality in the catchment carried out in 2002. The available land use data were transformed into a subset of 13 classes, selected to cover the range of land uses within the catchment and the expected microorganism sources (Table 5.2).

162

Figure 5.2 Map of the Wingecarribee catchment showing the 52 sub- catchments used in this study. The top panel shows the distribution of land use across the catchment, while the bottom panel shows the stream network and the sampling sites

163

Hydrologic and in-stream modules

The median flow rate in the Wingecarribee River during dry weather was 0.1 -1 m.sP ,P calculated from the previous ten years flow data for Berrima weir (sub-catchment 26) and Greensteads (sub-catchment 49) gauging stations. Estimates of mean daily flow for each sampling site within the Wingecarribee catchment (Table 5.6) were calculated by Ecowise Environmental as part of a previous study (Olley & Deere, 2003).

164

Table 5.6 Estimated mean daily flow for sampling sites in the Wingecarribee

Sample site Mean daily flow (ML) Sample site Mean daily flow (ML) 1 0.71 27 0.31 2 5.60 28 0.10 3 10.80 29 0.10 4 1.64 30 0.00 5 0.68 31 0.84 6 3.06 32 0.80 7 5.44 33 0.00 8 15.30 34 0.91 9 5.44 35 0.86 10 6.37 36 22.30 11 0.49 37 24.80 12 1.30 38 0.00 13 11.64 39 0.59 14 3.17 40 19.80 15 14.27 41 7.46 16 0.35 42 23.50 17 0.52 43 0.10 18 0.10 44 0.10 19 0.10 45 0.33 20 1.05 46 0.10 21 1.52 47 0.56 22 1.99 48 4.68 23 0.10 49 20.98 24 0.10 50 23.60 25 16.90 51 15.80 26 21.10 52 3.17

165

Land module

For each sub-catchment, data on the population, fraction of people using on-site systems and number of animals of each species present were extracted from the land use data (Table 5.2). The human population densities were assumed to be 2 400 people.km- 2 for urban residential (λ=10), 100 people.km-2 for rural residential (λ=11), and 10 people.km-2 for agricultural land uses (sum of λ=2, 3, and 5 to 9). Land uses allocated zero population included water (λ=1), commercial and industrial (λ=4) and forestry (λ=12 and 13). Comparison with population statistics from Australian Bureau of Statistics (ABS) suggested these were acceptable estimates of densities for the Wingecarribee catchment (www.abs.gov.au). The total population estimated with these land use allocations was approximately 40000, which is close to the NSW Department of planning estimate of 39453 persons in Wingecarribee catchment (June 2002: Planning NSW estimate).

The specific sub-catchment characteristics of the Wingecarribee catchment required to run the model are summarized in Table 5.7. Many of these variables were derived from the GIS land use layer (such as the proportion of each land use category present in a specific sub-catchment and sub-catchment area). However, other variables such as the location of the STP that an upstream sub-catchment is connected to (Hl) are identified and input manually.

166

Table 5.7 Subcatchment data file for the Wingecarribee catchment

l s a Rl Gl nl Hl fl Land use category (λ)§ 1 2 3 4 9 10 11 13 1 1 1.13 1.71 3 0 0 1.75 0.0044 0.5279 0.3048 0.0482 0 0 0 0.1146 2 1 3.25 1.80 3 0 0 1.75 0.0028 0.2732 0.0829 0.255 0 0.264 0.0567 0.0653 3 2 2.63 10.24 7 0 0 1.75 0.003 0.5213 0.2011 0.0292 0 0 0 0.2454 4 1 1.80 8.74 7 0 0 1.59 0 0.391 0.3113 0.0356 0 0 0 0.2622 5 1 1.11 3.67 6 0 0 1.42 0.0092 0.4886 0.2619 0.0593 0 0 0 0.181 6 2 5.31 9.77 9 0 0 1.40 0.0043 0.4969 0.3276 0.0375 0 0 0 0.1338 7 3 30.59 3.66 8 0 0 1.47 0.1486 0.3286 0.1949 0.0329 0 0.008 0.0033 0.2836 8 4 14.77 5.12 10 0 0 1.29 0.0244 0.6245 0.2212 0.0417 0 0 0 0.0882 9 3 41.26 2.06 10 0 0 1.22 0.0023 0.7174 0.1448 0.0267 0 0 0 0.1088 10 5 18.75 8.78 13 0.98 52 1.16 0.0116 0.5406 0.1199 0.0594 0 0.0972 0.091 0.0803 11 1 8.53 5.14 12 0 0 1.20 0.0055 0.491 0.1362 0.0195 0 0 0.0001 0.3477 12 2 12.34 2.44 52 0.98 52 1.05 0.0005 0.1362 0.0381 0.1336 0 0.317 0.3147 0.0599 13 6 30.40 0.49 15 0.98 17 1.09 0.018 0.5621 0.044 0.0973 0.0035 0.1213 0.0738 0.0801 14 4 0.83 0.79 15 0.98 52 1.04 0 0.3215 0.1117 0.0657 0 0.2673 0.0511 0.1827 15 7 1.03 4.91 25 0.98 52 1.04 0.0064 0.5602 0.0404 0.1227 0 0.0362 0.0644 0.1697 16 1 6.67 2.23 17 0.98 17 1.14 0.0011 0.2414 0.0522 0.261 0 0.2699 0.166 0.0085 17 2 4.28 7.11 22 0.98 17 1.11 0.0167 0.7495 0.1368 0.0479 0 0.0065 0.013 0.0298 18 1 4.65 2.08 19 0 0 1.21 0.0005 0.6652 0.2917 0.0184 0 0 0 0.0241 19 2 10.73 2.74 20 0 0 1.18 0.0052 0.6655 0.2101 0.0451 0 0 0.015 0.0592 20 3 8.08 1.52 21 0 0 1.17 0.0026 0.6918 0.1647 0.0406 0 0.0005 0.02 0.0799 21 4 9.88 5.99 22 0 0 1.12 0.0006 0.7607 0.099 0.0615 0 0 0.0248 0.0535 22 5 16.40 10.10 39 0 0 1.04 0.0024 0.7108 0.0816 0.0382 0 0 0 0.1671 23 1 11.63 10.13 39 0 0 1.09 0.0011 0.7706 0.0849 0.02 0 0.0065 0.003 0.1138 24 1 10.96 1.95 25 0.98 26 1.02 0.0064 0.7275 0.0941 0.102 0 0.0114 0.0031 0.0555 25 8 10.24 2.47 26 0 0 1.01 0.0366 0.5919 0.0608 0.0254 0 0 6E-05 0.2852 26 9 3.50 3.42 51 0.98 26 0.99 0.0014 0.0855 0.0078 0.3793 0 0.0657 0.3278 0.1325

167

Table 5.7 (cont’d.) Subcatchment data file for the Wingecarribee catchment

l s a Rl Gl nl Hl fl Land use category (λ)§ 1 2 3 4 9 10 11 13 27 1 9.28 10.90 42 0 0 1.00 0.0018 0.8013 0.0892 0.0284 0 0 0 0.0792 28 1 2.01 6.10 29 0 0 1.16 0.0109 0.7699 0.0961 0.0346 0 0 0 0.0884 29 2 18.65 11.23 31 0 0 1.10 0.0041 0.6392 0.1861 0.032 0 0 0 0.1385 30 1 0.37 8.04 31 0 0 0.93 0.0197 0.7343 0.1734 0.0602 0 0 0 0.0123 31 3 32.30 1.84 32 0 0 0.96 0.0049 0.3253 0.0443 0.0203 0 0 0 0.6051 32 4 5.06 1.41 34 0 0 0.87 0.0003 0.0662 0 0.0248 0 0 0 0.9088 33 1 35.35 0.93 34 0 0 0.87 0.0018 0.2392 0.0068 0.0273 0 0 0 0.7248 34 5 1.38 4.10 35 0 0 0.87 0 0.1182 0.0018 0.0296 0 0 0 0.8504 35 6 30.07 1.16 37 0 0 0.87 0.0017 0.123 0.0021 0.0341 0 0 0 0.8391 36 14 28.20 1.24 37 0 0 0.87 0.0013 0.2777 0.0352 0.0166 0 0 0 0.6692 37 15 4.52 10.44 49 0 0 0.86 0.0003 0.1028 0 0.0349 0 0 0 0.862 38 1 6.95 6.05 39 0 0 0.90 0 0.0341 0 0.0672 0 0 0 0.8987 39 6 44.38 0.81 40 0 0 0.97 0.0102 0.5527 0.0636 0.0466 0 0.0002 0 0.3266 40 13 1.33 7.23 36 0 0 0.90 0 0 0 0.0114 0 0 0 0.9886 41 12 7.85 0.88 40 0 0 0.92 0.0054 0.2788 0.0358 0.0175 0 0 0 0.6625 42 11 26.40 5.34 41 0 0 0.96 0.0042 0.4133 0.0143 0.0366 0 0 2E-05 0.5315 43 1 22.03 3.55 42 0 0 0.97 0.0072 0.5793 0.0569 0.0224 0 0 0 0.3342 44 1 9.36 3.26 45 0 0 1.02 0.0091 0.5791 0.0543 0.0409 0 0 0 0.3167 45 2 13.76 10.16 47 0 0 1.01 0.0047 0.3389 0.002 0.0407 0 0 0 0.6137 46 1 2.55 9.18 47 0 0 0.95 0.0008 0.9371 0.0109 0.0189 0 0 0 0.0323 47 3 67.41 3.63 48 0 0 0.94 0.0007 0.2154 0.0129 0.0268 0 0 0 0.7442 48 4 10.62 17.35 50 0 0 0.84 0.0035 0.1353 0.0058 0.0318 0 0 0 0.8236 49 16 42.08 18.01 50 0 0 0.85 0.0022 0.1333 0.0017 0.0189 0 0 0 0.8438 50 17 59.52 0.00 0 0 0.82 0.0032 0.1779 0.0009 0.0208 0 0 0 0.7972 51 10 3.18 7.31 42 0.98 26 0.99 0.025 0.3541 0.004 0.0835 0 0.0161 0.0152 0.5022 52 3 6.90 0.57 14 0.98 52 1.05 0.0018 0.1407 0.0298 0.1276 0 0.48 0.1444 0.0756 § Land use categories not included in the table were not present in the Wingecarribee catchment

168

Sewage treatment plant module

All but one of the sub-catchments containing urban land use were located upstream or near to an STP. The exception was sub-catchment 13, which is near two STPs (located in sub-catchments 17 and 52). Sub-catchment 13 contains two distinct urban areas; the northern one is assumed to be connected to Bowral STP (sub-catchment 52) and the southern area connected to Moss Vale STP (sub-catchment 17) (see Figure 5.2). In dry weather the loads are calculated using the concentrations of the microorganisms in the post-treatment effluent for each STP (Table 5.8). In wet weather the volume of effluent that may be released during an event is based on the buffer capacity for each STP (19.7 ML for Moss Vale, 11.1 ML for Berrima and none for Bowral STPs respectively), and available data on overflow volumes (1.6 ML for Moss Vale, 0.5 ML for Berrima and 24.5 ML for Bowral STPs respectively) (Paterson & Krogh, 2003).

Table 5.8 Post-treatment concentrations of microorganisms in Wingecarribee STP effluent

STP Microorganism concentration (.L-1)† (sub-catchment) Cryptosporidium Giardia E. coli Bowral (52) 9.58 22.5 56 000 Berrima (26) 0.03 1.17 28 000 Moss Vale (17) 0.66 3.92 87

† Using all available data combined and adjusted for recovery efficiency

On-site systems module

There are approximately 5 000 on-site systems in the Wingecarribee catchment. The most common on-site system consisted of a primary sedimentation tank with supernatant overflow to a gravel-filled absorption trench, and infiltration to the underlying soil layer. It is likely that effluent dispersal areas of most on-site systems in Robertson (unsewered township upstream of Wingecarribee swamp, sub-catchment 2) have a significant hydrological connection (overland or seepage) to local watercourses (K. Charles, UNSW pers. comm.). The geometric mean number of persons per

169

household in Robertson was 2.35 (SD 0.16, n = 16). The usage rate of community tanks could not be estimated. Septic tank pump outs are treated at Berrima STP. Runoff from unsewered urban areas is believed to be the fourth largest source of Cryptosporidium in the SCA catchment areas (Australian Water Technologies, 2002a).

Outputs from the model for the Wingecarribee catchment

Predicted loads

The model predicts for each microorganism a local generated source budget (input) and the routed downstream (export) budget. Figure 5.3 to Figure 5.5 show the input and export budgets for Cryptosporidium, Giardia and E. coli, respectively. The predicted Cryptosporidium and Giardia input budgets both show dry weather peaks at sub-catchments 52, 17 (Bowral and Moss Vale STPs) and 2 (Robertson township). Dry weather input load predictions in the remaining sub-catchments vary between 3 to 5 log10 for Cryptosporidium and 3 to 6 log10 for Giardia. The exceptions are the low predicted input loads of 2.7 log10 for Cryptosporidium in sub-catchment 30 and 1.1 log10 for Giardia in sub-catchment 40. Sub-catchment 30 is very small (only 0.3 km2 and 73% improved pasture) while sub-catchment 40 is only 1.3 km2 and is 99% native forest. Predicted wet weather inputs of Cryptosporidium are generally 3 to 4 log10 higher than the dry weather inputs whereas the wet weather input loads for Giardia are

2 to 3 log10 higher than the dry weather input loads. The E. coli dry weather input loads are several orders of magnitude higher than either Cryptosporidium or Giardia, ranging from 9 to 11 log10 with small peaks above 11 log10 for sub-catchments 2 and 52. The lowest predicted input load is again from sub-catchment 30, with low loads also predicted for sub-catchments 1, 3, 4, 5, 14, 15, 28, 34, 37, 40, 46 and 51. These sub- catchments are all small in size (<5 km2, except 4 = 8.7 km2) and the predominant land use is either native forest or improved pasture. In most sub-catchments wet weather events increase the predicted wet weather input loads by approximately 2 log10 compared to dry weather. However, the increase is only about 1 log10 in those sub- catchments with existing point sources of pollution (sub-catchments 2 and 52).

170

The predicted export loads of Cryptosporidium, Giardia and E. coli highlight the cumulative effect of routing contaminants downstream, particularly during wet weather events when the transport time is too short for microbial inactivation or stream settling. For example, the high input of Cryptosporidium and E. coli at sub-catchment 2 impacts on the predicted load exported from both sub-catchment 2 and 3. In wet weather events the total load predicted for sub-catchment 3 will be higher than from sub-catchment 2. However, in dry weather the total predicted export load of E. coli was lower in sub- catchment 3 due to the effect of bacterial inactivation and in-stream settling. These factors mitigate the transport of E. coli to downstream sub-catchments to a much greater extent than for either of the protozoan microorganisms. Thus the total export of E. coli from most sub-catchments is reasonably constant at approximately 9 log10 per day in dry weather with the exception of those sub-catchments impacted by on-site sewage discharges (2, 3) or STP effluent (52, 14, 15, 25 downstream of Bowral). During large wet weather events the predicted daily load of Cryptosporidium and Giardia exported from the Wingecarribee catchment (sub-catchment 50) is greater than 10 log10 while the predicted E. coli load is almost 15 log10.

171

10 9 8 7 load log10

6 5 4 3 2

Cryptosporidium 1 0 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Dry weather Intermediate Wet Large Wet

11 10 9 8 7 load log10

6 5 4 3 2 Cryptosporidium 1 0 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Dry weather Intermediate Wet Large Wet

Figure 5.3 Cryptosporidium loads log10 oocysts per day a) generated in each sub- catchment and b) exported from each sub-catchment and routed downstream (not in sequential downstream order)

172

11 10 9 8 7 6 load log10 5 4

Giardia 3 2 1 0 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Dry weather Intermediate Wet Large Wet

12 11 10 9 8 7

load log10 6

5 4

Giardia 3 2 1 0 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Dry weather Intermediate Wet Large Wet

Figure 5.4 Giardia loads log10 cysts per day a) generated in each sub-catchment and b) exported from each sub-catchment and routed downstream (not in sequential downstream order)

173

15 14 13 12 11 10 load log10

9 8 E. coli 7 6 5 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Dry weather Intermediate Wet Large Wet

16 15 14 13 12 11 load log10

10 9

E. coli 8 7 6 5 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Dry weather Intermediate Wet Large Wet

Figure 5.5 E. coli loads log10 cfu per day a) generated in each sub-catchment and b) exported from each sub-catchment and routed downstream (not in sequential downstream order)

174

Sub-catchment rankings

Table 5.9 shows the top ten sub-catchments responsible for the generation of pathogens and E. coli from the Wingecarribee catchment. The sub-catchments were ranked according to the total load generated (raw ranking) and by the total per unit area (km2). The dry weather rankings were dominated by sub-catchments receiving inputs from STPs and on-site sewage systems. Sub-catchments 52 and 17 receive effluent from Bowral and Moss Vale STPs respectively, these two sub-catchments were ranked as having the highest loads of Cryptosporidium and Giardia in dry weather by both raw and area rankings. Sub-catchment 52 was predicted to have the highest load of E. coli by the raw ranking but was second to sub-catchment 2 in the unit area ranking. Sub- catchment 2 contains the unsewered urban township of Robertson and recorded the highest load per unit area of E. coli in both dry and intermediate wet weather. Sub- catchment 26 receives effluent from Berrima STP and was ranked third or fourth for all of the microorganisms per unit area.

The wet weather rankings were much more varied than the dry weather rankings. The highest loads by unit area rankings for intermediate wet weather events were recorded for the small sub-catchments in the upper part of the Wingecarribee catchment. Sub-catchments 2, 3 and 4 include and are immediately downstream of Robertson. Sub-catchment 1 located upstream consists of improved pasture with diffuse impacts from cattle and sheep and this sub-catchment had the highest predicted load of Giardia per unit area for intermediate wet weather events. Bowral STP was still ranked third highest load of Cryptosporidium and Giardia per unit area for intermediate wet weather events, but larger events causing STP overflow would return this sub- catchment to the highest ranking for both Cryptosporidium and E. coli loads. In the larger wet weather event conditions Giardia and Cryptosporidium loads ranked by unit area become dominated by inputs from sub-catchments with agricultural land uses (46) and native forest land use (40, 47). Also, the impacts from upstream sub-catchments accumulate as the water is routed downstream since there is negligible inactivation/loss of the microorganisms during transit in these large events. This is most evident in the raster diagrams produced by the model (see Figure 5.6 to Figure 5.11).

175

Table 5.9 Ranking of Wingecarribee sub-catchments with the highest predicted pathogen and E. coli loads (raw load and per unit area)

Raw Cryptosporidium Giardia E. coli Ranking Dry Wet I Wet L Dry Wet I Wet L Dry Wet I Wet L 1 52 47 47 52 9 9 52 9 52 2 17 50 50 17 7 39 2 7 47 3 2 7 49 9 39 13 47 47 9 4 47 49 33 39 13 7 50 39 39 5 7 9 39 13 8 47 7 13 50 6 50 39 31 47 29 29 49 50 13 7 13 31 35 7 10 43 13 31 7 8 39 33 9 43 22 22 39 10 49 9 9 35 36 29 47 10 33 8 31 10 49 13 7 2 43 8 12 49 33 Area Cryptosporidium Giardia E. coli Ranking Dry Wet I Wet L Dry Wet I Wet L Dry Wet I Wet L 1 52 3 52 52 1 28 2 2 52 2 17 2 40 17 3 46 52 3 2 3 2 52 38 2 52 9 26 1 17 4 26 4 32 26 8 21 12 4 3 5 12 1 47 46 9 17 16 5 1 6 16 7 37 23 4 23 7 6 9 7 23 5 34 27 17 18 23 7 18 8 7 6 45 21 6 20 14 8 28 9 14 11 35 28 5 27 10 9 8 10 10 40 49 30 28 19 13 18 20

Wet I = intermediate wet weather event (30 mm) Wet L = large wet weather event (100 mm)

Raster diagrams

Both the local source (input budgets) and the exported budgets can be reproduced as spatial (raster) graphs. Figure 5.6, Figure 5.8 and Figure 5.10 show the predicted input load per unit area (km2) of Cryptosporidium, Giardia and E. coli generated within each sub-catchment in dry, intermediate and large wet weather events. Figure 5.7, Figure 5.9 and Figure 5.11 show the predicted total loads of Cryptosporidium, Giardia and E. coli exported from each of the Wingecarribee sub- catchments.

176

Figure 5.6 Cryptosporidium input loads log10 oocysts generated within each sub- catchment per km2 per day in a) dry weather, b) intermediate wet weather event and c) large wet weather event

177

Figure 5.7 Cryptosporidium loads log10 oocysts exported from each sub- catchment per day in a) dry weather, b) intermediate wet weather event and c) large wet weather event

178

Figure 5.8 Giardia input loads log10 cysts generated within each sub-catchment per km2 per day in a) dry weather, b) intermediate wet weather event and c) large wet weather event

179

Figure 5.9 Giardia loads log10 cysts exported from each sub-catchment per day in a) dry weather, b) intermediate wet weather event and c) large wet weather event

180

Figure 5.10 E. coli input loads log10 mpn generated within each sub-catchment per km2 per day in a) dry weather, b) intermediate wet weather event and c) large wet weather event

181

Figure 5.11 E. coli loads log10 mpn exported from each sub-catchment per day in a) dry weather, b) intermediate wet weather event and c) large wet weather event

182

Discussion

Few studies have attempted to model variations in pathogen concentrations or loads in drinking water catchments (Dorner, Huck & Slawson, 2004; Medema & Schijven, 2001; Walker & Stedinger, 1999). This model incorporates a land budget of diffuse pollution as well as STP and on-site systems modules to calculate source generation of pathogens and E. coli based on geographic information system (GIS) land use layers. These sources are then routed through the sub-catchments using hydrologic and in-stream modules. In dry weather the only source of manure reaching streams is from the direct deposition of faeces, primarily from cattle, plus a small component from wildlife. Additional dry weather inputs include on-site systems and STP effluents. Following rainfall, direct deposition will be supplemented by the mobilisation of manure deposited on land into surface water runoff. Walker & Stedinger (1999) noted that if oocysts were easily freed from manure due to runoff processes, then their model would considerably underestimate oocyst concentrations in surface waters. A recent study by Davies et al. (2004b) indicated that Cryptosporidium oocysts were easily freed from cow manure and that the oocysts were readily transported in the surface runoff, particularly in the absence of vegetation. The PCB model uses manure mobilisation rates to estimate the rate at which faeces are mobilised from land to water during wet weather flood events.

Fraser et al. (Fraser, Barten & Pinney, 1998) noted that the application of their model in the field was limited by a number of poorly understood factors, including the assumption of steady state conditions which did not allow for variation in environmental conditions, such as rainfall or the variation in the deposition of faecal material on pastures. The PCB model accounts for variations in rainfall-induced runoff by incorporating a surface rainfall layer from the GIS. The model also includes a parameter that estimates microorganism attachment to sediment particles and uses this along with stream velocity to estimate in-stream loss due to settling, although the model does not account for sediment re-suspension. Inputs from native and feral animals are also accounted for in the land budget. Insufficient data was available to account for non-random deposition of manure on pastures. Further research to investigate the behaviour of animal species that have access to the riparian zone would improve estimates of the rate of direct deposition, and also quantify the location and distribution

183

of manure on pastures. Future refinements to the model could include the development of additional point source modules for manure spreading and/or biosolids applications and improvement of the hydrologic component to account for time series flow data. If estimates of pathogen infectivity for Cryptosporidium oocyst isolates from these catchments were available, it would be possible to use the model for quantitative microbial risk assessment (QMRA) to predict the proportion of the total load that represents a potential risk for human infection.

The PCB model was developed to predict export loads of Cryptosporidium, Giardia and E. coli from drinking water catchments for dry weather and for two wet weather event scenarios. The fate of these organisms within storages can subsequently be predicted by using the outputs of the PCB model as the input parameters for the hydrodynamic model developed by Hipsey et al. (Hipsey et al., 2005). In addition to the load data, the raw and unit area rankings of sub-catchment inputs and exports (Table 5.9) can be used to prioritise catchment management activities at both catchment and sub-catchment scale. Scenario evaluation of various management activities can also be achieved by simply varying the input parameters to the model. For example land use changes can be modelled by altering the proportion of each land use category in the sub- catchment input file. The model could be augmented to include economic estimates of various management options, thus enabling catchment managers to perform cost-benefit analysis of catchment rectification scenarios.

Conclusions

The developed model includes experimentally derived and field-tested effects of microorganism sources, environmental processes and transport. This data was subsequently combined with existing published data, GIS-derived land use data and sub-catchment information. The development of this model was prompted by the lack of any existing commercially available pathogen export models, and the need to quantify the sources and fate of pathogens in the Sydney drinking water catchment. Although application of the model is constrained by the availability of appropriate data and by a number of assumptions, the outputs represent the first attempt to quantify pathogen loads, and hence, risk in the Sydney drinking water catchment.

184

Chapter 6 will describe a sensitivity analysis of the model and compare the predicted outputs to water quality data collected from the Wingecarribee catchment. This chapter also evaluates a range of molecular source tracing techniques for the discrimination of E. coli isolates collected from the Wingecarribee catchment to determine if these tools could be used to investigate pollutant sources identified by the pathogen model. Chapter 7 will discuss the application of the pathogen model (PCB) to predict pathogen and E. coli budgets for all 27 of the catchments in the SCA area of operations. The outputs will be used to rank catchments and sub-catchments with the highest generated loads of Cryptosporidium, Giardia and E. coli. The identification of these high-risk sub-catchments will enable catchment managers to prioritise the implementation of catchment control measures and to evaluate the potential benefits and improvements that may be attained, thereby achieving the greatest possible improvements to water quality.

185

Chapter 6 Testing the model

Parts of this chapter have been submitted as: Miller, K., Ferguson, C., Gillings, M., Mitchell, H., Pappayut, S., Angles, M., Cox, P., Brusentiev, S. and Neilan, B. Comparison of Tracing and Tracking tools for identifying bacterial contamination in drinking water catchments. Environ. Sci. Tech. (submitted).

C. Ferguson (SCA) and B. Croke (Australian National University) performed the sensitivity analysis of the model in ICMS and FORTRAN program languages, respectively. C. Ferguson designed the water quality monitoring program in the Wingecarribee and assisted with field-work. The majority of field samples were collected by C. Kaucner and P. Paterson. All samples were analysed and reported by C. Kaucner. Comparison of water quality results to model predictions was performed by C. Ferguson. The comparative trial of molecular tracing and tracking methods was designed and managed by C. Ferguson. E. coli strains were isolated by the bacteriological laboratory at Sydney Water Laboratories, West Ryde under the supervision of G. Ault. Molecular analyses were performed by S. Brusentiev, H. Mitchell, M. Gillings, S. Pappayut and B. Neilan. The paper for publication of the results was prepared by K. Miller, C. Ferguson, M. Angles, P. Cox and B. Neilan.

Sensitivity analysis

Sensitivity analyses are formalised procedures to identify the impact of changes in model inputs and components on a model’s output. Sensitivity analysis is an important tool for evaluating the level of uncertainty in a model and to identify those parameters that have the maximum effect on the output of the model. These parameters can then be the focus of further data gathering and research activities to improve the estimates of the model.

187

Analysis of the export loads

A sensitivity analysis was performed to evaluate the effect of single perturbations of selected input parameters on the predicted loads exported from each sub-catchment. Thirteen key parameters were selected from the 27 model parameters and varied from the model default values by multiplying them by the factors shown in Table 6.1. Rather than selecting extreme perturbation values for the parameters, the sizes of the perturbations were selected based on plausible estimates of their likely uncertainty. For example, the model default values for animal microorganism concentrations.kg-1 manure, which are likely to have a high degree of variability, were perturbed by an order of magnitude. But the volume of sewage effluent per person per day, which is likely to be less variable, was only perturbed by 5%.

Table 6.1 Perturbation factors for the sensitivity analysis

Parameter Perturbation factor

Population (Sl) 0.9

Population connected (nl) 0.8

Microorganisms per person per day (βj) 0.1

Animal density (As,l) 0.1

Animal access to streams (Xs) ^1.5

Likelihood of direct defecation (Ds) ^1.5 -1 Animal microorganism concentrations.kg (Pj,s) 0.1

Microbial decay rates (δj,i) ^1.5

Manure mobilisation (Ms) ^2 Connectivity of on-site systems (O) 0.5 Volume of sewage.person-1.day-1 (V) 0.95

Fraction bound (Fj) ^1.5 Flow velocity (m.s-1) (v) 0.5 ^ raised to the power of

188

Results of the sensitivity analysis were viewed in several categories as follows: for each of the three microorganisms; for dry and wet weather; for each of the 52 sub- catchments; and for each of the 13 parameters perturbed. This resulted in over 4,000 relative change outputs. Take for example the results of sensitivity analyses for the Cryptosporidium loads delivered in wet weather from four representative sub- catchments (Figure 6.1). Loads are most influenced by plausible variations in the number of animals assumed to be in the sub-catchments, the extent of mobilisation of microorganisms from manure and by the assumed concentration of microorganisms in the faeces, microbial decay rates are less influential.

Humans in subcatchment Proportion connected to sewer Pathogens/person•day Animals in subcatchment Animal access to streams Likelihood of direct defecation Pathogens/kg faeces Microbial decay rates Mobilisation of manure

Parameter perturbed Parameter Connectivity of septics to stream Volume of sewage/person•day Fraction of pathogens bound Flow velocity

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Relative change in total Cryptosporidium load

Sewage discharge (52) Unsewered urban (2) Cattle grazing (28) Woodland (40)

Figure 6.1 Relative change in total Cryptosporidium loads exported during wet weather events from four representative sub-catchments (52, 2, 28, 40). The dominant microorganism source for each sub-catchment is indicated in the legend

In synthesizing the results of the sensitivity analysis the following observations emerged. In dry weather reducing the population by 10% caused a 10% reduction in the predicted load of all microorganisms in the sub-catchments that included either unsewered urban areas (2) or STPs (17, 26, 52) and their downstream sites (3, 51, 14, 15). In wet weather the same sites also showed a reduction of E. coli loads, however there was no impact on the predicted loads of Cryptosporidium and Giardia. This suggests that mobilisation of manure from diffuse sources, rather than human inputs, were the dominant processes contributing to Cryptosporidium and Giardia loads in wet weather. In dry weather reducing the population connected to the STPs by 20% resulted

189

in up to a 20% reduction in the predicted load of all three microorganisms but only at sites impacted by STPs. In wet weather E. coli loads were reduced by approximately 5% at the same sites and by up to 15% downstream of Bowral STP (52, 14). Reducing the number of microorganisms excreted per person per day by 90% caused a 90% reduction of the estimated dry weather loads in the sub-catchments affected by on-site systems (2, 3, 7) and approximately a 70% reduction in the E. coli load downstream of Moss Vale STP (17). As with the population connected, the wet weather effect was limited to reductions in the estimated E. coli loads by approximately 30% downstream of STPs and on-site systems and by as much as 70% downstream of Bowral STP (52, 14).

In dry weather a reduction in animal density by 90% caused a 90% reduction in the predicted loads of Giardia in sub-catchments dominated by improved pasture or native forest, and of E. coli loads in sub-catchments dominated by native forest. In wet weather the decrease in animal density predicted a 90% decrease in Cryptosporidium and Giardia for all sub-catchments. E. coli loads were also reduced by 90% with the exception of those sub-catchments that were impacted by on-site systems and STPs. Reducing animal access to streams and the likelihood of animals defecating in streams reduced the predicted dry weather loads of Giardia by 60% and 80% respectively. While Cryptosporidium loads in dry weather were reduced by 30% and 50 - 80% respectively, mainly in sub-catchments dominated by improved pasture and native forest.

In dry weather reducing the load of microorganisms excreted by animals each day by 90% decreased the predicted loads of Giardia by approximately 90% and Cryptosporidium and E. coli loads by between 20 – 80%. In wet weather the reduction in animal microorganism loads produced a 90% decrease in the predicted load of Cryptosporidium and Giardia in all sub-catchments. E. coli loads were also reduced by 90% in most sub-catchments, with the exception of those sub-catchments that were impacted by on-site systems and STPs. In dry weather manure mobilisation rates are set to zero, however, in wet weather reducing manure mobilisation by raising the rates to the power of 2 resulted in >90% decrease in the predicted loads of Cryptosporidium and Giardia in all sub-catchments. E. coli loads were again reduced by >90% in most sub- catchments, excluding those impacted by on-site systems and STPs.

190

Perturbation of the microbial decay rates by raising them to the power of 1.5 had no impact on the predicted load of Cryptosporidium in either dry or wet weather. In fact the short time frame for wet weather event duration precluded any impact of decay rates on the predicted wet weather loads. However, in dry weather reducing the decay rate of E. coli and Giardia caused approximately a 30% reduction in predicted loads, with the magnitude of the impact increasing in the downstream catchments dominated by improved pasture and native forest. In dry weather, decreasing the proportion of microorganisms bound to particles by raising the parameter to the power of 1.5 approximately doubled the predicted loads of all three microorganisms. The effect was more pronounced in the downstream sub-catchments 47, 49 and 50 where predicted loads were more than six times higher for all three microorganisms.

Reducing the connectivity of on-site systems by 50% decreased the predicted loads of all microorganisms in sub-catchments 2 and 3 by 50% during dry weather. However, in wet weather a decrease of 50% connectivity predicted only a 20% decrease in the load of E. coli at sites 2 and 3. Reducing the volume of effluent produced per person per day by 5% had no impact on the predicted microbial loads in wet weather events but did result in a corresponding 5% decrease in Cryptosporidium and Giardia loads in sub-catchments impacted by STPs in dry weather. In dry weather reducing flow velocity by 50% decreased the predicted load of E. coli and Giardia by a similar amount. The same effect was evident in wet weather but the decreases were only approximately 10%.

Analysis of sub-catchment inputs by area rankings

Although the predicted model export loads are useful for predicting the magnitude of pathogen and E. coli delivery to the major reservoirs and storages within the Sydney catchment these outputs are not necessarily useful tools for catchment management. However, ranking of sub-catchments by input loads generated within each sub-catchment and particularly the unit area rankings are a useful mechanism for the identification and prioritisation of individual sub-catchments that require management interventions. The raw and unit area rankings of the input loads (Table

191

5.9)X and the raster diagrams (Figure 5.6, Figure 5.8 and Figure 5.10) show which sub- catchments within the Wingecarribee are generating the greatest pathogen and E. coli loads. These sub-catchments may have a large impact on water quality at a local sub- catchment scale and may therefore be of interest to local stakeholders as well as to catchment managers responsible for drinking water quality.

An additional sensitivity analysis was performed to examine the effect of single parameter perturbations on the ranking of sub-catchments based on the input loads per unit area. The analysis weighted the sub-catchments in the top (n) number of sub- catchments by summing the sub-catchment rankings that disappeared out of the top (n) with a value of n for the first, n-1 for the second and 1 for the nth ranking. The analysis was performed using the same perturbations that were applied in the previous analysis of the export loads (Table 6.1).

The greatest impact on sub-catchment rankings was observed in response to the perturbation of manure mobilisation rates, the number of microorganisms excreted per person per day, the concentration of microorganisms excreted by animals, and animal density (Table 6.2). Some impacts on the predicted dry weather loads of Giardia and E. coli were also observed in response to changes in the likelihood of direct faecal deposition. The connectivity of on-site systems was also an important parameter for the prediction of dry weather loads of both Cryptosporidium and E. coli, while animal access to streams affected the ranking of loads for all three microorganisms but only in dry weather. The fraction of microorganisms bound to sediment had some impact on the ranking of sub-catchment loads for Cryptosporidium in dry weather. Five of the parameters had negligible impact on the ranking of sub-catchment loads, they included; population, population connected to the sewerage system, microbial decay rates, the volume of effluent per person per day and flow velocity.

Sensitivity analysis indicates that the rate of manure mobilisation is an important parameter for determining pathogen and E. coli loads in sub-catchments with land uses that contain animals. But sub-catchments that contain on-site systems and STPs are more affected by the presence of these individual point sources. Quantification of manure mobilisation rates for a range of rainfall-runoff event scenarios would also enhance the predictive capability of the model and reduce the level of uncertainty.

192

Other important parameters in the model include animal access to streams and the rate of direct faecal deposition to the stream network. This highlights the need for quantitative data regarding these types of activities in water catchments. Animal access rates will be catchment specific. The likelihood of direct deposition, however, relates to animal behaviour and farm management practices and is thus a more generic research need.

193

Table 6.2 Sensitivity analysis of sub-catchment rankings for input per unit area budgets

Sum of sub-catchment rankings that disappear from top (n) sub-catchments n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Microorganisms per person per day (βj) C Dry 0 0 0 0 1 3 6 10 15 21 28 36 45 55 66 78 91 >100 >100 >100 G Dry 0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 0 0 1 3 E Dry 1 0 1 3 3 6 10 15 21 28 35 42 50 58 66 74 82 90 98 >100 C Int 1 2 3 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Int 0 1 3 5 4 5 6 7 8 9 11 11 12 13 14 15 16 17 18 19 C Wet 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 G Wet 0 0 0 0 0 0 0 0 0 0 0 1 2 3 0 0 0 0 0 0 E Wet 1 2 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 Manure mobilisation (Ms) C Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Int 0 1 3 3 6 10 10 15 15 15 14 13 13 13 10 10 15 16 15 15 G Int 1 3 6 6 5 4 3 3 3 3 4 3 3 5 4 3 3 3 6 5 E Int 0 0 0 1 3 6 10 15 21 28 35 40 48 47 47 52 56 57 49 42 C Wet 1 2 3 1 3 3 3 6 6 10 9 10 14 13 18 19 14 11 9 7 G Wet 1 3 5 4 3 5 3 4 1 1 2 0 0 0 0 0 1 3 3 4 E Wet 0 1 3 6 10 11 15 13 18 17 23 30 38 38 35 34 34 32 32 26

194

Table 6.2 (cont’d.) Sensitivity analysis of sub-catchment rankings for input per unit area budgets

Sum of sub-catchment rankings that disappear from top (n) sub-catchments n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 -1 Animal microorganism concentrations.kg (Pj,s) C Dry 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 1 3 6 6 G Dry 0 0 0 0 1 2 4 7 11 16 22 21 22 29 34 32 31 30 32 28 E Dry 0 0 0 0 0 0 0 0 0 0 1 3 5 8 12 17 23 30 38 47 C Int 0 0 0 0 1 1 1 1 1 1 2 1 1 1 1 1 1 1 1 1 G Int 1 3 3 3 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 E Int 0 0 0 1 3 6 10 11 10 10 14 13 14 12 14 11 11 8 12 17 C Wet 1 2 0 0 0 0 0 0 0 1 2 1 1 1 1 1 1 1 3 3 G Wet 1 3 2 1 3 3 2 1 1 1 1 0 0 0 0 0 0 1 0 0 E Wet 0 1 2 1 1 3 3 1 3 3 6 10 10 10 14 13 12 9 9 12 Animal density (As,l) C Dry 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 1 3 6 6 G Dry 0 0 0 0 1 2 4 7 11 16 22 21 22 29 34 32 31 30 32 28 E Dry 0 0 0 0 0 0 0 0 0 0 1 3 5 8 12 17 23 30 38 47 C Int 0 0 0 0 1 1 1 1 1 1 2 1 1 1 1 1 1 1 1 1 G Int 1 3 3 3 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 E Int 0 0 0 1 3 6 10 11 10 10 14 13 14 12 14 11 11 8 12 17 C Wet 1 2 0 0 0 0 0 0 0 1 2 1 1 1 1 1 1 1 3 3 G Wet 1 3 2 1 3 3 2 1 1 1 1 0 0 0 0 0 0 0 0 0 E Wet 0 1 2 1 1 3 3 1 3 3 6 10 10 10 14 13 12 9 9 12

195

Table 6.2 (cont’d.) Sensitivity analysis of sub-catchment rankings for input per unit area budgets

Sum of sub-catchment rankings that disappear from top (n) sub-catchments n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Likelihood of direct defecation (Ds) C Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 3 6 6 G Dry 0 0 0 0 1 2 4 7 11 16 22 21 22 29 34 32 31 30 32 28 E Dry 0 0 0 0 0 0 0 0 0 0 1 3 5 8 12 17 23 30 38 47 C Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Connectivity of on-site systems (O) C Dry 0 0 0 0 0 1 2 0 0 0 0 1 3 6 10 15 21 28 36 36 G Dry 0 0 1 0 0 0 0 1 0 0 0 0 1 0 1 0 0 0 0 1 E Dry 1 0 0 0 1 3 4 7 11 16 21 26 32 38 44 38 43 36 40 44 C Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Int 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 C Wet 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 G Wet 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 E Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

196

Table 6.2 (cont’d.) Sensitivity analysis of sub-catchment rankings for input per unit area budgets

Sum of sub-catchment rankings that disappear from top (n) sub-catchments n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Animal access to streams (Xs) C Dry 0 0 0 0 0 0 1 0 0 0 0 0 1 3 6 10 15 21 28 36 G Dry 0 0 0 0 1 0 1 1 1 1 3 6 9 8 7 6 7 6 5 7 E Dry 0 0 0 0 0 0 0 0 0 0 0 0 1 3 3 4 5 6 7 0 C Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Population (Sl) C Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 3 5 G Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 E Dry 0 0 0 0 0 0 0 0 0 0 0 0 1 2 3 4 5 6 7 8 C Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Wet 0 0 0 0 0 0 1 2 0 0 0 0 0 0 0 0 0 0 0 0 E Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

197

Table 6.2 (cont’d.) Sensitivity analysis of sub-catchment rankings for input per unit area budgets

Sum of sub-catchment rankings that disappear from top (n) sub-catchments n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Population connected (nl) C Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Dry 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Wet 0 0 1 2 3 0 1 2 3 0 0 0 0 0 0 0 0 0 0 0 E Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 Microbial decay rates (δj,i) C Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 E Int 0 0 0 0 0 0 0 0 1 2 0 0 0 0 0 0 0 0 0 0 C Wet 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 G Wet 0 1 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Wet 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 1 1 0

198

Table 6.2 (cont’d.) Sensitivity analysis of sub-catchment rankings for input per unit area budgets

Sum of sub-catchment rankings that disappear from top (n) sub-catchments n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Volume of sewage.person-1.day-1 (V) C Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Wet 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 E Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Fraction bound (Fj) C Dry 0 0 0 0 1 0 0 0 0 0 0 1 1 3 5 8 11 15 20 18 G Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Dry 0 0 0 0 0 1 0 0 0 0 0 0 1 1 3 3 2 1 1 1 C Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

199

Table 6.2 (cont’d.) Sensitivity analysis of sub-catchment rankings for input per unit area budgets

Sum of sub-catchment rankings that disappear from top (n) sub-catchments n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Flow velocity (m.s-1) (v) C Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Dry 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Int 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 C Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 G Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E Wet 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C = Cryptosporidium G = Giardia E = E. coli Dry = Dry weather Int = Intermediate wet weather event Wet = Large wet weather event

200

Comparison of model outputs to water quality data

Sample site locations

A previous report targeting rectification actions in the Wingecarribee catchment identified 52 locations (Figure 5.2) in the stream network of the Wingecarribee catchment (Olley & Deere, 2003). The same sample sites were maintained in this study. In early 2004 the sites were evaluated for their suitability for the collection of water quality samples. Sites not located on public property or that were difficult or dangerous to sample were generally excluded. The exceptions were sites 49 and 50 for which the SCA had prior permission to collect water samples. Two sites were replaced with a more accessible sampling site close by (<500 m). These new sites were numbered the same as the old site with a suffix ‘a’ at the end (site 22a and 27a). Some sites were marked as suitable for wet weather sampling only since they were either dry or were a series of waterholes and not flowing at the time of assessment. Additional sites became dry in the following months. One additional gauged site on the Wingecarribee River at Bong Bong weir (53) was included because it had flow gauging instrumentation and was easy to access.

Six sites (numbered 3a, 8, 13, 22a, 48, and 49) were chosen for regular sampling. These were the best possible sites that represented different subcatchments within the Wingecarribee catchment. They were also likely to continue to flow even if drought conditions continued. Site 3a represented water flowing into the Wingecarribee Reservoir, which was likely to be impacted by the upstream township of Robertson. Sites 8, 13 and 49 were all on the Wingecarribee River, with site 8 not far downstream of the release from the reservoir, site 13 being in the urban area of Burradoo and site 49 close to the bottom of the catchment with native bushland and livestock farming in the area. Site 22a was located on Medway Rivulet, a tributary to the Wingecarribee River between sites 13 and 49, and site 48 was located on Joadja Creek, joining the Wingecarribee River just downstream of site 49. Table 6.3 describes the location and accessibility of all sites.

201

Table 6.3 Site locations for water quality monitoring in the Wingecarribee catchment

Site Site description Comments GPS reading Comments East North 1 Calaang Creek arm Cattle in area, bucket 279117 6171097 Random sampling required for sampling site 2 Calaang Creek - Urban area 279047 6169940 Random sampling Robertson site 3 Calaang Creek Private property 277110 6171197 Not sampled 3a Calaang Creek - Easy access 277493 6170936 Routine site Maugers 4 Warreeah Gully No flow 26/2/04, 275097 6172169 Not sampled difficult to access 5 Kellys Creek - Engsta Cattle in area 269530 6168306 Random sampling site 6 Kellys Creek – Inaccessible 267847 6171433 Not sampled Illawarra Hwy 7 Wingecarribee River Easy access from 268696 6175239 Random sampling - Sheepwash Bridge northern end of bridge site on reservoir side 8 Wingecarribee River Easy access 267044 6177651 Routine site - Sproules Lane 9 Wingecarribee River Swamp & Private land 264728 6177491 Not sampled Jn Kellys Creek 10 Wingecarribee River Private property 263248 6177573 Not sampled 11 Mittagong Creek u/s Easy access. No flow 265286 6181457 Random sampling Bowral (dry) on 19/2/04 site 12 Mittagong Creek - Muddy access in wet 262078 6181263 Random sampling Brickworks weather, otherwise easy site 13 Wingecarribee River Easy access, u/s 260005 6179410 Routine site - Burradoo Mittagong Creek 14 Wingecarribee River Unable to locate 259944 6179567 Not sampled sampling site, possible incorrect GPS reading 15 Wingecarribee River Private property 259843 6179551 Not sampled - d/s Mittagong Creek 16 Whites Creek - u/s Easy access 257039 6174813 Random sampling Mossvale STP site 17 Whites Creek - d/s Easy access 255587 6174216/6 Random sampling Mossvale STP 173900 site 18 Medway rivulet - Bucket required for 257004 6168758 Random site Malahide sampling 19 Medway rivulet - Easy access 256299 6170909 Random site 19th hole 20 Paynes Creek - Possible wet weather 254452 6171944 Random site Sutton Forest site, bucket required 21 Whites Creek - Bong Private land 254445 6173030 Not sampled Bong 22 Medway Rivulet - Difficult access 251778 6176959 Not sampled Hume Hwy 22a Medway Rivulet - Easy access ND ND Routine site Oldbury Farm

202

Table 6.3 (cont’d.) Site locations for water quality monitoring in the Wingecarribee catchment

Site Site description Comments GPS reading Comments East North 23 Wells Creek - Hume Bucket required for 250785 6175664 Random site Hwy sampling, no direct access 24 Stony Creek - New No flow 9/3/04 256456 6178627 Random site Berrima 25 Wingecarribee River SCA key required 255870 61719623 Random site - u/s Berrima 26 Wingecarribee River Easy access, dirt road to 254957 6180783 Random site - d/s Berrima camping ground 27 Cordeaux Creek Access using bucket, ND ND Not sampled barbed wire, no direct access 27a Cordeaux Creek 255400 6182800 Random site 28 Black Bobs Creek Private land 250856 6167193 Not sampled 29 Black Bobs Creek, Relatively easy access 247895 6170714 Random site d/s weir 30 Rocky Creek – Small tributary of Black 244283 6170184 Not sampled Illawarra Hwy Bobs Ck 31 Black Bobs Creek - No access 241504 6175732 Not sampled Brethren Fire Trail 32 Black Bobs Creek No access 241171 6177274 Not sampled 33 Emu Creek - Emu No access 240894 6177827 Not sampled Creek Fire Trail 34 Black Bobs Creek - No access 241563 6178811 Not sampled Bangadilly Fire Trail 35 Black Bobs Creek - No access 242407 6182041 Not sampled u/s jn Wingecarribee 36 Wingecarribee River No access 242433 6182089 Not sampled - u/s Black Bobs Creek 37 Wingecarribee River No access 242414 6181978 Not sampled - d/s Black Bobs Creek 38 Belanglo Creek Private property 247993 6176289 Not sampled 39 Medway Rivulet - d/s Private property 248205 6179301 Not sampled reservoir 40 Wingecarribee River Private property, 4WD 244246 6181175 Not sampled required 41 Wingecarribee River Private property 248532 6180942 Not sampled 42 Wingecarribee River Private property 250177 6181722 Not sampled - McAurthurs Crossing 43 Mandemar Creek Private property 251258 6183874 Not sampled 44 Joadja Ck - Possible wet weather 253949 6189770 Random site Wombean Caves site - swampy puddles Road 26/2/04 45 Borehole Creek - Possible wet weather 253446 6190490 Random site Wombean Caves site Road

203

Table 6.3 (cont’d.) Site locations for water quality monitoring in the Wingecarribee catchment

Site Site description Comments GPS reading Comments East North 46 Barracks Creek - No flow 9/3/04 ponded 248995 6195675 Random site Wombean Caves water on one side of Road road. Easy access. 47 Joadja Creek - Joadja Easy access 245023 6189935 Random site Ford 48 Joadja Creek - u/s jn Easy access 242453 6188405 Routine site Wingecarribee 49 Wingecarribee River Private property 241877 6187872 Routine site - Greenstead 50 Wingecarribee River Private property, 235391 6192109 Random site - u/s jn Wollondilly permission to access, 4WD required 51 Wingecarribee River Private property 254096 6181961 Not sampled Before Cordeaux Creek 52 Mittagong Creek - d/s Private property 260597 6180172 Random site Bowral STP 53 Wingecarribee River Easy access 260954 6175344 Random site - Bong Bong Weir ND = not done

204

Site 3a (Maugers)

The site consists of 3 culverts running under a gravel road. Approximately 2 km upstream is the unsewered township of Robertson, and downstream is the Wingecarribee Reservoir. Figure 6.2 shows the site from the downstream side of the culverts. Each culvert was round with a diameter of 610 mm.

Figure 6.2 Calaang Creek at Maugers (site 3a)

205

Site 8 (Wingecarribee River, Sproules Lane)

At this site the Wingecarribee River passes under the Sproules Lane bridge approximately 5 km downstream of the Wingecarribee Reservoir wall. The river passes through improved pasture grazing land with various livestock including cattle, horses and sheep. Figure 6.3 shows the river immediately upstream of the sampling site.

Figure 6.3 Wingecarribee River at Sproules Lane (site 8)

206

Site 13 (Wingecarribee River at Burradoo)

This site is just a few metres upstream of the railway bridge across the Wingecarribee River for the Main Southern Railway. The site consists of a concrete surfaced bridge across the river with 6 large rectangular culverts. Each culvert was 3040 mm across. The site is downstream of Burradoo urban area and just upstream of the confluence of Mittagong Creek. Figure 6.4 shows sampling site 13 from the downstream side on the left bank.

Figure 6.4 Wingecarribee River at Burradoo (site 13)

207

Site 22a (Medway Rivulet at Oldbury Farm)

This site is where the Medway Rivulet crosses Oldbury Road. It is in a rural farming area approximately 7 km downstream of the Moss Vale Sewage Treatment Plant and Moss Vale urban area, and is also downstream of the confluence of Paynes Creek with the Wingecarribee River. Figure 6.5 shows the ford from the right bank.

Figure 6.5 Medway Rivulet at Oldbury Road (site 22a)

208

Site 48 (Joadja Creek)

Joadja Creek flows into the Wingecarribee River approximately 1 km downstream of site 49. It consists of a bitumen road crossing 6 circular culverts of 600 mm in diameter. Upstream of the sampling site are unsewered houses and farming land. Figure 6.6 shows the downstream side of the culverts from the left bank.

Figure 6.6 Joadja Creek (site 48)

209

Site 49 (Wingecaribee River, Greensteads)

This site is on private property, a farm called Greensteads, and is just upstream of the confluence of Joadja Creek with the Wingecarribee River. The river here passes through faming land with native vegetation. Figure 6.7 shows the downstream side of the sampling site from the wooden bridge which passes over the river. This site has a flow gauging station approximately 500 m upstream from the bridge.

Figure 6.7 Wingecarribee River at Greensteads (site 49)

Sampling Program

The dry weather sampling programme was developed so that, on average, water would be collected from each of the 6 routine sites once every fortnight for a minimum period of 12 months. The weeks were split into fortnight blocks with days from Monday to Friday on weeks 1 or 2 being chosen at random to create the sampling

210

programme. Water samples were collected from the 6 routine sites on each occasion, with an additional 3 sites being chosen at random from the pool of remaining random sites (Table 6.3).

Wet weather events were sampled when they occurred. Rainfall data was obtained from the Bureau of Meteorology website (http://www.bom.gov.au) and storm fronts were monitored from the same site. Samples for a wet event were collected the day after an event was identified. As such, samples were collected approximately 48 hours after a rainfall event.

Water samples were analysed for various physico-chemical and microbiological parameters. A 250 mL sample was collected for E. coli, pH, turbidity and conductivity analyses from all sites visited. Ten litre water samples were collected for Giardia and Cryptosporidium analysis from the 6 routine sites on most programmed dry weather sampling occasions and on all wet weather occasions. The exception was during a period when the recovery efficiency of Cryptosporidium oocysts had reduced in many samples to less than 10% from some sites. The method was assessed over several weeks and a modification developed with improved recovery efficiency. Occasionally additional 10 L samples were collected and tested for Cryptosporidium and Giardia from the randomly selected sites.

Materials and methods

Physicochemical analyses

All physico-chemical analyses were carried out using standard American Public Health Association methods (APHA, 1995). The methods used were 2130 Turbidity, pages 2-8 to 2-11; 2510 Conductivity, pages 2-43 to 2-46; and 4500-H+ pH value, pages 4-65 to 4-69.

211

Bacterial analyses

Samples were collected in sterile 250 mL schott bottles and stored on ice during transportation to the laboratory. E. coli was analysed on the day of collection and enumerated using Colilert-18TM (IDEXX, Australia) with the QuantitrayTM 2000 format according to the manufacturer’s instructions. Analysis of 100 mL samples commenced within 10 hours of sample collection. An additional sample volume of 1 mL was also included when higher counts were expected, such as during rain events.

Protozoan analyses

Analysis for the enumeration of Giardia and Cryptosporidium (oo)cysts were performed using 10 L volumes of water. The samples were stored at ambient temperature until returned to the laboratory on the same day of collection and then stored at 4°C until analysed. Prior to processing, the samples were weighed and thoroughly mixed and then the volume adjusted to 10 ± 0.05 kg. Typically, analysis of samples was completed within 5 days of sample collection.

Immediately prior to concentration of particulates using hollow fibre ultrafiltration (HemoflowTM, Fresenius Medical Care, Australia) a vial of ColorSeedTM (BTF, Sydney, Australia) containing 100 Giardia cysts and 100 Cryptosporidium oocysts was added to the sample. Samples were analysed using the method initially described by Simmons et al. (Simmons et al., 2001) and further evaluated by Ferguson et al. (2004). Briefly, hollow fibre ultrafiltration (Hemoflow™, Fresenius Medical Care, Australia) were used to concentrate water samples by recirculation until the sample bag was empty and the entire sample was retained in the tubing and filter cartridge. The sample bag was then rinsed with 1 L of reagent grade water and the particulates were again concentrated using the same filter. When the sample bag was empty, a 200 mL volume of elution buffer(10 g Laureth-12, Gelman Sciences, Sydney), 0.1 g skim milk powder and 0.5 mL Antifoam (Sigma, Sydney) per litre of phosphate buffered saline (Sigma, Sydney) was recirculated through the filter at a reduced pressure for approximately 10 min. The retained sample concentrate was collected into a 250 mL centrifuge tube and the sample pelleted by centrifugation at 2500 g for 30 minutes

212

with the brake off. The supernatant was vacuum aspirated until approximately 5 mL remained above the pellet. The pellet was resuspended in the remaining supernatant and transferred to a Leighton tube containing 1 mL of SL Buffer A, 1 mL of SL Buffer B and 0.1 mL of 10%(w/v) skim milk powder. Immunomagnetic separation (IMS) was performed according to the manufacturer’s instructions (Dynal Biotech, Oslo, Norway). Following IMS the collected oocysts were filtered onto membranes (diameter 13 mm, pore size 0.8 μm, Millipore, Australia), stained with DAPI (Sigma, Australia) and FITC-labelled monoclonal antibodies (Easystain, BTF, Australia) according to the manufacturer’s instructions. The membranes were mounted on glass slides, covered with a coverslip, and viewed by epifluorescence microscopy at x 250 magnification (450-490 nm excitation, 520 nm LP emission, (Nikon Optiphot-2 with fluorescence attachment EFD-3)). Cryptosporidium oocysts and Giardia cysts were confirmed at x 400 magnification by the presence of DAPI-stained nuclei (340-380 nm excitation, 425 nm LP emission).

The recovery efficiencies obtained for some of the samples from the first three programmed sampling occasions for Cryptosporidium oocysts were consistently low, with oocysts not being detected (0% recovery) in some samples from some sites (3a, 8 and 13). An investigation into the method resulted in a modification, where skim milk powder was added to the elution buffer to a concentration of 0.1%. In addition, 0.1 mL of skim milk powder (10% w/v) was also added to the IMS. The recovery efficiency was improved by these modifications which were thus applied to all future samples (from June 2004 onwards).

Statistical analyses

The recovery efficiency, determined as the percentage of ColorSeed™ (oo)cysts detected after processing the sample compared to the number added, was used to adjust the number of counted (oo)cysts to account for losses during processing. When Cryptosporidium oocysts or Giardia cysts were not detected in a sample, the sample was assigned a number of half the detection limit for statistical analysis. Results for all parameters were averaged and standard deviations calculated.

213

Results and Discussion

Physico-Chemical analysis

The results from the physico-chemical analysis indicated that the majority of sites tested have good water quality with neutral pH and low conductivity. The highest mean conductivity was recorded at site 22a which is in an area dominated by agricultural activities (improved pasture cattle) and is also downstream from the main urban areas in the catchment including inputs from both Bowral and Moss Vale STPs. Mean turbidity was highest at sites 8 and 13 in both dry and wet weather. Site 3a was the only site that showed increased turbidity in wet weather.

Table 6.4 Physico-chemical results for water collected from the 6 routine sites

Site n Weather pH Conductivity Turbidity Condition (µS.cm-1) (NTU) Mean SD Mean SD Mean SD 3a 12 Dry 6.93 0.35 84 10 3.8 0.85 2 Wet 7.12 0.13 59 31 17 15 8 11 Dry 7.15 0.37 93 30 17 9 2 Wet 7.20 0.16 100 54 18 5.7 13 11 Dry 7.31 0.26 96 32 23 28 2 Wet 7.00 0.45 110 30 21 7.1 22a 11 Dry 7.49 0.15 550 130 3.4 1.8 2 Wet 7.40 0.37 480 35 1.9 0.2 48 12 Dry 6.95 0.44 85 17 1.9 0.75 2 Wet 6.83 0.02 78 11 4.0 0.71 49 10 Dry 7.12 0.28 180 86 2.4 1.0 2 Wet 7.19 0.08 210 130 1.65 0.07

Bacterial and Protozoan analysis

Table 6.5 shows the mean concentrations of Cryptosporidium, Giardia and E. coli for water samples collected from each of the routine water quality monitoring sites. In dry weather the highest E. coli concentrations were consistently observed at site 3a downstream of the unsewered township of Robertson. It is probable that this contamination was caused by inputs from leaking or overflowing on-site systems. Sites 8 and 13 also had high mean concentrations of E. coli, with both sites being dominated 214

by inputs from cattle grazing on improved pasture and site 13 also containing urban areas. All sites showed increased mean concentrations of E. coli in wet weather reflecting the additional inputs derived from the washoff of faecal material from the land surface. The highest mean concentrations of both Cryptosporidium and Giardia were observed at site 3a in wet weather. The mean concentrations increased by an order of magnitude compared to the dry weather means. Elevated concentrations of Cryptosporidium were also observed at site 8 suggesting significant inputs from surface runoff from cattle grazing areas in wet weather. High concentration of Cryptosporidium and Giardia were observed at site 13 in both dry and wet weather while high mean concentrations of Giardia were also observed at site 22a.

Table 6.5 Bacteriological and Protozoan results from the 6 routine sites

Site Weather n E. coli cfu.L-1 p/n Cryptosporidium Giardia Condition oocysts.10 L-1 cysts.10 L-1 Mean SD Mean SD Mean SD 3a Dry 14 5000 6100 7/8 6.6 6.5 2.5 1.2 Wet 2 3.8 x 104 3.5 x 104 2/2 76.4 100 20 0.77 8 Dry 12 1400 2000 8/9 3.3 1.8 1.7 1.4 Wet 2 9000 3060 1/2 10 1 0 13 Dry 13 1300 800 8/9 12 16 3.2 1.3 Wet 2 1.1 x 104 3100 2/2 9.2 10 5.7 2.2 22a Dry 12 840 1400 8/8 4.0 4.3 2.7 2.2 Wet 2 5300 640 2/2 4.4 3.4 9.1 8.5 48 Dry 14 500 1200 9/9 1.2 0.86 1.2 0.86 Wet 2 3900 3100 1/2 0.6 0.79 0.10 49 Dry 12 270 200 7/7 3.8 5.8 0.89 0.43 Wet 2 1600 520 2/2 4.6 5.3 0.95 0.05

215

Table 6.6 Bacteriological and Protozoan results from randomly sampled sites

Site Weather n E. coli cfu.L-1 p/n Cryptosporidium Giardia Condition oocysts.10 L-1 cysts.10 L-1 Mean SD Mean SD Mean SD 1 Dry 3 1540 1424 1/1 - 2 2 Dry 3 1820 955 - 4 Dry 1 2700 - - 5 Dry 2 1630 1440 - 6 Dry 1 310 - - 7 Dry 6 342 357 1/1 2 1 9 Dry 1 1400 - - 10 Dry 1 200 - - 11 Dry 1 970 - - 12 Dry 1 430 - - 14 Dry 1 8800 - - 15 Dry 1 740 - - 16 Dry 3 1424 1101 1/1 0 2 17 Dry 4 2212 842 1/1 2 3 18 Dry 2 3125 4065 - 19 Dry 2 843 1070 1/1 1 2 19 Wet 1 2.9 x 104 - - 20 Dry 1 860 - - 21 Dry 1 520 - - 23 Dry 1 100 - - 24 Dry 1 3800 - - 25 Dry 3 303 2027 - 26 Dry 3 913 285 1/1 5 1 27 Dry 1 100 - - 28 Dry 1 1.6 x 104 - - 29 Dry 3 1028 1234 - 31 Dry 1 970 - - 32 Dry 1 630 - - 34 Dry 1 1700 - - 35 Dry 1 1700 - - 36 Dry 1 1100 - - 37 Dry 1 1700 - - 39 Dry 1 100 - - 40 Dry 1 410 - - 41 Dry 1 410 - - 42 Dry 1 100 - - 43 Dry 1 70 - - 44 Dry 1 310 - - 45 Dry 1 5100 - - 46 Dry 1 2300 - - 47 Dry 5 1.6 x 104 2.1 x 104 2 50 70 0.5 0.5 47 Wet 1 2010 - - 50 Dry 3 118 85 1/1 1 1 51 Dry 1 850 - - 52 Dry 2 1.3 x 104 1.4 x 104 - 52 Wet 1 2.6 x 104 - 1/1 19 -

216

Water quality data from Table 6.5 and Table 6.6 were used to calculate pathogen and indicator loads and compared to the PCB model predications. Figure 6.8 to Figure 6.13 show the pathogen and E. coli loads predicted by the PCB model compared to the measured pathogen and E. coli loads in the Wingecarribee catchment. At sites that had flow gauging equipment (3a, 8, 26, 49) the load was calculated using the mean daily flow measured in ML per day. At sites without flow gauging equipment the mean daily dry weather flow was estimated based on the data from the report by Olley and Deere (2003) (Table 5.6). Wet weather loads were only calculated for those sites that had flow gauging equipment.

The load was calculated as:

m 6 ,lj = ( , ZYE llj )10 (12)

m Where E ,lj is the measured export load of microorganism (j) from sub- catchment (l), and Yj,l is the measured concentration of microorganism (j) per litre at sub-catchment (l) and Zl is the average daily flow measured in ML at the same location in sub-catchment (l). Loads calculated in this way are plotted in Figure 6.8 to Figure 6.13 as measured flow. Loads calculated using the estimated daily flow are shown as estimated flow. Measured and estimated loads for sites 3a and 22a are plotted as sub- catchments 3 and 22 respectively.

217

10 9 8

load log10 7 6 5 4

Cryptosporidium 3 2 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Predicted Estimated flow Measured flow

Figure 6.8 Predicted versus measured daily dry weather Cryptosporidium loads exported from sub-catchments in the Wingecarribee

11 10 9

load log10 8 7 6

Cryptosporidium 5 4 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Predicted Measured flow

Figure 6.9 Predicted versus measured daily intermediate wet weather Cryptosporidium loads exported from sub-catchments in the Wingecarribee

218

Observed dry weather export loads of Cryptosporidium were generally within 1-

2 log10 of the predicted export load for each sub-catchment. Sites that showed good agreement between the observed loads and the predicted loads included; 7 Sheepwash Bridge, 17 Whites Ck, 19 Medway at the 19th hole, 22a Medway at Oldbury farm, 26 Berrima, 47 Joadja Ck at the Ford, 48 Joadja Ck u/s of the Wingecarribee, 49 Greensteads, 50 Wingecarribee R u/s of the junction with the Wollondilly R and 52 Mittagong Ck d/s of Bowral STP. The exceptions were site 8 (Sproules lane) and site

13 (Burradoo) where the measured export loads were consistently 1-3 log10 higher than the 5 log10 predicted by the PCB model. Both of these sub-catchments are dominated by improved pasture grazed by cattle suggesting that the contribution of cattle faeces in dry weather may be underestimated by the model. Factors that may contribute to this include the rate of direct deposition, animal density and the estimate of cattle access to the stream network. Site 8 is also affected by the controlled release of water by the SCA from the Wingecarribee reservoir located immediately upstream of sites 7 and 8. On four sampling occasions the measured flow was 238, 334, 105 and 145 ML per day instead of the usual dry weather flow of approximately 5 ML per day. At site 3a observed export loads were sometimes similar to the model prediction of 5 log10 but were observed to almost reach 7 log10 on some occasions. A possible cause for these observations would be a higher rate of dry weather connectivity of on-site systems to the stream network than currently used in the model (1%).

The observed intermediate wet weather export loads of Cryptosporidium at site

3a were similar to the predicted export load of 8 log10 for this sub-catchment. The estimated load for site 8 was 1-2 log10 lower than the model prediction of 8.6 log10.

While at site 49 the observed wet weather load was 2-3 log10 lower than the 9.5 log10 predicted by the PCB model. However, these results were not unexpected since the size of the actual rain events sampled (25 mm and 9 mm) were smaller than the size of the rainfall event used in the intermediate wet weather model simulation (30 mm in 24 h). The greater variation from the model predictions at site 49 compared to the upstream sub-catchments may indicate that more microorganisms were attached to particles and sedimented out during the routing process than is currently estimated by the model. The level of uncertainty in the model predictions could be expected to increase with increasing distance downstream due to the cumulative effect of sub-catchment routing.

219

10 9 8 7

load log10 6 5

Giardia 4 3 2 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Predicted Estimated flow Measured flow

Figure 6.10 Predicted versus measured daily dry weather Giardia loads exported from sub-catchments in the Wingecarribee

load log10 7

6 Giardia 5

4 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Predicted Measured flow

Figure 6.11 Predicted versus measured daily intermediate wet weather Giardia loads exported from sub-catchments in the Wingecarribee

220

Observed dry weather export loads of Giardia were very close to the predicted export load for each sub-catchment, usually within 1 log10. Sub-catchments that showed good agreement between the observed loads and the predicted loads included; 1 Calaang Ck, 3a Calaang Ck at Maugers, 7 Sheepwash Bridge, 13 Burradoo, 16 Whites Ck u/s of Moss Vale STP, 19 Medway at the 19th hole, 26 Berrima, 47 Joadja Ck at the Ford, 48 Joadja Ck u/s of the Wingecarribee, 50 Wingecarribee R u/s of the junction with the Wollondilly R and 52 Mittagong Ck d/s of Bowral STP. At site 8 (Sproules lane) the majority of the measured export loads were close to the 5.6 log10 predicted by the PCB model, while three observed loads were about 2 log10 higher. This is explained by the controlled release of water by the SCA from the Wingecarribee reservoir located immediately upstream of sites 7 and 8. On these three occasions the measured flow was 334, 105 and 145 ML per day instead of the usual dry weather flow of between 4-20 ML per day. At site 49, Greensteads, the measured dry weather load varied approximately 1 log10 either side of the predicted load. The largest variation was seen at site 17 Whites Ck d/s of Moss Vale STP where the estimated load was 2 log10 below the load predicted by the model.

The measured intermediate wet weather export load of Giardia at site 3a was very similar to the predicted export load of 8 log10 for this sub-catchment. However, the estimated loads for sites 8 and 49 were 3 to 3.5 log10 lower than the loads predicted by the PCB model. Again this is likely to be related to the smaller size of the sampled events compared to the rainfall values used in the model simulation.

221

14 13 12 11 10 load log10 9 8 E. coli E. 7 6 5 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Predicted Estimated flow Measured flow

Figure 6.12 Predicted versus measured daily dry weather E. coli loads exported from sub-catchments in the Wingecarribee

12 load log10

11 E. coli E. 10

9 0 5 10 15 20 25 30 35 40 45 50 55 Sub-catchment

Predicted Measured flow

Figure 6.13 Predicted versus measured daily intermediate wet weather E. coli loads exported from sub-catchments in the Wingecarribee

222

Numerous sites showed very good agreement (less than 1 log10 variation) between the observed dry weather loads of E. coli and the predicted loads including; 1 Calaang Ck, 2 Calaang Ck at Robertson, 3a Calaang Ck at Maugers, 4 Warreah Gully, 5 Kellys Ck Engsta, 6 Kellys Ck at Illawarra Hwy, 7 Sheepwash Bridge, 10 Wingecarribee R, 11 Mittagong Ck u/s of Bowral, 12 Mittagong Ck Brickworks, 14 Wingecarribee R, 20 Paynes Ck Sutton Forest, 21 Whites Ck Bong Bong, 24 Stony Ck, 26 Berrima, 28 Black Bobs Ck, 31 Brethren Fire Trail, 34 Bangadilly Fire Trail, 35 Black Bobs Ck u/s of Wingecarribee, 39 Medway d/s of the reservoir, 40 41 and 42 Wingecarribee R, 45 Borehole Ck, 46 Barracks Ck and 50 Wingecarribee R u/s of the junction with the Wollondilly R. Most of the remaining sites showed estimated dry weather export loads of E. coli within 2 log10 of the predicted export load for the sub- catchment.

Again at site 8 (Sproules lane) some of the measured dry weather export loads were similar to the 9 log10 predicted by the PCB model, while seven observations were about 3 log10 higher. Five of the seven measured loads can be explained by the controlled release of water by the SCA from the Wingecarribee reservoir located immediately upstream of sites 7 and 8. The other two observations occurred on days with intermediate daily flow of approximately 18 ML, suggesting that the water samples were collected several days after a controlled release. Although some variation in load can also be seen at site 7, the lack of flow gauging equipment means that the estimated loads will be incorrect on those dry weather days that controlled releases occur. Water quality sampling at site 7 coincided with a controlled release of water on one occasion during this study. At sites 23, 27 and 43 estimated dry weather loads were slightly more than 2 log10 lower than predicted by the model. These three sub-catchments are predominantly cattle grazing areas suggesting that in these locations the input of cattle faeces by direct deposition, and or stocking rate or animal access were lower than the values used in the model. At site 13 and 49 many of the estimated load observations were similar to the predicted load, but a few were up to 2-3 log10 higher, and one observation at site 49 was 4 log10 higher than the predicted load. These elevated loads all coincided with controlled releases of water from the Wingecarribee reservoir. The mean daily flow at Greensteads increased from typical dry weather day flows of 5-20 ML up to 210, 285, 131, 161 and 481 ML.

223

The measured intermediate wet weather export load of E. coli at site 3a was similar to the predicted export load of 12.6 log10 for this sub-catchment. However, the estimated loads for sites 8 and 49 were 2 to 4 log10 lower than the loads predicted by the PCB model. Again this is likely to be related to the smaller size of the sampled events compared to the rainfall values used in the model simulation.

Molecular source tracing tools

Introduction

Faecal pollution of water resources from human and animal waste is a major issue affecting water quality in receiving waters. Sources of faecal pollution include industrial and municipal effluents, surface run-off, indigenous wildlife, and manure from livestock. Faecal contamination of receiving waters are considered to be a high risk to human health and thus may also cause significant economic losses due to closures of beaches and shellfish harvesting areas (Simpson, Santo Domingo & Reasoner, 2002). Identification of, and targeting mitigative action towards, the dominant source of faecal contamination in the catchment is essential for effective management of water resources (Griffith, Weisberg & McGee, 2003).

Monitoring of faecal pollution has frequently been determined using either faecal coliform bacteria or E. coli. E. coli is considered a good faecal indicator due to its presence at high concentrations in faeces, the ease with which it can be cultured and the presence of some pathogenic sub-groups within the species. Phenotypic and genotypic typing methods have been used to assess faecal E. coli isolates and further development of typing methods presents their possible application in microbial source tracking (MST). MST is the process of tracing and tracking sources of faecal pollution within a catchment and this knowledge has the potential to improve the management of water resources. MST has utilised a variety of methods, primarily to try to distinguish human from non-human sources of faecal pollution, but also to try to identify and pin- point particular sources of contamination within catchments. It is an active area of research as evidenced by recent reviews in the field (Scott et al., 2002; Simpson, Santo Domingo & Reasoner, 2002).

224

MST studies so far have favoured the use of E. coli as a faecal indicator. In this study, profiling of E. coli isolates from the catchment was performed using the genotyping methods of random amplified polymorphic DNA (RAPD), repetitive element PCR (rep-PCR), PCR of the intergenic spacer region (IGS-PCR) and ribotyping and the phenotypic-based method API profiling. The genotypic methods used in this study produced a series of bands or patterns on an agarose gel, which is called a fingerprint. Different typing methods have advantages and disadvantages for their use in MST studies and they differ in cost, time, ease-of-use and discriminative ability. MST methods are based on the concept that human and non-human hosts harbour particular populations of microorganisms which contain some strains specific to the gut environment of that host. Fingerprints for the host specific isolates could then be associated with the host of origin when detected in an environmental sample. Validation of an appropriate typing technique that can distinguish these differences between strains would facilitate identification of pollution sources within catchments.

Rep-PCR has been employed in many genotyping and MST studies (Carson et al., 2003; Jonas et al., 2003; McLellan, Daniels & Salmore, 2003). One type of rep- PCR involves PCR with extragenic repeating elements (BOX-PCR). The presence of differing numbers of BOX elements and inter-BOX distances results in highly specific fingerprints produced by bacterial species and strains within a species. It has been shown that rep-PCR targeting the BOX element is effective at grouping 100% of E. coli isolates derived from cows and chickens and greater than 78% of human, goose, chicken, pig and sheep E. coli isolates (Dombek et al., 2000).

E. coli has seven rRNA operons. The ribosomal RNA IGS region lies between the 16S and 23S rRNA genes and is under minimal selective pressure, which contrasts to the highly conserved rRNA genes that flank the region. IGS-PCR is performed using universal primers targeted to conserved sites in the 16S and 23S rRNA genes. Differences between the IGS region of differing strains is visible through changing fingerprint patterns. This method has been used successfully to discriminate between the host origin of isolates and may have further application in MST studies (Seurinck, Verstraete & Siciliano, 2003).

225

Ribotyping generates DNA fingerprints by treatment of genomic DNA with restriction enzymes followed by probing for the highly conserved rRNA genes. The generated fingerprints are based on the occurrence of polymorphisms of enzyme restriction sites (Carson et al., 2003; Carson et al., 2001; Hartel et al., 2002; Parveen et al., 1999; Scott et al., 2003). In one study discriminate analysis of ribopatterns gave an 82% average rate of correct classification to host group (Parveen et al., 1999).

RAPD-PCR employs short random sequence primers, typically 10 bp, which at low annealing temperatures hybridise at random chromosomal locations and initiate DNA synthesis (Jonas et al., 2003). The number and location of primer binding sites vary for different strains, thus producing different products visible by banding patterns on an agarose gel. The random nature of the reaction can cause problems with inter- assay reproducibility of patterns (Khandka et al., 1997).

Comparisons of genotyping methods are required to determine methods that are appropriate for use in MST studies. Suitable methods will need to have a high discriminatory power so that differentiation between host sources can be determined and the groupings also need to be reliable and reproducible.

Materials and Methods

Bacterial Strains

The 101 strains examined (Table 6.7) were isolated from water samples taken from receiving waters in the Sydney and Shoalhaven river system in NSW, Australia. Each water sample was examined for faecal coliforms using membrane filtration (APHA, 1995). Presumptive E. coli counts were confirmed by examining for the presence of acid and gas formation using Lauryl Tryptose broth incubated at 44.5 °C with a Durham tube, and production of indole from Tryptone water at 44.5 °C. E. coli isolates were confirmed using the API 20E system (bioMérieux, Marcy l’Etoile, France). Isolates were stored at -70 °C in Luria-Bertina (LB) medium supplemented with 40% (v/v) glycerol.

226

Table 6.7 Location and strain numbers for E. coli isolates collected from raw waters in the Sydney and Shoalhaven catchments

Location Site Date Isolate Faecal coliforms Numbers (cfu.100 mL-1) Wingecarribee River at 26 9/2/02 180 - 184 20 Berrima Weir Wollondilly River at E450 13/2/02 210 - 235 20 Golden Valley Wollondilly River at E450 25/3/02 346 - 348 5 Golden Valley Wingecarribee River at 26 26/3/02 361 - 403 93 Berrima Weir Wollondilly River at E450 23/4/02 451 - 467 20 Golden Valley Wingecarribee River at 26 26/4/02 471 - 476 4 Berrima Weir Wollondilly Catchment 49 14/5/02 696 7 Wollondilly Catchment 49 14/5/02 726 10* Wingecarribee Catchment 49 22/5/02 795 10* Wollondilly River at E450 20/5/02 823 4 Golden Valley

* Result for sediment sample (cfu.g-1)

DNA Extraction

The total genomic DNA used for RAPD analysis was extracted using a method for the purification of DNA from Gram-negative bacteria and cyanobacteria (Neilan, 1995). DNA for BOX and IGS-PCRs was performed by resuspending 20 mg of bacterial cells from a Luria-Bertina agar (LA) plate into 150 μL of 10 mM Tris-HCl (pH.6.0)- 100 mM EDTA- 100 mM NaCl. DNA was extracted using a phenol/chloroform/SDS method as described by Gillings and Fahy (1993). DNA yields were assessed by electrophoresis.

227

BOX-PCR

DNA fingerprinting was performed on all bacterial isolates using the rep-PCR method outlined by Gillings and Holley (1997a; 1997b) and the BOXA1R primer (CTACGGCAAGGCGACGCTGACG) (Louws et al., 1994; Versalovic, Koeuth & Lupski, 1991; Woods et al., 1993). DNA fragments between the BOX elements were amplified using the thermal program of Louws et al. (1994). BOX- PCR products were detected by separation, of an 8 μL aliquot of the completed amplification reaction, using electrophoresis on a 2% agarose gel which was cast and run in Tris-borate-EDTA (TBE) buffer (pH 8.3) (Sambrook, Fritsch & Maniatis, 1989). After initial cluster analysis was performed, a second analysis of 2% agarose gels was performed. Classification of isolates into genogroups was performed by comparison of banding patterns on agarose gels.

IGS-PCR/Rsa 1

DNA was extracted using the method of Gillings and Fahy (1993) and the IGS region was amplified from isolates using the SPRRNAF (GAAGTCGTAACAAGG) and SPRRNAR (CAAGGCATCCACCGT) primers of Lane et al. (1991). PCR was performed as for BOX-PCR, except that the magnesium concentration was lowered to 2 mM and the thermal cycling protocol was altered to; 94 °C 3 min., 1 cycle; 94 °C 30 sec., 52 °C 30 sec., 72 °C 90 sec., 35 cycles; 72 °C 5 min., 1 cycle.

A 20 μL aliquot of successful IGS amplifications was digested with the restriction enzyme Rsa1 according to the manufacturer’s instructions (Promega, Madison, USA). A 15 μL aliquot of the completed digests was separated by electrophoresis on 2% agarose gel, stained and photographed. IGS-PCR products were detected by separation in an agarose gel, as for BOX-PCR products (Sambrook, Fritsch & Maniatis, 1989). Classification of isolates into genogroups was performed by comparison of banding patterns on agarose gels.

228

Random Amplified Polymorphic DNA (RAPD)

DNA was extracted as above and RAPD-PCR was performed using the primers Cra25 (AACGCGCAAC) and Cra26 (GTGGATGCGA) (Kresovich et al., 1992). PCR reactions were performed according to Neilan (1995). RAPD-PCR products were separated on 3% agarose gels and were run in Tris-acetate-EDTA (TAE) buffer. All gels were stained with ethidium bromide and photographed using transmitted UV light and polaroid film (Sambrook, Fritsch & Maniatis, 1989). Each band visualised on a gel (Neilan, 1995) was considered a RAPD marker and part of the RAPD fingerprint generated for the E. coli isolate. Isolates were classified into genotypic groups based on the possession of the same banding pattern.

Automated ribotyping

The RiboPrinter® system (Qualicon, Inc., Wilmington, USA) was used for the automated ribotyping of the isolates. Samples were prepared according to manufacturer’s instructions. Briefly, the automated process involved the lysis of the bacterial cells and cleavage of the DNA using the restriction enzyme EcoRI. The DNA fragments were then separated by size using gel electrophoresis and analysed using a modified southern hybridisation blotting technique. The DNA was hybridised with a labelled rRNA operon probe derived from E. coli and the bands were detected using a chemiluminescent substrate. An image was captured using a charge-coupled device camera and was then electronically transferred to the system computer. Each sample lane of data was normalised to a standard marker set, characterized and identified using similarity measurements to previously run isolates and reference patterns.

229

Results and Discussion

Analysis of DNA Fingerprints

MST investigations require a method that is reliable, relevant in its discrimination of sources and both cost-effective and logistically feasible. A variety of typing methods were evaluated for their suitability as MST tools by assessing their ability to type isolates of E. coli collected from raw waters from the Sydney and Shoalhaven drinking water catchments. The methods were found to be successful at differentiating between isolates of E. coli and due to varying discriminatory abilities, the different subgroups of homozygeneous strains were resolved by each method. A comparison of the groups formed by each method is shown in Table 6.8.

230

Table 6.8 Comparison of the groups produced by different typing methods for E. coli isolates

Isolate No. API 20E ID code Ribotyping BOX IGS RAPD 180 5144572 216-45-S-1 B9 I7 E1 181 5144572 216-45-S-2 B18 I18 E2 182 5144572 216-45-S-1 B7 I7 E1 183 5144572 216-45-S-1 B7 I7 E1 184 5144572 216-45-S-2 B15 I15 E4 210 5144572 216-45-S-2 B26 I26 E3 211 5144572 216-50-S-6 B25 I25 E13 212 5144572 216-50-S-6 B25 I25 E13 213 5144572 216-48-S-1 B27 I27 E13 214 5144572 216-48-S-1 B28 I28 E6 215 5144572 216-48-S-1 B40 I40 E16 216 5144572 216-54-S-7 B42 I42 E35 217 5144572 216-50-S-6 B45 I25 E12 218 5144572 216-45-S-2 B46 I46 E21 219 5144572 216-50-S-6 B25 I25 E16 220 5144572 216-55-S-3 B47 I7 E27 221 5144572 216-50-S-6 B25 I25 E13 222 5144572 216-50-S-6 B26 I26 E4 223 5144572 216-48-S-1 B28 I28 E5 224 5144573 216-50-S-6 B25 I25 E13 225 5144572 216-50-S-6 B25 I25 E13 226 5144572 216-45-S-2 B41 I41 E9 227 5144572 216-56-S-2 B48 I7 E29 228 Didn't grow API 229 5144572 216-50-S-6 B25 I25 E14 230 5144572 216-50-S-6 B25 I25 E14 231 5144572 216-50-S-6 B25 I25 E14 232 5144572 216-50-S-6 B25 I25 E14 233 5144572 216-56-S-8 B49 I7 E31 234 5144572 216-45-S-2 B26 I26 E31 235 Missing API 346 5044552 216-57-S-2 B50 I50 E32 347 5144572 216-45-S-2 B33 I33 E31 348 5144552 216-57-S-4 B35 I35 E34 361 5144572 216-45-S-2 B1 I1 E4 362 5044552 216-45-S-7 B6 I6 E16 363 Didn't grow API 364 5144572 216-45-S-2 B1 I1 E4 365 5144572 216-45-S-2 B1 I1 E6 366 5144572 216-45-S-2 B1 I1 E4 367 5144572 216-45-S-2 B16 I16 E6 368 5144572 216-46-S-5 B8 I8 E39 369 5144572 216-46-S-5 B8 I8 E39 370 5144572 216-45-S-2 B1 I1 E22 371 5144572 216-46-S-8 B2 I2 E6 372 5144572 216-46-S-8 B2 I2 E5 373 5144572 216-47-S-2 B10 I10 E2 374 5144572 216-45-S-2 B1 I1 E5

231

Table 6.8 (cont’d.) Comparison of the groups produced by different typing methods for E. coli isolates

Isolate No. API 20E ID code Ribotyping BOX IGS RAPD 375 5144552 216-45-S-2 B3 I3 E6 376 5144572 216-45-S-2 B1 I1 E6 377 5144552 216-47-S-6 B19 I7 E28 378 5144572 216-45-S-2 B1 I1 E7 379 5144572 216-45-S-2 B1 I1 E7 380 5144572 216-48-S-1 B3 I3 E8 381 5144572 216-45-S-2 B1 I1 E22 382 5144572 216-45-S-2 B1 I1 E22 383 5144572 216-47-S-2 B10 I10 E2 384 5144572 216-47-S-2 B5 I5 E17 385 5144572 216-45-S-2 B4 I4 E8 386 5044552 216-45-S-7 B6 I6 E18 387 5144572 216-45-S-2 B4 I4 E8 388 5144572 216-45-S-2 B1 I1 E3 389 5144572 216-45-S-2 B4 I4 E8 390 5044552 216-49-S-3 B20 I20 E9 391 5044552 216-45-S-7 B6 I6 E10 392 5144572 216-45-S-1 B9 I7 E30 393 5144572 216-45-S-2 B4 I4 E4 394 5144552 216-45-S-2 B12 I12 E24 395 5144572 216-47-S-2 B5 I5 E17 396 5144572 216-47-S-2 B5 I5 E17 397 5144572 216-45-S-2 B1 I1 E3 398 5144572 216-45-S-2 B1 I1 E3 399 5144572 216-45-S-2 B1 I1 E23 400 5144572 216-45-S-2 B11 I11 E23 401 5144572 216-45-S-2 B1 I1 E23 402 5144572 216-50-S-7 B21 I21 E38 403 5144572 216-50-S-8 B22 I22 E10 451 5144572 216-57-S-5 B39 I39 E22 452 5144572 216-50-S-6 B25 I25 E15 453 5144572 216-57-S-7 B25 I25 E15 454 5144572 216-48-S-1 B32 I33 E24 455 5144572 216-48-S-1 B36 I35 E3 456 5144572 216-50-S-6 B25 I25 E3 457 5144572 216-45-S-2 B37 I28 E3 458 5144572 216-57-S-5 B39 I39 E3 459 5144572 216-50-S-6 B31 I31 E37 460 5144572 216-57-S-6 B25 I25 E16 461 5144572 216-57-S-6 B37 I37 E3 462 5144572 216-50-S-6 B33 I33 E12 463 5144572 216-57-S-5 B37 I37 E3 464 5144572 216-45-S-2 B34 I34 E11 465 5144572 216-57-S-5 B43 I43 E3 466 5144572 216-45-S-2 B38 I38 E3 467 5244572 216-50-S-6 B31 I31 E37 471 5144572 216-45-S-2 B11 I11 E4 472 5044572 216-45-S-1 B23 I7 E33 473 5144553 216-48-S-1 B13 I13 E19

232

Table 6.8 (cont’d.) Comparison of the groups produced by different typing methods for E. coli isolates

Isolate No. API 20E ID code Ribotyping BOX IGS RAPD 474 5144552 216-51-S-4 B24 null E25 475 5144572 216-45-S-2 B17 I17 E20 476 5144552 216-48-S-1 B14 I14 E4 696 Missing Ribotype 726 5144572 216-59-S-6 B44 I7 E26 795 5144572 216-59-S-7 B29 I29 E22 823 5144572 216-50-S-6 B30 I30 E4 Number of clusters 7 24 50 39 37 null = no result

Complex banding patterns were generated when DNA fingerprinting of E. coli isolates was performed with BOX primers. Reactions with BOX primers produced approximately 20 amplification products, with some bands well conserved between strains. The presence of conserved bands allowed the presumptive identification of the species. Analysis of the BOX-PCR banding patterns resolved the isolates into 50 different clonal lines (Figure 6.14). Amongst the clonal patterns, there were two BOX- pattern groups, B3 and B8, which differed by only the possession of a single, high molarity band generated by BOX-PCR. This difference could be due to the presence of a plasmid in one of the strains, resulting in an additional set of primer binding sites and thus one additional band. An advantage to the use of the BOX-PCR method was the banding patterns produced were more obvious and thus easier to interpret than those produced by IGS-PCR/ restriction analysis.

The presence of seven rRNA operons in E. coli generated multiple products during IGS-PCR, sized between 400 and 800 bp (data not shown). Although direct amplification of the IGS region could distinguish between groups of isolates, increased resolution was generated when the IGS-PCR products were digested with restriction enzymes such Rsa 1. Restriction digest of the IGS-PCR products resulted in complex DNA fragments, with major bands ranging in size from 50 bp to 600 bp (Figure 6.15). These banding patterns were used for analysis and 39 clonal lines were identified. Isolate (#474) did not generate a product with the IGS primers and also produced an aberrant BOX-PCR pattern and therefore was probably not E. coli. Analysis of the

233

banding patterns by the different methods resolved many of the strains into the same groups. Seven of the subgroups were resolved by all of the genotyping methods and thus indicates that the genetic differences detected by these methods are true indications of strain differences. Reproducible subgroups were particularly noted between the methods of BOX-PCR and IGS-PCR, thus increasing confidence in the grouping ability of BOX-PCR and IGS-PCR. Previously, a different rep-PCR method targeting a different repetitive region had been shown to have considerably better discrimination than 16S-23S spacer region typing and ribotyping (Appuhamy et al., 1997; Vila, Marcos & Jimenez de Anta, 1996). The use of the rep-PCR primer from the study by Vila et al. (1996) may result in even better discrimination than the BOX primer used in this study.

234

Figure 6.14 BOX –PCR was performed on the 101 E. coli isolates from water samples and run on 2% agarose gel. Isolates were sorted into clonal lines and a representative fingerprint from each is shown above

Figure 6.15 IGS-PCR followed by restriction digestion with Rsa1. Digests were separated on 2% agarose. Isolates were sorted into clonal lines and a representative fingerprint from each line is shown above

235

Ribotyping has been used in many MST studies and has been reported to differentiate between human and non-human sources of E. coli (Parveen et al., 1999), and have excellent reproducibility and discriminatory power (Hartel et al., 2002). In this study ribotyping generated a simple pattern with products from 2 kb to 50 kb. Pattern analysis distinguished 24 different groups of E. coli (Figure 6.16). Method comparison showed that ribotyping had a lower discriminatory ability compared to the other genotypic methods tested in this study. Alternative methods, such as BOX-PCR or IGS-PCR, may be more suitable for MST studies. Other studies comparing DNA based typing methods have also shown the superior results produced by rep-PCR over ribotyping (Carson et al., 2001; Myoda et al., 2004). Carson et al. (2001) speculated that the reason why rep-PCR excelled over ribotyping could be due to the larger number of features, in this case bands, which are produced by the former method. The increased number of bands and thus greater availability of information allows better pattern discrimination. Although isolate #474 was suspected to not be E. coli due to an aberrant BOX pattern and absent IGS-PCR pattern, it did not produce an aberrant ribopattern and patterns produced by other isolates were found to be more dissimilar than that of isolate #474.

236

Figure 6.16 Representative patterns of each E. coli ribogroups. DNA extracts from the isolates was treated with EcoRI and automated ribotyping was performed with the RiboPrinter® system (Qualicon, Inc., Wilmington, USA)

RAPD primers generated 26 differential bands on an agarose gel (Figure 6.17). Between 7 and 15 amplification products were observed for each isolate and discriminant analysis of the RAPD banding patterns characterised isolates into 39 clonal lines. Previous issues with the reliability and robustness of RAPD strain typing were attributed to the low annealing temperature and arbitrary binding nature of the primers. These factors led to variable bands produced during different PCR reactions (Khandka et al., 1997) and thus could lead to the incorrect grouping of isolates and decrease the reliability of the method. This was evident in one study where poor inter-assay reproducibility prevented all strains from being distinguished (Jonas et al., 2003). This issue was not found to be a factor in our tests, with reproducible banding patterns between assays. In this study the discriminatory power of RAPD-PCR was found to be comparatively high and supports its applicability as a MST tool. 237

Figure 6.17 RAPD patterns produced by the environmental isolates of E. coli. RAPD PCR products were separated on 3% agarose gels with a 100bp marker (M) included in each

238

API profiling is a phenotypic typing method and, like other phenotypic methods investigated previously (Griffith, Weisberg & McGee, 2003; Vogel et al., 2000), it lacked the high level of discrimination that is necessary for use as an MST method. API profiling was primarily useful as a confirmatory method of identification to ensure water isolates were E. coli, although it appears that it incorrectly identified isolate #474 as E. coli.

Appropriateness of E. coli as an MST indicator

MST tools require the target organism to be relatively abundant in faecal material and also not become inactivated too quickly when dispersed from the faecal matrix into the environment. Throughout the study, the quantity and occurrence of E. coli in the environmental grab samples was found to be highly variable, and at some sites it was either absent or present in low numbers (< 30 cfu; Table 6.7). This highlights a problem with the use of E. coli based MST tools as it does not fulfil the essential characteristics of an indicator appropriate for MST studies. Use of an E. coli based MST tool may result in false negative results, with the assumption that a water sample was free from faecal pollution when perhaps the E. coli had become inactivated but more resistant human pathogens such as Cryptosporidium and Giardia may still have been present (Payment & Franco, 1993).

Strain Diversity within a Single Water Sample

The ability to fully characterize the E. coli population within an environmental sample requires accurate representation of strains in the population. Multiple isolates from a single water sample were analysed to determine the extent of diversity or clonal variation of E. coli isolates within the sample. The accumulation analysis of groups produced by BOX- PCR is shown in Figure 6.18. Isolates were identified into clonal line and then randomised for entry into the accumulation curve to ensure independence of observations. The number of new observations of a clone is plotted against the total number of observations in the sample. The curve suggests that between 30 and 40

239

isolates need to be identified to adequately represent the clonal diversity present within an individual water sample.

2 Number of New Observations of Clones

0 0 1020304050 Number of Isolates

Figure 6.18 Accumulation Curve for the isolates of the 26-03-2002 Collection. The number of new observations is plotted against the number of total observations. Isolates were randomized for entry into the curve to ensure independence of observation

When detected in the environment, E. coli populations were shown to be transient and diverse. Although some clonal lines were isolated on different collection dates, in general each collection date recovered a set of unique clonal lines not found on any other day. The accumulation curve, Figure 6.18, was produced using BOX-PCR patterns from isolates from one collection day. The clonal distribution of strains generated a typical accumulation curve, which rises steeply before reaching an asymptote. The curve shows that at least 30-40 isolates were needed to characterise a water sample taken at a certain location on a given collection date. A similar level of diversity was found by Seurinck et al. (2003) where Chao1 analysis showed that generally 32 isolates were needed to sufficiently characterise the richness of the E. coli population in a sample. The high diversity within the environment will result in increased time and materials necessary for testing each sample. The diversity of E. coli

240

populations compounds the issue that it may be an unsuitable indicator organism for MST studies as populations within a water sample were generally of low concentrations, and the clonal lines were so diverse that it is difficult to fully characterize a water sample at a particular location at a particular point in time.

The variation present within E. coli populations could be attributed to temporal variation, a consequence of changes in the environment between the dates of sampling. Changes in the climate, rainfall and diet of animals in the area may all be factors that affect an E. coli population at a site. Spatial variation between the sampling sites may account for few of the differences in E. coli populations due to differing inputs of faecal pollution in that area. Previously, the geographic structure appeared to account for a small amount of the genetic variation observed in E. coli, only 2%, from two populations of feral house mice 15 km apart (Gordon, 1997). Yet Gordon et al. (1997) only investigated population variation in one host type and in a water sample additional host sources would impact E. coli populations. In the environment various inputs would contribute to the E. coli population. Undoubtedly both spatial and temporal variation, contributed to the very diverse and rapidly changing E. coli population observed in the samples.

Another factor for consideration is the lack of evidence supporting the assumption that the dominant E. coli community present within the host will be reflected in the community found in the environment. Differing levels of nutrients and stresses between the primary and secondary environments would lead to the selection of different strains adapted to a certain environment and these strains may not be the same in both incidences. It has been shown (Whittam, 1989) that the clonal composition of E. coli communities changes substantially during the transition from host to the external environment and only 10% of clones are recovered from the sub-populations of both primary and secondary habitats of E. coli. The stability of genotypic markers in E. coli during the transition from the host to the external environment needs to be considered. This stability may be dependant on the genotypic targets of the MST method used and the presence of markers needs to be validated in environmental and faecal samples. Additionally, other species of enteric bacteria exhibit greater geographic structures and host specificity than has been shown by E. coli (Gordon, 2001) and thus may be more appropriate as indicators in MST studies.

241

Alternate indicator organisms may have potential as MST tools. An indicator of interest is Clostridium perfringens due to its ability to persist in the environment, reported correlation with Giardia cysts and Cryptosporidium oocysts (Payment & Franco, 1993), and the increased presence of C. perfringens spores in waters impacted by point-source contamination (Sorensen, Eberl & Diksa, 1989). Previous evaluations of MST methods have shown host-specific markers to have the greatest promise as MST tools due to their ability to successfully differentiate human from non-human sources of faecal contamination (Griffith, Weisberg & McGee, 2003). Future work in identifying host-specific markers in Clostridium perfringens could result in their successful application to MST. One possibility is that C. perfringens spores could be used as an indicator of long-term faecal contamination and E. coli as an indicator of recent contamination. This means that methods for host determination of both indicators would need to be developed and optimised.

Application of BOX-PCR, IGS-PCR, and RAPD PCR typing methods to faecal isolates will identify the level of variation present among isolates from hosts. Understanding genetic fingerprints for isolates from particular hosts may then help to distinguish the origin of faecal pollution in a catchment based on the host-specific markers. This study showed BOX-PCR as the most discriminatory typing method of those compared in the study. Investigation into the host specificity of BOX patterns to hosts may prove it as an appropriate tool for MST studies.

Conclusions

Sensitivity analysis identified that manure mobilisation, concentrations of microorganisms excreted per person per day and microorganism concentrations in animal faeces were the most important model parameters determining export loads and the ranking of sub-catchments based on input loads per unit area. Other parameters which were also influential in some instances were animal density, microbial decay rates and the likelihood of direct defecation. Further investigation of these parameters to improve the estimates used in the PCB model will help to reduce the level of uncertainty. Experiments are currently underway in the SCA catchments to estimate the

242

rate of direct defecation by cattle and future work is planned to quantify manure mobilisation rates for a range of the most common and abundant types of animal faeces.

Comparison of observed pathogen and indicator loads with the predicted loads for sub-catchments in the Wingecarribee showed that in dry weather Cryptosporidium and E. coli were generally within 1-2 log10 of the predicted load, while most Giardia loads were within 1 log10 of the model predictions. Some higher than predicted loads, particularly at site 8 and 49 were caused by the controlled release of water from Wingecarribee reservoir. Although these releases would also have impacted other sites downstream of the reservoir, such as site 7 and site 13, the lack of flow gauging equipment makes it difficult to quantify the impact at these other sites. Some variations from the predicted loads were also evident in sub-catchments dominated by improved pasture land use grazed by cattle. At sites 8 and 13 Cryptosporidium loads were 1-3 log10 higher than predicted by the model, while at sites 23, 27 and 43, E. coli loads were at least 2 log10 lower than predicted by the model. These differences may be related to local variations in stocking density and/or stream access or to variability in the rate of direct faecal deposition by cattle. These results support the outcome of the sensitivity analysis that indicated the rate of direct faecal deposition and animal density were both important parameters in the model. Few conclusions can be made regarding the comparison of the observed and predicted loads during wet weather due to the paucity of wet weather events sampled. Further testing of this component of the model will be constrained by the continuing drought conditions in the Sydney catchment.

Microbial source tracking (MST) is one tool that can be used to identify the source/s of faecal pollution inputs within a defined geographic area, such as a drinking water catchment. In this study, genotypic methods were used to generate identifying DNA banding patterns for strains of the commonly used faecal indicator bacteria, E. coli. The E. coli strains were isolated from raw water samples collected from the Wingecarribee catchment and were discriminated using four genotypic methods and one phenotypic method. Extragenic repeating element polymerase chain reaction (BOX- PCR) and intergenic spacer region polymerase chain reaction (IGS-PCR) were the most discriminatory methods with 50 and 39 sub-groups resolved respectively. Less discriminatory were RAPD PCR, ribotyping and phenotypic analysis which identified 38, 24 and 7 sub-groups, respectively. E. coli concentrations were low in catchment

243

water samples and populations were highly variable both spatially and temporally. These results indicated that although genotypic methods can distinguish between E. coli strains it may be an inappropriate indicator for MST studies due to its low level abundance in some areas and high variability within water samples.

244

Chapter 7 Application of the model to the entire Sydney drinking water catchment

Description of SCA catchments

The Sydney drinking water catchment is comprised of 27 individual catchments that have been divided into 196 sub-catchments on the basis of digital elevation mapping and the stream drainage network. The catchments are described in Table 7.1 and Figure 7.1. Each sub-catchment has been given a unique number with the first two digits indicating the catchment code and the last two digits representing the sub- catchment number within that catchment. For example, sub-catchment 1701 represents the Drapers Creek sub-catchment within catchment 17 (Nattai River). The total catchment area comprises approximately 16,000 km2 and has a population of approximately 105,000 people. There are six major storage reservoirs; Warragamba (1001), Woronora (2702), Nepean (2107), Avon (2102), Cordeaux (2106) and Cataract (2104). The last four are all located within the Upper Nepean River catchment (21).

245

Table 7.1 Sub-catchments within the Sydney drinking water catchment

l : Sub-catchment number Sub-catchment name Catchment name 101 Back Ck Back & Round Mountain Cks 102 Ballalaba Back & Round Mountain Cks 103 Bourkes Ck Back & Round Mountain Cks 104 Mount Ck Back & Round Mountain Cks 105 Upper Round Mountain Ck Back & Round Mountain Cks 106 Witts Ck Back & Round Mountain Cks 201 Big Burney Boro Ck 202 Larbert Boro Ck 203 Lower Boro Ck Boro Ck 204 Millendale Ck Boro Ck 205 Upper Boro Ck Boro Ck 301 Bombay Ck Braidwood 302 Boyle Braidwood 303 Gillamatong Ck Braidwood 304 Jembaicumbene Ck Braidwood 401 Barbers Ck Bungonia Ck 402 Bungonia Ck Bungonia Ck 403 Jerrara Ck Bungonia Ck 404 Morton Bungonia Ck 405 Shoalhaven Gorge Bungonia Ck 406 Upper Bungonia Ck Bungonia Ck 501 Bainbrig Ck Endrick River 502 Bulee Ck Endrick River 503 Endrick R Endrick River 504 Halls Hill Endrick River 505 Lower Endrick R Endrick River 506 Running Ck Endrick River 507 Upper Endrick R Endrick River 508 Water Race Ck Endrick River 601 Blackheath Grose River - Blue Mts 602 Katoomba Grose River - Blue Mts 603 Woodford Grose River - Blue Mts 701 Jerrabattagulla Ck Jerrabattagulla Ck 702 Wyanbene Jerrabattagulla Ck 801 Barrengarry Ck Kangaroo River 802 Brogers Ck Kangaroo River 803 Bundanoon Ck Kangaroo River 804 Fitzroy Kangaroo River 805 Kangaroo R Kangaroo River 806 Sandy Ck Kangaroo River 807 Yarrunga Kangaroo River 808 Yarrunga Ck Kangaroo River 901 Hollanders R Kowmung River 902 Lower Kowmung Kowmung River 903 Tuglow R Kowmung River 904 Upper Kowmung Kowmung River 905 Upper Tuglow R Kowmung River 1001 Warragamba reservoir Warragamba reservoir 1101 Blue Gum Ck Little River 1102 Little River Little River 1103 Upper Little R Little River 1201 Black Dog Lower Coxs River 1202 Cedar Ck Lower Coxs River 1203 Jamison Ck Lower Coxs River 1204 Kedumba R Lower Coxs River 1205 Leura Falls Ck Lower Coxs River 246

Table 7.1 (cont’d.) Sub-catchments within the Sydney drinking water catchment

l : Sub-catchment number Sub-catchment name Catchment name 1206 Mid Kedumba R Lower Coxs River 1207 Moodys Hill Lower Coxs River 1208 Upper Kedumba R Lower Coxs River 1301 Beefsteak Ck Mid Coxs River 1302 Blackheath Ck Mid Coxs River 1303 Cullenbenbong Ck Mid Coxs River 1304 Ganbenang Ck Mid Coxs River 1305 Gibraltar Mid Coxs River 1306 Jenolan R Mid Coxs River 1307 Kanangra Ck Mid Coxs River 1308 Long Swamp Ck Mid Coxs River 1309 Lowther Ck Mid Coxs River 1310 Mary Anns Ck Mid Coxs River 1311 Megalong Ck Mid Coxs River 1312 Pulpit Hill Ck Mid Coxs River 1313 River Lett Mid Coxs River 1314 Wild Dog Mid Coxs River 1315 Yorkeys Mid Coxs River 1401 Bindi Ck Mid Shoalhaven River 1402 Corang River Mid Shoalhaven River 1403 Jerricknorra Ck Mid Shoalhaven River 1404 Meangora Mid Shoalhaven River 1405 Ningee Nimble Ck Mid Shoalhaven River 1406 Oallen Ford Mid Shoalhaven River 1501 Cookanulla Mongarlowe River 1502 Lower Mongarlowe Mongarlowe River 1503 Mid Mongarlowe Mongarlowe River 1504 Tantulean Ck Mongarlowe River 1505 Upper Mongarlowe Mongarlowe River 1506 Warrambucca Ck Mongarlowe River 1601 Bangalore Ck Mulwaree River 1602 Bullamalito Ck Mulwaree River 1603 Covan Mulwaree River 1604 Covan Ck Mulwaree River 1605 Crisps Ck Mulwaree River 1606 Gundary Ck Mulwaree River 1607 Lake Bathurst Mulwaree River 1608 Lower Mulwaree Mulwaree River 1609 Mid Mulwaree Mulwaree River 1610 Mullengullenga Mulwaree River 1611 Quialigo Ck Mulwaree River 1612 Run O Waters Ck Mulwaree River 1613 Saltpetre Ck Mulwaree River 1614 Terranna Mulwaree River 1615 Upper Mulwaree Mulwaree River 1701 Drapers Ck Nattai River 1702 Jellore Ck Nattai River 1703 Mid Nattai Nattai River 1704 Nattai Ck Nattai River 1705 Nattai R Nattai River 1706 Rocky Waterholes Ck Nattai River 1707 Upper Nattai Nattai River 1708 Wanganderry Ck Nattai River 1801 Budjong Ck Nerrimunga River 1802 Jacqua Ck Nerrimunga River 1803 Nadgigomar Ck Nerrimunga River

247

Table 7.1 (cont’d.) Sub-catchments within the Sydney drinking water catchment

l : Sub-catchment number Sub-catchment name Catchment name 1804 Nerrimunga R Nerrimunga River 1805 Upper Jacqua Ck Nerrimunga River 1806 Windellama Ck Nerrimunga River 1901 Bruce Reedy Ck 1902 Durran Durra Ck Reedy Ck 1903 Lower Reedy Ck Reedy Ck 1904 Manar Ck Reedy Ck 1905 Reedy Ck Reedy Ck 1906 Sandhills Ck Reedy Ck 1907 Upper Reedy Ck Reedy Ck 2001 Ben Bullen Upper Coxs River 2002 Farmers Ck Upper Coxs River 2003 Lake Lyell Upper Coxs River 2004 Lidsdale Upper Coxs River 2005 Marrangaroo Ck Upper Coxs River 2006 Pipers Flat Ck Upper Coxs River 2007 Wallerawang Upper Coxs River 2008 Wangcol Ck Upper Coxs River 2101 Upper Nepean Upper Nepean River 2102 Avon reservoir Upper Nepean River 2103 Broughtons Pass Upper Nepean River 2104 Cataract reservoir Upper Nepean River 2105 Cordeaux downstream Upper Nepean River 2106 Cordeaux reservoir Upper Nepean River 2107 Nepean reservoir Upper Nepean River 2108 Pheasants Nest Upper Nepean River 2109 Upper Burke River Upper Nepean River 2201 Currambene Ck Upper Shoalhaven River 2202 Lower Jinden Ck Upper Shoalhaven River 2203 Upper Jinden Ck Upper Shoalhaven River 2204 Upper Shoalhaven Upper Shoalhaven River 2301 Baw Baw Upper Wollondilly River 2302 Dixons Ck Upper Wollondilly River 2303 Heffernans Ck Upper Wollondilly River 2304 Kialla Ck Upper Wollondilly River 2305 Mummel Upper Wollondilly River 2306 Pejar Upper Wollondilly River 2307 Sooley Ck Upper Wollondilly River 2308 Upper Wollondilly Upper Wollondilly River 2401 Clemsons Ck Werri Berri Ck 2402 Horse Ck Werri Berri Ck 2403 Lower Werriberri Werri Berri Ck 2404 Unnamed Ck Werri Berri Ck 2405 Upper Werriberri Ck Werri Berri Ck 2406 Werombi Werri Berri Ck 2501 Wingecarribee reservoir Wingecarribee River 2502 Kellys Ck Wingecarribee River 2503 Kellys Ck d/s Wingecarribee River 2504 Mittagong Ck Wingecarribee River 2505 Medway Rivulet Wingecarribee River 2506 Berrima Wingecarribee River 2507 MacArthurs and Belanglo Wingecarribee River 2508 Black Bobs Ck Wingecarribee River 2509 Joadja Ck Wingecarribee River 2601 Bangadilly Wollondilly River 2602 Chain of Ponds Wollondilly River

248

Table 7.1 (cont’d.) Sub-catchments within the Sydney drinking water catchment

l : Sub-catchment number Sub-catchment name Catchment name 2603 Cookbundoon Wollondilly River 2604 Eden Forest Wollondilly River 2605 Guineacor Wollondilly River 2606 Guineacor Ck Wollondilly River 2607 Jaorimin Ck Wollondilly River 2608 Jocks Ck Wollondilly River 2609 Jooriland River Wollondilly River 2610 Junction Ck Wollondilly River 2611 Kerrawary Ck Wollondilly River 2612 Leibnitz Wollondilly River 2613 Long Swamp Ck 2 Wollondilly River 2614 Lower Murruin Ck Wollondilly River 2615 Lower Tarlo Wollondilly River 2616 Lower Wollondilly Wollondilly River 2617 Mares Forest Ck Wollondilly River 2618 Murrays Flats Wollondilly River 2619 Myrtle Ck Wollondilly River 2620 Narambulla Ck Wollondilly River 2621 Norrong Wollondilly River 2622 Paddys River Wollondilly River 2623 Upper Murruin Ck Wollondilly River 2624 Upper Tarlo Wollondilly River 2625 Uringalla Wollondilly River 2626 Uringalla Ck Wollondilly River 2627 Woolshed Ck Wollondilly River 2628 Wooroondooroonbidgee Wollondilly River 2701 Waratah Woronora River 2702 Woronora River Woronora River

249

Figure 7.1 Map of the Sydney drinking water catchments

250

Input data requirements for all SCA catchments

Hydrologic module

All parameters in the hydrologic module were set at the same values as used in the application of the model to the Wingecarribee catchment. For example, all sub- catchments used the same initial moisture deficit. The rainfall surface layer for the entire Sydney catchment was derived from the SCA GIS data base and used to modify the event rainfall for each sub-catchment by the factor (fl).

Land module

The available GIS land use data for the Sydney drinking water catchment were transformed into a subset of 13 land use classes as described in Table 6.2. The same assumptions were made regarding the density of the human population for land use areas as were used for the Wingecarribee catchment. These were 2400 people.km-2 for urban residential (λ 10), 100 people.km-2 for rural residential (λ 11), and 10 people.km-2 for agricultural land uses. The specific sub-catchment characteristics of the catchments required to run the model are shown in Appendix 1. Many of these variables were derived from the GIS land use layer (such as the proportion of each land use category present in a specific sub-catchment and sub-catchment area). However, other variables such as the location of the STP that an upstream sub-catchment is connected to (Hl) were identified and input manually. The animal and microorganism data files for the model were the same as used for the Wingecarribee catchment.

Sewage treatment plant module

In dry weather the STP loads were calculated using the arithmetic mean concentrations of the microorganisms in the post-treatment effluent for each STP (Table 7.2). These inputs to the model were calculated from the existing data on microbial quality of sewage effluent (Krogh & Paterson, 2002; Paterson & Krogh, 2003) combined with the new data (Chapter 3). In wet weather the volume of effluent that

251

may be released during an event was based on the buffer capacity for each STP and available data on overflow volumes (Table 7.3) (Paterson & Krogh, 2003). STP design and treatment capacity are summarised in Table 2.14 with more detailed information on plant processes contained in the report by Paterson and Krogh (2003).

Marulan STP consists of an aeration pond and wet weather storage dam with a large buffer capacity. No effluent is discharged from the plant and an effluent reuse scheme supplies the treated effluent to nearby lands for pasture and woodlot irrigation. Irrigation lands have a slope of less than 5% and a twenty metre wide vegetation buffer zone was created between effluent irrigation areas and drainage lines. For these reasons the input to the stream network from this plant has been set to zero by reducing the microorganism concentration input variables to 0.

252

Table 7.2 Arithmetic mean post treatment concentrations of Cryptosporidium, Giardia and E. coli in sewage effluent from STPs in the SCA area of operations (Australian Water Technologies, 2002b; Krogh & Paterson, 2002; Paterson & Krogh, 2003)

STP l : Sub-catchment cj,l : Cryptosporidium cj,l : Giardia cj,l : E. coli (oocysts.L-1)† (cysts.L-1)† (cfu.L-1) n p/n Mean SD p/n Mean SD n Mean SD Braemar 1707 37 35/37 9.4 9.6 36/37 54.9 106.2 107 36917 56992 Berrima 2506 38 8/38 0.87 4.9 18/38 3.1 12.0 85 4628 25430 Bowral 2503 40 39/40 17.6 15.5 38/40 44.1 45.4 124 9380 15618 Moss Vale 2504 38 35/38 1.2 1.5 35/38 11.7 18.5 121 197 997 Bundanoon 2622 38 13/38 0.1 0.2 23/38 1.1 4.4 80 1254 1778 Goulburn 2301 36 29/36 22.6 30.7 27/36 4.2 6.4 37 29025 33060 Braidwood 303 36 25/36 8.0 11.1 13/36 2.1 9.5 37 66 59 Mt Victoria 1313 3 3/3 100.7 165.7 3/3 115.1 20.4 185 77661 343427 Wallerawang 2006 3 1/3 0.3 0.5 3/3 45.5 45.6 3 10467 4992 Lithgow 2002 36 8/36 0.2 0.4 31/36 11.4 18.7 38 1203 1539

† adjusted for recovery efficiency using ColorSeed™ as an internal control p = number of positive samples n = total number of samples

253

Table 7.3 Buffer capacity and overflow volumes for STPs in the SCA area of operations

l : Sub-catchment b : Buffer W : Volume of STP l capacity (ML) overflow (ML) Braemar 1707 2.6 2.0 Berrima 2506 11.1 0.5 Bowral 2503 0 24.5 Moss Vale 2504 19.7 1.6 Bundanoon 2622 0 1.0 Goulburn 2301 45 0 Braidwood 303 0 0.1 Mt Victoria 1313 0 0.1 Wallerawang 2006 0 0.1 Lithgow 2002 66 0 Marulan 401 44 0

On-site sewage systems module

There are approximately 18 000 on-site systems in the Sydney drinking water catchment with 5 000 of these located in the Wingecarribee catchment. The input from on-site systems for the entire SCA catchment was calculated in the same way that the model was applied to the Wingecarribee sub-catchment. The total population was estimated based on land use type, and then the proportion of the population that is not located in an urban area and thus not connected to an STP are assumed to be using on- site systems.

In-stream module

All parameters in the in-stream module were set at the same values as used in the application of the model to the Wingecarribee catchment. For example, flow velocities for dry weather were 0.1 ms-1, 0.3 m.s-1 for intermediate wet weather and 3 m.s-1 for large wet weather events.

254

Model Outputs

Predicted loads

The model predicts for each microorganism a local generated source budget (input) and the routed downstream (export) budget. The input and export budgets for

Cryptosporidium, Giardia and E. coli are shown as both plots of the output (Figure 7.2 X to Figure 7.7) and as spatial raster diagrams (Figure 7.8 to Figure 7.13) respectively. The raster diagrams for the input loads are shown as the load generated per unit area (km2) while the raster diagrams of the export loads represent the total load exported from each sub-catchment.

In dry weather Cryptosporidium input loads were predicted to range from approximately 4 log10 in sub-catchments 602 (Katoomba) and 1401 (Bindi Ck) to as high as 6.3 and 7.8 log10 in sub-catchments 2504 (Mittagong Ck) and 2503 (Kellys Ck) respectively. Both of these latter sub-catchments are located downstream of Bowral STP in the Wingecarribee catchment. Other sub-catchments with high predicted dry weather loads of Cryptosporidium were 1608 (Mulwaree), 1001 (Warragamba reservoir), 803 (Bundanoon Ck) and 904 (Upper Kowmung) (Figure 7.2). Similar trends were predicted for the exported loads of Cryptosporidium during dry weather with most sub-catchments predicted to export approximately 5 log10 oocysts per day (Figure 7.3 and Figure 7.9). However, with the routing of oocysts to downstream catchments the exported loads impact more of the downstream sub-catchments in the Wingecarribee catchment. In wet weather Cryptosporidium input loads were predicted to range from 7-7.5 log10 in sub-catchments 2506 (Berrima) and 1603 (Covan) to as high as 10.6 log10 in 1001 (Warragamba reservoir) and 10.4 log10 in 904 (Upper Kowmung). Other sub-catchments with high predicted wet weather loads of Cryptosporidium were 803 (Bundanoon Ck) and 1306 (Jenolan R) (Figure 7.2). The exported loads of Cryptosporidium during wet weather again showed similar trends to the predicted input loads except that the exported loads were spread over a slightly wider range than the input loads with the highest exported loads reaching 11.6 log10 in sub-catchment 1001 (Warragamba reservoir) (Figure 7.3).

255

In dry weather predicted Giardia input loads were spread over a much larger range than Cryptosporidium loads. The predicted loads varied from 1.4 log10 in sub- catchment 1202 (Black Dog) and 1.8 log10 in 902 (Lower Kowmung) to as high as 7-8

log10B B in sub-catchments 2504 (Mittagong CK) and 2503 (Kellys Ck). Figure 7.4 shows that the variation between sub-catchments within a catchment were also large in some cases, particularly 05 (Endrick R), 12 (Lower Coxs R) and 09 (Kowmung R). Sub- catchments with high predicted dry weather input loads of Giardia were 2301 (Baw Baw), 2627 (Woolshed Ck), 803 (Bundanoon Ck) and 304 (Jembaicumbene Ck). Sub- catchment 2301 receives effluent from Goulburn STP. Similar trends were predicted for the exported loads of Giardia during dry weather with most sub-catchments predicted to export between 4-6 log10 cysts per day (Figure 7.5 and Figure 7.11). In wet weather Giardia input loads were predicted to range from as low as 4-5 log10 in sub- catchments 1202 (Cedar Ck) and 1102 (Little R) up to 10 log10 in sub-catchments 2502 (Kellys Ck), 702 (Wyanbene), 2627 (Woolshed Ck) and 2304 (Kialla Ck) (Figure 7.4

and FigureX 7.10). The predicted export loads of Giardia during wet weather were also similar to the trends for the input loads except that the exported loads were spread over a 4 log10 range compared to 3 log10 for the input loads. The highest predicted export load was 11.3 log10 in sub-catchment 1001 (Warragamba reservoir) (Figure 7.5).

E. coli input loads in dry weather were predicted to range from 9 log10 in sub- catchments 2404 (Unnamed Ck in Werri Berri catchment) and 2702 (Woronora R) to 12

log10B B in sub-catchments 803 (Bundanoon Ck) and 1608 (Lower Mulwaree). There was relatively little variation between sub-catchments compared to Giardia loads, with most sub-catchments predicting input loads of approximately 11 log10 mpn per day (Figure

7.6 X and Figure 7.12). The exception was catchment 12 (Lower Coxs R) which had predicted input loads ranging from 9.3 to 11.9 log10 mpn per day (Figure 7.6). Export loads of E. coli during dry weather were usually 3 log10 lower than the input load, with most sub-catchments predicting export loads of approximately 8 log10 mpn per day. The lower predicted export loads reflect the rapid die-off of E. coli in dry weather conditions compared to the more robust survival of the pathogenic protozoans. In wet weather the predicted E. coli input loads ranged from 11.5 log10 in sub-catchments 2506

(Berrima) and 602 (Katoomba) to 14.5 log10 in 1001 (Warragamba reservoir), 904 (Upper Kowmung) and 803 (Bundanoon Ck). Other sub-catchments with high predicted wet weather loads of E. coli were 2101 (Avon Upper Nepean) and 2627

256

(Woolshed Ck) (Figure 7.7 and Figure 7.13). The predicted export loads of E. coli during wet weather again showed similar trends to the predicted input loads except that the exported loads ranged from 12-15 log10 mpn compared to the input loads which ranged from 12-14 log10 mpn per day (Figure 7.7).

257

11 10 9 8 load log10

7 6 5

Cryptosporidium 4 3 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 Sub-catchment

Dry weather Intermediate Wet Large Wet

Figure 7.2 Cryptosporidium loads log10 oocysts generated daily in each sub-catchment (not in sequential downstream order)

]

258

12 11 10 9 load log10 8 7 6 5

Cryptosporidium 4 3 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 Sub-catchment

Dry weather Intermediate Wet Large Wet

Figure 7.3 Cryptosporidium loads log10 oocysts exported daily from each sub-catchment and routed downstream (not in sequential downstream order)

259

11 10 9 8 7 6 load log10 5 4

Giardia 3 2 1 0 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 Sub-catchment

Dry weather Intermediate Wet Large Wet

Figure 7.4 Giardia loads log10 cysts generated daily in each sub-catchment (not in sequential downstream order)

260

12 11 10 9 8 7

load log10 6

5 4 Giardia 3 2 1 0 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 Sub-catchment

Dry weather Intermediate Wet Large Wet

Figure 7.5 Giardia loads log10 cysts exported daily from each sub-catchment and routed downstream (not in sequential downstream order)

261

12 load log10 11

E. coli E. 10

8 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 Sub-catchment

Dry weather Intermediate Wet Large Wet

Figure 7.6 E. coli loads log10 mpn generated daily in each sub-catchment (not in sequential downstream order)

262

17 16 15 14 13 12 11 load log10 load 10 9

E. coli E. 8 7 6 5 4 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800 Sub-catchment

Dry weather Intermediate Wet Large Wet

Figure 7.7 E. coli loads log10 mpn exported daily from each sub-catchment and routed downstream (not in sequential downstream order)

263

Sub-catchment rankings

TableX 7.4 shows the top ten sub-catchments responsible for the generation of pathogens and E. coli from the Sydney drinking water catchment. The sub-catchments were ranked according to the total input load (raw ranking) and by the input load per unit area (km2).

Raw rankings

Sub-catchments 2503 and 2504 receive effluent from Moss Vale and Bowral STPs respectively, these two sub-catchments were ranked as having the highest predicted load of Cryptosporidium and Giardia in dry weather by both raw and area rankings. Sub-catchment 803 (Bundanoon Ck, Kangaroo River) had the highest load of E. coli in dry weather by the raw ranking but not placed in the top ten for the unit area rankings. The highest E. coli load by unit area was sub-catchment 1608 (Lower Mulwaree, Mulwaree River) and three sub-catchments in the Lower Coxs River catchment (1205, 1208, 1203).

Similar to the results for the Wingecarribee the wet weather rankings were much more varied than the dry weather rankings. In wet weather the highest raw rankings for both Cryptosporidium and E. coli were attributed to 1001 (Warragamba reservoir) reflecting the fact that this sub-catchment receives inputs from many of the upstream sub-catchments. Sub-catchments 803 (Bundanoon Ck) and 805 (Kangaroo River) were also highly ranked for Cryptosporidium and E. coli loads in wet weather. The highest predicted raw ranking of wet weather loads for Giardia were from sub-catchment 2502 (Kellys Ck) and 2627 (Woolshed Ck) in the Wingecarribee and Wollondilly River catchments, respectively.

Per unit area rankings

The model predicted that in dry weather the highest Cryptosporidium and Giardia loads per unit area were generated from sub-catchments 2502 (Kellys Ck) and

264

2504 (Mittagong Ck) in the Wingecarribee catchment. The highest load of E. coli generated in dry weather was predicted in sub-catchments 1608, 1205 and 1208 (Lower Mulwaree, Leura Falls Ck and Upper Kedumba R, respectively). In intermediate wet weather events the greatest load of Cryptosporidium and Giardia per unit area were from sub-catchments 802, 805 and 2503 (Brogers Ck, Kangaroo R and Kellys Ck respectively). In intermediate wet weather the greatest load of E. coli was predicted for sub-catchments 1208 and 802 (Upper Kedumba R and Brogers Ck). In large wet weather events the highest loads of Cryptosporidium per unit area were predicted in sub-catchments 2109 and 1206 (Upper Bourke R and Mid Kedumba R). The highest loads of Giardia per unit area were predicted in sub-catchments 303 and 2308 (Gillamatong Ck in Braidwood and the Upper Wollondilly). The highest load of E. coli per unit area was predicted for sub-catchment 2503 and 1208 (Kellys Ck and Upper Kedumba R).

Table 7.4 Ranking of Sydney sub-catchments generating the highest predicted pathogen and E. coli loads (raw load and per unit area)

Raw Cryptosporidium Giardia E. coli Ranking Dry Wet I Wet L Dry Wet I Wet L Dry Wet I Wet L 1 2503 1001 1001 2503 2502 2627 803 805 1001 2 2504 904 904 2504 304 2502 1608 904 904 3 1001 805 803 2627 2627 304 1203 802 803 4 803 2102 405 2301 2101 2304 1205 803 701 5 1608 802 1306 2502 805 702 2618 1001 702 6 1203 2104 404 304 2505 2505 904 2101 405 7 1205 803 1705 2304 702 2301 405 701 2101 8 904 2107 2102 1902 2304 303 404 801 2502 9 2618 1306 2107 2307 701 1902 1302 807 304 10 1302 2106 902 303 303 2624 603 2502 2606 Area Cryptosporidium Giardia E. coli Ranking Dry Wet I Wet L Dry Wet I Wet L Dry Wet I Wet L 1 2503 802 2109 2503 2503 303 1608 1208 2503 2 2504 805 1206 2504 804 2308 1205 802 1208 3 1608 2109 1307 2506 2501 2304 1208 1205 1205 4 1205 2104 507 1608 2502 1902 1203 805 1203 5 1208 2106 806 1205 303 2502 603 1203 2406 6 2506 2102 2107 1208 2504 2306 601 801 2504 7 1203 2701 2105 1203 304 304 2501 2101 2101 8 603 808 2102 603 2101 2627 1311 2501 2304 9 601 1206 1202 2308 2308 2619 1302 603 1707 10 2501 801 808 303 1309 106 2503 804 2405 Wet I = intermediate wet weather event (30 mm) Wet L = large wet weather event (100 mm) 265

a) b) c)

2 Figure 7.8 Cryptosporidium input loads log10 oocysts generated daily within each sub-catchment per km in a) dry weather, b) intermediate wet weather event and c) large wet weather event

266

a) b) c)

Figure 7.9 Cryptosporidium export loads log10 oocysts exported daily from each sub-catchment in a) dry weather, b) intermediate wet weather event and c) large wet weather event

267

a) b) c)

2 Figure 7.10 Giardia input loads log10 cysts generated daily within each sub-catchment per km in a) dry weather, b) intermediate wet weather event and c) large wet weather event

268

a) b) c)

Figure 7.11 Giardia export loads log10 cysts exported daily from each sub-catchment in a) dry weather, b) intermediate wet weather event and c) large wet weather event

269

a) b) c)

2 Figure 7.12 E. coli input loads log10 mpn generated daily within each sub-catchment per km in a) dry weather, b) intermediate wet weather event and c) large wet weather event

270

a) b) c)

Figure 7.13 E. coli export loads log10 cfu exported daily from each sub-catchment in a) dry weather, b) intermediate wet weather event and c) large wet weather event

271

Discussion

The PCB model developed and tested in the Wingecarribee catchment was subsequently applied to the entire Sydney drinking water catchment. The model was used to quantify the predicted loads generated within and exported from the 196 sub- catchments. Sub-catchments were ranked to identify those that generated the highest daily loads of Cryptosporidium, Giardia and E. coli in dry, intermediate and large wet weather events. All of the model assumptions and default values used in the application of the model to the Wingecarribee catchment were applied unchanged to the Sydney catchment.

The outputs from the model show that in dry weather the highest daily loads of both Cryptosporidium and Giardia were predicted to be generated in sub-catchments 2503 (Kellys Ck) and 2504 (Mittagong Ck) in the Wingecarribee. These sub- catchments are heavily impacted by the effluent discharged from Bowral and Moss Vale STPs, respectively. However, in wet weather the wash off of faecal material into surface runoff means that large loads of Cryptosporidium and Giardia are generated in sub-catchments dominated by improved pasture grazed by cattle. The slow decay of protozoan pathogens combined with their rapid transport in water during wet weather events results in a cumulative export of Cryptosporidium and Giardia to downstream sub-catchments (Figure 7.9 and Figure 7.11). Thus the model predicts that sub- catchments receiving inflows from a number of upstream sub-catchments will receive large loads of these pathogens during large wet weather events. For example, Warragamba reservoir would receive 4 x 1011 Cryptosporidium oocysts and 1.2 x 1011 Giardia cysts following a 100 mm in 24 h rainfall event in the Sydney catchment.

In dry weather E. coli loads are generated across all sub-catchments that contain improved pasture and agricultural livestock at around 1 x 1011 mpn per day with additional inputs from sub-catchments receiving STP effluent. The rapid die-off and limited transport of this microorganism in dry weather conditions results in fairly localized impacts as evidenced in the spatial diagram of the exported loads (Figure 7.13a). However in wet weather significant loads of E. coli are mobilised to the stream network and transported to downstream sub-catchments (Figure 7.13b and c) with Warragamba reservoir (1001) and the Lower Wollondilly (2616) predicted to receive up

272

to 5.4 x 1015 mpn following a 100 mm in 24 h rain event in the Sydney catchment. The estimation of pathogen and indicator loads that may be delivered to the lakes and storages within the water supply chain facilitates the estimation of the risk of their being transported to the water storage offtake point. A previous study by Hipsey et al. (2005) developed a hydrodynamic model to predict the fate and distribution of pathogens in lakes and reservoirs. The exported load from the PCB model can be used as the input variable to the reservoir model, and the reservoir model has already been made available to the SCA.

The identification and ranking of sub-catchments generating the highest loads of pathogens and indicator organisms (both raw loads and per unit area) enables catchment managers to target control measures to where they are most needed. The SCA have indicated that the outputs from the model will contribute to the overall process of prioritising rectification action plans for the reduction and control of contaminant risks within the Sydney catchment. The data generated from the model can also be used to inform current SCA policy and to develop subsequent management and education strategies.

Further development of the model could be undertaken to improve the outputs. This might include reviewing and refining the model assumptions and improving the estimates of those parameters identified by the sensitivity analysis as being important parameters to the model e.g. manure mobilisation rates. While some assumptions and default values will hold valid across the whole catchment, e.g. microbial decay rates, others should be replaced with values that are more appropriate for the different sub- catchments. Parameters that should be reviewed for each sub-catchment include; the fraction of urban areas connected to the sewerage system, flow velocities in dry, intermediate and large wet weather events, the level of stock access to streams and animal density estimates by land use type. For example, the default flow velocities could be replaced with measured values for those sub-catchments that have flow gauging equipment installed. Although, not all sub-catchments are gauged, at least one sub-catchment per catchment should have measured flow estimates and application of these would be preferable to using the Wingecarribee values for all of the Sydney catchment.

273

Conclusions

The application of the PCB model to the entire SCA catchment represents the first quantitative identification of those sub-catchments that represent the highest pathogen (and indicator) risk to the quality of Sydney’s raw drinking water supply. The outputs of the model should be used as “first cut budgets” to enable catchment managers to prioritise the implementation of control measures, to inform public education strategies and drive best management practices. However, the model should not remain static, incorporation of new data and replacement of default values with actual data will reduce the level of uncertainty of the outputs. Also, review of the model inputs by catchment officers to assist with the refinement of the model assumptions would enhance the reliability of the model outputs.

274

Chapter 8 General discussion

The formation of the Sydney Catchment Authority in 1999 was a direct result of the detection of Cryptosporidium and Giardia at levels of concern in Sydney’s water supply in July - September 1998 (McLellan, 1998). This incident was the result of a series of large rainfall-runoff events within the catchment following a prolonged period of drought. Although community disease surveillance data indicated that there was no detectable increase in the reported level of illness during the period of the incident, neither the source nor the genotypes of the pathogens were identified. However, it was confirmed that poorer quality water arising from surface runoff in the catchment short- circuited the main reservoir and reached the treatment plant in a much shorter time than had previously been predicted (Hawkins et al., 2000). An important task for the SCA was thus to identify and quantify the potential sources of pathogen contamination to the raw water supply. The objective being to reduce the risk from these sources by implementing a multiple-barrier, risk management approach to the operation of the water supply system. The SCA needed a tool that would integrate all of the existing and new research data into a single framework, or model that would assist catchment managers to prioritise the implementation of pathogen control measures within the catchment. No such tool existed at the time and this model is the first GIS-based model to be developed that quantifies pathogen and faecal indicator (E. coli) loads generated within and exported from drinking water catchments.

One of the difficulties with using models to simulate reality is that frequently there is insufficient good quality data to use in them. This can severely limit the usefulness and reliability of the outputs. When developing a model it is important to design it with a specific question or objective in mind. This facilitates the identification of the key parameters of importance and thus enables the modeller to identify the extent of existing data that is available, and to minimize the number of assumptions that need to be made. Although it is tempting to include as many parameters as possible, increased complexity of the model creates an increased need for input data and can lead to excessive calculation times with little improvement in reliability or accuracy of the

275

outputs. Several data constraints were apparent in the development of this model including the need for local data regarding the post treatment concentrations of pathogens and indicators in STP effluent. The main reasons for this include the variability in treatment processes used as well as the variability in pathogen excretion rates in the host populations. As pathogen testing is not routinely conducted on sewage effluent the available data sets for Cryptosporidium and Giardia concentrations in post- treatment effluent at some STPs were very small. Expansion of these datasets should be targeted as a priority and would be particularly beneficial for those plants where effluent quality was both poor and highly variable, for example Goulburn and Mt Victoria STPs. Quantifying variations in effluent quality would facilitate a cost-benefit analysis of proposed treatment plant upgrades. Quantification of the existing post treatment concentrations and identification of the likely concentrations after the STP upgrade would enable the SCA to use the PCB model to calculate the magnitude of any reduction in pathogen and indicator loads.

The PCB model was developed by assigning computational relationships to the key catchment processes that were identified as important in determining the source, fate and transport of pathogens in drinking water catchments (Figure 1.1). The input data required by the model includes local water quality, rainfall, hydrologic and GIS- based land use information. The model also includes available published data and the assumptions used in the calculation of the model were carefully defined. The mathematical model was designed to predict total daily loads of Cryptosporidium, Giardia and E. coli generated within and also exported from catchments in dry, intermediate (30 mm in 24 h) and large (100 mm in 24 h) wet weather events.

Comparison of the model predictions for the Wingecarribee catchment with the measured and estimated loads of Cryptosporidium and E. coli indicated they were generally within 1-2 log10 of the predicted load. Giardia cyst loads were often within 1 log10 of the predicted load. Considering the complexity of the processes being modelled these results were better than expected. However, it must be acknowledged that the dataset for comparison was relatively small, particularly for wet weather conditions. The collection of wet weather samples was constrained by the lack of rainfall events within the Wingecarribee and the Sydney catchment as a whole due to the ongoing drought. However, the availability of flow gauging equipment at key sites in the

276

Wingecarribee catchment make it the preferred location for continued monitoring and future wet weather sampling of rainfall events. Maintaining the capability to collect and analyse wet weather samples from the Wingecarribee should be a priority for the SCA as such data is required to test and calibrate the wet weather component of the PCB model. Until adequate wet weather comparative data can be collected the wet weather predictions of the PCB model should be used with caution. This is important because pathogen risks to the water supply are acute-actions hazards, that is, it is the extent of the peak variations that are important not the average total load. The model predicts the greatest risk of pathogen delivery to the water supply during large wet weather events that deliver diffuse pathogen loads from the land surface to the stream network. It is therefore important to verify the peak loads predicted by the model by comparing the model predictions to data collected during large wet weather events.

The application of the PCB model to the entire Sydney catchment facilitates the identification and risk ranking of those subcatchments that generate the highest load of pathogens (both raw load and per unit area) with the potential to contaminate Sydney’s raw water supply. By examining the land use distribution and sub-catchment specific source data (Appendix 1) it is possible to determine the probable origin of the microorganisms within each sub-catchment. For example, sub-catchment 303 (Gillamatong Ck) is predicted to have the highest load of Giardia per unit area in large wet weather events. In addition to receiving effluent from Braidwood STP it also has a very high proportion (91%) of its land use designated as improved pasture grazed by cattle, which would account for a high input of Giardia cysts from the land surface in wet weather.

The model predicts that in dry weather Bowral and Moss Vale STPs in the Wingecarribee catchment were the most significant source of both Cryptosporidium and Giardia in the Sydney catchment. This outcome is in alignment with the existing perception by many of the catchment stakeholders both within and external to the SCA that the majority of pollutants, including pathogens are generated from the discharge of effluent from STPs. This perception persists because the identification and quantification of point sources of pollution is comparatively easy compared to the quantification of diffuse pollution. The amenability of point source pollution to the implementation of “engineered solutions” also makes it a preferred target for

277

rectification actions compared to diffuse pollution which is not easily addressed. Catchment management of diffuse pollution involves the more difficult role of educating, informing, influencing and/or regulating catchment stakeholders to modify their behaviour, a more difficult task to implement, but probably less expensive than upgrading an STP.

However, in contrast to dry weather, the wet weather predictions from the model indicate significant inputs of pathogens and E. coli from diffuse pollution, particularly from forested land and/or improved pasture grazed by cattle. For example, the sub- catchment loads generated per unit area (Table 7.4) show that the highest loads of Cryptosporidium are generated in sub-catchments 802 (Brogers Ck), 805 (Kangaroo R), 2109 (Upper Bourke R) and 1206 (Mid Kedumba R). These sub-catchments are all dominated by native vegetation land use. Similarly, E. coli loads predicted in wet weather were also highest in sub-catchments dominated by diffuse sources of pollution (with the exception of 2503 which generates the highest E. coli load in large wet weather events once the STP overflows). These results suggest that the remediation of point sources alone will not be sufficient to mitigate the risk to the water supply from pathogens and indicators transported during wet weather events.

The PCB model is simple to use and simply requires installation of the ICMS software which is freely available from the CSIRO website and can be run on a personal computer. A significant feature of the model is the collation and interpretation of a large amount of published data regarding pathogen and animal fate and transport characteristics which are coded as default values in the pathogen and animal input data files. The data in these input files can be easily modified if the user has more specific values that may be applicable for their geographic area. The sub-catchment spreadsheet is the only input file that has to be prepared by the user and the majority of data in this file can be downloaded from a GIS database. The model itself takes minutes rather than hours to run and produce the outputs, this time could be reduced by designing a dedicated user interface. In designing the model it was important to be able to maximise the use of existing datasets. Since many water utilities already have land use data available in a GIS database incorporation of this information was seen as the most useful way to describe land use features on a broad range of scales. The data were collated in Excel spreadsheets as this file format is easy to use and negates the need for

278

further use of specialized GIS software. The model can be run at a range of scales by simply creating new sub-catchment data files.

Various input parameters can be easily changed to allow the user to simulate changes in the catchment and thus test management scenarios. For example the proportion of sub-catchment land use in any category can be altered to simulate the effect of land use changes such as increasing urbanisation or the introduction of improved pasture. The model could thus be used to develop and test various management scenarios at different locations within the catchment. For example, to simulate the effect of connecting a population in a rural residential area to an STP compared to the impact of on-site systems. The model could also be used to predict the reduction in pathogen and indicator loads that may be generated by the introduction of new control measures. For example, the predicted change in loads exported from a sub- catchment before and after an STP upgrade.

The construction of a pathogen and indicator budget enables water utility managers to prioritise research needs and to focus control measures and barriers at the most relevant locations in the delivery chain. Although pathogen tests and quantitative risk assessment are not routinely used by water supply managers to assess the public health significance of their product, such investigations fulfil a due diligence role in assessing potential risk. Their incorporation into a pathogen budget facilitates an assessment of hazardous scenarios, in what has more recently been termed hazard analysis critical control point (HACCP) risk management (Deere et al., 2001). Even though first-cut pathogen budgets may not have sufficient data to enable quantitative microbiological risk assessments to be conducted, the iterative nature of the process will facilitate progress in this direction. Within catchments, the setting of priorities requires an understanding of the relative significance of different sources of pathogens as well as the relative effectiveness of control measures in terms of potential delivery of infective pathogens to customers. Knowledge of actual significance and effectiveness is not essential. The construction of a pathogen and indicator budget promotes understanding of relative significance since it requires an understanding of human-infective pathogen sources and loads, as well as their fate and transport as they move through a catchment from land to water, and ultimately through the water supply system.

279

Another advantage of the PCB model is that by incorporating the settling and inactivation processes into the model it facilitates prediction of the delivery potential of pathogen and indicator loads to the reservoirs and storages within the water supply chain. This enables the outputs of the PCB model to be used as inputs to the hydrodynamic model developed by Hipsey et al. (2005). This reservoir model has already been made available to the SCA and will thus link directly to the PCB model. By running the PCB model and the hydrodynamic model in series with a range of rainfall event scenarios the SCA could calculate the size of rainfall event that would be likely to generate pathogen loads that could reach the offtake at the reservoir wall. The hydrodynamic model can also predict the timing, location (depth) and dispersion of the pathogens and indicators within the storages. The application of the PCB model to each of the sub-catchments within the Sydney catchment could assist the SCA to set appropriate target values for local water quality objectives. The predicted dry weather loads could be used as the 90th percentile target for water quality concentrations, and evaluation of rainfall and wet weather scenarios could be used to derive concentration values and or to characterise rainfall event parameters to use as threshold alert levels.

Limited data was available regarding the prevalence of pathogens and E. coli in the faeces of wildlife and domestic animal species within the catchment. Even with the additional data collected as part of this study, the sample sizes are still too small to be considered definitive for the Sydney catchment and further data collection, particularly for the more abundant species is highly recommended. In particular it would be useful to examine and quantify changes in prevalence and concentrations between seasons and with animal age. A study by Power et al. (2004) has already identified the importance of this factor in determining prevalence of Cryptosporidium in Eastern Grey Kangaroos. Few studies have quantified the importance of seasonal variation for Cryptosporidium prevalence and concentrations in domestic animal faeces (Sturdee et al., 2003; Wade, Mohammed & Schaaf, 2000). Results from the study by Sturdee et al. (2003) suggest that such effects in the Sydney catchment would only be detected in a long term study. The effect of animal age has been well documented in domestic livestock with numerous studies indicating that prevalence and concentration of Cryptosporidium decrease with increasing age, and that the species and genotype may also change with the age of the animal (Atwill, Johnson & Pereira, 1999).

280

One aspect of this model that has not yet been investigated but which would add significant value to the outputs would be to determine the proportion of the total export load that is potentially infectious for humans. This would enable the outputs from the model to be used in a quantitative microbial risk assessment of the Sydney drinking water supply. The organism for which this is most likely to be achievable in the short- term is Cryptosporidium. It is unlikely that all of the Cryptosporidium oocysts predicted by the model to be exported from sub-catchments would be infectious for humans. However, insufficient data is available at this time to determine the proportion of infectious oocysts. An estimate of the likelihood of an isolate having the capability to be infectious for humans can be achieved by speciating and then genotyping environmental isolates from sub-catchment sources. Comparison of Cryptosporidium genotyping methods has been performed using standardized samples of DNA (Chalmers et al., 2005) to determine the best method/s for speciating and genotyping environmental isolates of Cryptosporidium. The implications of these findings for the water industry are described in a separate publication (Ferguson et al., in press). Characterisation of the Cryptosporidium species and genotypes present in catchment sources would enable the infectious pathogen unit (IPU) budget to be estimated.

Sensitivity analysis of the model indicated that microbial decay rates were not a critical factor affecting the outputs from the model. However, it is possible that large variations in temperature, particularly the application of the model in areas affected by freezing conditions and snow would require an adjustment of the microbial inactivation factors. This would require a fairly simple change to the pathogen input data file. However, incorporation of the effect of snow melt and the runoff occurring as a result of this would require adjustment to the hydrologic component of the model. These climate conditions are extremely rare in the Sydney catchment area and are therefore not currently accounted for in the model. Neither does the model account for the land application of biosolids or manure. These practices are not widely practiced in the Sydney catchment and were therefore not included in the model. Application of the PCB model to catchments that commonly receive inputs from manure spreading or biosolids application, for example, catchments in the United Kingdom or the USA, would require modification of the model. This would most easily be achieved by adding this activity as a specific land use category in the sub-catchment input data file

281

and assigning pathogen concentrations in the pathogen input data file that are appropriate for the quality of the material.

The model could be made more stochastic by allowing certain parameters to be fitted as a range or distribution, rather than as a fixed value in the model. This would require some recoding of the model and modification of the presentation of the outputs. However, this would offer considerable advantages, particularly if the model were to be used more as a scenario analysis tool. Another way of achieving this outcome would be to incorporate the pathogen model into an existing platform that has a stochastic modelling framework. CatchMODS is a modelling platform coded using ICMS software that predicts catchment loads for nitrogen, phosphorus and sediment. The advantage of this modelling platform is that the scenario building capacity is already built-in and includes an economic assessment of the costs thus enabling cost-benefit analysis of the various catchment management options. CatchMODS also has the advantage that it can store a number of different management scenarios and compare the cost benefits of each for all of the water quality variables simultaneously. Addition of the pathogen module would therefore allow scenarios to be evaluated for sediments, nutrients, pathogens and indicators in one analysis.

The hydrologic component of the PCB model could be modified to include time series flow data. This would facilitate a more detailed prediction of microbial transport in stream networks during wet weather events. However, the need for this to be incorporated into the PCB model is offset by the recent work undertaken by Haydon (2005). Haydon (2005) is developing a model to predict the transport of E. coli from catchments using hourly timestep flow data and a microbial export coefficient. The two models could be used together in a complimentary way. The PCB model uses GIS land use data to predict the likely indicator and pathogen loads in dry and wet weather events. The predicted export loads from the PCB model could be used to predict the microbial export coefficient for various types of land uses which could then be used in the hourly timestep model to predict rapid transport of microbial loads in a range of wet weather events. Comparison of the predicted loads with measured wet weather loads could be used to both directly test the outputs from both models and to indirectly assess the level of uncertainty and variation in the microbial export coefficients and

282

subsequently to calibrate the land use module and manure mobilization rates within the PCB model.

Although the model is easy to use it requires some basic knowledge of how to run models in ICMS. The primary goal of the project was to develop the model and thus much less effort was directed towards making the model user friendly. Thus although there is a generic guide available in ICMS, there are no user interfaces telling the user what to do step by step. Thus some relatively minor improvements to the model that would make it more accessible would include the development of a “front- end” and/or web-based user interface. An additional benefit would be to automate the saving of the outputs as separate files, particularly the spatial raster maps which are currently only accessible as “screen grabs”. Additional graphs would also help with the interpretation of the model outputs. In particular, it would be useful to automatically generate pie charts showing the proportion of the total input load generated within each sub-catchment that can be attributed to each module component of the model (STPs vs. on-site systems vs. land). This would assist catchment managers to quickly compare the various sources of the input loads generated within the various sub-catchments.

One of the limitations with testing the model was the small number of sites which had flow gauging equipment installed. This problem has been previously identified as a major issue in the study of contaminant fluxes in catchments (Sivapalan et al., 2003). However, the Wingecarribee, with 4 stations, was one of the best gauged catchments in the Sydney drinking water catchment. This is a reflection of the high infrastructure, installation, maintenance and ongoing sampling costs that are associated with the collection of good quality flow and water quality data. It needs to be recognized that to effectively test the model predictions requires the simultaneous collection of both flow data and samples for water quality analysis. In addition to collecting new data, application of the model to the entire SCA catchment will facilitate the collation and review of the historical dataset to identify more data that can be compared to the model outputs. In addition to the mining of existing historical datasets further testing of the model could also be achieved by performing more sophisticated sensitivity analysis using larger ranges of perturbations and including parameters not yet evaluated. For example, changing the initial catchment moisture deficit would affect the rate of runoff and the amount of effective runoff thus impacting on the rate of

283

manure mobilization. All of the approaches outlined above will be implemented in an extension project funded by AwwaRF (Project #3124) which will commence in July 2005.

Like all models the PCB model is constrained by the quality and quantity of the available input data. It is hoped that the model will be used and improved in an iterative manner so that as new data becomes available it replaces assumptions and default values in the model. Already such data is becoming available, a recent study by Characklis et al. (2005) reported that the rate of attachment of E. coli to particles in surface/storm runoff was in the range of 30-55% and that during dry weather the rate of attachment is lower and in the range of 20-35%. This new information could be incorporated into the model by replacing the current dry weather attachment and sedimentation rate of 50% with a range of 20-35% and applying a wet weather attachment rate of 50%. Similarly, the estimates of animal density within the SCA catchment are becoming more accurate as a result of new studies. For example, it may be appropriate to reduce cattle density from 500 per km2 to 250, based on the analysis of recent aerial photographs taken within the Wingecarribee catchment (Chris Chafer pers. comm.)

However, there are still some areas of research that need to be explored. The sensitivity analysis identified manure mobilisation rates as an important parameter in the model. There is a need to examine the rate of microbial transport from a variety of faecal matrices and to quantify the role of soil surface, vegetation and hydraulic factors in controlling the rate of this transport. Another area of importance where very little data is available is the role of animal behaviour in determining the distribution of faecal material over the land (and water) surface. The SCA has recently begun a series of experiments to assess the affect of animal behaviour using radio tracking satellite collars. The collars can be used to monitor the movement of cattle at set intervals and with an accuracy of within 5 m. This data will be used to quantify the likelihood of animals defecating into streams and help to assess the effectiveness of riparian zone control measures such as fencing and/or the provision of alternate sources of drinking water.

284

The PCB model is the first quantitative, process based model to use GIS land use data to predict pathogen and indicator loads generated within and exported from the Sydney drinking water catchment. The model is constrained by the data available but can be used to provide a relative measure of the pathogen risk associated with different sub-catchments. The model ranks the sub-catchments generating the highest pathogen loads enabling them to be prioritized for the implementation of catchment management activities. Modification of the input variables allows the model to be used as a scenario tool for the comparison of alternate management practices and to predict loads for a range of rainfall events. Iterative use of the model and replacement of assumptions and default values with new data as it becomes available will improve the reliability of the model outputs and reduce the uncertainty of the model.

285

REFERENCES

American Society of Agricultural Engineers. (1999). Manure production and characteristics. In, Vol. D384.1. American Society of Agricultural Engineers, St Joseph,

Michigan, USA.

Anderson, M.A., Stewart, M.H., Yates, M.V. & Gerba, C.P. (1998) Modeling the impact of body-contact recreation on pathogen concentrations in a source drinking water reservoir. Water Research, 32(11), 3293-306.

APHA. (1995) Standard Methods for the Examination of Water and Wastewater, 19th edn. American Public Health Association, Washington D.C.

APHA. (1998) Standard Methods for the Examination of Water and Wastewater, 20th edn. American Public Health Association, Washington D.C.

Appuhamy, S., Parton, R., Coote, J.G. & Gibbs, H.A. (1997) Genomic fingerprinting of

Haemophilus somnus by a combination of PCR methods. Journal of Clinical

Microbiology, 35(1), 288-91.

Aramini, J.J., Stephen, C., Dubey, J.P., Engelstoft, C., Schwanthje, H. & Ribble, C.S.

(1999) Potential contamination of drinking water with Toxoplasma gondii oocysts.

Epidemiology and Infection, 122(2), 305-15.

Artois, M., Delahay, R., Guberti, V. & Cheeseman, C. (2001) Control of infectious diseases of wildlife in Europe. The Veterinary Journal, 162, 141-52.

Ashbolt, N.J., Grabow, W.O.K. & Snozzi, M. (2001). Indicators of microbial water quality. In Water Quality: Guidelines, Standards and Health. Risk assessment and management for water-related infectious disease, Chapter 13 (eds L. Fewtrell & J.

Bartram), pp. 289-315. IWA Publishing, London.

287

Ashbolt, N.J. & Roser, D. (2003). Interpretation and management implications of event and baseflow pathogen data. In Watershed Management for Water Supply Systems (eds

M.J. Pfeffer, D.J.V. Abs & K.N. Brooks), pp. CD-ROM. American Water Resources

Association, New York City, New York June 30 - July 2, 2003.

Ashendorff, A.S., Principe, M.A., Seeley, A., LaDuca, J., Beckhardt, L., Faber, W.W.,

Jr. & Mantus, J. (1997) Watershed protection for New York City's supply. Journal of

American Water Works Association, 89(3), 75-88.

Atherholt, T.B., LeChevallier, M.W., Norton, W.D. & Rosen, J.S. (1998) Effect of rainfall on Giardia and Cryptosporidium. Journal of American Water Works

Association, 90(9), 66-80.

Atwill, E.R., Camargo, S.M., Phillips, R., Alonso, L.H., Tate, K.W., Jensen, W.A.,

Bennet, J., Little, S. & Salmon, T.P. (2001) Quantitative shedding of two genotypes of

Cryptosporidium parvum in California ground squirrels (Spermophilus beecheyi).

Applied and Environmental Microbiology, 67(6), 2840-43.

Atwill, E.R., Harp, J.A., Jones, T., Jardon, P.W., Checel, S. & Zylstra, M. (1998)

Evaluation of periparturient dairy cows and contact surfaces as a reservoir of

Cryptosporidium parvum for calfhood infection. American Journal of Veterinary

Research, 59(9), 1116-21.

Atwill, E.R., Hoar, B., das Gracas Cabral Pereira, M., Tate, K.W., Rulofson, F. &

Nader, G. (2003) Improved quantitative estimates of low environmental loading and sporadic periparturient shedding of Cryptosporidium parvum in adult beef cattle.

Applied and Environmental Microbiology, 69(8), 4604-10.

Atwill, E.R., Hou, L., Karle, B.M., Harter, T., Tate, K.W. & Dahlgren, R.A. (2002)

Transport of Cryptosporidium parvum oocysts through vegetated buffer strips and

288 estimated filtration efficiency. Applied and Environmental Microbiology, 68(11), 5517–

27.

Atwill, E.R., Johnson, E., Klingborg, D.J., Veserat, G.M., Markegard, G., Jensen, W.A.,

Pratt, D.W., Delmas, R.E., George, H.A., Forero, L.C., Philips, R.L., Barry, S.J.,

McDougald, N.K., Gildersleeve, R.R. & Frost, W.E. (1999) Age, geographic, and temporal distribution of fecal shedding of Cryptosporidium parvum oocysts in cow-calf herds. American Journal of Veterinary Research, 60(4), 420-25.

Atwill, E.R., Johnson, E.M. & Pereira, M.d.G.C. (1999) Association of herd composition, stocking rate, and duration of calving season with fecal shedding of

Cryptosporidium parvum oocysts in beef herds. Journal of the American Veterinary

Medical Association, 215(12), 1833-38.

Atwill, E.R., McDougald, N.K. & Perea, L. (2000) Cross-sectional study of faecal shedding of Giardia duodenalis and Cryptosporidium parvum among packstock in the

Sierra Nevada Range. Equine Veterinary Journal, 32(3), 247-52.

Atwill, E.R., Sweitzer, R.A., Pereira, M.D.C., Gardner, I.A., Vanuvren, D. & Boyce,

W.M. (1997) Prevalence of and associated risk factors for shedding Cryptosporidium parvum oocysts and Giardia cysts within feral pig populations in California. Applied and Environmental Microbiology, 63(10), 3946-49.

Australian Water Technologies. (2001). Sources of Cryptosporidium and Giardia in the

Warragamba water supply catchment. In, p 52. Sydney Catchment Authority, Sydney.

Australian Water Technologies. (2002a). Pilot study investigation of potential sources of pathogens in Sydney's water supply catchments. In, p 68. Sydney Catchment

Authority, Sydney.

Australian Water Technologies. (2002b). Spatial variation of pathogens within Sydney's water supply catchments. In, p 110. Sydney Catchment Authority, Sydney.

289

Ayres, M.P. & Lombardero, M.J. (2000) Assessing the consequences of global change for forest disturbance from herbivores and pathogens. The Science of The Total

Environment, 262(3), 263-86.

Bales, R.C., Li, S.M., Maguire, K.M., Yahya, M.T., Gerba, C.P. & Harvey, R.W. (1995)

Virus and bacteria transport in a sandy aquifer, Cape Cod, MA. Ground Water, 33(4),

653-61.

Becher, K.A., Robertson, I.D., Fraser, D.M., Palmer, D.G. & Thompson, R.C.A. (2005)

Molecular epidemiology of Giardia and Cryptosporidium infections in dairy calves originating from three sources in Western Australia. Veterinary Parasitology, 123, 1-9.

Belsky, A.J., Matzke, A. & Uselman, S. (1999) Survey of livestock influences on stream and riparian ecosystems in the Western United States. Journal of Soil and Water

Conservation, 54, 419-31.

Berrilli, F., Di Cave, D., De Liberato, C., Franco, A., Scaramozzino, P. & Orecchia, P.

(2004) Genotype characterisation of Giardia duodenalis isolates from domestic and farm animals by SSU-rRNA gene sequencing. Veterinary Parasitology, 122, 193-99.

Bodley-Tickell, A.T., Kitchen, S.E. & Sturdee, A.P. (2002) Occurrence of

Cryptosporidium in agricultural surface waters during an annual farming cycle in lowland UK. Water Research, 36(7), 1880-86.

Bonadonna, L., Briancesco, R., Ottaviani, M. & Veschetti, E. (2002) Occurrence of

Cryptosporidium oocysts in sewage effluents and correlation with microbial, chemical and physical water variables. Environmental Monitoring and Assessment, 75, 241-52.

Bos, M.G., Replogle, J.A. & Clemmens, A.J. (1993). Flow measuring flumes for open channel systems. In An Inventory of Irrigation Software for Microcomputers (eds K.J.

Lenselink & M. Jurriens), pp. 262-300. Wiley, New York.

290

Bottcher, A.B. & Hiscock, J.G. (2001). WAMView - A GIS/land source based approach to watershed assessment modeling. In TMDL Science Issues Conference, pp. 112-23.

Water Environment Federation, St Louis, Missouri.

Brenner, F.J., Brenner, E.K. & Schwartz, T.E. (1999) Use of plaque assay to detect enteric viruses in a rural watershed. Journal of Environmental Quality, 28(3), 845-49.

Brookes, J.D., Davies, C.M., Hipsey, M.R. & Antenucci, J.P. (submitted) Determining the size of Cryptosporidium-associated particles from cow faeces under simulated rainfall and risk reduction by settling within water supply reservoirs. Water and Health.

Bull, R.A., Hansman, G.S., Clancy, L., E., Tanaka, M.M., Rawlinson, W.D. & White,

P.A. (in preparation) Analysis of recombinant noroviruses reveals that recombination occurs within the ORF1/ORF2 overlap; a putative RNA subgenomic promoter.

Epidemiology and Infection.

Bull, S.A., Chalmers, R.M., Sturdee, A.P. & Healing, T.D. (1998) A survey of

Cryptosporidium species in Skomer bank voles (Clethrionomys glareolus skomerensis).

Journal of Zoology, 244(1), 119-22.

Camesano, T.A. & Logan, B.E. (1998) Influence of fluid velocity and cell concentration on the transport of motile and nonmotile bacteria in porous media. Environmental

Science & Technology, 32(11), 1699-708.

Carraro, E., Fea, E., Salva, S. & Gilli, G. (2000) Impact of a wastewater treatment plant on Cryptosporidium oocysts and Giardia cysts occurring in a surface water. Water

Science and Technology, 41(7), 31-37.

Carson, A.C., Shear, B.L., Ellersieck, M.R. & Schnell, J.D. (2003) Comparison of ribotyping and repetitive extragenic palindromic-PCR for identification of fecal

Escherichia coli from humans and animals. Applied and Environmental Microbiology,

69(3), 1836-39.

291

Carson, C.A., Shear, B.L., Ellersieck, M.R. & Asfaw, A. (2001) Identification of fecal

Escherichia coli from humans and animals by ribotyping. Applied and Environmental

Microbiology, 67(4), 1503-07.

Castro-Hermida, J.A., Gonzales-Losada, Y.A. & Ares-Mazas, E. (2002) Precalence of and risk factors involved in the spread of neonatal bovine cryptosporidiosis in Galicia

(NW Spain). Veterinary Parasitology, 106, 1118-20.

Causape, A.C., Quilez, J., Sanchez-Acedo, C. & Cacho, E.d. (1996) Prevalence of intestinal parasites, including Cryptosporidium parvum, in dogs in Zaragoza city, Spain.

Veterinary Parasitology, 67(3), 161-67.

Chalmers, R., Ferguson, C.M., Caccio, S., Gasser, R., Abs EL-Osta, Y., Heijnen, L.,

Xiao, L., Elwin, K., Hadfield, S., Sinclair, M. & Stevens, M. (2005) Direct comparison of selected methods for genetic categorisation of Cryptosporidium parvum and

Cryptosporidium hominis species. International Journal for Parasitology, 35(4), 397-

410.

Chalmers, R.M., Elwin, K., Reilly, W.J., Irvine, H., Thomas, A.L. & Hunter, P.R.

(2002) Cryptosporidium in farmed animals: the detection of a novel isolate in sheep.

International Journal for Parasitology, 32(1), 21-26.

Chalmers, R.M., Sturdee, A.P., Bull, S.A., Miller, A. & Wright, S.E. (1997) The prevalence of Cryptosporidium parvum and C. muris in Mus domesticus, Apodemus sylvaticus and Clethrionomys glareolus in an agricultural system. Parasitology

Research, 83(5), 478-82.

Characklis, G.W., Dilts, M.J., Simmons, O.D., Likirdopulos, C.A., Krometis, L.-A.H. &

Sobsey, M.D. (2005) Microbial partitioning to settleable particles in stormwater. Water

Research, 39, 1773-82.

292

Charles, K., Ashbolt, N., Roser, D., Deere, D. & McGuinness, R. (2001) Australasian standards for on-site sewage management: Application in the Sydney drinking water catchments. Water Journal of the Australian Water & Wastewater Association, 28(2),

58-64.

Charles, K., Schijven, J., Ferguson, C., Roser, D., Deere, D. & Ashbolt, N. (2003a).

Human enteric viruses: designing on-site sewage disposal systems to protect public health. In On-site 03. Future Directions of on-site systems: Best Management Practice,

30 September-2 October, pp. (CD-ROM). University of New England, Armidale, NSW.

Charles, K.J. & Ashbolt, N.J. (2004). Quantitative Microbial Risk Assessment: a catchment management tool to delineate setback distances for septic systems. In Young

Researchers Conference 2004 (eds P. Lens & R. Stuetz), pp. 139-46. IWA,

Wageningen, the Netherlands.

Charles, K.J., Ashbolt, N.J., Deere, D.A. & Roser, D.J. (2003b). Disinfection in Aerated

Wastewater Treatment Systems. In Ozwater AWA Annual Conference 2003, pp. [CD-

ROM], Perth.

Charles, K.J., Ashbolt, N.J., Ferguson, C., Roser, D.J., McGuinness, R. & Deere, D.A.

(2003c) Impacts of centralised versus decentralised sewage systems on water quality in

Sydney's drinking water catchments. Water Science & Technology, 48(11-12), 53-60.

Chauret, C., Armstrong, N., Fisher, J., Sharma, R., Springthorpe, S. & Sattar, S. (1995)

Correlating Cryptosporidium and Giardia with microbial indicators. Journal of

American Water Works Association, 87(11), 76-84.

Chmelik, V., Ditrich, O., Trnovcova, R. & Gutvirth, J. (1998) Clinical features of diarrhoea in children caused by Cryptosporidium parvum. Folia Parasitologica, 45(2),

170-72.

293

Cilimburg, A., Monz, C. & Kehoe, S. (2000) Wildland recreation and human waste: A review of problems, practices and concerns. Environmental Management, 25(6), 587-

98.

Clough, H.E., Clancy, D., O'Neill, P.D. & French, N.P. (2003) Bayesian methods for estimating pathogen prevalence within groups of animals from faecal-pat sampling.

Preventive Veterinary Medicine, 58(3-4), 145-69.

Cole, D.N. (1990). Ecological impacts of wilderness recreation and their management.

In Wilderness Management (eds J. Hendee, G. Stankey & R. Lucas), pp. 425-66. USDA

Forest Service, Washington, D.C.

Collins, R. & Rutherford, K. (2004) Modelling bacterial water quality in streams draining pastoral land. Water Research, 38, 700-12.

Conboy, M.J. & Goss, M.J. (2000) Natural protection of groundwater against bacteria of fecal origin. Journal of Contaminant Hydrology, 43(1), 1-24.

Cox, P., Griffith, M., Angles, M., Deere, D.A. & Ferguson, C.M. (2005) Concentrations of pathogens and indicators in animal feces in the Sydney watershed. Applied and

Environmental Microbiology, 71(10), 5929-5934.

Crane, S.R. & Moore, J.A. (1986) Modeling enteric bacterial die-off: A review. Water,

Air, & Soil Pollution, 27, 411-39.

Crane, S.R., Westerman, P. & Overcash, M.R. (1980) Die-off of faecal indicator organisms following land application of poultry manure. Journal of Environmental

Quality, 9(3), 531-37.

Craun, G.F. (1979) Waterborne disease outbreaks in the United States. Journal of

Environmental Health, 41(5), 259-65.

Croke, B.F.W. & Jakeman, A. (2004) A catchment moisture deficit module for the

IHACRES rainfall-runoff model. Environmental Modelling and Software, 19, 1-5.

294

Crowther, J., Wyer, M.D., Bradford, M., Kay, D. & Francis, C.A. (2003) Modelling faecal indicator concentrations in large rural catchments using land use and topographic data. Journal of Applied Microbiology, 94(6), 962-73.

Curriero, F.C., Patz, J.A., Rose, J., B. & Lele, S. (2001) The association between extreme precipitation and waterborne disease outbreaks in the United States, 1948-

1994. American Journal of Public Health, 91(8), 1194-99.

Davies-Colley, R.J., Nagels, J., Smith, R., Young, R. & Phillips, C. (2002). Water quality impact of cows crossing an agricultural stream, the Sherry River, New Zealand.

In 6th IWA International Symposium on Diffuse Pollution, pp. 671-78. International

Water Association, Amsterdam, The Netherlands.

Davies, C., Kaucner, C., Altavilla, N., Ashbolt, N., Hijnen, W., Medema, G., Deere, D.,

Krogh, M. & Ferguson, C. (2004a) Pathogen fate and transport in surface water flow.

Water (Australia), 31(3), 57-62.

Davies, C.M., Altavilla, N., Krogh, M., Ferguson, C.M., Deere, D.A. & Ashbolt, N.J.

(2005a) Environmental inactivation of Cryptosporidium oocysts in catchment soils.

Journal of Applied Microbiology, 98(2), 308-17.

Davies, C.M., Ferguson, C.M., Kaucner, C., Altavilla, N., Deere, D.A. & Ashbolt, N.J.

(2004b) Dispersion and transport of Cryptosporidium oocysts from fecal pats under simulated rainfall events. Applied and Environmental Microbiology, 70(2), 1151-59.

Davies, C.M., Kaucner, C., Altavilla, N., Ashbolt, N., Ferguson, C.M., Krogh, M.,

Hijnen, W., Medema, G. & Deere, D. (2005b). Fate and transport of surface water pathogens in watersheds. In, p 267. American Water Works Association Research

Foundation, Denver.

295

Davies, C.M., Kaucner, C., Deere, D.D. & Ashbolt, N.J. (2003) Recovery and enumeration of Cryptosporidium parvum from animal fecal matrices. Applied and

Environmental Microbiology, 69(5), 2842–47.

Davies, C.M., Logan, M.R., Rothwell, V.L., Krogh, M., Ferguson, C.M., Charles, K.,

Deere, D.A. & Ashbolt, N.J. (in press) Soil inactivation of viruses in septic seepage.

Letters in Applied Microbiology.

Davies, C.M., Long, J.A.H., Donald, M. & Ashbolt, N.J. (1995) Survival of fecal microorganisms in marine and freshwater sediments. Applied and Environmental

Microbiology, 61(5), 1888-96. de Graaf, D.C., Vanopdenbosch, E., Ortega-Mora Luis, M., Abbassi, H. & Peeters, J.E.

(1999) A review of the importance of cryptosporidiosis in farm animals. International

Journal for Parasitology, 29(8), 1269-87.

Deborde, D.C., Woessner, W.W., Lauerman, B. & Ball, P.N. (1998) Virus occurrence and transport in a school septic system and unconfined aquifer. Ground Water, 36(5),

825-34.

Deere, D.D., Stevens, M., Davison, A.D., Helm, G. & Dufour, A. (2001). Management

Strategies. In Water Quality: Guidelines, Standards and Health. Risk Assessment and

Management for Water-related Infectious Disease (eds L. Fewtrell & J. Bartram), pp.

257-88. IWA Publishing, London.

Dombek, P.E., Johnson, L.K., Zimmerley, S.T. & Sadowsky, M.J. (2000) Use of repetitive DNA sequences and the PCR to differentiate Escherichia coli isolates from human and animal sources. Applied and Environmental Microbiology, 66(6), 2572-77.

Dorner, S.M., Huck, P.M. & Slawson, R.M. (2003). Estimating pathogen source terms in a mixed-use watershed. In American Water Resources Association. American Water

Resources Association.

296

Dorner, S.M., Huck, P.M. & Slawson, R.M. (2004) Estimating potential environmental loadings of Cryptosporidium spp. and Campylobacter spp. from livestock in the Grand

River watershed, Ontario, Canada. Environmental Science & Technology, 38(12), 3370-

80.

Emerson, D.J. & Cabelli, V.J. (1982) Extraction of Clostridium perfringens spores from bottom sediment samples. Applied and Environmental Microbiology, 44(5), 1144-49.

Enemark, H.L., Ahrens, P., Lowery, C.J., Thamsborg, S.M., Enemark, J.M.D., Bille-

Hansen, V. & Lind, P. (2002) Cryptosporidium andersoni from a Danish cattle herd: identification and preliminary characterisation. Veterinary Parasitology, 107, 37-49.

Fayer, R., Fischer, J.R., Sewell, C.T., Kavanaugh, D.M. & Osborn, D.A. (1996)

Spontaneous cryptosporidiosis in captive white-tailed deer (Odocoileus virginianus).

Journal of Wildlife Diseases, 32(4), 619-22.

Fayer, R., Trout, J.M., Graczyk, T.K. & Lewis, E.J. (2000) Prevalence of

Cryptosporidium, Giardia and Eimeria infections in post-weaned and adult cattle on three Maryland farms. Veterinary Parasitology, 93(2), 103-12.

Feachem, R.G., Bradley, D.J., Garelick, H. & Mara, D.D. (1983) Sanitation and disease health aspects of excreta and wastewater management John Wiley & Sons (for The

World Bank), Chichester, England.

Fegan, N. & Desmarchelier, P. (2003). Use of AIMS to enumerate E. coli O157 in cattle faeces. In 90th Annual Meeting of International Association for Food Protection.

International Association for Food Protection, New Orleans.

Ferguson, C.M., Altavilla, N., Ashbolt, N.J. & Deere, D.A. (2003a) Prioritizing

Watershed Pathogen Research. Journal of American Water Works Association, 95(2),

92-102.

297

Ferguson, C.M., Ashbolt, N.J. & Deere, D.A. (2004) Prioritisation of catchment management in the Sydney catchment - construction of a pathogen budget. Water

Science and Technology: Water Supply, 4(2), 35-38.

Ferguson, C.M., Coote, B.G., Ashbolt, N.J. & Stevenson, I.M. (1996) Relationships between indicators, pathogens and water quality in an estuarine system. Water

Research, 30(9), 2045-54.

Ferguson, C.M., Davies, C.M., Kaucner, C., Krogh, M., Rodehutskors, J., Deere, D.A.

& Ashbolt, N.J. (in press) Field scale transport of Cryptosporidium parvum, E. coli and

PRD1 bacteriophage in surface runoff from bovine faecal pats under simulated rainfall.

Journal of Water and Health.

Ferguson, C.M., de Roda Husman, A.M., Altavilla, N., Deere, D. & Ashbolt, N.J.

(2003b) Fate and transport of surface water pathogens in watersheds. Critical Reviews in Environmental Science and Technology, 33(3), 299-361.

Ferguson, C.M., Deere, D.A., Sinclair, M., Chalmers, R.M., Elwin, K., Hadfield, S.,

Xiao, L., Ryan, U., Gasser, R., Abs EL-Osta, Y. & Stevens, M. (in press) Application of genotyping methods to assess pathogen risks from Cryptosporidium in watersheds.

Environmental Health Perspectives.

Ferguson, C.M., Kaucner, C., Krogh, M., Deere, D. & Warnecke, M. (2004)

Comparison of methods for the concentration of Cryptosporidium oocysts and Giardia cysts from raw waters. Canadian Journal of Microbiology, 50, 675-82.

298

Fleming, R., Hocking, D., Fraser, H. & Alves, D. (1999). Extent and magnitude of agricultural sources of Cryptosporidium in surface water. In National Soil and Water

Conservation Program, p 33. Agricultural Adaptation Council, West Guelph, ON.

Fraser, R.H., Barten, P.K. & Pinney, D.A.K. (1998) Predicting stream pathogen loading from livestock using a geographical information system-based delivery model. Journal of Environmental Quality, 27(4), 935-45.

Fraser, R.H., Barten, P.K. & Tomlin, C.D. (1996). SEDMOD: A GIS-based method for estimating distributed sediment delivery ratios. In American Water Resources

Symposium on GIS and Water Resources (ed C.A. Hallam), Vol. AWRA TPS-96-3, pp.

137-46. American Water Resources Symposium, Fort Lauderdale, Florida.

Freire-Santos, F., Oteiza-Lopez, A.M., Vergara-Castiblanco, C.A. & Ares-Mazas, E.

(2000) Study of the combined influence of environmental factors on viability of

Cryptosporidium parvum oocysts in water evaluated by fluorogenic vital dyes and excystation techniques. Veterinary Parasitology, 89(4), 253-59.

Garber, L.P., Salman, M.D., Hurd, H.S., Keefe, T. & Schlater, J.L. (1994) Potential risk factors for Cryptosporidium infection in dairy calves. Journal of the American

Veterinary Medical Association, 205(1), 86-91.

Garcia-Aljaro, C., Bonjoch, X. & Blanch, A.R. (2005) Combined use of an immunomagnetic separation method and immunoblotting for the enumeration and isolation of Escherichia coli O157 in wastewaters. Journal of Applied Microbiology,

98(3), 589-97.

Gary, H.L., Johnson, S.R. & Ponce, S.L. (1983) Cattle grazing impact on surface water quality in a Colorado front range stream. Journal of Soil and Water Conservation, 38,

124-28.

299

Geldreich, E.E. (1978). Bacterial populations and indicator concepts in feces, sewage, stormwater and solid wastes. In Indicators of Viruses in Water and Food (ed G. Berg), pp. 51-97. Ann Arbor Science, Ann Arbor, Mi.

Geldreich, E.E., Bordner, R.H., Huff, C.B., Clark, H.F. & Kabler, P.W. (1962) Type distribution of coliform bacteria in the feces of warm-blooded animals. Journal of the

Water Pollution Control Federation, 34(3), 295-301.

Gerba, C. (2000a). Domestic wastes and waste treatment. In Environmental

Microbiology (eds R.M. Maier, I.L. Pepper & C.P. Gerba), pp. 505-34. Academic Press,

San Diego.

Gerba, C.P. (2000b) Assessment of enteric pathogen shedding by bathers during recreational activity and its impact on water quality. Quantitative Microbiology, 2(1),

55-68.

Gerba, C.P., Riley, K.R., Nwachuku, N., Ryu, H. & Abbaszadegan, M. (2003) Removal of Encephalitozoon intestinalis, calicivirus, and coliphages by conventional drinking water treatment. J Environ Sci Health Part A Tox Hazard Subst Environ Eng, 38, 1259-

68.

Gillings, M. & Holley, M. (1997a) Amplification of anonymous DNA fragments using pairs of long primers generates reproducible DNA fingerprints that are sensitive to genetic variation. Electrophoresis, 18(9), 1512-18.

Gillings, M.R. & Fahy, P.C. (1993) Genetic diversity of Pseudomonas solanacearum biovars 2 and N2 assessed using restriction analysis of total genomic DNA. Plant

Pathology, 42, 744-53.

Gillings, M.R. & Holley, M.P. (1997b) Repetitive element PCR fingerprinting (rep-

PCR) using enterobacterial repetitive intergenic consensus (ERIC) primers is not necessarily directed at ERIC elements. Letters in Applied Microbiology, 25(1), 17-21.

300

Ginsberg, H.S. (1980). Reoviruses and epidemic acute gastroenteritis. In Microbiology

(eds B.D. Davis, R. Dulbecco, H.N. Eisen & H.S. Ginsberg), pp. 1205-16. Harper and

Row, Pensylvannia.

Gordon, D.M. (1997) The genetic structure of Escherichia coli populations in feral house mice. Microbiology, 143(6), 2039-46.

Gordon, D.W. (2001) Geographical structure and host specificity in bacteria and the implications for tracing the source of coliform contamination. Microbiology, 147, 1079-

85.

Graczyk, T.K., Evans, B.M., Shiff, C.J., Karreman, H.J. & Patz, J.A. (2000)

Environmental and geographical factors contributing to watershed contamination with

Cryptosporidium parvum oocysts. Environmental Research, 82(3), 263-71.

Graczyk, T.K., Fayer, R., Trout, J.M., Lewis, E.J., Farley, C.A., Sulaiman, I. & Lal,

A.A. (1998) Giardia sp. cysts and infectious Cryptosporidium parvum oocysts in the feces of migratory Canada geese (Branta canadensis). Applied and Environmental

Microbiology, 64(7), 2736-38.

Griffith, J.F., Weisberg, S.B. & McGee, C.D. (2003) Evaluation of microbial source tracking methods using mixed fecal sources in aqueous test samples. Journal of Water and Health, 1(4), 141-51.

Hackett, T. & Lappin, M.R. (2003) Prevalence of enteric pathogens in dogs of north- central Colorado. Journal of the American Animal Hospital Association, 39(1), 52.

Hafez, E.S.W., Schein, M.W. & Ewbank, R. (1969). The behaviour of cattle. In The behaviour of domestic animals (ed E.S.W. Hafez), pp. 279-81. The Williams and

Wilkins Company, Baltimore, MD.

Haith, D.A. & Shoemaker, L.L. (1987) Generalized watershed loading functions for stream flow nutrients. Water Resources Bulletin, 23(3), 471-78.

301

Hansen, J.S. & Ongerth, J.E. (1991) Effects of time and watershed characteristics on the concentration of Cryptosporidium oocysts in river water. Applied and Environmental

Microbiology, 57(10), 2790-95.

Hartel, P.G., Summer, J.D., Hill, J.L., Collins, J.V., Entry, J.A. & Segars, W.I. (2002)

Geographic variability of Escherichia coli ribotypes from animals in Idaho and Georgia.

Journal of Environmental Quality, 31, 1273-78.

Hawkins, P.R., Swanson, P., Warnecke, M., Shanker, S.R. & Nicholson, C. (2000)

Understanding the fate of Cryptosporidium and Giardia in storage reservoirs: a legacy of Sydney's water contamination incident. Journal of Water Supply, Research and

Technology AQUA, 496, 289-306.

Haydon, S. (2005). Trial of a coupled pathogen-stormflow model. In 29th Hydrology and Water Resources Symposium. Engineers Australia, Canberra, Australia.

Heitman, T.L., Frederick, L.M., Viste, J.R., Guselle, N.J., Morgan, U.M., Thompson,

R.C.A. & Olson, M.E. (2002) Prevalence of Giardia and Cryptosporidium and characterization of Cryptosporidium spp. isolated from wildlife, human, and agricultural sources in the North Saskatchewan River Basin in Alberta, Canada. Canadian Journal of Microbiology, 48(6), 530-41.

Hellard, M.E., Sinclair, M.I., Hogg, G.G. & Fairley, C.K. (2000) Prevalence of enteric pathogens among community based asymptomatic individuals. Journal of

Gastroenterology & Hepatology, 15(3), 290-93.

Helsel, D.R. & Hirsch, R.M. (1992) Statistical methods in water resources Elsevier,

Amsterdam.

Hibler, C.P. & Hancock, C.M. (1990). Waterborne giardiasis. In Drinking water microbiology:progress and recent developments (ed G.A. McFeters), pp. 271-93.

Springer-Verlag, New York.

302

Hipsey, M., Brookes, J.D., Antenucci, J.P., Burch, M.D., Regel, R.H., Davies, C.,

Ashbolt, N.J. & Ferguson, C. (2005). Hydrodynamic distribution of pathogens in lakes and reservoirs. In, p 206. American Water Works Association Research Foundation,

Denver, Colorado.

Hoar, B.R., Atwill, E.R., Elmi, C., Utterback, W.W. & Edmondson, A.J. (1999)

Comparison of fecal samples collected per rectum and off the ground for estimation of environmental contamination attributable to beef cattle. American Journal of Veterinary

Research, 60(11), 1352-56.

Horman, A., Korpela, H., Sutinen, J., Wedel, H. & Hanninen, M.-L. (2004) Meta- analysis in assessment of the prevalence and annual incidence of Giardia spp. and

Cryptosporidium spp. infections in humans in the Nordic countries. International

Journal for Parasitology, 34, 1337-46.

Hrudey, S.E. & Hrudey, E.J. (2004) Safe drinking water: lessons from recent outbreaks in affluent nations IWA Publishing, London.

Hunt, C.L., Ionas, G. & Brown, T.J. (2000) Prevalence and strain differentiation of

Giardia intestinalis in calves in the Manawatu and Waikato regions of North Island,

New Zealand. Veterinary Parasitology, 91(1-2), 7-13.

Hunter, P.R., Waite, M. & Ronchi, E. (2002) Drinking Water and Infectious Disease:

Establishing the Links IWA Publishing, London.

Hunter, P.R., Waite, M. & Ronchi, E. (2003) Drinking water and infectious disease: establishing the links CRC Press and IWA Publishing, London.

Hutchison, M.L., Walters, L.D., Avery, S.M., Synge, B.A. & Moore, A. (2004) Levels of zoonotic agents in British livestock manures. Letters in Applied Microbiology, 39,

207-14.

303

ISO. (1996). Water quality - Detection and enumeration of bacteriophages, Part 2:

Enumeration of somatic coliphages. In. International Organisation for Standardization,

Geneva.

Jakeman, A.J., Littlewood, I.G. & Whitehead, P.G. (1990) Computation of the instantaneous unit hydrograph and identifiable component flows with application to two small upland catchments. Journal of Hydrology, 177, 275-300.

Jakubowski, W. (1984). Detection of Giardia cysts in drinking water: state of the art. In

Giardia and Giardiasis, biology, pathogenesis and epidemiology, pp. 263-85. Plenum

Press, New York.

Jelliffe, P.A. (1997). Predicting stormwater quality from unsewered development. In

Clear Water, A technical response. Stormwater Industry Association Conference, Coffs

Harbour, Australia.

Jellison, K.L., Hemond, H.F. & Schauer, D.B. (2002) Sources and species of

Cryptosporidium oocysts in the Wachusett Reservoir watershed. Applied and

Environmental Microbiology, 68(2), 569-75.

Jenkins, A., Kirkby, M.J., McDonald, A., Naden, P. & Kay, D. (1984) A process based model of faecal bacterial levels in upland catchments. Water Science & Technology, 16,

453-62.

Jewett, D.G., Hilbert, T.A., Logan, B.E., Arnold, R.G. & Bales, R.C. (1995) Bacterial transport in laboratory columns and filters - Influence of ionic strength and pH on collision efficiency. Water Research, 29(7), 1673-80.

Jin, Y., Pratt, E. & Yates, M.V. (2000) Effect of mineral colloids on virus transport through saturated sand columns. Journal of Environmental Quality, 29(2), 532-39.

304

Joachim, A., Krull, T., Schwarzkopf, J. & Daugschies, A. (2003) Prevalence and control of bovine cryptosporidiosis in German dairy herds. Veterinary Parasitology, 112(4),

277-88.

Joergensen, R.G., Kuntzel, H., Scheu, S. & Seitz, D. (1998) Movement of faecal indicator organisms in earthworm channels under a loamy arable and grassland soil.

Applied Soil Ecology, 8(1-3), 1-10.

Johnson, S.R., Gary, H.L. & Ponce, S.L. (1978). Range cattle impacts on stream water quality in the Colorado front range. In. United States Department of Agriculture, Fort

Collins, CO.

Jonas, D., Spitzmuller, B., Weist, K., Ruden, H. & Daschner, F.D. (2003) Comparison of PCR-based methods for typing Escherichia coli. Clinical Microbiology and

Infection, 9, 823-31.

Jones, F. & White, W.R. (1984) Health and amenity aspects of surface waters. Water

Pollution Control, 83, 215-25.

Kaucner, C., Davies, C.M., Ferguson, C.M. & Ashbolt, N.J. (2004). Evidence for the existence of Cryptosporidium oocysts as single entities in surface runoff. In 4th

International Water Association World Congress and Exhibitions. International Water

Association, Marrakech, Mococco.

Kemp, J.S., Wright, S.E. & Bukhari, Z. (1995). On farm detection of Cryptosporidium parvum in cattle, calves and environmental samples. In Protozoan parasites and water.

Royal Society of Chemistry, Cambridge, UK.

Khandka, D.K., Tuna, M., Tal, M., Nejidat, A. & Golan-Goldhirsh, A. (1997)

Variability in the pattern of random amplified polymorphic DNA. Electrophoresis,

18(15), 2852-56.

305

Khatiwada, N.R. & Polprasert, C. (1999) Kinetics of fecal coliform removal in constructed wetland. Water Science and Technology, 40(3), 109-16.

Kistemann, T., Claßen, T., Koch, C., Dangendorf, F., Fischeder, R., Gebel, J., Vacata,

V. & Exner, M. (2002) Microbial load of drinking water reservoir tributaries during extreme rainfall and runoff. Applied and Environmental Microbiology, 68(5), 2188-97.

Kresovich, S.J., Williams, G.K., McFerson, J.R., Routman, E.J. & Schaal, B.A. (1992)

Characterisation of genetic identities and relationships of Brassica oleracea L. via a random amplified polymorphic DNA assay. Theoretical and Applied Genetics, 85, 190-

96.

Krogh, M. & Paterson, P. (2002). Sewage treatment plant effluent characterisation study. In, p 65. Sydney Catchment Authority, Sydney.

Lane, D.J. (1991). 16S/23S rRNA sequencing. In Nucleic acid techniques in Bacterial

Systematics (eds E. Stackebrandt & M. Goodfellow). John Wiley and Sons, London.

Langham, N.P. & Charleston, W.A. (1990) An investigation of the potential for spread of Sarcocystis spp. and other parasites by feral cats. New Zealand Journal of

Agriculture Research, 33, 429-35.

Larsen, R.E., Buckhouse, J.C., Moore, J.A. & Miner, J.R. (1988) Rangeland cattle and manure placement: a link to water quality. Oregon Academy of Science, 24, 7-15.

Larsen, R.E., Miner, J.C., Buckhouse, J.C. & Moore, J.A. (1994) Water-quality benefits of having cattle manure deposited away from streams. Bioresource and Technology, 48,

113-18.

Lau, M.M. & Ingham, S.C. (2001) Survival of faecal indicator bacteria in bovine manure incorporated into soil. Letters in Applied Microbiology, 33, 131-36.

Lefay, D., Naciri, M., Poirier, P. & Chermette, R. (2000) Prevalence of

Cryptosporidium infection in calves in France. Veterinary Parasitology, 89(1-2), 1-9.

306

LeJeune, J.T., Besser, T.E., Rice, D.H., Berg, J.L., Stilborn, R.P. & Hancock, D.D.

(2004) Longitudinal Study of Fecal Shedding of Escherichia coli O157:H7 in Feedlot

Cattle: Predominance and Persistence of Specific Clonal Types despite Massive Cattle

Population Turnover. Applied and Environmental Microbiology, 70(1), 377-84.

Lewis, A.L. & Stark, L.M. (1993) Virus study summary. Florida on-site septic disposal system research project Florida State Health Department, Florida.

Ley, D.H., Levy, M.G., Hunter, L., Corbett, W. & Barnes, H.J. (1988) Cryptosporidia- positive rates of avian necropsy accessions determined by examination of auramine O- stained fecal smears. Avian Diseases, 32(1), 108-13.

Lorenzo-Lorenzo, M.J., Ares-Mazas, E. & Villacorta Martinez de Maturana, I. (1993)

Detection of oocysts and IgG antibodies to Cryptosporidium parvum in asymptomatic adult cattle. Veterinary Parasitology, 47(1), 9-15.

Louws, F.J., Fulbright, D.W., Taylor Stephens, C. & deBruijn, F.J. (1994) Specific genomic fingerprints of phytopathogenic Xanthomonas and Pseudomonas pathovars and strains generated with repetitive DNA sequences and PCR. Applied and

Environmental Microbiology, 60(7), 2286-95.

Majewska, A.C., Werner, A., Sulima, P. & Luty, T. (2000) Prevalence of

Cryptosporidium in sheep and goats bred on five farms in west-central region of Poland.

Veterinary Parasitology, 89(4), 269-75.

Maldonado-Camargo, S., Atwill, E.R., Saltijeral-Oaxaca, J.A. & Herrera-Alonso, L.C.

(1998) Prevalence of and risk factors for shedding of Cryptosporidium parvum in

Holstein Freisian dairy calves in central Mexico. Preventive Veterinary Medicine, 36(2),

95-107.

307

Maluquer de Motes, C., Clemente-Casares, P., Hundesa, A., Martin, M. & Girones, R.

(2004) Detection of Bovine and Porcine Adenoviruses for Tracing the Source of Fecal

Contamination. Applied and Environmental Microbiology, 70(3), 1448-54.

Mawdsley, J.L., Brooks, A.E. & Merry, A.J. (1996) Movement of the protozoan pathogen Cryptosporidium parvum through three contrasting soil types. Biology and

Fertilisation of Soils, 21(1-2), 30-36.

Mawdsley, J.L., Brooks, A.E., Merry, R.J. & Pain, B.F. (1996) Use of a novel soil tilting table apparatus to demonstrate the horizontal and vertical movement of the protozoan pathogen Cryptosporidium parvum in soil. Biology and Fertilisation of Soils,

23(2), 215-20.

McGlade, T.R., Robertson, I.D., Elliot, A.D., Read, C. & Thompson, R.C.A. (2003)

Gastrointestinal parasites of domestic cats in Perth, Western Australia. Veterinary

Parasitology, 117, 251-62.

McLellan, P. (1998). Sydney Water Inquiry. Introduction recommendations and actions.

Volume 1. In. Premier's Department. New South Wales State Government., Sydney.

McLellan, S.L., Daniels, A.D. & Salmore, A.K. (2003) Genetic characterization of

Escherichia coli populations from host sources of fecal pollution by using DNA fingerprinting. Applied and Environmental Microbiology, 69(5), 2587-94.

McReynolds, C.A., Lappin, M.R., Ungar, B., McReynolds, L.M., Bruns, C., Spilker,

M.M., Thrall, M.A. & Reif, J.S. (1999) Regional seroprevalence of Cryptosporidium parvum-specific IgG of cats in the United States. Veterinary Parasitology, 80(3), 187-

95.

Medema, G.J. (1999) Cryptosporidium and Giardia: new challenges to the water industry. PhD, University of Utrecht, Utrecht.

308

Medema, G.J., Bahar, M. & Schets, F.M. (1997a) Survival of Cryptosporidium parvum,

Escherichia coli, faecal enterococci and Clostridium perfringens in river water - influence of temperature and autochthonous microorganisms. Water Science and

Technology, 35(11-12), 249-52.

Medema, G.J., Bahar, M. & Schets, F.M. (1997b) Survival of Cryptosporidium parvum,

Escherichia coli, faecal enterococci and Clostridium perfringens in river water: influence of temperature and autochthonous microorganisms. Water Science and

Technology, 35(11-12), 249–52.

Medema, G.J., Ketelaars, H.A.M., Hoogenboezem, W., Rijs, G.B.J. & Schijven, J.F.

(2001). Cryptosporidium and Giardia: ocurrence in sewage, manure and surface water.

In, p 171. Association of River Waterworks - RIWA.

Medema, G.J. & Schijven, J.F. (2001) Modelling the sewage discharge and dispersion of Cryptosporidium and Giardia in surface water. Water Research, 35(18), 4307-16.

Miller, K., Ferguson, C.M., Gillings, M.R., Mitchell, H., Pappayut, S., Angles, M., Cox,

P., Brusentiev, S. & Neilan, B. (submitted) Comparison of tracing and tracking tools for identifying bacterial contamination in drinking water catchments. Environmental

Science and Technology.

Moore, J.A., Smyth, J.D., Baker, E.S., Miner, J.R. & Moffitt, D.C. (1989) Modeling bacteria movement in livestock manure systems. Transactions of the ASAE, 32, 1049-

53.

Morgan, U.M., Buddle, J.R., Armson, A., Elliot, A. & Thompson, R.C.A. (1999a)

Molecular and biological characterisation of Cryptosporidium in pigs. Australian

Veterinary Journal, 77(1), 44-47.

309

Morgan, U.M., Xiao, L., Fayer, R., Lal, A.A. & Thompson, R.C.A. (1999b) Variation in

Cryptosporidium: Towards a taxonomic revision of the genus. International Journal for

Parasitology, 29(11), 1733-51.

Myoda, S.P., Carson, C.A., Fuhrmann, J.J., Hahm, B.K., Hartel, P.G., Yampara-Iquise,

L., Johnson, R.L., Kuntz, C.H., Nakatsu, S.M.J. & Sampadour, M. (2004) Comparison of genotypic-based microbial source tracking methods requiring a host origin database.

Journal of Water and Health, 1(4), 167-80.

Natural Resource Agriculture and Engineering Service. (2000). Managing nutrients and pathogens from animal agriculture. In, p 508. Natural Resource Agriculture and

Engineering Service, Ithaca, New York.

Neilan, B. (1995) Identification and phylogenetic analysis of toxigenic cyanobacteria by multiplex randomly amplified polymorphic DNA PCR. Applied and Environmental

Microbiology, 61, 2286-91.

Nicosia, L.A., Rose, J.B. & Stark, L. (2001) A field study of virus removal in septic tank drainfields. Journal of Environmental Quality, 30(6), 1933-39.

O'Keefe, B., D'Arcy, B.J., Davidson, J., Barbarito, B. & Clelland, B. (2003). Urban diffuse sources of faecal indicators. In 7th International Water Association Conference

Diffuse Pollution and Basin Management, pp. 6.1-6.6. International Water Association,

Dublin, Ireland.

O'Neill, R.A., Roads, G.K. & Wiese, R.N. (1993). On-site waste water treatment and disposal in New South Wales. In School of Civil and Environmental Engineering.

University of Technology, Sydney, Sydney.

Ogden, I.D., MacRae, M. & Strachan, N.J.C. (2004) Concentration and prevalence of

Escherichia coli O157 in sheep faeces at pasture in Scotland. Journal of Applied

Microbiology, 10.

310

Olley, J. & Deere, D.A. (2003). Targeting rectification action in the Wingecarribee catchment. In, p 87. CSIRO, Sydney.

Olson, M.E., Goh, J., Phillips, M., Guselle, N. & McAllister, T.A. (1999) Giardia cyst and Cryptosporidium oocyst survival in water, soil, and cattle feces. Journal of

Environmental Quality, 28(6), 1991-96.

Olson, M.E., Guselle, N.J., Ohandley, R.M., Swift, M.L., McAllister, T.A., Jelinski,

M.D. & Morck, D.W. (1997a) Giardia and Cryptosporidium in dairy calves in British

Columbia. Canadian Veterinary Journal - Revue Veterinaire Canadienne, 38(11), 703-

06.

Olson, M.E., Thorlakson, C.L., Deselliers, L., Morck, D.W. & McAllister, T.A. (1997b)

Giardia and Cryptosporidium in Canadian farm animals. Veterinary Parasitology,

68(4), 375-81.

Ong, C., Moorehead, W., Ross, A. & Isaac-Renton, J. (1996) Studies of Giardia spp. and Cryptosporidium spp. in two adjacent watersheds. Applied and Environmental

Microbiology, 62(8), 2798-805.

Ongerth, J.E. & Stibbs, H.H. (1987) Identification of Cryptosporidium oocysts in river water. Applied and Environmental Microbiology, 53(4), 672-76.

Ongerth, J.E. & Stibbs, H.H. (1989) Prevalence of Cryptosporidium infection in dairy calves in western Washington. American Journal of Veterinary Research, 50(7), 1069-

70.

Ortega-Mora, L.N., Requejo-Fernandez, J.A., Pilar-Izquierdo, M. & Pereira-Bueno, J.

(1999) Role of adult sheep in transmission of infection by Cryptosporidium parvum to lambs: confirmation of periparturient rise. International Journal for Parasitology,

29(8), 1261-68.

311

Parveen, S., Portier, K.M., Robinson, K., Edmiston, L. & Tamplin, M.L. (1999)

Discriminant analysis of ribotype profiles of Escherichia coli for differentiating human and nonhuman sources of fecal pollution. Applied and Environmental Microbiology,

65(7), 3142-47.

Paterson, P. & Krogh, M. (2003). Review of sewage treatment plants within the Sydney

Catchment Authority area of operations. In, p 75. Sydney Catchment Authority,

Sydney.

Payment, P. & Franco, E. (1993) Clostridium perfringens and somatic coliphages as indicators of the efficiency of drinking water treatment for viruses and protozoan cysts.

Applied and Environmental Microbiology, 59, 2418-24.

Pina, S., Buti, M., Cotrina, M., Piella, J. & Girones, R. (2000) HEV identified in serum from humans with acute hepatitis and in sewage of animal origin in Spain. Journal of

Hepatology, 33(5), 826-33.

Pitt, R. & Voorhees, J. (2003). SLAMM, the Source Loading and Management Model.

In Wet-Weather Flow in the Urban Watershed: Technology and Management (eds R.

Field & D. Sullivan), pp. 79-101. Lewis Publishers, Boca Raton.

Post, D.A. & Jakeman, A.J. (1996) Relationships between catchment attributes and hydrological responses characteristics in small Australian Mountain Ash catchments.

Hydrological Processes, 10, 877-92.

Power, M.I., Shanker, S.R., Sangster, N.C. & Veal, D.A. (2001). Prevalence of

Cryptosporidium paruvum in Eastern Grey kangaroos Macropus giganteus located in the Sydney catchment. In Proceedings of Cryptosporidium from Molecules to Disease,

7-12 October, 2001, Fremantle, Western Australia (ed A. Thompson), p 32. Murdoch

University, Perth.

312

Power, M.L., Slade, M.B., Shanker, S.R., Sangster, N.C. & Veal, D.A. (2004).

Cryptosporidium in Eastern Grey Kangaroos Macropus giganteus. In Cryptosporidium:

From molecules to disease (ed R.C.A. Thompson), pp. 207-09. Elsevier, Amsterdam.

Quilez, J., Sanchez-Acedo, C., Clavel, A., del Cacho, E. & Lopez-Bernad, F. (1996a)

Prevalence of Cryptosporidium infections in pigs in Aragon (northeastern Spain).

Veterinary Parasitology, 67(1), 83-88.

Quilez, J., Sanchez-Acedo, C., del Cacho, E. & Causape, A.C. (1996b) Prevalence of

Cryptosporidium and Giardia infections in cattle in Aragon (Northeastern Spain).

Veterinary Parasitology, 66(3-4), 139-46.

Ralston, B.J., Allister, T.A. & Olson, M.E. (2003) Prevalence and infection pattern of naturally acquired giardiasis and cryptosporidiosis in range beef calves. Veterinary

Parasitology, 114(113-122).

Ramsay, B. (1994) Commercial use of wild animals in Australia Australian Government

Printing Service.

Renter, D.G., Sargeant, J.M., Oberst, R.D. & Samadpour, M. (2003) Diversity, frequency, and persistence of Escherichia coli O157 strains from range cattle environments. Applied and Environmental Microbiology, 69(1), 542-47.

Rickard, L.G., Siefker, C., Boyle, C.R. & Gentz, E.J. (1999) The prevalence of

Cryptosporidium and Giardia spp. in fecal samples from free-ranging white-tailed deer

(Odocoileus virginianus) in the southeastern United States. Journal of Veterinary

Diagnostic Investigation, 11(1), 65-72.

Robertson, L.J., Paton, C.A., Campbell, A.T., Smith, P.G., Jackson, M.H., Gilmour,

R.A., Black, S.E., Stevenson, D.A. & Smith, H.V. (2000) Giardia cysts and

Cryptosporidium oocysts at sewage treatment works in Scotland, UK. Water Research,

34(8), 2310-22.

313

Robertson, L.J., Smith, H.V. & Paton, C.A. (1995). Occurrence of Giardia cysts and

Cryptosporidium oocysts in sewage influent in six sewage treatment plants in Scotland and the prevalence of cryptosporidiosis and giardiasis diagnosed in the communities served by those plants. In Protozoan parasites and water (eds W.B. Betts, D. Casemore,

C. Fricker, H. Smith & J. Watkins), pp. 47-49. Royal Society of Chemistry, Cambridge,

UK.

Robinson, S.E., Wright, E.J., Williams, N.J., Hart, C.A. & French, N.P. (2004)

Development and application of a spiral plating method for the enumeration of

Escherichia coli O157 in bovine faeces. J Appl Microbiol, 97(3), 581-89.

Rose, J., Huffman, D.E., Riley, K., Farrah, S.R., Lukasik, J.O. & Hamann, C.L. (2001)

Reduction of enteric microorganisms at the upper Occoquan Sewage Authority water reclamation plant. Water Environment Research, 73(6), 711-20.

Rose, J.B. & Carnahan, R.P. (1992). Pathogen removal by full scale wastewater treatment. In. Department of Environmental Regulation, State of Florida.

Roser, D., Ashbolt, N., Charles, K., Deere, D., Steffensen, D. & Ferguson, C. (2003).

Transforming pathogen water quality data and collection experiences into source water monitoring and control information products. Oral presentation for OzWater March,

2003 Paper oz094. In Australian Water Association Annual Conference, Perth,

Australia.

Rothwell, V., Angles, M.L., Ferguson, C.M., Deere, D. & Logan, M.R. (2004).

Monitoring of enteric viruses in the Sydney Catchment. In Australian Society for

Microbiology Annual Conference, p poster. Australian Society for Microbiology,

Sydney.

314

Rouquet, V., Homer, F., Brignon, J.M., Bonne, P. & Cavard, J. (2000) Source and occurrence of Giardia and Cryptosporidium in Paris rivers. Water Science and

Technology, 41(7), 79-86.

Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular cloning: a laboratory manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

Santin, M., Trout, J.M., Xiao, L., Zhou, L., Greiner, E. & Fayer, R. (2004) Prevalence and age-related variation of Cryptosporidium species and genotypes in dairy calves.

International Journal for Parasitology, 122, 103-17.

Sargent, K.D., Morgan, U.M., Elliot, A. & Thompson, R.C.A. (1998) Morphological and genetic characterisation of Cryptosporidium oocysts from domestic cats. Veterinary

Parasitology, 77(4), 221-27.

Schijven, J.F., de Bruin, H.A.M., Hassanizadeh, S.M. & de Roda Husman, A.M. (2003)

Bacteriophages and clostridium spores as indicator organisms for removal of pathogens by passage through saturated dune sand. Water Research, 37(9), 2186-94.

Schijven, J.F. & Hassanizadeh, S.M. (2002) Virus removal by soil passage at field scale and ground-water protection of sandy aquifers. Water Science and Technology, 46(3),

123–29.

Scott, C.A., Smith, H.V., Mtambo, M.A. & Gibbs, H.A. (1995) An epidemiological study of Cryptosporidium parvum in two herds of adult beef cattle. Veterinary

Parasitology, 57(4), 277-88.

Scott, T.M., Parveen, S., Portier, K.M., Rose, J.B., Tamplin, M.L., Farrah, S.R., Koo, A.

& Lukasik, J. (2003) Geographical variation in ribotype profiles of Escherichia coli isolates from humans, swine, poultry, beef and dairy cattle in Florida. Applied and

Environmental Microbiology, 69, 1089-92.

315

Scott, T.M., Rose, J.B., Jenkins, T.M., Farrah, S.R. & Lukasik, J. (2002) Microbial source tracking: current methodology and future directions. Applied and Environmental

Microbiology, 68(12), 5796-803.

Seurinck, S., Verstraete, W. & Siciliano, S.D. (2003) Use of 16S-23S rRNA intergenic spacer region PCR and repetitive extragenic palindromic PCR analyses of Escherichia coli isolates to identify nonpoint fecal sources. Applied and Environmental

Microbiology, 69(8), 4942-50.

Shere, J.A., Kaspar, C.W., Bartlett, K.J., Linden, S.E., Norell, B., Francey, S. &

Schaefer, D.M. (2002) Shedding of Escherichia coli O157:H7 in dairy cattle housed in a confined environment following waterborne inoculation. 68, 4(1947-1954).

Sherer, B.M., Miner, J.R., Moore, J.A. & Buckhouse, J.C. (1992) Indicator bacterial survival in stream sediments. Journal of Environmental Quality, 21, 591-95.

Simmons, O.D., Sobsey, M.D., Heaney, C.D., Schaefer, F.W. & Francy, D.S. (2001)

Concentration and detection of Cryptosporidium oocysts in surface water samples by method 1622 using ultrafiltration and capsule filtration. Applied and Environmental

Microbiology, 67(3), 1123-27.

Simpson, J.M., Santo Domingo, J.W. & Reasoner, D.J. (2002) Microbial source tracking: state of the science. Environmental Science & Technology, 36(24), 5279-88.

Sivapalan, M., Takeuch, K., Franks, S.W., Gupta, V.K., Karambiri, H., Lakshmi, V.,

Liang, X., McDonnell, J.J., Mendiondo, E.M., O'Connell, P.E., Oki, T., Pomeroy,

J.W.P., Schertzer, D., Uhlenbrook, S. & Zehe, E. (2003) IAHS Decade on predictions in ungauged basins (PUB), 2003-2012:Shaping an exciting future for the hydrological sciences. Hydrological Sciences Journal, 48(6), 857-80.

316

Snyder, D.E. (1988) Indirect immunofluorescent detection of oocysts of

Cryptosporidium parvum in the feces of naturally infected raccoons (Procyon lotor).

Journal of Parasitology, 74(6), 1050-52.

Sobieh, M., Tacal, J., Wilcke, B.W., Jr., Lawrence, W. & El-Ahraf, A. (1987)

Investigation of cryptosporidial infection in calves in San Bernardino County,

California. Journal of the American Veterinary Medical Association, 191(7), 816-18.

Sorensen, D.L., Eberl, S.G. & Diksa, R.A. (1989) Clostridium perfringens as a point source indicator in non-point polluted streams. Water Research, 23, 191-97.

Standards Australia. (2000). Water Microbiology- spores of sulfite reducing anaerobes

(clostridia) including Clostridium perfringens - membrane filtration method. In.

Standards Australia, Australia.

Steets, B.M. & Holden, P.A. (2003) A mechanistic model of runoff-associated fecal coliform fate and transport through a coastal lagoon. Water Research, 37(3), 589-608.

Stoddard, C.S., Coyne, M.S. & Grove, J.H. (1998) Fecal bacteria survival and infiltration through a shallow agricultural soil: Timing and tillage effects. Journal of

Environmental Quality, 27(6), 1516-23.

Sturdee, A.P., Bodley-Tickell, A.T., Archer, A. & Chalmers, R.M. (2003) Long-term study of Cryptosporidium prevalence on a lowland farm in the United Kingdom.

Veterinary Parasitology, 116(2), 97-113.

Sturdee, A.P., Chalmers, R.M. & Bull, S.A. (1999) Detection of Cryptosporidium oocysts in wild mammals of mainland Britain. Veterinary Parasitology, 80(4), 273-80.

Tate, K.W., Atwill, E.R., George, M.R., McDougald, M.K. & Larsen, R.E. (2000)

Cryptosporidium parvum transport from cattle fecal deposits on California rangelands.

Journal of Range Management, 53(3), 295-99.

317

Thelin, R. & Gifford, G.F. (1983) Fecal coliform release patterns from fecal material of cattle. Journal of Environmental Quality, 12, 57-63.

Tian, Y.Q., Gong, P., Radke, J.D. & Scarborough, J. (2002) Spatial and temporal modeling of microbial contaminants on grazing farmlands. Journal of Environmental

Quality, 31, 860-69.

Tiedemann, D.A., Higgins, D.A., Quigley, T.M., Sanderson, H.R. & Marx, D.B. (1987)

Responses of fecal coliform in streamwater to four grazing strategies. Journal of Range

Management, 40(4), 322-29.

Trask, J.R., Kalita, P.K., Kuhlenschmidt, M.S., Smith, R.D. & Funk, T.L. (2001).

Overland and near-surface transport of Cryptosporidium parvum. In 2001 ASAE Annual

International Meeting, pp. Paper No. 01-2104, Sacramento, California, USA.

Trask, J.R., Kalita, P.K., Kuhlenschmidt, M.S., Smith, R.D. & Funk, T.L. (2004)

Overland and Near-Surface Transport of Cryptosporidium parvum from Vegetated and

Nonvegetated Surfaces. Journal of Environmental Quality, 33(3), 984-93.

Trevisan, D., Vansteelant, J.Y. & Dorioz, J.M. (2002) Survival and leaching of fecal bacteria after slurry spreading on mountain hay meadows: consequences for the management of water contamination risk. Water Research, 36, 275-83.

Ungar, B.L.P. (1990). Cryptosporidiosis in humans (Homo sapiens). In

Cryptosporidiosis of man and animals. CRC Press Inc, Boca Raton, Florida, USA.

Upton, S.J. (1997). In vitro cultivation. In Cryptosporidium and Cryptosporidiosis (ed

R. Fayer), pp. 181-207. CRC Press, Washington, DC.

USDA. (1994). Cryptosporidium and Giardia in beef calves. In. National animal health monitoring system, United States Department of Agriculture, Fort Collins, CO.

318

USEPA. (1999). Method 1623 - Cryptosporidium and Giardia in Water by

Filtration/IMS/IFA. In, p http://www.epa.gov/nerlcwww/1623.pdf. Office of Water,

United States Environment Protection Agency, Washington D.C.

USEPA. (2001). Protocol for Developing Pathogen TMDLs. In, p 132. Office of Water

(4503F), United States Environmental Protection Agency, Washington, DC 20460.

USEPA. (2002). EPA Microbiological Risk Assessment Framework Workshop Tools,

Methods, and Approaches, August 27 29, 2002. In (ed I. ICF Consulting). U.S.

Environmental Protection Agency, Washington, DC. van Eerdt, M.M. (1998) Mestproductie en mineralenuitscheiding 1997.

Kwartaalberichten Milieu (CBS), 98(4), 41-46.

Versalovic, J., Koeuth, T. & Lupski, J.R. (1991) Distribution of repetitive DNA sequences in eubacteria and application to fingerprinting bacterial genomes. Nucleic

Acids Research, 19(24), 6823-31.

Vila, J., Marcos, M.A. & Jimenez de Anta, M.T. (1996) A comparative study of different PCR-based DNA fingerprinting techniques for typing of the Acinetobacter calcoaceticus - A. baumannii complex. Journal of Medical Microbiology, 44(6), 482-

89.

Vinten, A.J.A., Lewis, D.R., McGechan, M., Duncan, A., Aitken, M., Hill, C. &

Crawford, C. (2004) Predicting the effect of livestock inputs of E. coli on microbiological compliance of bathing waters. Water Research, 38, 3215-24.

Virginia Department of Environmental Quality. (2003). Guidance Memorandum No.

03-2012. Model calibration and verification for bacteria TMDLs. In, Vol. 2005, pp.

Guidance Memorandum No. 03-2012. Virginia Department of Environmental Quality,

Richmond, Virginia.

319

Vogel, L., van Oorschot, E., Maas, H.M.E., Minderhoud, B. & Dijkshoorn, L. (2000)

Epidemiologic typing of Escherichia coli using RAPD analysis, ribotyping and serotyping. Clinical Microbiology and Infection, 6(2), 82-87.

Wade, S.E., Mohammed, H.O. & Schaaf, S.L. (2000) Prevalence of Giardia sp,

Cryptosporidium parvum and Cryptosporidium muris (C. Andersoni) in 109 dairy herds in five counties of southeastern New York. Veterinary Parasitology, 93(1), 1-11.

Walker, F.R., Jr. & Stedinger, J.R. (1999) Fate and transport model of Cryptosporidium.

Journal of Environmental Engineering, 125(4), 325-33.

Walker, S.E., Mostaghimi, S., Dillaha, T.A. & Woeste, F.E. (1990) Modeling animal waste management practices: impacts on bacteria levels in runoff from agricultural lands. Transactions of the ASAE, 33(3), 807-17.

Warnecke, M., Weir, C. & Vesey, G. (2003) Evaluation of an internal positive control for Cryptosporidium and Giardia testing in water samples. Letters in Applied

Microbiology, 37(3), 244-48.

Weiskel, P.K., Howes, B.L. & Heufelder, G.R. (1996) Coliform contamination of a coastal embayment:sources and transport pathways. Environmental Science and

Technology, 30(6), 1872-81.

Whittam, T.S. (1989) Clonal dynamics of Escherichia coli in its natural habitat. Antonie

Van Leeuwenhoek, 55(1), 23-32.

Woods, C.R., Versalovic, J., Koeuth, T. & Lupski, J.R. (1993) Whole cell repetitive element sequence-based polymerase chain reaction allows rapid assessment of clonal relationships of bacterial isolates. Journal of Clinical Microbiology, 31, 1927-31.

Woolhouse, M.E.J. (2002) Population biology of emerging and re-emerging pathogens.

Trends in Microbiology, 10(10), s3-s7.

320

Xiao, L., Singh, A., Limor, J., Graczyk, T.K., Gradus, S. & Lal, A. (2001) Molecular characterization of Cryptosporidium oocysts in samples of raw surface water and wastewater. Applied and Environmental Microbiology, 67(3), 1097-101.

Xiao, L.H., Herd, R.P. & Bowman, G.L. (1994) Prevalence of Cryptosporidium and

Giardia infections on two Ohio pig farms with different management systems.

Veterinary Parasitology, 52(3), 331-36.

Xiao, L.H., Herd, R.P. & Rings, D.M. (1993) Diagnosis of Cryptosporidium on a sheep farm with neonatal diarrhea by immunofluorescence assays. Veterinary Parasitology,

47(1), 17-23.

Yates, M.V. & Gerba, C.P. (1998). Microbial considerations in wastewater reclamation and reuse. In Wastewater Reclamation and Reuse (ed T. Asano), Vol. 10 Water Quality

Management Library, pp. 437-88. Technomic Publishing Co. Inc., Lancaster,

Pennsylvania.

Zhao, T., Doyle, M.P., Shere, J. & Garber, L. (1995) Prevalence of enterohemorrhagic

Escherichia coli O157:H7 in a survey of dairy herds. Applied and Environmental

Microbiology, 61(4), 1290-93.

Zhou, L., Fayer, R., Trout, J.M., Ryan, U.M., Schaefer III, F.W. & Xiao, L. (2004a)

Genotypes of Cryptosporidium species infecting fur-bearing mammals differ from those species infecting humans. Applied & Environmental Microbiology, 70(12), 7574-77.

Zhou, L., Kassa, H., Tischler, M.L. & Xiao, L. (2004b) Host-adapted Cryptosporidium spp. in Canada Geese (Branta canadensis). Applied & Environmental Microbiology,

70(7), 4211-15.

321

APPENDIX 1 Subcatchment data file for all of the Sydney drinking water catchments

l s a Rl Gl nl Hl fl Land use category (λ) 1 2 3 4 5 6 7 8 9 10 11 12 13 101 1 58.59 20.957 302 0 0 0.99 0.0016 0.5434 0.0028 0.0150 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4372 102 4 99.21 28.344 302 0 0 0.95 0.0095 0.5622 0.0033 0.0188 0.0000 0.0000 0.0000 0.0000 0.0012 0.0000 0.0000 0.0000 0.4049 103 1 81.17 18.988 102 0 0 0.99 0.0002 0.4227 0.0037 0.0428 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5306 104 1 29.17 9.582 102 0 0 1.00 0.0000 0.1446 0.0004 0.0207 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8343 105 1 27.68 8.777 106 0 0 1.13 0.0001 0.1636 0.0169 0.0091 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8104 106 2 49.11 13.835 102 0 0 0.97 0.0009 0.6843 0.0036 0.0180 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.2933 201 8 51.07 16.862 1406 0 0 0.79 0.0148 0.0722 0.0000 0.0164 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8966 202 7 87.09 14.935 201 0 0 0.78 0.0110 0.2999 0.0000 0.0234 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6658 203 2 55.04 10.81 201 0 0 0.76 0.0040 0.1946 0.0000 0.0078 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7936 204 1 57.60 20.749 203 0 0 0.78 0.0006 0.2711 0.0000 0.0163 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7121 205 1 101.50 29.019 203 0 0 0.78 0.0019 0.5218 0.0001 0.0180 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4582 301 1 90.85 20.705 1901 0 0 0.99 0.0009 0.1471 0.0000 0.0119 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8401 302 5 60.85 5.714 1901 0 0 0.91 0.0084 0.5756 0.0020 0.0195 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3944 303 1 87.62 19.57 1901 0.98 303 0.91 0.0013 0.9147 0.0037 0.0457 0.0000 0.0000 0.0000 0.0000 0.0000 0.0018 0.0014 0.0000 0.0313 304 1 134.09 29.7 302 0 0 1.05 0.0031 0.6641 0.0039 0.0198 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3091 401 1 94.36 24.173 405 0.98 401 0.85 0.0078 0.4421 0.0071 0.0692 0.0000 0.0000 0.0000 0.0000 0.0012 0.0016 0.0002 0.0000 0.4708 402 2 103.26 27.88 405 0 0 0.80 0.0021 0.5419 0.0063 0.0463 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4034 403 1 96.66 29.273 402 0 0 0.84 0.0009 0.2877 0.0017 0.0107 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6989 404 12 207.71 25.565 405 0 0 0.89 0.0080 0.0419 0.0007 0.0049 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9444 405 13 219.79 35.536 807 0 0 0.97 0.0169 0.0565 0.0023 0.0064 0.0000 0.0000 0.0000 0.0000 0.0018 0.0000 0.0000 0.0034 0.9128 406 1 75.40 23.897 402 0 0 0.79 0.0008 0.2715 0.0023 0.0124 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7130 501 1 20.00 12.083 503 0 0 1.11 0.0009 0.0067 0.0004 0.0132 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9788 502 1 35.67 14.184 503 0 0 1.23 0.0001 0.0412 0.0023 0.0063 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9502 503 2 81.70 23.366 505 0 0 0.91 0.0039 0.2430 0.0009 0.0159 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7363 504 11 29.27 9.568 404 0 0 0.85 0.0162 0.0888 0.0056 0.0026 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8868 505 3 39.21 16.296 504 0 0 0.87 0.0023 0.0308 0.0000 0.0106 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9563 506 1 24.30 10.486 503 0 0 1.04 0.0000 0.0000 0.0000 0.0054 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9946 507 1 93.08 16.947 503 0 0 1.28 0.0002 0.0003 0.0000 0.0040 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9955 322

l s a Rl Gl nl Hl fl Land use category (λ) 1 2 3 4 5 6 7 8 9 10 11 12 13 508 1 15.51 8.792 504 0 0 0.93 0.0001 0.0000 0.0000 0.0010 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9989 601 1 7.08 5.71 0 0 1.43 0.0140 0.0001 0.0000 0.1107 0.0000 0.0000 0.0000 0.0000 0.0000 0.0394 0.0065 0.8260 0.0032 602 1 3.35 3.485 0 0 1.46 0.0843 0.0015 0.0000 0.1013 0.0000 0.0000 0.0000 0.0000 0.0000 0.0095 0.0000 0.7557 0.0477 603 1 10.60 5.282 0 0 1.42 0.0168 0.0022 0.0000 0.1791 0.0000 0.0000 0.0000 0.0000 0.0000 0.1349 0.0332 0.5925 0.0412 701 1 178.02 42.388 102 0 0 1.14 0.0004 0.2840 0.0021 0.0164 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6971 702 3 180.46 33.954 102 0 0 0.98 0.0035 0.4278 0.0012 0.0106 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5569 801 1 86.84 24.489 807 0 0 1.58 0.0030 0.2574 0.0958 0.0301 0.0000 0.0000 0.0000 0.0000 0.0000 0.0011 0.0005 0.0000 0.6122 802 1 98.90 23.233 807 0 0 2.01 0.0006 0.1098 0.0576 0.0207 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8114 803 1 272.21 31.58 807 0 0 1.14 0.0014 0.1465 0.0394 0.0244 0.0000 0.0000 0.0000 0.0000 0.0015 0.0092 0.0019 0.0000 0.7757 804 1 31.23 9.809 0 0 1.52 0.1702 0.3580 0.1974 0.0367 0.0000 0.0000 0.0000 0.0000 0.0000 0.0011 0.0013 0.0207 0.2147 805 1 130.25 24.218 807 0 0 1.90 0.0025 0.1654 0.0577 0.0207 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7537 806 1 35.36 12.358 807 0 0 1.27 0.0000 0.0066 0.0014 0.0073 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9846 807 14 143.72 32.186 0 0 1.30 0.0485 0.0985 0.0218 0.0334 0.0000 0.0000 0.0000 0.0000 0.0000 0.0002 0.0000 0.0312 0.7665 808 1 65.23 18.426 807 0 0 1.42 0.0010 0.0660 0.0280 0.0141 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0011 0.8898 901 1 98.96 21.44 904 0 0 1.14 0.0002 0.2359 0.0393 0.0204 0.0003 0.0003 0.0003 0.0003 0.0000 0.0000 0.0000 0.0000 0.7028 902 4 164.50 36.536 1201 0 0 0.99 0.0025 0.0000 0.0000 0.0027 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8982 0.0966 903 2 85.69 23.343 904 0 0 1.07 0.0000 0.2761 0.0086 0.0296 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6857 904 3 371.71 45.702 902 0 0 1.12 0.0008 0.0035 0.0002 0.0057 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3125 0.6774 905 1 48.56 12.487 903 0 0 1.06 0.0002 0.2865 0.0163 0.0472 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6498 1001 15 803.59 52.218 0 0 0.98 0.0855 0.0065 0.0005 0.0109 0.0000 0.0000 0.0000 0.0000 0.0003 0.0000 0.0000 0.8944 0.0018 1101 1 43.93 13.459 1102 0 0 0.98 0.0058 0.0360 0.0022 0.0240 0.0000 0.0000 0.0000 0.0000 0.0055 0.0000 0.0003 0.9101 0.0161 1102 2 102.86 16.314 1001 0 0 0.97 0.0006 0.0000 0.0000 0.0094 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7015 0.2885 1103 1 36.80 13.385 1102 0 0 1.07 0.0005 0.0201 0.0000 0.0307 0.0001 0.0001 0.0001 0.0001 0.0006 0.0000 0.0000 0.0879 0.8597 1201 9 36.27 9.891 1207 0 0 0.86 0.0339 0.0014 0.0001 0.0087 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9559 0.0000 1203 1 22.88 10.295 1204 0 0 1.44 0.0038 0.0078 0.0000 0.1103 0.0000 0.0000 0.0000 0.0000 0.0000 0.1377 0.0832 0.1783 0.4788 1204 3 71.01 17.702 1207 0 0 1.17 0.0035 0.0063 0.0000 0.0142 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9711 0.0047 1205 1 12.37 6.895 1206 0 0 1.48 0.0000 0.0006 0.0000 0.1700 0.0000 0.0000 0.0000 0.0000 0.0000 0.2034 0.0701 0.1243 0.4317 1206 2 21.02 10.058 1204 0 0 1.37 0.0000 0.0000 0.0000 0.0064 0.0000 0.0000 0.0000 0.0000 0.0000 0.0002 0.0085 0.8118 0.1731 1207 10 33.51 10.015 1001 0 0 0.86 0.1550 0.0045 0.0003 0.0056 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8346 0.0000 1208 1 6.53 5.589 1206 0 0 1.47 0.0001 0.0000 0.0000 0.1586 0.0000 0.0000 0.0000 0.0000 0.0000 0.1884 0.1987 0.0504 0.4037 1301 1 55.36 20.968 1314 0 0 1.20 0.0000 0.0096 0.0000 0.0310 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0018 0.9575 1302 1 68.26 19.503 1305 0 0 1.08 0.0012 0.2836 0.0298 0.0347 0.0000 0.0000 0.0000 0.0000 0.0102 0.0148 0.0093 0.0000 0.6163 1303 1 25.87 11.933 1305 0 0 1.20 0.0000 0.2913 0.0114 0.0294 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6679 1304 1 49.89 15.928 1315 0 0 1.12 0.0002 0.5040 0.0156 0.0279 0.0000 0.0000 0.0000 0.0000 0.0000 0.0015 0.0000 0.0000 0.4509

323

l s a Rl Gl nl Hl fl Land use category (λ) 1 2 3 4 5 6 7 8 9 10 11 12 13 1305 7 73.38 28.934 1314 0 0 1.05 0.0067 0.3604 0.0093 0.0163 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6072 1306 1 163.90 34.359 1314 0 0 1.20 0.0000 0.0047 0.0002 0.0138 0.0000 0.0000 0.0000 0.0000 0.0009 0.0000 0.0000 0.2539 0.7264 1307 1 114.73 26.07 1314 0 0 1.33 0.0001 0.0000 0.0000 0.0048 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.2603 0.7348 1308 1 32.92 15.351 1305 0 0 1.16 0.0001 0.2949 0.0134 0.0284 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6632 1309 1 38.49 16.177 1315 0 0 1.09 0.0003 0.5924 0.0748 0.0255 0.0000 0.0000 0.0000 0.0000 0.0000 0.0007 0.0000 0.0000 0.3064 1310 5 36.27 14.259 1315 0 0 1.03 0.0012 0.4654 0.0433 0.0246 0.0000 0.0000 0.0000 0.0000 0.0020 0.0000 0.0000 0.0000 0.4635 1311 1 45.26 12.614 1305 0 0 1.18 0.0003 0.1106 0.0197 0.0302 0.0000 0.0000 0.0000 0.0000 0.0000 0.0150 0.0039 0.0031 0.8173 1312 1 20.02 13.61 1305 0 0 1.19 0.0006 0.1199 0.0054 0.0360 0.0000 0.0000 0.0000 0.0000 0.0052 0.0088 0.0011 0.0049 0.8180 1313 1 101.56 21.257 1315 0.98 1313 1.12 0.0024 0.2411 0.0039 0.0416 0.0000 0.0000 0.0000 0.0000 0.0029 0.0045 0.0002 0.0000 0.7034 1314 8 151.78 29.189 1201 0 0 1.00 0.0037 0.0113 0.0014 0.0037 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8344 0.1456 1315 6 91.32 27.184 1305 0 0 0.93 0.0034 0.5698 0.0088 0.0351 0.0000 0.0000 0.0000 0.0000 0.0005 0.0004 0.0000 0.0000 0.3822 1401 1 11.15 10.047 1404 0 0 0.86 0.0010 0.3290 0.0068 0.0269 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6363 1402 2 133.13 37.308 1404 0 0 1.01 0.0015 0.0997 0.0000 0.0124 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8863 1403 1 63.08 20.91 1402 0 0 0.94 0.0010 0.1369 0.0000 0.0238 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8382 1404 10 159.54 37.792 504 0 0 0.80 0.0151 0.0673 0.0000 0.0084 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9093 1405 1 38.07 13.977 1404 0 0 0.84 0.0003 0.1344 0.0000 0.0132 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8520 1406 9 93.34 21.852 1404 0 0 0.78 0.0144 0.1302 0.0000 0.0171 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8382 1501 4 102.21 21.834 1406 0 0 0.86 0.0042 0.2280 0.0000 0.0216 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7461 1502 3 72.59 21.996 1501 0 0 1.02 0.0024 0.3200 0.0013 0.0132 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6631 1503 2 90.83 17.497 1502 0 0 1.00 0.0054 0.2088 0.0012 0.0118 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7728 1504 1 41.47 13.537 1503 0 0 0.92 0.0009 0.4957 0.0001 0.0137 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4896 1505 1 87.18 33.287 1503 0 0 1.09 0.0006 0.1233 0.0029 0.0197 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8535 1506 1 34.97 13.393 1503 0 0 1.00 0.0002 0.4738 0.0000 0.0158 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5102 1601 1 34.21 12.673 1614 0 0 0.79 0.0002 0.6697 0.0024 0.0093 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3183 1602 1 38.26 14.509 1606 0 0 0.77 0.0055 0.7384 0.0061 0.0318 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.2181 1603 3 21.98 6.259 1609 0 0 0.75 0.0033 0.8585 0.0167 0.0570 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0644 1604 1 43.45 14.853 1609 0 0 0.89 0.0028 0.4693 0.0010 0.0162 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5107 1605 1 72.23 14.193 1615 0 0 0.81 0.0018 0.7071 0.0000 0.0856 0.0000 0.0000 0.0000 0.0000 0.0004 0.0004 0.0000 0.0000 0.2047 1606 2 79.72 12.364 1608 0 0 0.74 0.0033 0.7774 0.0339 0.0228 0.0000 0.0000 0.0000 0.0000 0.0007 0.0005 0.0492 0.0000 0.1123 1607 1 113.85 11 0 0 0.75 0.1667 0.6874 0.0001 0.0235 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.1222 1608 6 18.72 5.071 2618 0 0 0.73 0.0005 0.0276 0.0000 0.3925 0.0003 0.0003 0.0003 0.0003 0.0033 0.2121 0.3583 0.0000 0.0044 1609 4 84.53 14.781 1614 0 0 0.76 0.0018 0.7150 0.0048 0.0231 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.2553 1610 1 35.05 12.757 1609 0 0 0.75 0.0007 0.5916 0.0005 0.0097 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3975 1611 1 39.39 18.502 1606 0 0 0.76 0.0015 0.8261 0.0294 0.0166 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.1264

324

l s a Rl Gl nl Hl fl Land use category (λ) 1 2 3 4 5 6 7 8 9 10 11 12 13 1612 1 41.33 18.885 1614 0 0 0.75 0.0016 0.7334 0.0421 0.0659 0.0000 0.0000 0.0000 0.0000 0.0004 0.0000 0.0082 0.0000 0.1483 1613 1 50.17 14.131 1614 0 0 0.73 0.0016 0.9063 0.0217 0.0270 0.0000 0.0000 0.0000 0.0000 0.0004 0.0000 0.0000 0.0000 0.0429 1614 5 60.21 18.783 1608 0 0 0.74 0.0087 0.6583 0.0113 0.0659 0.0000 0.0000 0.0000 0.0000 0.0005 0.0011 0.0160 0.0000 0.2382 1615 2 55.13 12.482 1603 0 0 0.79 0.0005 0.7940 0.0025 0.0477 0.0000 0.0000 0.0000 0.0000 0.0004 0.0000 0.0000 0.0000 0.1548 1701 1 14.74 6.403 1703 0 0 0.96 0.0032 0.1393 0.0029 0.0239 0.0000 0.0000 0.0000 0.0000 0.0000 0.0110 0.0293 0.0000 0.7904 1702 1 25.73 10.492 1705 0 0 1.01 0.0005 0.1112 0.0293 0.0146 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8444 1703 2 44.40 11.918 1705 0 0 0.93 0.0015 0.1427 0.0173 0.0167 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8218 1704 1 22.98 10.143 1703 0 0 0.96 0.0094 0.5471 0.0619 0.0610 0.0002 0.0002 0.0002 0.0002 0.0000 0.0054 0.0028 0.0000 0.3116 1705 3 188.08 36.647 1001 0 0 0.92 0.0007 0.0012 0.0001 0.0060 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6146 0.3774 1706 1 46.85 13.05 1705 0 0 0.99 0.0012 0.0234 0.0031 0.0181 0.0000 0.0000 0.0000 0.0000 0.0000 0.0084 0.0043 0.0000 0.9415 1707 1 37.91 10.065 1703 0.98 1707 1.13 0.0008 0.1737 0.0085 0.1316 0.0006 0.0006 0.0006 0.0006 0.0000 0.1387 0.1512 0.0030 0.3903 1708 1 65.10 13.113 1705 0 0 0.99 0.0002 0.1365 0.0074 0.0066 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3608 0.4884 1801 1 88.44 32.217 1804 0 0 0.74 0.0022 0.4889 0.0006 0.0140 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4944 1802 1 44.92 17.238 1804 0 0 0.77 0.0055 0.4042 0.0165 0.0092 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5645 1803 1 116.91 25.888 1804 0 0 0.75 0.0025 0.1378 0.0000 0.0155 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8442 1804 2 88.09 34.029 404 0 0 0.77 0.0038 0.1272 0.0000 0.0072 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8618 1805 15 83.53 19.804 0 0 0.78 0.0020 0.5659 0.0001 0.0197 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4123 1806 1 61.80 20.608 1804 0 0 0.76 0.0014 0.6446 0.0012 0.0148 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3380 1901 6 111.51 18.81 202 0 0 0.82 0.0073 0.4274 0.0002 0.0189 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5462 1902 1 93.20 18.182 202 0 0 0.84 0.0023 0.9146 0.0011 0.0148 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0673 1903 3 47.48 15.777 202 0 0 0.77 0.0046 0.3327 0.0000 0.0193 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6434 1904 1 50.05 14.651 1903 0 0 0.81 0.0003 0.5006 0.0000 0.0159 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4832 1905 2 118.32 35.136 1903 0 0 0.78 0.0014 0.5736 0.0006 0.0164 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4080 1906 1 57.94 14.418 1905 0 0 0.80 0.0007 0.7690 0.0000 0.0153 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.2151 1907 1 96.58 27.555 1905 0 0 0.95 0.0001 0.0364 0.0000 0.0082 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9553 2001 1 55.28 11.682 2007 0 0 0.89 0.0022 0.0951 0.0000 0.0189 0.0000 0.0000 0.0000 0.0000 0.0000 0.0021 0.0000 0.0000 0.8818 2002 1 86.15 26.874 2003 0.98 2002 1.05 0.0032 0.0987 0.0007 0.2393 0.0000 0.0000 0.0000 0.0000 0.0007 0.0535 0.0109 0.0000 0.5930 2003 4 17.91 5.665 1310 0 0 0.95 0.1006 0.4873 0.0018 0.0308 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3795 2004 3 37.49 12.336 2003 0 0 0.89 0.0028 0.0775 0.0004 0.0304 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8889 2005 1 54.38 19.395 2004 0 0 0.98 0.0009 0.1478 0.0011 0.0426 0.0000 0.0000 0.0000 0.0000 0.0000 0.0056 0.0000 0.0000 0.8019 2006 1 79.00 17.4 2007 0.98 2006 0.97 0.0238 0.5842 0.0033 0.0534 0.0000 0.0000 0.0000 0.0000 0.0000 0.0016 0.0000 0.0000 0.3337 2007 2 35.66 7.899 2004 0 0 0.87 0.0356 0.2571 0.0000 0.1292 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5781 2008 1 29.45 8.349 2007 0 0 0.92 0.0056 0.1051 0.0000 0.2169 0.0000 0.0000 0.0000 0.0000 0.0007 0.0000 0.0000 0.0000 0.6717 2101 1 122.91 31.407 2107 0 0 1.43 0.0025 0.3262 0.1188 0.0251 0.0008 0.0008 0.0008 0.0008 0.0000 0.0004 0.0001 0.5231 0.0007

325

l s a RlB B GlB B nlB B HlB B flB B Land use category (λ) 1 2 3 4 5 6 7 8 9 10 11 12 13 2102 1 142.91 26.578 2108 0 0 1.61 0.0751 0.0002 0.0000 0.0124 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9108 0.0015 2103 2 81.94 17.87 0 0 1.24 0.0015 0.0086 0.0000 0.0221 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9664 0.0014 2104 1 126.50 17.355 2103 0 0 1.70 0.0688 0.0036 0.0006 0.0312 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8957 0.0002 2105 2 77.01 28.263 0 0 1.27 0.0007 0.0020 0.0000 0.0200 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9772 0.0001 2106 1 86.78 17.475 2105 0 0 1.67 0.0924 0.0031 0.0012 0.0215 0.0000 0.0000 0.0000 0.0000 0.0031 0.0000 0.0000 0.8778 0.0009 2107 2 140.76 20.277 2108 0 0 1.35 0.0255 0.0044 0.0006 0.0140 0.0000 0.0000 0.0000 0.0000 0.0011 0.0009 0.0000 0.9534 0.0002 2108 3 58.92 11.959 0 0 1.10 0.0007 0.0042 0.0000 0.0278 0.0000 0.0000 0.0000 0.0000 0.0000 0.0004 0.0000 0.9667 0.0002 2109 1 54.27 12.614 2107 0 0 1.67 0.0004 0.0311 0.0173 0.0150 0.0001 0.0001 0.0001 0.0001 0.0000 0.0000 0.0000 0.9352 0.0004 2201 1 61.12 17.831 2202 0 0 1.17 0.0001 0.2018 0.0001 0.0113 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7867 2202 2 23.06 7.772 702 0 0 0.98 0.0013 0.6071 0.0017 0.0142 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3757 2203 1 55.23 22.661 2202 0 0 1.17 0.0000 0.0753 0.0000 0.0078 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9168 2204 1 77.57 24.69 702 0 0 1.13 0.0009 0.0955 0.0000 0.0067 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8970 2301 4 141.40 39.476 2618 0.98 2301 0.74 0.0042 0.7389 0.0402 0.0455 0.0000 0.0000 0.0000 0.0000 0.0024 0.0258 0.0289 0.0000 0.1141 2302 1 64.83 16.045 2301 0 0 0.76 0.0013 0.7844 0.0091 0.0218 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.1834 2303 1 93.24 18.642 2301 0 0 0.84 0.0004 0.7029 0.0209 0.0116 0.0003 0.0003 0.0003 0.0003 0.0000 0.0000 0.0000 0.0000 0.2628 2304 1 97.39 19.055 2305 0 0 0.87 0.0030 0.9039 0.0163 0.0157 0.0007 0.0007 0.0007 0.0007 0.0000 0.0000 0.0000 0.0000 0.0583 2305 3 76.14 19.909 2301 0 0 0.80 0.0023 0.8634 0.0040 0.0171 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.1133 2306 2 79.76 16.489 2305 0 0 0.83 0.0252 0.8740 0.0059 0.0266 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0683 2307 1 130.68 21.014 2301 0 0 0.78 0.0118 0.6386 0.0120 0.0227 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3149 2308 1 57.22 16.745 2306 0 0 0.87 0.0046 0.9486 0.0169 0.0166 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0133 2401 3 33.67 15.885 2403 0 0 1.02 0.0053 0.0575 0.0032 0.0154 0.0004 0.0004 0.0004 0.0004 0.0096 0.0000 0.0000 0.9075 0.0000 2402 2 45.20 6.896 2401 0 0 1.02 0.0068 0.1899 0.0132 0.0612 0.0009 0.0009 0.0009 0.0009 0.0285 0.0000 0.0000 0.6896 0.0074 2403 4 12.80 6.276 1001 0 0 0.93 0.0790 0.0349 0.0000 0.0176 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8645 0.0040 2404 1 18.01 10.177 2403 0 0 1.02 0.0000 0.0025 0.0004 0.0259 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9712 0.0000 2405 1 44.44 16.883 2402 0 0 1.00 0.0202 0.5440 0.0542 0.0531 0.0009 0.0009 0.0009 0.0009 0.0000 0.0000 0.0000 0.3231 0.0019 2406 1 10.31 9.175 2401 0 0 1.00 0.0056 0.2010 0.0034 0.0248 0.0021 0.0021 0.0021 0.0021 0.0000 0.0000 0.0000 0.7377 0.0189 2501 1 39.40 26.143 2502 0 0 1.48 0.1159 0.3455 0.1945 0.0516 0.0000 0.0000 0.0000 0.0000 0.0000 0.0280 0.0073 0.0000 0.2572 2502 2 134.65 40.023 2506 0.98 2503 1.12 0.0126 0.6246 0.1236 0.0565 0.0000 0.0000 0.0000 0.0000 0.0008 0.0438 0.0304 0.0000 0.1077 2503 1 27.77 8.153 2502 0.98 2503 1.13 0.0024 0.2463 0.0662 0.0970 0.0000 0.0000 0.0000 0.0000 0.0000 0.2601 0.1758 0.0000 0.1522 2504 1 10.95 9.343 2505 0.98 2504 1.09 0.0072 0.4400 0.0853 0.1777 0.0000 0.0000 0.0000 0.0000 0.0000 0.1669 0.1061 0.0000 0.0168 2505 2 112.69 39.411 2507 0 0 1.05 0.0052 0.6098 0.0982 0.0435 0.0000 0.0000 0.0000 0.0000 0.0000 0.0008 0.0053 0.0000 0.2371 2506 3 3.50 3.420 2507 0.98 2506 0.97 0.0014 0.0855 0.0078 0.3793 0.0000 0.0000 0.0000 0.0000 0.0000 0.0657 0.3278 0.0000 0.1325 2507 4 144.86 64.896 2509 0.98 2506 0.90 0.0037 0.3336 0.0264 0.0246 0.0000 0.0000 0.0000 0.0000 0.0000 0.0004 0.0003 0.0000 0.6110 2508 1 125.20 34.793 2507 0 0 0.92 0.0030 0.2948 0.0437 0.0280 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6306

326

l s a RlB B GlB B nlB B HlB B flB B Land use category (λ) 1 2 3 4 5 6 7 8 9 10 11 12 13 2509 5 163.22 43.581 2601 0 0 0.90 0.0026 0.2391 0.0095 0.0268 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7221 2601 11 121.56 26.446 2605 0 0 0.82 0.0088 0.4257 0.0003 0.0172 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0439 0.5041 2602 4 45.36 22.15 2615 0 0 0.80 0.0034 0.2116 0.0075 0.0068 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7707 2603 8 99.27 20.429 2625 0 0 0.79 0.0049 0.6059 0.0085 0.0196 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3611 2604 10 141.18 21.432 2601 0 0 0.81 0.0029 0.5044 0.0006 0.0179 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4742 2605 12 44.50 13.167 2612 0 0 0.85 0.0130 0.1803 0.0006 0.0215 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7846 0.0000 2606 2 176.41 33.954 2605 0 0 0.94 0.0003 0.3445 0.0003 0.0116 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0902 0.5532 2607 1 54.63 19.123 2625 0 0 0.81 0.0077 0.5503 0.0005 0.0378 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4036 2608 1 97.33 26.617 2605 0 0 0.96 0.0000 0.0740 0.0003 0.0112 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4030 0.5114 2609 1 111.38 30.385 1001 0 0 1.01 0.0000 0.0661 0.0007 0.0061 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9271 0.0000 2610 1 66.38 20.399 2615 0 0 0.82 0.0004 0.3949 0.0000 0.0094 0.0000 0.0000 0.0000 0.0000 0.0003 0.0000 0.0000 0.0000 0.5950 2611 1 81.81 24.092 2615 0 0 0.89 0.0006 0.6714 0.0031 0.0101 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3147 2612 13 168.50 21.014 2616 0 0 0.88 0.0059 0.1691 0.0004 0.0097 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8148 0.0001 2613 1 50.71 17.869 2604 0 0 0.94 0.0036 0.4178 0.0371 0.0335 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5081 2614 2 33.66 9.420 2612 0 0 0.93 0.0002 0.1952 0.0008 0.0082 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7839 0.0116 2615 5 91.31 38.728 2601 0 0 0.84 0.0031 0.2974 0.0003 0.0063 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6929 2616 14 95.53 15.531 1001 0 0 0.90 0.0085 0.1388 0.0008 0.0106 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8413 0.0000 2617 1 76.50 13.244 2606 0 0 0.98 0.0000 0.1169 0.0004 0.0147 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.1189 0.7490 2618 7 138.66 23.491 2603 0 0 0.77 0.0033 0.4164 0.0082 0.0519 0.0000 0.0000 0.0000 0.0000 0.0005 0.0087 0.0470 0.0000 0.4640 2619 1 92.58 20.915 2615 0 0 0.86 0.0012 0.8097 0.0079 0.0152 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.1660 2620 1 62.26 15.497 2603 0 0 0.78 0.0112 0.4221 0.0053 0.0634 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4980 2621 3 101.20 21.909 2602 0 0 0.80 0.0025 0.5994 0.0075 0.0135 0.0000 0.0000 0.0000 0.0000 0.0015 0.0000 0.0000 0.0000 0.3756 2622 2 129.09 41.601 2604 0.98 2622 1.02 0.0034 0.3824 0.0336 0.0575 0.0001 0.0001 0.0001 0.0001 0.0012 0.0212 0.0028 0.0000 0.4973 2623 1 151.58 30.503 2614 0 0 1.01 0.0000 0.0048 0.0001 0.0063 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8436 0.1452 2624 1 122.73 24.096 2628 0 0 0.86 0.0013 0.6341 0.0066 0.0145 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3436 2625 9 52.39 18.792 2604 0 0 0.80 0.0099 0.4742 0.0075 0.0145 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4939 2626 1 64.02 15.284 2622 0 0 0.85 0.0043 0.4411 0.0072 0.0288 0.0004 0.0004 0.0004 0.0004 0.0002 0.0016 0.0000 0.0000 0.5153 2627 1 146.26 22.699 2606 0 0 0.89 0.0006 0.7807 0.0021 0.0244 0.0000 0.0000 0.0000 0.0000 0.0000 0.0007 0.0000 0.0000 0.1913 2628 2 83.95 16.990 2621 0 0 0.84 0.0005 0.5907 0.0004 0.0212 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.3872 2701 1 49.11 18.373 0 0 1.50 0.0500 0.0066 0.0001 0.0405 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8981 0.0047 2702 1 24.97 13.344 0 0 1.27 0.0652 0.0003 0.0000 0.0126 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9206 0.0012

327