Inactivation of faecal indicator bacteria in a roof-captured rainwater system under ambient meteorological conditions

Authors

  • W. Ahmed,

    Corresponding author
    1. CSIRO Land and Water, Ecosciences Precinct, Dutton Park, Qld, Australia
    2. Faculty of Science, Health and Education, University of the Sunshine Coast, Maroochydore DC, Qld, Australia
    • Correspondence

      Warish Ahmed, CSIRO Land and Water, Ecosciences Precinct, 41 Boggo Road, Dutton Park, Qld 4102, Australia. E-mail: Warish.Ahmed@csiro.au

    Search for more papers by this author
  • K. Richardson,

    1. CSIRO Land and Water, Ecosciences Precinct, Dutton Park, Qld, Australia
    2. School of Population Health, University of Queensland, Brisbane, Qld, Australia
    Search for more papers by this author
  • J.P.S. Sidhu,

    1. CSIRO Land and Water, Ecosciences Precinct, Dutton Park, Qld, Australia
    2. Faculty of Science, Health and Education, University of the Sunshine Coast, Maroochydore DC, Qld, Australia
    Search for more papers by this author
  • P. Jagals,

    1. School of Population Health, University of Queensland, Brisbane, Qld, Australia
    Search for more papers by this author
  • S. Toze

    1. CSIRO Land and Water, Ecosciences Precinct, Dutton Park, Qld, Australia
    2. School of Population Health, University of Queensland, Brisbane, Qld, Australia
    Search for more papers by this author

Abstract

Aims

In this study, faecal indicator bacteria (FIB) namely Escherichia coli and Enterococcus spp. were seeded into slurries of possum faeces and placed on the roof and in the gutter of a roof-captured rainwater (RCR) system. The persistence of FIB in these circumstances was determined under ambient climatic conditions. FIB persistence was also determined under in situ conditions in tank water using diffusion chambers.

Methods and Results

The numbers of surviving FIB at different time intervals were enumerated using culture-based methods. Both FIB were rapidly inactivated on the roof under sunlight conditions (T90 = 2 h) compared with shade conditions (T90 = 9–53 h). Significant differences were observed between sunlight and shade conditions on the roof for both T90 values of E. coli (< 0·001) and Enterococcus spp. (< 0·001). E. coli showed biphasic inactivation patterns under both clean and unclean gutter conditions. Enterococcus spp., however, showed rapid inactivation (T90 = 2 h for the clean gutter and T90 = 6 h for the unclean gutter) compared with E. coli (T90 = 22 h for the clean gutter and T90 = 20 h for the unclean gutter). Significant differences were also observed between the T90 values of E. coli and Enterococcus spp. for both clean (< 0·001) and unclean (< 0·001) gutters. Both E. coli and Enterococcus spp. showed nonlinear biphasic inactivation in tank water. Significant difference was observed between the T90 value of E. coli inactivation compared with Enterococcus spp. (< 0·001) in the tank water.

Conclusions

In this study, FIB were observed to survive longer (T90 = 9–53 h) on the roof under shade conditions compared with sunlight conditions (T90 = 2 h). If there is a rainfall event within two to three days after the deposition of faecal maters on the roof, it is highly likely that FIB would be transported to the tank water. When introduced into the tank, a relatively slow inactivation process may take place (T90 = 38–72 h).

Significance and Impact of the Study

The presence of FIB in water indicates faecal pollution and potential presence of enteric pathogens. Therefore, the information on the resilience of FIB, as obtained in this study, can be used for indirect assessment of health risks associated with using roof-captured rainwater for potable and nonpotable purposes.

Introduction

Roof-captured rainwater (RCR) is considered as an alternative water source in many countries because it has the potential to replace significant volumes of potable water in and around domestic dwellings as well as industries (Evans et al. 2006; Despins et al. 2009). The most significant issue in relation to RCR use, however, is the potential public health risks associated with the exposure to pathogens that could be present in the faeces of birds, insects, mammals and reptiles (Marino et al. 1992; Chilvers et al. 1998; Ahmed et al. 2012a). Faeces from animals and other organic debris deposited on the roof and in the gutter can be introduced into the tank via roof runoff during rain events. The presence of potential bacterial pathogens, such as Aeromonas spp., Campylobacter spp., Salmonella spp., Legionella pneumophila, and protozoa pathogens such as Giardia spp. and Cryptosporidium spp., along with the traditional faecal indicator bacteria (FIB) namely Escherichia coli and Enterococcus spp. have been frequently reported in rainwater tank samples (Ahmed et al. 2008; Crabtree et al. 1995; Savill et al. 2001; Simmons et al. 2001).

The presence of FIB in water indicates the potential for presence of pathogens of faecal origin. A recent study reported that the microbiological quality of rainwater tanks in an Ecovillage in Southeast Queensland (SEQ), Australia was highly variable. The presence of FIB along with the occurrence of bacterial and protozoa pathogens in rainwater tank and related household tap water samples suggests that untreated RCR may not be suitable for drinking (Ahmed et al. 2012a). Recent studies have also reported the presence of clinically significant virulence genes associated with FIB in rainwater tank samples in SEQ (Ahmed et al. 2011, 2012b). Questions have, therefore, arisen regarding the persistence of FIB and pathogens in rainwater tank samples as well as in faecal deposits on the roof and in the gutter.

A range of climatic and biological factors have been shown to influence the inactivation of FIB and pathogens. These factors include temperature, moisture content, solar radiation, relative humidity, pathogen type, presence of biodegradable organic matter and interaction with other micro-organisms (Hurst et al. 1980; Jiang et al. 2002; Sidhu et al. 2008; Wanjugi and Harwood 2013). The FIB and pathogens from bird and animal droppings deposited on the roofs and in the gutters are expected to inactivate rapidly due to harsh environmental conditions such as temperature, UV radiations and loss of moisture. However, certain conditions such as the shade portion of the roof, instant rain, availability of biodegradable organic matter in the gutter may prolong the inactivation of FIB and pathogens. However, very little has been documented about the inactivation of FIB and pathogens on the roofs, in the gutters and in the tank water.

The primary aim of this study was to investigate the persistence of E. coli and Enterococcus spp. on the roof and in the gutter of a model RCR system. The specific aims of the study were to: (i) determine one log10 inactivation time (T90) of E. coli and Enterococcus spp. on the corrugated iron roof and in the gutters; (ii) one log10 inactivation time (T90) of FIB in the tank water.

Materials and methods

Study site and experimental RCR system construction

This study was conducted at the Ecosciences Precinct, Dutton Park, Qld, Australia (27º29′ 45·62″S; 153º01′ 46·41″E) from March to April 2012 (late summer to early autumn). A model RCR system (similar set up to RCR systems commonly in use in Australian domestic dwellings) comprising of a 5000 l polyethylene tank, a roof (2 m2) constructed with corrugated iron sheets and steel guttering and plastic downpipe leading water into the tank to simulate an urban RCR system. The tank was placed under direct sunlight (received minimum 5 h sunlight per day during the experiment).

Sources of faecal indicator bacteria (FIB) used for roof, gutter and tank water inactivation experiments

Escherichia coli and Enterococcus spp. were isolated from fresh faeces of possums and wild birds (crow, pigeon, honey eaters and magpie) as these were the most likely sources of faecal deposition on the roof and in the gutters of dwellings in SEQ. Possum and bird faeces were pooled separately, serially diluted with phosphate buffer saline (PBS) and were streaked on Chromocult™ coliform agar (Merck, KGaA, Darmstadt, Germany) and Chromocult™ enterococci agar (Merck) plates. Agar plates were incubated overnight at 37°C. Ten E. coli colonies (five from possum and five from birds) and 10 Enterococcus spp. colonies (five from possum and five from birds) were isolated from the respective agar plates and streaked twice on the agar plates to obtain pure colonies. All these colonies were confirmed as E. coli and Enterococcus spp. by PCR amplifications of 23S rRNA genes as described elsewhere (Frahm and Obst 2003; Haugland et al. 2005). Mixed E. coli and Enterococcus spp. colonies (pure) were inoculated in flasks containing nutrient broth (BD, Sparks, MD, USA) and brain heart infusion broth (Merck), respectively. The flasks were kept in a shaking incubator at 100 rpm overnight at 37°C. Prior to seeding, the mixed bacterial cultures were washed twice in 20 ml sterile phosphate buffer saline (PBS). The cultures were centrifugation at 6000 g for 3 min. followed by resuspension of the pellet in fresh PBS to remove culture media. The cultures were then acclimatized in PBS overnight at room temperature as described elsewhere (Gordon and Toze 2003). At the same time, the numbers of FIB were enumerated from the culture suspension using spread plate method. In brief, serial dilutions were made for the culture suspension and streaked on Chromocult™ coliform agar (Merck) and Chromocult™ enterococci agar (Merck) plates for the isolation of E. coli and Enterococcus spp., respectively. Agar plates were incubated overnight at 37°C for 24 and 48 h. Plates with 20–200 colonies were enumerated.

FIB inactivation on the roof and in the gutter

The FIB inactivation study was designed to simulate the event of roof and gutter contamination with possum and bird faecal materials by using homogenized possum faecal slurry. Possum faecal pellets were chosen as seeding matrix as they generally contain more organic matter and as a result dry out more slowly which potentially provide more favourable condition for pathogen survival on the roof and in the gutter compared with small-sized bird or lizard faecal droppings that desiccate more rapidly. Briefly, possum faecal pellets (= 8) were collected from the Currumbin Wildlife Sanctuary and Orphan Native Animal Rear and Release Association Inc., SEQ and transported to the laboratory on ice, where they were stored at 4°C. The collected pellets from several possums were pooled and homogenized into c. 900 ml slurry using sterile PBS. The numbers of E. coli and Enterococcus spp., in possum faecal slurry were determined within 24 h of collection using the spread plate method as described elsewhere (Sidhu et al. 2008). The numbers were c. 2 × 10ml−1 of slurry for both FIB. Because the numbers of FIB were low in possum faecal slurry, known numbers of mixed E. coli and Enterococcus spp. were seeded into the possum faecal slurry to a final numbers of c. 108 E. coli and 106 Enterococcus spp. ml−1 of slurry.

For the roof and gutter inactivation experiments, 5 ml of faecal slurry was poured into a series of 50 mm petri dishes (without lids) and placed on the corrugated iron roof and in the gutter of the experimental structure. The petri dishes were exposed to diurnal cycles of insolation. For the roof experiment, 30 petri dishes were kept directly under sunlight and another 30 were kept in the shade. The shade on the roof was artificially created by placing a tarpaulin over the petri dishes. Enough air space was provided between the petri dishes and the tarpaulin to ensure that sufficient airflow could still occur. For the gutter experiment, 30 petri dishes were placed in the clean segment of the gutter (free from faecal matters and organic debris), and another 30 petri dishes were placed in the unclean gutter. The gutter was made unclean by filling with moist sediment (similar to unclean urban household gutters) containing vegetation and organic debris. The petri dishes were kept under the vegetation and organic debris.

Roof and gutter experiments started in the early morning of the day and ended after 96 and 48 h, respectively. Triplicate petri dishes containing FIB were randomly collected at 0, 1, 2, 3, 4, 6, 8, 24, 48, 72 and 96 h time intervals from the roof and respective gutter segments, and the numbers of surviving FIB were enumerated. The collected petri dishes on each sampling occasion were placed on ice, transported to the laboratory and processed within 2–4 h. The volume of slurry was adjusted to 5 ml with PBS in petri dishes where desiccation was observed to correct for evaporation loss. Rehydrated materials were scraped carefully from the petri dishes and transferred to 15-ml tubes. Enumeration of E. coli and Enterococcus spp. was then performed in triplicate. Serial dilutions were made for each replicate, and the numbers of FIB were enumerated using a spread plate method as described earlier.

FIB inactivation in tank water

The inactivation experiment in tank water was undertaken using diffusion chambers as previously described (Toze et al. 2004; Sidhu and Toze 2012). The diffusion chambers with internal volume of 7 ml were made of Teflon and 25 mm diameter mixed cellulose esters membranes with a pore size of 0·025 μm (Millipore, Tokyo, Japan). The chambers were placed inside the rainwater tank under the water surface to rule out the influence of external factors such as sunlight or evaporation. The membranes on either side of the chamber allow passage of water and nutrients through the diffusion chamber, but prevent seeded micro-organisms escaping from the chambers (Pavelic et al. 1998; Toze et al. 2010; Sidhu and Toze 2012).

Prior to setting up diffusion chambers, a rainwater sample from the tank was collected in a sterile 1-l glass bottle and stored at 4°C. The background numbers of E. coli and Enterococcus spp. in the collected rainwater sample were determined by membrane filtration method (U.S. EPA 1997). The numbers were determined to be <10 CFU 100 ml−1 for both FIB. The numbers of mixed E. coli and Enterococcus spp. in the prepared PBS suspension were added to the rainwater sample matrix to a final number of c. 3·6 × 106 E. coli and 1·4 × 107 Enterococcus spp. ml−1 of water. The seeded rainwater sample was distributed equally into 24 diffusion chambers. This provided three replicate samples for collection on each of the eight sampling occasions (after time 0). The residual of seeded volume remaining after filling the chambers was retained for use as the time 0 sample. All of the assembled diffusion chambers were suspended in the tank at a depth of 1 m below the water level with a steel wire. Three replicates diffusion chambers were collected at on each sampling event (24, 48, 96, 144, 240, 408 and 800 h) intervals from the tank water, and the numbers of surviving FIB were enumerated. The collected samples were placed on ice, transported to the laboratory and processed within 2–4 h. The 7 ml seeded water sample was transferred from each diffusion chamber to sterile 15-ml polypropylene tubes using a sterile 10-ml syringe and 21 gauge needles. Sample serial dilutions were made, and the surviving numbers of E. coli and Enterococcus spp. were enumerated using the spread plate method as described above.

Meteorological data

Ambient temperature, rainfall, evaporation, relative humidity, wind speed and solar exposure data were collected from the Australian Bureau of Meteorology (BOM) web site during each of the inactivation experiments. Temperature data loggers (HOBO devices; Onset Computer Corporation, Pocasset, Mass.) were placed on the roof and in the gutter adjacent to the petri dishes to record the temperature at 1 h intervals for the duration of the roof and gutter experiment. A HOBO device was also suspended into the tank water to record the temperature at 4 h intervals for the duration of the tank water experiment.

Data analysis

For each FIB, all determined numbers in each replicate at each sampling occasion were converted to log10 values and plotted over time. One log10 reduction time (T90) for each FIB was determined from each plot using the following equation as previously described (Gordon and Toze 2003).

display math

where C0 is the number (CFU ml−1) at day 0, Ct is the final number (CFU ml−1) at day t. A linear regression was fitted to each plot and the slope was taken as the inactivation rate. The inverse of these calculated inactivation rates was then used as the determination of the one log10 reduction time (T90). The average T90 on each sampling occasion was determined from the replicates of each FIB. Where the inactivation of FIB in some experiments was biphasic, two T90 values were calculated, one for the initial inactivation (first phase) and the other for the second stage of the inactivation (second phase) (Gordon and Toze 2003). The number of data points used to calculate T90 values varied in the biphasic inactivation curves. Only data points above the assay limit of detection were used in calculating the inactivation times. An analysis of variance (anova) was performed on the T90 values of FIB on the roof and in the gutter under different conditions. For statistical comparison, T90 values derived from the first phases of various experimental conditions were used. The critical P-value for the test was set at 0·05.

Results

Climatic conditions

The average ambient minimum temperature during the experiments ranged from 15·5 ± 3·10°C (sunlight roof) to 18·0 ± 2·90°C (shade roof) (Table 1). The average maximum temperature ranged from 26·6 ± 2·02°C (shade roof) to 30·2 ± 2·20°C (tank water). The evaporation rates were high throughout the inactivation experiments with values higher during the roof experiments compared with the gutter. Solar exposure was higher (16·9 ± 3·12 MJ m−2) during the sunlight roof experiment compared with shade roof (12·6 ± 1·14 MJ m−2) and gutter (10·3 ± 1·12 MJ m−2) experiments.

Table 1. Average (± SD) ambient meteorological data during inactivation experiments
Meteorological dataExperiments
RoofGuttercTank waterd
SunlightaShadeb
  1. a

    12-04-2012.

  2. b

    17-04-2012 to 21-04-2012.

  3. c

    23-04-2012 to 25-04-2012.

  4. d

    19-02-2012 to 25-03-2012.

  5. SD: Standard deviation.

  6. NA: Not applicable.

Temperature minimum (ºC)15·5 ± 3·1018·0 ± 2·9016·2 ± 2·0215·6 ± 1·30
Temperature maximum (ºC)25·3 ± 4·1026·6 ± 2·0224·9 ± 3·0530·2 ± 2·20
Rainfall (mm)3·407·400·00NA
Evaporation (mm)5·40 ± 2·104·06 ± 1·603·32 ± 1·01NA
Relative humidity (%)74·5 ± 5·2679·3 ± 4·9877·2 ± 2·05NA
Wind speed (km h−1)34·5 ± 5·3023·6 ± 2·4021·6 ± 3·24NA
Solar exposure (MJ m−2)16·9 ± 3·1212·6 ± 1·1410·3 ± 1·12NA

The tank water temperature (measured by data logger) ranged from 21·4–28·5°C (average 24·1 ± 2·42°C) during the tank inactivation experiment. The temperature of the corrugated iron roof and gutter was also measured using the data loggers. For the roof sunlight experiment, the temperature ranged from 23·7 to 39·3°C (average 28·7 ± 7·42°C), and for the roof shade experiment, the temperature ranged from 19·3 to 26·0°C (average 24·4 ± 2·32°C). Gutter temperature ranged from 24·5 to 38°C (average 29·3 ± 5·35°C) for the duration of the gutter inactivation experiment.

FIB inactivation on the roof

The inactivation rates of FIB in possum faecal slurry placed on the roof were evaluated under sunlight and shade conditions (Fig. 1). Under direct sunlight, E. coli rapidly inactivated (T90 = 2 h) compared with shade where slow nonlinear (biphasic) inactivation rate [T90 = 53 h (first phase) and 9 h (second phase)] was observed (Table 2).

Table 2. T90 inactivation time of Escherichia coli and Enterococcus spp. on the roof, in the gutter and tank water
Faecal indicatorsExperimentsConditionsT90 (h) (R2)
First phaseSecond phase
  1. a

    Three data points were used to calculate T90 inactivation times.

  2. b

    Four data points were used to calculate T90 inactivation times.

  3. c

    Five data points were used to calculate T90 inactivation times.

  4. d

    Six data points were used to calculate T90 inactivation times.

  5. R2Coefficient of determination.

E. coli RoofSunlight2 (0·72) 
Shade53 (0·85)b9 (0·95)a
GutterClean22 (0·66)a3 (0·99)a
Unclean20 (0·73)b6 (0·88)a
Tank water 72 (0·86)a273 (0·84)c
Enterococcus spp.RoofSunlight2 (0·95) 
Shade9 (0·92)b18 (0·80)c
GutterClean2 (0·80) 
Unclean6 (0·97) 
Tank water 38 (0·99)a195 (0·94)d
Figure 1.

Log10 (mean ± SD) colony-forming units (CFU) of culturable Escherichia coli (EC) and Enterococcus spp. (ENT) during the course of inactivation on the roof under sunlight and shade conditions. (image_n/jam12342-gra-0001.png) EC sunlight roof; (image_n/jam12342-gra-0002.png) EC shade roof; (image_n/jam12342-gra-0003.png) ENT sunlight roof and (image_n/jam12342-gra-0004.png) ENT shade roof.

Similar results were also obtained for Enterococcus spp. which was inactivated faster under direct sunlight (T90 = 2 h) compared with shade condition where slow biphasic inactivation [T90 = 9 h (first phase) and 18 h (second phase)] was also observed. A rapid inactivation (1 log) was observed in the first 8 h followed by a slow decline. No significant (> 0·05) difference was observed in T90 value of E. coli inactivation compared with Enterococcus spp. for sunlight conditions. A significant (< 0·001) difference, however, was observed in T90 value of E. coli compared with Enterococcus spp. for shade conditions. A significant differences were also observed between sunlight and shade conditions for both T90 values of E. coli (< 0·001) and Enterococcus spp. (< 0·001).

FIB inactivation in the gutter

The inactivation rates of FIB in possum faecal slurry were evaluated in the clean and unclean gutters under direct sunlight with the organic matter and vegetation in the unclean gutter shading the faecal slurry from sunlight (Fig. 2). E. coli showed biphasic inactivation patterns under both clean and unclean gutter conditions [T90 = 22 h (first phase) and 3 h (second phase)] for the clean gutter conditions and [T90 = 20 h (first phase) and 6 h (second phase) for the unclean gutter conditions. Enterococcus spp., however, showed rapid inactivation (T90 = 2 h for the clean gutter and T90 = 6 h for the unclean gutter) compared with E. coli. No significant (> 0·05) difference was observed for the T90 values of E. coli inactivation between clean and unclean gutter conditions. A significant (< 0·001) difference, however, was observed in T90 value of Enterococcus spp. between clean and unclean gutter conditions. Significant differences were also observed between the T90 values of E. coli and Enterococcus spp. for both clean (< 0·001) and unclean (< 0·001) gutter conditions.

Figure 2.

Log10 (mean ± SD) colony-forming units (CFU) of culturable Escherichia coli (EC) and Enterococcus spp. (ENT) during the course of inactivation in the clean and unclean gutters. (image_n/jam12342-gra-0001.png) EC clean gutter; (image_n/jam12342-gra-0002.png) EC unclean gutter; (image_n/jam12342-gra-0003.png) ENT clean gutter and (image_n/jam12342-gra-0004.png) ENT unclean gutter.

FIB inactivation in tank water

The inactivation rates of FIB were determined under in situ conditions in the rainwater tank kept exposed to sunlight (Fig. 3). E. coli fell below detection limit after 576 h, whereas, Enterococcus spp. were detected up to 816 h. Both E. coli [T90 = 72 h (first phase) and 273 h (second phase)] and Enterococcus spp. [T90 = 38 h (first phase) and 195 h (second phase)] showed nonlinear biphasic inactivation. A Student's paired t-test was performed on the T90 values between E. coli and Enterococcus spp for tank water. The critical P-value for the test was set at 0·05. Significant difference was observed between the T90 value of E. coli inactivation compared with Enterococcus spp. (paired t-test, < 0·001) in the tank water.

Figure 3.

Log10 (mean ± SD) colony-forming units (CFU) of culturable Escherichia coli (EC) and Enterococcus spp. (ENT) during the course of inactivation in the tank water. (image_n/jam12342-gra-0001.png) EC tank water and (image_n/jam12342-gra-0003.png) ENT tank water.

Discussion

Large numbers of rainwater tank samples in SEQ were reported to have faecal indicators above the Australian drinking water guideline value (Ahmed et al. 2008, 2010). Furthermore, PCR analysis of clinically significant virulence genes associated with E. coli and Enterococcus spp. indicated the presence of a wide array of virulence genes in E. coli and Enterococcus spp. isolated from rainwater tank samples (Ahmed et al. 2011, 2012b). Certain E. coli strains from rainwater tank samples harbouring virulence genes were identical to those found in possum and bird faeces (Ahmed et al. 2012c). Pathogens such as Campylobacter spp., Salmonella spp., Giardia intestinalis and Cryptosporidium spp. have also been detected in rainwater tank samples (Ahmed et al. 2010). No information is available on the inactivation rates of E. coli and Enterococcus spp. on the roof, in the gutter and in the tank water. This study, therefore, was undertaken to obtain information on the inactivation of FIB under different scenarios that could assist in the assessment of health risks.

The inactivation of E. coli and Enterococcus spp. was investigated in this study due to their wide prevalence in RCR systems in SEQ and the former is the FIB recommended for monitoring microbiological quality of RCR in SEQ, Australia (Ahmed et al. 2010). The results showed that E. coli and Enterococcus spp. became inactivated more rapidly on the roof under sunlight conditions compared with shade conditions. The average ambient daily minimum and maximum temperature over the study period for sunlight and shade conditions were similar, and, therefore, appeared not to have played any significant role in FIB inactivation. Faecal indicator bacteria are part of the normal gut-flora of warm-blooded animals and the optimum growth temperature is 35°C for most enteric bacteria, although growth can occur at higher and lower temperatures (Sinton et al. 2002). However, the rate of inactivation at temperatures <35°C may have little impact on the inactivation (Sinton et al. 2007; Klein et al. 2011).

A biphasic inactivation was observed under shade conditions for both FIB with slow inactivation rates up to 48 h followed by rapid losses. An important factor leading to the rapid inactivation under sunlight could be high intensity of ultraviolet radiations associated with direct sunlight (16·9 ± 3·12 MJ m−2) compared with shade (12·6 ± 1·14 MJ m−2) where the faecal slurry was sheltered from the direct sunlight. This was in agreement with a previous study undertaken by Moriarty et al. (2011). In this study, the roof experiment was undertaken on the corrugated iron roof, which can have extremely high surface temperature under sunlight as previously observed by Bretz et al. (1998). The temperature logger kept on the roof under sunlight in this study recorded temperature as high as 39·3°C during the first few hours of the experiment. The combination of direct sunlight and high roof temperatures may have lead to a rapid inactivation of FIB. It should be noted that inactivation rates of FIB could yield different results on other roof types such as fibreglass, concrete or tiled roof, which were not included in this study. Loss of moisture through rapid evaporation (5·4 ± 2·10 mm per day) may have been another factor leading to the rapid inactivation observed for sunlight conditions compared to shade conditions as moist conditions are essential for the viability of metabolically active bacteria (Ward et al. 1981; Sinton et al. 2002). Under direct sunlight, the complete dessication (dried as evident by lack of moisture) of the slurry was observed to occur within 2 h compared with the shade conditions where the complete desiccation occurred in 8 h. Moriarty et al. (2011) reported a significant rise in inactivation rates of FIB in cow pats when the moisture content of the pats decreased from 80% to 40%.

For the gutter experiment, it was hypothesized that FIB would survive longer in the unclean gutter compared with the clean gutter due to microclimatic factors such as moisture, nutrients and protection from UV in the leaf debris. No significant differences were observed for the T90 values of E. coli inactivation between clean and unclean gutters. A significant difference, however, was observed in T90 value of Enterococcus spp. between clean and unclean gutters. Enterococcus spp. in both unclean and clean gutters survived a relatively shorter period of time than E. coli. This could be due to the fact that E. coli showed biphasic inactivation patterns in both types of gutters. The T90 values of E. coli in the first phases for clean and unclean gutters were higher (T90 = 20–22 h) compared to the second phases where T90 values were much lower (T90 = 3–6 h). The T90 values of E. coli in the second phase for both gutters, however, were similar to the T90 values of Enterococcus spp. We acknowledge that clean and unclean gutter experiments were undertaken under sunlight conditions only. It is highly likely that FIB would inactivate more slowly in the unclean gutter under shade conditions. Other conditions such as seasonal impacts and higher moisture levels in the leaf litter could also influence the inactivation of micro-organisms.

When comparing the inactivation of two FIB groups in the tank water, the results indicated that Enterococcus spp. had a faster T90 time than E. coli. This concurs with a previous study that showed that faecal coliforms had greater persistence in freshwater than Enterococcus spp. (Sinton et al. 2002; Anderson et al. 2005). Slower inactivation rates were observed for both indicators in the tank water compared to the roof and gutter experiments. This was not unexpected, considering the fact that FIB in tank water were not exposed to harsh meteorological conditions such as sunlight, desiccation that could attribute faster inactivation on the roof and in gutter experiments. We acknowledge that FIB inactivation in tank water was undertaken using diffusion chambers that have internal volume of 7 ml. The diffusion chambers were filled with water sample (collected from the experimental tank) seeded with FIB to allow predation by existing predators. The numbers of predators in 7 ml of water sample may be low that may have inflated the T90 values of FIB under the experimental conditions.

It is possible that certain strains of FIB survived better than others (Anderson et al. 2005) in the tank water, on the roof and in the gutter as mixed faecal strains were used for spiking in this study. The biphasic inactivation rates of FIB for the roof, gutter and tank experiments suggesting that perhaps certain strains of FIB survived better than others (Hellweger et al. 2009). Studies have shown that E. coli persistence in the environment can be strain dependent (Whittam 1989; Topp et al. 2003; Anderson et al. 2005). Several studies have also documented slow inactivation of FIB of faecal origin similar to the results reported in this study (Sommer et al. 1997; Sinton et al. 2002; Noble et al. 2004). It is also possible that other factors such as dessication of faecal slurry, moisture content and UV radiation may have contributed to the biphasic inactivation. It is noted that inactivation experiment was undertaken in the tank, which was kept under direct sunlight. Many urban rainwater tanks are located in the shade or underneath the house shielding them from high temperature, and in such conditions, FIB inactivation rates may differ from the results obtained in this study.

In conclusion, FIB especially E. coli can survive longer (T90 = 53 h) on the roof under shade conditions compared with sunlight conditions. This could have an impact on health risks associated with tank water use. If there is a rainfall event after the deposition of faecal matter on a shade roof, it is highly likely that FIB and other faecal pathogens could be transported to the tank water. When introduced to the tank, a slower inactivation process may take place (T90 = 38–72 h). Further research is required to understand the persistence of bacterial and protozoa pathogens on the roof and in tank water in relation to FIB because certain pathogens are known to be more persistent in the environment than FIB. Maintenance of good roof and gutter hygiene and elimination of overhanging tree branches and other mounted structures on the roof where possible to prevent the flocking of possums and birds should be considered to minimize chances of faecal contamination on the roof and in the gutter. The magnitude of faecal contamination in rainwater tank immediate after rain events can be minimized by installing first flush device as the first flush runoff may contain a large amount of the faecal contamination load. Approximately, 10% (ABS 2007) of Australian people use rainwater as a major source of their drinking water, and therefore, it is recommended that rainwater should be treated with effective treatment procedures such as filtration, ultraviolet disinfection or simply boiling the water prior drinking.

Acknowledgements

This research was undertaken and funded as part of the Urban Water Security Research Alliance, a scientific collaboration in Southeast Queensland, Australia, between the Queensland government, CSIRO, The University of Queensland and Griffith University. We thank Currumbin Wildlife Sanctuary and Orphan Native Animal Rear and Release Association Inc., SEQ for providing possum faecal samples.

Conflict of Interest

The authors declare no conflict of interest.

Ancillary