Survival of pathogenic and indicator bacteria in biosolids applied to agricultural land.
Abstract: Concentrations of surviving Escherichia coli, Clostridium perfringens, and Salmonella slap. were determined temporally in mechanically dewatered biosolids derived from anaerobic--mesophilic digestion and applied to agricultural land. Following applications in different seasons, repeated assessments of bacterial concentrations in biosolid clumps, using most-probable-number (MPN) techniques, found sustained high levels of these bacteria. Bacterial concentrations were often well above soil background levels at 6 months, and in some cases 11-12 months, after land application. Survival in surface-applied biosolids was similar to that for biosolids incorporated into the soil, and between application rates of 10 or 30 dry t/ha. Salmonella concentrations in applied biosolids were not predicted from, and could exceed those of, the indicator organism E. coli. Multiple plot analyses indicated regrowth of E. coli and Salmonella can occur within biosolids, up to several months after application. However, Salmonella serovars likely to pose a significant risk to animal health were not detected among isolates from the dewatered biosolids. Reduced accessibility for grazing livestock by soil incorporation, together with the time taken for normal pasture establishment practices, and the limited pathogenicity of the vast majority of salmonellae present in biosolids may significantly reduce the risk of spread of these organisms to the human food chain.

Additional keywords: sewage sludge, survival, splines.
Subject: Pathogenic microorganisms (Research)
Sewage sludge (Waste management)
Soil microbiology (Research)
Authors: Eamens, G.J.
Waldron, A.M.
Nicholls, P.J.
Pub Date: 11/01/2006
Publication: Name: Australian Journal of Soil Research Publisher: CSIRO Publishing Audience: Academic Format: Magazine/Journal Subject: Agricultural industry; Earth sciences Copyright: COPYRIGHT 2006 CSIRO Publishing ISSN: 0004-9573
Issue: Date: Nov, 2006 Source Volume: 44 Source Issue: 7
Topic: Event Code: 310 Science & research; 420 Pollutants produced & recycled
Geographic: Geographic Scope: Australia Geographic Code: 8AUST Australia
Accession Number: 155614684
Full Text: Introduction

Several studies have evaluated the presence and survival of microorganisms in sewage products (Yanko et al. 1978, Jones et al. 1980; Surampalli et al. 1994; Hu et al. 1995), but relatively few have studied their survival after land application of biosolids (Hess and Breer 1975; Ibiebele et al. 1985; Pepper et al. 1993), particularly dewatered biosolids (Gibbs et al. 1995, 1997). This product can be transported more easily than liquid biosolids, and is manufactured for short-term storage at a waste water treatment plant (WWTP) before land application. For dewatered biosolids, routine agricultural applications in New South Wales (NSW), Australia, currently require soil incorporation within 36h (NSW Environment Protection Authority 1997). However, surface application was also investigated in this study, as this may have future application for improving sites already sown with improved pastures.

Escherichia coli and Clostridium perfringens represent faecal indicator organisms, and can be enumerated to assess the survival of bacteria specific to faeces, and to compare with existing data from previous studies on biosolids. While there are some reports describing faecal coliform die-off in biosolids, there is a paucity of specific information on E. coil Of the faecal coliforms, E. coli is the main organism associated with human and animal disease, particularly strains that possess virulence factors for assisting enteric colonisation and for toxin production. A third bacterial group investigated in this study, Salmonella spp., represents an important bacterial risk factor in biosolids. While other important bacterial pathogens such as Campylobacter jejuni and related campylobacters can be found in faeces, these organisms survive poorly in digested sewage sludge (Jones et al. 1990; Stelzer and Jacob 1991).

In this paper we report quantitative microbiological results obtained from repeated sampling of dewatered biosolids after application to agricultural land using 2 different rates and 2 different methods of application. Die-off patterns of the 3 bacterial groups described above were investigated after biosolids application in different seasons.

Materials and methods

Sites

Mechanically dewatered biosolids produced by anaerobic mesophilic fermentation at the Cronulla WWTP, Sydney, NSW, were used in this study. Concentrations of E. coli, Salmonella spp., and C. perfringens spores were assessed in dewatered biosolids, either incorporated within 36h of application or left as a surface application on agricultural land at Goulburn, NSW. The biosolids were initially spread on the land, which is normally used for sheep and cattle grazing, with a rear discharge manure spreader. Incorporation to 0.15 m was undertaken using 2 passes of an offset disc plough, and domestic livestock were excluded from the study sites 6 months prior to, and throughout, the study period. The treatments are summarised in Table 1. Three separate application times, 2 rates of application, and 2 methods of application were studied during 1997-99 in a total of 12 plots, each measuring 20 by 3 m. Between each plot there was an untreated buffer zone of 2 m to assist the spreading rate accuracy for each plot. At each application time, 4 plots (A-D) were used in a 2 x 2 factorial design (rate x method). Each plot received only a single application of biosolids, in summer (Goulburn 1 plots), spring (Goulburn 2 plots), or the following summer (Goulburn 3 plots), at application rates of 10 or 30 dry t/ha. For comparative purposes, bacterial die-off was additionally measured in 2 large adjacent paddocks (E, F) that had received a surface application of 30 dry t/ha of the same biosolids and at the same time as the Goulburn 1 plots. The soil type in the plots and adjacent paddocks used in the study comprised a 0.12-m-deep sandy loam topsoil on a yellow medium clay subsoil. It was classified within the Solodic Great Soil Group (Charman 1991), and a duplex soil according to the Northcote Classification (Northcote 1974).

Immediately prior to land application, bacterial concentrations of E. coli, Salmonella spp., and C. perfringens spores in stockpiled biosolids were determined at each study site. In addition, soil baseline concentrations of the 3 bacteria at all sites were measured on plots prior to biosolids application, and during the study period on adjacent ungrazed land that had not received biosolids. The treated sites were monitored until bacterial concentrations in biosolids clumps reached or approached these baseline values. To investigate the consistency of bacterial concentrations of E. coli, Salmonella spp., and C. perfringens in product from the Cronulla WWTP, samples of fresh dewatered biosolids from this plant were collected on 6 occasions and similarly tested. These collections occurred during the winter after the Goulburn 1 application and before the Goulburn 2 application (on 3 occasions), and in the summer of the Goulburn 3 application (on 3 occasions).

Meteorological data (maximum and minimum daily temperature, rainfall) were collected at each study site for correlation with bacteriological findings. To remove excessive weed growth during the trials, some plots required a single mechanical slashing. For Goulburn 2 and 3, this was required at weeks 53 and 33, respectively.

Microbiology

Bacterial survival was monitored on each plot fortnightly and later monthly. Each plot was sampled by collecting visible biosolid clumps from 5 random sites to make a composite sample of at least 55 g. Random collections were made by selecting material located 5 m apart during zig-zag style sampling along each study site. Biosolid clumps (10-30 mm diam.) selected at random from the surface (from plots with surface-applied biosolids) or at a depth of 0.05-0.15 m (from plots with incorporated biosolids) were used for each composite sample. Where larger clumps were encountered during sampling, these were broken down by hand to portions no larger than 30 mm. For pre-application and comparison samplings from areas that had not received biosolids, random surface soil samples were collected to a depth of 0.15m. All bacterial counts were calculated by Most Probable Number (MPN) techniques and expressed per g sampled product.

For all bacteriological counts, duplicate subsamples of 27.5 g were diluted in buffered peptone water (BPW), from which 8 replicate subaliquots of 10 mL were taken. Five tubes of each subaliquot from each duplicate were used to inoculate 5-tube replicates of Ringer's solution and diluted 10-fold from [10.sup.-2] to [10.sup.-7]. Equivalent MPN series of tubes containing media for E. coli (lauryl tryptone broth with inverted Durham tubes) were inoculated (1 mL in 8 mL media). The residual Ringer's solution from the 35 tubes was heated (75[degrees]C, 12 min) prior to inoculation of 1 mL into 35 x 5 mL iron milk broth (St. John et al. 1982) for C. perfringens spore determinations by MPN. The remaining three replicates in BPW were used to inoculate a 3 tube dilution series of BPW (10-fold from [10.sup.-2] to [10.sup.-5]) for subsequent inoculation of Rappaport-Vassiliadis (RV) broth for Salmonella determinations.

E. coli isolates were identified by gas production in EC broth at 44[degrees]C, morphology on eosin methylene blue agar, and indole production in tryptone water (Standards Association of Australia 1987). C. perfringens isolates were identified by the production of a 'stormy clot' in iron milk broth associated with marked lactose fermentation and gas production (St. John et al. 1982), and by colony morphology on lactose egg yolk agar, and biochemical and motility testing in lactose--gelatin medium and in nitrate-motility medium (Standards Australia 1991b). Salmonella isolates were confirmed after selection in RV broth and examination of RV subcultures on serial Brilliant Green and Bismuth Sulfite agar plates for colony morphology at 24 and 48h, respectively (Standards Australia 1991a). Colonies morphologically consistent with Salmonella spp. were subcultured to Cystine Lactose Electrolyte Deficient (CLED) agar, and 16-24h colonies on CLED agar consistent with Salmonella spp. were selected for subculture to nutrient agar for testing of their reactivity with poly O and poly H antisera (Standards Australia 1991a) after a further 16-24h incubation period. Approximately 5% of isolates classified as Salmonella spp. for this study were poly O positive and poly H negative. This would have captured salmonellae that had lost fimbrial (H) antigens, but could have added up to 5% of a count as non-salmonellae.

Salmonellae were serotyped at the Australian Salmonella Reference Centre, Adelaide, to determine if strains found to have survived for extended periods are considered important human or animal pathogens. Sixty-six isolates of Salmonella spp., randomly selected from those cultured from biosolids during the course of the study at Goulbum 2 and 3 sites were submitted for serotyping.

Statistical analysis

Time trends of log E. coil and log Salmonella spp. concentrations at the 3 Goulburn sites were investigated using ASReml analyses (Gilmour et al. 1999) to assess whether pathogen regrowth may have occurred, and any correlation with rainfall. To establish underlying mean trends and overcome short-term variability of means within the overall die-off patterns (high for both pathogens), smoothing of the data was undertaken using cubic smoothing splines, applied as part of a linear mixed model (Verbyla et al. 1999). The model is represented as follows:

Y = Fixed [application method + rate + linear(time) + all interactions] + Random [spline(time) + deviation(time) + error]

where deviation(time) is a measure of lack-of-fit of the spline. All parameters were estimated using the Reml method (Patterson and Thompson 1971). The modelling started with a double spline to enable recognition of apparent increases in bacterial concentrations. If the difference in lack-of-fit between the double spline and a single spline was large, the double spline was considered more appropriate for modelling.

The association between E. coil and Salmonella spp. concentrations at each site and weekly rainfall between samplings was assessed by including a fixed effect of rainfall in the final model, and also in a model with all spline and lack-of-fit effects (i.e. all non-linearity) removed.

Results

E. coli ([10.sup.4] - [10.sup.5]/g), C. perfringens spores ([10.sup.3] - [10.sup.4]/g), and salmonellae ([10.sup.1] - [10.sup.2]/g) were detected in fresh dewatered biosolids from Cronulla WWTP, but concentrations of these bacteria were often substantially higher in stockpiled biosolids applied at Goulburn. Pre-application and non-treated site (baseline) MPN concentrations of E. coli and C. perfringens were 10/g or less, while Salmonella concentrations were <4/g. Meteorological data relevant to the times of biosolids application are shown in Fig. 1. Since die-off patterns were similar for plots receiving either rate of biosolids (10 or 30 dry t/ha) on all occasions, plots A and C and plots B and D were virtual replicates, and their results were pooled (as AC, BD) in presenting the bacteriological data.

[FIGURE 1 OMITTED]

Pathogen die-off patterns at Goulburn 1 (summer application)

Bacteriological results on the logarithmic scale for the 4 plots are shown in Figs 2-4. There were very high initial concentrations of E. coli and Salmonella spp. ([10.sup.6]/g) and concentrations persisted above baseline levels to weeks 40 and 51 (the final sampling), respectively. The initial concentration for C. perfringens was high ([10.sup.4]/g) but the mean declined at a slower rate than the other species and remained above the baseline level at week 51.

[FIGURES 2-4 OMITTED]

The single-spline models for log E. coli and log Salmonella spp. indicated declining trends in the mean with strong linear components and significant spline curvatures (Table 2), with no intermediate increases detected. The intercepts, slopes, and curvatures of the trends did not differ among the plots. The double-spline models used to assess regrowth of E. coli and Salmonella spp. similarly did not detect any change points, so the mean trends in the single-spline models which showed no evidence of regrowth were accepted for both pathogens. Samples from the 2 adjacent paddocks treated with 30 dry t/ha of dewatered biosolids at the same time as the plots showed similar trends (Figs 2 and 3, E1 and F1), confirming that the plots were good indicators of die-off patterns on more extensive areas treated concurrently.

Pathogen die-off patterns at Goulburn 2 (spring application)

Bacteriological results for the plots are shown in Figs 5-7 (AC2, BD2). There were high initial concentrations of E. coli (3 x [10.sup.5]/g) but lower initial concentrations of Salmonella spp. (2 x [10.sup.2]/g) and concentrations persisted above baseline levels to weeks 64 and 68 (the final sampling), respectively. The initial concentration for C. perfringens was > [10.sup.3]/g and the mean declined slowly and remained well above the baseline level at week 68.

[FIGURES 5-7 OMITTED]

The single spline model for log E. coli showed a decline in the mean that was strongly linear with no significant spline curvature (Table 2), and the intercepts and slopes of the trends did not differ among the plots. In the double spline model for log E. coli there was a change point estimated at week 39, but the resulting spline curvature was minor and from this model it was concluded that the data did not provide sufficient evidence that regrowth of E. coli occurred.

For Salmonella spp., the single-spline model for the log data suggested there were different trends for the plots with and without biosolids incorporation, with no significant linear component (Table 2). For the plots with biosolids incorporation (plots A, C) there were local minimum and maximum mean log Salmonella concentrations at week 10 and week 39, respectively, and the estimated mean increase over this period was significant (Table 2). This result suggested that regrowth of Salmonella spp. may have occurred between 10 and 39 weeks for the incorporated plots. For the plots without biosolids incorporation, the model suggested the mean was constant. For the double-spline model for log Salmonella spp. in the plots with incorporation, there was a change point estimated at week 10, and in the resulting model (Fig. 8a), spline curvature and the increase in bacterial concentration to week 39 was significant. This model provided strong evidence for the existence of regrowth of Salmonella spp. between weeks 10 and 39 on the plots with biosolids incorporation.

[FIGURE 8 OMITTED]

Pathogen die-off patterns at Goulburn 3 (summer application)

Bacteriological results for the 4 plots are shown in Figs 5-7 (AC3, BD3). The initial concentrations of E. coli (2 x [10.sup.3]/g) were lower than at the other sites but persisted above baseline levels to week 44, while those for Salmonella spp. (3 x [10.sup.2]/g) were similar to the initial concentrations at Goulburn 2 and persisted above baseline levels to the final sampling at week 52. The initial concentration for C. perfringens was high (4 x [10.sup.4]/g) but the mean declined slowly and remained above the baseline level at week 52.

For the single-spline model, the decline in mean log E. coli concentrations was mainly linear (Table 2) with no significant differences among plots in slopes or spline curvatures. In the double-spline model for E. coil there was a change point estimated at week 19, and in the resulting model (Fig. 8b), spline curvature was highly significant with a significant increase in log E. coil between weeks 19 and 25 (Table 2). This model provided strong evidence for the existence of regrowth of E. coli between 19 and 25 weeks.

The observed time pattern for log Salmonella spp. concentrations suggested an initial rise to about week 10 with a decline thereafter, but in the single-spline model, although the spline curvature was significant, the mean increase between weeks 2 and 10 was not significant (Table 2). The double spline model for Salmonella spp. also showed significant spline curvature, but again the mean increase between weeks 2 and 10 was not significant (Table 2). Therefore spline modelling of the log Salmonella spp. data did not provide sufficient support for a real increase in the mean before week 10.

Association between bacterial counts and rainfall

When weekly rainfall was included as a fixed effect in the final models described above, with and without the spline curvature component included, in each case the effect of rainfall was not detected as significant (P > 0.10).

Salmonella serotyping

Ten serovars of Salmonella spp. were found in biosolids in this study, with Tennessee, Ohio, Singapore, Rissen, and Cubana the most prevalent (Table 3). Some dewatered biosolids were shown to contain up to 7 different serovars. Salmonella typhimurium, the most important cause of salmonellosis in man and animals in Australia, was not detected.

Discussion

High counts of salmonellae were demonstrated in stockpiled dewatered biosolids at the time of land application for Goulburn 1, but lower concentrations were detected at the time of land application for Goulburn 2 and 3. This is consistent with variations in salmonellae in digester effluent from anaerobic digesters in NSW, with concentrations in effluent from a WWTP varying between <10 and [10.sup.5]/g on separate sampling occasions (Eamens et al. 1996).

In our study, bacteriological analyses of biosolids applied to agricultural land were based on clumps of biosolids, rather than on random samples of soil/biosolid mixtures. The results are therefore a direct measurement of the actual survival of bacteria within the biosolids, and avoid soil dilution factors that may have confounding effects on the measurement of bacterial die-off. Bacterial concentrations were also expressed on a per g fresh weight (wet weight) basis, for 2 reasons: firstly, for consistency with some prior studies of similar organisms in anaerobically digested, dewatered biosolids in both stockpiles (Pillai et al. 1996; Gibbs et al. 1997; Jepsen et al. 1997), and after application to agricultural land (Gibbs et al. 1995, 1997); and secondly because the pathogen risk (to humans or animals) relates to exposure to (or ingestion of) biosolids on a wet weight basis, rather than on a theoretical concentration adjusted for dry-matter content. While some environmental microbiologists express bacterial concentrations in soil, raw sludge, or biosolids on a dry matter basis, for comparison of concentrations between different samplings when moisture content may be changing, this was not considered critical in this study. Fresh dewatered biosolids from WWTPs in NSW (including Cronulla WWTP) typically contain 72-80% water content (Michalk et al. 1996; Robinson et al. 2002), which declines after field application to about 15% in a matter of weeks (Robinson et al. 2002). Other studies have found that surface applications in forests of dewatered biosolids with 20-40% solids fluctuated in moisture content by only 0.2 [log.sub.10] over 18 months (Edmonds 1976). Hence, on a logarithmic scale, an increase in bacterial concentration on a dry matter basis of up to 0.7 [log.sub.10] may be anticipated in the first few weeks after application in this study. Since the concentrations of indicator organisms at this time were relatively high, the relative adjustment of the concentration for solids content (i.e. on a dry matter basis) would be minor. Similarly some months after application, when the moisture content is relatively low and stable, the impact of water content on bacterial concentrations is still likely to be very limited, and much less than 0.7 [log.sub.10].

The results confirm dewatered biosolids contain potentially pathogenic bacteria that can survive in biosolid clumps for many months after soil incorporation or surface application. In some instances (Goulburn 2), 17 months was required to confirm that stable low or baseline levels of E. coli and Salmonella were reached in biosolids clumps, although counts at earlier samplings were at these levels on one or more occasions. C perfringens was very slow to decline in concentration, and viable spores remained at moderate levels even 17 months post-application. On agricultural land, overall there was no significant difference between survival in biosolids clumps with and without soil incorporation. However, the raw data suggested a trend to slightly greater survival in incorporated biosolids, and thus a possible protective effect under soil. The similar survival between application rates (of 10 or 30 dry t/ha) indicates that while application rates may affect total bacterial availability, the concentration per g biosolids is similar. This study also showed that concentrations of Salmonella and E. coil concentrations did not correlate, and in some cases, Salmonella spp. concentrations were found to exceed those of E. coli. In contrast to some findings in raw sludges (Yanko et al. 1978), these data demonstrate clearly that the concentration of one cannot be used to reliably predict the other after land application.

From the current study, an apparent regrowth phenomenon occurred in biosolids on some plots. The strongest evidence of regrowth occurred for Salmonella spp. in incorporated plots at Goulburn 2 between 10 and 39 weeks, and for E. coli in all plots at Goulburn 3 between 19 and 25 weeks after land application. Spring-applied biosolids showed greater bacterial survival (Goulburn 2, 68 weeks) than those applied in summer (Goulburn 1 and 3, 51-52 weeks), despite similar E. coli and Salmonella concentrations at 39-40 weeks. Of interest were variable time points and different species of bacteria associated with this regrowth, in terms of weeks after application. This suggests that environmental factors were present that influenced regrowth of the different bacterial species independently.

The investigation of regrowth of E. coli and Salmonella spp. using a mixed linear model with smoothing splines was applied on a double-spline model around the apparent time when bacterial concentrations began to increase. We found that application of the double-spline model enabled greater distinction between random fluctuation and apparent regrowth, and that there was evidence of regrowth under certain conditions, relevant to Salmonella and E. coli at Goulburn 2 and 3, respectively.

Studies of faecal coliforms in digested biosolids incorporated into soil have shown that numbers may decrease initially (to 12 weeks) but subsequently increase (to week 20) (Yanko et al. 1978) and sporadic increases in numbers are associated with storm events (Hagedorn 1980). Regrowth of faecal coliforms and faecal streptococci in sludge-amended soil has been previously associated with increasing soil moisture after rainfall events (Pepper et al. 1993). An association between rainfall and apparent environmental regrowth of enteric bacteria in biosolids-amended soil was also found by (Gibbs et al. 1995) after a 25-week dry period. In their study, there was an association between moisture content of the dewatered biosolids and regrowth when moisture content rose following low to moderate rainfall of 2-14mm/week between 26 and 35 weeks after soil application. They showed that a single high Salmonella recovery at week 36 and a single high faecal coliform recovery at week 29 occurred where the moisture content of biosolids rose from 1% (weeks 4-21) to 17% (week 29) and then fell to 4% (week 36) and 1% (week 37).

E. coli, C. perfringens, and Salmonella spp. were once considered unable to grow in the environment, and to proliferate naturally only within an animal or human host. Several studies have now shown that enteric organisms can multiply in the environment under certain conditions, both in stockpiled biosolids and in soil-incorporated biosolids. Early studies on raw sewage sludge subjected to evaporation at 21[degrees]C from 1-cm beds gave complex findings, where regrowth of enteric bacteria was not found in dry raw sludge ([greater than or equal to] 85% solids) (Yeager and Ward 1981), yet survival was high at both high and low moisture contents of 5 or 95%, but not at intermediate levels (Ward et al. 1981; Yeager and Ward 1981). Edmonds (1976) reported limited evidence of regrowth of faecal coliforms in dewatered biosolids applied in a forestry application, but had few samplings on which to base any reliable conclusions, and bacterial concentrations were quite low, unlike the current study. However, regrowth has been reported to be a potential problem with composted sewage sludge (Russ and Yanko 1981; Burge et al. 1987; Soares et al. 1995) where the indigenous microflora has been reduced and cannot suppress organisms such as salmonellae (Sidhu et al. 2001). Soares et al. (1995) identified key factors for regrowth as moisture, carbon availability, and microbial diversity. In previously composted sludge, inoculated salmonellae can grow readily at 20-40[degrees]C if the moisture content is > 20% and carbon exceeds nitrogen by 15:1(Russ and Yanko 1981). In contrast, low moisture content and temperatures outside this range do not support growth, although survival for at least 3 weeks at high inoculation rates will occur (Russ and Yanko 1981). Similarly, Burge et al. (1987) reported that Salmonella growth in composted biosolids required a moisture content > 20%. Stored dewatered biosolids with a constant moisture content of 70-80% can also support repopulation (regrowth) by faecal coliforms and salmonellae in association with rainfall after a hot, dry summer, when these bacteria reach undetectable levels (Gibbs et al. 1997).

An alternate explanation of the apparent regrowth in this trial is that it represents an artefact attributable to bacterial survival in a central core of the biosolids clump that only become exposed after weathering some months later. While our sampling method did break open and reduce the size of large biosolid clumps for culture, it is possible that the outer segments of such (larger) clumps contained fewer bacteria due to desiccation. However, we are not aware of any direct evidence to support this possibility in dewatered biosolids.

To investigate a correlation between bacterial regrowth and rainfall suggested by others (Hagedorn 1980; Pepper et al. 1993; Gibbs et al. 1997), data for Goulburn 2 and 3 were also assessed biometrically for an association between prior rainfall and bacterial counts, by 2 methods. Firstly, deviations away from the model described by the smoothing spline analyses were examined for any association with prior rainfall. Secondly, in a model ignoring curvature, deviations about the linear trend were examined for an association with prior rainfall. Neither model for Goulburn 2 or Goulburn 3 indicated a significant association between bacterial counts and rainfall. It may be that the response was due to moisture in the biosolids mass that did not correlate with rainfall, that the rainfall data failed to capture aspects of the local microclimate at the sampling site, or that additional or alternative factors (e.g. nutrient factors, nutrient supply, or other microorganisms) may be involved in regulating bacterial regrowth. Further clarification would require studies of bacterial populations in biosolids under controlled environment conditions, and correlation with additional variates such as moisture content, nutrient availability, and competition from other bacterial species.

Long-term survival and regrowth of E. coli and Salmonella spp. in biosolids may have also been influenced by populations of environmental protozoa that can harbour or influence the growth of these bacteria (Barker and Brown 1994). Laboratory studies have shown that E. coli and Salmonella enterica serovar Typhimurium have enhanced survival under adverse environmental conditions in co-culture with amoeboid or ciliated protozoa (Acanthamoeba spp., Tetrahymena spp.) (King et al. 1988). Further laboratory studies have confirmed that E. coli multiplies in co-culture with Acanthamoeba spp. (Steinert et al. 1998; Barker et al. 1999) and can do so within membrane-bound vacuoles of the protozoa at 4 and 25[degrees]C (Barker et al. 1999). If such organisms can provide both nutrients and protection for vegetative bacteria in biosolids clumps, a complex interaction of environmental physical and biological factors may exist to influence their survival and regrowth at certain times after land application. Further studies to investigate the effect of populations of free-living protozoa on the survival and regrowth of E. coli and Salmonella spp. in biosolids many weeks after land application are warranted.

Salmonella serotyping has been previously used to study sewage sludge and its risk to animal populations. Linklater et al. (1985) described 317 isolates from sewage sludge representing 34 different serotypes, only 5 of which occurred in animals. Their study confirmed the presence of S. enterica serovar Typhimurium of the same phage type as that recovered from incidents on 11 farms, 4 of which had received sewage sludge. These workers concluded that human populations are likely to carry a higher Salmonella background than farm animal populations. In our study, 10 serovars in 66 isolates were detected in dewatered biosolids, with serovars Tennessee and Ohio the most predominant. Notably, serovar Typhimurium was absent. Data from the Australian human population during the time of this study, with 7000-8000 human Salmonella isolates typed annually, indicates that Typhimurium is the major serovar causing human salmonellosis (36-49% of isolates), with a range of other serovars including Virchow, Saintpaul, and Birkenhead being next most prevalent (3-10% of isolates) (Anon. 1998, 1999a, 2000b). Among the Salmonella serovars detected in this study, some were typical of the human population of Sydney (including serovars Liverpool and Singapore), others were among those most often associated with animal feedstuffs (serovars Cubana, Havana, Ohio, Senftenberg, and Tennessee), and 2 serovars (Kentucky and Rissen) were strongly associated with isolations from humans who have travelled overseas (Anon. 1998, 1999a, 2000b). The 10 serovars detected in biosolids in this study account for only 2.9-3.6% of isolates typed from Australian humans annually, and none were among the ten most prevalent serovars (Anon. 1998, 1999a, 2000b).

The most important Salmonella serovars isolated from Australian cattle and sheep are Typhimurium, Bovismorbificans, and Dublin, which represent > 80% of salmonella serovars identified in these species during the time of this trial (1997-99) (Anon. 1999b, 2000a), and still represent the major serovars affecting grazing ruminants between 2000 and 2004 (Anon. 1999b, 2004). The absence of these important serovars associated with grazing animal salmonellosis indicates the direct risk of dewatered biosolids to animal health can be relatively low.

From the data presented, Salmonella spp. may be perceived as a risk factor in dewatered biosolids over an extended period. For humans, the bacteriological risk from biosolids is influenced by soil intake and age, the latter affecting the minimum infectious dose. There remains disagreement on what constitutes an infectious human dose for salmonellosis--varying from [10.sup.1] to [10.sup.5] organisms (Skanavis and Yanko 1994), and lowest when ingested salmonellae are protected against gastric acid by accompanying dietary components (Blaser and Newman 1982; Waterman and Small 1998). Modelling of the risk of salmonellosis from ingestion of soil amendments based on composted sewage sludge (containing < 4 x [10.sup.2] salmonellae/g) has been undertaken previously (Skanavis and Yanko 1994). Based on assumed soil intakes of 5 g/day for children with pica, 0.5 g/day for normal children, and up to 0.1 g/day for adults, that model predicted children may be at risk if relatively low infectious doses (10-100 Salmonella spp. organisms) occur. In contrast, it identified a detectable risk for such products only if adults were susceptible to a very low dose (10 organisms) (Skanavis and Yanko 1994). Based on our results for dewatered biosolids up to 6 months after land application, where [10.sup.4] sahnonellae/g were commonly present in biosolids, similar intakes of biosolids would equate to doses 50-100-fold higher than this. Thus, if infection was possible with up to [10.sup.3] salmonellae for adults, or up to 5 x [10.sup.3] for normal children, then a risk may be present from dewatered biosolids, and would be greatest < 6 months after land application.

However, factors other than bacterial concentrations in biosolids clumps, including dilution, risk of exposure, and strain pathogenicity, contribute to the real risk of human or animal infection. The model described is relevant to humans with direct access to biosolids products, such as bagged soil conditioners (Skanavis and Yanko 1994). In biosolids applied to agricultural land, there are significant dilution effects with soil, and the likelihood of direct exposure to incorporated biosolids is low, so intakes of soil amended with dewatered biosolids are likely to contain very low doses of salmonellae. In addition, many of the serotypes of Salmonella in biosolids are likely to have higher infectious doses than the common pathogenic serotypes (Glynn and Bradley 1992). Animal infection would similarly be markedly reduced by soil incorporation. Considering the time taken to establish pastures after land treatment with dewatered biosolids, a delay of several months after soil incorporation of biosolids before access by grazing livestock would equate to a low risk of uptake. The low concentration of the most important serovars responsible for animal and human salmonellosis is also likely to reduce the risk of carriage of such serovars to the human food chain. Although the risk of animal exposure and uptake of salmonellae may be low for incorporated biosolids, further studies to confirm this would be beneficial.

The results of this study confirm that a risk of infection would remain with exposure of animals to dewatered biosolids for many months, but particularly when access to biosolids clumps is unimpeded, as in surface applications in agricultural or forestry environments. For surface-applied applications, restricted access for 12 months would reduce the risk of wildlife infection, but may have limited practicality. The ability of native animals to harbour or become infected with salmonellae derived from surface-applied biosolids is worthy of further investigation.

Acknowledgments

This work was financially supported by the Sydney Water Corporation. The technical support of Kiley Seymour and Jocelyn Gonsalves during sample collection and bacteriological procedures is gratefully acknowledged. Dr Idris Barchia provided valuable assistance in the presentation of biometrical data.

References

Anon. (1998) 'Human Annual Report 1997.' p. 15. (National Enteric Pathogens Surveillance Scheme, Australia)

Anon. (1999a) 'Human Annual Report 1998? p. 15. (National Enteric Pathogens Surveillance Scheme, Australia)

Anon. (1999b) 'Non-human Annual Report 1998.' p. 14. (National Enteric Pathogens Surveillance Scheme, Australia)

Anon. (2000a) 'Annual Report 2000, Australian Salmonella Reference Centre.' p. 8. (Institute of Medical and Veterinary Science, Infectious Diseases Laboratories: Adelaide, S. Aust.)

Anon. (2000b) 'Human Annual Report 1999.' p. 13. (National Enteric Pathogens Surveillance Scheme, Australia)

Anon. (2004) 'Annual Report 2004.' Australian Salmonella Reference Centre, p. 7. (institute of Medical and Veterinary Science, Infectious Diseases Laboratories: Adelaide, S. Aust.)

Barker J, Brown MRW (1994) Trojan horses of the microbial world: protozoa and the survival of bacterial pathogens in the environment. Microbiology 140, 1253-1259.

Barker J, Humphrey TJ, Brown MRW (1999) Survival of Escherichia coil 0157 in a soil protozoan: implications for disease. FEMS Microbiology Letters 173, 291-295. doi: 10.1111/j.15746968.1999.tb13516.x

Blaser M J, Newman LS (1982) A review of human saimoneliosis: I. Infective dose. Reviews of Infectious Diseases 4, 1096-1106.

Burge WD, Enkiri NK, Hussong D (1987) Salmonella regrowth in compost as influenced by substrate. Microbial Ecology 14, 243-253. doi: 10.1007/BF02012944

Charman PEV (1991) Appendix II Identification of soil groups. In 'Soils--their properties and management', p. 329. (Sydney University Press/Oxford University Press: South Melbourne, Vic.)

Eamens GJ, Lavis AM, Ross AD (1996) Survival of pathogenic and indicator bacteria in biosolids applied to agricultural land. In 'Biosolids Research in NSW, Proceedings of the Biosolids Summit'. (Eds GJ Osborne, RL Parkin, DL Michalk, AM Grieve) pp. 129-138. (NSW Agriculture, Organic Waste Recycling Unit: Rydalmere, NSW)

Edmonds RL (1976) Survival of coliform bacteria in sewage sludge applied to a forest clearcut and potential movement into groundwater. Applied and Environmental Microbiology 32, 537-546.

Gibbs RA, Hu CJ, Ho GE, Philips PA, Unkovich I (1995) Die-off of human pathogens in sludge amended soil. In 'Proceedings of the Australian Water and Wastewater Association 16th Federal Convention'. pp. 667-674.

Gibbs R.A, Hu CJ, Ho GE, Unkovich I (1997) Regrowth of faecal coliforms and salmonellae in stored biosolids and soil amended with biosolids. Water Science and Technology 35, 269-275. doi: 10.1016/S0273-1223(97)00271-0

Gilmour AR, Cullis BR, Welham SJ, Thompson R (1999) ASReml Reference Manual. NSW Agriculture Biometric Bulletin No. 3, NSW Agriculture, Orange, NSW.

Glynn JR, Bradley DJ (1992) The relationship between infecting dose and severity of disease in reported outbreaks of Salmonella infections. Epidemiology and Infection 109, 371-388.

Hagedorn C (1980) 'Potential health hazard associated with the disposal of sewage sludge on agricultural soils in western Oregon.' (Water Resource Research Institute, Oregon State University: Corvallis, OR)

Hess E, Breer C (1975) Epidemiology of salmonellae and fertilizing of grassland with sewage sludge, (author's transl.). Zentralblatt fur Bakteriology [Orig B] 161, 54-60.

Hu CJ, Gibbs RA, Ho GE, Philips P, Unkovich I (1995) Pathogen densities in wastewater sludge treated by anaerobic digestion and dewatering. In '3rd International Conference on Appropriate Waste Management Technologies for Developing Countries'. pp. 23-31.

Ibiebele DD, Inyang AD, Lawrence CH, Coleman RL, Pees N (1985) Some characteristics of the behaviour of indicator bacteria in sewage-amended soil. Environmental Pollution Series A 39, 175-182. doi: 10.1016/0143-1471(85)90015-7

Jepsen SE, Krause M, Gruttner H (1997) Reduction of fecal Streptococcus and Salmonella by selected treatment methods for sludge and organic waste. Water Science and Technology 36, 203-210. doi: 10.1016/S0273-1223(97)00684-7

Jones K, Betaieb M, Telford DR (1990) Seasonal variation of thermophilic campylobacters in sewage sludge. Journal of Applied Bacteriology 69, 185-189.

Jones PW, Rennison LM, Lewin VH, Redhead DL (1980) The occurrence and significance to animal health of salmonellas in sewage and sewage sludges. Journal of Hygiene 84, 47-62.

King CH, Shotts EBJ, Wooley RE, Porter KG (1988) Survival of coliforms and bacterial pathogens within protozoa during chlorination. Applied and Environmental Microbiology 54, 3023-3033.

Linklater KA, Graham MM, Sharp JC (1985) Salmonellae in sewage sludge and abattoir effluent in south-east Scotland. Journal of Hygiene 94, 301-307.

Michalk DL, Curtis IH, Langford CM, Simpson PC, Seaman JT (1996) Effects of sewage sludge on pasture production and sheep performance. In 'Proceedings 8th Australian Agronomy Conference'. pp. 439-432. (Australian Society of Agronomy)

Northcote KH (1974) Schema for Divisions and Subdivisions. In 'A factual key for the recognition of Australian soils', p. 34. (Rellim Technical Publications: Glenside, S. Aust.)

NSW Environment Protection Authority (1997) Section 3.4-3.6. In 'Environmental guidelines: Use and disposal of biosolids products', pp. 11-18. (NSW EPA: Chatswood, NSW)

Patterson HD, Thompson R (1971) Recovery of interblock information when block sizes are unequal. Biometrika 58, 545-554. doi: 10.2307/2334389

Pepper I, Josephson K, Bailey R, Burr M, Gerba C (1993) Survival of indicator organisms in Sonoran Desert soil amended with sewage sludge. Journal of Environmental Science and Health, Part A: Environmental Science and Engineering 28, 1287-1302.

Pillai SD, Widmer KW, Dowd SE, Ricke SC (1996) Occurrence of airborne bacteria and pathogen indicators during land application of sewage sludge. Applied and Environmental Microbiology 62, 296-299.

Robinson MB, Poiglase PJ, Weston CJ (2002) Loss of mass and nitrogen from biosolids applied to a pine plantation. Australian Journal of Soil Research 40, 1027-1039. doi: 10.1071/SR01088

Russ CF, Yanko WA (1981) Factors affecting salmonellae repopulation in composted sludges. Applied and Environmental Microbiology 41, 597-602.

Sidhu J, Gibbs R, Ho G, Unkovich I (2001) The role of indigenous microorganisms in suppression of Salmonella regrowth in composted biosolids. Water Research 35, 913-920. doi: 10.1016/S0043-1354(00)00352-3

Skanavis C, Yanko WA (1994) Evaluation of composted sewage sludge based soil amendments for potential risks of salmonellosis. Journal of Environmental Health 56, 19-23.

Soares H, Cardenas B, Weir D, Switzenbaum M (1995) Evaluating pathogen regrowth in biosolids compost. BioCycle 36, 70-76.

St. John WD, Matches JR, Wekell MM (1982) Use of iron milk medium for enumeration of Clostridium perfringens. Journal--Association of Official Analytical Chemists 65, 1129-1133.

Standards Association of Australia (1987) Methods for the microbiological examination of food. Part 2--Examination for Specific Organisms. In 'Australian Standard 1766.2.3 Coliforms and Escherichia coli', pp. 1-3.

Standards Australia (1991a) Method 2.5. Examination for specific organisms--Salmonellae. In 'Australian Standard 1766.2.5. Food microbiology', pp. 1-8.

Standards Australia (1991b) Method 2.8. Examination for specific organisms--Clostridium perfringens. In 'Australian Standard 1766.2.8 Food microbiology', pp. 1-6.

Steinert M, Birkness K, White E, Fields B, Quinn F (1998) Mycobacterium avium bacilli grow saprozoically in coculture with Acanthamoeba polyphaga and survive within cyst walls. Applied and Environmental Microbiology 64, 2256-2261.

Stelzer W, Jacob J (1991) A study of Campylobacter in sewage, sewage sludge and in river water. Water Science and Technology 24, 117-120.

Surampalli RY, Banerji SK, Chen JC (1994) Microbiological stability of wastewater sludges from activated sludge systems. Bioresource Technology 49, 203-207. doi: 10.1016/0960-8524(94)90040-X

Verbyla AP, Cullis BR, Kenward MG, Welham SJ (1999) The analysis of designed experiments and longitudinal data by using smoothing splines (with discussion). Applied Statistics 48, 269-311.

Ward RL, Yeager JG, Ashley CS (1981) Response of bacteria in wasterwater sludge to moisture loss by evaporation and effect of moisture content on bacterial inactivation by ionizing radiation. Applied and Environmental Microbiology 41, 1123-1127.

Waterman SR, Small PL (1998) Acid-sensitive enteric pathogens are protected from killing under extremely acidic conditions of pH 2.5 when they are inoculated onto certain solid food sources. Applied and Environmental Microbiology 64, 3882-3886.

Yanko WA, Glass JS, Van Siuis RJ, Dahlgren JA, Easly RC (1978) Survival of pathogenic and indicator microorganisms in sewage sludge amended agricultural soils--winter and summer growing seasons, 1977. County Sanitation Districts of Los Angeles County, San Jose Creek Water Quality Lab, Whittier, CA.

Yeager JG, Ward RL (1981) Effects of moisture content on long-term survival and regrowth of bacteria in wastewater sludge. Applied and Environmental Microbiology 41, 1117-1122.

Manuscript received 2 February 2006, accepted 16 August 2006

G.J. Eamens (A,D), A.M. Waldron (A,B), and P.J. Nicholls (A,C)

(A) NSW Department of Primary Industries, Elizabeth Macarthur Agricultural Institute, PMB 8, Camden, NSW 2570, Australia.

(B) Present address: University of Sydney, PMB 3, Camden, NSW 2570, Australia.

(C) Present address: PO Box 20, Menangle, NSW 2568, Australia.

(D) Corresponding author. Email: graeme.eamens@dpi.nsw.gov.au
Table 1. Biosolids treatments, application times,
and plot designations (A-D)

Site           Time of      Total time
             application    examined
                            (weeks)

Goulburn 1      Summer          51
Paddock 12      Summer          51
Paddock 17      Summer          51
Goulburn 2      Spring          68
Goulburn 3      Summer          52

             Plot designation and treatment

                    10 dry t/ha

                             Surface-
Site         Incorporated    applied

Goulburn 1        A1            B1
Paddock 12
Paddock 17
Goulburn 2        A2            B2
Goulburn 3        A3            B3

             Plot designation and treatment

                      30 dry t/ha

                             Surface-
Site         Incorporated    applied

Goulburn 1        C1            D1
Paddock 12                      El
Paddock 17                      F1
Goulburn 2        C2            D2
Goulburn 3        C3            D3

Table 2. Statistical evaluation of E. coli and Salmonella spp.
survival curves relevant to linear, single spline and double
spline models of the data

Significant P values indicate a significant agreement between a
linear, single-spline, or double-spline trend and the raw data.
The significance of increases in mean bacterial concentrations
over certain time intervals (refer Figs 2, 3, 5 and b) are also
evaluated

                                        P value

Site         Organism          Biosolids      Linear
                                incorp.

Goulburn 1   E. coli              Yes         <0.001
                                  No          <0.001
             Salmonella spp.      Yes         <0.001
                                  No          <0.001
Goulburn 2   E. coli              Yes         <0.001
                                  No          <0.001
             Salmonella spp.      Yes          n.s.
                                  No           n.s.
Goulburn 3   E. coli              Yes         <0.001
                                  No          <0.001
             Salmonella spp.      Yes          n.s.
                                  No           n.s.

                                        P value

                                             Incr. in
                                           bact. conc.
                                            in single
Site         Organism           Single     spline (weeks
                                spline         (A))

Goulburn 1   E. coli             <0.01         n.a.
                                 <0.01         n.a.
             Salmonella spp.     <0.05         n.a.
                                 <0.05         n.a.
Goulburn 2   E. coli             n.s.          n.a.
                                 n.s.          n.a.
             Salmonella spp.     n.s.      <0.05 (10-39)
                                 n.s.          n.s.
Goulburn 3   E. coli             n.s.          n.s.
                                 n.s.          n.s.
             Salmonella spp.     <0.05         n.s.
                                 <0.05         n.s.

                                        P value

                                             Incr. in
                                           bact. conc.
                                            in double
Site         Organism           Double     spline (weeks
                                spline         (A))

Goulburn 1   E. coli             n.s.          n.a.
                                 n.s.          n.a.
             Salmonella spp.     n.s.          n.a.
                                 n.s.          n.a.
Goulburn 2   E. coli             n.s.          n.a.
                                 n.s.          n.a.
             Salmonella spp.    <0.05      <0.01 (10-39)
                                 n.s.          n.a.
Goulburn 3   E. coli            <0.001     <0.01 (19-25)
                                <0.001     <0.01 (19-25)
             Salmonella spp.    <0.01          n.s.
                                <0.01          n.s.

n.a., Not applicable; n.s., not significant (P > 0.05).

(A) Range in weeks after application where increase in
bacterial concentration was significant.

Table 3. Salmonella serovars among 66 isolates from dewatered
biosolids derived from Cronulla WWTP applied on plots at an
agricultural site at different times (Goulburn 2, Goulburn 3)

Serovar       Goulburn 2   Goulburn 3   Total

Cerro                           2         2
Cubana                          1         1
Havana            4                       4
Kentucky          1                       1
Liverpool                       3         3
Ohio                           19        19
Rissen                          4         4
Senftenberg       3                       3
Singapore                       5         5
Tennessee         12           12        24
Total             20           46        66
Gale Copyright: Copyright 2006 Gale, Cengage Learning. All rights reserved.