Cereal crops grown in the biosolids-amended soil can potentially become contaminated with pathogenic micro-organisms during growth and at the time of harvesting. There is small but unquantified potential risk of transfer of enteric pathogens to humans and animals from contaminated plants and grains. This study examined decay of Escherichia coli, Salmonella enterica serovar Typhimurium and bacteriophage MS2 on the wheat phyllosphere and on stored grains. This was done to assess the health implications of cereal crops contaminated from land application of biosolids. E. coli, S. enterica and MS2 were inoculated onto the leaves, spikelets and grains of wheat. The change in the numbers of inoculated micro-organisms was determined over time to calculate the respective 90% reduction time (T90) in each of these environments. A rapid inactivation (T90 <1–3 days) of E. coli and S. enterica and MS2 from the plant phyllosphere was observed, particularly from the spikelets. The decay rates were influenced by micro-organism type and location on the plant phyllosphere. Decay times on the stored grains were longer (T90 9–71 days), with some observed influence of grain variety on pathogen decay times.
Significance and Impact of the Study
Results of this study suggest that there is very limited potential of enteric pathogens survival on wheat phyllosphere and grains. Therefore, the risk of transfer of enteric pathogens from biosolids-amended soil to consumers of grain products is considered to be low. This study has important implications for the grains industry, as the results suggest that chances of preharvest contamination of grains with enteric pathogens from biosolids-amended soil are low.
Biosolids are a valuable resource that can be used sustainably as a fertilizer and a soil conditioner to improve the chemical and physical properties of soil (Sidhu et al. 2001; Zaleski et al. 2005). In Australia, over 360 000 dry tonnes of biosolids are produced annually, and the majority is applied to land in broad-acre agriculture (Gale 2007). However, biosolids are known to contain a diverse range of human pathogens such as adenovirus, polyomavirus, Salmonella enterica, Cryptosporidium and Giardia (Sidhu and Toze 2009). Pathogens such as S. enterica and enteric viruses may be dispersed from the biosolids-amended soil onto plant leaves by rain splash and wind which may then persist on the plant phyllosphere (Boyer 2008). There is a concern that enteric pathogens could then be transferred onto the grain heads of cereal crops and thus be transmitted to humans. This unknown health risk could, in turn, impact the marketing and trade of grain products.
In several studies, the presence of low levels of S. enterica, Bacillus cereus, E. coli, and various other food spoilage micro-organisms have been reported in wheat and flour in Australia and other parts of world (Eyles et al. 1989; Richter et al. 1993; Berghofer et al. 2003; Sperber 2007). The source of contamination has been reported to be due to both pre- and postharvest contamination. Most of the cereals are consumed after processing and are expected to be pathogen-free due to processing. However, there is a small residual risk if minimally processed foods containing cereals contaminated with pathogens are consumed (Berghofer et al. 2003). Regrowth of surviving bacterial pathogens is also possible following hydration with water or milk as regrowth of Salmonella surviving in low numbers on dry cereal flakes after addition of milk have been reported (Ui et al. 2009). Salmonellosis outbreaks have been linked to the consumption of cereal-based products (Zhang et al. 2007; Neil et al. 2012; McCallum et al. 2013). Such foods include uncooked breakfast cereals (e.g. muesli), infant foods containing cereals, or home-baking products such as cookie dough. Ingestion of raw flour may also occur while baking cakes and cookies at home.
In broad-acre cereal crop production, the risk to consumers from contaminated plant and grains is not fully understood, because very little research has been conducted on the quantification and survival potential of human enteric pathogens on the phyllosphere of plants (Ibekwe et al. 2004). Most of the available information on the survival potential of enteric pathogens and potential health risks come from the fresh fruit and vegetable industry, in particular from animal manure and treated effluent used as irrigation water (Beuchat 1996; Doyle 2000a,b; Buck et al. 2003; Johannessen et al. 2005). Contaminated soil, animal manure, compost and irrigation water have been shown to be responsible for the contamination of salad and vegetable crops (Solomon et al. 2002; Ibekwe et al. 2004, 2009; Islam et al. 2004c; Ibenyassine et al. 2006). Enteric pathogens such as Campylobacter jejuni, E. coli O157:H7 and S. enterica have been reported to be able to attach to plant surfaces and metabolize available nutrients in plant exudates (Brandl et al. 2004; Warriner and Namvar 2010). It has been postulated that enteric pathogens survive wastewater treatment process in low numbers (Sidhu and Toze 2009) and then, if introduced to soil at the time of biosolids application, subsequent contamination onto plant leaves may occur during the rainfall events (Brown et al. 1980; Boyer 2008). If this was to occur in food grain crops, contaminated grains and fodder could potentially transmit diseases to humans and livestock at consumption.
The plant phyllosphere, in particular cereal crops with longer growing times, present a hostile environment for bacterial pathogens to survive (Brandl 2006). The cereal phyllosphere is subjected to a rapid and large fluctuations in temperature, humidity and osmotic pressures (Wilson et al. 1999) which may adversely affect the survival of enteric pathogens (Cox 1993; Casanova et al. 2010). Environmental factors, such as ultraviolet (UV) radiation and desiccation, have been identified as important factors that influence pathogen survival on the phyllosphere (Heaton and Jones 2008). Competition for limited nutrients and moisture also makes enteric pathogen survival more difficult (Mercier and Lindow 2000). Plant-specific factors, such as waxes, may also restrict bacterial attachment to leaf surfaces (Aruscavage et al. 2006). However, pathogens may survive better under certain conditions such as in the shade or under increased moisture content, usually between the leaves and stems of the plants (Brown et al. 1980; Lindow and Brandl 2003; Ibekwe et al. 2004). Once deposited in the phyllosphere through rain splash, pathogens could also migrate into the biofilms established by autochthonous microflora which is reported to shield them from desiccation and UV (Elasri and Miller 1999; Fett 2000; Monier and Lindow 2005). However, it is unclear how these factors may influence the persistence of enteric pathogens introduced onto cereal crops.
The microbiological quality of cereal grains is considered to have an impact on the quality of the end product (Berghofer et al. 2003), and many processors monitor the microbial quality of the raw grains. Therefore, over time, the responsibility of consumer safety has gradually shifted from the consumer across the food supply chain back to the producer. Thus, from a marketing perspective, it is important to assess the biosafety of cereal crops grown in biosolids-applied land (Chaney et al. 1996; Tauxe 2002). There is currently very limited experimentally derived information that quantifies the extent of potential health risks from cereal crops grown in biosolids-amended soils.
The purpose of the present study was to assess the potential health risks from enteric pathogens that may survive on the plant phyllosphere during production of cereal crops in biosolids-amended soils. To examine this, the decay times of enteric micro-organisms (E. coli, S. enterica and bacteriophage MS2) inoculated on the wheat phyllosphere and grains were assessed in a glasshouse study.
Results and discussion
Decay of seeded micro-organisms on the plant phyllosphere
The inoculated E. coli, S. enterica and MS2 were observed to decay rapidly (a decrease in ~3–4 log10), particularly during the first 3 days on the leaves and spikelets of the wheat plants (Fig. 1). This was followed by slower inactivation during the rest of the study period (8 days or 192 h). Comparable rapid decay of E. coli inoculated on spinach leaves under ambient conditions has been reported (Wood et al. 2010). The anova results (Table 1) show that over the course of experiment, the overall decay rates of individual micro-organisms across two treatments (leaves vs spikelets) were significantly different from zero (represented by sdate, sdate2 and sdate3, respectively, for the first, second and third orders of reduction, P < 0·0001).
|Source of variation||Escherichia coli||Salmonella enterica||MS2|
|Mean square||P-value||Mean square||P-value||Mean square||P-value|
|Treatment (leaves vs spikelets)||0·32||0·6197||2·39||0·3858||5·88||0·092|
|Sdate*treatment (leaves vs spikelets)||0·75||0·446||3·97||0·2644||9·27||0·0348|
|Sdate2*treatment (leaves vs spikelets)||1·56||0·2708||6·68||0·1485||14·67||0·0081|
|Sdate3*treatment (leaves vs spikelets)||1·69||0·2525||6·84||0·1438||15·06||0·0074|
When the decay rate of E. coli and S. enterica on the spikelets was compared with the rate of the same micro-organism on the leaves, there was no significant difference between the decay on spikelets and leaves. This was clearly demonstrated by the nonsignificant interaction values in the anova results (Table 1, represented by sdate*treatment, sdate2*treatmen and sdate3*treatment, P > 0·05). The results indicate that the pathogens on the spikelets and leaves followed the same decay rate. The estimated average one log10 reduction time for micro-organism on leaves, spikelets and grains is presented in Table 2. The decay times (T90) of bacteria was 1–2 days on the spikelets and slightly longer (2–3 days) on the leaves. In contrast, much longer decay times for E. coli O157:H7 have been reported on lettuce (15–77 days), spinach (7–14 days), ryegrass (41 days) and grassland (99 days) in the literature (Sjogren 1995; Beuchat 1999; Bolton et al. 1999; Solomon et al. 2002, 2003; Islam et al. 2004a,c; Ibekwe et al. 2009). Longer decay times for S. enterica compared with E. coli on lettuce (63 days) and parsley (161 days) have also been reported (Islam et al. 2004b). The longer persistence of bacteria on plant surfaces such as parsley and lettuce may be attributed to crop-specific factors such as crop density and complex leaf structure (Islam et al. 2004a). Normally, complex leaf structures like the spikelets provide microclimatic factors such as reduced desiccation and shading which prolong the survival of enteric pathogens, particularly on crops such as alfalfa (El Hamouri et al. 1996). In contrast, a faster decay rate of pathogens would be expected on plants with flat wide leaves and simple leaf structures such as cereal crops due to a greater exposure to direct sunlight (UV) and desiccation (Brandl 2006). Wheat spikelets are shielded from rainfall or irrigation (fine mist) by plant physiological factors such as the awns (course hair-like protrusions), the angle of the spikelet (upright or downward) and the kernel density (Pekkarinen 2003). In the present study, the decay rate for both bacteria on the leaf and spikelet was statistically non significant (Table 1, Sdate*treatment, P > 0·05) which suggests that prolonged survival of bacteria on shaded/protected areas (leaves or spikelets) which are not exposed to direct sun light is unlikely.
|Micro-organism||T90 times (days)|
|Leaves||Spikelets||Noodle||Australian Soft White wheat|
MS2 showed faster decay compared to the bacteria with T90 of <1 day (Table 2); however, unlike the bacteria, there was a significantly different decay rate of MS2 (on the spikelets compared with the leaves (P values for sdate*treatment, sdate2*treatment and sdate3*treatment for MS2 in Table 1, P < 0·01). The precise reason for this is unclear; however, it is possible that lower recovery of MS2 due to irreversibly bound phage particle onto the spikelets might have contributed to the observed rapid decay. A similar inactivation rate of MS2 (T90 = <1 day) inoculated onto grass surfaces under ambient conditions has been reported previously (Sidhu et al. 2008). The results of this study are also similar to Choi et al. (2004) study where T90 of 1–2 day were reported for bacteriophage (MS2 and PRD1) on the surface of lettuce. The shorter decay times observed in the present study are most likely due to environmental stress caused by high UV radiation and desiccation under direct sunlight over the duration of the experiment.
Decay patterns of bacteria and bacteriophage on stored grains
The decay times (T90) of enteric bacteria on both grain varieties were <12 days (Fig. 2, Table 2) which were much shorter than MS2 (60–71 days). The anova results for the grain experiment (Table 1) show that the overall decay rates of all three micro-organisms across the experiment course were significantly different from zero (represented by sdate in Table 1, P < 0·0001). The longer decay time of MS2 on the stored grains could potentially be due to the absence of direct sunlight (UV) as well as no loss of moisture inside the tins (due to condensation) over the duration of the experiment because decay of bacteriophage is known to be influenced by UV radiation intensity and moisture content (Ehrlich et al. 1964; Iriarte et al. 2007). Bacteriophage MS2 and PRD1 have been reported to be more persistent in humid conditions as opposed to dry conditions (Choi et al. 2004). In this study, the moisture content of the stored grains remained almost constant with a small increase from 9·5% prior to inoculation to 11·6% toward the end of study period inside the tins. Higher initial moisture content and higher outer temperature have been known to result in an increase in moisture content in stored grains (Kusinska 2006). This suggests that maintenance of constant moisture might have prolonged survival of MS2 on the stored grains.
The decay times of enteric pathogens on the grains were also influenced by grain variety. When two grain varieties [Australian Soft White (ASW) and NN] were compared, there was a significant difference between them not only in pathogen count but also in the decay rate of E. coli or S. enterica (P values for Sdate*variety and variety in Table 1, P < 0·001). However, the opposite was true for MS2 (P > 0·05). The enteric bacteria decay times were shorter on the noodle (NN) grains (T90 < 10 days) compared with the soft wheat (ASW) grains (T90 > 10 days). Likewise, the MS2 decay times were also shorter on the NN grains (T90 = 60 days) compared with the ASW grains (T90 = 71 days). This means that the enteric micro-organisms decayed faster on the pasta variety of wheat (NN) compared with the biscuit variety (ASW). The reason for the different decay rates on the varieties remains unclear; however, autochthonous micro-flora such as antibiotic-producing actinomycetes on the grain surface may have influenced the decay times of seeded micro-organisms. Also, the comparative grain properties, such as grain strength or texture, may also have influenced decay via providing altered microclimatic conditions that needs to be further investigated. The observed bacterial decay times (T90 > 10 days) on stored grains suggests that there is minimal potential of bacterial pathogens surviving on the stored grains prior to processing as harvested grains are usually stored for a longer period of time (>3 months) prior to processing and consumption. Similarly, health risks from enteric viruses should also be low following storage of grains as the time to a one log10 reduction (90% reduction) in bacteriophage MS2 was observed to be <71 days.
General survival potential of enteric pathogens in the phyllosphere and on grains
The results of this study have shown that there is a limited survival potential of enteric pathogens on the wheat plants, particularly the spikelets. Further, research work needs to be carried out with other pathogens such as adenovirus and Campylobacter to further validate the findings of this study. It should be noted that in this study, enteric micro-organisms were inoculated onto the wheat phyllosphere at much higher numbers than would be expected to occur naturally through contamination of wheat plants in biosolids-amended soils. A small proportion of pathogens could be transferred from biosolids-amended soil onto the plants through factors such as rain-splash and as wind-borne particles; however, high levels of these pathogens would not be expected to penetrate the chaff casing on the spikelets and thus contaminate the grains. A rapid reduction in enteric bacteria (T90 < 3 days), observed in the present study from the phyllosphere of wheat, suggests that the risks to the consumer are much lower than that for salad and vegetable crops which are often consumed raw. In addition, the growing season for cereal crops is much longer than vegetable and salad crops, and the consumable parts (for wheat) are not grown close to the soil; therefore, the health risks would be expected to be much lower. Furthermore, wheat is ground, processed and cooked prior to consumption and for this reason, it can be surmised that the health risks to consumers are low. However, separate studies may be required to evaluate human health risks from yeast, fungi/micotoxins and other spoilage micro-organisms on the grains that were reported in a previous study (Berghofer et al. 2003).
In conclusion, rapid inactivation of the enteric micro-organisms occurred on the leaves and spikelets of wheat, while longer decay times were observed on the stored grains. The decay times were mostly influenced by micro-organism type, the location on the plant and the grain variety. In general, the risk of transfer of enteric pathogens from the use of biosolids in cereal crop production to the consumers of grain products is considered to be low.
Materials and methods
Two experiments were carried out in glasshouse facilities at the Commonwealth Scientific and Industrial Research Organisation (CSIRO) in Perth, Western Australia (WA). Air temperature in the glasshouse was maintained at 17°C (±0·25) by an air-conditioning unit and relative humidity maintained at 72% (±0·97). Daily air temperature and relative humidity were recorded inside the glasshouse every 20 min using a Tinytag Plus 2 (Gemini Data Loggers Ltd, West Sussex, UK).
The first experiment was a microbe-plant experiment and the second was a microbe-grain experiment. Wheat (Triticum aestivum cv. Calingiri) was selected to represent the cereal crops that are commonly sown in broad acre agriculture following biosolids application. The following micro-organisms were tested: E. coli (ACM 1803) as an indicator of enteric bacteria, Salmonella enterica serotype Typhimurium (ATCC 13311) as a human pathogenic bacteria and bacteriophage MS2 (ATCC 15597-B1) as an enteric virus surrogate.
Microbial inoculum preparation and sampling
Escherichia coli and Salm. enterica inoculants were prepared by culturing each in 100 ml of nutrient broth (Oxoid, UK) on a shaking platform incubator overnight at 37°C. Prior to inoculation, these overnight cultures were washed in phosphate buffer three times as described in Gordon and Toze (2003) and rested in phosphate buffer for 24 h prior to seeding to acclimatize cultures to low-nutrient environment (Sidhu and Toze 2012). The resulting suspensions were determined to have a final cell count of approximately 1 × 108 colony forming units (CFU) ml−1. The male-specific bacteriophage MS2 (ATCC 15597-B1) used in this study was cultured using E. coli host HS (pFamp)R (ATCC 700881) in Tryptone broth in a shaking incubator overnight at 37°C (Sidhu et al. 2008). The MS2 culture was then purified using centrifugation at 3000 g for 10 min to pellet the E. coli cells followed by filtering through a 0·2-μm membrane to remove bacterial cells and then stored at 4°C in phage buffer until required (72 h). The final suspension was determined to have a phage count of more than 1 × 108 plaque forming units (PFU) ml−1. Inoculum for application to wheat plants and grains was then prepared from the bacterial and MS2 cultures by adding 1 ml each of 1 × 108 per ml stock cultures to 100 ml of phosphate buffer.
Experimental design – plant experiment
This experiment was conducted over a 9-day period using two treatments in triplicate; wheat leaves and spikelets (i.e. the grain head). The pathogen counts of the selected micro-organisms were monitored at anthesis or the flowering stage (stage Z65, Zadoks scale (Zadoks et al. 1974). The experiment was undertaken using three pots (TerraBoxes™; Planterra, Detroit, MI, USA) of wheat sown with 10 grains per pot (Fig. 3). Soil was amended with nonsterile biosolids (Beenyup Wastewater Treatment Plant, Perth WA, USA) at a rate equivalent to 10 t DS ha−1 (dry wt/hectare) and incorporated into the topsoil (<10 cm). Each pot contained approximately 10 flowering plants with well-formed grain heads. Inoculum of E. coli, S. enterica and MS2 was applied to the leaves and the spikelets in three applications using a sterile commercial spray bottle in the biohazard cabinet. Approximately, 15 ml culture was applied during each event and 10 min of drying time between applications was allowed for an even distribution of inoculum. A control pot was set up by sparing equivalent volume of sterile water. After the final drying period of 30 min, the pots were moved to a bench in the glasshouse. Plant samples taken from the control pot were tested for the presence of E. coli, S. enterica and MS2 and were found to be not present at time 0, 28 h and 206 h.
Plant samples were collected at hours 0, 1, 2, 4, 6 and 8. Sampling frequency was then reduced to 26, 28, 53, 64, 206 h to a maximum of 9 days. At each sampling event, three representative plant leaves and three spikelet samples (i.e. ~3 g) from each wheat plant were taken using sterile secateurs for each treatment (n = 3 samples per treatment). All samples were placed directly into sterile stomacher Bag Filter® bags (Interscience, Nom la Bretêche, France), placed on ice and processed within 2 h of collection.
Experimental design – grains experiment
The grain experiment was conducted over 63 days using common wheat varieties sourced from the Co-operative Bulk Handling Limited (CBH), Forrestfield, WA, Australia. The two varieties tested were NN, typically used for pasta, and ASW, used for pasta flour and soft dough for biscuits and cakes. One portion (1·5 kg) of each variety seeds were inoculated with E. coli, S. enterica and MS2 inoculums in three applications using a fine-mist atomizer on grains evenly distributed across a tray placed in the biohazard cabinet. Approximately, 5–10 ml of culture was applied during each event. Grains were mixed thoroughly in order to evenly distribute the inoculums and then allowed to air-dry for 30 min between the three applications. Prior to inoculation, grain moisture content was determined to be approximately 10% (Infratec, CBH, Forrestfield, WA, Australia). The control portion (1·5 kg) was sprayed with sterile water to match the moisture content of the inoculated grains and placed in control tin. Seven metal tins (500 g) with lids were used to model grain storage silos (n = 3 tins containing inoculated grains, n = 3 tins containing control grains, n = 1 tin for monitoring grain moisture). Sealed tins were placed in the glasshouse under direct sunlight with the lids on to model grain stored in grain silos.
Grain samples were collected at days 0, 1, 2, 7, 14, 21, 30, 35, 50 and 63. At each sampling event, three representative samples (i.e. ~15–25 g) of grain from each of three tins were taken using a sterile spatula for each treatment (n = 9 samples per treatment). All collected grain samples were placed directly into sterile stomacher Bag Filter® bags (Interscience), placed on ice immediately and then processed within 2 h of collection.
Quantification of Escherichia coli, Salmonella enterica and MS2 from collected plant and wheat samples
The net weights of sample contents in each stomacher bag (Bag Filter®; Interscience) were obtained prior to processing. Phosphate buffer (P-buffer) at pH 7·2 was aseptically added (20 ml to plant samples and 50 ml to grain samples). All sample bags were placed in a stomacher (Bag Mixer®; Interscience) and mixed for 2 min (speed no. 7). The supernatant was transferred into sterile polypropylene tubes and serial 10-fold dilutions were subsequently made in sterile Phosphate buffer (P-buffer). The amount of dilution undertaken for each sample was determined based on the detectable microbial numbers determined at the previous sampling event. The numbers of E. coli, S. enterica and MS2 at each time interval were detected by direct culture on the respective selective media. All analyses for each micro-organism were performed in triplicate.
Escherichia coli and Salm. enterica were detected by spread plating 100 μl of the appropriate serial dilutions onto the selective agar plates. E. coli was detected on Chromocult™ coliform agar (Merck, Darmstadt, Germany) and S. enterica was detected on xylose lysine deoxychlorate agar (BBL) as outlined in Sidhu et al. (2008). Inoculated plates were incubated overnight at 37°C, and then, typical colonies were counted to determine the average number of colony forming units (CFU ml−1). The quantification of F-specific bacteriophage MS2 was carried out by the standard double layer agar method using E. coli HS(pFamp)R (ATCC 700881) as the host bacteria (Havelaar and Hogeboom 1984). Clear plaques were counted to determine the average plaque forming units (PFU ml−1) after overnight incubation at 37°C.
Prior to statistical analysis, pathogen counts were normalized from the raw data by transforming into log10 CFU g−1 using the initial sample weights obtained prior to analysis of each collected simple and the formula log10 total cell Count = log10 [((actual microbial number(CFU/PFU)*10Dilution*10volume plated)*ml phosphate buffer) (1/soil weight) + 1]. All statistical analyses were performed using SAS version 9.1 (SAS Institute Inc., 2005).
As the micro-organisms inoculated on the grain samples were observed to have a first order reduction, the generalized linear model of analysis of variance (anova) was used. The variation sources included the effect due to grain variety (ASW and NN), linear decay rate due to various sampling dates (sdate), the interaction between sdate and grain variety (sdate*grain variety, indicating a specific decay rate for a variety), and the tin effect nested within the grain variety. The decay time for one-log10 reduction in pathogen count (T90) for each micro-organism in the grain experiment (days) was determined using the equation T90 = 1/b, where b was the derived decay rate from the model.
In contrast, the microbial decay on the plant phyllosphere was observed to be a second order reduction; therefore, the generalized linear log-quadratic model of anova was used for these samples to identify significant variation sources affecting individual micro-organism pathogen counts (log10 Count). The variation sources included the treatment effect (leaves vs spikelets), linear and quadratic terms of sampling date (sdate and sdate2, corresponding to average first and second order decay rates across two treatments), the interactions between sdate and treatment (sdate*treatment and sdate2*treatment, indicating specific linear and quadratic decay rates for a treatment), and the storage tin effect nested within the treatment. The decay time for one log reduction of the virus count (T90 values) was estimated by solving the quadratic equation , where ‘a’ and ‘b’ were the corresponding estimated linear and quadratic inactivation rates for each treatment from the model.
The authors wish to acknowledge the Water Corporation in Perth, CSIRO Water for a Healthy Country Flagship, Water Quality Research Australia (WQRA) Limited and the Victorian Department of Human Services for project funding. Thank you to CBH Limited, Perth WA for providing grain samples and to Mr Sebit Gama (CSIRO Land and Water) for technical assistance.
Conflict of interest
No conflict of interest declared.