A significant majority of the human pathogens associated with irrigation water and fresh produce are enteric (gut-related) bacteria. Enteric means these microbes tend to be carried in the gut of livestock and wildlife and they can spread around the environment as animals deposit manure. Examples of important enteric bacteria which cause human illness are campylobacters, which are associated with poultry broilers and other livestock in the UK; Salmonella, which is often isolated from UK pigs, and toxin-producing E. coli O157, for which cattle are an important reservoir. The only non-enteric (i.e. not exclusively spread from an animal's gastrointestinal tract) human bacterial pathogens of significance to food safety in the UK are some strains of Listeria monocytogenes. Listeria species generally are routinely isolated from UK soils and the wider environment. Although not strictly classed as an enteric, Listeria can also be isolated from the manures of grazing sheep (Hutchison et al. 2004) because they tend to eat grass down to the roots and thus ingest small amounts of soil containing Listeria.
Why is testing to determine pathogen numbers rarely undertaken?
Although human pathogens can be isolated from animal manures, it is important to realise that not all livestock harbour enteric pathogens in their wastes (Hutchison et al., 2004). UK surveillance showed there was about a one in three chance of manure from UK livestock harbouring one of the four most important pathogenic bacteria with potential to cause an outbreak of foodborne illness in humans (Hutchison et al., 2004). Furthermore, even if pathogens are present in livestock manures, they are likely to be found in quite small numbers — often too few to count by conventional microbiological methods. Thus, laboratory tests generally do not attempt to count the numbers of bacterial pathogens in a sample, and normally an enrichment approach is taken. Enrichment involves trying to increase the numbers of pathogens in a sample, by incubation at a temperature and in growth media which promotes the multiplication the pathogen of interest and which inhibits the growth of other bacteria. Enrichment test results for pathogens are reported simply as “Detected” or “Not Detected”, with the vast majority of test results reported as “Not Detected”. For these reasons, many microbiologists do not consider that it is an effective spend of a testing budget to repeatedly test samples for pathogens and to generate a historical collection of lab reports all of which say “Not Detected” because such an approach tells a grower nothing about how their water sources or produce quality changes over time.
Testing for faecal indicator species
An alternative approach to microbiological testing involves the use of faecal indicator bacteria (i. e. general bacteria commonly found in faeces). In contrast to most pathogens, indicator bacteria are chosen because they are regularly isolated from a test sample in sufficient numbers that can be counted. Since most laboratory tests will return a numerical result, indicator numbers can be plotted onto a graph showing microbiological trends. Consequently, and in contrast to pathogen detections (see above), testing for indicator numbers provides growers with useful information describing general levels of faecal contamination in the production of fresh produce, and how that changes over time. A very important point to make clear about indicators is that there are no absolute correlations between the likelihood of a pathogen detection and numbers of indicator bacteria in a sample. Although there is one specific instance of a significant correlation between the presence of pathogens such as Salmonella and E. coli counts (Castro-Ibanez et al., 2015).
However, it is rarely true that high numbers of indicator equate with an increased likelihood of pathogen detection.
Commonly used indicators for water and fresh produce quality
E. coli. E. coli is an excellent indicator for the contamination of water or fresh produce (Allende et al., 2017) with human waste or animal manure (it is important to note that human disease-causing E. coli O157 is a specific subgroup of E. coli which is quite rarely isolated, and that most E. coli are harmless to humans). Since E. coli does not typically survive for extended periods in surface waters, on the surfaces of plants (Hutchison et al., 2008) or in soil (Avery et al., 2004), its presence is associated with a fairly recent contamination event. E. coli numbers in water samples tends to increase after heavy rainfall as livestock manure on hills gets washed into streams and rivers (Gil et al., 2015; Balleste et al. 2020; Lee et al. 2020; Lizaga et al. 2020). In addition, during very heavy rainfall, sewers and waste treatment facilities can become overloaded and overflow, further contaminating surface waters.
The Enterobacteriaceae. The Enterobacteriaceae are a large and diverse collection different species of bacteria. Despite the name, many members of the Enterobacteriaceae group can be isolated from surface waters and soil, as well as human and animal digestive tracts (i. e. the gut). The numbers of Enterobacteriaceae present in a test sample can be thought of as a general indicator of the degree of contamination acquired by fresh produce from contaminated water, insects, faecal material, wildlife, soil and other plants (including some plant pathogens).
Coliforms and Faecal Coliforms. Coliforms are a subgroup of the Enterobacteriaceae. They comprise mostly of the Escherichia, Klebsiella, Citrobacter, Enterobacte and Serratia genera. Membership of the coliform group is determined in a laboratory and includes an ability to ferment lactose sugar (milk sugar) with acidic and gaseous byproducts at 35oC. Alternative, rapid testing to establish membership of the coliforms can be by testing for the presence of the b-galactosidase enzyme. Coliforms can be isolated from the environment and also from faecal materials. Faecal coliforms are a subgroup of the coliforms that have an ability to grow at high temperatures of 44.5oC. The underlying strategy of the increased temperature for faecal coliform isolation is that when coliforms have been present in the environment for a while, they begin to lose the ability to grow at the temperatures found in mammalian and other digestive tracts. The use of a higher incubation temperature selects for bacteria that have been recently deposited into the environment and have not yet lost their ability to grow at mammalian gut temperatures. It is worth noting that a very high percentage (more than 80%) of a faecal coliform count is typically made up of E. coli.
Streptococcus and Enterococcus. Streptococcus and Enterococcus are closely related genera of bacteria that have been used as alternative indicators for faecal pollution of water. The isolation of different species of faecal Streptococcus have been used with some success to enable differentiation between animal and human faecal pollution, as a way of providing clues to the sources of water contamination. Human-derived faecal material contains a large percentage of enterococci as the dominant cocci whereas animal-derived material contains high numbers of Streptococcus. Monitoring the ratios of enterococci to other streptococci has potential for determining whether faecal contamination is derived from human or animal sources, however this approach is limited as Enterococcus and Streptococcus populations decline in the environment at different rates. Enterococcus, which are present at low levels in animal manures are more hardy than generic streptococci. In addition, the ratio of faecal coliforms to faecal streptococci (FC:FS) also has a potential for differentiating pollution sources. For fresh pollution, FC:FS ratios of >4 correlate well with human faeces, and <0.7 is more representative of animal faeces. However, these ratios are also subject to change with time because individual members of the coliform group, coliforms collectively, and enterococci decline in the environment at different rates.
Listeria. Most strains of Listeria are not human pathogens. Listeria species that have some medical significance are restricted generally to some strains of L. monocytogenes and L. ivanovii. Listeria have become increasingly important for food safety over the last 30 years because they are psychrotrophic (i. e. they can grow and multiply at refrigeration temperatures). Furthermore, listeriosis is a serious infection in humans because it results in death in around 35% of cases. Thus, undertaking counts of generic Listeria as a general hygiene indicator in food production and processing environments has become increasingly popular over the last decade. Listeria is ubiquitous in the environment and can readily be isolated from soil, water, and also from the drains, floors, walls and equipment of most production and processing areas. Consequently, Listeria is considered to be more an indicator of post-harvest hygiene and cleaning effectiveness as well as an index of likely product shelf life (under refrigerated conditions). There are some statutory criteria for L. monocytogenes for the entire shelf life of the produce in some foods that are considered to be ready-to-eat.
Total aerobic count. The total aerobic count (TAC), is known by a number of names including; the total mesophilic aerobic count (TMAC), the total viable count (TVC) and a standard plate count. A TAC attempts to quantify the number of bacteria and fungi in a test sample which can grow at 30oC (within the moderate temperature range of 5-45oC) in the presence of oxygen. Thus, despite the use of the word ‘total’ in the majority of names used for this test, a TAC does not measure the entire population of bacteria contained within the sample. A high TAC is associated with rapid food spoilage and a low product shelf life. TACs can be problematic when used as indicators in raw products such as fresh produce. These difficulties stem in part from the widely fluctuating range of counts that are likely to be encountered in uncooked foods. By way of example, consider the following. A grower tests each field of lettuce he produces by sending five samples for TAC testing and trending the average count. The five test results he gets are 250, 500, 120, 100000, and 25 colony forming units per gram (cfu/g) of lettuce. The mean of these counts is 20179 cfu/g, which is much higher than the majority of the test results because the single high count is so large, it disproportionately skews the average. In order to get a more representative mean, it is necessary to take logs of each of the counts and use the logged number for the calculation. For the above example, the logs for each of the values are 2.40, 2.70, 2.08, 5 and 1.40 and the average is 2.72. The antilog of 2.72 is 525 CFU/g which is a more representative average of the counts obtained. Microbiologists call a result calculated in the above manner a geometric mean. For many growers, the additional calculation stages are off-putting and are one reason why TACs are not widely used indicators for fresh produce.
A summary Venn diagram that shows graphically the relationship between the indicators discussed above is shown above as Figure 1.
In the UK, there are no legal criteria for indicator bacteria in irrigation waters. Unless a grower is attempting compliance with a microbiological standard contained within an assurance scheme, it is therefore appropriate the growers decide for themselves what an acceptable standard is for their growing activities. There are a number of international guidance standards for irrigation water and statutory criteria for UK drinking water that growers may find useful points of reference when assessing their irrigation water quality. These standards are summarised in Table 1 below.
(RTE – ready to eat crops eaten without cooking; cfu – colony forming units; MPN – a test result calculated from the most probable number microbiological test method)
Issuing body and commodity
World Health Organisation Irrigation water for crops to be eaten raw
≤ 1000 cfu per 100 mls (calculated as a geometric mean).
UK Drinking Water Inspectorate: Service reservoir water
E. coli and also Faecal coliforms
Zero cfu per 100 mls for both indicators in 95% of samples taken.
State of California, USA. Recycled irrigation water for RTE
≤ 2.2 MPN cfu per 100 mls in previous 7 days of test results. No sample to exceed 23 MPN CFU/100mls in previous month.
Canadian Agriculture Ministry Irrigation water guidelines
Faecal coliforms or E. coli and also Total coliforms
≤ 100 cfu of faecal coliforms or E. coli per 100 mls ≤ 1000 cfu of total coliforms per 100 mls.
Tesco Stores Nurture Scheme RTE irrigation water standards
E. coli and also Faecal coliforms
≤ 1000 cfu per 100 mls for both indicators (calculated as a geometric mean if multiple samples are taken).
Marks & Spencer Field to Fork RTE irrigation water standard
≤ 100 cfu per 100 mls for RTE, but the criteria become more relaxed for less risky crops.
State of California, USA. Recycled irrigation water for RTE
≤ 2.2 MPN cfu per 100 mls in previous 7 days of test results. No sample to exceed 23 MPN CFU/100mls in previous month.
MacDonalds' Good Agricultural Practice
Five test rolling geomean average of 126 MPN/100ml with no single test >235 MPN/100 ml.
Table 1 International standards and guidelines for selected indicator numbers in irrigation and potable water.
Laboratory testing and reporting
Water can be tested by laboratories in three main ways. These are:
filter-based protocol, where 100ml sample is typically tested,
most probable number (MPN) based test method where a number of smaller volumes are independently tested,
direct plating onto culture media.
The filter-based method passes water through a porous membrane with pores are that are too small to allow bacteria through; thereby concentrating the bacteria in a large volume of water on one side of the membrane. The membrane is then placed on top of a solid growth media to allow single cells to grow into visible colonies which can be counted. Pictures of typical membranes after culture is shown as Figure 2.
The most probable number (MPN) method is considered to be old fashioned compared with the filter method, although it has an advantage it can detect cells below the threshold for many microbiological methods such as direct plating. Furthermore, the MPN method however is still useful if the water sample has a large amount of dissolved solids which makes filtration difficult or for very low numbers of microbes. It is important to note that the MPN method is a statistical estimate of the bacterial numbers within a sample rather than the absolute count made by the filter method. In brief, when undertaking MPN, small volumes of supplied sample are diluted by x10, x100, x1,000, x10,000 and x100,000. A number of volumes (typically 3 or 5) are removed from each of the dilutions and then added to tubes of liquid growth media. After incubation, the number of tubes containing growth for each dilution used in conjunction with a set of lookup tables to estimate the statistically most probable number of bacteria contained within the original sample. It is usual for laboratories to report an MPN test result as MPN/100mls rather than cfu/100mls.
Direct plating is commonly used for a count of total mesophilic aerobes (microbes which tend to like moderate temperatures and require oxygen) contained in poor-quality water. The technique is simple and involves spreading small quantities of either neat or diluted water directly onto the surface of an agar plate or mixing the sample into molten agar and allowing it to set. Figure 3 shows a set of plates after sufficient incubation to allow single organisms to grow up to visible colonies for counting.
Irrespective of the test method used, water test results are normally always reported as cfu per 100mls (because it’s rare to find a single bacteria in 1 ml of good-quality water). When food is tested by a laboratory, a 25g sample is typically used for the test and the result is reported as the number of colony forming units per gram (cfu/g) of food.
Interpretation of test results
There are no statutory criteria for irrigation water quality in the UK, unless it is recycled wastewater. Given that there are a number of sets of guidance for acceptable indicator numbers, growers should decide which standard they want to adopt. In a number of cases, retailer requirements will dictate which standard is adopted. At periodic intervals determined by either a standard such as Red Tractor or by growers themselves that are not following a standard (once per month is considered a reasonable initial test frequency), water samples should be collected and sent for testing. In order to form a picture of how the water microbiology changes over time, many grower technical managers find plotting a trend of test results to be of benefit. A typical plot of test results for a surface river is shown as Figure 4. More information on trending and how to overcome some of the difficulties encountered when trying to make sense of trended data is available.
Out of specification results
The general strategy of fairly frequent monitoring of water sources with action (and expense) only when test results go out of specification, is the most cost-effective strategy for growers interested in producing microbiologically safe crops. Although it will not change the status of the water, many growers switch to more frequent sampling and testing if they have an out of specification result because it provides more information on how the irrigation water quality is changing during the period of concern.
If test results are out of specification, a number of approaches can be attempted to try and help bring the water back into specification. The course of action will be largely dictated by the finances available to the grower and the seriousness of the contamination problem. Growers should initially carry out detailed risk assessment to identify all potential sources of contamination and consequently inform the most appropriate action to stop or eliminate further contamination. The following options could then be considered as a way of reducing the risk of contaminating a crop:
Maximise the time between irrigation and harvest. Pathogens decline relatively quickly on the surface of crops, where they are exposed to sunlight and periods of dry weather. Fast growing leafy salad and herb crops pose a particular risk. Work on persistence of human pathogens on salad leaf surfaces has recommended that the water free of contamination should be used within 2 weeks of harvest. It is advised not to use irrigation to hydrate crops immediately prior to harvest in extremely hot conditions because if the water is contaminated, it can result in high E. coli levels. More information on water intervals here.
Change the water source. Often the simplest course of action is to switch to an alternative water supply which meets the specification. Although expensive, mains water is considered to be the highest quality for irrigation. It is not good practice to dilute out of specification water with mains water to bring it back into specification because the high test result indicates a potential problem with the water source that will still exist after dilution. More information on water sources here.
Change the method of irrigation. Generally speaking, contaminated water poses the highest risk to fresh produce where the irrigation water comes in direct contact with the edible portion of the crop. The use of irrigation methods that reduce or prevent water coming in to contact with the edible portion of the crop, such as drip or trickle tape could reduce the risks of using lower grade irrigation water contaminating crops. More information on water contact here.
Treat water before irrigation. The safest (and most expensive) course of action is to treat the water by buying (or short-term leasing) an ultraviolet light water treatment unit or a reverse osmosis filtration device. Alternatively, chemical treatments such as the use of a reactive radical chemical-based purifier (e.g. ozone or hypochlorite) can be used to cost-effectively reduce the bacterial population of smaller volumes of water before application. Water treatment may be the only option where the only potential water source is heaviliy contaminated and other mitigations are not possible. More information on water treatment here.
Important issues relating to laboratories and results interpretation
When trending test results over time, there are a number of considerations to be made. It is important that the laboratory does not change the test method it uses to count indicators or detect pathogens. A test report for a good-quality laboratory will always contain a test method reference, and this reference should be checked to make sure it’s the same as previous test before trending new data with historical data. If a single water sample is tested using two different test methods, the two results will be different (sometimes very different). Similarly, because test results are sensitive to the equipment used to measure liquids, the brand and type of growth media used and numerous other factors, changing the test laboratory can also result in large changes in the test results. Furthermore, if the same test method is used, but the tests are undertaken in different laboratories, this can also lead to unexplained differences between test results. It is not good practice to mix different lab’s results on a single trend graph. If the testing laboratory or testing method is changed, a new trend graph should be started.
Good-quality laboratories will have determined the detection limits for all of the microbiological tests that they undertake. Consequently, these laboratories will not report low numerical test results as 0 cfu/100ml. If the detection limit of the test result is 10 cfu per 100mls, and no bacteria grew during the test, the laboratory will report the result as <10 cfu/ 100mls; acknowledging that the test method in use is unable to detect below 10 cfu/100mls. It is a common practice when trending microbiological results to use half the limit of detection for any low counts reported. Thus, a result of <10 cfu/ 100mls would be trended as 5 cfu/ 100mls. The same approach and value would be used for the calculation of any averages (i.e. geometric means) from multiple sample tests.