Source tracking of Enterococcus moraviensis and E. haemoperoxidus

Removal of micro-organisms during soil passage in dune infiltration areas is often used as one of the treatment steps in drinking water production in the Netherlands. Recovered (abstracted) groundwater is the product of this process and is normally free of faecal indicator bacteria, and therefore considered to be free of faecal-associated pathogenic micro-organisms.

During regular water quality control, enterococci have occasionally been isolated from 100 L samples of abstracted water in the Castricum dune infiltration area (The Netherlands).

Enterococci are bacteria present in the gastro-intestinal tracts of humans and warm-blooded animals and are therefore used as indicators for determining the sanitary quality of water, indicating the possible presence of pathogens. Compared with Escherichia coli, the association of Enterococcus spp. (all species) with the presence of pathogens is not very well known.

Enterococcus spp. is not only associated with warm-blooded animals, but has also been detected in extra-intestinal habitats like invertebrates (Martin & Mundt 1972; Švec et al. 2002), plants (Müller et al. 2001), sediments (Grant et al. 2001; Le Fevre & Lewis 2003), soils (Fujioka et al. 1999), foods (Klein 2003; Foulquie Moreno et al. 2006) and water (Švec et al. 2001).

Current data on Enterococcus species isolated from faecal and non-faecal environments depend upon the identification methods used. Since the number of Enterococcus species described is still increasing, greater species diversity can be expected in sources already known. In the past decade, matrix assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) has increasingly been applied as an identification technique and has also been shown to be suitable for the identification of enterococci in water (Taučer-Kapteijn et al. 2013). The introduction of molecular techniques has provided greater insight into the genetic diversity within Enterococcus spp. and rapidly accelerated the characterization of new Enterococcus species isolated from enteric and extra-enteric environments.

In 2001, two new species of enterococci, Enterococcus moraviensis and E. haemoperoxidus were isolated from surface water and described by Švec et al. (2001). E. moraviensis has been observed as the most frequently identified species in water samples abstracted from the dunes. Laboratory experiments have shown that E. moraviensis is able to multiply under non-enteric circumstances in the presence of dune plant material at 15 °C (Taučer-Kapteijn et al. 2016). The observation that certain strains of Enterococcus spp. may be able to survive and replicate in non-enteric environments – for instance, E. casseliflavus in submerged aquatic vegetation (Badgley et al. 2010) and E. casseliflavus, E. faecalis, E. faecium, E. hirae, E. mundtii, E. sulfureus and many other strains resembling E. faecalis isolated from forage crops (Cai 1999; Müller et al. 2001; Ott et al. 2001) – strongly supports the existence of plant-associated enterococci (Byappanahalli et al. 2012). Furthermore, some enterococci species have been shown as able to grow and persist under non-enteric conditions (Mundt et al. 1962; Whitman et al. 2003; Badgley et al. 2010; Taučer-Kapteijn et al. 2016). These findings challenge the suitability of Enterococcus species for the indication of faecal pollution.

Until now, there have been no reports of faecal sources of E. moraviensis and E. haemoperoxidus. This study aimed to determine if E. moraviensis and E. haemoperoxidus are present in animal faeces in the dune infiltration area. Since this area is used for recreational purposes, human faecal samples were also included in this study. To gain an overview of the Enterococcus species associated with various animal species living in the vicinity of abstraction wells in the infiltration area, a series of faecal samples from these animals was analysed. In order to track the possible origin of the observed contamination of abstracted water, this study additionally focused upon similarities between species distributions in both abstracted water samples and the faeces from the different animals. Information on the sources would establish the reliability of E. moraviensis and E. haemoperoxidus as indicators of faecal pollution, help to interpret the presence of these enterococci in abstracted water and help in the development of effective preventive measures.

To find a faecal source of E. moraviensis and E. haemoperoxidus, and to determine the abundance of various enterococci species in faecal samples, animal faecal samples from highland cattle (Bos taurus), red foxes (Vulpes vulpes), dogs (Canis lupus familiaris), greylag geese (Anser anser), sheep (Ovis aries) and rabbits (Oryctolagus cuniculus) were collected between March and October 2014 in the Castricum infiltration area (The Netherlands). Additionally, faecal samples from 20 healthy persons ranging in age from 3 to 66 years were analysed for enterococci. The numbers of faecal samples per animal host are indicated in Table 1.

Faecal samples were collected in a sterile plastic jar and analysed within 24 hours after collection. Each sample was divided into two parts using two sterile forceps, with the inner part taken for the analysis in order to exclude contamination from other sources (sand, grass, etc.). An amount of 0.5 g of faecal material was placed in a sterile container with 3 mm glass beads (Boom, The Netherlands) and suspended using 9 ml of sterile drinking water. Dilution series (10⁻¹–10⁻⁵) were then prepared. An amount of 1 ml of each dilution was filtered using a 0.45 μm cellulose nitrate filter (Sartorius Stedim) and incubated on Slanetz and Bartley Agar (SBA) for 48 hours at 37 °C (as per ISO 7988-2:2000). After incubation, the total number of characteristic colonies was counted. Moreover, a maximum of 20 single colonies per sample was used to make pure cultures on SBA, which were subsequently identified using MALDI-TOF MS (Biotyper, Bruker) in accordance with the manufacturer's instructions.

A total of 195 abstracted water samples (14 of 1 L and 181 of 100 L) were filtered at locations in the Castricum infiltration area between July 2012 and August 2014. A total of 5,117 enterococci colonies were isolated from these samples using the filtration method (ISO 7988-2:2000) and 381 selected isolates (7.4%) were identified using MALDI-TOF MS (Biotyper). The number of randomly chosen identified isolates varied from one to eight per sample.

From the unprocessed measurements, seven Enterococcus species were selected. These were all observed in the water samples and in at least one of the faecal samples. Bacterial species that were unique to one of the animal classes or the water class were discarded since they do not convey information concerning the animal class of origin in the water samples. The rabbit measurements were also discarded, since we had only two Enterococcus species. Since determining the number and bacterial species for all animal and water samples is labour intensive and expensive, not all Enterococcus colonies were identified at species level. In this experiment, we assumed that the samples from the same class were independent and originated from the same underlying distribution. To improve numerical granularity and statistical power, the empirical bootstrap was used. For the smaller classes (the animal classes), all possible combinations were made using half of the number of samples per combination. For the larger water class, 10⁵ random permutations were drawn using half the number of water samples for each permutation. The probability of drawing the same combination twice is practically zero. All combinations and permutations were averaged and normalized, such that the sum over all seven Enterococcus species for all combinations and permutations equals one. Referring to the combinations and permutations as our bootstrap dataset, this is a seven-dimensional dataset (seven Enterococcus species). The only difference is that the number of samples per class is much higher and that each element is probably statistically more robust. To determine how the different classes relate to each other based upon their Enterococcus species composition, hierarchical clustering was used. The distance measure used for hierarchical clustering was the Mahalanobis distance (Mahalanobis 1936), which assumes normal distributions. The resulting dendrogram was generated using MATLAB (version 7.10).

As a measure for the diversity of Enterococcus species within animal hosts, Simpson's index D was calculated using the formula D=Σ n(n − 1)/N(N = 1), where n = the total number of enterococci of a particular species and N = the total number of enterococci of all species (Simpson 1949).

During the sanitary survey in the Castricum infiltration area, faecal sources in the vicinity of abstraction wells were recorded. Because it was observed that the number of geese and geese droppings in particular parts of this area were much higher than those of and from other animal hosts, the faecal load of geese was estimated.

Two areas of the same size (c. 340 m²) at different locations (A and B) 400 m from one another, both in the immediate vicinity of abstraction wells, were chosen for counts of droppings in order to estimate the faecal load contributed by geese in June 2014. Randomly chosen dropping samples (n= 15) were weighed and measured (length). The average number of enterococci (cfu/m²) was calculated from the quantity of geese droppings per square metre and the average enterococci density (cfu/g faeces) measured in geese droppings.

Animal and human faecal samples (101 in total) were analysed for enterococci. A total of 1,386 isolates were identified as Enterococcus species (Table 1).

The relative distribution of Enterococcus species among selected host species and in abstracted water samples is shown in Table 2. Considerable variation in species composition was found between faecal samples and abstracted water samples. E. faecalis was the enterococcal species most frequently identified in faecal samples, with the exception of those from sheep. The second most common species was E. faecium, with its highest frequency observed in humans (35.1%). E. faecium was not found in any faecal sample from rabbits or sheep. It is also noteworthy that a very high percentage of isolates from rabbits were identified as E. gallinarum (98.1%). While E. faecium was one of the most frequently represented species in human faeces, it was only sporadically isolated from abstracted water samples (3.9%). Ten Enterococcus species found in faecal samples were not isolated from any abstracted water. E. phoeniculicola was isolated from water, but not found in any of the animal hosts. E. moraviensis was most abundant in droppings from geese (23.8%), but also present in droppings from foxes (0.9%) and rabbits (0.3%). E. haemoperoxidus was isolated from geese (11.3%) as the only carrier of this species. These results demonstrate a faecal origin for E. moraviensis and E. haemoperoxidus.

Higher numbers of E. moraviensis and E. faecalis isolates were found in water samples and in geese droppings. Moreover, species distributions in water samples and geese droppings were similar. Seven species isolated from water samples corresponded with species found in droppings from geese; this is higher than the number of corresponding species in other animal hosts. In order to verify these similarities, statistical methods were applied.

Additionally, the diversity of Enterococcus species (D) was calculated for each animal host and for water samples using Simpson's index. The highest diversity was found in geese (D= 0.17), followed by humans (D= 0.24), red foxes (D = 0.28), water samples (D = 0.30) and dogs (D = 0.34). The lowest diversity was observed in rabbits (D = 0.96).

During a sanitary survey in the vicinity of abstraction wells, it was observed that, in a particular area of the dune filtration area, numbers of geese droppings were much higher than those from other animal hosts. In recent years, a distinct increase in the feral geese population has occurred near these abstraction wells, especially in the period March–June. Therefore, that population was considered to have made a substantial contribution to the faecal load in particular parts of the area. On average, the amount of enterococci isolated from geese droppings (n = 20) was 3.48 × 10⁵ cfu/g. The faecal loads for enterococci at two locations (A and B) were almost the same: 1.9 × 10⁷ cfu/m² and 1.8 × 10⁷ cfu/m² respectively (as shown in Table 3). Due to the absence of geese in other parts in the dune area, the faecal load from these birds is believed to be much lower in those areas. The same is true for other animals, with their droppings much less frequently present in the vicinity of the abstraction wells.

This study demonstrates faecal sources of E. moraviensis and E. haemoperoxidus, which means that occurrence of these two Enterococcus species in water samples indicates the possible presence of pathogens. It is not clear if geese, red foxes and rabbits are the only faecal sources of E. moraviensis and E. haemoperoxidus, because the samples had been diluted by means of membrane filtration and so species present in lower concentrations might have remained undetected. To avoid this disadvantage, the application of molecular techniques specific to these species would be useful. Until recently, E. moraviensis and E. haemoperoxidus may have been identified as the closely related E. faecalis, which – together with E. faecium – is the predominant Enterococcus species in human faeces and sewage (Murray 1990; Ruoff et al. 1990; Manero et al. 2002) but is also present in the faeces of non-human animals (Devriese et al. 1987; Aarestrup et al. 2002; Kühn et al. 2003), including wildlife (Mundt 1963). E. moraviensis and E. faecalis have been shown to be the species most frequently observed in abstracted water, together representing 74.2% of all isolates. Because the same two species were also those most frequently represented (53%) in geese droppings, which were regularly observed in the vicinity of abstraction wells (specific parts of infiltration area), and because geese have been observed to make a substantial contribution to the faecal load in specific parts of the Castricum infiltration area, especially during warmer periods of the year, it is assumed that geese droppings may be the source of the Enterococcus species found in the abstracted water. Also, the bacterial compositions of Enterococcus species found in abstracted water samples were much closer to those in geese droppings than those observed in any other animal host. Moreover, since the presence of geese in the area of study coincides with detection of enterococci in abstracted water, molecular techniques could be applied to confirm that the isolates found in geese faeces and in water samples are identical.

The numbers of enterococci isolates and the diversity of Enterococcus species found in geese were higher than in other herbivores like sheep or cattle, and comparable with or even higher than those found in humans or dogs (omnivores). Since the diet of geese consists mainly of plant material and is therefore much more monotonous than an omnivorous diet, these results remain unexplained.

When feral geese cause a heavy faecal load near these abstraction wells, the question arises as to whether human pathogenic micro-organisms may be present in geese droppings and so whether contamination from this source poses a risk to human health risk. The geese population in this dune infiltration area consists mainly of greylag geese (Anser anser), but also a small number of Canada geese (Branta canadensis). Few studies have demonstrated the presence of pathogens in faecal samples from greylag geese, but a high prevalence of Cryptosporidium spp. (Chvala et al. 2006; Plutzer & Tomor 2009), Salmonella (Lillehaug et al. 2005) and Campylobacter spp. (Colles et al. 2008) have been reported. Canada geese have been found to be carriers of Cryptosporidium spp. oocysts (Kassa et al. 2004; Zhou et al. 2004; Moriarty et al. 2011), the cysts of Giardia spp. (Graczyk et al. 1998), Salmonella spp. (Fallacara et al. 2001) and Campylobacter spp. (Pacha et al. 1988; Wahlstrom et al. 2003; Moriarty et al. 2011).

Geese may pollute water by defecating on pasture in the vicinity of abstraction wells, and contamination of groundwater might occur when there is insufficient removal during vertical infiltration through a relatively short unsaturated zone from the surface to the groundwater level. Because it has also been shown that E. moraviensis is able to grow on the same plant material (Taučer-Kapteijn et al. 2016) as geese feed on, growth of this indicator might also occur in geese faeces. New applications of techniques like whole genome sequencing might have potential as tools to determine whether faecal contamination is recent or comes from a secondary source (environmental growth), and could therefore facilitate the estimation of risks to human health.

The intestinal enterococci group (Enterococcus faecalis, E. faecium, E. durans and E. hirae) is described as an indicator of faecal pollution, because these species are typically excreted in the faeces of humans and other warm-blooded animals (World Health Organization (WHO) 2011). This study has shown that high numbers of E. moraviensis and E. haemoperoxidus can be isolated from the droppings of warm-blooded animals, particularly geese. Since these animals may harbour and excrete human pathogens, it is advisable to revise the guidelines and include E. moraviensis and E. haemoperoxidus as indicators of faecal pollution, pointing to animal/bird origin of the pollution.

The faecal contamination and load delivered by the geese in the vicinity of the abstraction wells, the presence of E. moraviensis and E. haemoperoxidus in geese faeces and abstracted water, the similarity of the Enterococcus species composition found in geese and abstracted water, and the potential presence of human pathogens in geese faeces were the basis for the water utility to design a preventive measure: fencing of the specific parts of the dune filtration area to keep geese away from the abstraction wells. This resulted in an improvement of the quality of abstracted water in this area.

In this study, faeces of geese, red foxes and rabbits have been shown to be the source of E. moraviensis. Geese have also been found to be carriers of E. haemoperoxidus. The Enterococcus species compositions in abstracted water samples and in geese droppings were very similar. Although the actual routes of the presumed contamination are not yet known, large quantities of E. moraviensis in geese droppings and frequent identification of E. moraviensis in abstracted water, the presence of geese in specific parts of the dune filtration area, and the evidently high faecal load contributed by geese all indicate a probable influence on the quality of the abstracted water.

We wish to thank PWN Drinking Water Supply Company for making data available for this research. In particular, we thank Bernadette Lohmann, Gerben Schuitema, Ronald Slingerland, Tycho Hoogstrate and Paul van der Linden for providing information on the infiltration area and their assistance in the collection of samples. We also thank Soumaya Hachti for conducting the analysis, the volunteers for their willingness to participate in this study and the management team of Het Waterlaboratorium for giving us the opportunity to carry out this survey.