Cryptosporidium and Giardia taxa in faecal samples from animals in catchments supplying the city of Melbourne with drinking water (2011 to 2015)

Background In a long-term program to monitor pathogens in water catchments serving the City of Melbourne in the State of Victoria in Australia, we detected and genetically characterised Cryptosporidium and Giardia in faecal samples from various animals in nine water reservoir areas over a period of 4 years (July 2011 to November 2015). Methods This work was conducted using PCR-based single-strand conformation polymorphism (SSCP) and phylogenetic analyses of portions of the small subunit of ribosomal RNA (SSU) and 60 kDa glycoprotein (gp60) genes for Cryptosporidium, and triose-phosphate isomerase (tpi) gene for Giardia. Results The prevalence of Cryptosporidium was 1.62 % (69 of 4,256 samples); 25 distinct sequence types were defined for pSSU, and six for gp60 which represented C. hominis (genotype Ib - subgenotype IbA10G2), C. cuniculus (genotype Vb - subgenotypes VbA26, and VbA25), and C. canis, C. fayeri, C. macropodum, C. parvum, C. ryanae, Cryptosporidium sp. “duck” genotype, C. suis and C. ubiquitum as well as 12 novel SSU sequence types. The prevalence of Giardia was 0.31 % (13 of 4,256 samples); all three distinct tpi sequence types defined represented assemblage A of G. duodenalis. Conclusions Of the 34 sequence types (genotypes) characterized here, five and one have been recorded previously for Cryptosporidium and Giardia, respectively, from humans. Novel genotypes of Cryptosporidium and Giardia were recorded for SSU (n = 12), gp60 (n = 4) and tpi (n = 1); the zoonotic potential of these novel genotypes is presently unknown. Future work will continue to monitor the prevalence of Cryptosporidium and Giardia genotypes in animals in these catchments, and expand investigations to humans. Nucleotide sequences reported in this paper are available in the GenBank database under accession nos. KU531647–KU531718. Electronic supplementary material The online version of this article (doi:10.1186/s13071-016-1607-1) contains supplementary material, which is available to authorized users.


Background
One of the toughest challenges facing the world's supply of clean drinking water is contamination from faeces and soil [1,2]. Diarrhoeal disease is responsible for 10.5 % of deaths in children of less than five years of age [3][4][5], having a greater impact than malaria and HIV/ AIDS combined [5]. Pathogens of concern include viruses, bacteria and protists [1]. Of the latter pathogen group, human-infective taxa (i.e. species and genotypes/assemblages) of Cryptosporidium and Giardia are highly significant [1,[4][5][6][7][8]. Cryptosporidium and Giardia are unique in that very small numbers of infective stages (oocysts and cysts, respectively) can cause disease in humans [9,10] and that these stages are resistant to chlorination and other common water treatments [1,11]. An example of the tremendous impact these parasites can have was demonstrated in 1993, with a major outbreak of cryptosporidiosis in Milwaukee, USA [12], which affected more than 400,000 people and resulted in 100 deaths. This case emphasizes the major public health importance of waterborne diseases and the need for their sustained prevention.
Melbourne (Victoria, Australia; population~4 million) is one of the few cities in the world that receives largely unfiltered drinking water from protected wilderness catchment areas. The management of Melbourne's ten main water catchment areas includes restricted access for humans, long water retention times and an intense program of testing and monitoring for pathogens in source water. These catchments represent habitat for native and feral animals, such that the monitoring of zoonotic pathogens is central to management and the prevention of outbreaks of waterborne disease. In 2008, we initiated a program to monitor Cryptosporidium and Giardia in faecal samples from various mammals and birds in the Melbourne's catchments [13]. To do this, we collected 2,009 fresh faecal samples (from June 2009 to June 2011) and tested them using an established and validated PCR-based mutation scanningcoupled sequencing approach (cf. [14][15][16]), combined with phylogenetic analyses of loci (SSU and gp60) in the small subunit (SSU) of ribosomal RNA and 60-kDa glycoprotein (gp60) genes to detect and characterise Cryptosporidium, and another locus (tpi) in the triose-phosphate isomerase (tpi) gene to identify and classify Giardia [13]. In total, Cryptosporidium and Giardia were detected in 2.8 and 3.4 % of all 2009 samples tested, respectively, and 35 previously undescribed genotypes were reported [13]. In spite of this relatively low prevalence, the findings from this study emphasized a need for a sustained program.
Therefore, from July 2011 to November 2015, we extended our monitoring program, and genetically characterised Cryptosporidium and Giardia from native and introduced animals in Melbourne's water catchments, in order to continually assess the prevalence and diversity of Cryptosporidium and Giardia taxa, evaluate their host affiliations, geographical distributions and zoonotic potential, and support catchment management. In the present article, we describe the results from this 4-year study and discuss the findings in a water industry context.

Melbourne's catchments
Greater Melbourne sources its municipal drinking water from ten main water catchment reservoirs. Approximately 80 % of Melbourne's drinking water is drawn from 'closed' catchments in the Yarra Ranges (~85 km east of Melbourne), which cover 157,000 hectares of eucalypt forest, with restricted human and domestic animal access, to minimise the risk of waterborne diseases. The remaining 20 % of Melbourne's water comes from 'open' catchments, in which some farming and human activities are permitted. All water undergoes treatment in accordance with national and international guidelines [17,18]. The nine reservoirs studied here ( Fig. 1) are located north and east of Melbourne's central business district (CBD), are less than 90 km apart and include: Cardinia (CA) 37°47'S, 145°24'E; Greenvale (GV) 37°37'S, 144°54'E; Maroondah (MR) 37°38'S, 145°33'E; O'Shannassy (OS) 37°40'S, 145°48'E; Silvan (SV) 37°50'S, 145°25'E; Tarago (TAR) 37°59'S, 145°55'E; Thompson (TH) 37°47'S, 146°21'E; Upper Yarra (UY) 37°40'S, 145°55'E; and Yan Yean (YY) 37°33'S, 145°08'E. Reservoirs MR, OS, TH and UY are situated in the densely forested Yarra Ranges catchment, whereas YY reservoir is a much smaller catchment north of the CBD and surrounded by residential and grazing land. The remaining reservoirs, including CA, GV and SV, act as storage facilities for the larger catchments and have eucalypt and/or pine forests. TAR is the one 'open' water supply catchment, which permits farming in the land surrounding the reservoir. All regions have small areas of grassland adjacent to water reservoirs, and it is here that faecal samples were collected.

Polymerase chain reaction (PCR) amplification of genetic loci
Genomic DNA samples were subjected to nested PCRbased analyses of three loci. For the specific identification of Cryptosporidium, a portion of the SSU gene (~240 bp) was used [13], and genotypic/subgenotypic classification was achieved employing part of the gp60 gene (250-350 bp) [13]. For the genetic characterisation of Giardia (to the level of assemblage), a portion of the tpi gene (~530 bp) was employed [21]. PCR was carried out in a volume of 50 μl containing 10 mM Tris-HCl (pH 8.4), 50 mM KCl (Promega, Madison, USA), 2.0-3.0 mM of MgCl 2 (depending on the locus), 200 μM of each deoxynucleotide triphosphate, 50 pmol of each primer and 1 U of either GoTaq (Promega) or MangoTaq™ (Bioline, USA) DNA polymerase.

Mutation scanning, sequencing and phylogenetic analyses
Single-strand conformation polymorphism (SSCP) analysis was used to scan for sequence variation within and among SSU and gp60 amplicons (e.g. [13,16]). In brief, 1 μl of each secondary amplicon (< 450 bp) was mixed with 5 μl of DNA sequencing-stop solution (Promega) and 5 μl of H 2 0, heat-denatured at 94°C/30 min, snapcooled on a freeze-block (-20°C) and then subjected to electrophoresis at 74 V at 7.4°C (constant) for 16 h in a GMA Wide Mini S-2x25 gel in a SEA 2000 rig (Elchrom Scientific AG) using TAE buffer (40 mM Tris base, 20 mM acetic acid, 1.0 mM EDTA, Bio-Rad, USA). A control sample (representing a known genotype) was included on each gel to ensure the reproducibility of profiles representing this sample among gels.
Following SSCP-based analysis, selected amplicons representing each distinct electrophoretic profile and all SSU, gp60 (Cryptosporidium) and tpi (Giardia) amplicons were treated with shrimp alkaline phosphatase and exonuclease I (ThermoFisher, Waltham, USA), according to the manufacturer's instructions, and then subjected to bi-directional automated sequencing (BigDye® Terminator v.3.1 chemistry, Applied Biosystems, USA) using the same primers employed in the secondary PCR. Sequence quality was verified by comparison with corresponding electropherograms using the program Geneious v.8 [28]. Sequences were aligned using the program MUSCLE [29], and alignments were adjusted manually using the program Mesquite v.2.75 [30]. Sequences were then compared with those available in the GenBank database using BLASTn.
Phylogenetic analysis of sequence data was conducted by Bayesian inference (BI) using Monte Carlo Markov Chain (MCMC) analysis in MrBayes v.3.2.3 [31]. The likelihood parameters set for BI analysis of SSU data were based on the Akaike Information Criteria (AIC) test in jModeltest v.2.1.7 [32]. For SSU (Cryptosporidium) and tpi (Giardia) data, the number of substitutions (Nst) was set at 6, with a gamma-distribution and a proportion of invariable sites. For the separate analyses of gp60 (Cryptosporidium) sequence data, the Nst was set at 6, with an equal rate among sites. Posterior probability (pp) values were calculated by running 2,000,000 generations with four simultaneous tree-building chains. Trees were saved every 100th generation. At the end of each run, the standard deviation of split frequencies was < 0.01, and the potential scale reduction factor approached one. A 50 % majority rule consensus tree for each analysis was constructed based on the final 75 % of trees generated by BI. Analyses were run three times to ensure convergence and insensitivity to priors. Outgroups used in the analyses were Giardia muris for G. duodenalis (tpi), C. hominis for gp60 and C. muris for SSU.

Molecular detection of Cryptosporidium, and taxon identity based on SSU
We conducted mutation scanning and sequence analyses of all amplicons (n = 69) produced from 4,256 (1.62 %) faecal DNA samples and identified them to species and/ or genotype of Cryptosporidium. A total of 64 samples were characterised by their SSU sequences; 52 were assigned GenBank accession nos. (KU531647-KU531698; Tables 1 and 2), of which 24 sequences were selected as representatives for phylogenetic analysis ( Fig. 2; Additional file 2: Table S4). In total, there were 12 novel sequences (i.e. < 100 % identity with a sequence on GenBank) for SSU. Samples that were test-positive for SSU were assessed according to catchment (Additional file 1: Table S2) and host (Additional file 1: Table S3). Overall, prevalence was assessed by catchment in Table 2.  In the following list, we assign individual taxa identified in the present study (based on the sequencing of SSU) to the most closely related species of Cryptosporidium in particular clades based on sequence identity: Members of the C. fayeri clade: Six of 74 (8.10 %, catchment MR) samples from wallabies were test-positive for C. fayeri. Two of 287 (0.70 %, YY) and one of 603 (0.17 %, GV) samples from kangaroos were test-positive for Cryptosporidium sp. EGK1 (eastern grey kangaroo type 1), two of which were novel genotypes (GenBank accession nos. KU531656 and KU531666). One of 76 (1.32 %) samples from wallabies from catchment CA was a novel genotype that was similar to C. fayeri (KU531658). One of the 287 (0.35 %) samples from kangaroos from catchment YY was a novel genotype similar to Cryptosporidium sp. kangaroo genotype I (KU531694). For the first time, C. fayeri and C. fayeri-like genotypes were reported from wombats. Members of the C. hominis, C. parvum and C.  Cryptosporidium subgenotypes based on gp60 All ten samples test-positive in PCR for gp60 (n = 10) were characterised to the level of subgenotype, and seven of them were assigned GenBank accession nos. (KU531699-KU531705). Based on a comparison with reference sequences from GenBank, six unique gp60 sequence types were characterised as C. hominis (genotype Ib -subgenotype IbA10G2), and C. cuniculus (genotype Vb -subgenotype VbA25 and VbA26) ( Fig. 3; Table 1 and Additional file 2: Table S5). One of 74 (1.35 %) samples from wallabies from catchment MR was test-positive for C. hominis IbA10G2 (accession no. KJ506839). In addition, Cryptosporidium cuniculus subgenotypes VbA25 and VbA26 were identified in seven samples from rabbits from catchments GV, YY, SV and MR, four of which were novel. Cryptosporidium cuniculus (VbA26) was identified in two samples from kangaroos from catchment YY (cf. [33]).

Giardia species and assemblages
Sequencing of all tpi amplicons identified 13 of 4,256 (0.31 %) individual faecal samples to contain Giardia representing the genetic assemblage A of G. duodenalis, based on direct sequence comparisons. From the 13 samples, we  Table S6). Giardia sub-assemblage AI was identified in samples from a rabbit and a deer in catchment TH, a wombat in catchment MR and a kangaroo in catchment SV. The genotype (sequence type) of Giardia from deer identified here was novel. Giardia sub-assemblage AIII was identified in nine samples from deer, including five of 536 (0.93 %, UY), two of 408 (0.49 %, OS) and two of 18 (11.1 %, TAR) samples from three different catchments.

Discussion
Extending our initial monitoring program [13], the present study provides a unique perspective of the epidemiology of zoonotic protists found in wildlife inhabiting the water catchment areas of a major metropolitan city (Melbourne). Worldwide, there have been very few comprehensive wildlife surveys in catchments or watershed, with a few notable exceptions (e.g. [34][35][36]). In Australia, Cryptosporidium and/or Giardia have been described from catchment regions in New South Wales [37][38][39][40][41][42], Queensland [43], Victoria [13,33,44] and Western Australia [45]. In the first three years of the project (June 2009 to June 2011), 2,009 faecal samples were collected and tested, resulting in a prevalence of 2.8 % for Cryptosporidium and 3.4 % for Giardia [13]. In contrast, over the last four years of the project (July 2011 to November 2015), we tested more than double the number of faecal samples (n = 4,256), yet the prevalence of Cryptosporidium and Giardia were 1.62 % and 0.31 %, respectively. Overall, the prevalence of these protists was usually less than that reported previously in New South Wales [38,39,42], Victoria [13] and Western Australia [45]. The low prevalence recorded here (compared with surveys in other states of Australia) might be a consequence of testing a greater number of samples, the host groups tested, differences in local habitats and/or catchment management practices and/or proximity to agricultural land. Many factors may account for the low prevalence of Cryptosporidium and Giardia in Melbourne's catchments over the six-year duration of the project, including animal culls, changing water levels of the reservoirs and the end of a nine-year drought [46]. One factor influencing the higher prevalence of Giardia in the earlier years (cf. [13]) was a hot-spot event recorded in catchment YY in April 2010, which did not recur the subsequent years. Indeed, no Giardia was found in catchment YY at any stage during the present investigation. Interestingly, the presence of the ungulate-specific G. duodenalis sub-assemblage AIII in the catchments was only recorded Fig. 4 Relationships among Giardia taxa inferred from the phylogenetic analysis of partial triose-phosphate isomerase gene (tpi) sequence data by Bayesian inference (BI). Posterior probabilities are indicated at all major nodes. Bold-type indicates Cryptosporidium species or genotypes characterized from faecal DNA samples in this study. In parentheses are the numbers of samples representing a particular species, genotype and sequence (GenBank accession numbers indicated). Novel genotypes (*). Scale-bar represents the number of substitutions per site once between 2009 and 2011, and the first record of Cryptosporidium from a wombat was recorded in 2013, highlighting the rarity and ephemeral nature of these protists in this catchment system.
In the present study, we used a PCR-based approach to genetically characterise 82 samples, which were assigned to seven recognised species of Cryptosporidium (represented by the GenBank accession nos. KU531647-KU531705), and to the genetic assemblage ' AI' and ' AIII' of G. duodenalis (accession nos. KU531706-KU531718). Of the recognised species of Cryptosporidium, only C. macropodum, detected here in kangaroos, has not been reported previously from humans. The remaining six potentially zoonotic species (C. hominis, C. parvum, C. cuniculus, C. ubiquitum, C. canis and C. fayeri) were recorded from deer, emu, kangaroo, rabbit, wallaby and wombat, respectively, and from catchments CA, GV, OS, MR, SV, UY and YY (see Additional file 1: Table S1). Of the 29 species and < 40 reported genotypes of Cryptosporidium currently recognised [8], the causative agents of human cryptosporidiosis are typically C. hominis or C. parvum (see [47][48][49][50][51]); these parasites have been linked to numerous waterborne outbreaks around the world (reviewed in [52,53]). Despite the detection of both species in catchments regions surrounding Melbourne, their low prevalence (0.07 % for C. hominis and 0.02 % for C. parvum) might suggest a low risk of waterborne transmission to humans. Nonetheless, other species and genotypes of Cryptosporidium may have some zoonotic significance. For example, C. cuniculus, with a prevalence of 0.26 %, was implicated in a zoonotic outbreak of cryptosporidiosis in humans in England in 2008 [54] and was linked to a number of sporadic human cases across the UK in 2007 and 2008 [55,56], and was detected for the first time in a kangaroo in the YY catchment [33]. Cryptosporidium ubiquitum, detected here in deer in catchments CA and UY, might also be a concern, as it has been proposed to present a potential public health risk due to its broad geographical and host ranges, including humans in industrialised nations [49,57,58]. It is plausible that C. canis (0.02 %) and C. fayeri (0.26 %) might also represent minor risks to humans; C. canis has been detected in humans around the world [59], and C. fayeri was detected in a patient in New South Wales suffering from a prolonged gastrointestinal illness [60].
In the present study, we were able to assign Cryptosporidium to particular clades of species based on sequence identity in SSU, and to genotypes/subgenotypes also based on their gp60 sequence. Within the C. fayeri clade, several novel genotypes characterised from samples from kangaroos were all very similar genetically to the marsupialspecific C. fayeri. Here, we also report the first molecularly characterized Cryptosporidium genotypes from wombats. To date, there is only one record of Cryptosporidium from a wombat [61]; however, this was not included in the review of Cryptosporidium of marsupials [62]. Wombat faeces have been tested for Cryptosporidium in multiple surveys [13,39,42,63], yielding no test-positive results. Over the six years of our monitoring of Melbourne Water catchment areas ( [13] and the present study), 609 faecal samples from wombats have been tested molecularly, and nine (1.48 %) were test-positive for Cryptosporidium and all were genetically very similar to C. fayeri ( Fig. 2 and Additional file 2: Table S4). For the C. hominis, C. parvum and C. cuniculus clade, we found relatively few samples (n = 14) to be test-positive (by either SSU and/or gp60) for members of this clade compared with our previous study (n = 32) [13]. Considerably fewer samples were collected from rabbits (n = 97) than in the previous study (n = 263) [13]. The C. hominis positive (IbA10G2) from a sample from a wallaby from catchment MR was identical in gp60 sequence to a novel C. hominis genotype found in a human in Tasmania (GenBank accession no. KJ506839; [64]), and is the first report of C. hominis from a swamp wallaby. The genotypes of C. cuniculus characterised from kangaroos have been reported recently [33]. Within the C. canis clade, we identified Cryptosporidium consistent with C. canis in a faecal sample from an emu, which might relate to pseudo-parasitism, whereby the parasite is ingested and passed through the gastrointestinal tract of the host without establishing an infection. Typically, emus eat plants and insects [65], which could have been contaminated with oocyst-containing faeces from feral dogs or foxes within catchment CA. Within the C. suis clade, we identified, for the first time, a novel C. suislike genotype in a sample from a deer. Cryptosporidium suis has been considered to be specific to pigs, but has also been found in cattle, rodents, humans and chimpanzees [8]. Cryptosporidium suis-like protists that differ by only a few bases in SSU have been detected in cattle and rats [66,67]. Within the C. ubiquitum clade, we report, for the first time, a novel C. ubiquitum-like genotype from wombats (Fig. 2). The samples were collected from catchments CA (July 2013) and OS (November 2014). Cryptosporidium ubiquitum is considered an emerging human pathogen that has been found in a wide range of wildlife, including canids, deer, primates and rodents [8].
Within the C. macropodum clade, we identified C. macropodum-like genotypes in six swamp wallabies from catchments CA, OS and SV. Three of the six genotypes were novel. Although there seems to be a differentiation within the C. macropodum clade between the wallabies and kangaroos in the catchments, this difference does not carry through to brush-tailed wallabies in New South Wales, where three such wallabies had a C. macropodum genotype and two shared C. macropodumlike genotypes [68]. More extensive sampling and testing of samples from wallabies and kangaroos throughout Victoria and New South Wales would be needed to clarify whether particular macropod host-affiliations exist. Within the C. ryanae clade, C. ryanae is usually found in cattle, but has also been recorded in other ruminants, such as water buffaloes and roe deer [8]. Although C. ryanae was not detected in this study, multiple closely related C. ryanae-like and Cryptosporidium sp. "deer" genotypes were identified in deer. The majority of the cervines in the catchments studied here are sambar deer, but it is not possible to confidently distinguish scats of sambar from those of fallow and red deer without using molecular tools. Using such tools, the specific identification of the cervine hosts would assist in assessing host affiliations and genetic diversity within the C. ryanae complex. The potential role of some Cryptosporidium genotypes (e.g. C. suis-like in deer and C. ubiquitum-like in wombat -GenBank accession nos. KU531660 and KU531681, respectively) as zoonotic agents remains to be proven (cf. [8,69,70]). Further investigations of the presence and/or distribution of Cryptosporidium genotypes in rabbits, deer (e.g. C. hominis, detected here, for the first time, in wallaby) and other native and introduced wildlife in Australia, particularly in areas surrounding water catchments, are necessary, not only to determine the significance of various host groups as primary sources, potential reservoirs and amplifiers of Cryptosporidium for transmission to humans, but also to establish the mode(s) of transmission among reservoir animal hosts and how infection is maintained in wild animal populations [8,62]. The present results emphasize the need for increased investigation into the true host ranges of all Cryptosporidium species infecting wild and domesticated animals not yet studied.
Of the currently eight recognised species of Giardia, G. duodenalis is responsible for human disease [71][72][73], with isolates linked to sub-assemblages AI, AII, BIII and BIV, considered to be most commonly infective to humans, whereas sub-assemblages AIII, AIV, BI and BII are recognised to be infective to animals other than humans [72,74]. Sub-assemblage AIII, in particular, is associated with deer and other wild ungulates [72,75,76]. All of the three genetic variants of tpi detected herein represented assemblage A; two have been reported previously (e.g. GenBank accession nos. KU531708 and KU531707) (cf. [13]), and one is novel (accession no. KU531709). Since assemblages A and B of G. duodenalis appear to represent the greatest zoonotic risk, given their presence in humans, livestock and companion animals [71,72,74,77], studying genetic variability within/among Giardia isolates is pivotal to inferring the zoonotic potential of distinct genotypes within this genus of parasite.

Conclusions
The present study has provided detailed insights into the taxa of Cryptosporidium and Giardia in animals in key water catchments in Victoria. The genetic analyses indicated that 1.92 % of the 4,256 faecal samples contained Cryptosporidium or Giardia that matched species, genotypes or assemblages with the potential to infect humans. In addition, a number of new sequence records, which did not match any previously published genotypes, were identified. As nothing is known about the zoonotic potential of these new variants of Cryptosporidium and Giardia, future work should establish whether they are found in humans in Australia. Although the focus of the present study was on vast water catchment areas in south-eastern Australia, the research findings and the approach taken have considerable implications for other protected wilderness catchment areas around the world that supply unfiltered drinking water to millions of people.

Additional files
Additional file 1: Table S1. Number of faecal samples examined from each host/location combination. Table S2. Total number of each Cryptosporidium genotype sampled from each catchment. Table S3. Total number of each Cryptosporidium genotype sampled from each host. (DOCX 47 kb) Additional file 2: Table S4. Pairwise comparisons of the SSU gene sequences among Cryptosporidium genotypes. Table S5. Pairwise comparisons of the gp60 gene sequences among Cryptosporidium genotypes. Table S6