Molecular and phylogenetic approaches for assessing sources
of Cryptosporidium contamination in water
Norma J. Rueckera,b, Joanne C. Matsuneb, Graham Wilkesc, David R. Lapenc,
Edward Toppd, Thomas A. Edgee, Christoph W. Sensenb, Lihua Xiaof,
Norman F. Neumannb,g,*
aDepartment of Microbiology and Infectious Diseases, University of Calgary, Calgary, Alberta, Canada
bAlberta Provincial Laboratory for Public Health, 3030 Hospital Drive NW, Calgary, Alberta, Canada
cAgriculture and Agri-Food Canada, Ottawa, Ontario, Canada
dAgriculture and Agri-Food Canada, London, Ontario, Canada
eNational Water Research Institute, Environment Canada, Burlington, Ontario, Canada
fCenter for Disease Control, Atlanta, GA, USA
gSchool of Public Health, University of Alberta, Edmonton, Alberta, Canada
a r t i c l e i n f o
Received 19 December 2011
Received in revised form
28 June 2012
Accepted 29 June 2012
Available online 10 July 2012
a b s t r a c t
The high sequence diversity and heterogeneity observed within species or genotypes of
Cryptosporidium requires phylogenetic approaches for the identification of novel sequences
obtained from the environment. A long-term study on Cryptosporidium in the agriculturally-
intensive South Nation River watershed in Ontario, Canada was undertaken, in which 60
sequence types were detected. Of these sequence types 33 were considered novel with no
identical matches in GenBank. Detailed phylogenetic analysis identified that most
sequences belonged to 17 previously described species: Cryptosporidium andersoni, Crypto-
sporidium baileyi, Cryptosporidium hominis, Cryptosporidium parvum, Cryptosporidium ubiq-
uitum, Cryptosporidium meleagridis, muskrat I, muskrat II, deer mouse II, fox, vole, skunk,
shrew, W12, W18, W19 and W25 genotypes. In addition, two new genotypes were identi-
fied, W27 and W28. C. andersoni and the muskrat II genotype were most frequently detected
in the water samples. Species associated with livestock made up 39% of the total molecular
detections, while wildlife associated species and genotypes accounted for 55% of the
Cryptosporidium identified. The human pathogenic species C. hominis and C. parvum had an
overall prevalence of 1.6% in the environment, indicating a small risk to humans from the
Cryptosporidium present in the watershed. Phylogenetic analysis and knowledge of host
eparasite relationships are fundamental in using Cryptosporidium as a source-tracking or
human health risk assessment tool.
ª 2012 Elsevier Ltd. All rights reserved.
* Corresponding author. School of Public Health, Room 357, Civil Electrical Engineering Bldg. (CEB), University of Alberta, Edmonton,
Alberta, Canada T6G 2G7. Tel.: þ1 780 492 8502; fax: þ1 780 492 9070.
E-mail addresses: email@example.com, firstname.lastname@example.org (N.F. Neumann).
0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved.
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
water research 46 (2012) 5135e5150
Phylogenetic trees based on distance, maximum-likelihood,
parsimony, and bootstrap analysis have been used to
demonstrate the evolutionary divergence, as well as the
host-adapted nature of Cryptosporidium species and geno-
types. Several phylogenetic trees have been published, using
different tools that show the relationships between species
and genotypes of Cryptosporidium for a variety of gene targets
such as COWP (Egyed et al., 2003; Xiao et al., 2000, 2002),
hsp70 (Egyed et al., 2003; Sulaiman et al., 2000; Xiao et al.,
2002), actin (Egyed et al., 2003; Sulaiman et al., 2002; Xiao
et al., 2002) and 18S rRNA (Egyed et al., 2003; Fayer et al.,
2010; Feng et al., 2007; Lv et al., 2009; Xiao et al., 1999a,
2002, 2004b). Regardless of which genetic loci are used, or
which phylogenetic models are applied, discrete cluster
patterns (clades) are clearly apparent within the genus
Cryptosporidium, with phylogenetic relationships shown to
hold true across the multiple gene targets and which are
reflective of host specificity.
Over the past decade an effort has been made to under-
stand the hosteparasite relationship and specificity between
Cryptosporidium and their avian, reptilian or mammalian
hosts. In 2002, 10 recognized species and 8 known genotypes
were known, by 2004 this had increased to 13 species and 31
genotypes, and more recently 23 species and close to 50
genotypes have been described (Feng et al., 2007; Jellison et al.,
2009; Jiang et al., 2005b; Nichols et al., 2010; Xiao et al., 2002,
2004b). This information provides a solid framework for
tracking the hosts associated with Cryptosporidium contami-
nation in the environment.
Waterborne outbreaks of cryptosporidiosis have caused
regulatory agencies in the United States and Europe to
implement regulatory monitoring of Cryptosporidium oocysts
in source water. It is felt by many that understanding the
distribution of Cryptosporidium species in source waters would
provide an improvement to these existing drinking water
regulatory frameworks. This is based on the fact that although
many water samples have been shown to contain Cryptospo-
ridium oocysts, not all the species and genotypes identified in
source water pose the same level of risk to humans (Chalmers
et al., 2010a; Jellison et al., 2002; Jiang et al., 2005a; Nichols
et al., 2010; Ruecker et al., 2007; Yang et al., 2008).
Source attribution of Cryptosporidium, based on DNA
sequence information, relies heavily on the development of
robust and standardized bioinformatic tools, a feature
particularly relevant for characterization of Cryptosporidium
from environmental samples in which novel genotypes are
often observed (Chalmers et al., 2010a; Jellison et al., 2002;
Jiang et al., 2005a; Nichols et al., 2010; Ruecker et al., 2007;
Yang et al., 2008). Although bioinformatics are referenced in
most studies, the details of the bioinformatic process are
limited and the criteria used for separating data into species
and genotypes are often poorly described. Herein we describe
the details of the bioinformatics process and outline the
criteria used to assign species or genotype designations for
sequences isolated from a long-term Cryptosporidium source
tracking study carried out within the South Nation River basin
in eastern Ontario, Canada. Probable host sources were
identified based on known hosteparasite relationships and
the associated risk to humans was assessed.
2.Materials and methods
Enumeration of Cryptosporidium oocysts from water was
(U.S.Environmental Protection Agency, 2005). Water collected
for protozoan analyses was refrigerated (4?C) overnight at
Agriculture and Agri-Food Canada’s Eastern Cereal and
Oilseed Research Centre (ECORC) in Ottawa, Ontario. The
following day, 20 L of water was filtered for parasites (Filta-
max? filters, IDEXX, Westbrook, MN) and the filters were
shipped by air transportation overnight in sealed coolers with
ice packs to the Alberta Provincial Laboratory for Public Health
in Calgary, Alberta. The elution of parasites from the filters
was carried out using phosphate-buffered saline with 0.01%
Immuno-magnetic separation (IMS) was carried out using
the DynaBeads G-C Combo Kit (Invitrogen Canada Inc., Bur-
lington, ON), with a modification to the standard IMS protocol
for improving clarity of slides for microscopic examination.
Briefly, two washes of the beadeparasite complex in the
Leighton tubes using phosphate buffered saline with 0.01%
Tween 20 at pH 7.4 (PBST) were carried out following the 1 h of
incubation. The beads were captured by the magnet, the
supernatant was poured off, the beads suspended in 12 mL of
PBST, and mixed by inverting the tube (3e5 times). The tube
was returned to the magnet for bead capture, with the process
repeated. After discarding the PBST from the second wash, the
standard protocol was resumed.
Microscopeslides weredriedat 35?C on a slidewarmerand
fixed with methanol. Slides were stained with 40,6-diamidino-
2-phenylindole (DAPI, SigmaeAldrich, St. Louis, MO) and
stained with fluorescently-labeled monoclonal antibodies
Australia) according to the manufacturer’s recommendations.
Parasites were identified by fluorescence microscopy and
confirmed to be Cryptosporidium based on DAPI staining
characteristics or by their size, shape, and morphology using
differential interference contrast microscopy (DIC).
2.2.Oocyst removal, lysis and DNA extraction
Any slide that was positive by fluorescent antibody staining
was processed for molecular analysis. This included all slides
that were reported positive and a few that were reported
negative, as some fluorescent material resembling oocysts
could not be confirmed by DAPI or DIC. There was no storage
of microscope slides as staining, counting and oocyst removal
were carried out on the same day, followed by storage of the
slide material (in lysis buffer) at ?80
oocyst lysis and DNAextraction were carried out as previously
described (Kim et al., 1992; Ruecker et al., 2005). DNA was
eluted from the Qiagen column with 100 mL of AE buffer
(QIAGEN, Mississauga, ON). Negative lysis buffer controls and
?C. Oocyst removal,
water research 46 (2012) 5135e5150
positive controls containing whole oocysts of Cryptosporidium
muris (Waterborne Inc., New Orleans, LA) in lysis buffer were
included in every run of lysis and DNA extraction.
2.3.Nested PCR-RFLP and DNA sequencing
Replicate nested PCR-RFLP was carried out according to
Ruecker et al. (2011) with primers and conditions previously
described by Xiao et al. (2001) and Ruecker et al. (2011). An
additional replicate (6th) was included for every sample and
seeded with 100 template copies of C. muris 18S rRNA plasmid
DNA to determine possible sample inhibition. Electrophoresis
of secondary PCR products was carried out on 1.2% agarose
gels and visualized using ethidium bromide staining.
All reactions positive by nested PCR were digested with the
restriction enzymes Ssp I, Vsp I and Dde I (Ruecker et al., 2005)
and the digested products were fractionated by electropho-
resis in 2% agarose and visualized by ethidium bromide
staining. PCR products that appeared to contain only single
species/genotypes, as indicated by the RFLP banding patterns
were selected for DNA sequence analysis. The secondary PCR
products were excised from the agarose gel and purified using
a QIAQuickgelextractionkit (QIAGEN). DNAwas eluted off the
QIAQuick column with 50 mL of EB buffer and concentrated by
a vacuum centrifuge to yield a concentration of between 20
and 50 ng/mL of DNA, as determined by a NanoDrop spectro-
photometer (Thermo Fisher Scientific Inc., Napean, ON). DNA
was sequenced bi-directionally in house using BigDye V3.1
(Applied Biosystems, Foster City, CA) with a Prism 3100-Avant
Genetic analyzer (Applied Biosystems) or a commercial DNA
sequencing company (Macrogen Inc., Seoul, Korea).
When low concentrations of DNA were recovered after gel
electrophoresis or only partial length sequences were ob-
tained during sequencing, re-amplification of the primary PCR
product was performed. All primary PCR products were stored
at ?80?C and could be re-amplified in 100 mL total volumes.
The secondary products were purified by QIAQuick PCR puri-
fication kit (QIAGEN) and eluted with 30 mL of EB buffer to
achieve higher yields of DNA for sequencing.
completely resolve mixtures of species/genotypes from the
water sample, the DNA extracts were re-analyzed by limiting
template dilution. Briefly, the repetitive nested PCR-RFLP was
repeated using 1 and 0.5 mL of DNA extract as template in the
primary PCR (Ruecker et al., 2011).
gene from GenBank sequences
Construction of a reference database for 18S rRNA
A database of reference sequences was created by extracting
>1600 Cryptosporidium 18S rRNA sequences from GenBank.
Due to the potential for erroneous sequence data generated
from cloning PCR products containing multiple species of
Cryptosporidium (Ruecker et al., 2011; Zhou et al., 2003), any
GenBank sequence originating from cloned PCR products was
eliminated from use. The sequences were aligned and any
sequences not covering the polymorphic region (nucleotide
position 613 to 810 of the Cryptosporidium parvum sequence
with GenBank accession number AF164102), short sequences
(<400 bp) and those containing more than 2 ambiguities
(uncalled bases, ¼N) were removed from the alignment.
Multiple GenBank entries with 100% identity were observed
for many sequences. One representative sequence was
selected based on the following criteria: i) the sequence had
previously been used as a reference sequence (Chalmers et al.,
2010a; Feng et al., 2007; Jiang et al., 2005b; Ruecker et al., 2007);
ii) the sourcefrom which it was obtainedwas well defined(e.g.
animal host before environmental); and iii) the length of the
sequence (longest available for the target region). The align-
ment was trimmed at both, the 50and 30ends to eliminate the
extra length resulting from full-length 18S rRNA sequences in
Phylogenetic analysis was carried out on the retained
sequences and the resulting trees were verified against
previously published phylogenetic analyses (Chalmers et al.,
2010a; Jiang et al., 2005b; Ruecker et al., 2007; Xiao et al.,
2004b; Xiao and Ryan, 2008; Yang et al., 2008). Sequences
that showed long branch lengths in the tree or formed
branches which did not fit with the previously described
taxonomic classification were manually examined in the
multiple sequence alignment. Some of these sequences dis-
playeda string ofconsecutive
conserved regions. Such sequences were omitted from the
database, as this is atypical within the genus and therefore
considered to be a result of sequencing errors. Following this
quality control or cleansing exercise, the reference database
consisted of 294 sequences, representing the known diversity
for 18S rRNA genes across the genus Cryptosporidium.
All raw sequence data was imported into Seqscape Version
2.6.1 (Applied Biosystems) and base calling was performed
using the KB base-caller. Bi-directional sequence data were
assembled and a consensus sequence was produced. High
quality consensus sequences were exported as FASTA-
formatted files for analysis using bioinformatics.
Multiple sequence alignments and phylogenetic analysis
were carried using software packages offered by the Sun
Center of Excellence for Visual Genomics (COE) at the
University of Calgary through web interface accessible via the
Secure Global Desktop. Multiple sequence alignments were
carried out using Clustal-W (Version 1.83) (Thompson et al.,
1994) or T-coffee (Version 1.37) (25). Phylogenetic analysis
was performed using the Phylogeny Inference Package (PHY-
LIP Version 3.69) (Felsenstein, 1989). Evolutionary distances
were calculated with DNADIST using the Kimura 2-parameter
model. Tree construction was carried by neighbor-joining
(NEIGHBOR program) using Eimeria tenella as the outgroup
(GenBank accession number AF026388). Bootstrapping was
carried out by creating 1000 replicates using SEQBOOT and
rooted trees were created by the CONSENSE program.
Consensus sequences and multiple sequence alignments
were visualized and edited using BioEdit Sequence Alignment
Editor (Version 184.108.40.206) (Hall, 1999). Phylogenetic trees were
viewed using TreeView (Version 1.6.6) (Page, 1996), with the
phylogram ladderized left and rooted at the outgroup.
Novel sequences generated in this study were deposited in
the GenBank database under accession numbers JQ178266 e
JQ178298.The sequence databaseusedforthis studyincluding
water research 46 (2012) 5135e5150
the GenBank sequences JQ178266 e JQ178298 is available upon
request to the corresponding author.
Between October 2004 and July 2009, 674 water samples
collected from multiple sites within the South Nation River
basin were evaluated for the presence of Cryptosporidium
oocysts. A total of 311 samples were positive by FITC staining
and confirmed by DAPI and/or DIC morphology with 196 of
these samples positive by molecular analysis. There were 42
water samples with Cryptosporidium-like FITC staining which
could not be confirmed as either positive or negative by DAPI
and/or DIC which were also processed by molecular analysis.
Of these and additional 6 samples were determined to have
Cryptosporidium present. USEPA Method 1623 and nested PCR
combined resulted in 317 samples with Cryptosporidium
present and an overall rate of occurrence of 47%. The average
frequency of occurrence amongthesitesvariedfrom 20to 72%
between sites, while the mean oocyst density ranged from
0.033 to 1.7 oocysts/L.
Molecular detection and sequence diversity from
A total of 717 positive nested PCR-RFLP results were obtained
from 202 water samples which had suspect Cryptosporidium
based on FITC staining. Under the conditions employed, PCR
inhibition was not observed in any of the environmental
samples analyzed. Of these, sequencing was not attempted
for 37 reactions, since RFLP patterns indicated an obvious
results were not obtained for 81 reactions due to multiple
sequence types in the same reaction, as determined by the
sequence electropherogram. Some sequences did not appear
mixed but were of poor quality due to low concentrations of
DNA recovered from the gel purification (typically < 20 ng/mL).
For these, primary products were re-amplified in 100 mL
reaction volumes to produce higher quantities of DNA for
sequencing. High quality bi-directional sequence was ob-
tained for 599 nested PCR reactions.
To determine the sequence diversity obtained from the
environment in this study, all sequences obtained were
aligned to each other. Clustal-W grouped sequences in the
alignment together based on similarity. Altering the input
order of the sequences and manual editing of the multiple
sequence alignment was often necessary to align all of the
homologous sequences to each other. Identical sequences
were identified in the alignment and a single sequence was
selected to represent the group of homologous sequences in
the subsequent analysis. The total sequence diversity across
all the reactions was 60 sequence types.
3.3. Phylogenetic analysis
The 60 sequences representing the diversity from the envi-
ronment were aligned with the reference database. The
Clustal-W alignment was manually edited to obtain a high
level of sequence homology by adjusting the alignment at the
50ends (addition of gaps to some sequences) and adjusting
homologous nucleotides around extended gaps. After the
manual alignment was complete, the ends of the alignment
were trimmed to form blunt ends on most of the sequences in
the alignment. Blunt ends of the alignment could not be ob-
tained for some of the sequences in the alignment due to the
fact that the original GenBank submissions were shorter than
the rest of the sequences in the alignment (i.e. Avian II and IV
genotypes), but their inclusion in the analysis was deemed
important in the analysis.
The final alignment covered726 nucleotides corresponding
to nucleotide positions 247 to 973 of C. parvum with GenBank
accession number AF164102. Any environmental sequence,
which displayed 100% homology to a sequence in the refer-
ence database, was removed from further analysis. The
accession number of the homologous sequence was traced
through the analysis to verify the species or genotype
To determine the phylogeny of the environmental isolates,
the alignment was subjected to distance analysis and tree
formation. The tree was examined for clustering of environ-
mental sequences onto clades with known species/genotypes.
Xiao et al. (2006) has previously described that variations
within a species or genotype (intragenotypic variation)
demonstrated less than 0.5% difference in their genetic
distances. The number of sequences was reduced to a single
representative on clades for species and genotypes with no
environmental isolates from this study (i.e. Cryptosporidium
bovis, Cryptosporidium xiaoi, Cryptosporidium suis, Cryptospo-
ridium ryane, Cryptosporidium felis, etc.). This reduced the
number of sequences in the alignment from 342 to 243 which
then had the gaps removed and again aligned. Manual editing
was followed by distance analysis and tree formation. It was
determined that 33 of the sequence types were novel, in that
they had no identical match in GenBank. This was confirmed
by BLAST search of the GenBank database.
The phylogenetic tree indicated that the majority (91%) of
the novel sequences clustered with known species and
genotypes, but there were some obvious grouping errors. In
particular, fox genotype sequences split the clade of muskrat
II genotype, and there was a repositioning of some previously
described vole, and muskrat II variant sequences onto other
clades, along with a few environmental isolates. Visual
inspection of the multiple sequence alignment indicated that
some variable positions were not aligned properly by Clustal-
W within the hypervariable region and it was concluded that
the most accurate tree was not obtained. For a more accurate
alignment the pair-wise T-coffee algorithm was subsequently
used. As T-coffee could not accommodate the large number of
sequences used in the Clustal-W alignment, the number of
sequences had to be reduced (w100 sequences) before the
pair-wise multiple sequence alignments could be performed.
To accurately identify all the unknown sequences, three
separate analyses were performed.
Fig. 1 shows the phylogenetic tree obtained from the
T-coffee alignment for 12 of the novel sequences (as indicated
by a CRY prefix) and a representative sequence from most of
the previously described species and genotypes (as indicated
water research 46 (2012) 5135e5150
by GenBank accession numbers). The 12 environmental
isolates (Fig. 1) retained the same branch positions as those
occupied in the two preliminary Clustal-W trees, which
included greater sequence diversity. Branch support for 9 of
the 12 isolates was indicated by bootstrap values greater than
50%. Nine sequences clustered with sequences of previously
described species or genotypes and the genetic distances
based on the Kimura-2-parameter model indicated that they
varied from the sequence they clustered with by less than
0.4%. A single variation of Cryptosporidium hominis, shrew,
W12, W18 and W19 genotypes were identified along with 4
variations of Cryptosporidium andersoni.
Two sequences (CRY1130 and CRY1532) clustered together
with skunk and chipmunk I genotypes as the nearest neigh-
bors. While these sequences have a high degree of similarity
to each other (99.7%), the genetic distance revealed CRY1130
to be 99.0% identical to either skunk or chipmunk I, which in
turn only show 99.0% similarity to each other. Genetic
distance supported the designation of a new genotype (W27).
Sequence for isolate CRY2319 was also designated a new
genotype (W28) as it resulted in the formation of a clade by
itself and showed a maximum of 96.3% sequence similarity to
any other species or genotype.
The preliminary phylogenetic analysis indicated that 21 of
the environmental sequences clustered on clades with a high
degree sequence heterogeneity. Sequences belonging to two
of the major nodes of the phylogenetic tree (noted in Fig. 1)
were partitioned out for separate multiple sequence align-
ments performed by T-coffee. Figs. 2 and 3 show the phylo-
genetic position of the remaining 21 environmental isolates.
Bootstrapping displayed less branch support for these highly
similar sequences than was observed in the more disparate
sequences used for T-coffee alignment in Fig. 1.
Seven of the unknown environmental isolates clustered on
the clade of muskrat II genotype as shown in Fig. 2. All novel
least one other member of the clade, based on the genetic
distance determined by the Kimura-2-parameter model. One
isolate (CRY1599), demonstrated movement between vole and
that were constructed. After the T-coffee alignment, CRY1599
was most similar to GenBank sequence AY737567 (isolated
from source water), displaying only a 0.3% difference based on
the genetic distance. Visual inspection of the alignment indi-
cated that there was greater sequence homology to the
muskrat II genotype in nucleotide positions outside the poly-
morphic region. Furthermore, sequence AY737567 differed
from GenBank accession number AY545548 by 0.1% and has
been previously identifiedas a muskrat II genotype (Jiang et al.,
2005b). As CRY1599 has at least 0.8% difference to any of the
sequences on the vole clade (Fig. 2) it was determined that
W alignments. The T-coffee alignment also eliminated the
splitting of the muskrat II clade by the fox genotype, which
often occurred in preliminary Clustal-W alignments.
Fig. 2 also shows the clustering of novel environmental
isolates with other species and genotypes of Cryptosporidium.
Isolate CRY1145 clustered with Cryptosporidium ubiquitum and
was 99.9% similar to known C. ubiquitum sequences (EF641018
and EU926598). Four isolates formed a clade with the W25
genotype, as shown in Fig. 2. These have a sequence similarity
of 99.9% to the environmental isolate describing this genotype
(EU825746). Isolate CRY2628 showed a 0.1% difference from
sequence EF641020 which originated from a vole.
The rest of the novel environmental isolates formed clades
with muskrat I or W12 genotypes, as shown in Fig. 3. Isolate
CRY1565 clustered with sequences of the W12 genotype and
demonstrated 99.9% sequence similarity to AY007254. Seven
novel environmental sequences clustered with muskrat I
genotype. These isolates demonstrated a maximum genetic
distance to other members of the clade by 0.14%. The W20
genotype (GenBank accession AY737588) appeared to cluster
with the muskrat I genotype. This demonstrates a phenom-
enon, which occurred most likely as a result of the limited
number of species and genotypes represented in the analysis.
The W20 was previously described as a unique genotype (Jiang
et al., 2005b), based on sequence polymorphisms near the 50
end, long branch length and a sequence similarity of only
99.1%. In this analysis the genetic distances ranged from 99.3
to 98.8% to the other members on the muskrat I clade.
Based on the sequence similarity and phylogenetic anal-
ysis, it was determined that the 60 sequence types identified
from the watersheds belonged to 17 known species or geno-
types, with the addition of two new genotypes (W27 and W28).
The species and genotype designations for all sequences ob-
tained from the South Nation River basin are summarized in
Table 1. The number in brackets behind the GenBank acces-
sion number or the laboratory ID in Table 1 indicates the
number of times that each sequence variation was isolated
from a nested PCR reaction and successfully sequenced.
3.4. Intragenotypic heterogeneity
Of the 60 sequences analyzed by phylogenetic analysis, 30
represented intragenotypic variations of known species and
genotypes. Variations of the same species were found in 31%
of the water samples, which tested positive by molecular
analysis. The species/genotype with the greatest number of
variations observed was the muskrat II genotype, with 11
variants. This increased the number of intragenotypic varia-
tions reported for the muskrat II clade to 18.
Muskrat II genotype sequences matching GenBank acces-
sion numbers AY545546, AY737567 and AY737571 were
detected in 40, 8 and 5 water samples respectively. Newly
identified variations CRY1492, CRY1599, CRY2341, CRY1059
and CRY1594 were detected in 6, 4, 3, 2 and 2 water samples
respectively. Isolates CRY1058, CRY1107 and CRY1112 were
each detected in a single water sample. More than 1 sequence
variation was detected in 10 of the 66 water samples, in which
the muskrat II genotype was detected.
Ten muskrat I sequences were observed in the study, with
7 sequences reflecting new heterogeneity within the geno-
type. Sequences matching AY737599, EF061293, and EF061288
were detected in 16, 4 and 3 water samples respectively.
Isolate CRY1115 was detected in 9 water samples, CRY1016 in
3, while CRY433, CRY441, CRY1019 and CRY1556 were each
detected in 1 water sample. Only 5 water samples had
multiple muskrat I genotype variations present.
There were 6 variations of C. andersoni observed. The two
most commonly observed sequences differed by a single T,
water research 46 (2012) 5135e5150
Mink genotype [EF641015]
Beaver genotype [EF641022]
Ferret genotype [AF112572]
C. wrairi [AF115378]
C. hominis [GQ983352]
C. cuniculus [FJ262725]
Opossum I genotype [AY120902]
C. fayeri [AF112570]
Horse genotype [FJ435962]
Deer mouse III genotype [EF641014]
C. parvum Iowa isolate [AF164102]
Mouse I genotype [AF112571]
Skunk genotype [AY120903]
Chipmunk I genotype [EF641026]
C. meleagridis [AF329187]
UK E4 [GQ183525]
Squirrel genotype [DQ295013]
Fox genotype [AY120907]
W25 genotype [EU825746]
Vole genotype [EF641020]
C. suis [EF489038]
C. ubiquitum [EF641018]
Deer mouse IV genotype [EF641019]
C. macropodum [AF513227]
Muskrat II genotype [AY545546]
Deer mouse I genotype [AY120905]
Deer mouse II genotype [EF641027]
Chipmunk III genotype [GQ121021]
W12 genotype [AY007254]
Opossum II genotype [AY120906]
C. canis [AB210854]
Bear genotype [AF247535]
Muskrat I genotype [AY120904]
W20 genotype [AY737588]
UK E2 [GQ183523]
SW3 genotype [HM015876]
C. felis [AF108862]
Hamster genotype [GQ121023]
W19 genotype [AY737585]
C. varanii [EF502042]
W18 genotype [AY737575]
Shrew genotype [EF641011]
W6 genotype [AF262331]
Rat II genotype [GQ121025]
Rat III genotype [GQ121026]
Rat I genotype [FJ205699]
Seal genotype [AY731234]
C. xiaoi [FJ896050]
C. bovis [AY741305]
C. ryanae [AY587166]
Pig II genotype [DQ182600]
Goose II genotype [AY504515]
Goose I genotype [AY120912]
C. baileyi [AF093495]
Avian I genotype [GQ227479]
Avian II genotype [DQ002931]
C. andersoni [AB089285]
C. muris [AF093498]
C. galli [AY737590]
Avian IV genotype [DQ650344]
Avian III genotype [HM116386]
C. serpentis [AF093502]
Tortoise genotype [AY120914]
W21 genotype [AY737589]
Chipmunk II genotype [EU096238]
Eimeria tenella [AF026388]
See Figure 2
See Figure 3
Scale = Substitutions per site
Fig. 1 e Phylogenetic relationship among Cryptosporidium isolates from the South Nation River basin and known sequences
from GenBank. Multiple sequence alignment was carried using T-coffee with evolutionary distances calculated using
water research 46 (2012) 5135e5150
representing the three or four T repeats which have been
previously described (Yang et al., 2008). The sequence
matching AB089285 (3 T repeat) was present in 60 water
samples, while the sequence matching FJ463171 (4 T repeat)
was present in 51 water samples. Thirty-four water samples
contained both of these sequence variations.
The sequence matching EU825746 (W25 genotype) was
detected in 24 water samples. In addition, four variations of
this genotype were observed. CRY984 was observed in five
water samples, CRY1468 was detected in four water samples
and CRY1547 and CRY1636 were detected in one water sample
each. Five water samples had multiple W25 genotype varia-
Three of the four variations of C. ubiquitum observed in the
watershed had matches in GenBank. The sequence matching
EF641018 was observed in nine water samples, with those
matching EU827413 and EF641017 each observed in two water
samples. A new variation (CRY1145) was identified in a single
The sequence matching GenBank accession EF061292 was
the most commonly observed W12 variation in the watershed,
present in 28 water samples. The sequence matching
AY007254 was presentin 4 water samples and isolate CRY1565
was detected in one water sample. We did not detect multiple
W12 genotype variations in any sample.
Four variations of the vole genotype were observed in the
watershed. The sequence matching EU641020 was observed in
five water samples, AY737562 in four water samples and
AY737563 in three water samples. A new variation CRY2628
was detected in one water sample. Multiple vole variations
were detected in three water samples.
CRY2330 represents the one detection of the shrew geno-
type and was identified as a new variant of the genotype. Two
sequence variations of the W19 genotype were found in
a single sample. One matched GenBank accession number
AY737585. The single W18 isolate (CRY2364) obtainedfrom the
watershed was identified as a new variant of the genotype.
C. hominis often varies in the length of the threonine repeat.
The single sequence (CRY1243) was obtained from 6 water
samples and contained an 11 T repeat which is common
among C. hominis sequence entries in GenBank.
Nation River basin
Prevalence of Cryptosporidium spp. in the South
A summary of occurrence of the species and genotypes
detected in the watershed is shown in Table 1. The number of
occurrences (column 3, Table 1) reflects the number of times
that a particular Cryptosporidium species or genotype was
observed in the 202 water samples that were positive for
molecular testing. The cumulative number of occurrences
identified across all species/genotypes exceeds 202 due to the
fact that multiple species and genotypes were often observed
in a single water sample. Seventy-eight of the 202 molecular-
positive water samples (38%) contained multiple species/
genotypes of Cryptosporidium. Two species or genotypes were
detected in 28% of the samples, while 10% had 3 or more
species or genotypes detected. Overall prevalence of a partic-
ular Cryptosporidium species or genotype in the basin was
calculated based on the number of water samples in which
a particular species/genotype was observed and dividedby the
total number of water samples analyzed from the South
Nation River (positive or negative for Cryptosporidium, n ¼ 674).
C. andersoni (commonly found in adult cattle) was detected
most often (75 water samples), and had an overall prevalence
rate in the basin of 11% (Table 1). The muskrat I and muskrat II
genotypes (muskrats and voles) were frequently detected in
the water samples and if combined the overall prevalence of
muskrat genotypes in the watershed was 15%, exceeding that
of C. andersoni (Table 1).
The genotypes W12 and W25 (no known hosts) also made
up a significant portion of the overall prevalence of Crypto-
sporidium, at 4.9% and 4.3%, respectively. Cryptosporidium bai-
leyi (birds) and C. ubiquitum (sheep and wild ruminants) had
a combined prevalence of 3.9% in water samples, with the rest
of the species and genotypes combining for only a few percent
of thetotal observations. The two speciesof particularinterest
to public health, C. hominis (humans) and C. parvum (calves or
humans) had low overall rates of prevalence (0.9 and 0.7%
respectively) in the source waters.
Generally Cryptosporidium species and genotype informa-
tion can be grouped based on broad categories of their major
known host, allowing for inferences regarding the most
probable contribution of Cryptosporidium to the watershed
(Fig. 4). Birds were identified as contributors to 8% of the total
Cryptosporidium species and genotypes detected (C. baileyi and
Cryptosporidium meleagridis). Humans were associated with 2%
(C. hominis), while cattles were the likely contributors of 35% of
the total species and genotypes (C. andersoni and C. parvum).
Wildlife comprised 55% of the Cryptosporidium identified,
indicating that wildlife were the major contributors of Cryp-
tosporidium within South Nation River basin.
The results can also be grouped based on the perceived risk
to humans (Fig. 5). C. hominis and C. parvum are the most
commonoutbreakspecies and posethe largest risk to humans
and together made up 4% of all molecular detections. The
potentially zoonotic species, C. meleagridis and C. ubiquitum
(reported in humans on a number of occasions) accounted for
8% of the species detected and pose a medium level of risk to
humans; however, these organisms have never been impli-
cated in a waterborne outbreak of cryptosporidiosis. The
lower risk species, C. andersoni and skunk genotype (infre-
quently reported in humans), made up 30% of the species and
genotypes detected. As the remaining species and genotypes
have not been associated with human infection, they are
currently considered to be of no risk to humans and were the
Kimura two-parameter model and the phylogentic tree was inferred by neighbor joining analysis. The outgroup (Eimeria
tenella AF026388) was used to root the tree. The isolates on the tree from this study are represented by a CRY prefix, while
the GenBank sequences are represented by their accession numbers. Numbers on the branches are percent bootstrap values
(>50%) using 1000 re-samples. Boxes with dotted borders indicate the nodes where more detailed diversity was included in
the phylogenetic analysis (See Figs. 2 and 3).
water research 46 (2012) 5135e5150
Deer mouse IV genotype
Muskrat II genotype
Chipmunk III genotype
Deer mouse II genotype
Deer mouse I genotype
Scale = Substitutions per site
Fig. 2 e Phylogenetic relationship among Cryptosporidium isolates from the South Nation River basin with increased
sequence diversity focusing on C. ubiquitum, muskrat II, vole and W25 genotypes. Multiple sequence alignment was carried
out using T-coffee with evolutionary distance calculated using Kimura two-parameter model and the phylogentic tree
inferred by neighbor joining analysis. The isolates on the tree from this study are represented by a CRY prefix, while the
GenBank sequences are represented by their accession numbers. The outgroup (C. baileyi AF093495) was used to root the
tree. Numbers on the branches are percent bootstrap values (>50%) using 1000 re-samples.
water research 46 (2012) 5135e5150
Muskrat I genotype
Opossum II genotype
Scale = Substitutions per site
Fig. 3 e Phylogenetic relationship among Cryptosporidium isolates from the South Nation River basin with increased
sequenced diversity focusing on muskrat I and W12 genotypes. Multiple sequence alignment was carried out using T-coffee
with evolutionary distance calculated using Kimura two-parameter model and the phylogentic tree inferred by neighbor
joining analysis. The isolates on the tree from this study are represented by a CRY prefix, while the GenBank sequences are
represented by their accession numbers. The outgroup (C. baileyi AF093495) was used to root the tree. Numbers on the
branches are percent bootstrap values (>50%) using 1000 re-samples.
water research 46 (2012) 5135e5150
predominant category detected (58%). Of the 202 water
samples from which genotyping results were obtained, 51%
contained only no risk species and genotypes of Cryptospo-
ridium. The overall prevalence of high-risk species (C. hominis
and C. parvum) in the watershed was 1.6%.
Numerous studies have been carried out to identify the
species and genotypes of Cryptosporidium present in source
waters (Jellison et al., 2002; Jiang et al., 2005b; Nichols et al.,
2010; Ruecker et al., 2007; Ward et al., 2002; Yang et al.,
2008). Source tracking Cryptosporidium from source waters
has become increasingly complex, due to the large genetic
diversity observed in the 18S rRNA gene sequences. Identifi-
cation of novel environmental isolates requires accuracy in
the sequence data, a non-redundant 18S rRNA sequence
reference database, properly interpreted phylogenetic anal-
ysis and knowledge of hosteparasite relationships. In this
study, speciesand genotypedesignations wereassignedto the
novel sequences based on a number of criteria: i) their posi-
tion on a phylogenetic tree; ii) branch length, iii) boot-strap
value, iv) physical alignment, and v) genetic distance.
Sequence repositories are only as accurate as the quality of
data that is deposited in them. Random sequencing errors
(poor assembly, degenerate sequence, and partial sequences)
and PCR mediated artifacts (chimeras, deletions and point
Table 1 e Rates of detection (PCR reactions), occurrence (water samples), prevalence in environment (%) and major hosts for
the species and genotypes of Cryptosporidium detected in the South Nation River basin.
match or lab ID
(number of detectionsa)
Number of occurrences
in water samples that
were positive for
by molecular testing
(n ¼ 202)
Major known hosts
C. andersoni AB089285 (119), FJ463171 (92),
CRY1024 (1), CRY1219 (1),
CRY1627 (1), CRY2384 (2)
EF641018 (13), EU827413 (4),
EF641017 (2), CRY1145 (2)
AF329187 (5), EU825750 (1)
AF164102 (7), AF308600 (2)
AY545546 (67), AY737567 (10),
AY737571 (6), CRY1492 (10),
CRY1599 (6), CRY2341 (3),
CRY1058 (2), CRY1059 (2),
CRY1594 (2), CRY1107 (1),
AY737599 (25), EF061293 (6),
EF061288 (3), CRY1115 (15),
CRY1016 (8), CRY1556 (2),
CRY433 (1), CRY441 (1),
CRY1019 (1), CRY1114 (1)
Sheep, deer, rodents
Muskrat II genotype
Muskrat I genotype
38 5.6Muskrats, voles
Deer mouse III
4 0.6 Deer mice
EF061289 (4), EF061290 (3)
EF641020 (6), AY737562 (6),
AY737563 (3), CRY2628 (1)
EF061292 (43), AY007254 (4),
EU825746 (31), CRY984 (13),
CRY1468 (3), CRY1547 (1),
AY737585 (1), CRY1515 (1)
CRY1130 (6), CRY1532 (1)
Novel genotype (W27)
Novel genotype (W28)
a Numbers of detections represents the number of times that sequence was obtained by nested PCR and is reflective of the 599 reactions that
b Prevalence is based on the number of occurrences of a particular species/genotype of Cryptosporidium in water samples (column 3) divided by
the total number of water samples analyzed in the South Nation Watershed (n ¼ 674).
water research 46 (2012) 5135e5150
mutations) are known to plague the public sequence reposi-
tories (Ashelford et al., 2006; vonWintzingerode et al., 1997). It
is the responsibility of those submitting sequence to public
repositories to verify and validate the accuracy of their
sequence data. As artifacts in sequences obtained from
a natural sample can never truly be known, the interpretation
ofsequencemustfocusonexperimental replication (Qiu et al.,
2001; Stiller et al., 2006).
Of the 33 novel sequences obtained from the study site, 15
of the sequences were obtained from multiple nested PCR
products. Replication of an identical sequence from inde-
pendent PCR reactions was used to verify the accuracy of the
sequence, by ruling out the possibility of random sequencing
errors or PCR mediated artifacts. For the 18 novel sequences
identified, which were detected in a single nested PCR reac-
tion, validation becomes difficult. These sequences were
replicated by at least two separate amplifications of the
primary PCR product. Sequence quality and assembly of the
forward and reverse sequencing reaction were carefully
scrutinized, along with visual observation of the electrophe-
rogram and the base-calling at the positions, where poly-
morphisms were detected. This process minimized the
occurrence of random sequencing errors or PCR artifacts that
might have occurred at the secondary PCR level.
Studies dealing with source tracking Cryptosporidium from
raw waters have frequently identified the existence of new
genotypes (Chalmers et al., 2010a; Jiang et al., 2005b; Nichols
et al., 2010; Yang et al., 2008). This study identified two new
genotypes designated W27 and W28. The W27 genotype was
identified in seven nested PCR reactions from five water
samples and therefore unlikely to be the result of Taq poly-
merase or sequencing errors. The W28 sequence was identi-
fied in a single nested PCR reaction from a water sample,
wherethe otherreplicateswereallof theC. andersoni type.PCR
mediated recombination of ribosomal rRNA resulting in
chimeric sequence has been documented (Ashelford et al.,
2006; Qiu et al., 2001; vonWintzingerode et al., 1997). Recom-
bination between multiple species of Cryptosporidium has been
shown through cloning experiments, but there is no evidence
of chimeric sequences obtainedfrom direct sequencing of PCR
products (Ruecker et al., 2011; Zhou et al., 2003). For such an
error to occur, it would mean that multiple templates must be
present in the initial PCR and the resulting sequence would be
either of poor quality with many mixed bases, have under-
lying sequence in the electropherogram or be unreadable. The
W28 sequence was amplified from the primary PCR product
and sequenced bi-directionally on three different occasions.
There was no discrepancy in the sequence on the three
occasions and the consensus sequence was of high quality,
with no mixed bases. As C. andersoni was repeatedly obtained
from the same water sample, a PCR mediated chimeric
sequence would likely contain a portion of the sequence,
which would be similar to C. andersoni. Distance analysis
determined that the W28 sequence was 8.6% different to C.
andersoni and examination
homology only in the regions that were highly conserved
within the entire Cryptosporidium genus.
After careful evaluation, the 33 sequences with no Gen-
Bank matches were determined to be accurate. This was
based on replicate sequence analysis, the redundancy of
sequence polymorphism, the location of insertions/deletions
and the improbability that newly identified genotypes were
a result of PCR mediated recombination.
Source water samples have contributed significantly to the
known diversity of 18S rRNA sequence in Cryptosporidium, as
well as the identification of new genotypes and intragenotypic
variations of known species or genotypes (Chalmers et al.,
2010a; Jiang et al., 2005b; Nichols et al., 2010; Ruecker et al.,
2007; Yang et al., 2008). In all cited studies, phylogenetic
analysis was used to partition novel Cryptosporidium sequence
data into intragenotypic variations of known species/geno-
types and/or assign novel genotype designations.
The sequence analysis process begins by comparing the
unknown sequence to a database of DNA sequence informa-
tion from public databases. The National Center for Biotech-
nology Information (NCBI), houses the sequence reference
database GenBank, and offers a web interface tool known as
BLAST. Databases such as GenBank pose a number of chal-
lenges for its users due to: i) rapid growth in number of
sequences, ii) sequence redundancy iii) inclusion of poor
quality and erroneous sequence data, iv) inaccuracy in
annotation and nomenclature, v) continued existence of
outdated taxonomic nomenclature, and vi) in some cases,
a lack of primary literature associated with the entry.
of the alignmentshowed
Fig. 5 e Distribution of human risk (high, medium, low and
no risk) associated with species and genotypes of
Cryptosporidium detected in the South Nation River basin
Avian associated species/genotypes
(20 water samples)
Human associated species (6 water
Livestock associated species (90
Wildlife associated species/genotypes
(144 water samples)
Fig. 4 e Distribution of human, avian, cattle and wildlife
hosts detected in the South Nation River basin 2004e2009.
water research 46 (2012) 5135e5150
BLAST is often the first approach used for analysis of
sequence data. BLAST is used to identify regions of local
similarity between the unknown sequence and those in the
database and calculates a level of significance for each match.
Although fast and simple to use, BLAST is plagued by the
challenges listed above and should not be used for more than
general screening purposes. In some analysis, the most
divergent regions are excluded in the sequence alignment
generated by BLAST. The most accurate methods involve
a DNA sequence alignment of the unknown sequence to
a comprehensive database of known good-quality sequences,
followed by phylogenetic analysis to determine the relation-
ship between species and genotypes. The result of this anal-
ysis is a phylogenetic tree, representing the “relatedness” of
the sequences included in the study.
The alignment of the DNA sequence highly affects the
phylogenetic position of a sequence. The role of the sequence
alignment is to organize sequences in a way that only
homologous residues appear in the same column of the
alignment (Olsen and Woese, 1993). When different align-
ments are compared, there are generally regions of agreement
and regions of disagreement, but what is important are the
regions of greatest conservation (Olsen and Woese, 1993).
Clustal-W, an older alignment tool, which does not take into
account structural properties of the sequences in the align-
ment, was used primarily when large numbers of sequences
were included in the alignment. This included both database
construction and the initial iterations of alignments used for
determining theidentity ofsequencesfrom thestudy. Thereis
really no such thing as a single correct alignment and all
computer-generated alignments require some manual editing
to minimize the number of multi-state positions within the
alignment (Clayton et al., 1995; Olsen and Woese, 1993).
In general, T-coffee alignments, which used a more
sophisticated alignment algorithm, required less manual
editing than the Clustal-W alignments, but it was impossible
to accommodate the large number of sequences included in
the database simultaneously. Therefore, sub-datasets were
created for use with T-coffee. Forcing the minimum number
of variable positions within an alignment allows for the most
conservative estimate of within- species variability (Clayton
et al., 1995). This is important in understanding intra-
genotypic variations within species of Cryptosporidium. Also
stated by Olsen and Woese, the sequences included in the
alignment or the order in which sequences are presented into
the alignment program can affect the ability to define the
regions of greatest alignment certainty (Olsen and Woese,
1993). It was for these reasons that many iterations of align-
ment were carried out, evaluated and the automatic align-
ments were subsequently edited manually to obtain the
minimal number of variable positions within the alignment.
Multiple sequence alignments are needed to generate
a reference sequence database. Altschul et al. (1994) describes
sequence redundancy as a major problem, which can obscure
novel matches in database searches and therefore recom-
mends the useof smaller
McMahon and Sanderson (2006) stated that different lengths
of sequences and degree of overlap in sequences (submission
from different primer sets) can be problematic in phylogenetic
analyses. For these and other reasons previously stated, it
became necessary to critically evaluate the Cryptosporidium
18S rRNAdata from GenBankin orderto identifyand eliminate
entries with possible sequence errors, taxonomy errors,
annotation errors, insufficient length and redundancy from
the dataset.The output ofthe processes used in this studywas
a ‘cleansed’ database, suitable for the purpose of multiple
sequence alignment of the sequences from environmental
Phylogeny can be used for the identification of environ-
mental isolates of Cryptosporidium. The approach of using
multiple sequence alignments and evolutionary distance has
been used by many researchers for the identification of
Cryptosporidium species and genotypes from environmental
samples (Chalmers et al., 2010a; Jellison et al., 2002>, 2009;
Jiang et al., 2005b; Nichols et al., 2010; Ruecker et al., 2007;
Ryan et al., 2005; Yang et al., 2008). Phylogenetic positions and
the degree of sequence similarity, with minimal consideration
of bootstrap values, have been primarily used to identify the
species or genotypes detected in the environment (Chalmers
et al., 2010a; Jiang et al., 2005b; Nichols et al., 2010).
Re-sampling by bootstrapping is commonly carried out as
a means of estimating the accuracy of the topology of a tree
(Felsenstein, 1988). The majority of trees in the scientific
literature used to describe the evolutionary relationships of
Cryptosporidium have bootstrap values attached, but poor
explanation of their interpretation. Early publications stated
that clades separated from each other with high bootstrap
values were statistically reliable (Xiao et al., 1999c, 1999a,
2002). In the early cases where only a few species or geno-
types were examined (15 or 16 species), the majority of the
branching of the tree was supported by bootstrap values of
greater than >50% (Xiao et al., 1999c, 1999a). By 2002, when
known species and genotypes had increased to 32 sequences,
fewer branches were supported by bootstrapping (Xiao et al.,
2002). Over the years, as the number of sequences used in
phylogenetic trees increased to between 60 and 90 represen-
tatives, there was an apparent decrease in the number of
branches of the phylogenetic tree supported by bootstrap
values (Chalmers et al., 2010a; Fayer et al., 2010; Feng et al.,
2007; Jiang et al., 2005b; Lv et al., 2009; Nichols et al., 2010;
Ruecker et al., 2007; Xiao and Feng, 2008). In comparison, Fig. 1
from this study demonstrates similar branch support by
bootstrapping as with a large number of representatives. The
inability to resolve internal branching using bootstrapping,
such as was observed on the trees from Figs. 2 and 3, has also
been demonstrated for the 16S rRNA of some bacterial species
(Cilia et al., 1996). It was concluded by Cilia et al. (1996) that
related bacterial species to support bootstrapping relatedness.
This may also be true in the case of Cryptosporidium.
The absence of a high proportion of branches supported by
high bootstrap values does not invalidate the use of phylo-
genetic trees for the identification of unknown sequences
from the environment. The tree topology from the phyloge-
netic analysis in this study is consistent with the overall
topologies of the trees of other published studies (Chalmers
et al., 2010a; Feng et al., 2007; Lv et al., 2009). Similar major
nodes are supported by bootstrapping including: intestinal
from gastric parasites; W18 genotype from shrew genotype; C.
ubiquitum from deer mouse IV genotype; Cryptosporidium fayeri
water research 46 (2012) 5135e5150
from opossum I genotype; and the separation of goose I and II
genotypes from C. ryane, C. bovis, C. xiaoi and pig II genotype.
On branches with no bootstrap support, there is some varia-
tion in the order of species and genotypes on a branch;
however groupings generally remain consistent. For example,
the clustering of C. hominis, C. parvum, Cryptosporidium wrairi,
C. meleagridis, Cryptosporidium cuniculus, horse, mink, ferret,
skunk, mouse I and chipmunk I genotypes observed on the
tree from this study is consistent with other phylogenetic
trees (Chalmers et al., 2010a; Feng et al., 2007; Lv et al., 2009).
The overall topologies of the trees displaying a high level of
intra-genotypic heterogeneity in this study (Figs. 2 and 3) are
consistent with the topology displayed in a tree from
a previous watershed study (Jiang et al., 2005a). There is a high
degree of uniformity which occurs between trees from inde-
Sequence diversity within species is known to exist for
many species and genotypes of Cryptosporidium. This diversity
is believed to be the result of heterogeneous copies of the 18S
ribosomal RNA gene, host adaptation and geographical
distribution. Different isolates of C. hominis, C. parvum, C. muris
and Cryptosporidium serpentis were shown to contain sequence
polymorphisms at the 18S rRNA gene location (Xiao et al.,
1999a). At the same time, it is accepted that heterogeneity
occurs among different copies of the 18S rRNA gene (Le Blancq
et al., 1997; Xiao et al., 1999b) which also adds to the sequence
diversity of the various species and genotypes. For all the
Cryptosporidium species or genotypes detected in the water-
shed at a high frequency, one sequence for a particular
species/genotype was often dominant, with the exception of
C. andersoni, where two sequences were found in almost equal
proportions. For 10 of the 13 species or genotypes, the most
frequently detected sequence matched one which originated
from a known animal host (C. andersoni, C. baileyi, C. ubiq-
uitium, C. hominis, C. meleagridis, C. parvum, muskrat II, deer
mouse III, vole and skunk genotypes).
sequences of muskrat I, fox and shrew genotypes have only
been detected in environmental samples. With limited
molecular epidemiology available for many species and
genotypes, it is difficult to know or hypothesize whether
sequence variation represents diversity at the species level or
if the variation simply reflects the phenomenon of heteroge-
neous copies of the ribosomal RNA (Le Blancq et al., 1997; Xiao
et al., 1999b). It is possible that the minor heterogeneous copy
of a gene is more likely to be isolated during watershed
studies, where DNA templates are often limiting, as opposed
to being quenched by the major copy, as could happen in
animal investigations. This is supported by the large number
of sequence variations, in this and other studies, which have
been associated with an animal host based on phylogenetic
analysis, but have not been previously identified in the host
(Jiang et al., 2005b; Yang et al., 2008). While Feng et al. (2007)
determined that sequence
different copies of the gene, Fayer et al. (2010) concluded that
there is limited biological significance to sequence variation
within a species at this single genetic locus. The sequence
diversity within a species/genotype has been shown to have
limited impact on the phylogenetic analysis (Xiao et al., 1999b)
and therefore should not deter its use in source tracking
efforts with Cryptosporidium.
During source tracking, animal sources are inferred based
on the major known hosts associated with a species or
genotype. It is generally agreedthat host-specificity withinthe
genus Cryptosporidium is a relative term. Some species/geno-
types appear to have strict host-specificity, whereas others
appear to be more promiscuous in their host preference (Fayer
et al., 2010; Feng et al., 2011; Xiao and Ryan, 2008). C. hominis is
considered to be largely specific for humans, while C. ubiq-
uitum demonstrates a broader host range (Xiao and Feng,
2008). The key concept of microbial source tracking is to
identify and differentiate putative point source and non-point
faecal pollution in source waters. This was carried out based
on broad categories or sources, which have been determined
based on the major host sources identified for the sequences
isolated from the watershed. As an animal species can play
host to more than one Cryptosporidium species, the uncertain
source of some Cryptosporidium species/genotypes can be
further inferred by the presence or absence of other Crypto-
sporidium species/genotypes in water samples.
Source tracking of Cryptosporidium is also important for
human health risk assessment. Both an understanding of the
current taxonomy with respect to major hosteparasite rela-
tionships and an understanding of the watershed must be
considered when grouping the species and genotypes based
on source. C. andersoni made up the largest portion of detec-
tion by a single host found in the watershed, typical of find-
ings reported in other watershed studies (Nichols et al., 2010;
Ruecker et al., 2007; Yang et al., 2008). In the fall of 2004
(Ruecker et al., 2007), we observed that C. andersoni accounted
for approximately 50% of Cryptosporidium detections in this
basin. In the long term water quality monitoring study
described here, the proportion of samples positive for C.
andersoni was reduced to 25%, indicating that there may be
temporal variation in the sources of the parasite. For the
purpose of this study C. parvum was grouped with the cattle
derived portion of the parasite, according to its major known
host, pre-weaned calves (Fayer et al., 2006; Santin et al., 2008).
However its detection in the watershed was minimal. C.
ubiquitum was considered in the livestock fraction as its major
host is sheep (Fayer et al., 2010). No other livestock-adapted
species were detected in the watershed.
C. hominis was considered specific to human source
contamination, while C. baileyi and C. meleagridis made up the
avian fraction. As has been done in other studies, the
remainder of Cryptosporidium species detected were consid-
ered to be of wildlife origin (Jiang et al., 2005b; Ruecker et al.,
2007; Yang et al., 2008). The genotypes of unknown host
sources were presumed to originate from wildlife, as the
Cryptosporidium species of cattle and other agricultural live-
stock species have typically been well characterized (Jiang
et al., 2005b; Santin and Trout, 2008).
The wildlife fraction of Cryptosporidium is considerably
higher in this study than in the Potomac River watershed or
a Scottish study (Nichols et al., 2010; Yang et al., 2008). The
major portions of the wildlife contributors were muskrat I and
II genotypes, as well as the W12 and W25 genotypes. The
muskrat I and II genotypes have been commonly found in
other environmental samples in the United States, Scotland,
Wales and Canada (Chalmers et al., 2010a; Jiang et al., 2005b;
Nichols et al., 2010; Ruecker et al., 2007; Yang et al., 2008). The
water research 46 (2012) 5135e5150
W12 genotype was found previously in the South Nation River
basin, as well as watersheds in New York and the Potomac
(Jiang et al., 2005b; Ruecker et al., 2007; Yang et al., 2008);
however this genotype has not been detected in Scotland or
Wales. Similarly, W25 was found rather commonly in this
study and also in the Potomac study, but not reported in
Scotland or Wales. The number of genotypes with no known
host, indicate that our understanding of hosteparasite rela-
tionships is far from complete.
C. parvum and C. hominis are associated with most water-
borne, food-borne and contact associated outbreaks of cryp-
tosporidiosis (Xiao and Ryan, 2008) and therefore these
species represent the highest risk to humans. C. hominis
represents a source of human to human transmission
(anthroponotic), while C. parvum could represent either an
anthroponotic or zoonotic (animal to human) source of
transmission (Xiao, 2010). These species had an overall prev-
alence rate in the watershed of 1.6%. The low rates of preva-
lence of Cryptosporidium presenting a high risk to humans is
consistent with other watershed studies carried out in the
United Stated (Jiang et al., 2005b; Yang et al., 2008).
Of the numerous species and genotypes detected in the
South Nation River basin only three species and one genotype
are considered to be zoonotic: C. meleagridis, C. ubiquitum, C.
andersoni and the skunk genotype. C. meleagridis has been
documented in human infections worldwide, but represents
only a fraction of the human infections as compared to C.
hominis and C. parvum (Xiao and Feng, 2008). With the major
host of C. meleagridis identified to be turkeys and other avian
hosts, this species is considered to be potentially zoonotic to
humans, but its relatively high prevalence in humans (third
after C. parvum and C. hominis) suggests that it presents
considerable risk to humans, if present in drinking water. C.
ubiquitum (formerly cervine genotype) has been identified in
humans in Canada, New Zealand, England, Slovenia, United
Kingdom and United States (Fayer et al., 2010). C. meleagridis
and C. ubiquitum were grouped as zoonotic species posing
a medium risk to humans displaying a low rate of occurrence
among the species detected. C. andersoni was grouped as
a low-risk potential zoonotic species, based on the detection
of this species three patients in England (Leoni et al., 2006).
However, clinical analysis of these infections relied on
amplification and sequencing of a 390 bp fragment of the 18S
rRNA target from cloned PCR products (two conditions which
exclude the sequence from analysis in this study). It should be
noted that in other long-term studies carried out in the United
Kingdom using similar PCR and sequencing methods to those
used here, no other occurrences of human-derived C. ander-
soni have been reported (Chalmers et al., 2009; Nichols et al.,
2006; Robinson et al., 2008). Nevertheless, the possibility that
C. andersoni carries some form of minimal human health risk
cannot be ignored, and therefore this species was grouped as
a low risk zoonotic parasite in this study. The skunk genotype
has also been identified in the feces of a few humans in the
United Kingdom (Robinson et al., 2008) and was therefore also
classified as a potential zoonotic pathogen of low risk to
humans. The species and genotypes considered to be poten-
tially zoonotic and of low-risk to humans make up 30% of the
overall proportion of Cryptosporidium detected in the water-
shed. All of the other genotypes detected in this study are
considered to be non pathogenic as there have, to date, been
noreportsof human infection
(Chalmers et al., 2010b; Chalmers and Davies, 2010; Elwin
et al., 2011; Xiao et al., 2004a; Xiao, 2010; Xiao and Fayer,
2008; Xiao and Ryan, 2008). This category poses no known
risk tohumansandmakesupthe largest portionofthe species
and genotypes detected at 58%.
This study demonstrates the utility of phylogenetic analysis
for identifying the species and genotypes of Cryptosporidium
from environmental samples. The data represents a long-
term study carried out in the South Nation watershed which
builds upon previous work published for the South Nation
watershed (Ruecker et al., 2007). The long-term study reflects
an increased diversity in the species and genotypes detected
and differences in prevalence rates, however the overall
conclusions remain consistent. Based on the overall preva-
lence of C. hominis and C. parvum (1.6%) the risk by Cryptospo-
ridium to human health in the South Nation watershed was
minimal. Also, known hosteparasite relationships indicate
that the majority of the species and genotypes found in the
watershed were derived from adult cattle and various wildlife
This study was funded by Agriculture and Agri-Food Canada
(Government of Canada) through the National Water Quality
Research Initiative and by Environment Canada (Government
of Canada) through the National Agri-Environmental Stan-
dards Initiative [NAESI]. Funding support was also obtained
through the Alberta Water Research Institute [AWRI]. NJR was
supported by a scholarship from the Natural Sciences and
Engineering Research Council.
r e f e r e n c e s
Altschul, S.F., Boguski, M.S., Gish, W., Wootton, J.C., 1994. Issues
in searching molecular sequence databases. Nature Genetics 6
Ashelford, K.E., Chuzhanova, N.A., Fry, J.C., Jones, A.J.,
Weightman, A.J., 2006. New screening software shows that
most recent large 16S rRNA gene clone libraries contain
chimeras. Applied and Environmental Microbiology 72 (9),
Chalmers, R.M., Davies, A.P., 2010. Minireview: clinical
cryptosporidiosis. Experimental Parasitology 124 (1), 138e146.
Chalmers, R.M., Elwin, K., Thomas, A.L., Guy, E.C., Mason, B.,
2009. Long-term Cryptosporidium typing reveals the aetiology
and species-specific epidemiology of human cryptosporidiosis
in England and Wales, 2000 to 2003. Euro Surveillance 14 (2).
Chalmers, R.M., Robinson, G., Elwin, K., Hadfield, S.J., Thomas, E.,
Watkins, J., Casemore, D., Kay, D., 2010a. Detection of
Cryptosporidium species and sources of contamination with
Cryptosporidium hominis during a waterborne outbreak in north
west Wales. Journal of Water and Health 8 (2), 311e325.
water research 46 (2012) 5135e5150
Chalmers, R.M., Smith, R., Elwin, K., Clifton-Hadley, F.A.,
Giles, M., 2010b. Epidemiology of anthroponotic and zoonotic
human cryptosporidiosis in England and Wales, 2004e2006.
Epidemiology and Infections, 1e13.
Cilia, V., Lafay, B., Christen, R., 1996. Sequence heterogeneities
among 16S ribosomal RNA sequences, and their effect on
phylogenetic analyses at the species level. Molecular Biology
and Evolution 13 (3), 451e461.
Clayton, R.A., Sutton, G., Hinkle Jr., P.S., Bult, C., Fields, C., 1995.
Intraspecific variation in small-subunit rRNA sequences in
GenBank: why single sequences may not adequately represent
prokaryotic taxa. International Journal of Systematic
Bacteriology 45 (3), 595e599.
Egyed, Z., Sreter, T., Szell, Z., Varga, I., 2003. Characterization of
Cryptosporidium spp. e recent developments and future needs.
Veterinary Parasitology 111 (2e3), 103e114.
Elwin, K., Hadfield, S.J., Robinson, G., Chalmers, R.M., 2011. The
epidemiology of sporadic human infections with unusual
cryptosporidia detected during routine typing in England and
Wales, 2000e2008. Epidemiology and Infections, 1e11.
Fayer, R., Santin, M., Macarisin, D., 2010. Cryptosporidium
ubiquitum n. sp. in animals and humans. Veterinary
Parasitology 172 (1e2), 23e32.
Fayer, R., Santin, M., Trout, J.M., Greiner, E., 2006. Prevalence of
species and genotypes of Cryptosporidium found in 1e2-year-
old dairy cattle in the eastern United States. Veterinary
Parasitology 135 (2), 105e112.
Felsenstein, J., 1988. Phylogenies from molecular sequences:
inference and reliability. Annual Review of Genetics 22,
Felsenstein, J., 1989. PHYLIP e phylogeny inference package
(Version 3.2). Cladistics 5, 164e166.
Feng, Y., Alderisio, K.A., Yang, W., Blancero, L.A., Kuhne, W.G.,
Nadareski, C.A., Reid, M., Xiao, L., 2007. Cryptosporidium
genotypes in wildlife from a New York watershed. Applied and
Environmental Microbiology 73 (20), 6475e6483.
Feng, Y., Lal, A.A., Li, N., Xiao, L., 2011. Subtypes of
Cryptosporidium spp. in mice and other small mammals.
Experimental Parasitology 127 (1), 238e242.
Hall, T.A., 1999. BioEdit: a user-friendly biological sequence
alignment editor and analysis program for Windows 95/98/NT.
Nucleic Acids Symposium Series 41, 95e98.
Jellison, K.L., Hemond, H.F., Schauer, D.B., 2002. Sources and
species of Cryptosporidium oocysts in the Wachusett Reservoir
watershed. Applied and Environmental Microbiology 68 (2),
Jellison, K.L., Lynch, A.E., Ziemann, J.M., 2009. Source tracking
identifies deer and geese as vectors of human-infectious
Cryptosporidium genotypes in an urban/suburban watershed.
Environmental Science & Technology 43 (12), 4267e4272.
Jiang, J., Alderisio, K.A., Singh, A., Xiao, L., 2005a. Development of
procedures for direct extraction of Cryptosporidium DNA from
water concentrates and for relief of PCR inhibitors. Applied
and Environmental Microbiology 71 (3), 1135e1141.
Jiang, J., Alderisio, K.A., Xiao, L., 2005b. Distribution of
Cryptosporidium genotypes in storm event water samples from
three watersheds in New York. Applied and Environmental
Microbiology 71 (8), 4446e4454.
Kim, K., Gooze, L., Petersen, C., Gut, J., Nelson, R.G., 1992.
Isolation, sequence and molecular karyotype analysis of the
actin gene of Cryptosporidium parvum. Molecular and
Biochemical Parasitology 50 (1), 105e113.
Le Blancq, S.M., Khramtsov, N.V., Zamani, F., Upton, S.J.,
Wu, T.W., 1997. Ribosomal RNA gene organization in
Cryptosporidium parvum. Molecular and Biochemical
Parasitology 90 (2), 463e478.
Leoni, F., Amar, C., Nichols, G., Pedraza-Diaz, S., McLauchlin, J.,
2006. Genetic analysis of Cryptosporidium from 2414 humans
with diarrhoea in England between 1985 and 2000. Journal of
Medical Microbiology 55 (Pt 6), 703e707.
Lv, C., Zhang, L., Wang, R., Jian, F., Zhang, S., Ning, C., Wang, H.,
Feng, C., Wang, X., Ren, X., Qi, M., Xiao, L., 2009.
Cryptosporidium spp. in wild, laboratory, and pet rodents in
china: prevalence and molecular characterization. Applied
and Environmental Microbiology 75 (24), 7692e7699.
McMahon, M.M., Sanderson, M.J., 2006. Phylogenetic supermatrix
analysis of GenBank sequences from 2228 papilionoid
legumes. Systematic Biology 55 (5), 818e836.
Nichols, G.L., Champion, A., Sopwith, W., Regan, M., Hunter, C.A.,
Grenfell, P., Harrison, F., Lane, C., 2006. Cryptosporidiosis:
a Report on the Surveillance and Epidemiology of
Cryptosporidium Infection in England and Wales. DWI 70/2/201.
Drinking Water Inspectorate, UK.
Nichols, R.A., Connelly, L., Sullivan, C.B., Smith, H.V., 2010.
Identification of Cryptosporidium species and genotypes in
Scottish raw and drinking waters during a one-year
monitoring period. Applied and Environmental Microbiology
76 (17), 5977e5986.
Olsen, G.J., Woese, C.R., 1993. Ribosomal RNA: a key to phylogeny.
FASEB Journal 7 (1), 113e123.
Page, R.D., 1996. TreeView: an application to display phylogenetic
trees on personal computers. Computer Applications in the
Biosciences 12 (4), 357e358.
Qiu, X., Wu, L., Huang, H., McDonel, P.E., Palumbo, A.V.,
Tiedje, J.M., Zhou, J., 2001. Evaluation of PCR-generated
chimeras, mutations, and heteroduplexes with 16S rRNA
gene-based cloning. Applied and Environmental Microbiology
67 (2), 880e887.
Robinson, G., Elwin, K., Chalmers, R.M., 2008. Unusual
Cryptosporidium genotypes in human cases of diarrhea.
Emerging Infectious Diseases 14 (11), 1800e1802.
Ruecker, N.J., Bounsombath, N., Wallis, P., Ong, C.S., Isaac-
Renton, J.L., Neumann, N.F., 2005. Molecular forensic profiling
of Cryptosporidium species and genotypes in raw water.
Applied and Environmental Microbiology 71 (12), 8991e8994.
Ruecker, N.J., Braithwaite, S.L., Topp, E., Edge, T., Lapen, D.R.,
Wilkes, G., Robertson, W., Medeiros, D., Sensen, C.W.,
Neumann, N.F., 2007. Tracking host sources of
Cryptosporidium spp. in raw water for improved health risk
assessment. Applied and Environmental Microbiology 73 (12),
Ruecker, N.J., Hoffman, R.M., Chalmers, R.M., Neumann, N.F.,
2011. Detection and resolution of Cryptosporidium species and
species mixtures by genus-specific nested PCR-restriction
fragment length polymorphism analysis, direct sequencing,
and cloning. Applied and Environmental Microbiology 77 (12),
Ryan, U., Read, C., Hawkins, P., Warnecke, M., Swanson, P.,
Griffith, M., Deere, D., Cunningham, M., Cox, P., 2005.
Genotypes of Cryptosporidium from Sydney water catchment
areas. Journal of Applied Microbiology 98 (5), 1221e1229.
Santin, M., Trout, J.M., 2008. Livestock. Cryptosporidium and
Cryptosporidiosis, second ed. CRC Press, Boca Raton, Florida,
Santin, M., Trout, J.M., Fayer, R., 2008. A longitudinal study of
cryptosporidiosis in dairy cattle from birth to 2 years of age.
Veterinary Parasitology 155 (1e2), 15e23.
Stiller, M., Green, R.E., Ronan, M., Simons, J.F., Du, L., He, W.,
Egholm, M., Rothberg, J.M., Keates, S.G., Ovodov, N.D.,
Antipina, E.E., Baryshnikov, G.F., Kuzmin, Y.V.,
Vasilevski, A.A., Wuenschell, G.E., Termini, J., Hofreiter, M.,
Jaenicke-Despres, V., Paabo, S., 2006. Patterns of nucleotide
misincorporations during enzymatic amplification and direct
large-scale sequencing of ancient DNA. Proceedings of the
National Academy of Sciences of the United States of America
103 (37), 13578e13584.
water research 46 (2012) 5135e5150
Sulaiman, I.M., Lal, A.A., Xiao, L., 2002. Molecular phylogeny and Download full-text
evolutionary relationships of Cryptosporidium parasites at the
actin locus. Journal of Parasitology 88 (2), 388e394.
Sulaiman, I.M., Morgan, U.M., Thompson, R.C., Lal, A.A., Xiao, L.,
2000. Phylogenetic relationships of Cryptosporidium parasites
based on the 70-kilodalton heat shock protein (HSP70) gene.
Applied and Environmental Microbiology 66 (6), 2385e2391.
Thompson, J.D., Higgins, D.G., Gibson, T.J., 1994. CLUSTAL W:
improving the sensitivity of progressive multiple sequence
alignment through sequence weighting, position-specific gap
penalties and weight matrix choice. Nucleic Acids Research 22
U.S.Environmental Protection Agency, 2005. Method 1623:
Cryptosporidium and Giardia in Water by Filtration/IMS/FA.
Office of Water. EPA 815-R-05e002.
vonWintzingerode, F., Gobel, U.B., Stackebrandt, E., 1997.
Determination of microbial diversity in environmental
samples: pitfalls of PCR-based rRNA analysis. FEMS
Microbiology Reviews 21 (3), 213e229.
Ward, P.I., Deplazes, P., Regli, W., Rinder, H., Mathis, A., 2002.
Detection of eight Cryptosporidium genotypes in surface and
waste waters in Europe. Parasitology 124 (Pt 4), 359e368.
Xiao, L., 2010. Molecular epidemiology of cryptosporidiosis: an
update. Experimental Parasitology 124 (1), 80e89.
Xiao, L., Bern, C., Sulaiman, I.M., Lal, A.A., 2004a. Molecular
epidemiology of human cryptosporidiosis. In:
Thompson, R.C.A., Armson, A., Ryan, U.M. (Eds.),
Cryptosporidium: from Molecules to Disease. Elsevier,
Amsterdam, pp. 122e146.
Xiao, L., Fayer, R., Ryan, U., Upton, S.J., 2004b. Cryptosporidium
taxonomy: recent advances and implications for public
health. Clinical Microbiology Reviews 17 (1), 72e97.
Xiao, L., Escalante, L., Yang, C., Sulaiman, I., Escalante, A.A.,
Montali, R.J., Fayer, R., Lal, A.A., 1999a. Phylogenetic analysis
of Cryptosporidium parasites based on the small-subunit rRNA
gene locus. Applied and Environmental Microbiology 65 (4),
Xiao, L., Limor, J.R., Li, L., Morgan, U., Thompson, R.C., Lal, A.A.,
1999b. Presence of heterogeneous copies of the small subunit
rRNA gene in Cryptosporidium parvum human and marsupial
genotypes and Cryptosporidium felis. Journal of Eukaryotic
Microbiology 46 (5), 44Se45S.
Xiao, L., Fayer, R., 2008. Molecular characterisation of species and
genotypes of Cryptosporidium and Giardia and assessment of
zoonotic transmission. International Journal for Parasitology.
Xiao, L., Feng, Y., 2008. Zoonotic cryptosporidiosis. FEMS
Immunology & Medical Microbiology 52 (3), 309e323.
Xiao, L., Limor, J., Morgan, U.M., Sulaiman, I.M., Thompson, R.C.,
Lal, A.A., 2000. Sequence differences in the diagnostic target
region of the oocyst wall protein gene of Cryptosporidium
parasites. Applied and Environmental Microbiology 66 (12),
Xiao, L., Morgan, U.M., Limor, J., Escalante, A., Arrowood, M.,
Shulaw, W., Thompson, R.C., Fayer, R., Lal, A.A., 1999c.
Genetic diversity within Cryptosporidium parvum and related
Cryptosporidium species. Applied and Environmental
Microbiology 65 (8), 3386e3391.
Xiao, L., Ryan, U.M., 2008. Molecular epidemiology. In: Fayer, R.,
Xiao, L. (Eds.), Cryptosporidium and Cryptosporidiosis. CRC
Press, Boca Raton, pp. 119e163.
Xiao, L., Singh, A., Limor, J., Graczyk, T.K., Gradus, S., Lal, A., 2001.
Molecular characterization of Cryptosporidium oocysts in
samples of raw surface water and wastewater. Applied and
Environmental Microbiology 67 (3), 1097e1101.
Xiao, L., Sulaiman, I.M., Ryan, U.M., Zhou, L., Atwill, E.R.,
Tischler, M.L., Zhang, X., Fayer, R., Lal, A.A., 2002. Host
adaptation and host-parasite co-evolution in Cryptosporidium:
implications for taxonomy and public health. International
Journal for Parasitology 32 (14), 1773e1785.
Xiao, L., Alderisio, K., Singh, A., AWWA Research Foundation,
American Water Works Association, and Publishing, I.W.A,
2006. Development and Standardization of a Cryptosporidium
Genotyping Tool for Water Samples. AWWA Research
Foundation/American Water Works Association/IWA Pub,
Yang, W., Chen, P., Villegas, E.N., Landy, R.B., Kanetsky, C.,
Cama, V., Dearen, T., Schultz, C.L., Orndorff, K.G.,
Prelewicz, G.J., Brown, M.H., Young, K.R., Xiao, L., 2008.
Cryptosporidium source tracking in the Potomac River
watershed. Applied and Environmental Microbiology 74 (21),
Zhou, L., Yang, C., Xiao, L., 2003. PCR-mediated recombination
between Cryptosporidium spp. of lizards and snakes. Journal of
Eukaryotic Microbiology 50 (Suppl), 563e565.
water research 46 (2012) 5135e5150