//img.uscri.be/pth/ff228c7f120e2930630cc6be5518992e5b7d6354
Cet ouvrage fait partie de la bibliothèque YouScribe
Obtenez un accès à la bibliothèque pour le lire en ligne
En savoir plus

Stirred, not shaken: genetic structure of the intermediate snail host Oncomelania hupensis robertsoniin an historically endemic schistosomiasis area

De
18 pages
Oncomelania hupensis robertsoni is the sole intermediate host for Schistosoma japonicum in western China. Given the close co-evolutionary relationships between snail host and parasite, there is interest in understanding the distribution of distinct snail phylogroups as well as regional population structures. Therefore, this study focuses on these aspects in a re-emergent schistosomiasis area known to harbour representatives of two phylogroups - the Deyang-Mianyang area in Sichuan Province, China. Based on a combination of mitochondrial and nuclear DNA, the following questions were addressed: 1) the phylogeography of the two O. h. robertsoni phylogroups, 2) regional and local population structure in space and time, and 3) patterns of local dispersal under different isolation-by-distance scenarios. Results The phylogenetic analyses confirmed the existence of two distinct phylogroups within O. h. robertsoni . In the study area, phylogroups appear to be separated by a mountain range. Local specimens belonging to the respective phylogroups form monophyletic clades, indicating a high degree of lineage endemicity. Molecular clock estimations reveal that local lineages are at least 0.69-1.58 million years (My) old and phylogeographical analyses demonstrate that local, watershed and regional effects contribute to population structure. For example, Analyses of Molecular Variances (AMOVAs) show that medium-scale watersheds are well reflected in population structures and Mantel tests indicate isolation-by-distance effects along waterways. Conclusions The analyses revealed a deep, complex and hierarchical structure in O. h. robertsoni , likely reflecting a long and diverse evolutionary history. The findings have implications for understanding disease transmission. From a co-evolutionary standpoint, the divergence of the two phylogroups raises species level questions in O. h. robertsoni and also argues for future studies relative to the distinctness of the respective parasites. The endemicity of snail lineages at the regional level supports the concept of endemic schistosomiasis areas and calls for future geospatial analyses for a better understanding of respective boundaries. Finally, local snail dispersal mainly occurs along waterways and can be best described by using cost distance, thus potentially enabling a more precise modelling of snail, and therefore, parasite dispersal.
Voir plus Voir moins

Hauswald et al. Parasites & Vectors 2011, 4:206
http://www.parasitesandvectors.com/content/4/1/206
RESEARCH Open Access
Stirred, not shaken: genetic structure of the
intermediate snail host Oncomelania hupensis
robertsoni in an historically endemic
schistosomiasis area
1 2,3 4 5 4 2Anne-Kathrin Hauswald , Justin V Remais , Ning Xiao , George M Davis , Ding Lu , Margaret J Bale and
1*Thomas Wilke
Abstract
Background: Oncomelania hupensis robertsoni is the sole intermediate host for Schistosoma japonicum in western
China. Given the close co-evolutionary relationships between snail host and parasite, there is interest in
understanding the distribution of distinct snail phylogroups as well as regional population structures. Therefore, this
study focuses on these aspects in a re-emergent schistosomiasis area known to harbour representatives of two
phylogroups - the Deyang-Mianyang area in Sichuan Province, China. Based on a combination of mitochondrial
and nuclear DNA, the following questions were addressed: 1) the phylogeography of the two O. h. robertsoni
phylogroups, 2) regional and local population structure in space and time, and 3) patterns of local dispersal under
different isolation-by-distance scenarios.
Results: The phylogenetic analyses confirmed the existence of two distinct phylogroups within O. h. robertsoni.In
the study area, phylogroups appear to be separated by a mountain range. Local specimens belonging to the
respective phylogroups form monophyletic clades, indicating a high degree of lineage endemicity. Molecular clock
estimations reveal that local lineages are at least 0.69-1.58 million years (My) old and phylogeographical analyses
demonstrate that local, watershed and regional effects contribute to population structure. For example, Analyses of
Molecular Variances (AMOVAs) show that medium-scale watersheds are well reflected in population structures and
Mantel tests indicate isolation-by-distance effects along waterways.
Conclusions: The analyses revealed a deep, complex and hierarchical structure in O. h. robertsoni, likely reflecting a
long and diverse evolutionary history. The findings have implications for understanding disease transmission. From
a co-evolutionary standpoint, the divergence of the two phylogroups raises species level questions in O. h.
robertsoni and also argues for future studies relative to the distinctness of the respective parasites. The endemicity
of snail lineages at the regional level supports the concept of endemic schistosomiasis areas and calls for future
geospatial analyses for a better understanding of respective boundaries. Finally, local snail dispersal mainly occurs
along waterways and can be best described by using cost distance, thus potentially enabling a more precise
modelling of snail, and therefore, parasite dispersal.
Keywords: China, Sichuan Province, Oncomelania hupensis robertsoni, schistosomiasis japonica, coevolution, AFLP,
cytochrome c oxidase subunit I (COI), phylogeny, phylogeography, watersheds
* Correspondence: tom.wilke@allzool.bio.uni-giessen.de
1Department of Animal Ecology & Systematics, Justus Liebig University,
Heinrich-Buff-Ring 26-32 IFZ, D-35392 Giessen, Germany
Full list of author information is available at the end of the article
© 2011 Hauswald et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.Hauswald et al. Parasites & Vectors 2011, 4:206 Page 2 of 18
http://www.parasitesandvectors.com/content/4/1/206
and Yunnan Provinces above the Three Gorges.Background
However, subspecies validityandassignmentremainsThe parasite species of the trematode genus Schistosoma
controversial [24,25], including the western subspeciescause human schistosomiasis, one of the most prevalent
O. h. robertsoni. Genetically, it is highly distinct from allparasitic diseases in the world, infecting more than 200
other known subspecies of O. hupensis [16,22,26,27],million people and leading toasubstantialburdenof
raising questions about whether it deserves full speciesdisease [1]. Schistosoma japonicum, the causal agent of
status. Moreover, based on mitochondrial (mt) DNAthe disease common in East and Southeast Asia, remains
sequence data, the existence of at least two major phy-a public health threat for millions of people living in the
logroups within O. h. robertsoni was demonstrated, withtropical and subtropical zones of China [2]. Though
pairwise K2P-distances for two mtDNA genes of up tooverall prevalence and intensity of infection have
0.12 (= 12%) [26]. Because such a high divergence typi-decreased greatly in recent decades [3], cases of re-
cally reflects genus-level relationships within the snailemergence remain of concern [4,5]. The considerable
superfamily Rissooidea [28], it was initially not clearmedical importance of S. japonicum has spurred numer-
whether the patterns observed are due to the presumedous parasitological, ecological and genetic studies.
long evolutionary history of this subspecies [16] or areGenetic variation between strains of S. japonicum from
simply artefacts. Note that O. hupensis is a dioeciouswidely separated geographical areas, for example, have
species (i.e., an individual specimen is distinctly male orbeen reported based on allozyme, nucleotide sequence,
female). Thus selfing cannot explain the overall highmicrosatellite and single nucleotide polymorphism ana-
diversity within this species either.lyses [6-15]. Indeed, the genetic distance between some
Only recently, an independent study based on DNAS. japonicum populations is so immense as to indicate
sequencing data of a nuclear (nc) gene confirmed thethat they represent distinct taxa [10]. In particular,
existence of two phylogroups [27]. However, the regio-researchers consider that parasites from Sichuan Pro-
nal distribution of these groups is not well understood.vince, which are transmitted by the nominal snail sub-
Moreover, possible co-evolutionary implications of thesespecies Oncomelania hupensis robertsoni, possibly
two genetically and sexually isolated snail host taxa forbelong to a separate strain [10-12].
the genetic structure of S. japonicum remain unknown.Given the demonstrated close co-evolutionary rela-
At the same time, little knowledge exists about smaller-tionships between S. japonicum and its intermediate
scale population structures in O. h. robertsoni,suchassnail host O. hupensis ssp. [16], there is an increased
those within schistosomiasis transmission areas. As ainterest in understanding snail phylogenetics and popu-
lation structure. The importance of the snail host for consequence, local effects of barriers (such as watershed
comprehending disease transmission is further rein- boundaries or mountain ranges) and means of dispersal
(such as along water networks or bird-mediated) onforced by three principal findings:
snail and thus parasite distributions are poorly under-1) Oncomelania hupensis is the sole intermediate host
stood. Given these knowledge gaps, there is a need forfor Schistosoma japonicum:unlike S. mansoni and S.
better understanding of the phylogroup distribution inhaematobium, host switching does not appear to occur.
O. h. robertsoni and regional population structures in2) Given the close genetic interactions between snail
time and space.and parasite in terms of co-evolutionary relationships, a
This study focused on these aspects in a formerly ende-snail population likely reflects population genetic para-
mic schistosomiasis area previously reported to harbourmeters of the parasite and vice versa [17,18].
representatives of two distinct phylogroups [26] - the3) Genetically diverse snail populations appear to be
Deyang-Mianyang area in Sichuan Province (Figure 1).more susceptible to infection with S. japonicum than
Based on a combination of mtDNA and genome-widehomogeneous populations [19].
ncDNA, we studied:Molecular and morphological analyses, together with
1)Thephylogeographyofthetwo O. h. robertsonibreeding experiments and biogeographic studies of O.
phylogroups, with an interest in understanding howhupensis indicate that there are three subspecies on the
these groups are distributed on a micro-scale, e.g.,mainland of China [16,20-22]. Oncomelania h. hupensis
whether they occur in sympatry.primarily occurs in the Yangtze River drainage below
2) Regional and local population structures in spacethe Three Gorges; it has spread to Guangxi Province via
and time, focusing on the degree of population admixingthe Grand Canal from Hunan (note that some authors
as well as the potential correlation of population struc-consider the latter populations to belong to a distinct
ture with physical barriers. Our working hypothesis issubspecies, O. h. guangxiensis [23,24]). Oncomelania h.
tangi is restricted to Fujian Province along the coast, that strong habitat fragmentation in the study area,
and O. h. robertsoni has a patchy distribution in Sichuan together with the long evolutionary history of theHauswald et al. Parasites & Vectors 2011, 4:206 Page 3 of 18
http://www.parasitesandvectors.com/content/4/1/206
structure of the intermediate snail host affects suscept-
ibility to schistosome infections, the means of snail, and
thus parasite, dispersal, and phylogenetic tracking in
Schistosoma japonicum.
Results
Phylogenetic and molecular clock analyses (COI data)
The Bayesian phylogenetic analysis of specimens of
Oncomelania hupensis robertsoni from throughout
its distribution area under the relaxed-clock assump-
tion resulted in the consensus tree shown in Figure 2
(left). The tree topology is very robust with all major
and most medium-level clades being highly sup-
ported (i.e., Bayesian posterior probabilities [BPP] ≥
0.95). In the tree, all O. h. robertsoni specimens form
a monophyletic clade, clustering as sister group to
O. h. hupensis (BPP = 1). Within O. h. robertsoni,
two major phylogroups (BPP = 1) are evident. Phy-
logroup I (numbering according to Wilke et al. [26])
is primarily distributed in central and southern
Sichuan as well as Yunnan. Specimens of phylogroup
II occur mainly in western, but also in southern
Sichuan.
Figure 1 Sampling sites of Oncomelania h. robertsoni. Sampling All specimens from the northern Deyang-Mianyang
sites in the Deyang-Mianyang area, Sichuan Province, China. For site area studied in the present paper cluster within phy-
codes see Table 3 (A = Anxian County, J = Jingyang County, Z =
logroup I, whereas all specimens from the southernmost
Zhongjiang County). LandSat base map taken from MrSID (NASA).
population in the study area (Z3, see Figure 1) belong toNote the Longquan Mountain Range (max. altitude > 1000 m)
separating the southernmost site Z3 from the remaining sampling phylogroup II. In our study area, these two phylogroups
sites. The insert shows the location of the area in China. do not spatially overlap. However, overlapping is evident
in southern Sichuan (see specimens GB-A8a, h vs.
GB-A8d, j, k in Figure 1).
All specimens from the northern Deyang-Mianyangsubspecies, causes a clear population structure reflecting
area (except for the extreme southern ones belongingthe spatial structure of watersheds.
to phylogroup II) form a monophyletic group (BPP =3) Patterns of dispersal under different isolation-by-
0.98). Note that specimen GB-M4, previously studieddistance scenarios to test whether isolation-by-distance
by Attwood et al. [29], also clusters within this groupcan explain the phylogeographical patterns observed and
as it comes from the same area. Within this monophy-if so, which dispersal scenario (i.e., straight line vs.
letic Deyang-Mianyang clade, two subclades (BPP = 1waterway distances, etc.) is supported by genetic data.
each) are evident. Subclade Ia is highly diverse, com-The hypothesis to be tested is that active or passive dis-
prising several deep lineages. Subclade Ib, however, ispersal along waterways best explains the phylogeogra-
relatively shallow with fewer distinct groups. The twophical patterns observed.
subclades do not show a clear geographical structure,Our study constitutes the first effort to combine mito-
chondrial and genome-wide nuclear data on the one with specimens from different groups occurring
hand, with phylogeographical, molecular clock, and phy- sympatrically.
logenetic analyses on the other hand, to unravel the The molecular clock analysis of the reduced data set,
microevolution of Oncomelania h. robertsoni.Thefind- comprising phylogroup I specimens from the Deyang-
ings of this study relate to the allopatric distribution of Mianyang area (see strict clock Bayesian tree in Figure
the two distinct snail phylogroups, the existence of pre- 2, right), indicates that the MRCA of the two subclades
sumably endemic lineages within the study area, the is approximately 1.09 My (95% CI: 0.69-1.58 My) old.
high degree of population admixing, the roles of large Note that CI estimation includes both the error of
and small scale effects on population structure, and pat- branch length variation in the Bayesian analysis (i.e.,
terns of snail dispersal along watersheds, all of which 0.73-1.54 My) and the error of the clock rate for the
-1HKY model (i.e., ± 0.22% Myhave important parasitological implications. The results ; see Methods section for
contribute to a better understanding of how population details).Hauswald et al. Parasites & Vectors 2011, 4:206 Page 4 of 18
http://www.parasitesandvectors.com/content/4/1/206
GB-A8kAnxian County 1 GB-Y1a
GB-Y1c
1 1Jingyang County GB-A8j
GB-A8d
J8-09Zhongjiang County 11 J2-10
A3-01
A4-03
Z1-03
J1-021
A1-01
A5-05
A4-06
A2-02
A5-01 Ia
A3-02.99
A1-04
J8-01
A6-01
A4-13 1.09 My
.98 A5-030.02
J5-01 (0.691.58)
A1-12
J4-01
GB-M4 IJ1-15 1
J2-12
J5-03
Oncomelania h. robertsoni 1 1 J3-09
J2-09
J7-01
J3-01
Z2-01
Z4-01
A4-02
Z1-01 Ib
J2-01
J1-01
J1-03
A5-04
J2-02
J6-01
J3-15
A1-10
A4-01
A2-03
1 Z3-02
Z3-04
Z3-01 0 1 2 My1 GB-M2e
1 GB-M2b1 1 GB-A8a
GB-A8h
GB-A2d1
GB-A2a1
GB-A3a
GB-A2b
GB-A5a.99
GB-A3c
GB-A6b II
GB-A5b
GB-A5c
GB-A6a
.99 GB-A3b
1 GB-A3d1
GB-A4a
GB-A1a
GB-A7a
1 GB-M2a
GB-M3c1
GB-M2j1 GB-M3a
GB-M1a
1 Oncomelania h. hupensis
Oncomelania h. tangi
Figure 2 COI phylogeny of Oncomelania h. robertsoni. Bayesian phylogenetic tree for Oncomelania h. robertsoni under the relaxed clock
model inferred from the mitochondrial COI gene (left). The outgroup taxon, O. minima, was removed a posteriori. The two major phylogroups
within O. h. robertsoni are labelled I and II. For specimen codes see Table 3 (ingroup sequences taken from GenBank are indicated by the prefix
“GB”). A strict clock Bayesian tree of selected specimens of O. h. robertsoni from the Deyang-Mianyang area is shown on the right. Haplotypes
from this area are color-coded according to counties (note that specimen GB-M4, previously studied by Attwood et al. [29], also originates from
this area). For reasons of clarity, outgroups were removed posteriori and the overall topologies of the two distinct subclades Ia and Ib are
indicated by triangles. The age of the MRCM of the two subclades and the respective 95% confidence interval are illustrated by a black bar. For
both trees, all Bayesian posterior probabilities ≥ 0.95 are given next to the nodes.A1-01,
A1-02,
A1-03
J1-03
Hauswald et al. Parasites & Vectors 2011, 4:206 Page 5 of 18
http://www.parasitesandvectors.com/content/4/1/206
Ia Ib
A6-02, A6-03, A6-04,
A6-05, A6-07, A5-06, A5-07,
A5-08, A5-09, A5-10, A5-12,
A4-14, A1-13, A1-12, A5-03,
A4-13, A1-05, A6-06, A6-01
A1-10,
A1-11, A2-03,
A2-05, A2-08,J8-02, J8-03, J8-04,
A4-01, A4-05,A1-09, J8-05, J8-06, J8-07, J8-08,
A4-08, A4-11,A2-02, A2-04, A2-06, J8-11, J5-02, J4-02, J4-03,
A4-12, A4-15A2-07, A3-02, A3-03, J8-01, J5-01, J4-01
A3-05, A3-06, A3-08,
A3-10, A3-11, A3-13,
A3-14, A3-16, A4-06, J2-12,
A4-07, A4-10, A5-01, J1-15J2-13,
A5-02 A5-05 J5-03
A4-02,
A4-04,
A5-04 J1-01, J1-05,
J1-08, J1-09,
Z1-01, Z1-02, J1-10, J1-11,
Z1-04, Z1-05, J1-12, J1-13, J2-02, J2-03,Z2-01, Z2-03, J1-14, J1-16, J2-04, J2-06,Z2-04, Z4-01, J2-01, J2-05, J2-08, J3-15,Z4-02, Z4-03, J2-09, J3-01, J6-01, J6-02,Z4-04, Z4-05, J3-02, J3-03, J6-03, J6-04Z4-06 J3-04, J3-05,
J3-06, J3-08, J3-11,
J3-12, J3-13, J3-14,
J3-16, J7-01, J7-03
J3-09
II
Anxian County A3-01, A3-04,
Z3-01,A3-07, A3-09,
Z3-03, Z3-05,A3-12,Jingyang County Z3-04J2-10 Z3-06, Z3-08,A4-03
Z3-09, Z3-10
Z1-03Zhongjiang County
Figure 3 COI network of Oncomelania h. robertsoni. Statistical parsimony haplotype networks for specimens from the Deyang-Mianyang area
based on the COI gene (connection limit 95%). The three separate networks correspond to phylogroup II and the two subclades (Ia and Ib) of
phylogroup I (see Figure 2). Haplotypes are color-coded according to counties. Areas of circles representing the haplotypes found are
proportional to the number of specimens sharing the respective haplotype. Missing haplotypes are indicated by black dots. Haplotypes with the
highest probability of being the ancestral haplotype in the individual networks are indicated by bold circles.
Network analyses (COI and AFLP data) structuring is weak, though specimens belonging to the
The COI-based TCS network (Figure 3) largely reflects same population tend to cluster together. This is parti-
the relationships seen in the phylogenetic trees pre- cularly evident for the representatives of the southern-
sented above. The analysis resulted in three individual most population Z3 (see Figure 1; also see phylogroup II
networks, which could not be connected by the statisti- in Figure 2), which form the most distinct cluster in the
cal parsimony network algorithm in a parsimonious network.
fashion (connection limit 95%).
The first network (Ia) consists of specimens from AMOVA and SAMOVA analyses (COI and AFLP data)
throughout the Deyang-Mianyang area (except for speci- A global AMOVA of populations by scaling order (SO)
mens from Z3) and corresponds to subclade Ia in the revealed for the COI data that about 73% of total mole-
phylogenetic analysis (see Figure 2). The network con- cular variation is distributed among, and only 27%
sists of several distinct haplotypes with the putative within, populations (Table 1). For the AFLP data, the
ancestral haplotype (bold circle) being shared by only highest portion of genetic variation (87%) was found
four specimens from three populations. In contrast, net- within populations and only 13% among populations.
work two (Ib) shows a star-like structure with the puta- When using watersheds as scaling order (see Method
tive ancestral haplotype being shared by 43 specimens section), the highest proportion of variance in the
from eight populations. This network corresponds to COI gene can be found at small to medium scale levels
subclade Ib in Figure 2. The third network (labeled II) (70-75%).
consists of nine specimens from the southernmost For the AFLP data, the variation among watersheds is
population Z3 (see Figure 1). Specimens belonging to much smaller: the highest value of 6% was found at the
this population are the only ones from the Deyang-Mia- medium scale level WS 3 (Table 1). However, the high-
nyang area studied that belong to phylogroup II (see est among-group variation in both the COI gene and
Figure 2). AFLP data was detected when using the two distinct
The AFLP-based network generated in SplitsTree 4.10 phylogroups (see Figure 2) as SO (86% and 11% in COI
(Figure 4) shows a star-like topology. Geographical and AFLP data, respectively; see Table 1).
A1-04
J1-02,
J1-04,
J1-06,
J1-07
Z3-02
A3-15
J4-04, J5-04,
J5-05
J8-09J1-15
A6-07
J8-11
J8-09
J1-11
J8-03
J7-02
A5-01
Z1-02
Z4-02
A4-07
A6-03
A6-02
A2-06
Z1-01
Z1-05
A2-05
Z2-03
A1-12
A4-03
A4-14
J2-05
Z2-01
J2-10
A2-02
Z2-04
J2-12
Z2-02
J1-07
J8-01
J2-03
J1-02
J8-02
J1-01
J8-06
J2-09
J1-12
Z1-04 J7-01
A1-13
J1-08
J1-05
A1-10
A1-11
A1-07
J1-06
J3-10
A4-09
A6-06J1-14
J3-03 A6-05
J3-08 J1-10
J2-01
J3-06
J3-13 J8-05
J8-10
J1-03
J3-01
J3-11
A1-05
J3-15
J3-16
Hauswald et al. Parasites & Vectors 2011, 4:206 Page 6 of 18
http://www.parasitesandvectors.com/content/4/1/206
J3-12
0.02
Anxian County
Jingyang County
Zhongjiang County
Figure 4 AFLP network of Oncomelania h. robertsoni. NeighborNet network for specimens from the Deyang-Mianyang area based on AFLP
data. The scale bar indicates ML distances. Specimen labels are color-coded according to counties. The distinct specimens from population Z3,
which correspond to phylogroup II, are encircled.
In contrast to the AMOVA, the spatial groupings sug- Exact test and mismatch distribution (COI data)
gested by the SAMOVA are less distinct. For the AFLP Exact test and mismatch distribution analyses were car-
data, none of the F values (i.e., the variance among ried out for specimens from the Deyang-Mianyang areaCT
geographically adjacent groups relative to the total var- belonging to the two distinct COI networks within phy-
iance) is supported with p ≤ 0.1. For the COI gene, the logroup I (i.e., Ia and Ib in Figure 3). The exact test of
highest F value (0.86) was obtained for a spatial clus- sample differentiation based on haplotype frequenciesCT
tering corresponding to the distribution of the two phy- rejected the global null hypothesis of panmixia for both
logroups. However, the respective p-value was only subclades (p < 0.001). Pairwise analyses of individual
0.058. populations indicated significant differences (p ≤ 0.05)
Z3-10
J3-09
J3-04
Z3-08
Z3-09
J3-14
Z3-07
J3-05
Z3-06
Z3-01
J3-07
J1-04
J1-13
A6-01A2-03
J1-16
A6-04
A5-12
A1-08
A1-09
A5-11
A1-06
J2-13
J7-04
J1-19
J8-07
J1-17
J1-18
J2-04
J8-04
J2-06
J2-07
J8-08
J2-11
J2-08
J3-02
A2-07
A4-15
A2-08
A4-12
A4-13
A2-01
A2-04
A3-04
A4-11
J7-03
A3-06
A4-08
A3-05
A3-03
A4-10
A3-01
A4-04
A3-13
A3-02
A3-16
A3-10
A3-08
A3-07
A3-14
Z4-04
A3-12
Z4-05
Z4-03
Z4-01
A4-02
A3-09
Z1-03
Z4-06Hauswald et al. Parasites & Vectors 2011, 4:206 Page 7 of 18
http://www.parasitesandvectors.com/content/4/1/206
Table 1 Results of AMOVAs in Oncomelania h. robertsoni
Source of variation PP WS 1 WS 2 WS 3 WS 4 WS 5 PG
COI
V - 75.16*** 71.72*** 70.21*** 64.65*** 10.79*** 86.31***a
(among groups of populations)
V 72.94*** -1.99 2.33 5.39*** 15.46*** 63.54*** 7.22***b
(within groups of populations)
V 27.06*** 26.83%** 25.94*** 24.40*** 19.88** 25.67 6.47c
(within populations)
AFLP
V - -27.48*** 3.77*** 5.66*** 3.89*** 1.57*** 10.69***a
(among groups of populations)
V 13.03*** 40.45*** 9.39*** 8.18*** 10.57*** 12.11*** 10.77***b
(within groups of populations)
V 86.97*** 87.03 86.84 86.16** 85.54* 86.32 78.54c
(within populations)
Partitioning of variation (V) for populations from the Deyang-Mianyang area derived from COI and AFLP data. Scaling orders are populations (PP), watersheds
(WS), and phylogroups (PG); all values in %. Statistically significant results are marked with asterisks:* p ≤ 0.05, ** p ≤ 0.01, ***p ≤ 0.001.
in 90 out of 132 comparisons for subclade Ia and in 80 spatial expansion model for this subclade is not rejected
out of 144 for subclade Ib (raw data not in favour of a demographic equilibrium. In contrast, the
shown here). Due to the rejection of panmixia, we could mismatch distribution for subclade Ib shows a relatively
use mismatch distribution analyses only for testing the uniform distribution with the maximum number of pair-
spatial extension model and not the demographic one wise differences being low (Figure 5, right). With a SSD
(see Method section). value of 0.017 (p ≤ 0.001), the spatial expansion model
The individual mismatch distribution for subclade Ia for this subclade is rejected (note, however, that the var-
shows a bimodal pattern with a maximal number of iation in Ib is very low, therefore, this result
pairwise differences of 12 (Figure 5, left). Particular has to be treated with caution).
comparisons involving pairs with high nucleotide differ-
ences cluster outside the 95% confidence intervals of the Mantel tests (COI and AFLP data)
coalescent simulations. Nonetheless, with a sum of Mantel tests based on the COI gene showed a signifi-
squared deviation (SSD) value of 0.057 (p = 0.262), the cant correlation between genetic and geographic
2000 Subclade Ia 2000 Subclade Ib
1800 1800
1600 1600
1400 1400
1200 1200
1000 1000
800 800
600 600
400 400
200 200
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12
No. of pairwise nucleotide differences No. of pairwise nucleotide differences
Figure 5 COI mismatch distributions of Oncomelania h. robertsoni. COI-based mismatch distributions of specimens from the Deyang-
Mianyang area belonging to subclades Ia and Ib (see Figure 2). The distributions of pairwise nucleotide differences (circles) were tested against
the spatial expansion model (black line). The 95% confidence intervals of the coalescent simulations are indicated by dashed lines.
FrequencyHauswald et al. Parasites & Vectors 2011, 4:206 Page 8 of 18
http://www.parasitesandvectors.com/content/4/1/206
Table 2 Results of Mantel tests in Oncomelania h. other limnological parameters. Instead, the two groups
robertsoni remain separated by the largest mountain range in the
Geographic distance variable r (COI) r (AFLP) area, the Longquan Mountains, reaching altitudes of >M M
1000 m (Figure 1). This geographical barrier might beEuclidean distance 0.454*** 0.217
the reason why phylogroup II specimens have not (yet)Isotropic least-cost distance
reached the northern Deyang-Mianyang area, while weCost distance 0.588*** 0.249*
didfindasympatricpopulationinsouthernSichuanProportions 0.103 0.126**
(see GB-A8a, h vs. GB-A8d, j, k in Figure 1).Onstream segments 0.480*** 0.240*
The observed geographical patterns could be theTotal path 0.495*** 0.249*
result of increasing secondary contact of phylogroups IAnisotropic least-cost distance
and II. However, whether time of isolation has beenLinear 1:10 0.490*** 0.297*
long enough to allow for the development of pre- orLinear 1:10 proportion 0.115 0.054
postzygotic barriers to speciation [30] remains unclearBinary 1:10 0.490*** 0.246*
as the age of the MRCA of phylogroups I and II couldBinary 1:10 proportion 0.129 0.166**
not be estimated due to the rejection of the strict mole-Linear 1:100 0.350*** 0.131
cular clock model in the full data set.Linear 1:100 proportion 0.141** 0.228***
Nevertheless, given the topology of the phylogeneticBinary 1:100 0.396*** 0.133
tree (Figure 2), this MRCA must be much older thanBinary 1:100 proportion 0.132** 0.264***
0.69-1.58 My, possibly dating back to the early PlioceneIndices for correlation of genetic (COI and AFLP data) distances and
geographic distances in Oncomelania h. robertsoni from the Deyang-Mianyang or even late Miocene.
area (r ). Statistically significant results are marked with asterisks: * p ≤ 0.05,M The generally patchy distribution of the phylogeneti-
** p ≤ 0.01, *** p ≤ 0.001
cally old phylogroups I and II throughout Yunnan and
Sichuan might be largely due to the highly fragmented
distances for 10 out of 13 different geographic distance and often isolated habitats of O. h. robertsoni in the hilly
variables (Table 2). The best correlations were found for and mountainous regions of western China [31,32]. In
the following variables: Cost distance (r = 0.59), total this regard, O. h. robertsoni differs considerably from itsM
path (r = 0.50), linear 1:10 and binary 1:10 (both r = eastern Chinese sister taxon, O. h. hupensis.Beingdis-M M
0.49), as well as isotropic least-cost on-stream segments tributed mainly within and along the floodplains of the
(r = 0.48). Yangtze River, the latter taxon is strongly affected byM
The Mantel tests for the AFLP data showed very simi- annual flooding [16,22], leading to extensive gene flow
and population admixing [19,33].lar results. Here, 9 out of 13 different geographic dis-
Unfortunately, the two phylogroups are not as clearlytance variables showed significant correlations with
separated in the AFLP data set compared to thegenetic distance (Table 2). Moreover, the five best values
mtDNA data, further complicating decisions as to thewere obtained for the very same variables as in the COI
taxonomic status of these two groups. While phylogroupgene with values of r = 0.25, 0.25, 0.30, 0.25 and 0.24,M
II specimens form a distinct cluster in the AFLP-basedrespectively. The distance variable with the overall best
network (Figure 4) and show a significant grouping incorrelation in both data sets was cost distance.
the AMOVA analyses (Table 1), overall divergenceThe Mantel test performed on genetic distances calcu-
remains low. Possible explanations include the exchangelated for the COI and AFLP data sets also showed a sig-
of genes and the differential performance of AFLP.nificant relationship between these two different sets of
While this method works well for closely related geno-genetic data (r = 0.486, p = 0.048).M
types, unrelated genotypes may contribute considerable
Discussion noise to the data set [34]. This is due to the fact that
Phylogeny of Oncomelania h. robertsoni AFLP only uses fragment length as a criterion and not
From the present study it becomes evident that the two the actual DNA sequences. Accordingly, co-migrating
deviant phylogroups in O. h. robertsoni have no clear fragments (i.e., those of the same length) are considered
spatial structure. Though phylogroup II appears to have to be homologous. This assumption, however, is an
amoresouthernandeasterndistribution[26],ranges oversimplification [35] and the fraction of non-identical
are adjacent (e.g. in the southern part of our study area) co-migrating (i.e., homoplasious) fragments largely
or even overlap as previously shown for an area in depends on phylogenetic distance. Values can range
southern Sichuan [26]. from approximately 10% in closely related genotypes to
as much as 100% in distantly related taxa [34]. GivenInterestingly, the disjunctive distribution of the two
these findings, it appears possible that the lack of a clearphylogroups in the Deyang-Mianyang area cannot be
differentiation of the two ancient phylogroups based onexplained with watershed classifications (Figure 6) orHauswald et al. Parasites & Vectors 2011, 4:206 Page 9 of 18
http://www.parasitesandvectors.com/content/4/1/206
Figure 6 Classification of watersheds in the Deyang-Mianyang area. Watershed (WS1-5) classification of 18 study sites of Oncomelania h.
robertsoni. For county and site codes see Table 3.
AFLP data is due to noise problems (also see Caballero together in the network analysis (Figure 4) and the
& Quesada [36]). However, at a lower taxonomic level results of the Mantel test showing concordance of COI
(i.e., within phylogroups), this problem seems to be less and AFLP data.
severe as indicated by, for example, the tendency of spe- No matter whether the two O. h. robertsoni phy-
cimens originating from the same population to cluster logroups represent distinct species or not, implicationsHauswald et al. Parasites & Vectors 2011, 4:206 Page 10 of 18
http://www.parasitesandvectors.com/content/4/1/206
for understanding disease transmission might be sub- Figure 3). In contrast, the pattern seen in Ib points to
stantial due to the close co-evolutionary relationships young lineages with no sudden spatial expansions being
between S. japonicum and its hosts, both intermediate detectable. The most parsimonious explanation for these
and definitive. Not only are there distinct parasite striking differences in age, structure and spatial history
strains that correlate to snail subspecies [10-12], but between the two subclades is that both groups were pre-
also to definitive host species (i.e., buffalo, cattle and viously isolated and only recently came in contact. In
humans vs. goats, pigs, dogs and cats [13,37]; for a fact,ourgeneticdataindicatealonganddiverseevolu-
tionary history of O. h. robertsoni populations in thereview of host species see McManus et al. [38]). These
Deyang-Mianyang area starting some 0.69-1.58 My agodata indicate a distinct hierarchical genetic structure in
S. japonicum: 1) genotype groups according to inter- (see Figure 2).
mediate snail host taxa, 2) subgroups corresponding to Overall, the combined results of our phylogenetic
definitive hosts, and 3) possibly MLGs within strains. and phylogeographical analyses indicate, for the first
This presumably high degree of host specificity calls time, a hierarchical population structure in O. h.
for a closer investigation of the parasites within the robertsoni, caused by small to medium scale factors.
respective O. h. robertsoni phylogroups. Of interest is Scales of concern are: 1) the population level charac-
also whether snails belonging to different phylogroups terized by local demographic processes, 2) the
show different rates of susceptibility to infection or even watershed level with watershed boundaries acting
different scales of resistance ("Red Queen” [39]). Finally, against gene flow to different degrees, and 3) the regio-
cases of schistosomiasis re-emergence in Sichuan and nal level of a formerly endemic schistosomiasis area,
Yunnan provinces should be studied in respect to snail supporting a high degree of endemic snail lineages. In
phylogroups (i.e., phylogroup admixing and/or addition, we see large scale effects such as increasing
replacement). secondary phylogroup contact, possibly caused by
long-range dispersal of O. h. robertsoni throughout
Regional and local population structures in space and Sichuan and Yunnan provinces.
time Of these scale effects, watershed effects are important
Our phylogenetic analyses show that specimens of all but do not appear to be the dominant factor. Therefore,
but the southernmost population from the Deyang-Mia- our null hypothesis that the strong habitat fragmenta-
nyangarea(Z3)belongtophylogroupI.Moreover, tion in the study area and the long evolutionary history
these specimens form a relatively old and well supported of the subspecies are best reflected in a population
monophyletic group in our analyses (Figure 2), indicat- structure corresponding to watershed distribution, has
to be rejected.ing distinct geographical pattern at a regional scale and
corroborating the status of the area as being historically Instead, larger scale effects together with considerable
endemic for schistosomiasis. Within this clade (i.e., population admixing might also contribute to the
within the Deyang-Mianyang area), the geographical genetic patterns observed. Using the words of Ian Fle-
structure appears to be weaker. Nonetheless, the results mings’s fictional British Secret Service agent James Bond
of the AMOVAs show significant partitioning of genetic as a metaphor, which were first introduced in the 1956
variance according to watersheds (Table 1), both for the novel “Diamonds are Forever“, the population structure
COI and, to a lesser, extent, the AFLP data sets. In addi- of O. h. robertsoni within this formerly endemic schisto-
tion, local demographic processes appear to act heavily somiasis area is stirred, but not shaken.
on individual populations, causing medium (COI gene)
to high (AFLP data) partitioning of variation within Patterns of dispersal under different isolation-by-distance
populations (Table 1). scenarios
Interestingly, the Deyang-Mianyang clade consists of For both data sets AFLP and COI, we see clear isola-
two well supported subclades (see subclades Ia and Ib in tion-by-distance effects, that is, a significant correlation
Figure 2). These two groups do not appear to be spa- between geographic and genetic distance. This is a
tially structured as specimens belonging to both groups rather unexpected finding as, to our knowledge, such an
occur in sympatry. However, individual mismatch ana- effect has not been shown before for a currently or his-
lyses of subclade Ia and Ib specimens reveal contrasting torically endemic schistosomiasis area in China. Though
patterns in terms of ruggedness and number of maxi- this might be partly due to the fact that only few
mum pairwise nucleotide differences (Figure 5). detailed regional studies exist, previous studies mainly
Acknowledging that mismatch patterns are difficult to suggested either large scale admixing effects caused by
interpret [40], the pattern seen in Ia might suggests the long-range dispersal [26] or strong fragmentation effects
existence of old and diverse lineages and possibly indi- at the landscape level [25,32] as the prevailing biogeo-
cates a relatively stable population structure (also see graphical processes in O. h. robertsoni.