Content uploaded by Konstans Wells
Author content
All content in this area was uploaded by Konstans Wells on Mar 19, 2018
Content may be subject to copyright.
BIODIVERSITY
RESEARCH
The importance of parasite geography
and spillover effects for global patterns
of host–parasite associations in two
invasive species
Konstans Wells
1
*, Robert B. O’Hara
2
, Serge Morand
3
,
Jean-Philippe Lessard
4,5
and Alexis Ribas
6
1
The Environment Institute, School of Earth
and Environmental Sciences, The University
of Adelaide, Adelaide, SA, Australia,
2
Biodiversity and Climate Research Centre
(BiK-F), Frankfurt, Germany,
3
Centre
d’Infectiologie Christophe M
erieux du Laos,
CIRAD AGIRs, CNRS ISEM, Vientiane, Lao
People’s Democratic Republic,
4
Qu
ebec
Centre for Biodiversity Science, McGill
University, Montr
eal, QC, Canada,
5
Department of Biology, Concordia
University, Montr
eal, QC, Canada,
6
Biodiversity Research Group, Faculty of
Science, Udon Thani Rajabhat University,
Udon Thani, Thailand
*Correspondence: Konstans Wells, School of
Earth & Environmental Sciences and the
Environment Institute, The University of
Adelaide, North Terrace, Adelaide SA 5005,
Australia.
E-mail: konstans.wells@adelaide.edu.au
ABSTRACT
Aim Geographic spread and range expansion of species into novel environ-
ments may merge originally separated species assemblages, yet the possible
drivers of geographic heterogeneity in host–parasite associations remain poorly
understood. Here, we examine global patterns in the parasite assemblages of
two rat species and explore the role of parasite acquisition from local pools of
host species.
Location Global.
Methods We compiled a global data set of helminth parasites (n=241 spe-
cies) from two rat species (Rattus rattus species complex, R. norvegicus) and,
concomitantly, from all other mammal species known to be infected by the
same parasites. We used an inverse Bayesian modelling approach to explicitly
link species-level to community-level infestation probabilities at different geo-
graphic scales and alleviate the shortcoming of sampling bias.
Results Patterns of species richness and turnover of parasites in the two focal
rat species revealed clear biogeographic structure with lowest species richness
and most distinct assemblages in Madagascar and highest species richness and
least distinct assemblages in the Palaearctic region. Parasite species richness and
turnover across regions were correlated for the two focal hosts, although they
were associated with distinct assemblages within regions. Infection probability
of a focal host with any given parasite was clearly related to infection probabil-
ity of the local species pool of wildlife hosts with that same parasite. Infection
probability of other mammal species infected with these parasite species, in
turn, decreased with their taxonomic distance to the genus Rattus.
Main conclusions Our study demonstrates the importance of spillover of par-
asites from local wildlife hosts to invasive rats on global patterns of host–para-
site associations. Considering both changes in local pools of host species and
the global distributions of parasite and pathogen diversity in consistent model
frameworks may therefore advance the forecasting of species-level infestation
patterns and the possible risk of disease emergence from local to global scale.
Keywords
Biogeographic regions, biological invasions, geographic mosaics, global diver-
sity, helminths, host–parasite associations, inverse modelling, parasite spread,
species distribution, zoonoses.
DOI: 10.1111/ddi.12297
ª2014 John Wiley & Sons Ltd http://wileyonlinelibrary.com/journal/ddi 477
Diversity and Distributions, (Diversity Distrib.) (2015) 21, 477–486
A Journal of Conservation Biogeography
Diversity and Distributions
INTRODUCTION
As much as 60% of human diseases are of zoonotic origin
(Taylor et al., 2001), but our knowledge of how parasites
are distributed and shared among wildlife, commensal and
domestic animal species is inevitably incomplete given the
challenge to exhaustively document possible host–parasite
combinations for thousands of species. Moreover, while it is
evident that environmental change alters conditions for para-
site persistence and transmission (Patz et al., 2000), we lack
a solid understanding of how global patterns in host–parasite
associations are shaped by geographic range limits of para-
sites and interactions between invasive hosts and native
assemblages of wildlife hosts (Morand & Krasnov, 2010;
Estrada-Pe~
na et al., 2014).
During historical dispersal and invasions of new environ-
ments, host species are likely to escape from some associated
parasite species and thus harbour fewer parasites in newly
colonized regions compared to the associated parasite assem-
blages in their native range (Poulin & Mouillot, 2003;
Torchin et al., 2003). Moreover, a local assemblage of para-
sites (i.e. all parasites found in a host species in a region)
infecting a widely distributed host species (e.g. commensal
rat) may be strongly influenced by acquisition from the local
pool of wildlife hosts, that is a gain of parasites that origi-
nated in local wildlife species (Daszak et al., 2000). Geograph-
ical structure in host–parasite associations is thus likely to
track patterns of wildlife diversity such as those observed
along broad-scale environmental gradients (Jenkins et al.,
2013) and on global maps of zoogeographic regions (Holt
et al., 2013). The total species richness of parasites in local
host communities often correlates positively with the species
richness of hosts (Krasnov et al., 2004; Thieltges et al., 2011).
As such, an invasive host species colonizing an area with a
high diversity of wildlife species is likely to be exposed to a
high diversity of potentially suitable parasite species. How-
ever, increasing the diversity of host species may also cause
unfavourable conditions for parasites if host species differ in
quality. In such cases, increasing host species richness can
reduce parasite transmissibility due to more encounters with
unfavourable hosts (Ostfeld & Keesing, 2012). The strength
and generality of the relationship between the number of par-
asites in an invasive host species and the diversity of local
wildlife assemblages as potential reservoirs over large geo-
graphic scales remain therefore elusive (Morand, 2012).
Uncertainty persists as to whether parasite diversity on
any given species of host in a local community is positively
related to local host diversity. Presumably, the parasite spe-
cies richness of any given host species should be highest in
its ancestral centre of origin (i.e. South and Southeast Asia
for commensal rats of the genus Rattus; Robins et al., 2008;
Aplin et al., 2011). The sharing of parasite species with other
species from local host species pools can be expected to be
highest if species have a long history of sharing the same
biogeographical space: the longer domestic and commensal
animals are associated with humans, for example the more
parasites they share with them (Morand et al., 2014).
In this study, we explored changes in parasite species rich-
ness and turnover at global scale and the role of parasite
acquisition from local pools of wildlife hosts of two of the
most cosmopolitan invaders and important commensal rat
species. The black rat Rattus rattus (species complex) and the
Norway rat Rattus norvegicus have been introduced in most
regions of the world as a result of human activities (Aplin
et al., 2011), have a long history of disease transmission to
humans (Meerburg et al., 2009) and cause considerable eco-
nomic loss (Singleton et al., 2003; Stenseth et al., 2003).
R. rattus invades a large range of semi-natural and natural
environments, where it is likely to interact with various wild-
life species (Goodman, 1995; Harris et al., 2006; Wells et al.,
2014). Such human-induced mixture of anthropogenic and
natural habitats and animal species are likely to enhance the
exchange of parasite species across environments (Hoberg,
2010). R. norvegicus is more strongly associated with urban
environments that generally harbour fewer wildlife species
(Wells et al., 2014). We may therefore expect parasite assem-
blages of R. rattus to reflect the higher richness of reservoir
hosts in their environment relative to that of R. norvegicus.
The two rat species could be expected to share similar parasite
assemblages and exhibit similar patterns of spatial turnover
across zoogeographic regions if we take into account that they
occur in sympatry in urban environments and parasite may
frequently shift between these two closely related species.
Not only do we know very little about global geographic
trends of host–parasite associations; there are important
methodological obstacles that can preclude obtaining a clear
picture. Species distributional data commonly include bias
towards heterogeneous sampling efforts and incomplete sam-
pling (Lomolino, 2004; Hortal et al., 2007; Boakes et al.,
2010). Incomplete inventories introduce ‘false’ zeros into
data (Martin et al., 2005), and there is uncertainty as to
whether host–parasite associations are lacking or have simply
been unobserved (Hopkins & Nunn, 2007). Especially in
comparative studies, sampling bias and incomplete invento-
ries may lead to misleading conclusions about host–parasite
associations if not accurately accounted for in analyses (Wells
et al., 2013). We must therefore develop analytical tools that
will minimize how sampling biases influence our perception
of geographic patterns in host–parasite associations.
Addressing our study question with incomplete observa-
tions inevitably calls for statistical approaches that take
uncertainty and unknown measures into account (Keating &
Cherry, 2004; Reese et al., 2005; Ward et al., 2009). We fitted
an inverse modelling approach in a Bayesian hierarchical
framework to estimate possible host–parasite associations
from a limited set of observations, while also accounting for
the possible links between parasite species and local species
pools of wildlife hosts.
We therefore used the flexibility of a hierarchical Bayesian
approach for estimating parasite occurrence at poorly sampled
478 Diversity and Distributions, 21, 477–486, ª2014 John Wiley & Sons Ltd
K. Wells et al.
locations by ‘borrowing strength’ from more intensively sam-
pled locations, while also acknowledging that locations are not
identical in all aspects. The hierarchical model structure fur-
ther allows to model the variation of parasite occurrence in
wildlife hosts according to species and population attributes
and environmental variables (Fig. 1). For example, we can ask
whether species of conservation concern are particularly sensi-
tive to share parasites with invasive (focal) species, fostering
our understanding for informed wildlife management and pest
control (Daszak et al., 2000). We systematically combined
information at the species level (i.e. parasite associations in
the focal rats species) with those at the community level (i.e.
wildlife hosts linked to rats by sharing the same parasites) into
a hierarchical model that optimizes inference by maximizing
the use of all available information and simultaneously assess-
ing the influence of ecological processes expected to operate
across levels of organization.
METHODS
Database on host–parasite records
We compiled a database of recorded associations between
the focal rat species and their helminth parasites from the
host–parasite database of the Natural History Museum Lon-
don (NHML) (Gibson et al., 2005), which includes host–par-
asite records from more than 28,000 references up to 2003
(accessed in June 2013).
For each field record (excluding experimental and captive
records), we characterized the geographic location based on
current country-level geographic borders. We specified this
characterization in subregions for some locations such as
China (which encompasses multiple zoogeographic regions;
for all records from China which could not be identified to
subregion, we used an extra category that specified zoogeo-
graphic region as missing data). Additionally we separated
records from different islands in Indonesia (e.g. we consid-
ered Borneo as a separate location irrespective of whether
records were made in the Indonesian or Malaysian part of
the island). For countries with few records, we merged
neighbouring countries into larger units such as Scandinavia
(Finland, Norway, Sweden). We are aware that this classifica-
tion is coarse and arbitrary. Nevertheless, we consider this
approach to be acceptable in order to systematically assign
all records to geographical units while accounting for the
global topography and zoogeographic structure of a large set
of records with no detailed geographic positions available.
Our data set for analysis included 144 geographic locations.
Figure 1 Illustration of the inverse Bayesian model for inferences on parasite geography and spillover effects from global species lists.
The illustration represents a focal host species (dark rat) in three different regions (R1–R3; illustrated wildlife species are examples from
the Palaearctic, Afrotropical and Australian zoogeographic regions), which can be divided into any number of different locations (R1: l1
and l2; R2: i1 and i2). Rats and also other mammals species have been sampled for a parasite species, which has been only found in a few
species and localities, with presence recorded as ‘1’ (nematode drawn on top of mammals) and absence as ‘0’. Records are considered
random draws from a Bernoulli distribution (blue arrows) with probabilities wfor the focal host species and probability ϑfor all other
host species. Estimates of ϑfor any local host assemblage are used for the estimation of w, linking infection probability of local wildlife
hosts to the focal host species (green arrows). The parasite has not been sampled from the focal host species in location i2 and R3.
However, given the overall model framework, there is a certain probability that the focal species is also infected by the parasite in these
areas: the intercept l
w
denotes an average global infection risk independent of region and location, while the parameter l
Φ
estimates
regional infection probability independent of location. Thus, l
Φ
R2>0 (parasite is recorded in i1inR2) and l
Φ
R3=0 (no parasite
recorded in R3), and there is a higher probability that the parasite is present in i2 than in R3 given the data and parameter estimates.
Diversity and Distributions, 21, 477–486, ª2014 John Wiley & Sons Ltd 479
Parasite geography and spillover effects
We assigned all locations to one of the 11 zoogeographic
regions recently defined by Holt et al. (2013). We further
assigned locations to the main climate zones (equatorial,
arid, warm-temperate, snow, polar) based on an updated
world map of the K€
oppen–Geiger climate classification (Kot-
tek et al., 2006); if locations were covered by various climate
zones (28 of 144), we assigned the relative proportion of the
area covered by each climate zone and considered the uncer-
tainty in which climate zones parasites were recorded with
multiple data imputation as part of the Bayesian analysis and
sampling procedure.
With the same approach, for each helminth species in our
database we compiled the full range of host species for all
locations from the NHML host–parasite database. For all
mammal species in our database, we calculated the taxonomic
distance to the genus ‘Rattus’ based on the number of nodes
in a taxonomic tree (Wilson & Reeder, 2005) resulting from
the species’ genus, family and order classification, indexed
between 1 and 5. We further classified the IUCN conservation
status of all mammal species (categories: least concern, near
threatened, vulnerable, endangered, critically endangered)
based on the 2001 assessment (version 3.1, http://www.
iucnredlist.org). Note that we termed regional assemblages of
mammals as ‘wildlife hosts’ in this study, but these assem-
blages also included humans and domestic mammals.
For data cleaning, all records not identified to species level
were excluded, except those genera for which only single
unidentified species were recorded. Scientific names were
revised and standardized with the aid of a literature search
in Thompson Reuters Web of Science (http://apps.webof-
knowledge.com/; latest searches performed in September
2013), personal literature collections and the mammal online
database at http://vertebrates.si.edu/msw/mswCFApp/msw/
index.cfm (Wilson & Reeder, 2005).
Our final data set for analysis included a total of 12,405
records of host–parasite association from different locations.
Missing data were handled in our model approach by multi-
ple data imputation. We are aware that our database is
incomplete and lacks recently discovered helminth species.
However, we do not consider this to be a problem, as we
were interested in inference on geographic structure in host–
parasite interactions from a finite data set, rather than
complete lists of records. Species lists and classification of
sampling locations are provided in Appendix S1 in the
Supporting Information.
Inferring host–parasite associations with an inverse
modelling approach
We used an inverse hierarchical modelling approach in a
Bayesian framework to ask how likely it was for any parasite
species to occur in a focal host species (Rattus rattus and
R. norvegicus) in different locations inferred from a finite set
of observations. To make inferential summary statistics on
modelled estimates rather than observations, we estimated
the probability of having a parasite species associated with a
host species in any sampled location.
For all locations l, at which at least one parasite species p
has been recorded in at least one focal host species h,we
assumed that all records y(h, p, l) of host–parasite associa-
tions were random draws based on the true but unknown
distribution of host–parasite associations such that
yðh;p;lÞBernoulliðwðh;p;lÞÞ (1)
The probability of local host–parasite association w(h, p, l)
can be modelled further. In particular, we assumed w(h, p, l)
to be linked to the odds of the average occurrence probabil-
ity of the respective parasite species Φ(p, r) within the zoo-
geographic region rwhere lis located (based on records
from all kind of host species, irrespective of host species
identity), given that locations from the same region are likely
to harbour similar parasite assemblages. Likewise, we
assumed w(h, p, l) to be linked to the odds of the average
occurrence probability of the respective parasite species Ω(p,
c) within the climate zone cwhere lis located. We also
assumed w(h, p, l) to vary with the average infestation prob-
ability of any mammal species from local assemblages with
the same parasite, given as l
ϑ
(p, l) (the odds of the infesta-
tion probability ϑ(p, l)). Using a logit-link function, this
gives:
logitwðh;p;lÞ¼lwðh;pÞþa1ðh;pÞlUðp;r½lÞ
þa2ðh;pÞlXðp;c½lÞ þ a3ðh;pÞl0ðp;lÞ(2)
where l
w
(h, p) is the species-specific intercept and a
1
to a
3
are coefficient estimates.
The covariates l
Φ
(p, r), l
Ω
(p, c) and l
ϑ
(p, l) are them-
selves considered as random variables (i.e. modelled proba-
bilities from finite sets of observations), for which we
assumed all observations, Φobs and ϑobs respectively, as
random draws out of the true but unknown parasite distri-
butions and host associations. We thus assumed
Uobsðp;lÞBernoulliðUðp;lÞÞ and
0obsðp;l;xlÞBernoullið0ðp;lÞÞ (3)
where x
l
indexes all mammal species examined in location l
for parasites.
We assumed again logit-link functions to model Φ(p, l)
and ϑ(p, l) based on random intercepts such as
logitUðp;lÞ¼lUðp;r½lÞ þ lXðp;c½lÞ and
logit0ðp;lÞ¼l0ðp;lÞþc1TðmÞþc2CðmÞ:(4)
Here, we modelled ϑ(p, l) further as a function of species-
specific taxonomic distance Tand their IUCN conservation
status Cof mammal species m;c
1
and c
2
are the respective
coefficient estimates.
Given the estimated probability of local host–parasite asso-
ciation w(h, p, l), we can express our uncertainty in the
derived state variable z(h, p, l) of whether a host species his
480 Diversity and Distributions, 21, 477–486, ª2014 John Wiley & Sons Ltd
K. Wells et al.
We assigned all locations to one of the 11 zoogeographic
regions recently defined by Holt et al. (2013). We further
assigned locations to the main climate zones (equatorial,
arid, warm-temperate, snow, polar) based on an updated
world map of the K€
oppen–Geiger climate classification (Kot-
tek et al., 2006); if locations were covered by various climate
zones (28 of 144), we assigned the relative proportion of the
area covered by each climate zone and considered the uncer-
tainty in which climate zones parasites were recorded with
multiple data imputation as part of the Bayesian analysis and
sampling procedure.
With the same approach, for each helminth species in our
database we compiled the full range of host species for all
locations from the NHML host–parasite database. For all
mammal species in our database, we calculated the taxonomic
distance to the genus ‘Rattus’ based on the number of nodes
in a taxonomic tree (Wilson & Reeder, 2005) resulting from
the species’ genus, family and order classification, indexed
between 1 and 5. We further classified the IUCN conservation
status of all mammal species (categories: least concern, near
threatened, vulnerable, endangered, critically endangered)
based on the 2001 assessment (version 3.1, http://www.
iucnredlist.org). Note that we termed regional assemblages of
mammals as ‘wildlife hosts’ in this study, but these assem-
blages also included humans and domestic mammals.
For data cleaning, all records not identified to species level
were excluded, except those genera for which only single
unidentified species were recorded. Scientific names were
revised and standardized with the aid of a literature search
in Thompson Reuters Web of Science (http://apps.webof-
knowledge.com/; latest searches performed in September
2013), personal literature collections and the mammal online
database at http://vertebrates.si.edu/msw/mswCFApp/msw/
index.cfm (Wilson & Reeder, 2005).
Our final data set for analysis included a total of 12,405
records of host–parasite association from different locations.
Missing data were handled in our model approach by multi-
ple data imputation. We are aware that our database is
incomplete and lacks recently discovered helminth species.
However, we do not consider this to be a problem, as we
were interested in inference on geographic structure in host–
parasite interactions from a finite data set, rather than
complete lists of records. Species lists and classification of
sampling locations are provided in Appendix S1 in the
Supporting Information.
Inferring host–parasite associations with an inverse
modelling approach
We used an inverse hierarchical modelling approach in a
Bayesian framework to ask how likely it was for any parasite
species to occur in a focal host species (Rattus rattus and
R. norvegicus) in different locations inferred from a finite set
of observations. To make inferential summary statistics on
modelled estimates rather than observations, we estimated
the probability of having a parasite species associated with a
host species in any sampled location.
For all locations l, at which at least one parasite species p
has been recorded in at least one focal host species h,we
assumed that all records y(h, p, l) of host–parasite associa-
tions were random draws based on the true but unknown
distribution of host–parasite associations such that
yðh;p;lÞBernoulliðwðh;p;lÞÞ (1)
The probability of local host–parasite association w(h, p, l)
can be modelled further. In particular, we assumed w(h, p, l)
to be linked to the odds of the average occurrence probabil-
ity of the respective parasite species Φ(p, r) within the zoo-
geographic region rwhere lis located (based on records
from all kind of host species, irrespective of host species
identity), given that locations from the same region are likely
to harbour similar parasite assemblages. Likewise, we
assumed w(h, p, l) to be linked to the odds of the average
occurrence probability of the respective parasite species Ω(p,
c) within the climate zone cwhere lis located. We also
assumed w(h, p, l) to vary with the average infestation prob-
ability of any mammal species from local assemblages with
the same parasite, given as l
ϑ
(p, l) (the odds of the infesta-
tion probability ϑ(p, l)). Using a logit-link function, this
gives:
logitwðh;p;lÞ¼lwðh;pÞþa1ðh;pÞlUðp;r½lÞ
þa2ðh;pÞlXðp;c½lÞ þ a3ðh;pÞl0ðp;lÞ(2)
where l
w
(h, p) is the species-specific intercept and a
1
to a
3
are coefficient estimates.
The covariates l
Φ
(p, r), l
Ω
(p, c) and l
ϑ
(p, l) are them-
selves considered as random variables (i.e. modelled proba-
bilities from finite sets of observations), for which we
assumed all observations, Φobs and ϑobs respectively, as
random draws out of the true but unknown parasite distri-
butions and host associations. We thus assumed
Uobsðp;lÞBernoulliðUðp;lÞÞ and
0obsðp;l;xlÞBernoullið0ðp;lÞÞ (3)
where x
l
indexes all mammal species examined in location l
for parasites.
We assumed again logit-link functions to model Φ(p, l)
and ϑ(p, l) based on random intercepts such as
logitUðp;lÞ¼lUðp;r½lÞ þ lXðp;c½lÞ and
logit0ðp;lÞ¼l0ðp;lÞþc1TðmÞþc2CðmÞ:(4)
Here, we modelled ϑ(p, l) further as a function of species-
specific taxonomic distance Tand their IUCN conservation
status Cof mammal species m;c
1
and c
2
are the respective
coefficient estimates.
Given the estimated probability of local host–parasite asso-
ciation w(h, p, l), we can express our uncertainty in the
derived state variable z(h, p, l) of whether a host species his
480 Diversity and Distributions, 21, 477–486, ª2014 John Wiley & Sons Ltd
K. Wells et al.
occurrence probability of parasites across climate zones l
Ω
(p,
c) had a positive impact on the infection probability for only
7 of 241 parasite species in R. rattus and for eight parasite
species in R. norvegicus (lower limits of CI >0 for a
2
).
Average infection probability of other mammal species
with helminths decreased considerably with taxonomic dis-
tance from the genus Rattus (Fig. 4), and it also decreased
with increasingly endangered status (according to their IUCN
status) (Fig. 4). However, the species turnover in overall
mammal assemblages in different zoogeographic regions was
not correlated with the species turnover of parasite assem-
blages in the two rat species (both Mantel tests with Pear-
son’s correlation coefficients r<0.27).
DISCUSSION
Inferring host–parasite associations for two of the most com-
mon and invasive commensal rat species at a global scale
showed that species richness and assemblage composition of
parasitic helminths varied over zoogeographic regions. Geo-
graphic variation in parasite species richness and assemblage
composition was correlated between the two focal host
species (Rattus rattus and R. norvegicus), although locally
they were associated with distinct parasite assemblages. Fur-
ther, our hierarchical model framework showed a clear influ-
ence of local species pools of wildlife hosts on parasite
Table 1 Summary of species richness and spatial turnover
(meanb
sim
) of helminth parasite assemblages in the two host
species Rattus rattus and R. norvegicus in different zoogeographic
regions as defined by (Holt et al., 2013). For species richness,
recorded numbers are given as S
Rec
, while posterior estimates are
given as S
Est
. Spatial turnover estimates of meanb
sim
are
calculated as the mean of all pairwise b
sim
values from different
locations within regions. 95% credible intervals for posterior
estimates are given in parenthesis
Region S
Rec
S
Est
Meanb
sim
R. rattus
Afrotropical 27 40 (35–47) 0.47 (0.34–0.53)
Australian 11 17 (14–21) 0.49 (0.38–0.56)
Madagascan 0 1 (0–3) 0.99 (0.2–1)
Nearctic 3 15 (9–21) 0.55 (0.4–0.67)
Neotropical 16 24 (19–30) 0.51 (0.38–0.58)
Oceanian 9 15 (11–19) 0.58 (0.46–0.68)
Oriental 64 71 (67–76) 0.38 (0.24–0.43)
Palaearctic 48 67 (59–74) 0.39 (0.25–0.45)
Panamanian 6 15 (9–22) 0.61 (0.47–0.69)
Saharo-Arabian 25 30 (27–36) 0.53 (0.4–0.59)
Sino-Japanese 15 28 (22–34) 0.56 (0.42–0.63)
R. norvegicus
Afrotropical 0 24 (17–33) 0.58 (0.45–0.68)
Australian 13 19 (16–23) 0.49 (0.36–0.55)
Madagascan 0 1 (0–3) 0.99 (0.3–1)
Nearctic 27 34 (30–43) 0.54 (0.41–0.59)
Neotropical 19 29 (24–35) 0.59 (0.44–0.65)
Oceanian 0 11 (5–16) 0.54 (0.38–0.69)
Oriental 21 41 (33–48) 0.54 (0.4–0.62)
Palaearctic 97 100 (95–107) 0.26 (0.21–0.4)
Panamanian 6 14 (9–20) 0.58 (0.44–0.67)
Saharo-Arabian 26 33 (28–38) 0.53 (0.4–0.58)
Sino-Japanese 30 43 (36–49) 0.55 (0.41–0.61)
●
●
●
●
●
●
●
●
●
●
●
●
020406080
020406080100
●
●
●
●
●
●
●
●
●
●
●
R. rattus −parasite number
R. norvegicus− parasite number
Helminth species richness
Figure 2 Relationship in the estimated numbers of helminth
species associated with the two host species Rattus rattus and
R. norvegicus in different zoogeographic regions given as
posterior estimates of modes (points) and 95% credible intervals
(bars). The dashed line indicates a 1 : 1 relationship.
●
●
●
●
●
●
●
●
●
●
●●
0.2 0.4 0.6 0.8 1.0
0.2 0.4 0.6 0.8 1.0
●
●
●
●
●
●
●
●
●
●●
R. rattus −β
Sim
R.norvegicus −β
Sim
Uniqueness of helminth assemblages
Figure 3 Distinctness of parasitic helminth assemblages
associated with the two host species Rattus rattus and
R. norvegicus in different zoogeographic regions as calculated
from averaged spatial turnover estimates (modes of posterior
samples are plotted as points and 95% credible intervals as
bars). The dashed line indicates a 1 : 1 relationship.
482 Diversity and Distributions, 21, 477–486, ª2014 John Wiley & Sons Ltd
K. Wells et al.
associations in the two focal host species, which supports the
importance of spillover effects (Daszak et al., 2000). More-
over, in non-focal host species, taxonomic distance to the
genus ‘Rattus’ and conservation status was related to the
probability of being infected with a parasite species that had
also infected one of the focal hosts.
Commensal rats have escaped several helminth parasites in
regions such as Madagascar or Australia, where estimates of
the species richness of parasites associated with the focal
hosts are very small (see also Torchin et al., 2003). Only in
the Palaearctic region were estimates of parasite species rich-
ness higher (R. norvegicus) than in the Oriental region, where
the host genus Rattus originated and diversified (Robins
et al., 2008; Aplin et al., 2011). At a global scale, total num-
bers of recorded parasite species were considerably higher
than those in the Oriental region for both focal host species,
emphasizing that a considerable proportion of parasite spe-
cies are linked to non-focal host species and were likely to
have been acquired by the focal rat species during their inva-
sion and colonization history. However, despite the clear link
between focal and non-focal host–parasite associations, we
do not know specifically which parasite species co-evolved
with the rat species or any other host species. Moreover, with
only general relationships in species richness and turnover
examined, the underlying mechanisms that cause loss and
acquisition of host–parasite association across geographic
gradients remain unexplored.
Besides the likely impact of geographically varying regional
wildlife host assemblages on parasites, there are likely to be
other factors impacting parasite transmission and survival
according to parasites’ life histories. Parasitic helminths with
either free-living stages in their life cycles or indirect trans-
mission (e.g. via vectors) may be particularly sensitive to cli-
mate changes and other ecological perturbations (Brooks &
Hoberg, 2007), and variable conditions may result in geo-
graphic mosaics of species associations in time and space
(Thompson & Cunningham, 2002). Geographic patterns in
host–parasite associations and other species interactions are
most likely structured by multiple drivers of species and
environmental attributes (Sheppard et al., 2010; Guilhaumon
et al., 2012). Correlations in species richness and spatial
turnover of parasites in the two focal host species, despite
different associated assemblages, is an important result.
However, additional studies are required to explore possible
drivers of such relationships.
Contrary to our expectations, R. rattus was not associated
with more parasite species than R. norvegicus nor did its
associated parasite assemblages show more zoogeographic
variation. Along with the findings that more closely related
mammalian host species were more likely to be associated
with the same parasite species, we conclude that parasite
assemblages do evidently change with different conditions in
zoogeographic regions but not necessarily with different hab-
itat use of the focal host species, nor with their affinity for
near-natural habitats shared with local wildlife host species.
We found mammal species of least conservation concern
were more likely to be infected with the parasites of the two
rat species than endangered species. Most endangered species
can be found in natural habitats that are at continuous
decline due to human impact (Rondinini et al., 2011),
whereas a large proportion of mammal species of least con-
cern, including domestic species, are well able to persist in
anthropogenic landscapes, where the focal hosts also occur.
The stronger links between wildlife species of least conserva-
tion concern and the parasites recorded from the two com-
mensal rats provide a first indication that habitat overlap
and species ecological traits may impact the sharing of para-
sites between invasive species and local wildlife. However, we
currently lack further detailed information to incorporate
them into our analysis. Likewise, it is desirable to incorpo-
rate more geographic attributes of sample locations in future
analysis to better partition the role of geography and ecology
on the sharing of parasites by different host species (Davies
& Pedersen, 2008; Cooper et al., 2012).
Spillover and acquisition of parasites and pathogen are
important in many ecological systems of wildlife and domes-
tic or commercial species (Colla et al., 2006; Wood et al.,
2012). Understanding the underlying mechanism for better
predicting how particular species are under threat is typically
challenged by disentangling geographical and ecological
aspects. Pathogen transmission and spillover among species
may include complex dynamics of the ‘geographic’ compo-
nent: variation in the attraction of interacting species can
12345
–0.5 0.0 0.5
Taxonomic distance
Effect size
–0.5 0.0 0.5
LC NT VU ED CD
IUCN conservation status
Figure 4 Posterior estimates of the relative impact of
taxonomic distance from the genus ‘Rattus’ and the IUCN
conservation status on the infestation probability of mammal
species with the parasitic helminth species recorded in the two
focal rat species Rattus rattus and R. norvegicus. Posterior modes
are plotted as squares; 95% credible intervals as bars. Taxonomic
distance indexed between 1 and 5 is based on species’ genus,
family and order classification;IUCN conservation status ranges
from least concern (LC) to critically endangered (CD).
Diversity and Distributions, 21, 477–486, ª2014 John Wiley & Sons Ltd 483
Parasite geography and spillover effects
occurrence probability of parasites across climate zones l
Ω
(p,
c) had a positive impact on the infection probability for only
7 of 241 parasite species in R. rattus and for eight parasite
species in R. norvegicus (lower limits of CI >0 for a
2
).
Average infection probability of other mammal species
with helminths decreased considerably with taxonomic dis-
tance from the genus Rattus (Fig. 4), and it also decreased
with increasingly endangered status (according to their IUCN
status) (Fig. 4). However, the species turnover in overall
mammal assemblages in different zoogeographic regions was
not correlated with the species turnover of parasite assem-
blages in the two rat species (both Mantel tests with Pear-
son’s correlation coefficients r<0.27).
DISCUSSION
Inferring host–parasite associations for two of the most com-
mon and invasive commensal rat species at a global scale
showed that species richness and assemblage composition of
parasitic helminths varied over zoogeographic regions. Geo-
graphic variation in parasite species richness and assemblage
composition was correlated between the two focal host
species (Rattus rattus and R. norvegicus), although locally
they were associated with distinct parasite assemblages. Fur-
ther, our hierarchical model framework showed a clear influ-
ence of local species pools of wildlife hosts on parasite
Table 1 Summary of species richness and spatial turnover
(meanb
sim
) of helminth parasite assemblages in the two host
species Rattus rattus and R. norvegicus in different zoogeographic
regions as defined by (Holt et al., 2013). For species richness,
recorded numbers are given as S
Rec
, while posterior estimates are
given as S
Est
. Spatial turnover estimates of meanb
sim
are
calculated as the mean of all pairwise b
sim
values from different
locations within regions. 95% credible intervals for posterior
estimates are given in parenthesis
Region S
Rec
S
Est
Meanb
sim
R. rattus
Afrotropical 27 40 (35–47) 0.47 (0.34–0.53)
Australian 11 17 (14–21) 0.49 (0.38–0.56)
Madagascan 0 1 (0–3) 0.99 (0.2–1)
Nearctic 3 15 (9–21) 0.55 (0.4–0.67)
Neotropical 16 24 (19–30) 0.51 (0.38–0.58)
Oceanian 9 15 (11–19) 0.58 (0.46–0.68)
Oriental 64 71 (67–76) 0.38 (0.24–0.43)
Palaearctic 48 67 (59–74) 0.39 (0.25–0.45)
Panamanian 6 15 (9–22) 0.61 (0.47–0.69)
Saharo-Arabian 25 30 (27–36) 0.53 (0.4–0.59)
Sino-Japanese 15 28 (22–34) 0.56 (0.42–0.63)
R. norvegicus
Afrotropical 0 24 (17–33) 0.58 (0.45–0.68)
Australian 13 19 (16–23) 0.49 (0.36–0.55)
Madagascan 0 1 (0–3) 0.99 (0.3–1)
Nearctic 27 34 (30–43) 0.54 (0.41–0.59)
Neotropical 19 29 (24–35) 0.59 (0.44–0.65)
Oceanian 0 11 (5–16) 0.54 (0.38–0.69)
Oriental 21 41 (33–48) 0.54 (0.4–0.62)
Palaearctic 97 100 (95–107) 0.26 (0.21–0.4)
Panamanian 6 14 (9–20) 0.58 (0.44–0.67)
Saharo-Arabian 26 33 (28–38) 0.53 (0.4–0.58)
Sino-Japanese 30 43 (36–49) 0.55 (0.41–0.61)
●
●
●
●
●
●
●
●
●
●
●
●
020406080
020406080100
●
●
●
●
●
●
●
●
●
●
●
R. rattus −parasite number
R. norvegicus− parasite number
Helminth species richness
Figure 2 Relationship in the estimated numbers of helminth
species associated with the two host species Rattus rattus and
R. norvegicus in different zoogeographic regions given as
posterior estimates of modes (points) and 95% credible intervals
(bars). The dashed line indicates a 1 : 1 relationship.
●
●
●
●
●
●
●
●
●
●
●●
0.2 0.4 0.6 0.8 1.0
0.2 0.4 0.6 0.8 1.0
●
●
●
●
●
●
●
●
●
●●
R. rattus −β
Sim
R.norvegicus −β
Sim
Uniqueness of helminth assemblages
Figure 3 Distinctness of parasitic helminth assemblages
associated with the two host species Rattus rattus and
R. norvegicus in different zoogeographic regions as calculated
from averaged spatial turnover estimates (modes of posterior
samples are plotted as points and 95% credible intervals as
bars). The dashed line indicates a 1 : 1 relationship.
482 Diversity and Distributions, 21, 477–486, ª2014 John Wiley & Sons Ltd
K. Wells et al.
Goodman, S.M. (1995) Rattus on Madagascar and the
dilemma of protecting the endemic rodent fauna. Conser-
vation Biology,9, 450–453.
Guilhaumon, F., Krasnov, B.R., Poulin, R., Shenbrot, G.I. &
Mouillot, D. (2012) Latitudinal mismatches between the
components of mammal-flea interaction networks. Global
Ecology and Biogeography,21, 725–731.
Harris, D.B., Gregory, S.D. & MacDonald, D.W. (2006)
Space invaders? A search for patterns underlying the coex-
istence of alien black rats and Gal
apagos rice rats. Oecolo-
gia,149, 276–288.
Hoberg, E.P. (2010) Invasive processes, mosaics and the
structure of helminth parasite faunas. Revue Scientifique et
Technique,29, 255–272.
Holt, B.G., Lessard, J.-P., Borregaard, M.K., Fritz, S.A., Ara
ujo,
M.B., Dimitrov, D., Fabre, P.-H., Graham, C.H., Graves,
G.R., Jønsson, K.A., Nogu
es-Bravo, D., Wang, Z., Whittaker,
R.J., Fjelds
a, J. & Rahbek, C. (2013) An update of Wallace’s
zoogeographic regions of the world. Science,339,74–78.
Hopkins, M.E. & Nunn, C.L. (2007) A global gap analysis of
infectious agents in wild primates. Diversity and Distribu-
tions,13, 561–572.
Hortal, J., Lobo, J.M. & Jim
enez-Valverde, A. (2007) Limita-
tions of biodiversity databases: case study on seed-plant
diversity in Tenerife, Canary Islands. Conservation Biology,
21, 853–863.
Jenkins, C.N., Pimm, S.L. & Joppa, L.N. (2013) Global pat-
terns of terrestrial vertebrate diversity and conservation.
Proceedings of the National Academy of Sciences of the Uni-
ted States of America,110, E2602–E2610.
Keating, K.A. & Cherry, S. (2004) Use and interpretation of
logistic regression in habitat selection studies. Journal of
Wildlife Management,68, 774–789.
Kottek, M., Grieser, J., Beck, C., Rudolf, B. & Rubel, F.
(2006) World map of the K€
oppen-Geiger climate classifica-
tion updated. Meteorologische Zeitschrift,15, 259–263.
Krasnov, B.R., Shenbrot, G.I., Khokhlova, I.S. & Degen, A.A.
(2004) Relationship between host diversity and parasite
diversity: flea assemblages on small mammals. Journal of
Biogeography,31, 1857–1866.
Lachish, S., Gopalaswamy, A.M., Knowles, S.C.L. & Sheldon,
B.C. (2012) Site-occupancy modelling as a novel frame-
work for assessing test sensitivity and estimating wildlife
disease prevalence from imperfect diagnostic tests. Methods
in Ecology and Evolution,3, 339–348.
Lennon, J.J., Koleff, P., Greenwood, J.J.D. & Gaston, K.J.
(2001) The geographical structure of British bird distribu-
tions: diversity, spatial turnover and scale. Journal of Ani-
mal Ecology,70, 966–979.
Lomolino, M.V. (2004) Conservation biogeography. Frontiers
of Biogeography: New Directions in the Geography of Nature
(ed. by M.V. Lomolino and L.R. Heaney), pp. 293–296. Si-
nauer Associates, Sunderland, MA.
Lunn, D., Spiegelhalter, D., Thomas, A. & Best, N. (2009)
The BUGS project: evolution, critique and future direc-
tions. Statistics in Medicine,28, 3049–3067.
MacKenzie, D.I., Nichols, J.D., Lachman, G.B., Droege, S.,
Royle, J.A. & Langtimm, C.A. (2002) Estimating site occu-
pancy rates when detection probabilities are less than one.
Ecology,83, 2248–2255.
Martin, T.G., Wintle, B.A., Rhodes, J.R., Kuhnert, P.M.,
Field, S.A., Low-Choy, S.J., Tyre, A.J. & Possingham, H.P.
(2005) Zero tolerance ecology: improving ecological infer-
ence by modelling the source of zero observations. Ecology
Letters,8, 1235–1246.
McArt, S.H., Koch, H., Irwin, R.E. & Adler, L.S. (2014)
Arranging the bouquet of disease: floral traits and the
transmission of plant and animal pathogens. Ecology Let-
ters,17, 624–636.
Meerburg, B.G., Singleton, G.R. & Kijlstra, A. (2009)
Rodent-borne diseases and their risks for public health.
Critical Reviews in Microbiology,35, 221–270.
Morand, S. (2012) Phylogeography helps with investigating
the building of human parasite communities. Parasitology,
139, 1966–1974.
Morand, S. & Krasnov, B.R. (2010) The Biogeography of
Host-Parasite Interactions. Oxford University Press, Oxford.
Morand, S., McIntyre, K.M. & Baylis, M. (2014) Domesti-
cated animals and human infectious diseases of zoonotic
origins: domestication time matters. Infection, Genetics and
Evolution,24,76–81.
Ostfeld, R.S. & Keesing, F. (2012) Effects of host diversity on
infectious disease. Annual Review of Ecology, Evolution, and
Systematics,43, 157–182.
Patz, J.A., Graczyk, T.K., Geller, N. & Vittor, A.Y. (2000)
Effects of environmental change on emerging parasitic dis-
eases. International Journal for Parasitology,30, 1395–1405.
Poulin, R. & Mouillot, D. (2003) Host introductions and the
geography of parasite taxonomic diversity. Journal of Bioge-
ography,30, 837–845.
Reese, G.C., Wilson, K.R., Hoeting, J.A. & Flather, C.H.
(2005) Factors affecting species distribution predictions: a
simulation modeling experiment. Ecological Applications,
15, 554–564.
Robins, J.H., McLenachan, P.A., Phillips, M.J., Craig, L.,
Ross, H.A. & Matisoo-Smith, E. (2008) Dating of diver-
gences within the Rattus genus phylogeny using whole
mitochondrial genomes. Molecular Phylogenetics and Evolu-
tion,49, 460–466.
Rondinini, C., di Marco, M., Chiozza, F., Santulli, G., Baise-
ro, D., Visconti, P., Hoffmann, M., Schipper, J., Stuart,
S.N., Tognelli, M.F., Amori, G., Falcucci, A., Maiorano, L.
& Boitani, L. (2011) Global habitat suitability models of
terrestrial mammals. Philosophical Transactions of the Royal
Society B: Biological Sciences,366, 2633–2641.
Sheppard, S.K., Colles, F., Richardson, J., Cody, A.J., Elson,
R., Lawson, A., Brick, G., Meldrum, R., Little, C.L., Owen,
R.J., Maiden, M.C.J. & McCarthy, N.D. (2010) Host associ-
ation of Campylobacter genotypes transcends geographic
variation. Applied and Environmental Microbiology,76,
5269–5277.
Diversity and Distributions, 21, 477–486, ª2014 John Wiley & Sons Ltd 485
Parasite geography and spillover effects
Singleton, G.R., Sudarmaji & Brown, P.R. (2003) Compari-
son of different sizes of physical barriers for controlling the
impact of the rice field rat, Rattus argentiventer, in rice
crops in Indonesia. Crop Protection,22,7–13.
Stenseth, N., Leirs, H., Skonhoft, A., Davis, S., Pech, R., An-
dreassen, H., Singleton, G., Lima, M., Machang’u, R., Mak-
undi, R., Zhang, Z., Brown, P., Shi, D. & Wan, X. (2003)
Mice, rats, and people: the bio-economics of agricultural
rodent pests. Frontiers in Ecology and the Environment,1,
367–375.
Taylor, L.H., Latham, S.M. & Woolhouse, M.E.J. (2001) Risk
factors for human disease emergence. Philosophical Trans-
actions of the Royal Society B Biological Sciences,356, 983–
989.
Thieltges, D.W., Hof, C., Dehling, D.M., Br€
andle, M., Brandl,
R. & Poulin, R. (2011) Host diversity and latitude drive
trematode diversity patterns in the European freshwater
fauna. Global Ecology and Biogeography,20, 675–682.
Thompson, J.N. & Cunningham, B.M. (2002) Geographic
structure and dynamics of coevolutionary selection. Nature,
417, 735–738.
Torchin, M.E., Lafferty, K.D., Dobson, A.P., McKenzie, V.J.
& Kuris, A.M. (2003) Introduced species and their missing
parasites. Nature,421, 628–630.
Ward, G., Hastie, T., Barry, S., Elith, J. & Leathwick, J.R.
(2009) Presence-only data and the em algorithm. Biomet-
rics,65, 554–563.
Wells, K. & O’Hara, R.B. (2013) Species interactions: esti-
mating per-individual interaction strength and covariates
before simplifying data into per-species ecological net-
works. Methods in Ecology and Evolution,4,1–8.
Wells, K., O’Hara, R.B., Pfeiffer, M., Lakim, M.B., Petney,
T.N. & Durden, L.A. (2013) Inferring host specificity and
network formation through agent-based models: tick–
mammal interactions in Borneo. Oecologia,172, 307–316.
Wells, K., Lakim, M.B. & O’Hara, R.B. (2014) Shifts from
native to invasive small mammals across gradients from
tropical forest to urban habitat in Borneo. Biodiversity and
Conservation,23, 2289–2303.
Wilson, D.E. & Reeder, D.M. (2005) Mammal Species of the
World. A Taxonomic and Geographic Reference, 3rd edn.
Johns Hopkins University Press, Baltimore.
Wood, J.L.N., Leach, M., Waldman, L., MacGregor, H., Fo-
oks, A.R., Jones, K.E., Restif, O., Dechmann, D., Hayman,
D.T.S., Baker, K.S., Peel, A.J., Kamins, A.O., Fahr, J., Ntia-
moa-Baidu, Y., Suu-Ire, R., Breiman, R.F., Epstein, J.H.,
Field, H.E. & Cunningham, A.A. (2012) A framework for
the study of zoonotic disease emergence and its drivers:
spillover of bat pathogens as a case study. Philosophical
Transactions of the Royal Society B,367, 2881–2892.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the
online version of this article:
Appendix S1 List of parasite species and sampled locations.
Table S1 List of parasitic helminth species and associated
number of mammal hosts.
Table S2 Assignments of sampled locations to zoogeographic
regions.
Appendix S2 Model code in BUGS language.
BIOSKETCHES
The authors of this study have various interests linked to
biodiversity, parasitology, conservation, species distributions,
ecohealth, biotic interactions, and hierarchical models, and
assembled as a multidisciplinary team. All authors contrib-
uted jointly to this study, mostly by asking na€
ıve questions
to each other that helped to critically scrutinize approaches
and synthesize different views into this study.
Editor: Jacqueline Beggs
486 Diversity and Distributions, 21, 477–486, ª2014 John Wiley & Sons Ltd
K. Wells et al.
