Invasions and Infections
The effects of invasion on parasite dynamics and communities
Correspondence author. E-mail: firstname.lastname@example.org
- Invasions can impact on parasite communities through both the introduction of exotic parasite species and effects of invading hosts on native parasite dynamics. However, our understanding of the factors that influence the invasive process and mediate impacts on native hosts and parasites is limited.
- Using models of host–parasite dynamics as a framework, we explore how both the probability of spread for an exotic parasite and impacts of introduced species on native parasite dynamics depend on key parameters related to rates of encounter, transmission, mortality and recovery. We examine how invasions may interact with the diverse range of underlying biological mechanisms that can affect these rates. We specifically highlight the potential role of interactions between parasites, which has largely been ignored.
- For introduced parasites, high abundance of competent hosts and vectors within native communities can greatly facilitate spread. Introduced host species can cause amplification or dilution effects for native parasite dynamics, with the direction and magnitude of the effect determined by how the invasion influences the competency and abundance or relative abundance of the host community (community capacity).
- Invasions by exotic parasites and changes to endemic parasite dynamics following invasions may reflect numerical and functional processes in multihost single-parasite systems (e.g. influence of host and vector community structure on encounter rates). However, as co-infection can influence factors such as susceptibility and infection length, effects may also be mediated by within-host interactions between parasites. The ultimate effect of an invasion will depend on the community-wide summed direct and indirect impacts.
- Future studies should aim to further elucidate the key processes influencing disease dynamics in multihost (and multivector) communities, thereby informing predictions of how invasive host and parasite species and changes in biodiversity will influence disease risk. Theoretical studies should incorporate host interspecific competition and relax assumptions regarding the relationship between intra- and interspecific contact rates and density. Empirical and experimental studies should not only quantify the relative importance of host (and vector) density and diversity, but also consider other community interactions such as those between parasites.
Provide feedback or get help You are viewing our new enhanced HTML article.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Introduced host and parasite species can drive changes in community structure and biodiversity loss (Mack et al. 2000; McGeoch et al. 2010). Understanding those factors that influence the invasive process and mediate impacts on native communities will become increasingly important as the number of introductions of non-indigenous species continues to accelerate. Parasites may play an important role, either as invasive species or by mediating interactions between other community members. Indeed, the increasing evidence that parasites can drive changes in host population abundance (Hudson, Dobson & Newborn 1998; Kohler & Hoiland 2001; Albon et al. 2002) and influence food web stability (Lafferty et al. 2008; Chen et al. 2011) highlights the potential importance of host–parasite interactions at the population and community level. In turn, community interactions and structure can influence parasite dynamics. Although many studies of parasites in the context of invasions focus on the impact on native host populations or whether release from parasites facilitates the spread of introduced species (Cleaveland et al. 2002; Mitchell & Power 2003; Torchin et al. 2003; Colautti et al. 2004), in this review, we examine the impact of invasions on parasite dynamics and communities.
Global trade and travel can have direct impacts on parasite communities by the introduction of non-indigenous parasite species. Coincident introduction of both host and parasite can be followed by simultaneous spread, with potentially catastrophic consequences in cases where the introduced parasite ‘spills over’ to infect native species (Cleaveland et al. 2002). Squirrelpox virus was introduced to the United Kingdom with the grey squirrel (Sciurus carolinensis), and both empirical and modelling studies indicate that spillover infection is driving the grey squirrel's replacement of native red squirrel (S. vulgaris) populations (Tompkins, White & Boots 2003; Rushton et al. 2006). However, parasite species may persist even where the introduced host species fails to spread, or, for parasites with free-living stages, without the concurrent introduction of host species (‘pathogen pollution’; Kelly et al. 2009a). West Nile Virus (WNV) was introduced to the USA in 1999, probably by an infected mosquito or bird host, but rapidly established within existing communities of passerine birds and mosquitoes, spreading throughout the country within 4 years (Hayes et al. 2005), and ballast water from transoceanic shipping has been implicated in the introduction of marine parasites such as parasitic isopods (Chapman et al. 2012). Invasive host species may themselves facilitate subsequent invasion of exotic parasites. The introduction of the sibling vole (Microtus rossiaemeridionalis) to Svalbard has enabled the local establishment of the tapeworm Echinococcus multilocularis by providing a suitable intermediate host, where none previously existed (Hentonnen et al. 2001).
Native parasite dynamics may also be affected by invasions, with introduced host species acting to increase prevalence (amplification effects) or decrease prevalence (dilution effects). Amplification effects may occur where the introduced species acts as an alternative host for the parasite, with subsequent ‘spillback’ of infection to native host species (Daszak, Cunningham & Hyatt 2000; Kelly et al. 2009b). In North America, the presence of barley and cereal yellow dwarf viruses (B/CYDV) appears to have been essential for the widespread invasion by Mediterranean grass species (Malmstrom et al. 2005). The incidence of B/CYDV infections in native grass species more than doubled when they were grown in plots with introduced annual grass species owing to spillback of infections (Malmstrom et al. 2005). In contrast, in New Zealand, infections by several helminth parasites in two native fish species decreased with increasing abundance of introduced brown trout (Salmo trutta) (Kelly et al. 2009a).
Thus, host and parasite invaders can have diverse effects on parasite communities. Here, to explore and understand these effects, we consider how and why invasions may influence the key parameters of host–parasite models, leading to changes in the abundance of infected hosts. A similar approach has proved useful in examining why high levels of biodiversity may lead to a dilution effect for pathogens (Keesing et al. 2010) and parasites with complex life histories (Johnson & Thieltges 2010). Specifically, we use this framework to examine (i) what factors influence establishment success in introduced parasites, (ii) what factors determine whether the invasion of novel hosts (or vectors) leads to the amplification or dilution effects for native parasites, and (iii) what mechanisms underlie observed patterns. Although mechanisms may be linked to host–parasite interactions themselves, other community interactions such as interspecific competition and predation may also play a role. Indeed, there is increasing evidence that interactions between parasite species can influence infection dynamics at individual host and population levels (Box ), and such interactions could be important in the context of invasions. We also consider the role of additional factors such as seasonality, spatial structure and evolution and examine to what extent we can predict outcomes. We hope to generate a more complete understanding of key processes and inform attempts to predict and manage the impact of host and parasite invaders on disease prevalence.
Box 1 Within-host parasite communities and interactions
Macroparasites and microparasites are integral components of ecosystems and communities (Pedersen & Fenton 2006). As most hosts are likely to be co-infected or sequentially infected with several parasite species, interactions frequently occur. These interactions can be synergistic or antagonistic and mediated by direct competition for resources or indirect competition via host immune responses (see reviews in Pedersen & Fenton 2006; Graham et al. 2007). Interactions between parasites can have significant consequences for host and parasite fitness through impacts on host susceptibility, infection length, infection intensity, morbidity and mortality rates (Lello et al. 2004; Cattadori, Boag & Hudson 2008; Jolles et al. 2008; Telfer et al. 2008). The magnitude of these effects can be considerable. In an analysis of infection risk for a microparasite community infecting natural populations of the field vole Microtus agrestis, the effect of other infections was typically of greater magnitude and explained more variation in risk than the effects associated with host and environmental factors (Telfer et al. 2010). In turn, there is mounting evidence that these individual-level effects can influence host–parasite dynamics at the population level (Lello et al. 2008; Ezenwa & Jolles 2011).
Key processes influencing host–parasite dynamics
Host–parasite model structure and key parameters depend on parasite life cycle and transmission route. However, models for directly transmitted pathogens, vector-borne pathogens and macroparasites (Randolph et al. 2002; Swinton et al. 2002; Wilson et al. 2002) all highlight the importance of rates of encounter between susceptible individuals and infectious hosts, vectors or particles, transmission rates following an encounter and mortality and recovery rates of infected hosts. Here, we take each of these parameters in turn and consider (i) key predictions from theoretical models regarding parasite establishment and dynamics, and (ii) how interactions between existing communities and non-indigenous parasites, vectors and hosts may influence the parameter. In Table 1, we use this framework to summarize potential effects of introduced hosts on native parasite dynamics.
Potential effects of introduced hosts on native parasite dynamics through impacts on the four key parameters in parasite models (encounter, transmission, mortality and recovery rates). Possible mechanisms are described, as well as highlighting how mechanisms and impacts may depend on transmission mode and transmission function (the relationship between host abundance and contact rates)
Encounter rates: host competence, host abundance and contacts
Crucial for transmission, and therefore disease incidence, is exposure rate, determined by the encounter rate between susceptible hosts and infected hosts, vectors or infective stages. The encounter rate depends on the abundance of susceptible and infected hosts, vectors or particles and on patterns of mixing and contact.
Central to the issue of host abundance within a community is host competence. Although some parasites are specialist, infecting one host species, many species can infect multiple host species. However, competence, as assessed by the capability of a host species to disseminate the infective stages of the parasite, can vary greatly. In a study of 25 North American bird species experimentally infected with WNV, a reservoir competence index, based on susceptibility, infectiousness and infection duration, varied from 0 to 2·55 (Komar et al. 2003).
Clearly, for exotic parasites, invasion is much more likely if native species prove competent hosts. Examination of WNV transmission in 31 suburban sites in the USA revealed that the American robin (Turdus migratorius) was the key host, accounting for nearly 87% of the aggregate force of infection (calculated as the fraction of Culex mosquito blood meals taken from the host multiplied by the WNV competence index for that host) (Hamer et al. 2011). Similarly, the impact of an introduced species on dynamics of a native parasite will be strongly influenced by the extent to which it proves to be a competent host (Table 1). At one extreme, the introduced species could be completely unsuitable with R0 = 0 (R0 = basic reproduction number, the mean number of expected new infections caused by a single infected individual), whilst at the other the introduced species could be more competent than native hosts. Although trematodes may use a variety of species as secondary intermediate and final hosts, they are often constrained to a single first intermediate host species. In New Zealand, eggs from native Microphallus species that develop following ingestion by the native snail Potamopyrgus antipodarum cannot develop inside the invasive snail Lymnea stagnalis (Kopp & Jokela 2007). In contrast, in North America, the introduced house sparrow (Passer domesticus) develops higher viraemia after infection with Buggy Creek virus (BCRV) than the original host of the virus, the North American cliff swallow (Petrochelidon pyrrhonota) (O'Brien et al. 2011).
Contact Rates and Host Abundance
In host–parasite models, encounter rates between susceptible and infective hosts or stages are typically a function of either the density of infected hosts (density-dependent) or the proportion of the population that is infected (frequency-dependent). Directly transmitted parasites and parasites with free-living stages are normally assumed to have density-dependent encounter rates, whilst frequency-dependent models are typically used to model systems where the number of contacts made between individuals does not increase with density, for example sexually transmitted diseases or animal populations with strong territorial behaviour (McCallum, Barlow & Hone 2001; Begon et al. 2002). Vector-borne pathogens can also often be modelled under assumptions of frequency dependence (Thrall, Antonovics & Hall 1993).
Under density-dependent transmission, the density of susceptible hosts must exceed a threshold for pathogen establishment and persistence (R0 > 1), whilst no such threshold exists for diseases with frequency-dependent transmission (McCallum, Barlow & Hone 2001; Begon et al. 2002). More realistic stochastic models highlight the importance of stochastic fade-out of infection, with increased risk of non-persistence in small populations, irrespective of the mode of transmission (Swinton et al. 2002). Extensions of deterministic models to multihost systems with density-dependent transmission similarly demonstrate the importance of a critical community size. In general, any increase in the number of host species that leads to an increase in the total density of hosts will increase contact rates, increasing R0 and amplifying disease incidence (Holt et al. 2003; Dobson 2004; Rudolf & Antonovics 2005). Such an effect will occur even if the additional host is not particularly competent, as long as interspecific transmission is not zero. The level of amplification will depend on both the competence of the additional host and the relative rates of inter- and intraspecific transmission, with highest amplification at high rates of interspecific transmission (Dobson 2004).
In contrast, multihost models with frequency-dependent transmission suggest that an increase in the number of host species can inhibit parasite establishment (Dobson 2004). Crucial to this result are two assumptions: the additional host is not of higher competence and between-species transmission is less than within-species transmission. Consequently, as contact rates are constant, the decline in the relative frequency of within-species contacts leads to a decline in R0, and the larger the asymmetry in inter- and intratransmission rates, the greater the dilution effect (Dobson 2004; Rudolf & Antonovics 2005). A more recent study with frequency-dependent transmission, which incorporated both species competency and relative abundance into measures of reservoir capacity, highlighted that an increase in disease prevalence with increasing species diversity can occur if the additional species increases mean capacity of the host community (Brooks & Zhang 2010).
These results from theoretical studies inform predictions regarding invasions and parasite dynamics. In terms of parasite species introduced with their host, those that exhibit a density-dependent mode of transmission may find it harder to establish and spread, as initial low host densities may be under persistence thresholds (Torchin et al. 2003; Colautti et al. 2004). However, if such a parasite successfully spills over into one or more competent native host species, spread may prove easier than for a parasite with frequency-dependent transmission that is faced with a host community of varying competence levels (Table 2).
Table 2. Summary of factors predicted to favour (i) establishment and invasion of introduced parasites, (ii) introduced hosts amplifying native parasite prevalence and (iii) introduced hosts diluting native parasite prevalence
|Parasite traits||Simple life cycle; chronic infection; vertical transmission|| || |
|Host Competence||High competence of native hosts/vectors||High competence of introduced host: low host resistance; low investment in immune defences||Low competence of introduced host relative to native hosts: parasite adaptation to local hosts|
|Transmission rates|| |
|Density-dependent; interspecific transmission rates high relative to intratransmission rates||Frequency-dependent; interspecific transmission rates low relative to intratransmission rates|
|Impacts of introduction on native hosts||Low virulence among native hosts or simultaneous invasion by introduced host||No negative impacts of introduced host on abundance of key native hosts||Negative impacts of introduced host on abundance of key native hosts|
|Site/community traits||Global connectivity of site; climatic similarity to native range||Introduced host phylogenetically similar to native host community||Introduced host phylogenetically dissimilar to native hosts|
Concerning the impact of introduced hosts on native parasite dynamics, for parasites with density-dependent transmission, introduced species with higher than zero competence are likely to increase transmission rates and prevalence (Table 1). Decreases in incidence following invasion (dilution effects) are predicted to be more likely for parasites with frequency-dependent transmission. Bank voles (Myodes glareolus) were introduced to south-west Ireland around 60 years ago and are continuing to spread. We studied small mammal populations behind and ahead of the invasion front and found that the prevalence of two species of flea-borne Bartonella in wood mice (Apodemus sylvaticus) populations declined with increasing density of the uninfected bank voles (Telfer et al. 2005). Dilution effects following introductions have also been observed in trematode infections of mussels and helminth infections in fish (Thieltges et al. 2008; Kelly et al. 2009a). It has been argued that dilution effects are particularly likely in vector-borne pathogens (Ostfeld & Keesing 2000) and parasites with complex life cycles (Johnson & Thieltges 2010). In the former, non-competent or poor hosts may reduce the proportion of infected vectors that feed on competent hosts, whilst in the latter attempts by free-living stages to infect unsuitable hosts may also lead to wasted transmission events. However, native parasites with frequency-dependent transmission can also exhibit amplification effects if the competency of the introduced species, weighted by its relative abundance, increases mean community capacity (Brooks & Zhang 2010). A recent study on BCRV has demonstrated the potential impact of the introduction of a highly competent host species (O'Brien et al. 2011). In cliff swallow populations, BCRV is transmitted between birds by the swallow bug (Oeciacus vicarius). Over the last century, cliff swallow nesting colonies have been invaded by the introduced but highly competent house sparrow. In a survey of 22 colonies, virus prevalence in sparrows was eight times higher than in swallows, and the prevalence of infected swallow bugs at a site was related to prevalence in sparrows, but not prevalence in swallows, indicating that the sparrow now drives infection dynamics within mixed colonies.
It is important to note, however, that most theoretical explorations of multihost systems assume no competitive or predatory interactions between the host species. Clearly, any interspecific interaction that alters the absolute or relative abundance of key host species can indirectly influence parasite dynamics through changes to the encounter rate (Keesing, Holt & Ostfeld 2006; Keesing et al. 2010) (Table 1). Thus, even for parasites with density-dependent transmission, an invasive species that reduces the abundance of a key native host species may lead to a dilution effect (Rudolf & Antonovics 2005). The potential for drastic, indirect effects on native parasite prevalence and parasite communities is illustrated by trematodes from the western coast of North America that infect the native snail Cerithidea californica as a first intermediate host. The invasive Japanese mud snail (Batillaria cumingi) is replacing C. californica (Byers & Goldwasser 2001) and, as native trematode species are unable to infect the invader, the decline and extinction of C. californica populations will ultimately lead to local extinction of this parasite community of at least ten species (Torchin, Byers & Huspeni 2005).
Vector-Borne Pathogens: Vector Abundance and Competence
For vector-borne pathogens, transmission depends on the encounter rate between infected vectors and susceptible hosts (Randolph et al. 2002). Thus, both the proportion of vectors infected and vector abundance are important. Moreover, as with host communities, vector species richness and the proportion of competent vectors can have important implications for pathogen dynamics. Recent theoretical studies have predicted that whereas increased host biodiversity can decrease pathogen prevalence because of a general tendency to increase the proportion of host species with low susceptibility, increases in vector species richness will tend to amplify transmission, as, assuming no negative interspecific interactions between vectors, even vector species with low competence will increase summed exposure in reservoir hosts (Brooks & Zhang 2010; Roche & Guégan 2011). The successful establishment and spread of several exotic vector-borne diseases, including WNV in North America and bluetongue virus in Europe, emphasizes that the existence of a large, competent resident vector community can facilitate disease outbreaks in novel areas (Randolph & Rogers 2010).
For native vector-borne pathogens, the outcome following invasion by a potential new host species will depend not only on the new host's competency, but also on whether the introduction influences vector feeding behaviour and abundance through, for example, predation (Keesing et al. 2009) or the provision of additional feeding opportunities (Dobson 2004) (Table 1). The increase in B/CYDV prevalence observed in native bunchgrass species when grown together with introduced grass species appears to be driven by an increase in aphid abundance. Aphids both prefer to feed on the introduced annuals and achieve significantly higher fecundity on these plants (Malmstrom et al. 2005). Such increases in vector abundance induced by the presence of a preferred host species can lead to increased pathogen prevalence, even where the host is not a competent reservoir for the pathogen (Gilbert et al. 2001; Swei et al. 2011).
Free-Living Infective Stages
Many macroparasites and some microparasites have free-living infective stages that survive outside the host, for example fungal spores or trematode cercariae. Models for the dynamics of such parasites emphasize that encounter rates between susceptible hosts and infective stages depend on rates of propagule production and survival (Wilson et al. 2002). There is large potential for factors that influence propagule production to have population-level effects, as production can vary by orders of magnitude between hosts. For native parasites, invasion of a non-indigenous host species may affect the dynamics of free-living stages through a variety of mechanisms, including predation, interference competition and attempted infections of low-competence hosts (Johnson & Thieltges 2010). In addition, impacts could be mediated by changing interactions between co-infecting parasite species. In a study of gut helminths in wild rabbit populations, Lello et al. (2004) showed that infection intensity was influenced by a network of positive and negative interactions between the parasites. If a biological invasion alters the relative prevalence of different macroparasite species, such within-host interactions could drive further changes in the parasite community through impacts to the production rate of infective stages (Box ).
Parasite dynamics will also be affected by the probability that successful transmission occurs after contact between a susceptible host and an infective host, vector or stage. Indeed, intraspecific variation in susceptibility (and infectiousness) is crucial in determining the probability of parasite invasion and persistence (Lloyd-Smith et al. 2005). Host susceptibility can be influenced by a range of factors, including genetics and the biotic environment. Changes to the diversity and structure of communities following an invasion could influence susceptibility through a range of mechanisms (Table 1). Increased resource availability may enhance immunological defence mechanisms against infection (Lafferty & Holt 2003). Prior or current infections with other parasite species or strains can also either increase or decrease susceptibility (Box ) (Cattadori, Boag & Hudson 2008; Telfer et al. 2010), with increases in susceptibility mediated by specific immune responses or general impacts of infection on host condition and immunocompetence (the ‘vicious circle’; Beldomenico & Begon 2009).
Bovine tuberculosis (Mycobacterium bovis BTB) in South Africa provides an interesting example of the potential importance of interspecific parasite interactions for invasion success of an exotic parasite. BTB was first introduced with cattle from Europe and has since become endemic in African buffalo (Syncerus caffer) populations (Michel et al. 2006). At population, herd and individual levels, TB infections are negatively correlated with gastrointestinal nematode infections, and co-infections also occur less often than expected (Jolles et al. 2008). Buffalo with higher resistance to worms (determined by lower faecal egg counts) exhibit a weaker T-helper type1 (Th1) immune function and a stronger T-helper type2 (Th2) response (Ezenwa et al. 2010). Th1 cells play a key role in triggering immune responses to intracellular pathogens, including TB, whilst Th2 cells promote mechanisms targeting macroparasites such as helminths (Mosmann & Sad 1996). The variation in nematode resistance was partly associated with genotypic variation, but enhanced Th1 immunity following antihelminthic treatment also indicated that nematode infections suppressed Th1 responses. Models parameterized by the empirical data predict that TB will fail to invade in the absence of nematodes, whilst increasing nematode transmission leads to increases in the R0 for TB to above 1, allowing invasion (Ezenwa et al. 2010). Interactions between parasites could also be responsible for some amplification and dilution effects observed in native parasites following community change (Box ).
Box 2 Parasite interactions and community structure change
Few studies have considered the potential for parasite interactions to mediate the impact of community structure change on parasite dynamics (Johnson & Thieltges 2010). It is therefore useful to highlight how altering parasite interactions may have profound indirect consequences for the dynamics of the parasite community as a whole. For example, consider two species of parasites that compete within a native host, an introduced host that increases the prevalence of one of the parasites through spillback could increase the competitive advantage of this parasite in the native host through priority effects (increasing the proportion of individuals that are exposed to this parasite first), further increasing prevalence of the parasite in the native species (enhanced amplification effect) and decreasing prevalence of the second parasite (giving rise to a dilution effect, but mediated by interspecific parasite interactions). Similarly, synergistic interactions between parasites could lead to several species all apparently showing a spillback effect or a dilution effect, even if the introduced host only causes such an effect directly for one of the parasite species. Although exotic parasites successfully introduced with their host often show similar prevalence in native and naturalized ranges (Torchin et al. 2003), changes to abiotic or biotic environments, including parasite community diversity, could result in changes in prevalence in invaded areas. Release from competitive interactions could result in increased prevalence in the invaded range. We do not know of any studies that have investigated such a possibility, although increased prevalence in the invaded range has been observed for a monogenean gill ectoparasite of the rabbit fish Siganus rivulatus (Pasternak, Diamant & Abelson 2007).
Mortality and recovery rates
Factors that affect mortality and/or recovery rates can influence parasite dynamics via effects on the loss rate of infected individuals. In terms of invasions, parasites introduced with their host may be more likely to persist if they have little impact on survival of the introduced host or induce chronic infections (Torchin et al. 2003). Invasive host species may influence native parasite dynamics if they directly or indirectly alter the loss rate of infected individuals (Table 1). Such effects could be mediated by a range of community and food web interactions. Potential examples include predation or interspecific competition that preferentially impacts infected individuals, or changes to parasite communities and within-host interactions that alter disease pathology or clearance rates. Thus, by removing infected individuals or prolonging infectious periods, community interactions may decrease or increase R0 and directly impact on disease dynamics. In the BTB and buffalo system described above, the observed negative correlation between TB and nematodes appears to depend on both the heterogeneity in susceptibility caused by immune trade-offs and increased mortality of co-infected hosts (Jolles et al. 2008). In contrast, interactions between the globally invasive chytrid fungus Batrachochytrium dendrobatadis (BD) and non-pathogenic skin microbes can reduce morbidity and mortality (Harris et al. 2009). Although this could reduce negative effects on some hosts (Kilpatrick, Briggs & Daszak 2009), the ultimate impact on disease dynamics is unclear and will depend on how the skin microbes influence the cumulative release of zoospores from infected individuals.
Spatial structure and seasonality
Spatial structure in host populations can have significant impacts on parasite dynamics by introducing heterogeneity into encounter rates (Hess et al. 2002). In fragmented populations, levels of between-population coupling will determine parasite dynamics and persistence (Keeling & Gilligan 2000; Keeling, Bjornstad & Grenfell 2004; Ostfeld, Glass & Keesing 2005). Recent spatial models investigating landscape-level infection dynamics for parasites that infect wild and domesticated hosts have emphasized that epidemics are driven by the host most abundant in the landscape (Fabiszewski, Umbanhowar & Mitchell 2010). Whilst in managed landscapes this may often be a domesticated species, as many invasive species are generalists, occupying a range of habitats, such species could play a similar role as ‘epidemiological bridges’, effectively reducing host spatial structure and linking infection dynamics across landscapes. In North America, introduced populations of muskoxen Ovibos moschatus have acquired the lungworm Protostrongylus stilesi from the native Dall's sheep (Ovis dalli dalli). The lungworm is now found in Dall's sheep populations further east than previously, in areas where the expanding muskoxen population also occur, suggestive of epidemiological linking of previously isolated populations facilitated by the introduced host (Hoberg et al. 2002). Alternatively, in situations where an invasive species is an incompetent host for a native parasite and reduces the abundance and distribution of native hosts through competition, the invasion may result in decreased parasite prevalence because of the fragmentation of competent host populations.
Host spatial structure can also play an important role in determining the landscape epidemiology of invading parasites. Sudden oak death, an emerging forest disease in North America, is caused by the invasive pathogen Phytophthora ramorum. Both empirical work and spatially explicit disease models indicate that the distribution and abundance of susceptible and resistant trees at both local and landscape scales is critical to disease spread and the ultimate impact on populations of the susceptible tanoak Lithocarpus densiflorus. Disease severity is greatest in landscapes with a high proportion of contiguous forest (Condeso & Meentemeyer 2007), whilst at local scales, low densities of tanoak within a matrix of non-susceptible species may reduce disease transmission sufficiently to enable the persistence of overstorey tanoak trees (Cobb et al. 2012).
Seasonality can also play an important role in the dynamics of parasites, with effects driven through the variation in encounter rates (caused by changes in behaviour, abundance or survival of free-living stages) or susceptibility of hosts (changes in immunocompetence) (Altizer et al. 2006). The ability of an introduced parasite to invade may depend on how season influences the abundance of susceptible individuals. In a multihost–multivector model of plague dynamics in California, heterogeneity in the breeding seasons of two competent host species appears to be crucial for disease persistence (Foley & Foley 2010). Similarly, introduced host species may influence native parasite dynamics by altering seasonal patterns in encounter rates. In the BCRV system, there is evidence of a spillback effect on vector abundance, with more bugs in swallow nests from mixed colonies of swallows and sparrows than in nests from swallow-only colonies. This is despite the fact that many more bugs are found in swallow nests than in sparrow nests. O'Brien et al. (2011) speculate that as sparrows are resident year-round, the increased opportunity for blood feeds may increase bug survival compared to swallow-only colonies, which are only occupied during the summer months.
Seasonality can also alter the strength of community interactions and, consequently, influence their effect on community structure. It is interesting to note that although nematode infections appear to have facilitated TB invasion in South African populations of buffalo through effects on susceptibility, the trade-off between Th-1 and Th-2 parts of the immune system is only detectable in the dry season, when resources may be limiting (Jolles et al. 2008).
Most multihost models are non-spatial and non-seasonal (Holt et al. 2003; Dobson 2004; Rudolf & Antonovics 2005). However, recent models for B/CYDVs based on a patch framework highlight how spatial and seasonal processes can influence probabilities of parasite invasion and persistence, as well as the outcome of parasite-mediated competition between host species (Moore et al. 2011). Increased patch occupancy by perennials leads to higher regional prevalence of B/CYDV as only perennials can maintain the viruses through the winter. However, a high proportion of patches with perennials reduces the spread of the virus during the growing season as aphid transmission rates are positively related to the abundance of annuals. Consequently, spatial structure can interact with host community composition to influence the probability of parasite invasion and perennial-only patches that are isolated enough to prevent annual re-introduction of the disease by dispersing aphids remain disease free and therefore resistant to invasion by introduced annuals (Moore et al. 2011).
Long-term effects and evolution
By definition, invasions have both temporal and spatial dimensions. The nature of interspecific interactions within an invaded community may change over time as introduced hosts acquire native parasites (Torchin et al. 2003). In Canada, whilst newly established populations of the Eurasian round goby harboured far fewer species of parasite than native fish species, a population established over 15 years before had similar parasite richness and abundance to native fish species (Gendron, Marcogliese & Thomas 2011).
Evolutionary processes may also change host–parasite interactions in invaded communities. In Madagascar, the black rat Rattus rattus, present since the eleventh century (Hingston et al. 2005), has facilitated the spread and persistence of plague, introduced in 1898 (Duplantier et al. 2005). Rats from the endemic plague zone have developed increased resistance to infection, associated with genetic changes. This change has occurred in less than 100 years, with high virulence resulting in strong selective pressure (Tollenaere et al. 2010; Tollenaere et al. 2011). Invasions may also drive parasite evolution. In the swallow–sparrow–BCRV system, two virus lineages are found: one predominantly circulates in the vector, the swallow bug, whilst the other is preferentially transmitted to birds. Divergence of these two lineages dates to approximately the time when sparrows began invading swallow colonies, and the introduction of the highly competent sparrow may have driven selection for virus genotypes adapted to replicate in birds (Brown et al. 2009). Such evolutionary changes can obviously have profound effects on host–parasite dynamics and disease prevalence.
Predicting disease outcomes from invasions
Clearly, it would be useful for the effective prevention and management of disease outbreaks associated with invasive species if we could use our understanding of host–parasite dynamics to predict what parasites are most likely to invade and where, and how introduced species will impact on native disease dynamics (Jones et al. 2008; Randolph & Rogers 2010; Poulin et al. 2011).
As highlighted above, certain parasite characteristics are predicted to influence the probability that an introduced parasite will establish and spread (Table 2); notably, parasites without complex life cycles, parasites that induce chronic infections of low pathogenicity and parasites with vertical transmission are more likely to persist within introduced host populations (Torchin et al. 2003). A parasite survey of plant species introduced to the United States from Europe found that more viral parasites from native ranges were successfully introduced than fungal parasites, possibly as viral parasites are more likely to be seed-transmitted and have lower virulence (Mitchell & Power 2003).
Patterns of invasion by non-indigenous species are related not just to species traits but also site-specific properties, with some communities more invasion prone than others (Thuiller et al. 2006; Gassó et al. 2009). Although invasion and disease emergence often depend on complex nonlinear interactions between biotic and abiotic factors (Jones et al. 2008; Randolph & Rogers 2010), for specific vectors or diseases, risk mapping has sometimes proved useful in identifying vulnerable areas. For example, both global connectivity based on shipping traffic volume and climatic similarity appear to have determined the global spread of the Asian tiger mosquito Aedes albopictus (Tatem, Hay & Rogers 2006).
In terms of the impact of introduced host species on native parasite dynamics, a key question is whether invasive species are more likely to act as amplifying or diluting hosts. Central to this is the issue of competency (Kelly et al. 2009b; Poulin et al. 2011), and a lack of a co-evolutionary history could result in either high competence (as a result of low host resistance) or low competence (because of a lack of suitable parasite adaptations) (Table 2). In studies of the inverse relationship between biodiversity and disease, it has been argued that parasite evolution favours specialization on widespread hosts, leading to a positive relationship between competence and typical host abundance (Ostfeld & Keesing 2000). As hosts with low relative abundance are more likely to be lost, declines in biodiversity tend to increase mean community capacity and disease prevalence (Brooks & Zhang 2010). An alternative explanation is that resilient species that persist as biodiversity declines often have high intrinsic rates of increase, invest minimally in adaptive immune responses and are consequently more likely to act as amplifying species (Keesing et al. 2010). Similar predictions have been proposed for invading host species. For example, fecundity-related characteristics such as number of offspring or seeds per breeding season are hypothesized to influence invasive propensity (Kolar & Lodge 2001; Sakai et al. 2001; van Kleunen, Weber & Fischer 2010), with successful invasion favoured in species that preferentially allocate resources to growth and reproductive effort over costly immune defence mechanisms (Lee & Klasing 2004). If traits associated with invasive behaviour are correlated with low resistance and increased competency, introduced species should tend to have amplifying effects. In vertebrates, species with high reproductive output tend to have reduced investment in some aspects of adaptive immunity (Martin, Hasselquist & Wikelski 2006; Martin, Weil & Nelson 2007), whilst fast-growing plant species can be more competent hosts for arthropod vectors and pathogens than other plant species (Cronin et al. 2010). However, the evidence that species that allocate more resources to growth and reproductive output are more invasive appears equivocal. In a recent cross-taxa review that considered the characteristics associated with two stages of invasion, the establishment stage, where the introduced species has a self-sustaining population, and the invasive stage, where the introduced species has spread and become abundant (sensu Kolar & Lodge 2001), the only characteristic consistently associated with both establishment success and invasive propensity was a climate/habitat match between native and naturalized ranges (Hayes & Barry 2008). The study only found strong evidence that fecundity-related traits were associated with success for plants at the invasive stage. Thus, the issue of whether or not traits associated with invasiveness may also influence propensity to amplify disease requires further investigation.
In addition to species-specific competency, community capacity depends on the abundance or, for frequency-dependent transmission, relative abundance of the different host species (Brooks & Zhang 2010). Provided the introduced species shows some competency and, crucially, does not reduce the abundance of key native hosts through interspecific interactions, an invasion will tend to amplify parasites with density-dependent transmission, especially if the introduced host reaches high densities. For frequency-dependent transmission, the relative competency and relative abundance of the introduced species compared to the native species will also play a role. Spillback of disease from introduced hosts appears relatively common, occurring in diverse parasite systems from viruses to helminths (Kelly et al. 2009b; Poulin et al. 2011). However, dilution effects have also been observed (Telfer et al. 2005; Kopp & Jokela 2007; Thieltges et al. 2008; Kelly et al. 2009a), despite the likelihood that attention and research effort is preferentially drawn to increases in disease prevalence. Interestingly, apart from one exception, the parasites demonstrating a dilution effect appear completely incapable of infecting the introduced species. The exception was Telogaster opisthorchis, a trematode that infected introduced brown trout in New Zealand but at much lower prevalence than native fish species (Kelly et al. 2009a). Thus, dilution effects may generally only be detectable where introduced species have close to zero competency.
It has been suggested that introduced species are more likely to be competent hosts, and therefore act as amplifying species, if there are relatively short phylogenetic distances between the introduced species and native host fauna, and the native parasite fauna and the parasite fauna of the invader in its area of origin (Poulin et al. 2011). Whilst this appears intuitive and deserves further investigation, our work on Bartonella dynamics in Ireland suggests local adaptation of parasite and/or host populations may lead to unexpected outcomes. Despite the two Bartonella species detected in Ireland being highly prevalent in both bank voles and wood mice in the United Kingdom, in Ireland they were found exclusively in native wood mice (Telfer et al. 2005). Analyses of mitochondrial DNA are compatible with the hypothesis that Irish bank voles were introduced from Germany (Stuart et al. 2007), and so it is not clear whether this unexpected host specificity is driven by differences in parasite or host.
Thus, whilst there is some encouragement that general predictions may be made regarding parasite introductions and impacts of invasions on native parasite dynamics, the complexity of the factors involved means that often specific predictions will rely on a detailed understanding of the system. Further elucidation of mechanisms underlying observed patterns will be helpful, with consideration given to the full potential range of community interactions. Whilst competence and the abundance of introduced hosts are important, the ultimate effect will depend on the summed direct and indirect impacts of the invasion, including effects on vector and native host abundance and the prevalence of competing parasite species (Box ).
Conclusions and further work
Both the probability of spread for an exotic parasite and impacts of introduced species on native parasite dynamics depend on rates of encounter, transmission, mortality and recovery. Managing disease outbreaks caused by both exotic and endemic parasites requires a fuller understanding of how diverse interspecific interactions within communities, spatial structure and seasonality influence these rates and therefore parasite prevalence. For multihost parasites, control initiatives aimed at reducing disease prevalence in a target population often need to consider whether management should address the target population and/or a reservoir population [defined as populations in which the pathogen can be permanently maintained and from which infection is transmitted to the target population (Haydon et al. 2002)]. Combining empirical data on spatial and seasonal transmission patterns between target and reservoir populations with appropriate disease models can inform the development of effective strategies. In the United States, the rust fungus Phakopsora pachyrhizi infects many legume species, including soybeans (Glycine max) and kudzu (Pueraria montana var. lobata), an invasive host species (Christiano & Scherm 2007). Although kudzu is critical for overwintering survival of the fungus, parameterizing a spatially implicit model with empirical data revealed that localized control of kudzu is unlikely to prove effective, as even low densities of infected kudzu surviving in landscape patches separate from soybean crops, is sufficient to cause epidemics (Fabiszewski, Umbanhowar & Mitchell 2010). In addition to informing control programmes, additional theoretical and empirical studies will facilitate a wider appreciation of the mechanisms by which community structure and parasite dynamics interact, with relevance for ecosystems experiencing invasions or biodiversity loss. Below, we suggest some avenues for research that may be particularly fruitful.
As highlighted above, changes to community structure can result in different effects on transmission and parasite prevalence depending on whether encounter rates exhibit frequency or density dependence. However, empirical evidence indicates that more complex transmission functions may be common, including functions intermediate between the two and functions that vary seasonally (Rachowicz & Briggs 2007; Smith et al. 2009; Tompkins et al. 2011). Moreover, most models assume that interspecific transmission has the same functional form as intraspecific transmission, an assumption that has already been questioned, especially for diseases without vectors or free-living stages (Rudolf & Antonovics 2005). More realistic forms of transmission incorporated into multihost models could have important implications for predicted dynamics and the circumstances under which community change is expected to result in dilution or amplification effects. Similarly, models that explore how changes to community interactions influence parasite dynamics through effects on host or vector abundance or host susceptibility would also be useful, for example models that incorporate competition between host or parasite species.
The mean capacity of the reservoir community, a metric that incorporates information on relative abundance and host competence, has recently proved helpful in understanding how community change may influence the dynamics of a vector-borne disease with frequency-dependent transmission (Brooks & Zhang 2010). Extensions of this approach could be developed for theoretical studies of parasites with other functional forms for transmission (including density-dependent) or for systems with interspecific competition between hosts. Similarly, the force of infection is a useful metric that has been used to examine the relative importance of different host species and variation in disease risk between communities in theoretical and empirical studies (Dobson 2004; Hamer et al. 2011).
Linking theoretical, empirical and experimental studies is essential to advance our understanding. One practical approach may be to exploit natural experiments established by the invasion process itself, examining disease dynamics in populations ahead of and behind the invasion front (e.g. Telfer et al. 2005). Analyses of disease dynamics from empirical studies should aim to quantify the relative importance of host density and diversity and, where relevant, vector density and diversity. In addition, the potential role of interspecific parasite interactions in explaining observed effects should be assessed. A combination of theoretical and empirical studies that elucidate the key processes influencing disease dynamics in multihost (and multivector) communities will inform predictions of how environmental change, including biodiversity loss and invasive species, may influence disease risk, as well as facilitating appropriately targeted management.
This review benefited from comments made by two anonymous reviewers. ST was supported by a Research Career Development Fellowship from the Wellcome Trust 081705/B/06/Z.