Volume 88, Issue 3 p. 416-426
RESEARCH ARTICLE
Free Access

Your infections are what you eat: How host ecology shapes the helminth parasite communities of lizards

Tommy L. F. Leung

Corresponding Author

Tommy L. F. Leung

School of Environmental & Rural Sciences, University of New England, Armidale, NSW, Australia

Correspondence

Tommy L. F. Leung

Email: [email protected]

Search for more papers by this author
Janet Koprivnikar

Janet Koprivnikar

Department of Chemistry and Biology, Ryerson University, Toronto, ON, Canada

Search for more papers by this author
First published: 03 December 2018
Citations: 14

Abstract

  1. Understanding how parasite communities are assembled, and the factors that influence their richness, can improve our knowledge of parasite–host interactions and help to predict the spread of infectious diseases. Previous comparative analyses have found significant influences of host ecology and life history, but focused on a few select host taxa.
  2. Host diet and habitat use play key roles in the acquisition of parasitic helminths as many are trophically transmitted, making these attributes potentially key indicators of infection risk. Given the paucity of comparative studies with non-piscine, non-avian or non-mammalian hosts, it is critical to examine the degree to which host ecology influences parasite communities in other host taxa in order to identify common drivers.
  3. We examined helminth diversity in over 350 species of lizards in relation to their body mass, ecology (diet and habitat use) and life history (clutch size, and ovo- or viviparity) using previously published data.
  4. Overall, lizard species with herbivorous diets harboured fewer types of helminths (especially larval stages), with similar results for traits that were ultimately strongly associated with diet (host mass and habitat use). Large hosts tended to be herbivores with few helminth types, whereas species utilizing arboreal habitats typically consumed some animal matter and hosted more helminths.
  5. Understanding how host ecology and life history are related to their parasite assemblages has significant implications for the risk of acquiring novel parasites. Our results indicate an overwhelming influence of host diet such that many helminths may be relatively easily acquired by hosts in new ranges, or through dietary shifts.

1 INTRODUCTION

One of the major questions in community ecology concerns which factors are most influential in determining the composition and diversity of different communities (Brown, 1995; Gaston & Blackburn, 2000; Rosenzweig, 1995). Ecological communities are formed as a result of various abiotic and biotic processes, such as temperature and primary productivity, which can influence the resulting pattern of species composition (Gaston & Blackburn, 2000; Hawkins et al., 2003; Willig, Kaufman, & Stevens, 2003). Consequently, free-living communities often assemble in non-random and somewhat predictable ways, and the same principle can be applied to assemblages of symbiotic organisms such as parasites (e.g., Guégan, Morand, & Poulin, 2005; Poulin, 2007).

In certain respects, parasite communities are comparable to assemblages found on islands, with each host species representing an isolated and insular habitat, and the composition of the parasite community resulting from a series of colonization, speciation and extinction events that are mediated by factors such as niche availability (Kuris, Blaustein, & Alio, 1980). The theory of island biogeography and its variants predict that various characteristics of the available habitat can determine the composition and diversity of its community (MacArthur & Wilson, 1967). While host species are not completely comparable to islands (Kuris et al., 1980), this general theory can provide a conceptual basis for investigating host characteristics which can influence the richness of their parasite communities (Poulin, 2007).

Understanding how parasite communities are assembled, and the factors that influence their richness, can improve our knowledge of parasite–host interactions and help to inform aspects of disease ecology (Rigaud, Perrot-Minnot, & Brown, 2010). For instance, most host organisms are infected by multiple parasite species at any given time, and the diversity of parasites within a given individual (i.e., the parasite infracommunity) can affect its ability to mount an effective defence (Bordes & Morand, 2009, 2011). At the host population and community levels, parasite richness (the component and compound parasite communities, respectively) can also impact infection and transmission success (Johnson & Hoverman, 2012; Johnson, Preston, Hoverman, & LaFonte, 2013). Additionally, parasites can also play a role in mediating the establishment of exotic species in their newly introduced range (e.g., Kelly, Paterson, Townsend, Poulin, & Tompkins, 2009; Prenter, MacNeil, Dick, & Dunn, 2004), and understanding the key drivers behind parasite assemblages can help us determine the factors that facilitate parasite exchanges between introduced and native species (Dunn et al., 2012).

Over the last few decades, many studies on parasite communities, especially endoparasitic worms (i.e., endohelminths), have focused on identifying the factors that shape their diversity and composition, including host phylogeny and ecology (Kamiya, O'Dwyer, Nakagawa, & Poulin, 2014; Korallo et al., 2007; Poulin, 1997; Poulin & Morand, 2004). It may be expected that a host's diet plays a key role in shaping its endohelminth community because many are acquired through ingestion of infective larval stages, either incidentally from the environment or from infected prey items which serve as intermediate or paratenic hosts, particularly considering that trophic transmission is required for many endohelminths (Lafferty, 1999; Poulin & Morand, 2004). While some studies have found host diet to influence helminth diversity (Choudhury & Dick, 2000; Gutiérrez, Rakhimberdiev, Piersma, & Thieltges, 2017; Leung & Koprivnikar, 2016; Lindenfors et al., 2007), that has not always been found to be the case (Gregory, Keymer, & Harvey, 1991; Poulin, 1995; Sasal, Morand, & Guégan, 1997).

Another host trait which may influence a host's helminth community is its body mass. It is expected that hosts with greater body mass may have higher parasite diversity as they would have more microhabitats available for colonization, and could also support a higher abundance of parasites (Kamiya et al., 2014; Kuris et al., 1980; Poulin & Morand, 2004). Body mass has been found to positively correlate with parasite richness across different vertebrate groups (Gregory, Keymer, & Harvey, 1996; Poulin, 1995) including mammals (e.g., Lindenfors et al., 2007, but see Morand & Poulin, 1998), birds (Gregory et al., 1991), fish (Guégan, Lambert, Lévêque, Combes, & Euzet, 1992) and amphibians (Campião, de Aquino Ribas, Morais, da Silva, & Tavares, 2015). But host body mass itself may be inherently related to host diet given that larger hosts also ingest food at a higher rate, and are thus exposed to more trophically transmitted helminths (Poulin & Morand, 2004).

Host habitat type has also been found to be an important factor in determining the richness and composition of parasite communities for many vertebrates including amphibians, fish, and reptiles (e.g., Aho, 1990; Biserkov & Kostadinova, 1998; Bolek & Coggins, 2003; Brito et al., 2014; Poulin, Blanar, Thieltges, & Marcogliese, 2011; Sharpilo, Biserkov, Kostadinova, Behnke, & Kuzmin, 2001). Generally, vertebrate hosts that use aquatic habitats have greater parasite diversity than their terrestrial counterparts (Bush, Aho, & Kennedy, 1990), possibly because many parasites have larval stages which are aquatic and/or use aquatic organisms as hosts to complete their life cycles. For example, Gregory et al. (1991) found that bird species utilizing aquatic habitats have greater trematode (flukes) richness. Similarly, a recent study found that bird species across three orders which use aquatic habitat have greater nematode richness than species that do not (Leung & Koprivnikar, 2016). Comparable results have also been found for amphibians, with aquatic and semi-aquatic species harbouring helminth communities that are significantly different to those in terrestrial species (Bolek & Coggins, 2003; McAlpine, 1997). However, host diet can also be linked to habitat use because habitat can determine the availability of prey which can serve as intermediate hosts for trophically transmitted helminths.

In addition to host diet, body mass and habitat use, their life histories could also influence parasite assemblages. Notably, investment into anti-parasite defences comes at a cost to life-history traits such as reproduction (Lochmiller & Deerenberg, 2000; Sheldon & Verhulst, 1996), and vice versa. Greater reproductive investment can be associated with higher parasite richness and load (e.g., Fargallo & Merino, 2004; Gregory et al., 1991), indicating a trade-off, or hosts capable of producing larger clutches can also support more parasites without compromising survival. Furthermore, it has been documented that mounting an immune response during pregnancy can result in reduced litter mass in lizards (Meylan, Richard, Bauer, Haussy, & Miles, 2012), and parasitism has also been associated with pregnancy failure in viviparous lizards (Hare, Hare, & Cree, 2010). Thus, there may be an association between parasite load or richness with host clutch size and reproductive mode.

Comparative analyses of parasite communities have been useful in determining host traits which can influence parasite richness in different host species (Kamiya et al., 2014). However, there is a clear taxonomic bias in the host taxa which are the focus of such studies, with a concentration on fish, birds and mammals (e.g., Gregory et al., 1991; Guégan et al., 1992; Lindenfors et al., 2007; Poulin, 1995). In contrast, with the exception of the study by Aho (1990), there have been few comparative analyses on parasite communities found in herptiles (amphibians and reptiles), and as far as we are aware, large-scale comparative analyses of parasite communities have yet to be done for reptiles. The absence of reptile hosts in such comparative analyses means that parasite community patterns for a wide segment of vertebrate biodiversity have been neglected, leaving us with an incomplete picture of broad host characteristics which can potentially influence parasite richness across host taxa.

The parasite communities of reptiles are generally considered to be depauperate for various reasons including their relatively low vagility, food ingestion rate and metabolism in comparison with birds and mammals (Aho, 1990). Such traits mean fewer opportunities for parasite infection and transmission compared with birds and mammals; however, while there is considerable range in diet, body mass, habitat and life history among extant reptiles, little is known about how such variations shape their parasite communities. For this study, we compare the endohelminth communities of lizards (reptiles within the order Squamata excluding Serpentes). Lizards are highly diverse; with over 6,000 living species they constitute about 58% of extant reptile diversity (Uetz, Freed, & Hošek, 2017) and are found in a wide variety of ecosystems (Pianka & Vitt, 2003). Like many other reptiles, lizards are particularly susceptible to changes in climatic conditions, and a large percentage of lizards are at risk of extirpation and extinction due to climate change and other anthropogenic activities (Gibbon et al., 2000; Sinervo et al., 2010) even though many are considered to be data deficient in regards to their ecology and conservation status (Bland & Böhm, 2016).

As parasites can have significant influences on the conservation of threatened and endangered species (e.g., Dunham, Peper, Baxter, & Kendall, 2014; Godfrey, Moore, Nelson, & Bull, 2010; Thompson, Lymbery, & Smith, 2010), it is important to consider the role that parasite diversity can play in exacerbating or mitigating the impact of disease outbreaks in host populations. Additionally, there is a growing recognition regarding the significance of taking co-extinctions, such as that of parasites, into account when prioritizing conservation targets (Gómez & Nichols, 2013; Spencer & Zuk, 2016; Strona, 2015). Many parasites are specialists that only infect a single host species and are found only in a particular region. For example, it has been documented that almost one third of the helminth parasites harboured by the reptiles of Mexico are endemic to the region (Pérez-Ponce de León, García-Prieto, & Razo-Mendivil, 2002), and comparable levels of endemism are likely found in reptile hosts in other regions. Therefore, in addition to being an important part of free-living biodiversity, the vulnerability of lizards also presents a risk of co-extinction for their parasites, many of which have complex life cycles and may play integral roles in their ecosystems (Hatcher & Dunn, 2011).

Here we examine the influences of host ecology and life history, specifically body mass, diet, habitat use, clutch size, and ovo- or viviparity, in shaping the richness and composition of endoparasitic helminth communities in lizards. Beyond insight into the drivers of parasite community richness in this ecologically important and evolutionary diverse group of reptiles, the study of more host taxa will aid in elucidating whether there are consistent and broad influences on parasite assemblages that have important predictive and management applications.

2 MATERIALS AND METHODS

2.1 Data collection

To examine the influences of host body mass, diet, habitat use and select life-history parameters on patterns of endohelminth infection in lizards, we opted to focus on measures of parasite species as the relative consistency in their community composition is more representative of long-term selection pressure by parasites that may affect species-level host traits (see Supporting Information). In addition, recent studies with birds have demonstrated the importance of host diet and habitat use on the diversity of their helminth fauna (Gutiérrez et al., 2017; Leung & Koprivnikar, 2016). Consequently, we gathered data regarding lizard hosts and their endohelminth parasites using the search string “squamat* AND parasit*” in the ISI Web of Science (WoS) search engine with no date restrictions. This approach generated results for all hosts in the order Squamata; however, here we only focus on lizards because hosts in the order Serpentes represent unique lineages in which certain host traits are confounded with taxonomy (e.g., all snakes are carnivorous). Although this does not guarantee that all relevant studies of lizard parasite communities would necessarily be found, the resulting compilation should yield a large, representative and unbiased sample of studies dating from those published several decades ago to the present. Based on taxonomic revisions to the order Squamata, we used updated host species names from Feldman, Sabath, Pyron, Mayrose, and Meiri (2016) to conduct a supplemental WoS search with the same query string to ensure that we captured as many parasite records as possible where there were changes.

We only included studies which provided clear results for the following endohelminth groups in lizard host species: trematodes (flukes), cestodes (tapeworms), nematodes (roundworms), acanthocephalans (thorny-headed worms) and pentastomids (tongue-worms). These endohelminths include species that have either simple or complex (multi-host) life cycles requiring trophic transmission, as well as those that are more common in aquatic or terrestrial habitats. Following the approach of similar comparative analyses (Koprivnikar & Leung, 2015; Leung & Koprivnikar, 2016), we counted the number of unique endohelminth species for each host after considering taxonomic issues and those related to study bias (see Supporting Information). The body mass for each retained lizard species was additionally obtained from Feldman et al. (2016) given that larger hosts may have more parasites (Gregory et al., 1996; Morand & Poulin, 1998; Kamiya et al., 2014, but see also Poulin, 1997), but host size may also be confounded with diet, habitat, and life-history traits.

We collected information regarding diet, habitat use and life history (clutch size and ovo- or viviparity) for each retained host species from a variety of sources (see Dryad data). We designated each lizard species as an herbivore, omnivore or carnivore based on recent meta-analyses of squamate evolution that included such diet categorization (e.g., Meiri et al., 2013), but also supplemented this with individual reports of consumed organisms for those hosts lacking such information. For the latter, we used the following categorization: herbivore = no animal matter, omnivore = some animal matter, carnivore = no plant matter. We similarly categorized habitat use for each lizard species based on recent meta-analyses (e.g., Mesquita et al., 2015) or individual descriptions as one of: terrestrial (ground-dwelling), arboreal (including semi-arboreal), fossorial (burrowing) or aquatic (including semi-aquatic or riparian). We also gathered information regarding clutch size and ovo- or viviparity for each host species from a variety of sources (e.g., Pyron & Burbrink, 2014). Lastly, we assigned a categorical code for the family to which each host species belonged that allowed us to account for their relatedness in our analyses (357 species across 30 families).

2.2 Statistical analyses

To examine the relative influences of select lizard host traits on their helminth parasite fauna, we used a series of generalized linear mixed models (GLMM) that included the following as fixed effects: host body mass, diet (three categories: herbivore, omnivore or carnivore), habitat (four categories: terrestrial, arboreal, fossorial or aquatic), clutch size, and ovo- or viviparity (categorical). Separate models were run with different dependent variables to explore various aspects of helminth diversity: endohelminth richness, number of endohelminth types, each major endohelminth type (excluding pentastomids owing to their rarity) and larval endohelminth richness (see Supporting Information for details). We included host study effort (i.e., WoS hits) as a fixed effect given its expected influence and accounted for host relatedness by using the categorical code assigned to each family as a random effect. Although our complete dataset included 357 host species, information for all five of the host traits described above was available for only 187 of these. We ran a series of models, dropping those containing non-significant main effects, resulting in the best single model containing only significant terms. Post hoc (LSD) tests were used with the categorical fixed effects. To examine whether there was collinearity between habitat and diet as fixed effects, we carried out diagnostics with an analysis of helminth richness using our full dataset (N = 266 representing host species with data for both traits).

Given that information on the five host attributes was complete for only 187 of our host species, we conducted another set of GLMMs for each fixed effect separately using the same dependent variables as above in order maximize our power to detect any influences of these host attributes on their parasite fauna. To correct for multiple hypothesis testing, we applied the Benjamini–Hochberg procedure, with an accepted false discovery rate of 0.25, for this set of analyses in order to assess the significance of our fixed effects. This was complemented by a chi-square analysis to determine whether lizards differed in their frequency of diet type based on their habitat association, and another to see whether the number of host species among the 12 possible factorial combinations of our diet and habitat categories differed significantly. To determine whether the random effect of family was influential on the measures of host parasitism described above, we carried out a multiple analysis of variance (MANOVA) using a general linear model (GLM) procedure, including WoS hits as a covariate. We also conducted two separate GLMMs to examine whether the occurrence of host species among our diet and habitat categories, respectively, was influenced by their family, using a multinomial probability distribution and generalized logit link function, and including body mass as a fixed effect. Lastly, another set of GLMMs were performed with host mass (log10-transformed) as the dependent variable to determine whether this was affected by host diet and habitat given the likely influence of mass on host parasitism (Gregory et al., 1996; Morand & Poulin, 1998). All analyses were performed using SPSS 24.0.

3 RESULTS

Collinearity diagnostics found tolerance (0.982) and variation inflation factor (1.018) values for habitat and diet that were not problematic. There was no significant difference in frequency of lizard species occurrence among cells of a factorial diet by habitat grid (Pearson Χ2 = 10.840, df = 6, p = 0.093); however, certain combinations dominated (Supporting Information Table S1). For instance, the majority of host species (148) were carnivorous and terrestrial, and there were no herbivorous arboreal or fossorial species (Figure S1).

The GLMMs utilizing data for the 187 lizard host species with complete information for all five host attributes (body mass, diet, habitat, clutch size, and ovo- or viviparity) only retained host study effort in the models for endohelminth richness and number of endohelminth types (richness: F1,188 = 13.230, p < 0.001; types: F1,188 = 6.852, p < 0.001; see Tables in Supporting Information for full model outputs). However, the overall model for number of larval endohelminth species was significant (F6,183 = 5.317, p < 0.001), retaining habitat type (F3,183 = 2.849, p = 0.039), diet category (F2,183 = 4.597, p = 0.011) and WoS hits as fixed effects (F1,183 = 14.970, p < 0.001; Figure 1). Arboreal lizards harboured significantly more species of larval endohelminth than terrestrial species (p = 0.006), with a similar trend compared to aquatic lizards (p = 0.078). Herbivorous hosts had significantly fewer larval helminth species compared to omnivores (p = 0.005) and carnivores (p = 0.004). When considering the richness of each endohelminth type separately, WoS hits was the only fixed effect retained for trematode (F1,188 = 29.841, p < 0.001) and nematode richness (F1,188 = 9.854, p = 0.002), with no significant overall model achieved for cestode richness (F1,188 = 1.237, p = 0.267 when only WoS hits retained). However, acanthocephalan richness showed a negative relationship with host clutch size (F1,187 = 6.090, p = 0.014), with host study effort (F1,187 = 4.265, p = 0.040) also retained in a significant overall model (F2,187 = 4.559, p = 0.012).

Details are in the caption following the image
Effects of host (a) diet and (b) habitat use on larval helminth species richness for 187 lizard host species

The set of GLMMs that analysed each of the five host traits separately in order to maximize the number of host species included also resulted in significant models for some, but not all, attributes of lizard endohelminth fauna. Similar to the smaller dataset, host diet was not associated with endohelminth richness (F2,267 = 1.376, p = 0.254), but had a strong effect on number of larval helminth species (F2,267 = 4.954, p = 0.008), with carnivores and omnivores hosting significantly more species of larval helminth compared to herbivores (p = 0.003 and p = 0.005, respectively; Figure 2). In contrast to the smaller dataset, lizard diet was significantly associated with the total number of helminth types (F2,267 = 3.341, p = 0.037), and again herbivores exhibited reduced diversity relative to omnivores and carnivores (p = 0.012 and p = 0.027, respectively). Diet did not influence any other aspects of endohelminth fauna (see Table S9 for full model outputs).

Details are in the caption following the image
Effects of host diet on: (a) larval endohelminth species richness; and, (b) number of endohelminth types (trematode, cestode, acanthocephalan, nematode and pentastomid) for 202 lizard host species

Lizard habitat use was significantly associated with trematode richness in the larger dataset (F3,345 = 3.204, p = 0.023), with aquatic hosts harbouring the most species compared to those categorized as arboreal, terrestrial or fossorial (all p < 0.05). Clutch size was negatively associated with acanthocephalan species richness in the larger dataset as well (F1,201 = 6.352, p = 0.013) and showed the same relationship with larval helminth richness (F1,201 = 4.314, p = 0.039). Viviparity was not significantly associated with any of our helminth measures (all p > 0.188). Owing to the Benjamini–Hochberg procedure, host mass showed a significant negative association with larval helminth richness (F1,352 = 3.818, p = 0.052). Unsurprisingly, host study effort (WoS hits) showed a strong positive association with all of our parasite measures (all p < 0.034).

Host family had a significant overall effect on our parasite measures (Wilks’ λ = 0.251, p = 0.001), driven by a significant difference among families in cestode richness (F25,163 = 2.029, p = 0.005), with marginally insignificant differences in the number of trematode (F25,163 = 1.489, p = 0.075) and acanthocephalan species (F25,163 = 1.489, p = 0.074). However, the occurrence of lizard species in the different diet and habitat categories was not influenced by their family (diet: F56,211 = 0.491, p = 0.999; habitat: F56,211 = 0.320, p = 0.999). Our last set of GLMMs revealed a strong effect of diet type on host mass (F2,266 = 31.744, p < 0.001; Figure 3) because herbivorous lizards were significantly larger than either omnivores or carnivores (p < 0.001 for both contrasts). Finally, there was a trend for the frequency of lizard host diet type to differ among the three habitat categories (X2 = 11.979, df = 6, p = 0.062) because 86% of the arboreal species were classified as carnivores (none as herbivores), whereas this was 76% for terrestrial species (4% herbivores, and 20% omnivores).

Details are in the caption following the image
Differences in body mass among different diet categories for 266 lizard species

4 DISCUSSION

This study is the first large-scale comparative analysis on the parasite communities of any reptile group since the study by Aho (1990) almost three decades ago that considered 115 host species compiled from across 16 families of reptiles and 14 families of amphibians. Using the comparative analysis approach which has been commonly used for birds, mammals and fish, we were able to discern some key ecological factors associated with the endohelminth communities of over 350 species of lizard across 30 families. While previous studies have shown that the helminth communities of vertebrate hosts are influenced by a wide variety of different factors (Poulin & Morand, 2004), we found diet to be the host trait most strongly associated with the range of helminth types found in lizards, with the other significant host traits likely showing some association with host diet as well. This association between diet and composition of helminth communities in lizards has been noted in previous studies of specific host species (e.g., Brito et al., 2014; Carretero, Jorge, Llorente, & Roca, 2014; Carretero et al., 2006; Martin & Roca, 2004; Roca, Carretero, Llorente, Montori, & Martin, 2005), and our comparative analysis demonstrates its broad influence.

Compared with omnivorous and carnivorous species, we found that herbivorous lizards have fewer helminth types, that is, lower parasite richness and less diverse communities. This parallels results found by Leung and Koprivnikar (2016) in birds where herbivorous hosts were shown to have less diverse nematode fauna than omnivorous and carnivorous birds, as well as results from a study on helminths in 12 Brazilian lizard species which found hosts that consume a greater variety of prey have richer parasite communities (Brito et al., 2014). Furthermore, we found that host families containing carnivorous species which largely consume arthropods also harboured the most number of cestode species. Since cestodes that infect lizards are known to use arthropods as intermediate hosts for trophic transmission (Biserkov & Kostadinova, 1997; Conn, 1985; Hickman, 1963), it would be expected that they are more common in predominantly insectivorous host families.

Another notable finding is the absence of larval helminths in herbivorous hosts from our dataset. As mentioned above in relation to cestodes, endohelminths that infect lizards as larval stages are often trophically transmitted and acquired through ingestion of infected arthropods (Cabrera-Guzmán & Garrido-Olvera, 2014; Goldberg, Bursey, & Holshuh, 1994; Jones, 2010), and thus, the absence of such food items in the diet of herbivorous lizards would exclude them from acquiring most helminths with complex life cycles (Carretero et al., 2006; Roca et al., 2005). Additionally, the helminth community of a host species can be influenced by its trophic level and the types of predator–prey interactions in which it is engaged. Network analysis suggests that species which serve as intermediate hosts for larval parasites tend to have a variety of different predators (Chen et al., 2008). In addition, while some studies have found that the diversity of adult parasites harboured by fish was associated with host diet, the proportion of larval helminths present in those communities was correlated with the diversity of the host's predators and/or vulnerability to predation (Locke, Marcogliese, & Valtonen, 2014; Muñoz, Grutter, & Cribb, 2006; Poulin & Leung, 2011). Consequently, host species which are preyed upon by few predators may be infected by fewer or no larval helminths as these would effectively represent “dead ends” for the parasite. Similarly, Carretero et al. (2014) suggested that the absence of larval cestodes in a population of Gallotia atlantica (Atlantic lizard) may be due to its low predation pressure. As we found that herbivorous host species in our dataset also tend to have greater body mass, this would make them less susceptible to predation or have fewer predators, which may additionally play a role in explaining the absence of larval helminths.

In contrast to our expectations, we did not find any association between lizard host mass and helminth diversity. Previous studies have found body mass to be strongly associated with parasite richness (Campião et al., 2015; Gregory et al., 1991; Lindenfors et al., 2007), but at the same time, others have found that the influence of body mass on parasite richness becomes non-significant once host phylogeny has been taken into consideration (Poulin, 1995, 1997), and thus, body mass should not be considered as a universal predictor (Poulin, 2007).

It is considered that for energetic reasons, herbivory in lizards favours the evolution of larger body size, as well as specialized morphological and physiological adaptations associated with having a mostly plant-based diet (Cooper & Vitt, 2002; Dearing, 1993). This corresponds with our dataset, as host species with greater body mass were composed mostly by herbivorous species. As discussed above, herbivorous hosts are exposed to a narrower range of helminths than host species which include animals in their diet because herbivory puts an ecological filter on many trophically transmitted parasites which would be found in omnivorous and carnivorous lizards. Therefore, it is possible that the comparatively lower helminth diversity in herbivorous species offsets any potential effects of greater body mass in influencing the helminth communities of lizards.

Our analysis also found that larval helminths were most common in arboreal lizard species. Previous studies have demonstrated that differential use of microhabitat types can influence endoparasite infections in lizards (Araujo Filho et al., 2017; Brito et al., 2014), and that the association between habitat type and the proportion of larval helminths found in lizard parasite communities (Biserkov & Kostadinova, 1998; Sharpilo et al., 2001) seems to be correlated with the presence of the definitive hosts for those larval helminths.

While an arboreal lifestyle as such would not seem to greatly facilitate parasite transmission, it is possible that it is not the use of arboreal habitats per se which is contributing to the higher proportion of larval helminths found in arboreal lizards, but rather, their diet. In our dataset, 86% of the lizard species which were categorized as arboreal were carnivorous, and the rest omnivorous, with no herbivorous representatives among the arboreal species. Leung and Koprivnikar (2016) suggested that the greater nematode richness in Accipitriformes birds which use aquatic habitats was actually due to feeding on water-associated prey which often serve as intermediate or paratenic hosts, and while Brito et al. (2014) found that microhabitat type was an important determinant of parasite community composition in lizards, host species that consume a greater variety of prey also have greater endoparasite richness.

Our analysis also found that lizard species which inhabit or use aquatic habitats have richer trematode fauna than those that live predominantly in terrestrial environments. This may be due to the nature of digenean trematode life cycles, which have free-living aquatic stages that require the presence of water, or at least moisture, for successful transmission (Galaktionov & Dobrovolskij, 2003). Our results corroborate with previous studies on the effects of water availability on the trematode communities of squamates. For example, rainfall is a major factor in determining trematode abundance in the garden lizard Calotes versicolor as it regulates the population of snails that serve as the first intermediate host and produce infectious stages (Madhavi, Nirmala, & Subbalakshmi, 1998). Biserkov and Kostadinova (1998) found that Lacerta viridis (European green lizard) living in habitats with permanent water bodies harboured trematodes, whereas those living in drier habitats did not. Furthermore, host species that use aquatic habitat may also be more likely to feed on animals which act as second intermediate hosts for trematodes and contain their infective larvae. For instance, Santos, Martínez-Freiría, Pleguezuelos, and Roca (2006) concluded that the abundances of trematodes in some species of Vipera that inhabit wetter habitats are due to the higher number of infected amphibians serving as intermediate hosts and prey. Habitat and diet are thus often correlated, and the former may dictate the latter.

Overall, our study found diet to be an influential factor in determining the breadth of endohelminth communities of lizards, and that other host traits which initially seemed to reflect key associations were likely associated with diet in some way. In hindsight, it should not be surprising that diet plays such an important role given that the parasites in our analyses (endohelminths) are mostly acquired by lizards through ingestion of infective stages either from the environment or in prey items. Perhaps because lizards have comparatively lower parasite diversity than other vertebrate host groups such as fish, birds and mammals, any patterns associated with host traits such as diet can be seen more clearly. Our results also highlight the importance of taking the life cycles of different parasite groups into account when determining which ecological factors can influence their presence and richness in particular hosts. For future studies, it will be insightful to examine whether different patterns of association between host traits and parasite diversity will be found for communities of ectoparasites and blood parasites considering that the former transmit mostly through host contact and proximity to conspecifics or microhabitat usage, while the latter are mostly transmitted by haematophagous arthropods. Such studies will shed further light on how host traits influence parasite infracommunities, and how these relate to parasites with different transmission modes and life cycles.

When contrasted with the results of comparative analyses conducted on other host groups, one observation which can be drawn from this study is that there appears to be no universal predictor for determining the diversity of vertebrate parasite communities—each broad taxon differs in the variables which shape its parasite communities, and it is important to untangle and incorporate different variables as their influences are not necessarily mutually exclusive. As our results demonstrate, in some cases the association between helminth and habitat might actually be attributable to diet, and the positive correlation between body mass and helminth richness (as found in previous studies) might be negated by the host's herbivorous or narrow diet. Taken together, the effects of host habitat on helminth community richness and composition may in fact be based on their diet, which has important implications. However, the dominance of particular habitat and diet combinations within our dataset, especially that by arboreal and terrestrial carnivores, may have influenced our findings, demonstrating the need to collect parasite community information from hosts occupying a wide range of ecological niches to discern broad patterns. While there were 30 lizard families in our dataset, and 13 of these spanned more than one diet category, this was dominated by six families which each contributed >30 species (accounting for 256 of the 355 host species). Furthermore, we had to use broad scale categories (herbivore, omnivore, carnivore) for host diet because detailed information on specific food items was often unavailable for most host species. In future, such data would be valuable for clarifying the relationship between helminth transmission and specific food items. In addition, spatial and intraspecific variations in host traits could have an influence on their parasite communities and should be explored in future studies.

Given the important role that diet plays in shaping helminth communities, dietary flexibility may be an important factor in the likelihood that an introduced species will acquire native parasites (e.g., Criscione & Font, 2001), and dietary shifts associated with climate change (Butt et al., 2015; Jamieson, Trowbridge, Raffa, & Lindroth, 2012) and other anthropogenic activities (Oro, Genovart, Tavecchia, Fowler, & Martínez-Abraín, 2013; Robb, McDonald, Chamberlain, & Bearhop, 2008) may result in cascading impact on helminth community richness and composition.

Our comparative analysis of endohelminth richness in over 350 species of lizard supports the idea that host–parasite communities may often assemble in predictable ways at a host species (suprapopulation) level based on both bottom-up and top-down ecological processes (Guégan et al., 2005; Poulin, 2007). Importantly, we found that host diet is the best predictor of helminth diversity and community composition either directly or indirectly, such that other host traits, including body mass and habitat use, may instead reflect the underlying driving force of diet. That carnivorous and omnivorous lizard species harbour the richest endohelminth fauna indicates these hosts have an essential role in maintaining parasite biodiversity, but also suggests that they can acquire new species of trophically transmitted parasites rather easily under the right circumstances (e.g., human-assisted geographic range expansions). If so, changes to host helminth communities could occur over relatively short period mediated by changes to the “encounter filter” (Combes, 2001) rather than being mediated by the “compatibility filter” (which is largely based on physiology) over longer periods of co-evolutionary interactions. Future studies incorporating more host and parasite taxa will be needed to elucidate the extent to which host diet is a universal predictor of helminth communities, as well as careful considerations of the extent to which other host traits actually correlate with diet.

ACKNOWLEDGEMENTS

We thank Sarah Herbert for assistance in collecting data. Funding was provided by a NSERC Discovery Grant to JK. Silhouettes of lizards in the Figures were obtained from PhyloPic (http://phylopic.org/) or the Public Domain. The authors declare no competing interests.

    AUTHORS’ CONTRIBUTIONS

    T.L.F.L. and J.K. conceived the ideas and designed methodology; T.L.F.L. and J.K. collected the data; J.K. analysed the data; T.L.F.L. and J.K. led the writing of the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

    DATA ACCESSIBILITY

    Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.qk361v8 (Leung & Koprivnikar, 2018).

      Journal list menu