SEARCH

SEARCH BY CITATION

Keywords:

  • Cardinium;
  • endosymbiosis;
  • reproductive parasite;
  • spider

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Spiders have recently emerged as important diversity hot spots for endosymbiotic bacteria, but the consequences of these symbiotic interactions are largely unknown. In this article, we examined the evolutionary history and effect of the intracellular bacterium Cardinium hertigii in the marbled cellar spider Holocnemus pluchei. We showed that Cardinium infection is primarily transmitted in spider populations maternally via egg cytoplasm, with 100% of the progeny from infected mothers being also infected. Examination of a co-inherited marker, mitochondrial DNA (mtDNA), revealed that Cardinium infection is associated with a wide diversity of mtDNA haplotypes, showing that the interaction between Cardinium and H. pluchei has a long-term evolutionary dimension and that horizontal transmission among individuals could also occur. Although Cardinium is well known to exert sex ratio distortion or cytoplasmic incompatibility in various arthropod hosts, we show, however, that Cardinium does not interact with the reproductive biology of H. pluchei. A field survey shows a clear geographical structuring of Cardinium infection, with a marked gradual variation of infection frequencies from ca. 0.80 to 0. We discuss different mechanistic and evolutionary explanations for these results as well as their consequences for spider phenotypes. Notably, we suggest that Cardinium can either behave as a neutral cytoplasmic element within H. pluchei or exhibit a context-dependent effect, depending on the environmental conditions.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Although historically symbiosis has received less attention than other interactions such as predation or competition, it is increasingly recognized as an important selective force of evolution (Moran et al., 2008; Werren et al., 2008). Many terrestrial arthropods host various types of bacterial endosymbionts that are vertically inherited from mother to progeny through the egg cytoplasm (Duron et al., 2008a; Hilgenboecker et al., 2008). Molecular and experimental approaches have yielded an immensely rich understanding of the biological roles of these endosymbionts: obligate mutualists provide nutrients, facultative mutualists confer protection against natural enemies or abiotic stress, and reproductive parasites manipulate the host reproductive systems (Haine, 2008; Moran et al., 2008; Werren et al., 2008; Saridaki & Bourtzis, 2010). Endosymbionts have thus profound effects on arthropod phenotypes and are now regarded as major drivers of arthropod ecology and evolution (Engelstadter & Hurst, 2009; Fellous et al., 2011; Jiggins & Hurst, 2011).

Perhaps, one of the most remarkable observations of these last few years has been that the spiders harbour one of the widest ranges of endosymbionts found in arthropods (Goodacre et al., 2006; Duron et al., 2008a,b). These symbionts belong to phylogenetically diverse lineages of bacteria, including Cardinium hertigii (Bacteroidetes), Wolbachia pipientis (alpha-proteobacteria), Rickettsia sp. (alpha-proteobacteria) and Spiroplasma ixodetis (Mollicutes). It is not yet well understood how these bacteria affect the spider’s biology. Outside the spider order (Araneae), these bacteria are generally known as ‘reproductive parasites’ in the sense that they increase their own transmission by manipulating the reproductive phenotype of their hosts (Werren et al., 2008; Engelstadter & Hurst, 2009). Because males represent a dead end for transmission, manipulations frequently involve biasing the sex ratio (SR) of infected females towards the production of daughters, via the induction of thelytokous parthenogenesis (production of all female progeny from unfertilized eggs), feminization of genetic males or male-killing. However, some reproductive parasites do not distort the host SR but induce cytoplasmic incompatibility (CI), a sperm-egg incompatibility between infected males and uninfected females. Each of these manipulations is of ecological and evolutionary importance to the particular host species that is infected, potentially inducing reproductive isolation or driving changes in sexuality (Werren et al., 2008; Engelstadter & Hurst, 2009; Saridaki & Bourtzis, 2010; Jiggins & Hurst, 2011).

The importance of bacterial endosymbionts in shaping spider biology has rarely been investigated: Wolbachia has been shown to be involved in SR distortion in two spider species (Gunnarsson et al., 2009; Vanthournout et al., 2011), and the role of Rickettsia in the dispersal capacity has been established in one species of ballooning spider (Goodacre et al., 2009). Current knowledge of other inherited bacteria, such as Cardinium, is weaker. This bacterium is exceptionally frequent in spiders, infecting ca. 20% of species (Duron et al., 2008b; Martin & Goodacre, 2009; Perlman et al., 2009), but was only discovered in the last decade (Zchori-Fein et al., 2004). Cardinium was then isolated from mites (Weeks et al., 2001), ticks (Kurtti et al., 1996), parasitic wasps (Zchori-Fein et al., 2001, 2004), biting midges (Nakamura et al., 2009), planthoppers (Weeks et al., 2003; Zchori-Fein & Perlman, 2004; Marzorati et al., 2006) and plant-parasitic nematodes (Noel & Atibalentja, 2006). The list of manipulations induced by Cardinium is impressive: feminization in mites (Weeks et al., 2001), thelytokous parthenogenesis in parasitic wasps (Zchori-Fein et al., 2001, 2004) and scale insects (Provencher et al., 2005), and CI in parasitic wasps (Hunter et al., 2003) and mites (Gotoh et al., 2007; Ros & Breeuwer, 2009). In addition, Cardinium spread is favoured by an adaptive manipulation of the oviposition behaviour of a parasitic wasp (Zchori-Fein et al., 2001; Kenyon & Hunter, 2007) and by enhancing the fecundity of a predatory mite (Weeks & Stouthamer, 2004). This strongly suggests that Cardinium may also drive spider biology.

In this study, we explore how Cardinium interacts with the marbled cellar spider Holocnemus pluchei (Aranea: Pholcidae). First, we examined whether Cardinium alters reproductive traits of H. pluchei, by testing for SR distortion, CI and fecundity effects. Second, we report on the geographical variation of Cardinium prevalence in natural populations of H. pluchei. Third, because Cardinium depends on maternal transmission for spreading within H. pluchei populations, we also report on the diversity of mitochondrial DNA (mtDNA) haplotypes. We analyse how mtDNA variation is partitioned by infection status, as a means to investigate the evolutionary history of the Cardinium infection. This is an especially useful tool as some intracellular endosymbionts have lower rates of molecular evolution than their host’s mtDNA (e.g. James & Ballard, 2000; Shoemaker et al., 2003; Dyer & Jaenike, 2004). We discuss the importance of possible mechanisms leading to our results as well as the potential adaptive significance of the presence of Cardinium for spider evolution.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Spider collection

Holocnemus pluchei specimens were field-collected from various sites in the native distribution range of the species, principally in the Mediterranean area (including France, Corsica, Spain, Crete, Israel and Jordan; 2009–2011). Specimens from California (2010) where H. pluchei has been recently introduced were also included in the analysis. The populations are presented in Table S1. Within each population, there were 2–159 individuals separately investigated for analysis. Spiders were fixed in 75% ethanol and stored at 4 °C until analysed.

Effect of Cardinium infection

The effect of Cardinium infection on H. pluchei reproduction was examined using field-caught males and females from Montpellier (population I in Table S1). The infection status was determined using polymerase chain reaction (PCR) as described below. Females holding egg clutches in their chelicerae were captured alive and individually housed in plastic cups (0.7 dm3). Egg clutches were checked daily until hatching; the eggs were then counted, and the hatching rate was recorded. The mother’s body size was estimated by measuring the tibia-patella length of leg 1 with a micrometre NIKON Digital Counter CM-6S. Spiderlings were reared individually to adulthood, and their sex was recorded. Spiderlings were reared as follows: after their first moult, each spiderling was individually transferred with a paintbrush to its own plastic cup and fed ad libitum twice a week with living prey, that is, Drosophila melanogaster flies and Nemobius sp. crickets. All specimens were kept at ca. 25 ± 2 °C with 12-h/12-h light/dark cycle. PCRs of a random sample of 20 flies and crickets indicated that they were not infected by Cardinium.

Screening and sequencing

Spider DNA was extracted using a cetyltrimethylammonium bromide (CTAB) protocol (Rogers & Bendich, 1988). DNA extraction from adult specimens was performed on abdominal tissue to reduce the risk of missing infection with reproductive parasites when they are present (false negatives); DNA extraction from juvenile specimens was performed on the entire body. The DNA quality was routinely tested by PCR amplification of a region of the mitochondrial cytochrome oxidase I (COI) gene. Two independent assays for Cardinium infection were performed by PCR amplification of two genes, the 16S rRNA gene and the DNA gyrase b (gyrb) gene, using specific primers. Additional PCRs were conducted on a subsample of individuals to detect infections by four other endosymbionts found in other spider species: Wolbachia, Rickettsia, Arsenophonus and Spiroplasma (Goodacre et al., 2006; Duron et al., 2008a). Independent assays for detecting the infection by each of these endosymbionts were performed using specific PCR amplification as described by Duron et al. (2008a). Gene and primer features are listed in Table S2.

Polymerase chain reactions were performed under the following conditions: initial denaturation at 93 °C for 3 min, 35 cycles of denaturation (93 °C, 30 s), annealing (50–54 °C, depending on primers, cf. Table S2, 30 s), extension (72 °C, 1 min) and a final extension at 72 °C for 5 min. The PCR products were electrophoresed in a 1.5% agarose gel. Direct sequencing of PCR products was performed on an ABI Prism 3130 sequencer using the BigDye Terminator Kit (Applied Biosystems; Foster City, CA, USA) after purification with the QIAquick gel extraction kit (Qiagen, Valencia, CA, USA). The chromatograms were manually inspected and cleaned with Chromas Lite (http://www.technelysium.com.au/chromas_lite.html), and sequence alignments were performed using mega (Kumar et al., 2004). The sequences are deposited in GenBank (accession numbers, JN202549JN202557).

Phylogenetic analyses

Phylogenetic relationships were evaluated (i) for Cardinium sequences of the 16S rRNA and gyrb genes and (ii) for H. pluchei mitochondrial COI sequences. For Cardinium, phylogenetic trees were constructed using our own data combined with other Cardinium sequences (groups A–C; Nakamura et al., 2009) found in GenBank. Before analysing the sequences, the evolutionary model most closely fitting the data was determined using hierarchical likelihood ratio tests and Akaike information criterion with the program modeltest version 3.7 (Posada & Crandall, 1998). For the 16S rRNA, gyrb genes and COI data sets, the best-fit approximation was the general time reversible model with invariant sites (GTR + I). To analyse phylogenetic relationships, maximum likelihood (ML) analyses were conducted using paup version 4.0 (Swofford, 2002). Model parameters were first estimated by ML on a neighbour-joining topology and then used in optimal tree searches, which consisted of heuristic searches with Tree-Bisection-Reconnection (TBR) branch swapping. Clade robustness was assessed by bootstrap analysis using 1000 replicates. The phylogenetic trees were visualized and edited in mega (Kumar et al., 2004).

Statistical analysis

All statistical analyses were carried out using the r statistical package (http://www.r-project.org/). To explain inter-population variation of Cardinium frequency, we built a statistical model by including distance of each population from a reference location as explanatory variable using a quasibinomial error distribution (lmer, lme4 package of r; http://lme4.r-forge.r-project.org) to correct for overdispersed errors. Differentiation among geographical mtDNA groups and heterogeneity between infected and uninfected hosts were tested using amova (ade4 package of r; http://pbil.univ-lyon1.fr/ADE-4).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Effect of Cardinium infection

We first examined the mechanisms possibly used by Cardinium to spread through H. pluchei populations, that is, its mode of transmission, its effect on fecundity and on reproductive phenotype. That the Cardinium infection could be maternally transmitted was checked by examining the presence of infection in neonates from infected mothers. Thirteen infected mothers and five uninfected mothers, all holding clutches of eggs, were collected in Montpellier (population I), and 10 neonates per clutch were randomly sampled for PCR screening using Cardinium-specific primers. We found that all neonates (n = 130) from infected mothers were themselves infected, whereas no Cardinium infection was found in neonates (n = 50) from uninfected mothers. The mean transmission rate can be thus estimated at 1 (95% confidence interval, 0.972–1), demonstrating a very efficient maternal transmission of Cardinium in H. pluchei.

Next, we measured fecundity and body size of 67 females from Montpellier but found no significant effect of infection. The mean number of eggs produced by infected and uninfected females was almost identical (mean ± SE of infected, 25.2 ± 1.3 eggs, n = 50 females; uninfected, 23.9 ± 1.5 eggs, n = 17 females; Wilcoxon two-tailed test, = 419, = 0.94). There were no significant differences either between patella-tibia length of infected and uninfected females (infected, 11.7  ± 0.2 mm, n = 50 females; uninfected, 11.6 ± 0.2 mm, n = 17 females; Wilcoxon two-tailed test, = 483, = 0.41). Because patella-tibia length is correlated with body size in H. pluchei (Skow & Jakob, 2003), this suggests that Cardinium does not affect the host body size. The number of eggs was highly correlated with mother’s patella-tibia length (F1,65 = 23.60, R2 = 0.25, = 8.10−6), showing that larger female body size translates into increased reproductive success, but this correlation was not influenced by the infection status (F1,65 = 0.25, = 0.62) (Fig. S1). Overall, these data show that infection by Cardinium does not alter body size and fecundity in H. pluchei.

We examined the Montpellier population for sex bias in infection prevalence, as an indication of potential SR distorting activity. However, Cardinium showed no evidence of sex-biased prevalence in adult hosts: 73 of 87 males (83.9%) and 53 of 72 females (74.6%) were infected showing that there was no evidence to reject the null hypothesis of equal prevalence in male and females (Fisher’s exact test, = 0.63). In addition, field females carrying egg clutches were assayed for two phenotypic indicators of the presence of SR distorter: low egg hatch rates (indicating male-killing) and female-biased progenic SR (indicating male-killing, feminization or thelytokous parthenogenesis). No hatch rate decrease was observed in the broods of 50 Cardinium-infected females compared with the broods of 17 uninfected females (infected, 0.99 ± 0.01, n = 1258 eggs; uninfected, 0.97 ± 0.01, n = 407 eggs; Wilcoxon two-tailed test, = 468; = 0.41). The progeny of six infected and of four uninfected mothers were also reared to adulthood to record their SR. The brood from infected females resulted in 64 : 60 male-to-female ratio, and the brood from uninfected females resulted in 40 : 45 male-to-female ratio, neither of which differ significantly from a 1 : 1 SR (binomial exact test, = 0.78 and 0.66, respectively). Hence, Cardinium-infected females laid eggs with a high hatch rate, and the broods produced had a 1 : 1 SR, with resulting males infected. Overall, this shows that Cardinium does not exert a SR distorting activity in H. pluchei, excluding the possibility of parthenogenesis, feminization or male-killing.

It is more difficult to establish the presence or absence of CI. Unfortunately, our prior assays failed to obtain mating in laboratory population cages, preventing a direct test for CI through crossing experiments. However, the presence of CI is not corroborated by the field data given above. In its simplest form, CI results in increased embryonic mortality (up to 100%) in crosses between infected males and uninfected females. Hence, the mean hatching rate obtained from uninfected females should be lower than the mean hatching rate obtained from infected females, but as stated above, we did not observe such a pattern in H. pluchei. In addition, assuming random mating, one should expect that ca. 80% (i.e. the observed frequency of infected males in Montpellier) of the clutches from uninfected females suffer CI mortality. However, all clutches from uninfected females (n = 17) exhibit high hatch rate values (from 0.88 to 1). It is thus likely that crosses between infected males and uninfected females are fully fertile, a result not expected from a CI phenotype. Generally speaking, Cardinium does not act as a reproductive manipulator in H. pluchei.

Distribution of Cardinium infection

To further investigate the population biology of Cardinium, we assayed for its presence in 510 H. pluchei individuals from 26 populations encompassing the native and introduced distribution area of this spider (Table S1 and Fig. 1). All spider DNA retained for analysis was positive for PCR amplification using the COI arthropod universal primers, indicating satisfactory DNA template quality. Of the 510 specimens examined, 16S rRNA and gyrb PCR assays indicated the occurrence of Cardinium infection in 210 specimens (41%). Additional PCR screening did not reveal other endosymbionts (i.e. Wolbachia, Rickettsia, Arsenophonus and S. ixodetis) than Cardinium in H. pluchei (68 individuals were examined with 1–3 randomly sampled individuals per population).

image

Figure 1.  Prevalence of Cardinium across the populations of Holocnemus pluchei. 1, France, Spain and Corsica; 2, Eastern Mediterranean basin; 3, California. The bars show the prevalence of infection (white: uninfected, black: infected) and the 95% confidence interval calculated from the binomial distribution. Letters represent study sites listed in Table S1. Details on sample size and prevalence are given in Table S1.

Download figure to PowerPoint

Of the 26 populations examined, Cardinium was found in 16 populations from France (15 infected populations of 17 screened) and Israel (one of one) (Table S1 and Fig. 1). Cardinium was not found to infect populations from Spain (one population), Corsica (1), Crete (1), Jordan (1) and California (4). Where Cardinium infection was observed in a host population, a medium prevalence of infection was observed in all cases, and infection was never observed to be at fixation (Cardinium prevalence ranged from 7% to 86% of individuals; Table S1 and Fig. 1). Infection frequency is not homogeneous between the 16 infected populations as significant variation occurs between them (Fisher’s exact test, > 10−6). It appears obvious that Cardinium infection is exceptionally frequent in the south of France, prompting us to examine how prevalence varies with geographical distance between populations.

We thus built a statistical model by including distance as an explanatory variable of Cardinium frequency. We arbitrarily used Montpellier as a reference location from where the distance of each population (populations A–R on the Fig. 1) was measured. The populations from Corsica (population S), Crete (T), Jordan (U), Israel (V) and California (W–Z) were not included in the analysis because of their geographical isolation. We found that the Cardinium frequency shows a gradual geographical variation: Cardinium is frequent around Montpellier (observed prevalence, 0.79; 95% confidence interval, 0.72–0.85; Table S1), but progressively declines from this area, and finally completely disappeared (t16 = −2.822; = 0.012) (Fig. 2). The reliability of this observation was confirmed using some reference locations other than Montpellier (Fig. S2). The variation of Cardinium frequency is not yet strictly regular as exemplified by Saint Gilles (population J), which is close to Montpellier but exhibited a relatively low Cardinium frequency (observed prevalence, 0.07; 95% confidence interval, 0.01–0.24; Table S1). However, no Cardinium-infected specimen was found in populations more than 270 km from Montpellier (except one infected specimen from Israel).

image

Figure 2.  Variation of Cardinium frequencies in the south of France and Spain. The figure encompasses 18 populations (from A to R on Fig. 1); distances are measured from Montpellier (population I).

Download figure to PowerPoint

Evolutionary history of Cardinium infection

To ascertain Cardinium DNA variation, we sequenced the 16S rRNA and gyrb gene fragments (954 and 483 bp, respectively) from 1 to 4 randomly sampled individuals per infected population. No 16S rRNA and gyrb sequence variation was found between infected individuals (n = 39), and the 16S rRNA sequences we obtained were also strictly identical to the one deposited in GenBank (accession number, EU333930). Phylogenetically, this strain belongs to the A group within the Cardinium genus and clusters with the strains found in other spider species as well as with strains known to induce reproductive manipulations in mites and insects (Fig. S3a,b). Overall, this suggests that one Cardinium strain – or an assemblage of very closely related strains – occurs widely in H. pluchei, indicating that all of these Cardinium infections are derived from a single ancestral infection.

We also examined diversity in the co-inherited marker, H. pluchei mtDNA, and the partitioning of this between individuals of different infection status (Fig. S4). Partial COI sequences were obtained from 116 H. pluchei specimens, spanning 22 populations, and comprising 39 Cardinium-infected specimens and 77 uninfected specimens. Across these 116 COI sequences, 16 polymorphic sites were observed within the 488 bp sequenced that allowed us to distinguish eight different mtDNA haplotypes. The identity between pairs of COI sequences is moderate to high, ranging from 97.5% to 99.8%. The Cardinium infection was found in association with a wide diversity of haplotypes within H. pluchei (Figs 3 and S4). Indeed, five haplotypes were retrieved from the 39 Cardinium-infected specimens examined. No haplotype is confined to Cardinium-infected specimens (haplotypes found in association with Cardinium are also observed in uninfected individuals), except haplotype 5, which was observed in only one specimen. Most commonly, Cardinium is found associated with haplotype 6 (31 specimens) but is also found with haplotypes 3 (three specimens), 5 (1), 7 (1) and 8 (3).

image

Figure 3.  Phylogeny and distribution of Cardinium infections among Holocnemus pluchei mitochondrial DNA (mtDNA) haplotypes. The left part of the figure shows the phylogeny of the 8 H. pluchei mtDNA haplotypes via maximum likelihood. Numbers on branches indicate percentage bootstrap support for major branches (1000 replicates; only bootstrap values of 60% or more are shown). GenBank accession numbers are given in parenthesis. The scale bar is in units of substitutions/site. The right part of the figure gives the distribution of Cardinium infection observed with each mtDNA haplotype (black: infected specimens, n = 77; white: uninfected specimens, n = 39).

Download figure to PowerPoint

We used amova to formally assess and test for the association between Cardinium infection status and mitochondrial sequence variation. In our data set, infection status explains only 11% of the mtDNA variance (= 0.02 based on 1000 permutations), although a large fraction of the variation is found within populations (54%, = 0.001) and between populations (35%, = 0.001). No significant association was found between mtDNA haplotypes and Cardinium infection (Fisher’s exact test, = 0.09). This is not surprising given most mtDNA haplotypes found in association with Cardinium are also observed in uninfected individuals. There is thus no clear preferential infection of some mtDNA haplotypes, meaning that Cardinium infection is randomly distributed among mitochondrial lineages.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We used molecular variation and epidemiological variables to infer the evolutionary history and effect of Cardinium in H. pluchei. The incidence of infection is highly variable across the distribution range of H. pluchei: 16 populations of 26 harboured Cardinium and, when present, its prevalence ranged from 7% to 86%. Cardinium infection was, however, never observed at fixation as shown by the presence of a substantial number of uninfected H. pluchei specimens in each population examined. Although Cardinium infection was observed in geographically distant populations (France and Israel, i.e. ca. 2000 km apart), no Cardinium DNA variation was found, suggesting that only one Cardinium strain (or an assemblage of closely related strains) widely infects H. pluchei specimens. Three main conclusions can be drawn that we will discuss in more detail. First, the interaction between Cardinium and H. pluchei has a long-term evolutionary dimension, suggesting that both partners could have evolved complex interactions. Second, although Cardinium is primary described as a reproductive parasite of arthropods, it is not maintained in spider populations by manipulating reproductive phenotype or by enhancing fecundity. Third, the gradual variation of Cardinium frequencies across H. pluchei populations rather suggests either a neutral effect on or an adaptive response to geographical variation in natural selection.

Cardinium is transmitted through a very efficient maternal transmission in H. pluchei meaning that mtDNA variation can be used to infer the evolutionarily history of the infection. In a wide diversity of arthropods, maternally inherited symbionts are known to alter mtDNA genetic diversity through the linkage disequilibrium resulting from their common mode of inheritance (see Hurst & Jiggins, 2005; for review). If the Cardinium invasion is recent in H. pluchei populations, only a single mtDNA haplotype should occur among infected individuals, and a variety of haplotypes should persist among uninfected individuals. Through cytoplasmic hitchhiking, this will increase the frequency of the Cardinium-associated mtDNA haplotype and thus decrease overall mtDNA diversity. Precisely, this pattern is found associated with some symbiont infections in arthropods (e.g. von der Schulenburg et al., 2002; Jiggins, 2003; Gueguen et al., 2010; Verne et al., 2012), but the pattern found in H. pluchei stands in striking contrast to this scenario. This pattern revealed that this symbiotic association is evolutionarily ancient. First, there is no association between infection status and mtDNA haplotypes, indicating that incomplete maternal transmission can occur. Second, there are a substantial number of polymorphic sites in the mtDNA of H. pluchei, signifying that the infection is sufficiently old for many mutations to have occurred since the initial Cardinium invasion. This suggests that Cardinium is not expanding through H. pluchei populations but rather persists at intermediate frequencies, probably for a long evolutionary time. An alternative, but nonexclusive, hypothesis to account for the lack of disequilibrium between Cardinium and mtDNA is based on horizontal transmission among H. pluchei individuals. Horizontal transmission of Cardinium infection, which is thought to be rare (Weeks et al., 2003; Zchori-Fein & Perlman, 2004), can explain the lack of disequilibrium: although horizontal transmission is probably not a primary factor driving symbiont spread, it could also explain the presence of Cardinium in diverse mtDNA lineages over a sufficient number of host generations.

The perfect maternal transmission of Cardinium implies that its transmission success should broadly depend on its effect on spider fitness. Surprisingly, Cardinium does not persist in H. pluchei populations using one of its known phenotypes. In insects and mites, Cardinium is well known to spread either through reproductive manipulation (Weeks et al., 2001; Zchori-Fein et al., 2001, 2004; Hunter et al., 2003; Provencher et al., 2005; Gotoh et al., 2007; Ros & Breeuwer, 2009) or enhancing fecundity (Weeks & Stouthamer, 2004). In our case, we can be definite that Cardinium infection in H. pluchei does not fall into this evolutionary scheme. Although this Cardinium strain is genetically close to known reproductive manipulator strains, it does not either distort the SR or cause CI: males and females were found equally infected, and no variation in hatching rate was observed according to the infection status. Cardinium has also no detectable effect on fecundity and body size and is not required to support host development and reproduction. Although a more subtle effect can, however, exist, we thus found here no direct evidence for a Cardinium effect in H. pluchei. To our knowledge, very few similar cases have been reported in the literature (Gotoh et al., 2007; White et al., 2009). In the parasitic wasp Encarsia inaron, Cardinium was not able to induce CI, progeny SR distortion or clear fitness benefit (White et al., 2009, 2010). It was therefore suggested that the maintenance of Cardinium within E. inaron could be directly attributable to co-infection with the CI-inducing Wolbachia also present in this host (White et al., 2009). Perfect co-transmission of the two symbionts should confer the same CI transmission advantage to Cardinium as its Wolbachia partner. However, we rule out this possibility in the case of H. pluchei: we did not find infection by Wolbachia– or any other bacterial symbionts – which might induce a high frequency of Cardinium.

What then are the evolutionary forces governing Cardinium frequencies? A first possibility is that Cardinium has no effect, behaving as a neutral cytoplasmic element whose frequency is determined by stochastic processes. Although the relatively small sample sizes do not allow any definitive conclusion, it is thus clear that spatially close populations differ in their infection frequencies. The gradual variation of infection frequencies across the south of France is thus compatible with the isolation by distance (IBD) effect: the tendency for infected individuals to migrate between neighbouring populations, which results in decreasing Cardinium frequencies with increasing geographical distance. Under the hypothesis that the infection is not recent in H. pluchei, it is also likely that present-day Cardinium does not have the same effect as the Cardinium that initially invaded the spider populations. One possible process is the evolution of Cardinium towards weaker levels of manipulation: reproductive parasites may initially spread to reach high prevalence but then evolve towards an asymptomatic status, leading to the loss of infection (Hurst & McVean, 1996). This system is then expected to proceed to clearance of infection, at which point the population will have gone full circle, a process called ‘reversible evolution’. Loss of manipulation implies a decrease in drive, which will be reflected in reduced prevalence, preceding infection loss. The progress towards the loss of infection is not expected to be rapid after the loss of reproductive manipulation: when the host population is dominated by a high frequency of a parasite without the capacity to modify host reproduction, parasite frequency can decline to extinction only if the infection is costly and/or imperfectly vertically transmitted. If not, a long-term coexistence of uninfected and infected individuals may then occur, and their frequencies would be driven by stochastic processes. This process thus fits well with the infection pattern observed within H. pluchei.

Alternatively, a second possibility is that Cardinium exhibits a context-dependent effect in H. pluchei. Facultative endosymbionts are generally believed to carry an intrinsic cost and, therefore, evolve compensatory adaptations to maintain their frequency (Oliver et al., 2008; Jaenike et al., 2010). Depending on environmental conditions, Cardinium could be a conditional mutualist, and its distribution would thus reflect an adaptive response to geographical variation in natural selection. Recent literature provides striking examples with bacterial symbionts allowing environment-dependent effects, such as thermal tolerance (Russell & Moran, 2006) or resistance against natural enemies (Oliver et al., 2003; Scarborough et al., 2005; Hedges et al. 2008; Teixeira et al., 2008; Jaenike et al., 2010). In flies and aphids, experiments using population cages showed that, in host populations under an environmental stress, the symbiont frequency can rapidly increase, but in populations without this stress, the symbiont is not favoured and declines (Oliver et al., 2008; Jaenike & Brekke, 2010). In this context, environmental stresses could directly influence Cardinium frequencies in H. pluchei populations. If true, the exact nature of this selective pressure remains to be characterized. Laboratory assay may, however, miss the critical selective agent (e.g. a combination of climatic variables), and it may be arduous to capture the factor maintaining infection. Another way to indirectly test this hypothesis would be to compare the Cardinium distribution with genetic differentiation at neutral H. pluchei markers: infection prevalence should reflect, at least partly, the effect of local selection, whereas the differentiation at H. pluchei loci will only be the result of neutral evolutionary processes, such as founder events or migrational patterns. The comparison of genetic differentiation levels at neutral H. pluchei loci with infection distribution could thus indicate if selection acts on Cardinium.

In conclusion, we would emphasize that symbiosis represents an important but insufficiently studied component of spider evolution. An exceptionally wide diversity of endosymbionts is present in spiders; what is unclear is what role do they play, calling into question the ecological processes that govern their distribution and abundance. Although reproductive manipulations have been reported for spider Wolbachia (Gunnarsson et al., 2009; Vanthournout et al., 2011), we showed here that the consequences of spider symbiosis are actually more diverse. The nature of the Cardinium effect remains now to be formally characterized to explain its wide distribution in spiders. Furthermore, the presence in the Cardinium genus of lineages with and without effects on host reproduction would be of great interest to study the evolutionary transitions that shape symbiotic associations.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We are very grateful to S. Fellous, A. Rivero and F. Rousset for commenting on an early version of this manuscript; C. Atyame, S. Charlat, E. Dumas, V. Durand, L. Duron, C. Fraisse, P. Labbé, A. Lecler, M. Lecler, Y. Lubin, S. Marcoux, I. Olivieri, M. Raymond, M. Thierry, A. Tsagkarakou and F. Zélé for collecting spiders; A. Berthomieu, P. Makoundou and S. Unal for their technical help; and two anonymous referees for their comments. All sequence and morphological data were obtained on the Environmental Genomic Platform and on the Morphometry Platform of the IFR Montpellier-Environnement-Biodiversité, respectively. This is the contribution 2012.041 of the Institut des Sciences de l’Evolution de Montpellier (UMR 5554 CNRS – Université Montpellier 2).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Table S1 Description of the Holocnemus pluchei samples: location, sample size and prevalence of Cardinium.

Table S2 Genes and primer features.

Figure S1 Numbers of eggs regressed against mother’s tibia-patella length.

Figure S2 Variation of Cardinium frequencies in the south of France and Spain.

Figure S3 Cardinium phylogeny constructed via maximum-likelihood using (a) 16S rRNA and (b) gyrb sequences.

Figure S4 Distribution of mtDNA haplotypes among populations, partitioned by Cardinium infection status.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
JEB_2535_sm_FigS1.tif97KSupporting info item
JEB_2535_sm_FigS2.tif75KSupporting info item
JEB_2535_sm_FigS3A.tif177KSupporting info item
JEB_2535_sm_FigS3B.tif177KSupporting info item
JEB_2535_sm_FigS4.tif1209KSupporting info item
JEB_2535_sm_TableS1-S2-FigS1-S4.doc1053KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.