CROSS-SPECIES INFECTION TRIALS REVEAL CRYPTIC PARASITE VARIETIES AND A PUTATIVE POLYMORPHISM SHARED AMONG HOST SPECIES

Authors


Abstract

A parasite's host range can have important consequences for ecological and evolutionary processes but can be difficult to infer. Successful infection depends on the outcome of multiple steps and only some steps of the infection process may be critical in determining a parasites host range. To test this hypothesis, we investigated the host range of the bacterium Pasteuria ramosa, a Daphnia parasite, and determined the parasites success in different stages of the infection process. Multiple genotypes of Daphnia pulex, Daphnia longispina and Daphnia magna were tested with four Pasteuria genotypes using infection trials and an assay that determines the ability of the parasite to attach to the hosts esophagus. We find that attachment is not specific to host species but is specific to host genotype. This may suggest that alleles on the locus controlling attachment are shared among different host species that diverged 100 million year. However, in our trials, Pasteuria was never able to reproduce in nonnative host species, suggesting that Pasteuria infecting different host species are different varieties, each with a narrow host range. Our approach highlights the explanatory power of dissecting the steps of the infection process and resolves potentially conflicting reports on parasite host ranges.

Understanding the host range of a parasite is critical, as it has important implications for numerous ecological and evolutionary phenomena. It can affect selection on both parasite and host traits (Woolhouse et al. 2001; Kirchner and Roy 2001; Regoes and Nowak 2000). Host range can also affect species interactions thereby altering community structure (e.g., parasite-mediated coexistence Janzen 1970; Connell 1971; for a review see Hatcher et al. 2006). Furthermore, it is an important predictor for the likelihood of host switches (Parker and Gilbert 2004) and therefore plays an important role in conservation and biocontrol.

Determining a parasite's host range is often difficult because apparent host ranges may overestimate actual host range. A common problem is that parasites with an apparently broad host range may consist of multiple cryptic species, each with a narrow host range (e.g., Bucheli et al. 2000; Mccoy et al. 2001; Steinauer et al. 2007). A further problem in determining host ranges is the method used to test for host susceptibility. To incorporate a host into a parasite's host range, a parasite must be successful in all steps of the infection process. Failure to overcome a single step in the defence cascade results in failure of infection, regardless of the ease with which the parasite is able to overcome the other steps in the infection process (Solter and Maddox 1998; Schmid-hempel and Ebert 2003). Thus, assessments made based on only part of the infection process may not represent the outcome of the entire process. For example, infecting animals by injecting parasites into the body cavity circumvents possible resistance mechanisms related to the first steps of host–parasite encounter, such as attachment or penetration into the host. Here we test the hypothesis that a single step in the infection process is responsible for determining the host range of a parasite.

The bacterial parasite Pasteuria ramosa and its host Daphnia magna have become a model system for antagonistic coevolution (Ebert 2005; Decaestecker et al. 2007; Luijckx et al. 2013), and recent advances allow for the assessment of parasite's success at different steps of the infection process (Duneau et al. 2011). The infection process begins with ingestion by the host of the parasite's environmental transmission stage (endospore) followed by activation of the endospore (shedding of the outer spore coating). If host and parasite genotypes are compatible, the activated parasite spores may then attach to the host esophagus, penetrate the esophagus wall, and proliferate within the host (Duneau et al. 2011). Eventually, the parasite kills the host, and spores of the parasite are released into the environment from the decaying host cadaver. By separately testing the spore activation step, the attachment step, and the proliferation steps, one can assess which of these steps is critical in determining Pasteuria's host range.

The host range of Pasteuria is currently unresolved. However, Pasteuria has been reported from field samples of D. magna, Daphnia longispina, Daphnia pulex (Stirnadel and Ebert 1997), Daphnia dentifera (Duffy et al. 2010), Daphnia curvirostris (Goren and Ben-Ami 2012), and even cladocerans in other genera (Green 1974; Sayre and Wergin 1977; Goren and Ben-Ami 2012). These Daphnia species belong to different subgenera and are unlikely to have exchanged genes in the last 100 million years (Colbourne and Hebert 1996). However, within natural communities Pasteuria has been found to infect only a single Daphnia species, even if other sympatric Daphnia species are present (Ebert et al. 2001; J. Andras unpubl. data). Furthermore, individual genotypes of Pasteuria are known to be extremely specific, infecting only a few genotypes of D. magna (Luijckx et al. 2011).

One possible explanation for the high specificity of infection within species and apparent low specificity between host species is that Pasteuria infecting different host species are separate genetic entities with similar morphology, each with a narrow host range (here defined as a variety; Ebert 2005). Alternatively, Pasteuria could be specific for some host genotypes within each host species but unspecific across species (as suggested by Duneau et al. 2011). Under this scenario, Daphnia species might share the polymorphism coding for attachment that was identified in D. magna, which is believed to be responsible for the extreme specificity observed between D. magna and Pasteuria (Luijckx et al. 2013).

Here, we assess the host range of Pasteuria by assessing both spore attachment to the host esophagus (which can be visualized with fluorescent-labeled spores as described by Duneau et al. 2011) and by exposing animals to spores in infection trials. Spore activation has previously been described to be unspecific (spores are always activated upon contact with any Daphnia; Duneau et al. 2011). Spore attachment governs host–parasite compatibility in one step of the infection process, whereas infection trials quantify the outcome of the entire infection process (activation, attachment, penetration, and proliferation). Thus, using both tests allows us to determine if failure of infection is due to failure of spore attachment or due to failure during the post-attachment steps. We sampled a natural Finnish meta-population, where the three host species, D. magna, D. pulex, and D. longispina occur in sympatry and tested for attachment using two naturally co-occurring Pasteuria isolated from D. magna and D. longispina. In addition, we use four lines of Pasteuria to survey attachment and infection patterns in these three Daphnia species from various locations in Europe. We will use the term “native” here to refer to combinations of Pasteuria with the Daphnia species it was sampled from.

Materials and Methods

SYSTEM DESCRIPTION

Pasteuria is an obligate bacterial parasite of Daphnia and has a wide geographic distribution in the northern hemisphere (Ebert 2005). Pasteuria castrates its hosts (Ebert et al. 2004) and prevalence can be very high in natural populations (e.g., Duncan et al. 2006), making it a potentially strong selective agent (Little and Ebert 2000). Daphnia are cladocerans that occur in a variety of standing water bodies (e.g., rock pools, ponds, lakes and swamps). Multiple species of Daphnia can occupy the same body of water, and the species D. magna, D. longispina, and D. pulex have been found to occur in sympatry in several localities in England (Stirnadel and Ebert 1997), Sweden (Bengtsson 1986), Finland (Ebert et al. 2001), and Switzerland (J. Andras, unpubl. data).

ATTACHMENT TO HOST ESOPHAGUS

We tested if the polymorphism for spore attachment that was found for native combinations of D. magna and Pasteuria (Duneau et al. 2011) was also present in native combinations of D. longispina and Pasteuria. In total, we tested 623 native and 883 nonnative host–parasite combinations with the attachment test.

The attachment test is described in full detail by Duneau et al. (2011). In short, individual hosts were placed singly in 24-well plates in 1 mL of artificial medium (ADaM, modified by using only 5% of the recommended selenium dioxide concentration, Klüttgen et al. 1994). Twenty thousand fluorescently labeled spores were added to each well, and animals were incubated for 1 hour at room temperature. Spore attachment was determined by examining exposed Daphnia with a Leica fluorescent microscope and checking for the presence of fluorescently labeled spores attached to the esophagus.

We sampled rock pools on islands along the Baltic coast of Finland, where Pasteuria and the three Daphnia species, D. longispina, D. magna, and D. pulex, naturally co-occur (Ebert et al. 2001). Twenty individuals were sampled from each of six D. pulex, 28 D. magna, and 23 D. longispina populations from 53 different rock pools located on 21 islands. Most samples were taken on islands close to Tvärminne Zoological station, but additional samples were collected on islands approximately 150, 225, and 250 km to the East (GPS coordinates of sampling locations can be found in Table S1). Sampled individuals were split into two groups of 10, and spore attachment to the esophagus was assessed using Pasteuria clone C14 native to D. magna (C14magna) for the first group and isolate P10 native to D. longispina (P10longispina) for the second group. Both parasites were originally collected close to Tvärminne Zoological station. C14magna is a single genotype of Pasteuria (see Luijckx et al. 2011). P10longispina is a field isolate (potentially containing more than one genotype) of Pasteuria that was obtained by the propagation of spores from 3 infected D. longispina individuals in D. longispina clone FS-30-1.

To test for consistency of the attachment test within Daphnia clones, we produced clonal lineages by collecting asexual offspring produced by a single mother from eight D. magna, seven D. longispina, eight sexual D. pulex (resting eggs are produced sexually), and eight asexual D. pulex (resting eggs are produced asexually) from the same metapopulation (see Table S2 for GPS coordinates). Clonal lineages were propagated and maintained under standard laboratory conditions (fed three times per week with 50 × 106 of the chemostat-cultured algae Scenedesmus obliquus, 16 h:8 h light dark cycle, 20°C, in artificial medium), and four individuals of each host clone were tested for spore attachment with Pasteuria. As found previously, attachment was a binary trait with spores either attaching in all individuals of one host clone or none (Duneau et al. 2011).

We extended our sampling to include Daphnia and Pasteuria from additional locations in Europe to verify the generality of our results. We sampled additional individuals from D. magna, D. longispina, and D. pulex from Finland (20 clones), England (6), Germany (29), Switzerland (12), France (3), Italy (6), Hungary (1), Iran (3), and Russia (5) and established clonal lineages that were propagated and maintained under standard laboratory conditions (for GPS locations, see Table S3). In addition to Pasteuria isolate P10longispina and clone C14magna, we tested each of these host clones with Pasteuria clone C1 from a Russian D. magna population (C1magna) and Pasteuria clone C20 from an English D. magna population (C20magna). At least four individuals for each host–parasite combination were tested.

INFECTION TRIALS

Attachment of spores to the host esophagus in native combinations of D. magna and Pasteuria leads to successful infection (Duneau et al. 2011). We tested whether this also holds in native combinations of D. longispina and Pasteuria and if this holds in nonnative combinations. To do this, we performed infection trials with subsets of the same host clones used to determine spore attachment.

For infection trials, we took Daphnia from mass cultures that were kept under standard laboratory conditions. We placed groups of four 1-week-old individuals of the appropriate host clones into separate 100-mL jars containing 20 mL of artificial medium. Jars received a first dose of 100,000 spores from the appropriate Pasteuria clone or isolate and a second dose of 100,000 spores the next day. This dose is known to cause 100% infection without lethal effect in D. magna (Regoes et al. 2003). Spore suspensions were produced by crushing dead infected hosts and assessing the spore density with a hemocytometer. Negative controls received placebo solution (crushed uninfected Daphnia). A week after exposure, jars were filled with 60 mL of fresh medium, and medium was replaced on a weekly basis thereafter. Daphnia were fed 2 × 106 cells per day at the start of the experiment, and food levels were increased gradually to 4 × 106 cells per day for D. pulex and D. longispina and to 5 × 106 cells per day for D. magna to accommodate for the increase in food demand of the growing individuals. Dead individuals were recorded every other day, but only those that died 14 or more days after exposure were checked for presence of Pasteuria spores. Individuals that died earlier cannot be reliably checked for infection and were excluded from the analysis. Forty days after exposure, remaining Daphnia were pooled per replicate, crushed, and checked for presence of Pasteuria spores using a phase contrast microscope (400×). When one or more of the four individuals in a jar was infected, we considered the replicate susceptible, when none were infected we considered it resistant.

We tested for infection by P10longispina and C14magna in host clones of eight D. magna, seven D. longispina, eight sexual D. pulex, and eight asexual D. pulex used above for determining the polymorphism of attachment (eight replicates per host clone). In addition, we performed infection trials with Pasteuria and host clones from other locations across Europe where we had previously observed attachment in nonnative combinations and a subset in which we had observed no attachment (eight to 10 replicates for each combination of host and parasite). In total, we tested 18 native and 83 nonnative host–parasite combinations with infection trials.

GENETIC IDENTIFICATION OF PASTEURIA CLONES AND ISOLATES

To verify that the clones and isolates of Pasteuria used in this study were genetically unique and to infer patterns of relatedness among them, we sequenced 1346 base pairs from the Pasteuria genome, spanning two genes (ribosomal protein RpsA—797 bp; predicted sporulation protein—374 bp) and 175 noncoding bases. Genomic DNA was extracted from resting spores of Pasteuria as described in McElroy et al. (2011). PCR reactions were performed in a volume of 50 μL containing 2 U of FastStart Taq Polymerase (Roche, Basel, Switzerland), 1× reaction buffer, 1.5 mM MgCl2, 20 pmol of each primer (h115F: AGCAGCTTCTTGGAAAGGAATTGGT; h115R: TGGGGTGAGAGGATTTGTACCTGC), 200 μM dNTP, and ∼100 ng of template DNA. The PCR profile consisted of an initial denaturation step at 95°C for 10 min, and 35 cycles of 95°C for 45 sec, annealing at 60°C for 1 min, and extension at 72°C for 60 sec, followed by a final extension at 72°C for 10 min. Amplicons were column purified (QIAquick PCR Purification Kit; Qiagen, Basel, Switzerland) and sequenced directly (Microsynth, Balgach, Switzerland).

Results

ATTACHMENT TESTS

Attachment of the four Pasteuria lines used in this study was observed in all native host species and almost all nonnative host species (Table 1). Attachment was in all cases specific for some host genotypes and showed within population variation. Within the Finnish meta-population where Pasteuria and the Daphnia species, D. longispina, D. magna and D. pulex, naturally co-occur, we found that isolate P10longispina is able to attach to its native host D. longispina as well as to D. magna. Clone C14magna attached to its native host D. magna as well as D. pulex. The other two nonnative combinations did not result in any attachment (Fig. 1). However, in our extended sampling we found that P10longispina was also able to attach to three of D. pulex (from France; Table 1). We also found that C1magna showed a similar pattern as C14magna, able to attach to its native host D. magna (Finland, Italy, Hungary) as well as D. pulex (Finland and Iran). Attachment of C20magna was observed in its native host D. magna (Germany, Italy, and Hungary) and nonnative hosts D. longispina (Finland) and D. pulex (Finland and Iran).

Table 1. Summary of attachment-tests with four Pasteuria on Daphnia magna, Daphnia pulex, and Daphnia longispina from different locations. In most locations, multiple Daphnia genotypes were sampled per pond and in some location multiple ponds were sampled. Numbers represent the number of host clones were Pasteuria attached to the esophagus over the total number of host clones tested per location. Each host clone was tested with at least four replicates and measurements within host clone were always consistent (all individuals either showing attachment or no attachment). A detailed representation of individual host clones can be found in Tables 2 and S4
  Pasteuria
  Native to D. magna 
   Native to D. longispina
SpeciesLocation(number of ponds)C1C14C20P10
  1. a

    Represent cases were infection trials were conducted (with either eight or 10 replicates). The consistency between attachment and infection only held in native combinations of Daphnia and Pasteuria (in gray). In nonnative host–parasite combinations no infections are observed.

D. magnaGermany (Stuttgart) (2)0/60/61/60/6
 Germany (Gaarzerfeld)0/20/21/20/2
 Finland (archipelago) (8)0/1a3/8a0/1a1/8a
 Italy (San Rossoro)3/33/32/30/3
 Russia (Moscow)0/20/20/20/2
 Switzerland (Winterthur)0/30/30/30/3
 Hungary (Bogárzó Tó)1/1a1/1a1/1a0/1a
D. longispinaGermany (Stuttgart)0/20/20/20/2
 Germany (Gaarzerfeld)0/40/40/40/4
 Germany (Rümmingen) (3)0/110/110/110/11
 Germany (Märkt)0/20/20/20/2
 Finland (archipelago) (17)0/10a0/7a4/10a3/7a
 Italy (San Rossoro)0/30/30/30/3
 (Oxford)0/30/30/30/3
 Switzerland (Winterthur)0/30/30/30/3
 Switzerland (Belenzona)0/30/30/30/3
D. pulexGermany (Stuttgart)0/20/20/20/2
 Finland (archipelago) (28)4/12a5/16a4/12a0/16a
 Russia (Moscow)0/30/30/30/3
 England (Oxford)0/30/30/30/3
 Switzerland (Winterthur)0/30/30/30/3
 France (Connaux)0/30/30/33/3a
 Iran (Tabriz)3/3a3/3a3/3a0/3a
Figure 1.

Daphnia from rockpools of the islands of the Finnish Skerry archipelago, where the species Daphnia magna, Daphnia longispina, and Daphnia pulex co-occur, were tested for attachment of Pasteuria spores to the esophagus with two Pasteuria that were isolated from the same metapopulation. Pasteuria clone C14 native to D. magna and the isolate P10 native to D. longispina. Spores of C14 attached to individuals of D. magna and D. pulex but never in D. longispina. Spores of P10 attached to D. longispina and D. magna but never in D. pulex (attachment in nonnative combinations represented by black box). Each population is represented by a circle diagram with on the left half proportion of attachment of C14 out of 10 host clones and on the right P10. Black indicates that attachment was observed, white indicates no attachment. Letters M, L, and P indicate D. magna, D. longispina, or D. pulex. Total number of observations per host species can be found in the summary table (bottom right, nonnative combinations in bold).

INFECTION TRAILS

All Pasteuria lines infected only certain genotypes of their native host and failed in all other native and nonnative genotypes. The perfect correlation previously observed between attachment and infection (Duneau et al. 2011) thus only holds in native combinations of Daphnia and Pasteuria. All Pasteuria types that were able to attach to their nonnative hosts failed to produce infection (Tables 1, 2).

Table 2. Attachment-tests and infection trials with C14 from Daphnia magna and P10 from Daphnia longispina on clones of D. longispina, D. magna, and Daphnia pulex (both asexual and sexual clones), all from one metapopulation in Finland. Attachment to the esophagus of some genotypes of D. magna and D. pulex (sexual) was observed for C14 and for P10 in D. longispina. Both methods, attachment-tests and infection trials, were perfectly consistent when the parasite was tested on the Daphnia species they were isolated from (gray “cases”), however none of the nonnative host–parasite combinations that showed attachment became infected. Each cell for the attachment-test represents the percentage of observed attachment in four individuals and each cell for the infection trial represents the percentage out of eight replicates were infection was observed in infection trials. Attachment and infection data from other locations in Europe (and Iran) showed similar patterns (Table S4) and the total number of host clones per location where attachment was observed/the total number of tested host clones per location can be found in Table 1
  Pasteuria
  C14 native to D. magnaP10 native to D. longispina
SpeciesHost cloneAttachmentInfectionAttachmentInfection
D. magnaAL1-4-410010000
 FS-13-c10010000
 FS-26-b0000
 N-16-d0000
 N-44-c10010000
 SP1-2-30000
 Xinb30000
 FHS2-11-3001000
D. longispinaF-7-a0000
 FS-13-c0000
 FS-30-100100100
 FS-6-140010088
 G-102-b0000
 G-122-100100100
 N-20-d0000
D. pulex (asex)FU1-57-a0000
 FU2-83-c0000
 M-69-b0000
 SK-39-d0000
 SK-44-b0000
 SK-45-d0000
 SYN-3-d0000
 SYN-6-c0000
D. pulex (sex)AB2-1-b0000
 KV1-1-d100000
 LH-3-c100000
 M-60-a100000
 N-69-b100000
 RO1-3-c0000
 SK-47-a0000
 VIN-4-b100000

HAPLOTYPE NETWORK

A total of 19 single nucleotide polymorphisms (SNPs) were observed in the 1346 bases sequenced from the Pasteuria genome. These SNPs resolved all Pasteuria types used in this study, indicating that they were indeed distinct parasite genotypes. A median-joining network of haplotypes (Fig. 2) illustrates that the level of divergence between Pasteuria from different host species (P10 vs. C1, C14, C20) is greater than the divergence among Pasteuria isolated from a single host (C1, C14, C20).

Figure 2.

Haplotype network of clones/isolates of Pasteuria. A median-joining maximum parsimony network based on 19 SNPs across 1346 base pairs of sequence illustrates the relatively high divergence of P10 originating from Daphnia longispina relative to Pasteuria types originating from Daphnia magna. Each node is labeled with the clone/isolate name and branch lengths are scaled to the number of mutational steps (reported as numbers above the branches).

Discussion

Attachment to the host esophagus, a necessary step in the infection process, was found in native and nonnative host–parasite combinations. This may indicate that the mechanism responsible for attachment to the esophagus is shared between host species that diverged more than 100 million years ago. Thus, with regard to the attachment step, Pasteuria has a wide host range. However, despite successful attachment, none of our Pasteuria was able to infect nonnative hosts. This finding indicates that the narrow host range of Pasteuria is determined by postattachment steps. Reports of Pasteuria infecting different host species thus suggest the existence of several parasite varieties, possibly different species, each with a narrow host range.

ATTACHMENT OF PASTEURIA IS CONSERVED BETWEEN HOST SPECIES

Attachment to the host esophagus is believed to be responsible for the strong specificity of DaphniaPasteuria interactions within D. magna (Duneau et al. 2011). Here we show that although attachment of the parasite spores is genotype-specific within species, spores can also attach to specific genotypes of nonnative host species (Table 1, 2 and Fig. 1). Furthermore, we show that the mechanism for attachment is polymorphic in all three Daphnia species studied, as is evident from specific attachment of Pasteuria to all three host species (Tables 1, 2 and Fig. 1). We offer three explanations for the observed patterns of attachment.

First, parasite spore attachment across different host species suggests that the same genetic mechanism operates in the different host species. It was recently shown that the genetics underlying specific attachment of Pasteuria in D. magna is based on few loci (one or two) that follow a matching allele model (Luijckx et al. 2013). A dominant allele provides resistance against Pasteuria C1, whereas homozygotes for the recessive allele were resistant against Pasteuria C19 (which has an infection pattern identical to the here used C20 Luijckx et al. 2011). The alleles that control specific attachment of Pasteuria may thus be shared between species of Daphnia. This would imply that Daphnia has an ancient polymorphism for attachment of Pasteuria predating the split of the subgenus Ctenodaphnia (with D. magna) and the subgenus Hyalodaphnia (with D. longispina; >100 million years). Trans-species polymorphisms for defense against parasites have been found in other taxa, for example, vertebrate MHC (>65 million years; Flajnik et al. 1999), TRIM5α in Old World monkeys (>8 million years; Newman et al. 2006), and members of the interferon pathway in mice (3 million years; Ferguson et al. 2008). If Daphnia indeed has a trans-species polymorphism for spore attachment, it may be among the oldest trans-species polymorphisms for host defense ever recorded and the first outside the vertebrates. To maintain trans-species polymorphism, some form of balancing selection is needed. Both coevolution by negative frequency-dependent selection and selection that varies in space and time can generate balancing selection (Ferguson et al. 2008) and are likely to operate in the DaphniaPasteuria system. With high virulence (Ebert et al. 2004), highly specific infection (Luijckx et al. 2011) and host resistance congruent with a matching allele model (Luijckx et al. 2013), the prerequisites for coevolution by negative frequency-dependent selection are met, and indeed empirical evidence consistent with negative frequency-dependent selection has been found for D. magna interacting with Pasteuria (Decaestecker et al. 2007; J. Andras unpubl. data). Differences in selection for resistance against Pasteuria in time and space are also likely to occur. The Finnish meta-population in this study consists of a large number of ponds scattered over thousands of islands, yet Pasteuria has been reported from only a few ponds (Ebert et al. 2001), indicating that potential selection for resistance only occurs locally. Coupled with frequent extinction and recolonization (Pajunen 1986; Altermatt et al. 2008), this may create fluctuating selection regimes in time and space (Thompson 2005). In accordance with this, we find large differences in resistance profiles of populations on a small geographic scale (Fig. 1).

The second explanation for the specific attachment across species is based on convergent evolution. Similar to the results of Ashfield et al. (2004), who found that Soybean and Arabidopsis thaliana evolved separate resistance genes recognizing the same bacterial avirulence protein (AvrB), the different species of Daphnia may have independently evolved mechanisms to prevent attachment of Pasteuria.

The third explanation posits that Pasteuria interacts with a genetic polymorphism that is maintained by another process, not related to host parasite coevolution. We are not aware of examples for such a scenario. Given the apparent strong selection of Pasteuria on the host population (Little and Ebert 2000), it seems unlikely that the polymorphism is not influenced by Pasteuria–Daphnia interactions.

PASTEURIA CONSIST OF MULTIPLE HOST VARIETIES

One of the aims of this study was to test whether Pasteuria is composed of different varieties, which are specific to different host species. The presence of multiple host varieties within parasites “species” has been reported for other parasites, and the difficulties in distinguishing cryptic species is usually attributed to the highly derived and reduced morphology of parasites. For example, anther smut fungus Microbotryum violaceum (Bucheli et al. 2000), the seabird tick Ixodes uriae (Mccoy et al. 2001), and the spiny-headed worm Leptorhynchoides thecatus (Steinauer et al. 2007) were shown to be composed of cryptic host-specific species. Pasteuria seems to be another example. We found that three Pasteuria clones native to D. magna and an isolate native to D. longispina showed specificity for infection in their native host species. Attachment is perfectly consistent with infection within native combinations, but not in nonnative combinations where attachment sometimes, but infection never occurs (Tables 1, 2). Attachment in nonnative hosts without infection indicates that the parasite fails in subsequent steps of the infection process (host penetration or proliferation). Thus, although spore attachment and activation seem to be conserved, later steps in the infection process appear to have diverged among the different host species. This is consistent with the notion of Decaestecker et al. (2007) and Duneau et al. (2011) who suggested that genes underlying the different stages in the infection process follow different evolutionary dynamics. Yet, most theoretical studies pertaining to evolutionary dynamics of infection consider infection to be a single-step process coded by a single set of genes. Studies that have considered a multistep process have found that the outcome of the evolutionary process can be dominated by a single step in the infection process (Agrawal and Lively 2003), although another study using an alternative genetic basis of resistance found many parameter settings where evolution happened across multiple steps in the infection process (Fenton et al. 2012). Our study and others (Duneau et al. 2011; Luijckx et al. 2013) strongly suggest that, within native combinations of Daphnia and Pasteuria, the attachment step dominates the outcome of the infection process. However, it is also clear that subsequent steps of the infection process limit the host range.

If our finding of a narrow host range is generally valid, it would imply that Pasteuria consists of a complex of multiple co-occurring cryptic host varieties. The findings of Stirnadel and Ebert (1997) that Pasteuria infected three Daphnia species in populations in South England, could then be explained by the presence of multiple Pasteuria varieties within a multihost community. This hypothesis is consistent with the clear differences in the epidemiological dynamics of Pasteuria infections in the three host species observed in these populations (H. A. Stirnadel and D. Ebert, unpubl. data). The haplotype network of Pasteuria lines used in this study gives further support to the hypothesis of cryptic varieties. The four lines of Pasteuria used in our study were morphologically indistinguishable, but showed substantial genetic differentiation (Fig. 2). Co-occurrence of multiple host varieties is supported by our data from Finland where the three Daphnia species live in sympatry (Tables 1, 2), yet Pasteuria lines infect only their native host species. Consistent with the newly emerging picture, are findings from ongoing field studies in Swiss and Finish ponds with sympatric D. magna, D. pulex, and D. longispina where over periods of several years only one Daphnia species was found to be infected with Pasteuria (J. Andras, unpubl. data; D. Ebert, unpubl. data). Cross-infection of Pasteuria native to D. magna to the South African species D. dolichocephala, which is in the same subgenus as D. magna was previously reported (Duneau et al. 2011). However, the finding that parasites are often host-specific, but occasionally are able to infect so-called nonhosts, is well known for other systems (see Antonovics et al. 2013, for a conceptual review).

The high specificity of Pasteuria may have implications for host and parasite demographic and evolutionary dynamics, which can depend on the degree of host specialization (Barrett et al. 2008). In contrast to generalists, parasites with a narrow host range are more likely to be locally adapted (Lajeunesse and Forbes 2002). In accordance with this, local adaptation for spore production was suggested for Pasteuria infecting D. magna (Ebert et al. 1998). In addition, parasites restricted to a single host species are expected to experience more frequent local extinction and recolonization events, which may influence genetic structure and effective population size (Barrett et al. 2008). The host range of a parasite may also have consequences for the evolution of virulence (Woolhouse et al. 2001). According to theory, parasites with a narrow host range can evolve to high levels of virulence whereas generalist parasites need to trade-off virulence across different host species (Regoes and Nowak 2000). High virulence is indeed observed in the D. magna–Pasteuria system (Ebert et al. 2004). Finally, our finding of narrow host range supports previous evidence for antagonistic coevolution by negative frequency-dependent selection in the D. magna and Pasteuria system (Decaestecker et al. 2007; Ebert 2008) and may suggest that coevolution among the different combinations of Daphnia and Pasteuria can maintain genetic variation and sexual reproduction (Jaenike 1978; Red Queen Theory, Hamilton 1980).

The narrow host range of Pasteuria may also have implications for community dynamics. The occurrence of Pasteuria may alter the potential for coexistence of Daphnia species by parasite-mediated competition similar to the Anolis lizard parasite Plasmodium azurophilum. In areas where this parasite is absent Anolis gingivinius outcompetes Anolis wattsi, whereas the presence of P. azurophilum reduces the competitive ability of A. gingivinius allowing A. wattsi to coexist (Schall 1992). Similar evidence is missing for Daphnia, but Daphnia are known to differentially alter phytoplankton composition (Kasprzak and Lathrop 1997) and a change in Daphnia composition could thus have consequences for the entire community.

Conclusion

Our study highlights the explanatory power of separating the different steps in the infection process for understanding the evolution of host–parasite interactions. The three easily distinguishable steps in the infection of Daphnia by Pasteuria (activation, attachment, and proliferation) show contrasting patterns. The activation step showed neither within nor between host species variation (Duneau et al. 2011), the attachment step shows polymorphism both within and between host species, and the postattachment step(s) reveals host species specificity. Thus, it is likely that different forms of selection act on the genes underlying these steps. Using a step-wise approach, our study resolved the conflicting reports on host range in the DaphniaPasteuria system. This has consequences for our understanding of local adaptation, evolution of virulence, parasite-mediated competition, and coevolution in this model system. Similar approaches may help to gain explanatory power in other host–parasite systems.

ACKNOWLEDGMENTS

The authors thank M. Lori, L. Theodosiou, P. Flury, and J.Wetzel for help with experiments and J. Hottinger, N. Boileau, and U. Stiefel for assistance in the laboratory. The authors thank C. Metzger, E. Ho, J. Dudycha, and two anonymous reviewers for comments on earlier version of the manuscript. The authors thank the staff of the Tvärminne Zoological field station of Helsinki University for help and support during the work on the metapopulation. J. Pokki was instrumental in providing some of the Daphnia samples. This study was supported by the Swiss National Science Foundation.

Ancillary