Discussion
Using detailed multihost, multiparasite time series data and network analysis approaches, we estimated potential cross-species transmission in Daphnia communities. We found evidence that parasites in these lakes differ in their ability to infect and transmit between host species; three parasites showed high host breadth and strong potential for cross-species transmission (based on high overlap of epidemics in different host species in a given lake on a given date). Focusing on these three common multihost parasites, two hosts, D. retrocurvaand the well-studied D. dentifera , are most implicated in cross-species transmission. This was particularly true for the bacteriumP. ramosa . In contrast, for another common multihost parasite, the bacterium S. cienkowskii , cross-species transmission seemed less reliant on particular species.
Basing our estimates of plausible cross-species transmission on synchrony of single pathogen epidemics, we found that the identity of both the pathogen and host species were important to our networks. Pathogens varied in their specificity of host species and in their apparent ability to transmit among host species. With one notable exception, pathogen species with greater host breadth (implying lesser specificity) displayed greater potential for cross-species transmission. The exception is G. vavrai; this parasite was found to infect multiple host species, but epidemics were separated in time leaving no potential for cross-species transmission. Notably, it has been suggested that an intermediate, non-Daphnia host, is required for theG. vavrai lifecycle (Refardt et al. 2002), which would preclude cross-species transmission among Daphnia in our communities for this pathogen. This is also the case for the other pathogen for which we found no potential cross-species transmission, L. obtusa (Refardt et al. 2002). These results thus provide support for our method of cross species transmission detection.
For our three most common pathogens, our estimates of network centrality indicate that host identity plays a strong role in potential for cross species transmission, with some host species having a disproportionate influence on community transmission. However, the degree to which host identity influenced cross species transmission – and the particular host(s) that most strongly impacted it – varied by pathogen. For two of our pathogens, a subgroup of hosts were more central within the network (supporting greater cross species transmission), while for the third,S. cienkowskii, no host displayed greater cross species transmission potential. Thus, this approach allows us to identify pathogen species for which some hosts might have an outsized impact on disease dynamics in these communities, and also gives us insight into multihost-multiparasite communities as a whole. The potential patterns of interspecific transmission for P. ramosa can be more readily studied in lab experiments, as compared to those for S. cienkowskii ; while experimental infections with S. cienkowskiiare possible, they are rare and experiments are challenging. Thus, instead, molecular analyses of infections in these species seem more promising. A study of P. ramosa in these lake populations genotyped the parasite primarily from D. retrocurva and D. dentifera (based on the availability of infected hosts in the field), but also included several infected D. parvula and oneCeriodaphnia dubia (Shaw 2019; Shaw et al. in review). In this analysis, the genotypes found in D. parvula were always also found in D. retrocurva , but the same genotype was not found infecting both D. dentifera and D. retrocurva . Thus, the molecular results partially support the analyses here that indicate thatP. ramosa moves between host species, but also suggests there are barriers to doing so that are not captured by these analyses. Molecular analyses of S. cienkowskii -infected hosts from natural populations would be interesting; based on the network analyses, we predict less genetic structuring by host species for S. cienkowskii .
Parasites differ in their ability to infect and transmit among different host species, though we cannot yet say why some parasite species commonly infect more host species in the same site; both pathogen life history traits and taxonomic group have been suggested as potential mechanisms (Pedersen et al. 2005). Our two bacterial parasites, S. cienkowskii and P. ramosa , generally had higher host breadth and potential cross-species transmission compared to other parasites; however, the oomycete parasites, B. paedophthorum and ‘Spider’, significantly differed in both host breadth and potential cross species transmission. Moreover, more recent work on a microsporidian gut parasite that was not included in this study, O. pajunii , has found substantial transmission across host species (Dziuba et al. 2023). Overall, these results make it difficult to conclude whether taxonomic group strongly influences the likelihood of cross species transmission. We know that these parasites differ in key aspects of their transmission, with some (e.g., B. paedophthorum ) being continually shed and others (e.g., P. ramosa, S. cienkowskii , andM. bicuspidata ) being obligate killers. Moreover, within a parasite species, particular aspects of phenotype or genotype can influence transmission. In P. ramosa , only certain genotypes can attach to the esophagus of the host, a key first step in infection (Duneau et al. 2011; Ebert et al. 2015); when this occurs across species, infections often fail at a later (but currently unknown) stage in the infection process (Luijckx et al. 2014). In contrast, withM. bicuspidata , interspecific transmission seems to be asymmetric and to depend on parasite genotype and spore size (Shaw et al. 2021; Sun et al. 2023). Future studies that aim to uncover the traits that promote – or prevent – cross species transmission will help us better understand and predict infections in complex natural communities.
Host species differed in the pathogens they hosted and in their relative potential importance for cross species transmission (Fig 4). Phylogeny can influence patterns of cross-species transmission (Streicker et al. 2010; Parker et al. 2015; Mollentze and Streicker 2020). In particular, hosts might be more susceptible to parasites transmitted from more closely related host species. Our P. ramosa network suggests thatCeriodaphnia are significantly less likely to be involved in cross species transmission than Daphnia species, supporting a phylogenetic influence for this pathogen. This is consistent with the limited experimental evidence we have, which suggests that cross species transmission between C. dubia and D. dentifera is possible, but rare (Auld et al. 2017). It is also consistent with genetic evidence collected in 2015 (overlapping with this study), which found that the P. ramosa strain in C. dubia was an outgroup to the strains in Daphnia (Shaw 2019; Shaw et al. in review)). In contrast, our results for B. paedophthorum do not suggest challenges traversing greater phylogenetic distances, asC. dubia did not differ from the Daphnia hosts displaying the greatest transmission potential. This is also consistent with experimental studies, which were readily able to transmit this parasite across host species (Duffy et al. 2015). As P. ramosa is known to have a matching allele interaction with hosts (Bento et al. 2017), it is probable that pathogens with this infection strategy are more likely to have phylogenetically regulated cross-species transmission.
Within our communities, we saw variation in epidemic size within our pathogens. Focusing on the three pathogens for which we were able to create networks, we found that two (P. ramosa and B. paedophthorum ) displayed significant differences in infected host density among host species (Fig 5). The patterns in infected host density closely replicated the patterns for centrality among hosts (Fig 4, Fig 5). Our remaining pathogen, S. cienkowski , displayed no differences in epidemic size by host and also no differences in host centrality among hosts. Previous work has demonstrated a positive (though nonlinear) causal relationship of infected host density to within species transmission (Rachowicz and Briggs 2007; Roberts and Hughes 2015). As centrality, here, represents potential cross-species transmission, our results suggest that infected host density (or epidemic size) within host species may also influence cross-species transmission. Thus, these results extend known within species transmission mechanisms to the multi host species scale.
We are beginning to disentangle the ecological and epidemiological drivers of cross species transmission (Lloyd-Smith et al. 2009; Plowright et al. 2015; Plowright et al. 2017). However, we still lack the ability to accurately predict novel transmission events and host shifts, or the degree to which cross species transmission maintains epidemics within communities (Mollentze and Streicker 2020). By applying network analysis techniques to a rich dataset from a well-studied planktonic host-parasite system, we were able to uncover host species that seem particularly important for maintaining levels of infection in a community, as well as differences among parasite species in the degree to which they infect multiple host species. Because of its correlative nature, this work cannot definitively say whether certain host or parasite species are more involved in cross-species transmission, but it can inform hypotheses about transmission patterns within this community. For example, it suggests that parasites differ in the degree to which host phylogeny constrains their ability to spillover between hosts (e.g., P. ramosa vs. S. cienkowskii ). In doing so, it points to additional studies that will help us better understand this system (e.g., studies characterizing the genetic diversity of P. ramosa vs. S. cienkowskii in different host species during epidemic outbreaks). More broadly, it provides additional support for the value of network approaches for providing insights into patterns of transmission in nature and for generating hypotheses regarding transmission in complex multihost-multiparasite communities.