Pathogens as potential selective agents in the wild



Mélanie Dionne, Fax: 418-646-6863; E-mail:


Pathogens are considered a serious threat to which wild populations must adapt, most particularly under conditions of rapid environmental change. One way host adaptation has been studied is through genetic population structure at the major histocompatibility complex (MHC), a complex of adaptive genes involved in pathogen resistance in vertebrates. However, while associations between specific pathogens and MHC alleles or diversity have been documented from laboratory studies, the interaction between hosts and pathogens in the wild is more complex. As such, identifying selective agents and understanding underlying co-evolutionary mechanisms remains a major challenge. In this issue of Molecular Ecology, Evans & Neff (2009) characterized spatial and temporal variation in the bacterial parasite community infecting Chinook salmon (Oncorhynchus tshawytscha) fry from five populations in British Columbia, Canada. They used a 16S rDNA sequencing-based approach to examine the prevalence of bacterial infection in kidney and looked for associations with MHC class I and II genetic variability. The authors found a high diversity of bacteria infecting fry, albeit at low prevalence. It was reasoned that spatial variability in infection rate and bacterial community phylogenetic similarity found across populations may represent differential pathogen-mediated selection pressures. The study revealed some evidence of heterozygote advantage at MHC class II, but not class I, and preliminary associations between specific MHC alleles and bacterial infections were uncovered. This research adds an interesting perspective to the debate on host–pathogen co-evolutionary mechanisms and emphasizes the importance of considering the complexity of pathogen communities in studies of host local adaptation.

In the context of frequent environmental changes, pathogens may represent selective agents promoting local adaptation in host populations through adaptive genetic variation. Growing evidence now exists for pathogen-mediated selection on genes of the major histocompatibility complex (MHC) (reviewed in Sommer 2005; Piertney & Oliver 2006). However, the underlying mechanisms involved in pathogen resistance are not well understood and the identification of selective agents in the wild remains in its infancy. In this study, Evans & Neff (2009) examined the diversity and structure of bacterial communities infecting Chinook salmon fry populations in the wild (Fig. 1). They found a high diversity of bacteria infecting fry, with 55 unique phylotypes identified over the five populations studied. This estimate, which is probably conservative because only certain types of bacteria can grow under the studied conditions (agar plates under aerobic conditions), suggests the occurrence of a higher than expected diversity of bacteria infecting Chinook salmon at the beginning of its life cycle when freshwater mortality can be high (Healey 1991). The prevalence of bacteria was relatively low, which is similar to results from other juvenile salmonids (Dionne et al. 2009) and could represent epizootic pathogens, affecting a limited number of fish but having the capacity to spread rapidly in populations under favourable environmental conditions. The freshwater pathogenic or potentially pathogenic bacteria identified in Chinook salmon from this study constitute a necessary first step to understand host-pathogen interactions in the wild.

Figure 1.

 Infected male Chinook salmon (Oncorhynchus tshawytscha). Elwha River, Washington State. (Photo: John McMillan.)

Evans & Neff (2009) identified spatial variability in both the number of bacterial phylotypes and infection rates across populations (phylotypes: 4–21; infection rates: 10–29%), with no apparent geographical trend. Interestingly, infection rates in salmon inhabiting the main stem and one tributary of the Skeena River were mildly different, but not significantly, potentially suggesting some variability in host-pathogen interactions within a river system. The spatial variability in infection rates and bacterial diversity detected in kidneys of Chinook salmon in this study could be mediated by environmental selection pressures differing among habitats or by variance among populations in immune resistance, potentially related to genetic variability at the MHC. Indeed, high infection rates in host individuals could be the result of a high selection pressure from the environment or a low immune resistance capability, or both as previously observed in Atlantic salmon (Salmo salar; Dionne et al. 2009). In future studies, it will be interesting to differentiate between these two hypotheses by quantifying selection pressure from the host environment through the identification of pathogens in water and soil, for example, in the case of fish populations.

A UniFrac metric of phylogenetic community similarity was used to compare bacterial communities infecting fish across populations and years (Evans & Neff 2009). This adds a new and interesting perspective on how pathogen community composition, in addition to pathogen diversity and richness, might vary according to geographical distance. They found relatively low weighted UniFrac estimates (spatial scale: 0.094–0.161; temporal scale: 0.128 to 0.155) suggesting a certain level of similarity in pathogen communities that infect Chinook salmon through space and time. Interestingly, an isolation by distance pattern in bacterial community phylogenetic similarity was uncovered across infected populations, which could suggest differential selection pressure as geographical distance increases. However, as southern populations were sampled earlier in the summer than northern populations, short-term variation in bacterial community cannot be excluded to explain this pattern. On the other hand, no significant change in bacterial community composition was observed between the two consecutive years of the study in three populations, suggesting a certain stability over a longer temporal scale. Clearly, this shows how diverse and complex bacterial communities might be in the wild and emphasizes the need to consider pathogen community composition through space and time when studying host adaptation.

When relating infection rates and MHC genetic variability, Evans & Neff (2009) found some evidence of heterozygote advantage at MHC class IIβ, but not at MHC class Iα. Indeed, the proportion of infected individuals was significantly lower for MHC class II heterozygotes than for MHC class II homozygotes. These findings support the role of MHC class II in bacterial resistance and are concordant with balancing selection imposed by pathogens. The authors also found associations between MHC class I and II alleles and the prevalence of specific bacteria in kidney through a co-inertia (COIA) analysis showing significant covariance between MHC alleles and individual bacterial abundance matrices (12.6% variance explained by the first two axes of the COIA model). However, these associations were not all cross-validated by a second analysis, possibly because of the low bacterial prevalence and infection rate observed within the studied populations, as suggested by the authors. Nevertheless, these represent preliminary analyses identifying potential susceptibility alleles towards some infections and could help orient future work on pathogen resistance in a context of multiple pathogen exposure. Such associations between specific MHC alleles and the prevalence of infections was previously observed in multiple vertebrates including humans (e.g. Hill et al. 1991), chickens (e.g. Briles et al. 1977) and salmonids (e.g. Langefors et al. 2001), and support the frequency-dependent selection and the variable selection in time and space hypotheses (Nei & Hughes 1991; Hedrick 2002). In the literature so far, frequency-dependent selection has received more support as a mechanism of balancing selection maintaining MHC diversity in wild populations (Sommer 2005). This study adds another perspective to the debate on host-pathogen interactions and underlines the importance of conducting such studies in the wild to complement existing controlled laboratory experiments.

Understanding mechanisms underlying host-pathogen interactions remains a major challenge and will certainly continue to offer exciting opportunities for future research. The study of Evans & Neff (2009) brings an interesting perspective on how diverse and complex pathogen communities infecting fish might be in the wild and underlines the importance of considering this complexity in host local adaptation studies. This research also identified potential selective agents and related their prevalence to host adaptive genetic variability, and as such, took us one step closer to reaching a global understanding of pathogen resistance and host adaptation in the wild.


I thank Kristina M. Miller and Andrew Whiteley for insightful comments on this article. John McMillan provided the Chinook salmon photograph.