Introduction

In nature, all plants and animals are colonised by microbes (Barriere, 2006; Ley et al., 2006; Vántus et al., 2014). The composition of these microbial communities is highly diverse and includes harmful, neutral, and beneficial microbial species (Ley et al., 2006), including those that can be important players in host defence against parasites, a phenomenon referred to as ‘defensive mutualism’ (King, 2019; May and Nelson, 2014). Recognised for over a century, defensive mutualism has been observed in plants (Mendes et al., 2011) and in a range of animals (Dillon et al., 2000; Dong et al., 2009; Jaenike et al., 2010; Koch and Schmid-Hempel, 2011), including humans (Kamada et al., 2013; Ley et al., 2006; Maynard et al., 2012) wherein microbes can supplement host immune systems (Abt and Artis, 2013; Hooper et al., 2012; McFall-Ngai et al., 2013).
The net benefits of defensive mutualism are dependent upon the presence of pathogens (Clay et al., 2005; King and Bonsall, 2017; Lively et al., 2005). Whilst hosts can benefit from microbe-mediated protection, defensive symbionts can be less beneficial to the host in the absence of enemies, due to metabolic and physiological costs (King, 2019). For example, in the interaction of aphids and the bacteriumHamiltonella defensa , the host tissue is harmed by defensive toxins that protect against infection from parasitoids (Vorburger and Gouskov, 2011). In some cases, possessing protective microbes might be more beneficial to the host than investing in its own immune system (Martinez et al., 2016). From the perspective of the symbiont, it is most useful to its host under high pathogen prevalence, and thus can persist in the host population (Palmer et al., 2008). Nevertheless, a stable symbiotic interaction is hypothesized to be evolved and maintained (Kwiatkowski and Vorburger, 2012) only when the host benefit of carrying defensive symbionts outweighs any costs. The interactions of obligate and defensive symbionts and hosts can be stable for millions of years (Moran et al., 2005).
Not all environments are constantly pathogen rich which might shift the balance of costs and benefits during defensive mutualisms, particularly during coevolutionary interactions (King and Bonsall, 2017). Pathogen prevalence can be spatially (King et al., 2009) or temporally variable, the latter in the case of seasonal epidemics (e.g., flu peaks each winter in the northern hemisphere (Finkelman, 2007) or rabies in North American skunks which peaks in Autumn (Gremillion-Smith and Woolf, 1988)). Different environmental factors can influence disease transmission such as an increase in malaria risk in warmer regions after rainfall (Altizer et al., 2006), or an increase in contact rate and thus higher flu infection rate during the winter months (London and Yorke, 1973). The impact of other temporally heterogeneous factors on the strength and direction of selection on species interactions have been explored (oxygen concentration (Dey et al., 2016), resource availability (Friman et al., 2011; Friman and Laakso, 2011; Hiltunen et al., 2012), environmental productivity (Harrison et al., 2013)). Whether the varied presence of pathogens can similarly alter selection for symbiotic interactions has been explored theoretically (Fenton et al., 2011), but remains to be empirically tested.
Here, we examined the impact of temporal variation in pathogen infection on the evolution of microbe-mediated protection. We usedCaenorhabditis elegans as a worm host and allowed it to be colonised by a bacterium (Enterococcus faecalis ) that protects against infection by Staphylococcus aureus (King et al., 2016).Enterococcus faecalis has been shown to be protective across animal microbiomes (Kommineni et al., 2015; Martín-Vivaldi et al., 2010). It has been previously shown that E. faecalis can evolve to provide enhanced protection when residing inC. elegans hosts during constant pathogen infection (King et al., 2016; Rafaluk-Mohr et al., 2018). From this, we predict that variation in pathogen infection might limit the evolution of microbe-mediated protection. In the present study, we experimentally co-passaged C. elegans with protective E. faecalis and infected the host with evolutionary static pathogenic S. aureus at different intervals of host evolution. We also examined whether pathogen presence at the initial formation of the coevolving interaction is crucial to the evolution of protection. We show that enhanced microbe-mediated protection emerged out of novel coevolutionary host-microbe interactions and during pathogen infection, regardless of its temporal variability or the time point of first infection. Enhanced protection was only effective during pathogen infection. If hosts survived infection, they could recover and had the same longevity and reproductive output across treatments. These results thus suggest that even occasional pathogen infection can select for defensive mutualism, revealing the potential for this phenomenon to be widespread in nature.