Infection with Gyrodactylus turnbulli reduced the distance covered by small adult gup- pies during the escape response by an average of 20.1%, whereas the escape response of larger fish was unaffected (Fig. 2.2). Wild guppy females are significantly larger than males (Fig. 4.2), so this result provides evidence that parasite-induced vulnerability to predation is sex-biased in natural populations. As well as a size effect, there appears to be an inherent sex difference: the distance covered by infected males, but not females, was negatively correlated with infection load (Fig. 2.3). The weaker escape response of infected males relative to infected females supports our hypothesis that, rather than being an exception to the rule, guppies reveal a further cost for the sicker sex in the form of male-biased parasite-induced vulnerability to predation. The resulting higher mortality among infected males, potentially correlated with their infection load, explains why observational studies record female-biased Gyrodactylus spp. infection in high predation populations (Chaper 4; Gotanda et al., 2013).
Infection with G. turnbulli reduced the distance covered by small adult guppies during the escape response, indicating that these parasites increase guppy vulnerability to pre- dation (Domenici and Blake, 1997; Bergstrom, 2002; Ghalambor et al., 2004). Parasites can affect fish swimming performance in a number of ways (reviewed by Barber et al., 2000). Gyrodactylus spp. infection causes changes in the guppy epidermis (Gheorghiu
et al., 2012), reduced feeding and activity levels (van Oosterhout et al., 2003; Kolluru et al., 2009), and at higher infection loads the pathology includes clamped fins (Cable et al., 2002) with obvious negative impacts on swimming (Hockley et al., 2013). The
effect of infection on host escape response could also be due to activation of a costly immune response (Lochmiller and Deerenberg, 2000), independent of the effects of the parasite: Eraud et al. (2009) and Janssens and Stocks (2014) found that stimulation of the immune response using non-pathogenic methods increased the vulnerability to predation of Eurasian collared doves (Streptopelia decaocto; using lipopolysaccharide injection) and larval damselfly (Coenagrion puella; using non-pathogenic bacteria) re- spectively. The immune response of our test fish would have been activated: we tested the guppies on days 8 and 9 post infection and at this stage the growth of the parasite infrapopulation on many hosts slows, and may become negative (Fig. 6.1), indicative of an active immune response (Bakke et al., 2007).
We found that the effect of infection on guppy escape response was strongest among small adults, and that the distance covered by males was negatively correlated with infection load. We suggest this size- and sex-dependence is due to smaller and male fish being less tolerant of infection than larger and female fish. Positive correlations between body size and resistance or tolerance of parasitism have been recorded in a number of species, for example Soay sheep, Ovis aries (see Coltman et al., 2001) and garter snakes,
Thamnophis elegans (see Sparkman and Palacios, 2009). Although we failed to detect a
size- or sex-dependent change in body condition (a common measure of tolerance; Råberg
et al., 2009) in our experimental fish over the nine days of their infection, data from
wild populations indicate that Gyrodactylus spp. tolerance is lower among male than female guppies (Chapter 3). Additionally, Gyrodactylus spp. infections are energetically costly (van Oosterhout et al., 2003; Bakke et al., 2007; Kolluru et al., 2009), and it is likely that differences in tolerance between these groups of fish would be manifest in other physiological costs that explain the decrease in escape response (Barber et al., 2000; Lochmiller and Deerenberg, 2000). In support of this hypothesis, smaller and male guppies select habitats with lower velocity, less turbulent flow, particularly when infected with G. turnbulli, suggesting they are more energetically limited generally (Hockley
et al., 2013), as has been found for smaller sticklebacks (Krause et al., 1998), and are
more affected by infection (Hockley et al., 2013). As males are smaller than females (Fig. 4.2), this size effect, in addition to inherent sex differences in Gyrodactylus spp.
Chapter 2. A further cost for the sicker sex 25 tolerance (Chapter 3) and swimming ability (Hockley et al., 2013), will likely translate into male-biased parasite-induced vulnerability to predation in wild populations. The sex- and size-biased parasite-induced vulnerability to predation indicated by our findings will have implications for the evolutionary ecology of this host-parasite inter- action. For example, our results suggest that Gyrodactylus spp. parasites select for larger body size in the host, in opposition to the selection imposed by predators, which select for smaller body size (Reznick et al., 1990). While previous work has indicated that smaller fish may be more resistant to Gyrodactylus spp. in that they develop smaller infection loads (van Oosterhout et al., 2003, 2008), the present study suggests that infection load is less important than the size of the fish in determining the effect of infection on escape response. Parasitism could therefore be an important force in driving the evolution of larger body sizes in upper course, low predation populations. Further, removal of the most heavily infected males will have implications for the epidemiology of the disease. Due to their low shoal fidelity and high contact rates resulting from the search for mating opportunities (Croft et al., 2003a,b), males are likely to be key in both intra- and inter-shoal transmission of the parasite. Consequently, selective predation on males, particularly if it is correlated with their infection load, may well reduce parasite prevalence and intensity in natural populations (Packer et al., 2003). It is also likely that this process will exacerbate the male-biased predation already implicated in driving female biased sex ratios in this species (McKellar et al., 2009; McKellar and Hendry, 2011).
We have demonstrated the importance of the direct (present study) and indirect (Chap- ter 4) effects of predators on the distribution of parasites among host populations. In the guppy-gyrodactylid system at least it is likely that the relative importance of these effects changes through time: guppy population densities, and therefore the presence of predators at a given site is highly temporally variable (Magurran, 2005). Because the indirect effect of predators is mediated by an evolved guppy trait, shoaling, that differs between high and low predation populations even after several generations in the laboratory (Huizinga et al., 2009), it will drive constantly higher transmission rates among lower course guppies (Chapter 4). By contrast, patterns driven by selective predation on infected guppies will only be evident when predators are present. It is clear from Fig. 4.2 that both these processes are important. Relative to low predation sites, small infected guppies are underrepresented in high predation sites, probably due to the direct effect of predators: these are the fish most prone to parasite-induced vulnerability to predation (present study). Large uninfected guppies are similarly relatively scarce in high predation sites, probably due to the indirect effects of predators: increased transmission due to increased shoaling results in higher re-infection rates (Chapter 4). In summary, these findings clearly demonstrate that the community in which the host and
parasite interact can have profound implications for that interaction: we support recent calls for further investigation of the impact of community interactions, particularly predation, on disease ecology (Packer et al., 2003; Hatcher et al., 2006; Lafferty et al., 2008; Raffel et al., 2008, 2010).