Discussion
Host selection (= choice or preference) is a trait of high relevance for the study of infection dynamics, which has drawn much attention for parasitoids and vectors (e.g. Henry et al., 2009; Campbell et al., 2013), but which has been often neglected for other parasites. Parasites differ in their ability to drive the encounter of hosts. While some parasites rely on contact between hosts or passive encounter of a free-living stage to reach compatible hosts, motile and vector-borne parasites are able to find hosts in a less random manner. Discriminating among hosts provides the opportunity of maximising fitness by selecting the most suitable options. Little is known about how parasites select hosts or the degree to which such choices are linked to their success, but some empirical evidence supports the notion that host preference is adaptive. Ectoparasitic mites (Spinturnix spp.) are more attracted by female and subadult bats than by male bats, and they survive longer in the former, indicating an adaptive selection of the more beneficial type of host (Christe et al. 2007). Similar evidence of host preference in relation with host competence was observed in multi-host systems. Cercariae of trematodes in the family Plagiorchiidae were put in selection chambers to choose among tadpoles of four host species, and the choice pattern observed was negatively correlated with the level of resistance of each host species. The preferred host species was that with the greatest proportion of metacercariae encysted, whereas the least chosen host was the one in which most cercariae failed (Sears et al. 2012). In the Philornis- multihost system, larvae that feed on the main host have higher chances of surviving than those that parasitise alternative hosts (Manzoli et al. 2018).
Most previous work on host selection has been conducted in experimental settings, under unnatural circumstances and providing limited opportunities. These experiments can establish differential attractiveness among hosts, but are unable to evaluate the degree to which parasites can reliably differentiate among an entire assemblage of alternative hosts in natural contexts. The Philornis- multihost system provides the rare opportunity to evaluate host attractiveness in the real world (for a detailed description of the advantages of this system see Suppl. Mat. in Manzoli et al. 2018). It is noteworthy that the longitudinal studies conducted here not only provide correlational evidence (very strong associations between putative cause and alleged effect), but also demonstrate robust temporal coherence, and the analyses controlled for potential confounders that may cause spurious associations (Höfler 2005).
Gravid Philornis females can fly relatively long distances seeking bird broods that could be hosts of their larvae. Being the encounter the result of such an active search, there is opportunity for evolving capacity to discriminate among hosts and select the ones which maximise larval fitness (Forbes et al. 2017). Our results showed that, in this system, the parasite has an enormous ability to discriminate between hosts of differing quality, and that host selection depends on the structure of the host assemblage and on the size of the parasite population (host demand). The probability and intensity of parasitism in alternative hosts increased as better hosts were less available, especially with growing host demand. It is worth highlighting that the results observed were drastic: bad hosts were virtually not used at all when better alternatives were sufficiently available, even at high host demand levels.
Another study in a different system did not find such context-dependency of host selection. Under experimental conditions, motile cercariae of the trematode Ribeiroia ondatrae were attracted differentially by five alternative hosts (Johnson et al. 2019). Unlike the experiment by Sears et al. (2012), the attractiveness was not correlated with host competence. The most preferred host was the bullfrog (Rana catesbiana ), a non-native and large-bodied species in which most cercariae failed to encyst. The attractiveness of the most competent host in that experiment, Peudacris regila , was not altered by variations in the assemblage composition, including presence/absence of bullfrogs. However, bullfrog presence decreased infection for the other 3 alternative host species, which is in coincidence with the effect of Great Kiskadees on Thornbirds observed in this work.
In response to environmental variability, parasites can evolve from generalism to specialism or vice versa (Lievens et al. 2018). Dietary specialisation is predicted to evolve when the fitness obtained from feeding on different resources (i.e. hosts) is lower than that obtained by consuming (i.e. parasitising) a limited subset, whereas generalism should evolve when differences in energetic gains between resources are not large (Lyimo & Ferguson, 2009). The robust dataset analysed here shows that specificity may also be plastic within individuals, swinging conveniently between generalism and specialism depending on the structure of the host community. In presence of sufficient broods of optimal hosts, ’P. torquans c. A.’ behaves as a specialist, being found almost exclusively on Great Kiskadees. When the optimal host is absent or very little available, however, the parasite can broaden its host range, beginning by choosing the most suitable options, leaving bad hosts for moments when good alternatives cannot be found. This sort of facultative generalism is highly advantageous for the parasite, as it may accommodate its choice so it uses suboptimal resources only at times when it is profitable. In the study area, Great Kiskadees are present only during a limited window of time, while Little Thornbirds (bad alternative hosts) breed actively for much longer (Fig. 1). Hence, being able to decide using bad hosts only when more competent ones are missing may prove essential for the parasite to be able to persist and thrive in this ecosystem.
From the host’s perspective, the degree of host selection observed in this system may favour the evolution of resistance. The strategy of tolerance of Great Kiskadees appears to be very successful (Manzoli et al. 2018), but it implies that this host is targeted and therefore highly exposed to the parasite. Resisting is highly costly, but it may be beneficial not only because it would reduce the parasite burden, but also because of its potential to dissuade host selection, as parasites would be driven to select hosts in which their fitness is maximised.
The level of host discrimination observed for Philornis flies rivals the cognitive decisions made by vertebrates when selecting their food items, their refuge or their partners. Further, it is in line with the optimal foraging theory (Pyke et al. 1977) which predicts that foragers will be selective when a high-quality food type is abundant, but less selective when that high-quality item is scarce (Emlen 1966, Stephens and Krebs 1986). Underlying these selection processes there are supply-demand dynamics. For example, avian parents should include lower quality foods in the nestling diet only when brood demand is higher than expected or when there is a shortage of food availability (Wright et al. 1998; Sauter et al. 2006). Similarly, in our system, Philornisflies select low quality hosts when host demand is high and better hosts are lacking.
The remarkable capability of ’P. torquans c. A.’ of discriminating among hosts should be enabled by a substantial amount of time and energy dedicated for host searching, coupled by cues that allow reliable differentiation among hosts. It is noteworthy that the parasite can definitely distinguish between Little Thornbird (bad alternative host) and Greater Thornbird (good alternative host), which is outstanding considering that both Phacellodomus spp. are phylogenetically very close to each other, and both build nests of identical materials and similar shape. The mechanisms of host selection by dipteran parasites have been studied mainly for haematophagous species. The cues involved in them are thermal, visual and chemical (Takken and Knols, 2010). Nonetheless, chemical cues seem to prevail in highly selective species (Jossart et al., 2019). Anophelesgambiense is highly anthropophilic, and this preference is driven by specific components found in human sweat (Braks et al., 2001; Zwiebel & Takken, 2004) and skin microbiota (Verhulst et al., 2009). Variability in the chemical composition of the sweat determines differential attractiveness among humans (Smallegange et al., 2011). Likewise, species of Glossinia (tse-tse fly) show variable attraction to different odoriferous compounds (Torr & Solano, 2010). Oviposition site selection by Glossina brevipalpis was found to be independent of the salinity and pH of the substrate, and instead was driven by presence of con-specific or hetero-specific puppae (Renda et al., 2016), which would be mediated by odoriferous compounds (Saini et al., 1996). Chemical cues have also been shown to drive host selection in other arthropods, such as ticks (Ferreira et al., 2019).
Our findings have important implications for the epidemiology of multi-host parasites, specifically for community scale patterns of disease transmission. The concept of facultative generalism adds pertinent elements to the notion of a ’dilution effect’ and the relevance of biodiversity on infection dynamics (Schmidt & Ostfeld 2001). Shifts in biodiversity and community composition can influence the likelihood of a parasite or vector encountering a highly competent host species rather than a low-quality host. The ’dilution effect’ implies that greater biodiversity results in less encounters with highly competent hosts and, therefore, in less parasite success and transmission (Schmidt & Ostfeld 2001). However, if parasites or vectors selectively infect the most competent hosts as observed in this study, the presence of alternative ones (which increases with biodiversity) may have little effect on transmission (Levine et al., 2017). Nevertheless, when the alternative hosts are humans, domestic animals, or an endangered wildlife species, the plastic host selection observed here, influenced by biodiversity, can be of high importance for public health, animal husbandry, and biodiversity conservation, respectively. For example, humans might not be targeted by a given parasite as long as more suitable hosts are available. This kind of ’deflection effect’ is related to the ’dilution effect’ because they are both a function of biodiversity. More biodiversity may dilute the infection of highly competent hosts, or it can also divert it from a host of interest. Reductions in biodiversity might therefore cause that humans (or domestic animals, or endangered species) become increasingly chosen by parasites. Thus, although the abundance of a given parasite is highly dependent on the abundance of its ’reservoir’ host (e.g. mean abundance of ’P. torquans c. A.’ is low in the community in years with few Great Kiskadee broods; Manzoli et al. 2013), lack of main hosts would paradoxically increase infection risk in an alternative host of interest.