INTRODUCTION
To maximise reproductive success, most angiosperms are tasked with attracting pollinating mutualists whilst also evading herbivore antagonists (Strauss 1997; Lucas-Barbosa 2016; Kessler & Chautá 2020). However, plant interactions with pollinators and herbivores, and the traits that mediate these interactions, are often inter-linked (Theiset al. 2007; González-Teuber & Heil 2009; Galen et al.2011; Kessler et al. 2019; Rusman et al. 2019). Such linkages typically lead to ecological trade-offs for plants. For instance, pollinators often preferentially forage on larger, more apparent flowers and inflorescence displays (Conner & Rush 1996; Parachnowitsch & Kessler 2010). Yet greater visual or olfactory apparency can also come at the cost of increased herbivore damage when co-opted as shared signals (Halitschke et al. 2008; Sletvold & Grindeland 2008; Theis & Adler 2012; Knauer & Schiestl 2017; Santangelo et al. 2019). Plant-herbivore interactions can likewise bear large consequences for pollination success (Kessleret al. 2011; Muola et al. 2017). The presence and action of herbivores, and in particular their damage to leaves (folivory) and flowers (florivory), can deter pollinators via a range of direct and indirect mechanisms (Jacobsen & Raguso 2018; Moreira et al.2019; Haas & Lortie 2020). These include visually (e.g. modification of floral aesthetics; reduced resource allocation to floral rewards and displays (McCall & Irwin 2006)), olfactorily (e.g. release of herbivore-induced plant volatiles (Kessler & Chautá 2020)), and gustatorily (e.g. upregulated expression of unpalatable defensive compounds in floral nectar (Adler et al. 2006)). Hence given the multiple routes by which these ecological effects can manifest for plants, the net outcome for fitness is thereby expected to be highly context-specific e.g. (Gegear et al. 2007).
Yet despite growing appreciation of the ecology of plant-herbivore-pollinator systems, it remains poorly understood how these interactions ultimately affect phenotypic selection on traits at the microevolutionary scale. The high degree of ecological linkage and specificity inherent in these systems suggest that pollinator and herbivore selection may often be conflicting (i.e., exerted in opposing directions) and diffuse (i.e., context-dependant on the presence or ecological effects of the other) (Strauss et al. 2005; Sletvoldet al. 2015; Knauer & Schiestl 2017; Ramos & Schiestl 2019; Sletvold 2019). However, although herbivory is known to potentially constrain floral evolution (Johnson et al. 2015; Jogesh et al. 2017; Ramos & Schiestl 2019; Santangelo et al. 2019), a recent meta-analysis by Caruso et al. (2019) revealed an almost categorical lack of studies which simultaneously quantify pollinator- and herbivore-mediated selection on the same trait. Further studies are hence required to examine the relative importance and dynamics of pollinator and herbivore selection on traits that link pollination and herbivory. Beyond floral traits, these should also include direct and indirect (i.e., natural enemy-mediated) defences. Few studies to date have examined the potential for pollinators to select on defence traits related to herbivory (Kessler & Halitschke 2009; Egan 2015; Ramos & Schiestl 2019). Yet such selection may in fact be commonplace in plants (Egan et al. 2016), especially as an adaptation to mitigate herbivore deterrence of pollinators.
In this study we manipulated pollination and herbivory in a common garden experiment with woodland strawberry (Fragaria vesca L.). We examined phenotypic selection on several chemical traits previously identified as markers of direct and indirect defence for this species (see below and Weber et al. , 2020a; Weber et al. , 2020b), alongside several morphological traits potentially important for attraction of pollinators and herbivores. We tested and found at least partial support for the hypotheses that: 1) pollinators and herbivores positively select on defence-related traits – which in this system are expected to aid against the direct and pollinator-mediated costs of herbivory (Muola et al. 2017; Muola & Stenberg 2018; Weberet al. 2020a); 2) pollinators and herbivores impose conflicting selection on plant attractive traits – as shared host-selection cues potentially used by both agents; and 3) that the above selection regimes are diffuse – i.e., that pollinator-mediated selection is modified (in strength, and possibly direction) when herbivory is manipulated, and vice versa. Investigations of this kind can thereby provide greater insight into the eco-evolutionary dynamics of plant-herbivore-pollinator interactions.