Geographic mosaics of selection.
Despite the seemingly large spatial variation in the intensity and effects of HP receipt the potential role of HP transfer interactions as a force generating geographic mosaics of selection is mostly unknown (but see Kay & Schemske 2008, Hopkins & Rausher 2012, Arceo-Gómez et al. 2016a). HP receipt can act as a selective force driving the evolution of floral strategies that mitigate female (Morales & Traveset 2008) and male fitness costs (conspecific pollen loss; Moreira-Hernandez & Muchhala 2019). Examples include adaptations to enhance pollen placement (e.g. Armbruster et al. 1994. Minnaar et al. 2019a), shifts in flowering phenology (e.g. Waser 1978) and adaptations to minimize pollinator sharing such as flower trait divergence (e.g. Hopkins & Rausher, 2012), pollinator specialization (Muchhala et al. 2010) and increased selfing (e.g. Fishman & Wyatt, 1999). Specifically, Ashman and Arceo-Gómez (2013) proposed two main evolutionary strategies to mitigate female fitness effects, i.e. tolerance or avoidance of HP receipt. Although these strategies were proposed as a means to explain among-species variation in HP receipt (Ashman & Arceo-Gómez 2013), plants within the same species that are exposed to different HP transfer environments may experience the same evolutionary pressures and outcomes. For instance, Hopkins and Rausher (2012) showed evidence for divergent selection pressures on flower color in Phlox drummondiipopulations as a result of HP transfer from Phlox cuspidata.Selective pressure on genes that affected floral pigmentation occurred only in sympatric Phlox populations to prevent hybridization (i.e. HP avoidance; Hopkins & Rausher 2012), or maybe even direct HP effects on reproductive success (e.g. stigma clogging), thus generating spatial mosaics of selection. In a similar study, Kay and Schemske (2008) found pollen-pistil incompatibilities had evolved only in sympatric populations of two Costus species, and not in isolated populations, thus providing evidence for geographic mosaics of selection on HP tolerance strategies (also see Arceo-Gómez et al. 2016a). Furthermore, Mimulus guttatus growing in serpentine seeps in California showed an increase in flower longevity as an adaptive response to minimize effects of high levels of HP receipt with increasing co-flowering diversity (Arceo-Gómez & Ashman 2014a). In this case, population-level differences in HP receipt likely led to changes in the adaptive value of flower longevity hence influencing spatial patterns of floral evolution (Arceo-Gómez & Ashman 2014a). Evolutionary adaptations that minimize male fitness costs have also been shown and these include traits that enhance accuracy in pollen placement (e.g. Muchhala & Potts 2007, Armbruster et al. 2014), increase in pollinator specialization (e.g. Muchhala et al. 2010) and floral constancy (Moreira-Hernandez & Muchhala 2019). Nonetheless, evidence of within-species variation in these male-driven evolutionary responses across a species’ distribution range is still limited. In addition to HP tolerance and avoidance strategies, selection can favor mechanisms that increase the degree of autonomous self-pollination in order to preemptively minimize HP effects (e.g. Fishman & Wyatt 1999, Randle et al. 2018). For instance, Fishman and Wyatt (1999) demonstrated that selection favored selfing and selfing-related traits only inArenaria uniflora  populations that grew in sympatry with congeneric A. glabra. They further  show that HP transfer rather than pollinator competition was the main driver of selection (Fishman & Wyatt 1999). Thus, HP transfer not only has the potential for generating geographic mosaics of selection in floral traits, but also on patterns of mating system evolution.
Changes in HP transfer dynamics as a result of changes in community species composition (e.g. Arceo-Gómez & Ashman 2014a, Johnson & Ashman 2019) can further contribute to generate selection mosaics via diffuse selection (e.g. Iwao & Rausher 1997; Stinchcombe & Rausher 2001). For instance, Iwao and Rausher (1997) proposed that diffuse co-evolution would occur if, 1) susceptibility to different selective pressures (e.g. HP donors) are genetically uncorrelated, 2) the presence/absence of one species (e.g. HP donor) does not mediate the incidence of effects caused by another, and 3) the fitness effects of one species (e.g. HP donor) do not depend on the presence/absence of another. So far, we know that at least one of these conditions is likely violated in the context of HP receipt. Specifically, Arceo-Gómez and Ashman (2011) showed that the fitness effects of HP receipt can strongly depend on the number and identity of HP donor species present on the stigma (violating condition three). Furthermore, it is possible that HP recipient susceptibilities to different HP donor species will be correlated if they all depend on the efficiency of recipient’s self-incompatibility system, that is, self-compatible plants may be similarly susceptible to a wide array of HP donors (violating condition one; Hiscock & Dickinson 1993, Murfett et al. 1996). It has also been shown that the presence of one species can cause a rearrangement of HP interactions in the community, mediating the incidence of HP effects from other species, and thus violating condition two sated above (Johnson & Ashman 2019, Daniels & Arceo-Gómez 2020). In other words, the effects and responses to selection via one HP donor are likely non-independent of the presence/absence of other HP donors in the community, thus setting the stage for diffuse selection (e.g. Stinchcombe & Rausher 2001). While the role of multispecies interactions in mediating diffuse evolutionary processes is a topic of ongoing study (Johnson & Stinchcombe 2007), the potential importance of diffuse selection via HP transfer interactions has been entirely overlooked. Nevertheless, the above evidence suggests that the potential for within-species variation in HP receipt to act as a driver of microevolutionary processes is strong. Yet, its contribution to generating geographic mosaics of selection remains largely unexplored.