How shrub encroachment under climate change could threaten pollination services for alpine wildflowers: A case study using the alpine skypilot, Polemonium viscosum

Abstract Under climate change, shrubs encroaching into high altitude plant communities disrupt ecosystem processes. Yet effects of encroachment on pollination mutualisms are poorly understood. Here, we probe potential fitness impacts of interference from encroaching Salix (willows) on pollination quality of the alpine skypilot, Polemonium viscosum. Overlap in flowering time of Salix and Polemonium is a precondition for interference and was surveyed in four extant and 25 historic contact zones. Pollinator sharing was ascertained from observations of willow pollen on bumble bees visiting Polemonium flowers and on Polemonium pistils. We probed fitness effects of pollinator sharing by measuring the correlation between Salix pollen contamination and seed set in naturally pollinated Polemonium. To ascertain whether Salix interference occurred during or after pollination, we compared seed set under natural pollination, conspecific pollen addition, and Salix pollen addition. In current and past contact zones Polemonium and Salix overlapped in flowering time. After accounting for variance in flowering date due to latitude, Salix and Polemonium showed similar advances in flowering under warmer summers. This trend supports the idea that sensitivity to temperature promotes reproductive synchrony in both species. Salix pollen is carried by bumble bees when visiting Polemonium flowers and accounts for up to 25% of the grains on Polemonium pistils. Salix contamination correlates with reduced seed set in nature and when applied experimentally. Postpollination processes likely mediate these deleterious effects as seed set in nature was not limited by pollen delivery. Synthesis: As willows move higher with climate change, we predict that they will drive postpollination interference, reducing the fitness benefits of pollinator visitation for Polemonium and selecting for traits that reduce pollinator sharing.


| INTRODUCTION
Pervasive shrub encroachment into herb-dominated ecosystems of high altitudes and latitudes can be attributed to climate change (Formica, Farrer, Ashton, & Suding, 2014;Huss et al., 2017). Warmer temperatures promote woody encroachment in alpine tundra ecosystems through infilling of openings and advancement of tree-line (Myers-Smith et al., 2011). Shrubs create living snow fences and a thick layer of leaf litter, impacting resident flora and fauna by altering microclimate and nutrient cycling (Becklin, Pallo, & Galen, 2012;Dona & Galen, 2007;Myers-Smith et al., 2011). Paleo-ecological evidence suggests that shrub species are preadapted to invade the open conditions of tundra biomes and have extended their ranges into high altitude and latitude ecosystems under past periods of favorable climate (Naito & Cairns, 2011). Opportunistic growth, clonal spread, and high allocation to reproduction increase biomass and cover ("shrubification") for these woody gap species, allowing them to rapidly spread into meadow openings (Myers-Smith et al., 2011;Fig. 1). These same life history traits have the potential to alter consumer behavior and reconfigure negative (herbivory) and positive (pollination) plant-animal interactions (e.g., Lara-Romero, Garcia, Morente-Lopez, Iriondo, & Brody, 2016;Muñoz, Celedon-Neghme, Cavieres, & Arroyo, 2004).
Here we ask whether encroachment of shrubby subalpine willows (Salix glauca and brachycarpa; hereafter Salix) into alpine ecosystems interferes with pollination and thus reduces fitness in a native alpine wildflower, the skypilot, Polemonium viscosum (hereafter, Polemonium).
In the Rocky Mountains of North America, willows along with other woody species are responding to warmer temperatures by expanding their altitudinal range and encroaching into herb-dominated alpine vegetation (Formica et al., 2014;Petry et al., 2016;Fig 1). Willows are dioecious and plasticity in eco-physiological traits of male shrubs and trees in particular, may promote invasiveness under climate change (Dudley & Galen, 2007;Petry et al., 2016;Tognetti, 2012). Petry et al. (2016) propose that broad environmental tolerance in males of dioecious species facilitates their upward migration. Male willows of the Rocky Mountain subalpine complex, Salix glauca-brachycarpa sustain shoot growth over a broad range of environmental conditions with extension of new shoots in one year giving rise to prolific flowering in the next (Dudley, 2006;Sakai, Sasa, & Sakai, 2006). Their massive floral displays and numerous, small pollen grains exemplify traits associated with negative heterospecific pollen receipt (hereafter, HP; Ashman & Arceo-Gómez, 2013;Loughnan, Thomson, Ogilvie, & Gilbert, 2014;Carvalheiro et al., 2014).
Because willows are important early-season flower resources for pollinators (Formica et al., 2014;Moquet, Mayer, Michez, Wathelet, & Jacquemart, 2015) altitudinal migration of males may also negatively affect pollination regimes of resident alpine wildflower species. At high altitudes and latitudes, Salix acts as a magnet species, drawing pollinators away from the more diminutive floral displays of tundra plants (Lara-Romero et al., 2016;Mosquin & Martin, 1967). Here we investigate the potential for willow shrubification of alpine meadows to reduce reproductive success of resident wildflowers due to HP via pollinator sharing, focusing on the skypilot, P. viscosum, an obligate outcrosser that depends primarily on bumble bees for pollination (Galen, 1996).
The extent of pollinator sharing between insect-pollinated species depends on their spatial proximity. Polemonium occurs over a 500 m altitudinal gradient from tree-line to summit in the Rockies (Galen, Zimmer, & Newport, 1987). With altitude, this landscape decreases in surface area and thus habitable space (Elsen & Tingley, 2015). This geomorphological pattern implies that all else being equal, Polemonium populations will have broader ranges in lower parts of the alpine zone where shrubification is ongoing than on the upper tundra slopes that serve as refugia from this process. Separation of flowering times could moderate the impacts of spatial proximity in these low altitude contact zones (Poole & Rathcke, 1979). Flowering synchrony is promoted when species share environmental controls on timing of flower development (Post, Pedersen, Wilmers, & Forchhammer, 2008). Thus, if altitudinal immigrants converge with resident species in use of cues for budburst then competitive interactions may ensue among them (Brunet & Larson-Rabin, 2012). Conversely, asynchrony may arise when close neighbors differ in cues for onset and tempo of flowering (CaraDonna, Iler, & Inouye, 2014;Galen & Stanton, 1995 Polemonium is a model system for studies of pollinator-mediated selection (Galen, 1996;Galen, Kaczorowski, Todd, Geib, & Raguso, 2011). If woody encroachment alters pollinator foraging behavior, it may select for functional traits that limit pollen interference before (e.g., corolla tube length; Galen & Cuba, 2001) (Fig. 1b). Salix and Polemonium flower early in the season, typically in mid to late June in these four low alpine contact zones.
Both species provide early food resources for flower-visiting insects in high altitude habitats and are heavily frequented by bumble bee queens during nest establishment (Byron 1978). Individuals depend on insect pollination for outcrossing and reproductive success. The small dioecious flowers of Salix bloom in a mass display, offering pollen (male flowers) and nectar (both sexes) to bee and fly pollinators.

| Flowering phenology
We noted incidence of flowering time overlap in the four contact

| Pollinator sharing
In 2016, we surveyed the frequency with which primary pollinators of

Polemonium, (queens of the bumble bee species Bombus balteatus) carry
Salix pollen when foraging from Polemonium flowers in contact zones.

ety of insect pollinators including queens of the bumble bee species
Bombus balteatus whose visits account for 60%-95% of pollination in Polemonium (Galen, 1996). Although these queens are relatively scarce and difficult to survey, Salix pollen was found on all ten individuals flower for up to a two-week span (Galen & Kevan, 1980;Post et al., 2008), specimen collection date is a coarse indicator of flowering time.
Nonetheless, data suggest that willows and skypilots show similar advancement in flowering period with warming (Fig. 3). Temperature is a driver of phenotypic plasticity in flowering schedule in wildflowers of high altitude environments and has also been reported to advance flowering in Salix glauca (Brunet & Larson-Rabin, 2012;Inouye, 2008;Post et al., 2008). Our results are consistent with these findings though experimental temperature gradients are needed (e.g., Brunet & Larson-Rabin, 2012) to eliminate confounding factors in alpine ecosystems (e.g., snowmelt, growing season length, and soil nutrient flux (Ernakovich et al., 2014).
In historically stressful alpine environments, plant recruitment is rare, and insect pollinators have a crucial role in promoting outcrossing and seedling establishment (Geib & Galen, 2012;Utelli & Roy, 2000).
Our results indicate that the widespread "shrubification" of these high altitude ecosystems under climate change has potential negative indirect impacts on the quality of insect pollination for a widespread alpine wildflower. Given the implications of these effects for population recruitment, efforts to identify topographic conditions buffering Polemonium from willow HP are warranted. The rate of woody encroachment into alpine ecosystems shows well documented spatial trends, correlating with slope aspect across Polemonium's latitudinal range. These relationships suggest that such refugia from willow HP exist (Elliott & Cowell, 2015). Our results imply they may have considerable conservation value.

ACKNOWLEDGMENTS
Field and laboratory assistance were provided by Austin Lynn, Carlie

AUTHORS CONTRIBUTIONS
JAK carried out field experiments, collected data, and contributed to the design of the study; NMS contributed to data analysis and presentation, writing and microscopy; ZM carried out herbarium and field surveys; and CG led the writing of the manuscript and design of the study. All authors contributed critically to the drafts and approved the final draft for submission to the journal.

CONFLICT OF INTEREST
None declared.

DATA ACCESSIBILITY
Data and SAS Code for all statistical analyses are archived at Dryad… Digital Repository. doi:10.5061/dryad.2p2bh