Honey bees are the dominant diurnal pollinator of native milkweed in a large urban park

Abstract In eastern North America, the field milkweed, Asclepias syriaca L. (Asclepiadaceae), is used in planting schemes to promote biodiversity conservation for numerous insects including the endangered monarch butterfly, Danaus plexippus (Linnaeus) (Nymphalidae). Less is known about its pollinators, and especially in urban habitats where it is planted often despite being under increasing pressure from invasive plant species, such as the related milkweed, the dog‐strangling vine (DSV), Vincetoxicum rossicum (Kleopow) Barbar. (Asclepiadaceae). During the A. syriaca flowering period in July 2016, we surveyed bees in open habitats along a DSV invasion gradient and inspected 433 individuals of 25 bee species in 12 genera for pollinia: these were affixed to bees that visited A. syriaca for nectar and contain pollen packets that are vectored (e.g., transferred) between flowers. Of all bees sampled, pollinia were found only on the nonindigenous honeybee, Apis mellifera (43% of all bees identified), as well as one individual bumblebee, Bombus impatiens Cresson. Pollinia were recorded from 45.2% of all honeybees collected. We found no relationship between biomass of DSV and biomass of A. syriaca per site. There was a significant positive correlation between A. syriaca biomass and the number of pollinia, and the proportion vectored. No relationship with DSV biomass was detected for the number of pollinia collected by bees but the proportion of vectored pollinia declined with increasing DSV biomass. Although we find no evidence of DSV flowers attracting potential pollinators away from A. syriaca and other flowering plants, the impacts on native plant–pollinator mutualisms relate to its ability to outcompete native plants. As wild bees do not appear to visit DSV flowers, it could be altering the landscape to one which honeybees are more tolerant than native wild bees.


| INTRODUCTION
Bees are essential pollinators, and their long-term persistence, especially in human-dominated landscapes, requires an abundance and diversity of flowers that serve as pollen and nectar sources (Kearns & Inouye, 1997;Müller et al., 2006). Bee populations are declining globally (Kerr et al., 2015;Potts et al., 2010), and subsequently, actions to support pollinators have increased substantially in habitats around the world and include a range of participants from government to researchers to citizens (Vanbergen, 2013). One plant often used in planting schemes aimed at supporting pollinator diversity in northeastern North America is the field milkweed, Asclepias syriaca L.
(Asclepiadaceae). Asclepias syriaca is a native perennial herb and the oviposition site for the endangered monarch butterfly, Danaus plexippus Linnaeus (Nymphalidae), as well as a food plant for countless other insect species (Urquhart, 1960). Although once considered a "weed" in Ontario, it was removed from the noxious weed list in 2014 (Shahani, del Río Pesado, Schappert, Serrano, & Oberhauser, 2015).
Milkweed flower morphology and pollination are unusual compared to other flowering plants (Wyatt, 1976). Within each flower, nectar is secreted inside five stigmatic chambers (Galil & Zeroni, 1965). The accumulating nectar attracts many pollinators; however, no pollen is consumed. Milkweed pollen is massed into packets onto structures called pollinia, which include two pollen masses (packets) joined via translator arms to a corpusculum ( Figure 1). The corpuscula of each pollinium is set above one of the five stigmatic chambers and when a pollinator visits, it inadvertently removes the pollinia with their appendages when seeking nectar from a flower and deposits them in the same manner into different flowers (Fritz & Morse, 1981). When the pollinator visits a second milkweed flower, the pollen packets split off from the corpuscula, which remain on the pollinator and can be counted. Thus, vectored pollen loads and the proportions potentially transferred to flowers can be estimated from the number of "complete" pollinia and "deposited" pollinia (corpusculum and translator arms only) remaining on insect body parts (Kephart & Theiss, 2004;Robertson, 1929).
Asclepias syriaca has been shown to attract diverse floral "generalist" pollinators, including bumblebees (Bombus spp.) and the nonindigenous Western Honey bee (Apis mellifera L.) (Ivey, Martinez, & Wyatt, 2003;Kephart & Theiss, 2004;Robertson, 1929). For example, Morse (1981) showed that the native Bombus vagans Smith and Bombus terricola Kirby made up 50.2% and 21.5% of all visits, respectively, and honeybees accounted for 12.3% of visits in undisturbed rural fields in Maine. In the same study region, bumblebees accounted for 44%-73% of A. syriaca pollination that resulted in mature seed pods, and in areas where nonindigenous honeybees were absent, this increased to as high as 95% (Morse, 1985).
Although urbanization and agriculture are known threats to milkweed populations (Brower et al., 2012;Oberhauser et al., 2001) Sheeley & Raynal, 1996). Once established, DSV displaces most surrounding herbaceous vegetation indiscriminately through rapid growth that quickly overtakes and "strangles" surrounding plants, including tree saplings, as they compete for access to sunlight (Antunes & Sanderson, 2013). DSV is also classified as a milkweed (Bongard, Navaranjan, Yan, & Fulthorpe, 2013) and so the potential competitive interactions for habitat are amplified as DSV establishes and exploits niches similar to those occupied by A. syriaca.
In this study, we were interested in the potential direct and indirect impacts of DSV on the pollination of A. syriaca by bees, which can be inferred by the collection of pollinia on the bodies of bee specimens obtained in field surveys. DSV could directly impact A. syriaca abundance through competition for space and nutrients, and indirectly by impacting the number and type of pollinators visiting A. syriaca, thereby increasing pollen limitation and reducing seed set. We were interested in identifying the community of diurnal pollinia-carrying bee visitors to A. syriaca and DSV flowers in our study area to address two hypotheses: (i) that greater DSV abundance will be negatively correlated with A. syriaca abundance, as DSV directly displaces native vegetation and (ii) that increasing DSV invasion will indirectly lead to lower rates of pollinia vectored between flowers of A. syriaca.

| Site
All sampling occurred in 2016 in the Rouge National Urban Park (http://www.pc.gc.ca/eng/pn-np/on/rouge/index.aspx) (total area: 79.1 km 2 ), located in the northeast of Toronto, which is Canada's largest city (metro Toronto population: >5 million). The park has a multitude of different habitats and uses, including agricultural and industrial activity. While ecological restoration initiatives are ongoing F I G U R E 1 A honeybee affixed with numerous pollinia. Inset image: picture of Asclepias syriaca pollinium; (a) pollen packet, (b) translator arm, (c) corpusculum. Photograph credit: Marylouisse Feliciano A B C throughout the park, there remain many heavily degraded habitats that include established populations of DSV and other invasive species. Honeybees are not feral in the park; they are managed in hives within the region due to agricultural activities in the region, but information on the locations or numbers of hives is not available. In the southern edge of the park within the boundaries of the city of Toronto, we sampled at seven open meadow sites experiencing different levels of DSV invasion. All sites were 50 × 50 m, at least 300 m apart, and with a ground layer of grasses, herbaceous flowers, and small shrubs.

| Plant biomass
Each 50 × 50 m site was broken down into 1 × 1 m grid cells, and five 1 × 1 m cells (hereafter referred to as "plots") were randomly selected for measuring plant biomass, which required complete above-ground plant removal. In late June 2016, all individuals of DSV and A. syriaca per plot were identified, removed, and put in a standing drying oven at 60°C for a minimum of 48 hr until constant weight was reached. The dried samples were then weighed (in grams) to determine dry biomass per plot (g/m 2 ). For each species, site-level biomass was determined by taking the average biomass value from each of the five plots.

| Analysis
To examine relationships between the abundance of A. syriaca and DSV, we used a linear mixed effects model to compare biomass of each plant species, calculated from the five plot-level samples from each of the seven sites (N = 35) with site included as a random factor. We used linear regression analysis to examine the relationship between site-level biomass of A. syriaca and biomass of DSV with wild (nonhoney) bee species richness and abundance from samples summed over all three dates (July 4, July 11, and July 19). Mixed effects models were used to examine the total number of "complete" pollinia (including pollen packets) and "deposited" pollinia (corpusculum only) collected by honeybees, and the proportion of pollinia vectored (calculated as "deposited" pollinia divided by the total; e.g., "complete" + "deposited") with site-level A. syriaca biomass and DSV biomass. Site and the time (each 30-min block) during which the site was sampled for bees were included as random factors in each model. Analyses were completed using the lmer and lmerTest package (Kuznetsova, Brockhoff, & Christensen, 2015) with the R statistical program v3.2.2 (R Core Team, 2015).

| RESULTS
Over the three sampling periods during which A. syriaca was in bloom, hand netting yielded 433 bees from 25 species in 12 genera (Table S1). All bees collected were female. Of this total, 186 individuals were honeybees, and of these, 43.0% had at least one

| DISCUSSION
In a large urban park, we found that >99% of diurnal pollination of the field milkweed, A. syriaca was carried out by the nonindigenous Western Honey bee, Apis mellifera. Our findings suggest that A. syriaca is pollinated predominantly by honeybees in human-impacted region, unlike in similar studies carried out in natural areas decades earlier in which bumblebees were the dominant visitor (e.g., Macior, 1965;Morse, 1985). In this study, 55% of all bees sampled were wild (nonhoney) bees, but none had pollinia attached. Many of these bee species are presumably small enough to obtain nectar while avoiding the pollinia, while others might avoid A. syriaca to minimize interactions with honeybees. Asclepias syriaca does however provide significant habitat for numerous other important insects, including the endangered monarch butterfly, which uses the plant for oviposition (Inamine, Ellner, Springer, & Agrawal, 2016). Moreover, bumblebees are known visitors to A. syriaca but are experiencing declines in North America (Cameron et al., 2011;Kerr et al., 2015), and so for these purposes should be promoted in planting initiatives around metropolitan Toronto and throughout the region.

| Impacts of invasive DSV on Asclepias syriaca biomass
We found a negative but nonsignificant relationship between increasing biomass of DSV and A. syriaca biomass at the plot-and site level and do not have sufficient evidence to accept our first hypothesis that the invasive plant has a negative impact on the presence of A. syriaca. Both of these plants are milkweeds and so likely overlap in their fundamental niche space leading to high competition between these species, although our data do not display evidence of such negative interaction. DSV has been shown to have negative impacts on an assortment of taxa. For example, Ernst and Cappuccino (2005) found that DSV supported the lowest number of phytophagous insects, as well as stem-and ground-dwelling arthropods, and supports almost no pollinators.
We found a negative but nonsignificant effect of DSV biomass on the number of pollinia affixed to bees in our study system. As we did not find any DSV pollinia on any bees in this study, these findings support previous observations that little to no insects are attracted to DSV flowers (St. Denis & Cappuccino, 2004); however, Lumer and Yost (1995) noted 14 species of fly visiting the related Vincetoxicum nigrum, which is also introduced to northeastern North America and occurs in similar habitats (Gleason & Cronquist, 1991)

| Pollinia as an indicator of milkweed pollinators versus milkweed visitors
We found that more than half of all bees identified in this study did not have pollinia attached-including all of the nonmanaged wild beesdespite the milkweed A. syriaca providing rich and abundant nectar as an attractant to pollinators. Pollinia are big structures (Figure 1) compared to individual pollen grains, and we found that only honeybees and one bumblebee were identified as A. syriaca pollinators in our study. Other bee species identified (Table S1) could include A. syriaca visitors seeking nectar but are too small and not nearly as hairy as compared to honeybees and bumblebees for pollinia attachment.
More work is needed to identify and differentiate bee pollinators from bee visitors of A. syriaca, as the plant may have great value in pollinator management planning as a nectar resource.

| Honeybees as primary pollinators of native Asclepias syriaca
Honeybees have been present in North America for centuries and are common visitors to A. syriaca. In our study, we found that the introduced honeybee is the primary pollinator of the native A. syriaca.
T A B L E 1 Site-level summary of bees collected, average, and total complete Asclepias syriaca pollinia and corpusculum (mean ± SE) and the proportion vectored, as well as the biomass of A. syriaca and DSV This finding has been repeated in other recent studies of the plant; for example, in a grassland remnant in the city of Bloomington, Indiana, Theiss, Kephart, and Ivey (2007) found that honeybees visited A. syriaca the most and foraged on it longer than native bee species. Also, in disturbed and remnant areas near Urbana, Illinois, Willson and Bertin (1979) found that honeybees accounted for 26%-49% of vectored  Fritz and Morse (1981) found 97% of bumblebees recorded had pollinia affixed to them compared to only 9% of honeybees. Further, in Morse (1982), the author describes more than 88% of bumblebees recorded having more than one pollinia affixed. It is possible that honeybees visit A. syriaca in higher proportions in and around urban areas where they are managed more abundantly, but more work is needed.
Although these findings illustrate the role of honeybees as pollinators of native plants (Aizen, Morales, & Morales, 2008), the high number of honeybees found with pollinia affixed and the complete lack of other diurnally active pollinator species could reflect interference competition (Huryn, 1997;Schaffer et al., 1983), or strong selection for A. syriaca to attract honeybees over bumblebee species. Our evidence indicated that most honeybee visits to A. syriaca (80.5%) occur before 12:00 p.m. Willson and Bertin (1979) have suggested that honeybees have a selective pressure on A. syriaca floral biology such that the timing of nectar availability and peak flowering is more aligned with honeybee activity instead of that of other native pollinators. For example, they found that the abundance of honeybee visitors to A. syriaca coincided with a shift in the temporal patterns of total pollinator visits (Willson & Bertin, 1979). Many visitors to A. syriaca are nocturnal, and the authors found a shift in nectar production to earlier in the day, when honeybees are foraging, from previously typical production in early evening for attracting nocturnal moths. Honeybees are abundant and good dispersers and thus could bring higher quantities of pollen to A. syriaca, thereby replacing some of the services that native pollinators provide to A. syriaca plants.
With populations of bumblebees declining across North America (Kerr et al., 2015), the importance of managed honeybees could become increasingly important as pollinators of A. syriaca, especially for species that depend on the plant for nesting, such as the Monarch butterfly. Milkweed requires nonselfed pollinia transfer for reproductive success, and in natural areas they can be pollinia-limited. For example, Morse & Fritz (1983) determined that over two different study years, only 25%-50% of A. syriaca pods matured compared to what would be expected had there been adequate nonselfed pollinia provided by pollinators. Prioritization of wild bees such as bumblebees to sustain native plants in natural areas is thus essential when evaluating plans for conservation of plant-pollinator mutualisms and diverse pollination networks. Promoting honeybees to pollinate native milkweed will also lead to potentially lessened pollination service as many of the pollinia are wasted because pollen packets detach and accumulate at the entrance to honeybee hives (Willson & Bertin, 1979). Nevertheless, Identifying honeybees as primary pollinators of A. syriaca where bumblebees are less abundant indicates potential short-term solutions to ensure these plants are effectively pollinated to support other native insect species of conservation concern (e.g., Monarch butterflies).
However, this strategy could have unintended consequences for wild bee diversity through enhanced competition for other (nonmilkweed) flowering plants (Cane & Tepedino, 2017).