Assessing Ecological Impacts of Cultivated Hybrids of Milkweed (Asclepias) Open Access

There are questions as to what extent new cultivated lines of native plants should be assessed for ecological consequences through their use in managed landscapes. Cultivated lines of native plants add complexity for breeding projects, as both ecological and ornamental value could be considered. Here we provide an example of an ecological trial by testing new interspecific hybrids of milkweed for their ecological impacts, prior to market release. We investigated the reproductive biology of the hybrids to understand the likelihood of outcrossing of hybrid genes into natural populations of A. tuberosa. Further, we investigated the likelihood that the interspecific hybrids will establish in natural environments and compete with natural populations. We then evaluated the ecosystem services for pollinating species, as selecting for horticulturally preferable traits could inadvertently select away from traits contributing to the coadapted plant-pollinator relationship. We suggest that new cultivars of native species should be shown to support diverse groups of pollinator species, with minimal loss of floral rewards, and with minimal likelihood of introgression, outcrossing, and establishment into natural populations.

There has been increased interest in the use of native plants in urban landscapes, due to the ability of native plants to provide ecosystem services (Burghardt et al. 2009, Campbell et al. 2017, Isaacs et al. 2009, Narango et al. 2017). Native plants currently are underrepresented in horticulture markets due to insufficient propagation protocols, as well as many native plants not having undergone breeding efforts to comply with market trends, despite having ecological value (Wilde et al. 2015). This has led to increased breeding efforts for native plants with ornamental value (Ault 2003, Lewis et al. 2021). Breeding of native plants provides particular challenges for plant breeders, as both horticultural interests and ecological impacts should be considered (Kramer et al. 2019, Wilde et al. 2015).

In the case of the use of native plants to support pollinator species in urban landscapes, sudden changes to plant traits for increased ornamental value, could alter the ability for these cultivars to provide ecosystem services, as artificial trait selection may not coincide with the coadapted relationship between pollinators and plants (Erickson et al. 2022, Garbuzov and Ratnieks 2014, Rollings and Goulson 2019). Prior studies have compared cultivated plants, selected for horticulturally-preferable traits, to either their wild-type counterpart or related species to test how selection for horticultural traits affects the ability for cultivars to provide ecosystem services to pollinators (Hoadley 2020, Mach and Potter 2018, Baker and Baker 1975, Ricker et al. 2019). These studies found many cases where cultivars were adequate to support pollinator species. However, there also were cases where a reduction in floral rewards and subsequent reduction in pollinator visitation was associated with significant changes in floral morphology, especially in double-blooming varieties where reproductive floral parts are converted to petals, reducing or eliminating pollen rewards and/or reducing nectar rewards (Hoadley 2020, Mach and Potter 2018).

Regarding conservation of biodiversity, gene flow between cultivars and natural populations could reduce the genetic diversity of natural populations (Gámez-Virués et al. 2015, Olden and Rooney 2006). As cultivars of native plants are often produced asexually at scale, cultivated lines of native plants are often genetically uniform. Repeated outcrossing of genetically uniform plant stock could inhibit adaptation if cultivars were inadvertently selected away from fitness optima found under natural conditions (Frankham et al. 2011, Slatkin 1987, Wilde et al. 2015). Cultivars of native plants should therefore be assessed for their outcross likelihood into natural populations and in some cases, efforts should be made to reduce fertility, leaning toward sterility (Lattier and Contreras 2022), to reduce the ecological impact of cultivars of native plants. This may be especially important for interspecific hybrids, as outcrossing of hybrids into natural populations could introduce genes from related, but non-sympatric species. These examples demonstrate additional considerations that accompany breeding efforts for cultivated lines of native plants, compared to traditional breeding efforts dedicated purely to ornamental horticulture.

Cultivars of native species provide a market-driven means to increase ecosystem services in urban landscapes. However, they can be associated with positive and/or negative consequences. For this reason, ecological impacts should be considered prior to market release, not after new cultivars have been released and are widespread in urban landscapes. While we recognize that not every ecological impact can be assessed, as the scale is too grand, we advocate for increased ecological awareness regarding the release of cultivars of native plants. Here we assess the ecological impact of four interspecific hybrids of milkweed (Asclepias tuberosa × hirtella, Asclepias tuberosa × syriaca, Asclepias tuberosa × speciosa, Asclepias tuberosa × purpurascens), developed by Lewis (2021), prior to market release.

First, we assessed the reproductive biology of the new interspecific hybrids to understand the likelihood of genetic outcrossing into natural populations. This is done by performing one-way controlled crosses of the interspecific hybrids to the maternal parent, A. tuberosa. Further, controlled crosses were performed to see if the interspecific hybrids are capable of self-fertilization or are capable of crossing with their full-sibling. Self-compatibility and full-sibling-compatibility conventionally are important for breeding purposes and for creating subsequent generations for selection of horticulturally preferable traits. However, additional information about the fertility of the interspecific hybrids is useful in understanding the likelihood that the interspecific hybrids could establish in natural environments and compete with naturally-occurring populations of milkweed.

We were also interested in understanding whether the new interspecific hybrids of milkweed provide similar ecosystem services to pollinator species as the maternal parent, A. tuberosa. As the new interspecific hybrids were selected for horticulturally-favorable traits, we were unsure if any traits had been inadvertently selected away from those that contribute to coadapted plant-pollinator relationships. To investigate this, we first compared nectar properties among the interspecific hybrids and A. tuberosa. Further, we compared ecosystem services provided by Asclepias tuberosa × hirtella, Asclepias tuberosa × syriaca and A. tuberosa by collecting pollinating insects on the flowers to investigate if ecological function is conserved in the hybrids.

Four populations of interspecific hybrids of Asclepias, developed by Lewis et al. (2021), were used for this study. Hybrids were selected for novel coloration, stout and branching structure, increased bloom periods and number of blooms, as well as adaptability to commercial growing protocols. Asclepias tuberosa (butterfly weed) (yellow/orange blooms) was the maternal parent for all hybrid lines and served as a reference for the ecological value of the new hybrids. Paternal parents were A. hirtella (green blooms), A. syriaca (pink blooms), A. speciosa (pink/white blooms), and A. purpurascens (pink/purple) blooms) (Lewis et al. 2021). Interspecific hybrid populations therefor consisted of A. tuberosa × hirtella, A. tuberosa × syriaca, A. tuberosa × speciosa, and A. tuberosa × purpurascens. Interspecific hybrid blooms expressed floral colors that were intermediate values of their respective parents (Lewis 2021) (Fig. 1). A. tuberosa (butterfly weed) is common in native nursery production throughout the southeastern US and provides nectar rewards for both hymenopteran and lepidopteran pollinator species (Fishbein and Venable 1996, Kephart 1983, Stoepler et al. 2012). All paternal parents are native to certain locations within the United States—A. speciosa being found in the northwestern U.S., A. hirtella the central U.S., A. purpurascens in the eastern U.S., and A. syriaca native to the eastern U.S. but having naturalized throughout much of the U.S.— and part of the temperate North American clade of milkweed (Fishbein et al. 2011, Fishbein et al. 2018).

Growing regiments of new interspecific hybrids and A. tuberosa mimicked those of Lewis et al. (2021) for growing Asclepias under greenhouse conditions. All species were grown at the College Station Greenhouse Complex (latitude: 33.9480 °N, longitude: 83.3773 °W) at the University of Georgia. Plant stock was three to four years old, with the exception of some A. tuberosa stock that was two years old. Species were grown in 4 L (1 gal) containers (Classics 400; Nursery Supply, Agawam, MA) with a substrate mix of 80% 1 cm (3/8 in) pine bark and 20% milled peat (Foothills Compost, Gainesville, GA). Species were given supplemental light using LED lighting (Fluence Spyder, Fluence Technologies Inc. Austin, Tx) with a photoperiod of 14 hours and a daily light integral (DLI) of 20 mol·m⁻²·d⁻¹. Photoperiod and DLI were maintained using an inhouse lighting control system to adjust supplemental light levels throughout the day. The lighting control system works by continually monitoring light levels in the greenhouse using ePAR light sensors (SQ-610-SS Series, 400-750 nm) (Apogee Instruments, Inc., Logan, UT). Light readings are taken by a data logger (CR1000, Campbell Scientific, Inc., Logan, UT) that calculates the appropriate amount of supplemental light to be provided by the LED fixtures. The datalogger then sent a voltage signal (SDM-AO4A, Campbell Scientific, Inc., Logan, UT) to control the LED fixtures light output. Daytime temperatures had a set point of 25 C (77 F) and night time temperatures had a set point of 18 C (64 F). Plants were fertigated biweekly using a 20N–4.4P–16.6K water-soluble fertilizer (Peter’s 20-20-20, Scotts Co., Marysville, OH) at a concentration of 100 ppm N. Biweekly fertigation was used as Asclepias is sensitive to over-fertilization.

To investigate the reproductive biology of the interspecific hybrids, three types of controlled crosses were made. Controlled crosses refer to the manual insertion of a pollinium, the specialized pollen structure of Asclepias, into the stigmatic slit of another flower to attempt to induce fertilization (Wyatt 1976, Lewis et al. 2021). The three types of controlled crosses investigated were self-compatibility, full-sibling-compatibility, and backcross-compatibility of the hybrid populations. Self-compatibility was tested by inserting a pollinium from a single individual into the stigmatic slit of the same individual. Full-sibling-compatibility was trialed by conducting crosses between two individuals derived from the same seed pod. As pollen is packaged into pollinia, all ovules of a single ovary are fertilized by pollen from a single individual. Therefore, all seeds from a single seed pod are full-siblings, sharing the same maternal and paternal individual (Wyatt and Broyles 1994). Backcross-compatibility was tested by performing crosses with the interspecific hybrids as the paternal species and A. tuberosa as the maternal species.

Controlled crosses were conducted under greenhouse conditions to better control plant bloom cycles and maintain plant health (Queller 1985, Wilson and Price 1980, Wyatt and Broyles 1994). Crosses were performed from 2019 to 2022. Hybrids used to attempt self-compatibility and full-sibling-crosses were A. tuberosa × hirtella, A. tuberosa × syriaca, A. tuberosa × speciosa, and A. tuberosa × purpurascens (Fig. 1). To assess likelihood of backcrossing, A. tuberosa × hirtella, A. tuberosa × syriaca, A. tuberosa × speciosa, and A. tuberosa × purpurascens were backcrossed to the maternal species, A. tuberosa. Possible crosses that could be made at any time were limited to what species or siblings were concurrently blooming. When species overlapped extensively in bloom time, additional crosses were conducted. Crossing methods were derived from those described by Lewis et al. (2021) and Wyatt (1976). A pollinium from the paternal parent was removed using forceps, rotated perpendicular to the stigmatic slit of the maternal parent and inserted, preferentially with the convex outer side of the pollinia inserted into the stigmatic chamber as to minimize the distance from the terminal end of the pollinium to the stigma and ensuring the correct orientation of pollen tube germination. A total of 401 intraspecific crosses were conducted between individuals of A. tuberosa to verify whether this method could yield seed pods from known fertile plants.

Two garden sites were established at the University of Georgia’s Durham Horticulture Farm (33.8869539 °N, 83.4205012 °W) in the spring of 2021, with the two sites located on the furthest northern and southern edges of the farm, approximately 600 m (0.37 miles) apart. Both sites are located in open farmland, surrounded by other agricultural land and sparse, forested landscapes. Each site was designed to have all plant types (A. tuberosa and the four hybrids) represented with three plants of a single plant type included as one planting. Plantings were arranged following a randomized complete block design with one replicate at each site. Plantings were 1 m (3 ft) by 1 m (3 ft) in size with 2 m (6 ft) space between each planting. Black nursery fabric was laid between the plantings for weed control and ease of work. At each planting site, three plants of a single plant type were planted in spring of 2021 from nursery stock grown at the College Station greenhouse. Plant stock was in 4 L (1 gal) pots at the time of planting and plantings were top-dressed with pine bark mulch (The Stone Store, Watkinsville, Ga). Plants were watered twice a week during the first three months of establishment using micro-sprinklers, then by hand as necessary during any period of drought. In the spring of 2022 all plants that did not establish over the winter were replaced with plants from the greenhouse.

Prior to assessing ecosystem services to pollinator species under field conditions, gas chromatography was used to assess nectar composition and concentration of sugars so as to investigate deviations in nectar properties between the interspecific hybrids and the wild-type maternal parent, A. tuberosa. Nectar was sampled from greenhouse stock. Four individuals were selected from each interspecific hybrid line as well as from the maternal parent, A. tuberosa. Due to the small floral structures present in A. tuberosa × hirtella, only two individuals produced enough nectar for extraction and processing. All plants were sampled on the same morning at 0700 hr to reduce abiotic influences that effect nectar concentration through the day. Nectar was extracted by a combination of micropipettes and scraping nectar into a 1ml Eppendorf centrifuge vial using a curette. Approximately 50 μl of nectar was extracted from each plant. Flowers were sampled across the plant, preferentially sampling nectar from flowers in peak bloom. Samples were taken for immediate processing to be analyzed using gas chromatography. Methods were derived from Chapman and Hovat (1990) and Beshir et al. (2017). Each sample was divided into two 20 μl technical replicates. Samples were then mixed with 80 μl of distilled water and added to 900 μl of an internal standard (IS) consisting of 100% methanol and 0.125 mg·ml⁻¹ phenyl-β-D-glucopyranoside and vortexed until thoroughly mixed. Samples were then centrifuged at 6,000 rpm for 20 minutes. A 100 μl sample of the solution was then transferred to a gas chromatography (GC) vial with an added liner. The GC vial was placed in a vaporizer and nitrogen gas was sent through the vial for 10 minutes at 45 C (113 F). If all liquid was not vaporized, the solution was rerun for two minutes intervals until no liquid remained. A 50 μl quantity of methoxyimino hydrochloride was added to the GC vial and covered immediately with micropore tape. The GC vial was then heated for 30 minutes at 50 C (122 F). Micropore tape was then removed and 100 μl MSTFA+1%TMCS (N-Methyl-N-(trimethylsilyl)trifluoroacetamide with 1% trimethylchlorosilane) was added and a fitted cap was attached to the GC vial and reheated for 30 minutes at 50 C. Samples were then run through a gas chromatograph (Shimadzu GC-2014, Shimadzu Corporation, Kyoto, Japan) to determine quantities of sucrose, glucose, and fructose.

To assess ecosystem services to pollinator species, pollinators were collected from interspecific hybrids and A. tuberosa from May to July of 2022. Both locations were sampled up to three times per week, always leaving at least a full day between collections. Collections were conducted preferentially on sunny days and never on rainy days. All treatments that were in bloom on the day of collection were sampled. Collections occurred between 1200 hr and 1400 hr (Packer and Darla-West 2021). A single collection consisted of five minutes of collecting pollinators with a sweep net, where any floral visitors that landed on an inflorescence within the planting was collected for the duration of the collection. Immediately following the sweep net collection, another five minutes of collection was conducted using an aspirator to catch any smaller insects that were not captured during the initial collection—as many small pollinator species are not provoked to fly in response to disturbances from a sweep net. Captured insects were then transferred into a 4L (1 gal) Ziploc bag and placed into an ice chest and transferred as soon as possible to a freezer. Only hymenopteran and lepidopteran visitors were included in the analysis, as they comprise the two major groups of pollinating species in the southeastern United States and are a common topic of pollinator conservation for urban landscapes (Tepedino 1979). Hymenoptera and Lepidoptera were identified using a combination of print and online keys (Mitchell 1960, Ascher and Pickering 2017) to the lowest possible taxonomic resolution (Table 1). Insect abundance, richness, and diversity data was assigned to each collection according to genus level resolution. Abundance is the total number of pollinators collected during the collection period. Richness refers to the total number of genera represented in the collection period. Diversity was calculated using the Shannon-Weiner diversity index, which is a measurement of the number of genera in a community and their relative abundance (Nolan and Callahan 2006).

When a planting was sampled for pollinator visitation, a photo was taken of the planting using an iPhone 11 mounted 178 cm (70 in) above the ground. A ruler was placed on a tripod and adjusted to floral canopy height to provide a scale in each image. Images were processed by creating a macro in ImageJ to quantify total floral area in the planting. (Schneider et al. 2012). Macros are dedicated programs that run a series of commands in ImageJ (Schneider et al. 2012). The macro first uses the ‘Brightness/Contrast’ and ‘Color Threshold’ function to distinguish pixels that fall into color ranges represented by flowers—pink and orange flowers were mainly represented within our tested genotypes of milkweed. The ‘Make Binary’ function then creates a binary image by converting all pixels in the image to either white, if the pixel is within a designated range of color values, or black, if the pixel is not within the designated range of color values. To reduce any unintended noise that was generated by the background of the image, a box was manually drawn around the area of the image that was represented by the plant. the ‘Analyze Particle’ function was used to sum all white pixels; all pixels that represent floral area in the image. A scale was set manually for each image by measuring the 30 cm ruler using the ‘Measure’ and ‘Set Scale…” function to convert pixel number to square centimeters so the output is represented in square centimeter of floral area.

We used an analysis of variance (ANOVA) when comparing nectar compositions and concentrations using R statistical software version 4.3. Diversity (Shannon-Weiner) was calculated using the vegan package in R (Oksanen et al. 2017). Abundance, richness, and diversity were assessed using a stepwise regression model including the effects of floral area, genotype, site, and date and the floral area × genotype interaction. Only factors with P < 0.05 were left in the final model. Statistical analysis for the stepwise regression was completed using JMP version 16.2.

The 401 intraspecific controlled crosses of A. tuberosa used to verify the crossing method yielded 33 seed pods, a success rate of 8.2% (Table 2). This indicates that our method to conduct crosses can result in successful fertilization and seed set in milkweed. This success rate is comparable to Lewis et al. (2021), who had a success rate of 11.7%. For the three cross types to assess self-compatibility, full-sibling-compatibility, and backcross compatibility, 2,634, 644, and 1602 crosses were conducted respectively. None of these 4,880 crosses performed across the four interspecific hybrids, between the three cross types, resulted in seed set (Table 2). The number of crosses was limited by what species were blooming concurrently as well as time limitations; to perform one cross requires about 1 minute (Lewis et al. 2021). This resulted in some variability in the number of crosses for each respective category. However, that there was not a single occurrence of seed set from 4,880 crosses suggests incompatibility across the three cross types.

Our observation that none of the new interspecific hybrids of Asclepias successfully set seed, despite nearly 5,000 crosses, gives confidence that the ecological impact on the genetics of the native A. tuberosa will be minimal. Specifically, our finding that 1,602 attempts to cross the interspecific hybrids to the maternal parent, A. tuberosa, resulted in no successful fertilization gives confidence that introgression of the hybrids into surrounding natural populations of A. tuberosa will not ensue. Hybridization is known to cause complications in reproductive biology, reduce fertility, or induce hybrid sterility (López-Caamal and Tovar-Sánchez 2014, Rieseberg and Carney 1998). Under natural conditions, sympatric hybrids of Asclepias have been shown to be capable of backcrossing to parent species, contributing to introgression of hybrid genes to parental species (Broyles 2002, Kephart et al. 1988, Wyatt and Broyles 1992). However, studies of artificially-derived interspecific hybrids of milkweed species have shown hybrid sterility or reduced hybrid fertility (Klips and Culley 2004, Wyatt and Broyles 1994). The interspecific hybrids developed by Lewis et al. (2021) seem to follow the latter and exhibit levels of reduced fertility or sterility, inhibiting these hybrids from backcrossing to A. tuberosa.

Further, self-compatibility and full-sibling-compatibility were tested with no successful fertilization and subsequent seed set. Self-fertilization and full-sibling crosses are generally used to continue breeding efforts as to explore the genome in subsequent generations for masked recessive traits and continue to select for horticulturally-preferable traits. That these hybrids cannot self-fertilize and will not cross with their full-sibling means that from a breeder’s perspective, these plants cannot undergo further breeding efforts using traditional crosses. This information has implications for ecological impacts as well. If these new interspecific hybrids enter horticulture markets, it is of concern that the hybrids themselves may establish in natural environments and compete for niche space in natural areas with naturally occurring populations of milkweed. An introduction of interspecific hybrids developed from Lewis et al. (2021) in urban landscapes would likely be genetically uniform or closely-related individuals as nursery stock is typically vegetative propagated. Conspecific pollen transfer in urban spaces would likely occur between individuals from the same genotype or closely-related individuals. While Asclepias is generally considered self-incompatible, A. tuberosa as well as A. syriaca have shown low, but present, levels of self-compatibility (Kahn and Morse 1991, Wyatt 1976). Self-compatibility and full-sibling compatibility do not seem to be conserved in the new interspecific hybrids. This gives confidence that these interspecific hybrids will not establish in natural areas.

Under open pollination in field and nursery settings, we have observed the formation of seed pods on interspecific hybrids. This is likely through conspecific pollen transfer of less related (non-sibling) individuals of the same hybrid populations, where the barriers for full-sibling fertilization of self-pollination are not present. These seedpods, however, aborted prior to full development. This late-acting post-zygotic abortion is similarly seen in attempts to create interspecific hybrids of Asclepias (Wyatt and Broyles 1994), but we are unsure why. Our objective was to investigate the ecological impact of these new interspecific hybrids of Asclepias. For the new interspecific hybrids to not be able to successfully set seed with A. tuberosa, self-fertilize, or cross with their full-siblings demonstrates the low levels of ecological impact to natural populations of milkweed, should these hybrids be released in urban landscapes.

Nectar sugar concentration (F_4,13 = 1.42, P = 0.3) and composition (F_4,13 = 1.43, P = 0.3) were similar among the interspecific hybrids and A. tuberosa (Fig. 2). Nectar sugar composition—the relative proportion of sucrose, glucose, and fructose—was predominantly sucrose in all hybrids and A. tuberosa, making up no less than 98% of the sugar profile. This is consistent with prior assessments of Asclepias nectar (Southwick 1983, Wyatt et al. 1992) and it is not surprising that interspecific hybrids derived from species whose nectar sugar composition is primarily sucrose would also have a predominate sucrose profile. Nectar sugar composition affects preference for certain pollinator group, as well as influences nectar properties, such as viscosity (Baker and Baker 1975). Nectar concentration, the total quantity of sucrose, glucose, and fructose per volume nectar, is more variable and can depend on environmental conditions, plant health, and time of day (Southwick 1983, Wyatt et al. 1992). Nectar sugar concentrations ranged from 640 to 924 g^.L⁻¹ (Fig. 2). Variability in nectar concentrations was seen within the hybrid genotypes and A. tuberosa, but not among the different hybrid genotypes and A. tuberosa. Nectar concentration influences pollinator preference, as it affects sugar uptake rates for pollinators and can prevent certain groups of pollinators from successfully gaining a nectar reward (May 1985). Butterflies, for example, use a proboscis to uptake nectar and thus cannot take up as high concentrations of nectar as a bee, which uses its tongue (May, 1985, Willmer 2011). We attempted to determine nectar volume using the greenhouse plants, as total nectar volume can influence pollinator preference and foraging time, the time a pollinator will spend interaction with a flower. (Parachnowitsch et al. 2019, Willmer 2011). However, our sampling method did not yield usable data.

Flowering is an energy-intensive process. As hybrids were selected for increased bloom number, a total increase in floral production could reduce the available resources for nectar rewards. Similarity, the nectar concentration data shows both that there is not a reduction in total sugars found in nectar due to increased floral displays, as well as that the new interspecific hybrids should not exclude pollinator groups that previously had access to Asclepias floral rewards. Similarly, as nectar composition is similar among hybrids and A. tuberosa, preference of pollinators when foraging should not differ due to the nectar sugar profile.

Pollinator visitors during summer 2022 trials totaled 508 across 55, ten-minute, sampling periods. Samples comprised of 11 genera of Lepidoptera and 31 species (16 genera) of Hymenoptera (Table 1). While moths are generally considered nocturnal pollinators, we found cases where moths appeared to be actively pollinating during our collection period, so they were included in our count data. Due to limited collections from A. tuberosa × speciosa and A. tuberosa × purpurascens, they were excluded from the analysis. This left A. tuberosa, A. tuberosa × hirtella and A. tuberosa × syriaca to be analyzed. Of the 513 pollinators collected from the three genotypes, 150 pollinators were collected from A. tuberosa from 18 collection periods, 175 pollinators were collected from A. tuberosa × hirtella from 16 collection periods, and 188 pollinators were collected from A. tuberosa × syriaca from 21 collection periods. The most abundant genus collected from A. tuberosa and A. tuberosa × hirtella was Lasioglossum (sweat bees) with a relative abundance of 39% and 41%, respectively, followed by Apis (honeybees) with a relative abundance of 25% and 30% respectively. For A. tuberosa × syriaca, the most abundant visitor was Apis, consisting of 52% of the collection followed by Lasioglossum at 15% (Table 1). It is important to note that there are multiple bee hives located near this study site. Honeybees could therefore be overrepresented in the pollinator community, compared to more natural or habitats without managed honeybee hives.

Multiple regression was used to compare abundance, richness and diversity among hybrids and A. tuberosa. There was no difference in abundance among both hybrid genotypes, A. tuberosa × hirtella and A. tuberosa × syriaca. However, pollinators were more abundant on hybrids than on A. tuberosa (P = 0.0285); the two hybrids on average having 1.3 more visitors that A. tuberosa. Pollinator abundance increased with increasing floral area for A. tuberosa, A. tuberosa × hirtella and A. tuberosa × syriaca (P > 0.001), with an additional visitor per 5 cm² increase in floral area (Fig. 3A). It would be expected that insect abundance increases with increased floral area, as the plants can support more pollinators simultaneously. Further, pollinator species forage by recognizing contrasts in floral display against a green vegetative background (Kevan 1972, Willmer 2011). As floral area increases, there is more area contrasting green vegetative backgrounds, which could make the planting easier to recognize when a pollinator is foraging. This too could have contributed to the increase in pollinator abundance with increasing floral area. It is not clear why the hybrids attracted more pollinators than A. tuberosa. The interspecific hybrids of Asclepias were selected for novel coloration. It is possible that this increased the ability for pollinators to distinguish flowers from vegetative backgrounds.

Richness increased with floral area (Fig. 3B), with an additional genus represented for each 313 cm² (123 in²) increase in floral area. Richness was similar among A. tuberosa and both hybrids. Increased floral area may reduce competition for floral resources among foraging pollinators. This could allow for more genera of pollinator species to simultaneously interact with the inflorescences.

Diversity was not affected by floral area and was similar among the hybrids and A. tuberosa (Fig. 3C). Similarity in richness and diversity among the hybrids and A. tuberosa demonstrates that the hybrid genotypes can support a diverse group of pollinator visitors, similar to A. tuberosa. Selection for horticulturally-preferable traits in this case does not hinder the ability for the hybrids to support diverse pollinators.

There is still the question as to what extent new cultivars of native plants should be assessed for unintended ecological consequences. This poses a difficult challenge for the intersection of horticultural interest and ecological value, as there are market incentives to get new introductions to market as efficiently as possible. This is not conducive to account for environmental consequences associated with new introductions. With rising interest in native plants for gardens and ecologically-considerate landscapes, the horticulture industry could increase efforts to assess ecological impacts of new introductions prior to market release.

This project encompassed two years of research regarding the ecological impacts of new interspecific hybrids of Asclepias intended for urban landscapes. Genetic outcrossing from the new hybrids is unlikely, due to inability for the hybrid genotypes to backcross to the maternal parent, A. tuberosa. Spread through natural areas is unlikely due to lack of seed set for crosses from self-fertilization or crosses between full-siblings.

Similar nectar properties were seen among hybrids and A. tuberosa for nectar composition and concentration. Further, pollinator diversity and richness were similar among the tested hybrids and A. tuberosa. Pollinator abundance was higher for hybrids than for A. tuberosa, indicating that hybrids would be able to provide ecosystem services to at least a similar degree as A. tuberosa, if introduced into horticultural markets. Is this enough to recommend these new selections as viable introductions as we demonstrated minimal impact to native genetics and possible positive impacts for pollinator communities? We investigated topics regarding conservation of genetics of natural populations and ecosystem services for pollinator species. However, with two years of investigating the topic we were unable to breach the subject of specialist interactions with the interspecific hybrids. The monarch butterfly and ten other insect species have obligate relationships with milkweed (Agrawal 2017) and those interactions are important, but were not studied. This raises the question as to what is viable for the horticulture industry when considering ecological impacts? We suggest that new cultivars of native species should be shown to support diverse groups of pollinator species, with minimal loss of floral rewards, and with minimal likelihood of introgression, outcrossing, and establishment into natural populations.