Introduction

Monarch butterflies (Danaus plexippus L.) migrate annually from overwintering sites in the oyamel fir forests of central Mexico to broad regions across North America, mostly east of the Rocky Mountains, a migratory cycle typically requiring four generations (Malcolm and Zalucki 1993; Agrawal 2017). Monarch larvae feed exclusively on milkweeds (family Apocyaceae, subfamily Asclepiadoideae), including true milkweeds in the genus Asclepias (Agrawal and Fishbein 2006) and their close relatives; e.g., Cynanchum laeve (Michx.) Pers. (Bartholomew and Yeargan 2001; Yeargan and Allard 2005). The eastern migratory population of monarchs has declined by > 80% since systematic censuses of numbers of overwintering adults began in the 1990s, falling to the lowest level ever recorded in winter 2013–2014 (Brower et al. 2012; Rendón-Salinas et al. 2015). Concerns about its long term viability have mobilized scientists, federal and state agencies, non-governmental organizations, and private citizens into actions to safeguard and restore monarch populations (Pollinator Health Task Force 2015; Gustafsson et al. 2015; Monarch Joint Venture 2018).

Although surveys suggest that milkweed populations have been relatively stable in more natural and semi-natural areas (Hartzler 2010; Pleasants and Oberhauser 2013; Zaya et al. 2017), when croplands and loss of natural habitat to urbanization are considered, there has been substantial loss of milkweeds in the monarch flyways (Pleasants et al. 2017; Zaya et al. 2017). Despite some scientists’ questioning of a causal link between milkweed loss and monarch decline (Davis and Dyer 2015; Dyer and Forister 2016; Inamine et al. 2016; Agrawal 2017), the milkweed-limitation hypothesis has gained traction because it suggests a plausible strategy by which diverse stakeholders can work together in actionable science (Palmer 2012; Gustafsson et al. 2015) to help conserve the monarch and its migration. Planting milkweeds on public and private lands has emerged as a central conservation strategy (Thogmartin et al. 2017; Monarch Joint Venture 2018; National Pollinator Garden Network 2018; US Fish and Wildlife Service 2018).

Public interest in monarch conservation is reflected in the more than 18,600 Monarch Waystation habitats (managed gardens containing milkweeds and nectar plants) that have been registered with MonarchWatch as of January 2018 (MonarchWatch 2017a), and the countless other similar gardens that have been planted in residential landscapes, at schools, businesses, parks, zoos, golf courses, nature centers, and other public and private places. Irrespective of the ecological value for monarch populations, pollinator gardening provides opportunities to engage large numbers of citizens in reconciliation ecology (Rosenzweig 2003; Colding et al. 2006; Lundholm and Richardson 2010), which in turn can foster a deeper interest in nature conservation (Miller 2005; Goddard et al. 2010; Bellamy et al. 2017).

Natural stands of milkweeds are generally scarce in residential areas (Cutting and Tallamy 2015), so it seems intuitive that planting milkweed in urban or suburban butterfly gardens will attract monarch adults to oviposit. That assumption, which previously was supported mainly by observational data, was validated in experiments that compared monarch colonization and survival in small plots of common milkweed, Asclepias syriaca L., planted in managed landscapes in residential neighborhoods and equivalent plots planted in minimally managed native meadows (Cutting and Tallamy 2015). In that study, oviposition was significantly higher on plants in residential settings than in natural areas, with no difference in subadult survival between the two types of habitats.

Milkweed species vary in growth form, height, leaf shape and size, floral morphology and bloom time, and extent to which they spread vegetatively via rhizomes (Woodson 1954; Borders and Lee-Mäder 2015), so some species may be better suited than others for use in garden-type settings. Asclepias syriaca is the most important host for monarchs in their summer breeding range within eastern North America (Malcolm and Zalucki 1993; Flockhart et al. 2013, 2015), and nearly all habitat restoration models and recommendations are based on that species (Thogmartin et al. 2017; Pleasants 2017). However, because of its height (up to 2 m) and propensity to spread, A. syriaca may be horticulturally less suitable than some other native milkweeds for managed gardens that, in addition to supporting monarchs, are designed to be aesthetically attractive while also providing resources for other pollinators.

Previous studies have examined monarch oviposition preference and larval growth and survival in relation to defensive characteristics of the host plant (Agrawal et al. 2015), on closely related species (Yeargan and Allard 2005), and in experimental garden plots differing in botanical composition (either native or exotic plants) and weed management intensity (Majewska et al. 2018). Other studies examined larval growth on excised leaves of different milkweed species, and young plants in a greenhouse (Pocius et al. 2017a, b). Monarch oviposition is influenced by the height, age, and condition of milkweed plants, as well as their spatial dispersion and other factors (Cohen and Brower 1982; Zalucki and Kitching 1982), so usage of different milkweeds in garden settings can not necessarily be inferred from laboratory or greenhouse trials. To date, no published studies have compared monarch colonization and performance on different milkweed species in a replicated, common garden experiment in the field.

Gardening for pollinators is promoted by prominent conservation organizations (National Pollinator Garden Network 2018; National Wildlife Federation 2018; Pollinator Partnership 2018), and many gardeners are interested in growing plants that attract bees as well as butterflies (Garbuzov and Ratnieks 2014a, b). With native bee populations declining in North America and globally due to agricultural intensification and loss and degradation of natural habitats (Biesmeijer et al. 2006; Koh et al. 2016; Potts et al. 2016), urban butterfly gardens can play a role in supporting native bee biodiversity (Hernandez et al. 2009; Baldock et al. 2015; Hall et al. 2017). Milkweed flowers produce abundant nectar and are highly attractive to bees (Borders and Lee-Mäder 2015). Research on bee visitation to milkweeds has focused mainly on determining which types of floral visitors are most effective at extracting and transferring pollinia (Kephart 1983; Betz et al. 1994; Maclvor et al. 2017), as opposed to documenting different milkweed species’ relative attractiveness to bees or differences in the bee assemblages that visit them as a nectar resource in garden settings. Planting milkweed species that attract and sustain bees as well as monarchs could boost the conservation value of gardens at no additional cost.

In this paper, we describe a two-year study comparing suitability of eight species of milkweed for such use in managed gardens. We assessed colonization and usage by wild monarchs over two growing seasons, compared larval performance, and documented abundance of other milkweed specialist insect herbivores. The milkweeds’ extent of tillering, growth characteristics, and bloom periods were evaluated. Finally, we assessed visitation by bees, and composition of bee assemblages associated with six of the eight milkweed species.

Materials and methods

Milkweed characteristics, monarch usage, and other herbivores in replicated gardens

Eight species of milkweed (Asclepias spp.) were selected for evaluation based on their suitability for use in low-maintenance sites, in full sun with limited supplemental irrigation. Five of the milkweeds, A. syriaca L. (common), A. incarnata L. (swamp), A. tuberosa L. (butterfly), A. viridis Walter (green antelopehorn), and A. verticillata L. (whorled), are native to Kentucky, whereas the other three, A. speciosa Torr. (showy), A. fascicularis Decne. (Mexican whorled), and A. latifolia (Torr.) Raf. (broadleaf), are native to the central or western United States (Woodson 1954; USDA 2018). Seed was purchased from Prairie Moon Nursery, Winona MN, cold-moist stratified before planting, and planted in tree pots (3.8 cm diameter, 20 cm deep; Stuewe and Sons, Tangent, OR) containing commercial potting medium (Promix BX, Premier Tech Horticulture, Quakertown, PA) in late February. The seedlings were grown in a greenhouse, fertilized (Osmocote 5-9-12, Scotts, Marysville, OH), and transplanted to replicated garden plots on 16 May 2016, 1 week after the 90% probability of last frost date for Lexington, KY (National Oceanic and Atmospheric Administration 2018).

The main study was conducted at University of Kentucky Arboretum and State Botanical Garden of Kentucky, Lexington, KY (GPS coordinates: 38.0139, − 84.5052). This arboretum was an ideal site for this research because it reflects a typical residential setting consisting of a mixture of ornamental trees, shrubs, and gardens surrounded by low-maintenance turfgrass lawns located within a medium-sized city. Five milkweed gardens, each 1.22 × 9.75 m, were tilled and covered with landscape fabric. We subdivided each garden into eight plots (1.22 × 1.22 m), one for each of the milkweed species that were arranged in a randomized complete block. Four seedlings were planted 0.6 m apart within each plot (five replicates; 20 total plants of each species). Height of the seedlings ranged from 16 to 30 cm at planting. The gardens were covered with shredded hardwood mulch (5 cm depth) and watered to aid plant establishment. The gardens were oriented in an east–west direction and separated from one another by at least 20 m.

We conducted counts of monarch eggs and larvae on all plants in the gardens once every 2 weeks from May to October 2016 and from April to September 2017. In addition, plants were measured for height, bloom presence, and colonization by milkweed specialist herbivores including Aphis nerii Boyer de Fonscolombe (oleander aphid), Oncopeltus fasciatus (Dallas) (large milkweed bug), and Tetraopes spp. (milkweed longhorn beetles) in July and August during each of the growing seasons. For aphids, each plant was rated by two independent observers for the percentage of surface area of leaves and stems infested, using a 1–5 scale with (0 = no infestation, 1 = < 20%, 2 = 21–40%, 3 = 41–60%, 4 = 61–75%, 5 = > 75%). For the other herbivores, actual numbers were counted. In 2017, tiller production was recorded by counting ramets that had escaped from the original garden plots. Those counts were taken in September near the end of the growing season.

Performance of monarch larvae on milkweed

Monarch larval growth and survival on the different milkweed species was compared in two field trials done on plants in the aforementioned garden plots, and in a third trial on young potted plants in the greenhouse. Cohorts of larvae (mostly late first instars, a few newly-molted second instars) were obtained as-needed from Idlewild Butterfly Farm (Louisville, KY) where monarchs are bred only seasonally, rotating with breeder and wild stock that is screened to exclude the protozoan parasite Ophryocystis elektroscirrha (McLaughlin and Myers) (MonarchWatch 2017b), and where only selected individuals from different parents are put into the breeding house. Initial mass of individual larvae averaged about 8 mg, based on collective mass of 20 representative larvae weighed at the start of each trial.

For Field Trial 1, the larvae were caged in white fine mesh bags (25 × 40 cm) on two plants per plot, with one bag per plant and two larvae per bag, using a similar proportion of first and second instars for each plot. The larvae were placed on the plants on 19 August 2016 and left to feed for 9 days, after which we recorded final weight, instar, and survival. By the start of Field Trial 2 (15 September 2016), some plants had begun to senesce. We therefore caged larvae on nine healthy plants of each species distributed across the gardens, using three larvae per bag, and analyzed that trial as a completely randomized design with plants as replicates. For that trial, larval weight, instar, and survival were evaluated after 7 days. The differences in the number of larvae started per plant and duration of the two field trials were determined by larval availability, and by how quickly the survivors attained the 4th or 5th instar at which point they were weighed before they could void their gut contents prior to pupation.

A third trial, conducted in 2017, compared larval growth on the aforementioned milkweed species in the greenhouse under standardized conditions, i.e., without possible variation in shading from neighboring plants, soil moisture, or other factors that might influence plant quality or larval performance in the field. Procedures for growing the milkweeds were as described for the replicated garden study, except that larger (10.1 cm diameter, 36 cm deep) pots were used. The seeds were planted in May 2017, and monarch larvae were placed on the resulting plants on 18–23 August, by which time the milkweeds were 30–50 cm tall, depending on species. Each plant received a single neonate (< 1 day old) caterpillar confined in a fine mesh bag (25 × 40 cm) that covered most of the plant. We used a single larva per plant to ensure that they were not food limited. Asclepias viridis seeds planted for this assay failed to germinate, so that species was dropped from the trial. There were 10 plants (replicates) each of seven milkweed species in a randomized complete block with 1 m between plants to avoid shading. Positions on the greenhouse benches were re-randomized twice per week to reduce site variation. Larvae were allowed to feed for 5 days, after which their instar and weight were assessed. Greenhouse temperatures while the larvae were on the plants ranged from 26 to 28 °C.

Bee assemblages on milkweeds in gardens and at other field sites

Six of the eight milkweed species produced enough blooms in the replicated gardens allow us to assess their extent of bee visitation and bee assemblages in 2017. The exceptions were A. viridis and A. latifolia, which bloomed too sparsely for such sampling to be possible. Relative attractiveness of the six sampled species was compared by 2-min “snapshot” counts (Garbuzov and Ratnieks 2014a) taken twice at each plot during peak bloom (June to early July). Snapshot counts were taken on clear warm days (temperature > 20 °C, wind < 20 km/h), with one count in late morning (1100–1200 h) and another in afternoon (1400–1600 h), to help account for possible diurnal variation. At each visit, we counted the number of bees actively foraging on blooms of the milkweed plants in a given plot, trying to not to count individuals more than once. Counts from the two visits were averaged and plants were assigned a rating where < 5 = low, 5–10 = moderate, and > 10 bees = high, in addition to the mean count per plot.

After snapshot counts were completed, we collected a 30-bee sample from milkweed flowers in each plot that had enough blooms to make such a collection possible. Sampling involved walking from plot to plot during mid-day (1100–1600 h) and knocking the first 30 bees observed on open flowers into plastic containers partially filled with 75% ethanol. It was not possible to collect 30 bees at a single sampling because the numbers of inflorescences per plot were limited, and because bees were disturbed and flew off, so sampling of particular milkweed species was completed over 1–3 successive days, depending on extent of bloom, and throughout the aforementioned time period to capture diurnal variability. The bees were cleaned and prepared for identification according to guidelines in Droege (2015), pinned, and identified to genus using online keys (Packer et al. 2007). Honey bees (Apis mellifera L.), bumble bees (Bombus spp.), and carpenter bees (Xylocopa spp.) were identified to species (Colla et al. 2011).

To assess if bee assemblages visiting milkweeds in the replicated gardens were representative of those associated with milkweeds at other central Kentucky field sites, we collected additional 50-bee samples from natural stands or plantings of A. incarnata (five sites), A. syriaca (four sites), and A. tuberosa (five sites) in parks, golf course naturalized roughs, butterfly gardens, and other locations in or near Lexington. Those samples, collected during peak bloom (16 June to 5 July) in 2016 or 2017, were prepared and identified as described above.

Data analyses

Plant characteristics (tillers, height) and insect abundance in the main garden study were compared among milkweed species by two-way analysis of variance (ANOVA) for a randomized complete block design with mean separation by Fisher’s least significant difference (LSD) test when the overall treatment effect was significant (P < 0.05). Single degree of freedom contrasts were used to further compare monarch abundance between selected sets of milkweeds, e.g., tall versus shorter species, and narrow-leaved versus broad-leaved ones. Log- or square root- transformations were applied in cases where raw data failed to meet the assumptions of parametric statistical tests for normality or homogeneity of variance. Data from larval monarch performance trials were similarly analyzed, except that for second field trial, we used a completely randomized design with individual plants as replicates as described earlier.

Chi square tests for heterogeneity were used to test for differences in proportional representation of different bee taxa in collections from different milkweed species. Bee genus richness and diversity (Simpson 1-D; Magurran 2004) were compared between milkweed species by ANOVA for a randomized complete block (garden data) or completely randomized design (data from sites other than the replicated gardens). Statistical analyses were performed with Statistix 10 (Statistix Analytical Software 2013). Data are reported as original (non-transformed) means ± standard error (SE).

Results

Milkweed characteristics in gardens

The eight milkweed species differed in height, form, and propensity to spread via rhizomes (Table 1). Asclepias fascicularis, in particular, produced numerous above-ground tillers, on average > 100 per plot. Ascelpias verticillata and A. speciosa also produced rhizomes and tillers, the latter species spreading several meters beyond the plot borders. The other milkweed species produced relatively few or no tillers. The milkweeds also varied in height (Table 1), with the taller species (A. syriaca, A. speciosa, A. incarnata, and A. fascicularis) attaining 1–1.7 m height by the second growing season after transplanting. All of the milkweed species except A. latifolia, A. speciosa, and A. syriaca produced blooms in 2016, and all eight species bloomed in the following year. Bloom periods varied from May to August, and the different species varied in their extent of visitation by bees (see below).

Table 1 Growth and bloom parameters, and overall attractiveness to bees, of the eight milkweed species evaluated in the replicated garden plots
Full size table

Monarch usage and abundance of other herbivores on milkweeds in gardens

All of the gardens attracted monarchs, with eggs and larvae found throughout the 2016 and 2017 growing seasons (Fig. 1). In 2016, the first monarch progeny were found in May, within a few weeks after the seedlings had been transplanted. In that first year, colonization of the gardens peaked in July and persisted until October, even after the plants had begun to senesce. In 2017, high numbers of monarchs found in our gardens beginning in April (Fig. 1). Usage by monarchs continued throughout the summer, peaking in August. No eggs or larvae were found past mid-September, reflecting the earlier senescence of the plants in 2017 compared to 2016.

Fig. 1
figure 1

Seasonal abundance of naturally-occurring Danaus plexippus eggs and larvae on milkweeds in the experimental gardens in 2016 and 2017. Counts are means (SE) per garden for all eight milkweed species combined. *Young milkweeds were not transplanted until 16 May 2016, 1 week after the 90% probability frost-free date for Lexington, KY

Full size image

Numbers of monarch progeny found in the garden plots differed significantly between milkweed species in both years (Table 2). The taller species (A. incarnata, A. syriaca, A. speciosa, and A. fascicularis) recruited more monarchs than did the four shorter ones (t = 9.9, 8.6 for 2016 and 2017, respectively; P < 0.001; single degree of freedom contrasts). Milkweeds that were both tall and broad-leaved (A. syriaca and A. speciosa) were colonized more than all other species as a group (t = 6.9, 6.4 for 2016 and 2017, respectively; P < 0.001). In 2016, when A. incarnata, A. syriaca, and A. speciosa were of similar height (Table 1), more eggs and larvae were observed on A. incarnata, but the following summer, when A. syriaca and A. speciosa were taller than A. incarnata, more eggs and larvae were found on the former two species (Table 2). Compared to 2016, A. tuberosa recruited relatively more eggs and larvae in 2017, possibly reflecting their similar size to the other milkweed species during the monarchs’ early arrival in April 2017. In total, we found 474 naturally-occurring monarch eggs and larvae on milkweeds in the gardens over the two growing seasons.

Table 2 Abundance of Danaus plexippus eggs and larvae and other specialist herbivores found in biweekly inspections of milkweed plots in five replicated gardens during the 2016 and 2017 growing seasons
Full size table

Aphid (A. nerii) populations also differed significantly among milkweed species (Table 2). Asclepias incarnata and A. latifolia supported relatively high infestations of aphids in both years, whereas A. fascicularis had relatively few. On A. incarnata, which has relatively narrow leaves, most of the aphids were on stems and petioles. Aphids on A. latifolia, which has broad leaves, were mainly on abaxial leaf surfaces. Large milkweed bugs were found on all milkweed species but were particularly abundant on A. syriaca, A. tuberosa, and A. fascicularis, which had pods throughout much of the growing season. Milkweed longhorn beetles tended to be found mostly on A. speciosa and A. fascicularis (Table 2).

Performance of monarch larvae on milkweeds

Monarch larvae survived and developed on all milkweed species (Table 3). In Field Trial 1, there was no difference in survival, but the final weight and instar attained differed significantly between the milkweed species, with relatively large size and faster development on A. verticillata, A. tuberosa, and A. speciosa, and slower growth on A. fascicularis. Growth and survival were similar on all milkweed species in the other two trials (Table 3). Average daily temperatures ranged from 20.6 to 27.8 °C (mean, 25 °C) during Field Trial 1 and from 23.3 to 27.2 °C (mean, 24.4 °C) during Field Trial 2 (Kentucky Climate Data; http://www.agwx.ca.uky.edu/cgi-bin/ky_clim_data_www.pl).

Table 3 Performance (weight and instar attained; percentage survival) of cohorts of first or early second instar Danaus plexippus confined on living plants of eight milkweed species in two field trials in the garden plots, and one greenhouse trial
Full size table

Bee assemblages on milkweeds in gardens and at other field sites

Six of the eight species of milkweeds in our gardens produced enough blooms to compare their overall attractiveness via snapshot counts (Fig. 2). By that measure, A. tuberosa and A. fascicularis were particularly attractive to bees, followed by A. syriaca, A. verticillata, and A. incarnata. Asclepias speciosa attracted relatively few bees. Five families of bees were collected from milkweeds in our gardens (Table 4). The number of bee genera collected from particular milkweed species ranged from six on A. speciosa, to 13 on A. verticillata (Table 4). Bee genus richness (Simpson 1-D) differed significantly between milkweed species (F5,15 = 2.93; P < 0.05) and was significantly higher for A. tubersosa, A. verticillata, and A. fascicularis than for common milkweed, A. syriaca. The brown-belted bumble bee Bombus griseocollis (DeGeer), a common native species, and Apis mellifera, the European or western honey bee, were the most abundant bees sampled from milkweeds in our garden plots. Proportions of bees belonging to different taxa (A. mellifera, Bombus spp., Xylocopa virginica (L.), Megachilidae, Halictidae, and combined other groups) differed significantly among milkweed species (Chi square test for homogeneity; χ2 = 316, df = 25; Fig. 3). Bombus spp. dominated the bee assemblages visiting A. syriaca, A. incarnata and A. tuberosa. Asclepias fascicularis and A. speciosa were particularly attractive to A. mellifera, whereas A. verticillata attracted proportionately more Halictidae and other relatively small bees (Fig. 4).

Fig. 2
figure 2

Relative attractiveness of different milkweeds to bees as measured by two 2-min “snapshot counts” in the late morning and mid-afternoon during each species’ peak bloom, 2017. Means (SE) not topped by the same letter differ significantly (F5,20 = 7.62; LSD, P < 0.005). Snapshot counts were not taken for A. viridis and A. latifolia because they did not sufficiently bloom

Full size image
Table 4 Composition of bee assemblages visiting the six milkweed species in the main gardens that produced enough blooms to make such collections possible
Full size table
Fig. 3
figure 3

Composition of bee samples from the six milkweed species in the main gardens for which data are available. Proportions of different taxa (A. mellifera, Bombus spp., Xylocopa virginica, Megachilidae, Halictidae, and combined other groups) differed significantly among milkweed species (Chi square test for homogeneity; χ2 = 316, df = 25). See text and Table 4 for genera collected, and genus diversity and richness data

Full size image
Fig. 4
figure 4

Composition of bee samples from A. incarnata, A. syriaca, and A. incarnata at urban or peri-urban field sites other than the main experimental gardens based on five sites per milkweed species, and 50–55 bees per site. Proportions of taxa differed significantly among milkweed species (χ2 = 104, df = 10). See text and Table 5 for genera collected, and genus diversity and richness data

Full size image
Table 5 Composition of bee samples collected on A. incarnata, A. syriaca, and A. incarnata at urban or peri-urban field sites other than the main experimental gardens
Full size table

Assemblages of bees collected from A. syriaca, A. incarnata and A. tuberosa at the additional field sites were generally similar to those from the replicated gardens (Table 5). Asclepias tuberosa supported higher genus diversity than did either of the other two milkweeds (F2,12 = 4.36; P < 0.05; Table 5). Proportionate abundance of the different bee taxa differed significantly among those milkweed species (χ2 = 104, df = 10; Fig. 3). Bombus spp. and A. mellifera dominated the samples from A. syriaca and A. tuberosa, whereas A. incarnata attracted a somewhat higher proportion of Halictidae.

Discussion

This study demonstrates that small urban gardens planted with milkweed in Kentucky are readily found and colonized by monarch butterflies. It supports the premise that planting Monarch Waystations (MonarchWatch 2017a) or similar gardens is effective for augmenting monarch habitat in urban settings, and extends knowledge of how gardeners can best deploy milkweeds for conservation value. To our knowledge, this is the first published study comparing usage of different milkweed species by monarchs and bees in a replicated outdoor common garden setting. Milkweeds in our gardens also recruited other specialist insects including aphids, milkweed bugs, and longhorn beetles. Although high densities of those herbivores can sometimes negatively affect seed production and become pests of milkweed crops (Borders and Lee-Mäder 2015), their presence in butterfly gardens is more likely to contribute interest, educational value, and ecological perspective.

Our gardens included eight milkweed species varying in height, growth form, leaf morphology, and propensity to spread by tillering (Borders and Lee-Mäder 2015; Lady Bird Johnson Wildflower Center 2018). All species were successfully established from transplants and regenerated in the second year, and all of them supported monarch larval growth and development. However, based on numbers of eggs and larvae found on the plants, they were not equally colonized by wild monarchs.

Host-finding and oviposition by monarchs are influenced by species, height, age, developmental stage, and condition of milkweed in the field (Cohen and Brower 1982; Zalucki and Kitching 1982; Fischer et al. 2015). Females encountering single-species stands of milkweed tend to lay more eggs on taller plants than on shorter ones (Cohen and Brower 1982; Zalucki and Kitching 1982) which is consistent with our observations of more eggs and larvae on taller milkweed species (A. syriaca, A. incarnata, and A. speciosa) than on relatively shorter-statured ones in both years. Monarchs tend to lay more eggs per plant on isolated plants compared with milkweed in patches, and on plants on the edge of a patch as opposed to ones in a patch center (Zalucki and Kitching 1982). Although the extent to which they use visual cues in host finding is unknown, other specialist butterflies (e.g., swallowtails, Papilio spp.) use search imaging to orient to host plants standing out against background vegetation (Rausher 1978, 1981). Short-statured milkweeds may go unnoticed by butterflies in mixed gardens because they are less apparent than taller milkweeds when surrounded by non-host plants. In a related study, milkweeds that were planted around the perimeter of small, mixed-plant gardens recruited more than twice as many monarchs as did same-sized milkweeds in the garden interior (Baker and Potter unpublished). Female monarchs also tend to lay more eggs on younger plants with newer growth, or on regenerating plants with fresh regrowth of leaves (Zalucki and Kitching 1982; Fischer et al. 2015), but all milkweeds in our gardens were of the same age.

Monarch eggs and larvae were first observed in our bi-weekly inspections in late May 2016, only two weeks after planting, indicating how rapidly the adults can find and utilize small gardens. At that time, the plants were < 30 cm tall. The milkweeds reached their maximum height by late July (Table 1), which in 2016 coincided with peak abundance of eggs and larvae. We continued to find sub-adult life stages in the gardens in September and October after many of the plants had begun to senesce. In 2017, the large number of eggs and larvae found in the gardens in April coincided with the unusually early arrival of northward flying adults observed in many parts of the eastern flyway (Journey North 2017; MonarchWatch 2017c).

North American Asclepias species have evolved a suite of defensive traits that has been strongly implicated in providing protection against herbivores (Agrawal et al. 2009). Milkweed species vary in cardenolide content, extent of latex exudation, and leaf surface traits (trichomes and wax crystals), all of which influence monarch oviposition and larval performance (Cohen and Brower 1982, Malcolm 1992, 1995; Zalucki et al. 1990, 2001; Agrawal et al. 2009, 2015; Agrawal 2017) despite the monarch’s ability to circumvent or attenuate the negative effects of those defenses through vein-cutting (Doussourd and Eisner 1987), target site insensitivity (Holzinger et al. 1992), and sequestration (Malcolm and Brower 1989; Petschenka and Agrawal 2015). Milkweed defenses are dynamic and change with plant age and condition, and in response to herbivory, and they interact with plant height, architecture, and leaf morphology, and with natural enemies, to influence monarch behavior and performance (Agrawal and Fishbein 2006; Rasmann et al. 2009; Agrawal 2017).

Given the complexity of the aforementioned interactions, we can only speculate on the reasons why milkweed species in our gardens were not equally utilized by wild monarchs. All eight supported generally similar larval growth and development, which is consistent with earlier studies (Erickson 1973; Pocius et al. 2017a, b) that also found relatively little difference in larval performance on excised leaves or young potted plants of some of the same milkweed species. Our counts of wild monarch eggs and larvae on milkweeds in gardens also are consistent with Pocius et al. (2018) who found that in field cages, individual females given a four-way choice between young (10–30 cm tall) potted plants laid many more eggs on A. incarnata and A. syriaca than on A. verticillata or on A. tuberosa. Therefore, the difference in abundance of wild eggs and larvae among milkweed species in our gardens more likely reflects ovipositional preference as opposed to nutritional suitability.

Milkweed flowers are long-lived, produce copious amounts of nectar (Wyatt and Broyles 1994), and are highly attractive to native bees, honey bees, butterflies, and other nectar-feeding insects (Fishbein and Venable 1996; Borders and Lee-Mäder 2015). Because milkweed pollen is enclosed within pollinia and is probably inaccessible as food, nectar is the only reward that milkweeds offer to their pollinators (Kephart 1983; Wyatt and Broyles 1994). Large bees in the family Apidae (honey bees, bumble bees, and carpenter bees), and some large wasps, moths, and butterflies are the most effective milkweed pollinators (Willson and Bertin 1979; Willson et al. 1979; Kephart 1983; Betz et al. 1994; Fishbein and Venable 1996; Ivey et al. 2003; Maclvor et al. 2017), whereas most of the smaller visitors are nectar thieves that do not provide pollination services to milkweed. Milkweeds, nevertheless, support a diversity of native bees that pollinate other cultivated and wild plants in urban habitats.

Large-bodied, eusocial Apidae dominated the bee assemblages of A. syriaca, A. incarnata, and A. speciosa in our gardens. Such bees have high energy demands (Heinrich 1976), so they may favor milkweeds having large flowers and abundant nectar rewards. Three other milkweeds, A. tuberosa, A. verticillata, and A. fascicularis, which have relatively small flowers, tended to attract proportionately more relatively small native bees. Asclepias viridis and A. latifolia produced few blooms in the two growing seasons of our study so we were unable to evaluate their extent of bee visitation. Asclepias viridis does attract bees, mainly Apidae (Bombus, Xylocopa, and Anthophora spp.), in natural settings within its native range in the southcentral and southeastern United States (Liaw 1999). Our literature search found no published scientific information concerning bee visitation to A. latifolia. It is possible that those two species take longer to establish and produce flowers, and that given additional growing seasons, they too would help support bees in garden settings.

Across their breeding range, and within different geographic zones and habitats, monarchs may encounter and exploit a variety of milkweed species that differ in chemical and physical characteristics (Malcolm et al. 1992; Zaya et al. 2017). Three of the eight milkweed species evaluated in our gardens are not native to Kentucky. The native ranges of A. latifolia and A. speciosa extend from the central Great Plains to the western United States, whereas A. fascicularis is restricted to the western states (Woodson 1954; USDA 2018). Asclepias speciosa and A. fascicularis grew well in our garden plots, and both attracted and supported monarchs and bees, so if they are not invasive, they may be suitable for use in urban monarch and pollinator conservation plantings in our region. Asclepias latifolia, which bloomed sparsely and attracted relatively few monarchs, may perform better in gardens within its native range than it did in Kentucky. The best milkweeds for use in urban gardens will doubtless vary geographically, so studies such as this one are needed to support recommendations for other regions.

Conclusions and applications

Our findings will help gardeners and land managers to choose the milkweed species that best match their conservation goals. Milkweeds such as A. incarnata, A. tuberosa and A. viridis that “stay put” will integrate well with other plants in managed gardens, whereas tillering species such as A. syriaca, A. speciosa, A. fascicularis and A. verticillata may be less well suited for formal gardens because of their tendency to spread into neighboring plant beds or lawns, but better for filling in larger land areas dedicated to monarch habitat restoration and pollinator conservation. Milkweed species should be screened for potential invasiveness before being used outside of their native ranges.

There is evidence, at least from field cage trials (Pocius et al. 2018), that monarchs may lay more total eggs when several milkweed species are available than when restricted to a single species, so planting several species, including taller and shorter-statured ones, and combining milkweeds that preferred for oviposition by monarchs with those that are particularly attractive to bees, may be a strategy for increasing the conservation value of managed gardens. Where such gardens are located; e.g., whether in urban, peri-urban, or more rural settings, their proximity to trees, buildings, and hardscape, and the spatial configuration of milkweeds and nectar plants within them may also influence usage by monarchs, as well as their mortality risk from natural enemies (e.g., Majewska et al. 2018). Our study confirms that small urban gardens containing milkweeds are readily found and colonized by monarch butterflies, so further research to determine how to optimize their value as part of a larger conservation strategy to save the monarch and its migration is warranted.