Pacific Northwest native plants and native cultivars, part I: pollinator visitation

Abstract

Planting native flora is a popular conservation strategy for pollinators. When searching for native plants, consumers may encounter cultivars of native plants, which can have different phenotypic traits than plants found in wild populations (“wild-type native plants”). Previous research evaluating pollinator visitation to wild-type native plants and native cultivars has yielded mixed results, in terms of whether their visitation rates are similar or distinct. We established a garden experiment in Corvallis, Oregon, to examine pollinator visitation and utilization of Pacific Northwest native plant species and cultivars. Over 3 years, we collected and observed bees (Hymenoptera: Apoidea), butterflies (Lepidoptera: Papilionoidea), and syrphid flies (Diptera: Syrphidae) to understand (i) if plant pairs had different visitation rates, (ii) whether any pollinators were associated with differential visitation, and (iii) if specialist taxa preferred wild types over cultivars. Pollinator visitation rates varied by plant and pollinator groupings, but in comparisons between native plant and cultivar pairs, native plants were preferred 37.2% of the time (n = 29 comparisons), cultivars 7.7% of the time (n = 6), and there was no difference in 55.1% of comparisons (n = 43). Our pollinator community data found native plants had greater observed total pollinator richness (except for 1 tie) and bee richness than cultivars, though predicted richness varied. Specialist bees were collected more often from wild types. Cultivars with high visitation rates were minimally developed selections, as opposed to interspecific hybrids. Our results join a growing body of literature in suggesting wild-type native and minimally developed plants should be emphasized for supporting pollinator fauna.

butterflies, bees, syrphid flies, community composition, ornamental plants

Introduction

Media coverage of global insect decline has garnered public support of pollinators and interest in their conservation. Restoration and supplementation of floral resources is one way to tackle habitat and forage loss, which are considered primary drivers of pollinator population decline (Potts et al. 2010, Vanbergen and the Insect Pollinators Initiative 2013, Goulson et al. 2015, Kerr et al. 2015). Planting for pollinators is an accessible strategy used among gardeners, farmers, universities, and businesses to support pollinating insects in fragmented landscapes. High-quality plantings can support diverse assemblages of pollinators, including bees (Hymenoptera: Apoidea), butterflies (Lepidoptera: Papilionoidea), and syrphid flies (Diptera: Syrphidae; e.g., Di Mauro et al. 2007, Fukase 2016, Garratt et al. 2017, Baldock et al. 2019, Majewska and Altizer 2020, Nabors et al. 2022, and others).

Research on pollinator plantings has suggested an array of approaches to improve their ability to support pollinators, such as increasing floral diversity (Blaauw and Isaacs 2014, Pardee and Philpott 2014, Makinson et al. 2017, Baldock et al. 2019, Lanner et al. 2020, and others), tailoring plantings to specific pollinator taxa (Batáry et al. 2011, Kremen and M’Gonigle 2015, Montgomery et al. 2020), and ensuring late-season forage is available (Pywell et al. 2011, Filipiak 2019). Careful plant selection is an important component of pollinator plantings, as different pollinator species have differential relationships with their host and forage plants (Frankie et al. 2013) and many commercially available plants are unattractive to pollinators (Garbuzov et al. 2017). To conserve pollinators, numerous studies emphasize the use of native flora (e.g., Morandin and Kremen 2013, Pardee and Philpott 2014, Egerer et al. 2019, Lanner et al. 2020, Fukase 2016, Staab et al. 2020, Nabors et al. 2022, Prendergast et al. 2022, Threlfall et al. 2015 and others).

Native plants are relatively specialized products in the US green industry: they represent only 9–13.4% of annual nursery plant sales (Hall et al. 2011, Khachatryan et al. 2020), despite around 43% of industry firms selling some native plant species (Rihn et al. 2022). Native species diversity in the commercial trade represents only 26% of US native vascular plant taxa (White et al. 2018). This is consistent with a survey of mid-Atlantic wholesale nurseries, which found 25% of plant species to be native to the region, though most species were sold in the form of cultivars and/or hybrids of wild-type native taxa (Coombs et al. 2020, but see Zinnen and Matthews 2022). In a survey of Eastern US native plant users, 74% preferred to use local ecotypes, 21% preferred straight species (e.g., no associated cultivar name), and 0.3% preferred cultivars (Tangren et al. 2022). Consumers are interested in and willing to pay premiums for locally sourced plant material (Hooper et al. 2008, Curtis and Cowee 2010, Tangren et al. 2022), yet limited availability and diversity (Hooper 2003, Basey et al. 2015, White et al. 2018) impact the ability of landscapers, gardeners, and researchers to acquire native plants (Brzuszek et al. 2010, Anderson 2019).

There are no criteria that a plant needs to meet to be called native in the US horticultural industry, though origin certification programs do exist (Young 1995, McCormick et al. 2021). Individual definitions of “native” are highly subjective (Kitchen and McArthur 2001), such that both natural and genetically modified cultivars of native plants may be marketed as native (Jones and Young 2005, Wilde et al. 2015, Coombs et al. 2020). Some cultivars are named selections of naturally occurring phenotypic variations (e.g., Phlox paniculata ‘Jeana’, which was found along the Harpeth River in Tennessee; Coombs 2017). These phenotypes may be rare and resulting from chance genetic mutations (e.g., Cercis canadensis ‘Arnold Banner’, Fabales: Fabaceae, a periclinal chimera; Friedman et al. 2023), or they may be common in wild populations, though representative of less dominant phenotypes (e.g., white flower morphs in Camassia species, Asparagales: Asparacaeae; Watson 1884). Other cultivars have been bred for traits not found in wild-type plants (e.g., Coreopsis ‘Show Stopper’, Asterales: Asteraceae, which has magenta inflorescences, rather than the yellow that typifies Coreopsis L. species; Korlipara 2012). To be consistent with the plant material that is available commercially, we refer to plants from here on as “natives” or “wild types” if they are straight species and their phenotype is consistent with those observed in wild populations, and “cultivars” if they have an associated cultivar name.

There is growing concern over the use of native plant cultivars, as consumers and researchers raise questions about their capacity to provide the same ecological benefits as wild-type native plants (Wilde et al. 2015, Kramer et al. 2019, Tangren et al. 2022). Current guidance in native plant breeding emphasizes that selections should be guided by the goal of maintaining adaptive and ecological function (Wilde et al. 2015), yet primary emphases in ornamental breeding include the improvement of novel floral traits, such as color, form, size, and flower number (Ault 2003, Heywood 2003, De 2017). While plants with ornamental traits are capable of supporting pollinator communities (Mach and Potter 2018, Honchar and Gnatiuk 2020, Erickson et al. 2021, Marquardt et al. 2021), they may have less abundant and diverse communities when compared to native flower forms (Comba et al. 1998, Corbet et al. 2001, Salisbury et al. 2015). Natural variation in plant color and morphology exists in wild species, but pollinator visitation is mediated by complex interactions with floral traits, and novel traits vary in whether they exclude or attract certain pollinator species (Erickson et al. 2022a). For example, a single gene changes Petunia Juss. (Solanales: Solanaceae) flowers from white to pink, and the resulting plants have significantly different pollinators (Hoballah et al. 2007). Consumers want native plants that support pollinators (Halleck 2015, Khachatryan et al. 2017, Narem et al. 2018, Bennett 2019, Tangren et al. 2022), yet it is unclear if cultivars are as valuable to pollinators as wild-type native species.

Insect pollinators (especially native species) benefit from native plants, as they provide habitat, forage, and nesting resources (Free 1993, Isaacs et al. 2009). Native plants may be most important to specialist pollinators, which have obligate plant hosts. Specialist bees, for example, collect pollen from a narrow range of plant taxa to feed their larvae (Cane and Sipes 2006). Most caterpillars are specialist herbivores which cannot successfully develop without their host plants (Futuyma 1976, Forister et al. 2015). Native plants support more specialist lepidopterans than introduced plants (Burghardt et al. 2010). The same has been found for specialist bees in gardens (Seitz et al. 2020) and in native plant nurseries, compared to conventional nurseries (Cecala and Wilson Rankin 2021). Generalist bees, adult lepidopterans, and syrphid flies are more flexible in their forage choices but may still prefer native plants (Frankie et al. 2013, Pardee and Philpott 2014, Fukase 2016). Native plants may be more attractive to pollinators due to phenology (Mitchell et al. 2022), nectar quantity (Branquart and Hemptinne 2000, Mallinger and Prasifka 2017), pollen nutrition (Roulston et al. 2000), their abundance (Williams et al. 2011), or ease of accessing their floral rewards (Branquart and Hemptinne 2000). The traits of both pollinators and plants influence plant–pollinator interactions.

Research evaluating pollinator preference for native plants and native cultivars has yielded mixed results. Two studies of pollinator use and visitation to woody native plant species and cultivars concluded that cultivars need to be evaluated on a case-by-case basis, with results ultimately depending on the traits cultivars have been selected for (Baisden et al. 2018, Ricker et al. 2019). Three studies of perennial plants found that cultivars were generally equally attractive or less attractive to pollinators than native plants (Ellis et al. 2013, White 2016, Dibble et al. 2020b). Four additional studies found cultivars to be as suitable for pollinators as wild-type natives (Baker et al. 2020, Bjørklund 2022, Peterman et al. 2023, Torrez et al. 2023). Another study, which focused on plants from the genus Phlox, found cultivars could be more, less, and equally attractive as native plants (Nevison 2016). These results confirm a need for additional research and examination of cultivars with varying plant traits across a suite of pollinator species.

We thus selected 8 plant species native to the Pacific Northwest and 1–3 cultivars of each species to evaluate for pollinator preference, based on pollinator abundance, richness, and community composition, in a common garden experiment. Through visual observations and collections of bees, butterflies, and syrphid flies over a 3-yr period, we sought to answer 2 primary research questions: Is there a difference in pollinator visitation to native plants and cultivars? And, which, if any, pollinator groups are associated with changes in visitation? We hypothesized that changes in pollinator preference for native plants or their cultivars will vary by plant species group, with 3 possible outcomes, based on pollinator abundance and/or richness: (i) preference for native plant types, (ii) preference for cultivar types, or (iii) no preference. In addition, few studies have considered native plant and cultivar use in the context of generalist and specialist taxa, though 2 studies have considered the impact of cultivated milkweed (Asclepias spp.) on pollinators, including monarch butterflies (Baker et al. 2020, Peterman et al. 2023). We hypothesized that specialist foragers were more likely to visit their native host plants in higher abundances, compared to native cultivars.

Materials and Methods

Experimental Design

This study took place over 3 field seasons (2020–2022) in Corvallis, Oregon, United States, at Oregon State University’s Oak Creek Center for Urban Horticulture. In November of 2019, two 3 × 30 m rows were tilled, mulched, and labeled in preparation for a common garden experiment. Each row contained 2 trial beds (blocks) with 30 consecutive 1 × 1 m² plots and a 1 × 30 m² inter-row of mulch, for a total of 4 blocks containing 120 plots (Supplementary Fig. A1). Unmulched areas around the garden rows contained turf grass. The garden layout follows the criteria for a randomized complete block design and was guided by previous studies (White 2016, Rollings and Goulson 2019).

We selected 8 species of Pacific Northwest native plants (annuals and perennials) that were previously evaluated for their attractiveness to bees, in terms of bee abundance and diversity (Anderson et al. 2022). We selected plants exhibiting varying levels of visitation and diversity to see how ornamental selection for plant traits might increase or decrease pollinator use of cultivars, relative to the wild-type native plant species. For each species, we located 1–3 cultivars, or hybrid cultivars, available in the horticultural market that represented an array of ornamental trait variations (e.g., changes in flower color, vegetation color, flower size, or growth habit; Supplementary Table A1) and had our native plant species in their pedigree (Supplementary Table A2). We did not attempt to capture genetic diversity when sourcing plant material. Instead, we used commercially available plant material from a single source when possible (Supplementary Table A3).

Five replicate 1-m² plots of each plant were established across the 4 blocks, with each individual plant type at least 1 m away from another replicate (Supplementary Fig. A1). Seeds and bulbs were planted in the fall of 2019 and 4-inch pots of plants were established in the spring of 2020; each plant type was seeded or planted based on nursery recommendations to achieve full plot coverage. Plots were watered during plant establishment and then on an as-needed basis over each growing season. If a plant in a perennial plot failed to overwinter, a reserve plant was planted in its place. Annuals were reseeded each fall to ensure sufficient germination and true-to-type plants. Additional plot maintenance included removal of weeds, pruning, pulling plants that expanded beyond their 1-m² area, and deadheading blooms. All bulbs were pulled and replanted in plots lined with 1″ hexagonal wire mesh in fall of 2020, in response to heavy rodent activity.

Sidalcea A. Gray ex Benth. (Malvales: Malvaceae) plants were removed in fall of 2020 when the wild-type species was reclassified as Sidalcea asprella ssp. virgata Greene (Halse 2020), thus yielding it distinct from the study cultivars derived from S. malviflora (DC.) A. Gray ex Benth. Additional cultivars of Achillea millefolium L. (Asterales: Asteraceae), Clarkia amoena (Lehm.) A. Nelson & J.F. MacBr. (Myrtales: Onagraceae), and Eschscholzia californica Cham. (Ranunculales: Papaveraceae) were added to the study to replace the Sidalcea plants. Nemophila Nutt. plants were intolerant of the clay soil at our site and 1 A. millefolium cultivar (‘Calistoga’) failed to overwinter. These plants were thus removed from the study in 2021. Due to consistent rain during the season, Camassia leichtlinii (Baker) S. Watson (Asparagales: Asparagaceae) was in flower, we were unable to obtain sufficient pollinator samples for comparison of the native plant against its cultivars. In this article, we thus report results of pollinator preference for 5 native taxa and 11 corresponding cultivars from 4 botanical families that performed well in our experimental garden over at least 2 field seasons (Table 1; Supplementary Table A1).

Table 1.

The native plant species and cultivars included in this study

Common name	Native species	Cultivar 1	Cultivar 2	Cultivar 3
Yarrow
	Achillea millefolium	Achillea × ‘Moonshine’	Achillea × ‘Salmon Beauty’
Western red columbine
	Aquilegia formosa	Aquilegia × ‘XeraTones’
Farewell-to-spring
	Clarkia amoena lindleyi	C. amoena ‘Aurora’	C. amoena ‘Dwarf White’	C. amoena ‘Scarlet’
California poppy
	Eschscholzia californica	E. californica ‘Mikado’	E. californica ‘Purple Gleam’	E. californica ‘White’
Douglas’ aster
	Symphyotrichum subspicatum	S. subspicatum ‘Sauvie Snow’	S. subspicatum ‘Sauvie Star’

Two photos of Aquilegia × ‘XeraTones’ are included to show its multiple color forms. The native C. amoena is from the subspecies lindleyi, while the cultivars were likely developed from C. amoena ssp. whitneyi. All plants were started in 2020, except for 3 plants added in 2021: Achillea × ‘Moonshine’, C. amoena ‘Scarlet’, and E. californica ‘Purple Gleam’. All photos were taken by J. Hayes.

Pollinator Surveys

Observations

We monitored the number of open flowers in each 1-m² plot at least once per week from April through October (2020–2022). We used different flowering units to account for variations in floral morphology across plant groups. For Aquilegia, Clarkia, and Eschscholzia groups, we measured the number of simple flowers. We used capitula as our flowering units for Symphyotrichum, and peduncle divisions for Achillea. Following flower counts, trained observers conducted 5-min pollinator observations (1–2× per week) between April and September.

Since flower abundance can impact pollinator visitation (Fowler et al. 2016, Erickson et al. 2022a, but see Essenberg 2012, Garbuzov et al. 2015), we structured our observations to reduce the likelihood that a zero count resulted from sparse flowering. In the first year of the study (2020), we started conducting observations at the initiation of flowering, and later removed observations on plots if they occurred when the flower count was below the 25th percentile for the year. In 2021, the 25th percentile flower counts from 2020 were used as guiding counts for initiating observations, and in 2022 we used the counts from 2021. Due to changes in bloom counts as plants established and the tendency for observers to conduct observations on plots with few flowers, we calculated the true 25th percentile thresholds for each season retroactively and removed any observations with flower counts that fell below them.

Observations occurred on days with weather conditions favorable to pollinator activity (temperature ≥15.5°C, average wind speed less than 3.5 m/s, and cloud cover ≤50%) between 09:00 and 16:00 PST. Protocols were adjusted to accommodate extreme weather events during the study: in 2020, fieldwork was suspended after September 4th due to excessive smoke from nearby wildfires, which made field conditions hazardous. The starting time for observations was temporarily changed to 08:00 during extreme heat events in 2021 and 2022.

Observers haphazardly chose 1 block and a subsequent plot to begin their observations each day, and rotated blocks with other observers to avoid watching the same set of plants multiple times within a week. During observations, we monitored one 1 × 1 m² plot for insect activity, recording and sight-identifying all insect visitors that entered the plot, and listing all interactions between the insect(s) and the study plant (e.g., foraging, basking, resting, nectar-robbing). We considered 2 types of interactions as foraging: (i) an insect directly interacting with the nectaries or anthers of a flower, and (ii) leafcutting bees collecting leaf or petal segments from plants. We did not differentiate pollen foraging from nectar foraging during observations. If an insect left the observation area and returned later, it was considered a new individual.

When analyzing pollinator visitation data, it is common to organize taxa into groups to understand visitation trends among and within taxa. Common groups include all pollinators, honey bees, bumble bees (Bombus spp., Hymenoptera: Apidae), other bees, syrphid flies, and butterflies (e.g, Garbuzov and Ratnieks 2015, White 2016, Nabors et al. 2022, Torrez et al. 2023, Dibble et al. 2020a, and others). Though we identified some pollinator taxa to more specific morphospecies during field observations, such as placing bumble bees in color groups, we present 8 pollinator groups here. We include the “All pollinators” group to address our first research question, whether there is a difference in visitation to native plants and cultivars, regardless of pollinator type. In addition to the groups listed previously, we include long-horned bees (Apidae: Eucerini) and leafcutting bees (Hymenoptera: Megachilidae) for their high abundance and ease of identification from other bees in our field site. These 2 groups contain entirely solitary bees (Danforth et al. 2019), with long-horned bees having a high incidence of specialization (Wright 2018), and leafcutting bees foraging from plants for pollen, nectar, as well as nesting materials.

Collections

After 5-min observations were completed across the entire garden on a given date, we vacuum-sampled bees and syrphid flies from plots using a modified RYOBI ONE+ 18-V battery-powered hand vacuum (Bioquip Products, Gardena, California). Plots were vacuumed one at a time in a haphazard order until no pollinators, except honey bees and butterflies, remained. We only vacuumed pollinators visiting flowers. Butterflies present during vacuum sampling were identified to species and their total abundance on a given plot was recorded. One honey bee worker was collected per plot to confirm identification; the remaining workers were counted and recorded in the same way as butterflies. Vacuumed specimens and a label with counts for recorded but uncollected specimens (butterflies and additional honey bees) were transferred to a jar of ethanol with a collection label and brought to the lab to be pinned, identified, and curated. J. Hayes and L. Best identified bee specimens and J. Hayes identified syrphid fly specimens. A list of the keys, guides, and other resources used for identification can be found in Hayes et al. (2024).

We categorized the native status and diet breadth of each of our pollinator species, to understand how pollinators with these traits may differentially visit our study plants. For flies and butterflies, we classified adult and larval diet breadth separately. We categorized bee diet based on female pollen foraging and descriptions from Cane and Sipes (2006). Bees are described as polylectic (collecting pollen from 4 or more plant families), mesolectic (collecting from 1 to 3 families or big tribes), eclectic oligolectic (foraging from 2 to 3 families or tribes and 2 to 4 genera), or as narrow oligolectic (foraging from a single family, tribe, or genus). For bee species without sufficient records in the literature, we classified diet using visitation records of female bees to plants from our own data set, records from the Oregon Bee Atlas (Best et al. 2021a, 2021b, 2022a, 2022b), and inferred traits from related species. Voucher specimens, at least 1 per caste and sex, as available, were deposited in the Oregon State Arthropod Collection (Hayes et al. 2024).

Analyses

We examined abundance per 5 min and richness data across years to investigate the main effect of plant type on pollinator visitation to native plants and native cultivars. All statistical analyses were conducted in R version 4.3.1 (R Core Team 2024).

Observations

We used generalized linear mixed-effects models (GLMMs) to model fixed and random effects of the abundance of our 8 pollinator groups (Supplementary Table B1). We limited our data set to foraging observations only. To account for differences in flower counts within plant groups, we included bloom count as a covariate and removed observations if they fell below the 25th percentile of flower counts for each season. We modeled our response variable, pollinator abundance, separately for each of our 8 pollinator groups, due to issues with overfitting and model convergence when including an interaction between pollinator group and plant type. The “All pollinators” model included the fixed effects of plant type, year, observer, bloom count, the interaction between plant type and time of day, and the random effect of plot number. We included year as a fixed effect to account for the high interannual variation in pollinator communities (Lázaro et al. 2010, Aldercotte et al. 2022). Observation start times were categorized into 2 periods, morning (11:59 and earlier) and afternoon (12:00 and later). We included time of day as a covariate to account for differential temporal visitation, since patterns of pollinator foraging activity vary by time of day (Boyle-Makowski and Philogène 1985, Knop et al. 2018, Xu et al. 2021, Karbassioon and Stanley 2023).

Like Erickson et al. (2020), we developed simplified models to examine visitation by specific pollinator groups. These models contained a limited set of predictor variables: plant type, bloom count, and the random effect of plot number (Supplementary Table B1). We fitted models to negative binomial and zero-inflated negative binomial distributions using the “glmmTMB” function from the R package “glmmTMB” (Brooks et al. 2017). To assess model fit and assumptions, we examined visuals of residuals from the R packages “performance” (Lüdecke et al. 2021) and “DHARMa” (Hartig 2022). We also used functions from the “DHARMa” package to test for distributional assumptions (Kolmogorov–Smirnov test), dispersion, and outliers (Hartig 2022). We used these tests to guide model adjustments (e.g., changing the zero-inflation parameter) and select the best-fitting model (Supplementary Table B1; Supplementary Figs. B1–B8). Some plants were removed from models when they were never visited by a particular pollinator group and their presence in models led to an excess of zeros, even when accounting for zero-inflation (Supplementary Table B1).

Estimates for each of the fixed effects from our final models were calculated and back-transformed using the “emmeans” function from the R package “emmeans” (Lenth 2023). We then used the Šidák method (Šidák 1967) to adjust within group P values and confidence intervals of our preplanned pairwise comparisons between native plants and cultivars (native–cultivar) by pollinator group (78 total comparisons). To understand the overall ecological trends between native plants and cultivars, which may not be statistically significant, we also used a liberal approach to evaluate comparisons between native plants and cultivars. We examined the mean ± standard error (SE) from comparisons, equivalent to a 68% interval around the mean: if the interval was greater than zero, we interpreted that as preference for the native plant. Intervals less than zero indicated preference for cultivars, and intervals that contained zero indicated no change in visitation. We conducted additional post hoc analyses on the fixed effects from the “All pollinators” model, including year, observer, the interaction between plant type and time of day via pairwise comparisons with Šidák adjustments (Šidák 1967).

Collections

Vacuum samples of pollinators across all years and all plots, including from plants not reported on in this study (e.g., Sidalcea, Nemophila, and Camassia groups, and A. millefolium ‘Calistoga’), were pooled to establish the garden-level species richness. The garden-level species pool serves as a comparative metric for species richness across individual plants and reveals the types and traits of pollinators that are absent from our study plants. We recorded raw pollinator richness counts for each plant type in addition to using the “iNEXT” function in the R package “iNEXT” (Chao et al. 2014, Hsieh et al. 2022) to build species accumulation curves and estimate pollinator richness and sample coverage. We also calculated the Sørensen–Dice Index (SDI; Dice 1945, Sørensen 1948) for native–cultivar plant pairs, using the “vegdist” function from the “vegan” package (Oksanen et al. 2022). To better understand the specific pollinator communities visiting native plants and cultivars, we also created 2 nonmetric multidimensional scaling (NMDS) plots (one for each: bees and syrphid flies) to understand how plant types correspond with pollinator community structure. To construct the NMDS plots, we used the “metaMDS” function in the “vegan” package (Oksanen et al. 2022) with Bray–Curtis distance on abundance counts of our collected specimen.

Results

Observations

Over 3 field seasons, we watched 12,440 individual pollinators visit plants in our experimental garden during 5-min observations across 85 sampling dates. Forty-six of these individuals were observed collecting petals from C. amoena plants, and the rest were observed foraging for pollen or nectar. The number of 5-min observations across plant types ranged from 32 (Aquilegia × ‘XeraTones’, shortest bloom period) to 210 (native E. californica, longest bloom period).

In the “All pollinators” model, we examined the effects of plant type, year, observer, bloom count, and the interaction between time of day and plant type on total pollinator abundance during our 5-min plant observations. Values are reported as estimated marginal means ± SE unless otherwise noted. We found an effect of time of day for 4 plant types: 2 had higher visitation in the afternoon (Achillea × ‘Salmon Beauty’ and C. amoena) and 2 had greater visitation in the morning (Aquilegia formosa Fisch. ex DC., Ranunculales: Ranunculaceae and C. amoena ‘Scarlet’; Fig. 1; Supplementary Table A4). Achillea and Symphyotrichum Nees (Asterales: Asteraceae) tended to have greater visitation in the afternoon, and Aquilegia plants tended to be visited more often in the morning, though their means were not significantly different (Fig. 1).

Fig. 1.

Interaction between plant type and time of day on total pollinator visitation. Visitation rates are reported as estimated marginal means for each time of day. Asterisks indicate a significant time of day effect for individual plant types (P < 0.05) from post hoc pairwise comparisons. We did not compare time of day across plant (genus) groups.

Total visitation per 5 min by individual observers ranged from 4.03 ± 0.19 (observer E) to 6.85 ± 0.98 (observer F). There were significant differences between observer visitation rates, but each observer shared at least 1 letter with at least 2 other observers in the compact letter display (Supplementary Fig. A2A; Supplementary Table A5). No single observer, then, was significantly different from all the others. The effect of individual observers may be explained by when they participated in the study. Observer F, for example, only conducted observations during the height of summer in 2022 (Supplementary Fig. A2A). Year influenced total visitation per 5 min: 2021 had a higher visitation rate (5.55 ± 0.27) than either 2020 or 2022 (Supplementary Fig. A2B).

Across all pollinator groups, visitation per 5 min by plant type ranged from 0.00 ± 0.00 (2 plant–pollinator combinations) to 12.87 ± 1.02 (“All pollinators” visiting S. subspicatum ‘Sauvie Snow’; Fig. 2). Outside of “All pollinators”, “Other bees”, “Honey bees”, and “Bumble bees” had some of the highest visitation rates, while “Butterflies” and “Leafcutting bees” had some of the lowest rates. In native–cultivar comparisons, native plants were statistically significantly preferred 29 times, cultivars 6 times, and 43 comparisons had no significant outcome (Fig. 3). All pollinator groups except “Leafcutting bees” had at least 1 statistically significant comparison. “All pollinators”, “Bumble bees”, and “Honey bees” exhibited preferences for native plants and cultivars, and the 4 remaining groups only had significant preferences for native plants. The direction of pollinator preference was consistent within plant groups with significant results, except for the Achillea and Symphyotrichum groups, where preferences varied by pollinator group (Fig. 3). By adjusting for 78 comparisons between native plants and cultivars, we reduced the probability of a Type I error, but our results are more conservative, and may conceal ecologically significant trends. If we examine just the means and SEs of these comparisons, a liberal approach, we found at least a slight preference for native plants in 40 comparisons, for cultivars in 13 comparisons, and no preference in 25 comparisons (Fig. 3).

Fig. 2.

Mean pollinator visitation rates by plant type and pollinator group. Means are reported as estimated marginal means ± SE. Values in each subplot were drawn from the models for their respective pollinator group. Black bars on the y-axis indicate means of 0.00 ± 0.00 (e.g., ‘XeraTones’ on the “Leafcutting bees” subplot). Gaps on the y-axis are indicative of plants that were removed from models (e.g., the Aquilegia group on the “Long-horned bees” subplot).

Fig. 3.

Native minus cultivar comparison results. Means are reported as estimated marginal means (native minus cultivar) ± SE (black bars) with 95% confidence intervals (colored bars). Values in each subplot were drawn from models for their respective pollinator group. The dotted line at x = 0 represents the threshold for a nonsignificant result. Points that fall to the left of the zero line indicate a preference for the cultivar compared (C.P.) and points that fall to the right indicate a native preference (N.P). Points with asterisks below them indicate a statistically significant comparison result (P < 0.05). Points without confidence intervals represent pairs where neither the native plant nor the cultivar received any visitation by the listed pollinator group, and gaps in subplots represent plant groups removed from that model.

The effect of bloom count was consistent across all 8 pollinator groups: when all other variables were held constant, a 1-unit increase in bloom count increased pollinator abundance per 5 min by a factor of 1.00 ± 0.00. This effect was statistically significant for half of the pollinator groups: “All pollinators” (z = 10.67, P = <0.001), “Honey bees” (z = 3.49, P = <0.001), “Bumble bees” (z = 3.02, P = 0.003), and “Other bees” (z = 7.22, P = <0.001). We were unable to statistically test the interaction between bloom count and plant type, but an examination of average bloom counts over time suggests that plants with more flowers may not always be more attractive to pollinators (Fig. 4). For example, Achillea × ‘Moonshine’ and Aquilegia × ‘XeraTones’ tended to have higher flower counts than their respective native plant pairs (Fig. 4), but neither cultivar was ever significantly preferred by pollinators (Fig. 3).

Fig. 4.

Average bloom count over time for each plant type. All positive counts (count > 0) across years and replicates were averaged by week and are displayed using local polynomial regression with an unbounded distribution.

Collections

We collected 4,691 pollinators and recorded an additional 1,998 butterflies and honey bees from our study plants. From here onwards, we use “collected” to refer to the specimens physically collected and recorded during vacuum samples. From these specimens we identified 88 pollinator species: 4 butterflies, 14 syrphid flies, and 70 bees. Our garden-level sample coverage was estimated at 99.67%, with a predicted species richness of 104 (Table 2, Supplementary Fig. C1). Our most abundantly collected taxa included A. mellifera (1,446 individuals), Halictus ligatus Say (Hymenoptera: Halictidae, 1,016 individuals), and B. vosnesenskii Radoszkowski (504 individuals). Sixteen taxa were represented by a single individual.

Table 2.

Observed and estimated pollinator species richness by plant type

Native species	Plant type	Richness
		All pollinators	Bees	Butterflies	Syrphid flies
Garden-level		88 (104)	70	4	14
Achillea millefolium	Native	21 (41)	16	1	4
	‘Moonshine’	15 (20)	8	1	6
	‘Salmon Beauty’	14 (41)	8	3	3
Aquilegia formosa	Native	9 (29)	7	0	2
	‘XeraTones’	3 (8)	3	0	0
Clarkia amoena ssp. lindleyi	Native	32 (104)	29	2	2
	‘Aurora’	17 (28)	16	1	0
	‘Dwarf White’	23 (24)	22	1	0
	‘Scarlet’	9 (13)	9	0	0
Eschscholzia californica	Native	31 (40)	26	0	5
	‘Mikado’	19 (57)	14	0	4
	‘Purple Gleam’	16 (26)	15	0	1
	‘White’	21 (23)	18	0	3
Symphyotrichum subspicatum	Native	29 (141)	24	3	2
	‘Sauvie Snow’	29 (57)	20	3	6
	‘Sauvie Star’	25 (55)	18	3	4

All butterfly and syrphid fly species present in the garden-level species pool were observed on at least 1 focal plant, but 13 bee species were unique to the garden-level pool (Fig. 5; Hayes et al. 2024). Additional diversity metrics are reported in Supplementary Fig. C2. Observed richness is reported, followed by Chao1 richness estimates in parentheses for the total richness (“All pollinators”) column.

Across our study plants, observed pollinator species richness ranged from 3 (Aquilegia × ‘XeraTones’) to 32 (native C. amoena) (Table 2). No plant was visited by all the species in a given pollinator group, but the native C. amoena had the greatest bee richness (29 species), there was a 4-way tie for butterfly richness (3 species), and a 2-way tie for greatest syrphid fly richness (6 species). In all groups except Symphyotrichum, the native plant had a greater total observed richness than any of the cultivars. The native Symphyotrichum subspicatum (Nees) G.L. Nesom had the greatest estimated species richness (141 species), and Aquilegia × ‘XeraTones’ had the lowest (8 species). Syrphid flies were collected from every native plant in our study, but not from every cultivar (Table 2; Fig. 5). Species accumulation curves and sample coverage are reported per plant type in Supplementary Fig. A7.

Fig. 5.

Heat map of pollinator species collected from each plant. Morphospecies represented only by male individuals may inflate some estimates of species richness.

We used the SDI to examine total pollinator community similarity between native plants and cultivar pairs (Supplementary Fig. A3). The mean overlap in pollinator communities was 0.43 ± 0.05 within plant pairs. The native Aquilegia and its cultivar had the greatest similarity (SDI = 0.83), with C. amoena ‘Dwarf White’ and E. californica ‘White’ sharing the least similar communities with their respective native plants (SDI = 0.27). The lowest number of shared species was 1 (Aquilegia), and the greatest was 20 (C. amoena ‘Dwarf White’ and S. subspicatum ‘Sauvie Snow’ with their respective native plants). Native plants tended to have a greater number of unique species, which is consistent with the higher richness seen in Table 2. The one exception was Symphyotrichum, where 9 species were unique to both the native and S. subspicatum ‘Sauvie Snow’ (Supplementary Fig. A3).

Two of the butterfly species collected are considered specialists at the larval stage, Ochlodes sylvanoides (Boisduval; Lepidoptera: Hesperiidae) on Poaceae (Scott 1992) and Pieris rapae L. (Lepidoptera: Pieridae) on Brassicaceae and Capparidaceae (Jones 1977) (Fig. 5; Supplementary Fig. A4). We collected 6 specialist bee species from our focal plant taxa, including 4 eclectic oligoleges on Asteraceae: Megachile apicalis Spinola (Hymenoptera: Megachilidae; Müller and Bansac 2004), Megachile fidelis Cresson (Wilson et al. 2010), Melissodes microstictus Cockerell (Hymenoptera: Apidae), and Melissodes lupinus Cresson (LaBerge 1961, Wright 2018). Two narrow oligoleges on Clarkia were collected: Megachile gravita Mitchell and Melissodes clarkiae LaBerge (LaBerge 1961, MacSwain et al. 1973; Fig. 5; Supplementary Fig. A4).

The most common specialist was M. lupinus (Fig. 5). Most females of this species were collected from Symphyotrichum plants, with the majority collected from the native (113), followed by S. subspicatum ‘Sauvie Snow’ (77) and S. subspicatum ‘Sauvie Star’ (20; Fig. 5). Most M. clarkiae females were collected from the native C. amoena (145), compared to only 13 females collected on C. amoena ‘Dwarf White’, which was second to the native species. M. gravita, another Clarkia specialist, was only collected from the native C. amoena (6 females). We collected more M. microstictus from the native Symphyotrichum (18 females) and the native Achillea (3 females) than either of their respective cultivars. Most pollinator species, regardless of diet, were not collected evenly within plant groups (Fig. 5).

Bee community composition within plant groups, based on the NMDS ordination, was most similar in Symphyotrichum, followed by Eschscholzia (Supplementary Fig. A5). Three of the Clarkia plants had similar bee communities, although the community visiting C. amoena ‘Scarlet’ was more similar to that of the native Aquilegia and Achillea × ‘Moonshine’ than any other Clarkia plants. The Aquilegia plants are the farthest from one another in the ordination, which is unexpected from their SDI value of 0.83 (Supplementary Fig. A3), but the NMDS considers similarity across plant groups and within plant pairs. The abundant specialists (n > 10; M. clarkiae, M. lupinus, and M. microstictus) tended to be closely aligned with their respective native host plants (Supplementary Fig. A5). Two petal-cutting bees (M. brevis Say and M. montivaga Cresson) were collected from C. amoena plants, including 5 M. brevis individuals collected while actively harvesting flower petals (Hayes et al. 2024). Both petal cutters were associated with the native C. amoena and collected from it in greater abundance than the cultivars (Supplementary Fig. A5, Fig. 5).

Syrphid fly community composition was less consistent (Supplementary Fig. A6). No syrphid fly taxa were collected from cultivars of the Aquilegia or Clarkia groups. The closest ordinations within a plant group occurred between the native E. californica and E. californica ‘Purple Gleam’. The most abundant syrphid species, Eristalis arbustorum L., was closely associated with the same Eschscholzia plants. This result is surprising, considering E. arbustorum was collected more often from Achillea plants (Fig. 5).

Discussion

Our results are the first to demonstrate variation in pollinator visitation and utilization of native plants and cultivars for Pacific Northwest flora and pollinator fauna. These results may only be indicative of the specific suite of native plants and cultivars examined and may be constrained by the specific pollinator community and environmental conditions found at our study site. While we did not identify a consistent trend in visitation to native plants and cultivars, when a preference did occur, pollinators were more likely to prefer native plants than cultivars (Fig. 3). This finding is consistent with previous research across multiple plant and pollinator taxa (Ellis et al. 2013, White 2016, Ricker et al. 2019, Dibble et al. 2020b, but also see Bjørklund 2022, Torrez et al. 2023).

Our results may differ from other studies based on regional differences, the number of plant taxa we examined (e.g., Baker et al. 2020, Peterman et al. 2023), the traits exhibited by cultivars, and by methodology. Evaluations of native plants and cultivars may also vary when results are based solely off of specimen collections (Bjørklund 2022, Peterman et al. 2023) or observations (Nevison 2016, White 2016, Dibble et al. 2020b), compared to a combination of methodologies (Ellis et al. 2013, Ricker et al. 2019, Baker et al. 2020, Torrez et al. 2023). Some studies compared visitation rates across their entire suite of plants (Ricker et al. 2019, Torrez et al. 2023), whereas we followed White (2016) and conducted pairwise comparisons between native plants and cultivars. Our study system may also be unique in that specialist pollen foragers were highly abundant: we collected 466 female specialist bees in the genus Melissodes, which were likely the dominant individuals in our long-horned bees group. Other studies have collected specialist bees (Bjørklund 2022, Torrez et al. 2023), but in significantly lower quantities.

Studies often report pollinator visitation to native plants and cultivars, but few identify pollinators to the species level, which precludes finer scale analyses of pollinator community composition. We found native plants always had a greater bee richness and almost always equal or greater pollinator richness than cultivars, though estimated cultivar richness was equal or greater in 2 cases (Achillea × ‘Salmon Beauty’ and E. californica ‘Mikado’, respectively). Across their 8 native–cultivar plant pairs, Torrez et al. (2023) had 3 native plants with greater bee species richness, compared to 5 cultivars with greater richness. In Asclepias trials, the native A. tuberosa attracted more species (bees and butterflies) than 2 hybrid plants, though there was only a difference of 2 species between one of the cultivars and the native (Peterman et al. 2023).

Examining the species present and those absent from our focal taxa may further explain trends in pollinator visitation. Butterflies, for example, were never collected from the Eschscholzia group, and were uncommon overall, save for P. rapae and O. sylvanoides. Eschscholzia californica produces a negligible quantity of nectar (Hicks et al. 2016), such that most researchers claim there is none provided at all (Cook 1962, Thorp 2011, Becker et al. 2023). The absence of butterflies on Eschscholzia plants, then, may be best explained by the lack of this resource. Most bees collect nectar from a wide range of plant species (Wcislo and Cane 1996), yet the Sidalcea specialist Diadasia nigrifrons Cresson (Hymenoptera: Apidae) is rarely collected from plants outside the Sidalcea genus (Linsley and MacSwain 1958, Ram 1969, Best et al. 2022a, 2022b). Though our other focal taxa produce nectar, D. nigrifrons may satisfy its energy requirements by primarily visiting its host plant. Annual variation in pollinator populations (Lázaro et al. 2010, Aldercotte et al. 2022) or low population abundance may also preclude visitation. These factors interact with inherent species traits and plant traits to predict plant use. We thus echo White (2016), Baisden et al. (2018), and Ricker et al. (2019) suggesting that pollinator preference for native plants and cultivars depends on the particular flora and fauna observed, in addition to the traits exhibited by cultivars.

Visitation by specialist pollinators could act as an indicator of the ecological value of cultivars. Specialist organisms are of particular ecological concern due to their greater sensitivity to exotic introductions (Valdovinos et al. 2018) and land use change (Biesmeijer et al. 2006, Cane et al. 2006, Kleijn and Raemakers 2008, Winfree et al. 2011, Concepción et al. 2015). Their losses can lead to functional homogenization (Clavel et al. 2011) and even loss of their host plants (Weiner et al. 2014, Mathiasson and Rehan 2020). Here, we found 3 abundant specialist species more often on native plant taxa than native cultivars. The proportion of female specialists collected from native plants varied by host specificity: 85% for M. clarkiae (narrow oligolege), 67% for M. microstictus, and 53% for M. lupinus (eclectic oligoleges). Specialists that are more specific in their preferences (e.g., narrow oligoleges and specialist Lepidoptera) may be most sensitive to changes in plant traits. Most cultivars, even if visited by specialists, would not be suitable replacements for wild-type native taxa in restoration (Gann et al. 2019), but they may be ecologically valuable in gardens, especially when wild-type plants are unavailable.

Another plant–pollinator relationship that requires further exploration is the nontrophic use of plants by pollinators. Some leafcutting bees are generalized in their nest substrate choice, though others have preferences (Krombein et al. 1979, Genaro 1996, Eigenbrode et al. 1999, MacIvor 2016, Soh et al. 2019). We observed M. brevis and/or M. montivaga collecting petals from all C. amoena plants, but they were collected in greater abundance from the native (Fig. 5). Mead et al. (2023) examined petal-cutting bee usage of the native C. amoena, ‘Aurora’, ‘Dwarf White’, and ‘Scarlet’ over 1 field season and found that the bees collected petals far more often from the native plant. This finding was significant, even after accounting for bloom count, which only played a minor role in predicting the total number of flowers with petal cuts. In a study of pollinators visiting native shrubs in Connecticut, petal-cutting bees collected yellow petals of Dasiphora fruticosa L., as well as the native cultivar D. fruticosa ‘Goldfinger’ (a yellow-flowered tetraploid), but not the pink-flowered D. fruticosa ‘Pink Beauty’ (Ricker et al. 2019). Other specialist and generalist herbivores reduced their use of native plant cultivars with significantly altered vegetation color (Tenczar and Krischik 2007, Baisden et al. 2018). Changes in flower and foliage color, then, may impact nesting resources that plants provide to leafcutting bees.

It is noteworthy that the only cultivars preferred multiple times over native plants, S. subspicatum ‘Sauvie Star’ and S. subspicatum ‘Sauvie Snow’, were some of the least developed cultivars in the study. These cultivars are named selections from a population of S. subspicatum on Oregon’s Sauvie Island (Shepherd, personal communication). Seeds were collected from wild morphs and then grown out to select for the most vigorous purple and white flowers (Shepherd, personal communication). In Torrez et al. (2023), Symphyotrichum oblongifolium (Nutt.) G.L. Nesom had greater visitation by wild bees than S. oblongifolium ‘Raydon’s Favorite’ across both their sites, but was less visited than ‘October Skies’ and ‘Dream of Beauty’ (at 1 site and 2 sites, respectively). Symphyotrichum oblongifolium ‘Dream of Beauty’ (Barr 2015) and ‘Raydon’s Favorite’ (Barnette 2016) also originate from wild populations, and ‘October Skies’ is a selection of ‘Raydon’s Favorite’ (Lurie Garden 2024). In our study and White (2016), highly developed cultivars (e.g., interspecific hybrids) were less likely to be preferred over their native counterparts. These results suggest that plants with wild-type native phenotypes should be emphasized when planting for pollinator fauna.

Honey bees are often the most abundant pollinator taxa in plant–pollinator studies (Hung et al. 2018, 2019, Dibble et al. 2020a, Pei et al. 2023, Peterman et al. 2023, Torrez et al. 2023), suggesting that examining “all bees” or “all pollinators” as composite groups may not be indicative of true wild pollinator activity in honey bee-abundant landscapes (Paini 2004, Hung et al. 2019, but see Garibaldi et al. 2021). Viewing honey bees as a distinct pollinator group improves our understanding of the foraging behavior by unmanaged pollinators. Honey bee visitation data may reveal plants that consolidate highly abundant, eusocial bees in garden environments, and those that remain accessible for a diversity of pollinators. Syrphid flies, for example, may leave flower patches in response to high densities of pollinators (Jauker and Wolters 2008). By planting species that act as magnets for honey bees, in addition to wild-type native plants, gardens could potentially support greater pollinator abundance and diversity (Gilpin et al. 2019). Future research should explore the capacity of cultivars to diminish competition for floral resources in gardens between abundant, eusocial and solitary, small-bodied pollinators.

Historically, pollinator syndromes (e.g., patterns in floral phenotypes) have been used to predict plant use by different pollinator taxa (Willmer 2011). Current research continues to conclude that pollinator preference is not constrained in the way pollinator syndromes may suggest. Bees, which are supposed to be primarily drawn to blue, white, and yellow flowers (Faegri and Van Der Pijl 1980), visited purple, orange, and even pink flowers in great abundance in our study (S. subspicatum ‘Sauvie Star’, E. californica, C. amoena, respectively), and the flower with the greatest bee richness was pink with polymorphic red petal spots (C. amoena). Pollinator choice is influenced by a multitude of floral traits, many of which are selected for and/or altered when breeding cultivars (Wilde et al. 2015, Erickson et al. 2022a). Floral resource production, for example, is temporally variable (Real and Rathcke 1988, Herrera 1989, Fowler et al. 2016), and plant breeding can impact resource quality and quantity (Garbuzov and Ratnieks 2015, Egan et al. 2018, Erickson et al. 2022a). Cultivars, even within a single species, may vary in pollinator visitation rates across times of day, as seen here and in Erickson et al. (2020). Differences in visitation to native plants and cultivars across the day could thus result from synchrony, or a lack thereof, in pollinator activity cycles and floral resource production, or the quality of resources provided.

Since we did not identify consistent trends in visitation, we cannot conclude that wild-type native plants or native cultivars are universally preferred by pollinators. Quantifying pollinator visitation rates and richness is the first step in improving our understanding of pollinator relationships with native plants and cultivars. Pairing visitation data with evaluations of floral traits, including nutrition (pollen and nectar), size of floral displays, bloom duration, floral morphology, floral height, and floral color will enrich our understanding of choice for pollinators with different life history traits (Erickson et al. 2022a, b, Wang et al. 2024). These evaluations will reveal plant traits that differ between native plants and their cultivar pairs, and may explain differences in visitation rates. This is the focus of part II, our upcoming manuscript.

Guidance on floral traits for pollinators can be used by plant breeders and the horticultural industry to improve techniques to develop plants that perform well in retail environments while retaining their ecological value (Wilde et al. 2015), such as their capacity to support pollinators. What remains is how to improve cohesion between the green industry’s use of plant labels (e.g., “pollinator friendly” and “native plant”), and research-based support for their application. When paired with standardized protocols to measure attractiveness to pollinators (Erickson et al. 2022a), cultivars can be evaluated for their ability to support different pollinator groups. Whether cultivars with ornamentally derived phenotypes should be marketed as “native” is still up for debate, especially as conservation professionals emphasize the importance of local ecotypes in the production of native plant material (McKay et al. 2005, Basey et al. 2015, Altrichter et al. 2017, Gettys 2023, Lawson-Canning 2023, Ren et al. 2023).

Supplementary data

Supplementary data are available at Environmental Entomology online.

Acknowledgments

We would like to acknowledge that Oregon State University and the Oak Creek Center for Urban Horticulture, where our study took place, are “within the traditional homelands of the Mary’s River, or Ampinefu Band of Kalapuya. Following the Willamette Valley Treaty of 1855 (Kalapuya etc. Treaty), Kalapuya people were forcibly removed to reservations in Western Oregon. Today, living descendants of these people are a part of the Confederated Tribes of Grand Ronde Community of Oregon (https://www.grandronde.org) and the Confederated Tribes of the Siletz Indians (https://ctsi.nsn.us)” (Whitebear et al. 2023). We thank Angelee Calder, Elliot Ariel, Isabella Messer, Jay Stiller, Aaron Anderson, Clifford Brock, and Mericos Rhodes in assisting with the initial phases of this research. We are also grateful to Rebecca Agatstein, Cara Still, and Nina Ferrari for providing comments on figure layouts. We also thank Annie’s Annuals and Perennials, Bluestone Perennials, Eden Brothers, Gray’s Garden Center, Heritage Seedlings, Xera Plants, Inc., Northwest Meadowscapes, Seven Oaks Native Nursery, Silver Falls Seed Co., Willamette Gardens LLC, Easy to Grow Bulbs, and Native Foods Nursery for helping us source the plants used in this research.

Author contributions

Jen Hayes (Conceptualization [equal], Data curation [lead], Formal analysis [lead], Funding acquisition [supporting], Investigation [lead], Methodology [lead], Project administration [equal], Software [lead], Supervision [lead], Validation [lead], Visualization [lead], Writing—original draft [lead], Writing - review & editing [lead]), Nicole Bell (Investigation [supporting], Writing - review & editing [equal]), Lincoln Best (Investigation [supporting], Resources [supporting], Writing - review & editing [supporting]), Svea Bruslind (Investigation [equal]), Devon Johnson (Investigation [supporting]), Mallory Mead (Investigation [equal], Writing - review & editing [supporting]), Tyler Spofford (Investigation [equal]), and Gail Langellotto (Conceptualization [equal], Funding acquisition [lead], Methodology [equal], Project administration [equal], Resources [lead], Supervision [equal], Validation [equal], Writing - review & editing [equal])

Funding

This work was supported by Y. Sherry Sheng and Spike Wadsworth and grants from the Garden Club of America and the Native Plant Society of Oregon.

Data availability

All data reported in this manuscript will be provided upon reasonable request to the corresponding author.