This study was conducted at the University of Georgia Research and Education Garden on the UGA Griffin Campus (Spalding Co.; 33°24'67”N, 84°26'40”W). Cultivar trial plots were established during fall 2016. Salvia, coreopsis, and aster cultivars were each established in their own separate plots within the 24.3-hectare Research Garden. Plants were chosen to reflect new series and standards in the horticulture industry and commercial availability. Salvia cultivars included in this study were Salvia greggii “Radio Red”, Salvia guaranitica “Black and Blue”, Salvia microphylla x greggii “Heat Wave Blast”, “Heat Wave Blaze”, “Heat Wave Glitter”, “Heat Wave Sparkle” and Salvia nemorosa “Steel Blue”. Coreopsis cultivars were Coreopsis lanceolata hybrid “Desert Coral”, a hybrid cross of Coreopsis auriculata “Zamfir” (female parent) and Coreopsis lanceolata “Early Sunrise” (male parent) “Jethro Tull”, Coreopsis verticillata Sizzle and Spice? series “Hot Paprika”, Coreopsis Solanna? “Golden Sphere” and Coreopsis verticillata “Sylvester”. Aster cultivars were Ampelaster carolinianus “Climbing Aster”, Symphyotrichum grandiflorum “Wild Blue”, Symphyotrichum ericoides “Heath Aster”, Aster oblongifolius “Jane Bath”, Aster oblongifolius “Rachel Jackson”, Aster tataricus “Jindai” and Symphyotrichum novae-angliae “English Countryside”. Plants were provided by regional plant nurseries as trade-gallon size transplants and were planted on 0.9 m centers. There were three plants per replication and six to eight replications per cultivar (six asters, seven coreopsis and eight salvia replications × three plants per replication) planted in a randomized complete block design. Blocks were 4.6 m apart. Plots= blocks were mulched with pine bark and drip irrigated with water being applied at first sign of wilt.

Insect observations began at first flowering and were made weekly during the flowering period for 2 years. Observations were made between 1,000 and 1,400 h unless rain or high wind impeded observations. While some pollinators are active before and after this window, it is a standard period for assessment when flower visitation is most frequent. All plants that had reached anthesis were observed on the same day. Number of insect visitors during a 1-min time span per replication was recorded in six categories: honey bees (Apis mellifera), bumble bees (Bombus spp.), carpenter bees (Xylocopa sp.), small (all other) bees (Hymenoptera), hoverflies (Diptera: Syrphidae) and butterflies (Lepidoptera). Visual observations were recorded on 14 dates for asters in September and October over 2 years, 25 dates for coreopsis from May-August, and 40 dates for salvia from April-October spanning 2 year's growing seasons. Additional hand netting approximately monthly allowed finer taxonomic resolution of some bees visiting the plants. Bees were mounted and identified to the lowest possible taxonomic resolution (Table 1) using a combination of print and online keys (Mitchell, 1960; Bouseman and LaBerge, 1978; http://www.discoverlife.org; Ascher, 2017).

In mid-June 2017, five of the salvias were subjected to nectar sampling. Salvia nemorosa “Steel Blue” was excluded because by that time, it had passed peak anthesis. For comparison purpose, an additional cultivar, Salvia microphylla x greggii “Hot Lips” in nearby demonstration gardens was also included in the analysis. Nectar was allowed to accumulate for 24 h in flowers from which insects were excluded by means of fine netting (organza bags placed over five flower spikes per plant). A hand-held refractometer was used to measure sugar content as degrees Brix (°Bx, grams of sugar in 100 g solution). Following methodology by Hicks et al. (2016) microcapillary tubes (5 μL) were used to remove nectar, with individual flowers yielding 2–3 μL of fluid. The refractometer was rinsed with deionized water and dried after each sample. The number of open flowers was counted and recorded for each plant and total sugar content per plant calculated as a product of sugar and number of flowers.

Visual observations of insect visitors to the cultivars were analyzed for each main plant taxon. The data were analyzed using a generalized linear mixed model (PROC GLIMMIX, SAS Institute Inc., 2013). Differences in least square means were determined by pairwise t-tests (alpha = 0.05) as the multiple comparisons post-hoc test to determine significant differences between levels of all factors. Data from coreopsis, salvia and asters were analyzed separately, and no direct comparison among these main taxa was attempted. Data analysis on salvia flower nectar and number of flowers was performed using ANOVA with mean separation through Tukey's Honestly Significant Difference Test.

Data collected during this two-year study comprised 6,911 pollinator observations across all plant taxa with bees, butterflies and syrphids well-represented (Figures 1–3). Relative abundance of insect taxa across all plant taxa and both years (Figure 4) revealed 62% bees, 4.2 % butterflies, and 33.8% hover flies. Among the bees, 21.7% were honey bees, 7.6% were bumble bees, 9.3% carpenter bees and 23.5% other or small bees. This diversity of insect visitors was represented differently within the broad plant taxa salvia, coreopsis and asters (Figure 4). The most frequent visitors to coreopsis were the small bees with over 77% of visitors falling into this category. Salvia was most frequently visited by honey bees (36.4%) and carpenter bees (24%), although all the groups were represented. Syrphids were the group most commonly observed on asters (58.5%) with nearly 40% of the visitors being bee species.

Of the 325 individual bees collected for greater taxonomic resolution, 23 bee species representing four families and 12 genera were collected between February and October from the aster, salvia, and coreopsis flower cultivars and plots (Table 1). Six species were collected from coreopsis cultivars, 20 species from salvia cultivars and eight species from asters. Additional species collected during preliminary sampling in the plot area prior to regular sampling included Svastra obliqua (Say) on asters, Lasioglossum (Dialictus) imitatum (Smith) on asters and coreopsis, Lasioglossum (Dialictus) mitchelli Gibbs on Coreopsis and Colletes americanus Cresson on asters.

Relative abundance of the six insect groups evaluated varied significantly among cultivars within plant taxa (Figures 5–7). “Hot Paprika” coreopsis was the most frequently visited cultivar by bees and by total flower visitors (P < 0.0001; Figure 5) although all cultivars were visited by the insects of interest. Total bees and total flower visitors were most often and equally observed on “Jane Bath” and “Rachel Jackson” asters among the cultivars evaluated (P < 0.0001; Figure 6). Total bees and total flower visitors were most frequently observed on the Heat Wave series cultivar “Blaze” salvia (P < 0.0001; Figure 7), with “Glitter” being visited least often by total flower visitors.

While the above-mentioned cultivars were the most or least- frequently visited across the trial period, expected seasonal variation did occur. All salvia cultivars, for example, were visited by pollinators during the course of the 2-year study, with frequency of visitation by cultivar not surprisingly also varying by date (P < 0.0001; Figure 8). Cultivar “Steel Blue”, for example, was often most frequently visited earlier in the season, corresponding with bloom. “Blaze”, the overall most frequently visited cultivar, was more often visited as the season progressed compared to early visitation observed on “Steel Blue”.

Nectar sugar content in salvias ranged from 22.9 to 30.4°Bx, and did not differ statistically among the six cultivars (P > 0.05; Figure 9). Number of open flowers ranged from 10 to 39, and was highest in S. “Radio Red”, and lowest in S. “Hot Lips”. When total amount of nectar (product of number of flowers and Bx) was calculated, S. “Hot Lips” had the lowest value, while S. “Radio Red” had the highest (data not shown). Yet, S. “Blaze,” which had the highest number of flower visitors, did not differ significantly from either cultivar with respect to total amount of nectar. Therefore, the higher attraction of pollinators to S. “Blaze” could not be explained by volume of nectar produced per plant. Flower tube length has been shown to be important in impacting the type of pollinators; nectaries at the bottom of longer corolla typically can be accessed by insects with longer tongues (e.g., long-tongued bees and lepidopterans). The corolla width is also important as wider corollas allow access of smaller bees which crawl inside to reach the nectaries. In our study, flower number did vary significantly among salvia cultivars, with “Radio Red” having the highest number, and “Hot Lips” the lowest number. While “Blaze” had the highest number of visitors, it did not differ significantly from either of these cultivars in terms of total nectar volume it produced. The number of flowers were not significantly different between “Blaze” and “Radio Red”. Based on our findings, flower number could not explain the different number of pollinators observed on the salvia cultivars.