Successful reproduction and fruit set of most flowering plants, including valuable food crops and fruit trees that make up 35% of global food production, are dependent on animal pollination, primarily bee pollination (Klein et al., 2007). Pollinating insects can increase fruit or seed quality or quantity of many major crops (Klein et al., 2007). Yields of insect-pollinated crops are often managed by bringing hives of honeybees (Apis mellifera) to the fields to increase pollination. However, wild bee species (Bombus spp.) are also vital to our cropping systems. For over 40 globally important crops, wild pollinators improved pollination efficiency and increased fruit set compared to honeybees alone (Garibaldi et al., 2013). Floral strips can be used to increase plant biodiversity within and surrounding crop fields by providing diverse nutritional resources for wild bee species (Bombus spp.) and to increase pollination in agroecosystems as long as potential negative consequences of insecticide contact are considered (Garibaldi et al., 2013, 2021; Nicholls and Altieri, 2013; Blaauw and Isaacs, 2014). Better understanding pollination cues of major crops could help inform future decisions regarding conservation of wild pollinator abundance and diversity.

Flowers have adapted diverse advertising strategies to attract bees, including flower size, shape, scent, color, and pattern, which often correspond to morphological and sensory characteristics of pollinators. Bees can associate visual cues such as flower color and shape with nectar and pollen rewards and will use these cues to optimize foraging times (Spaethe et al., 2001). Bees are sensitive to ultraviolet (UV), blue, and green wavelengths, and their detection of large flowers (about 15 mm in diameter or larger) is dependent on high contrast between flower and green foliage background color (Spaethe et al., 2001). For detection of small flowers (around 8 mm in diameter or less), wild bees fly closer to the ground and use a different neuronal channel focused on green contrast only. The switch between neuronal channels is best exemplified by search times for yellow and UV-absorbing, white flowers, since yellow flowers have a higher color contrast to the background but about two-thirds the amount of green contrast. Spaethe et al. (2001) found that bumblebees (B. terrestris) took 22% longer to detect large, UV-absorbing white flowers than large, yellow flowers but 72% longer to identify small, yellow flowers than small, UV-absorbing white flowers due to using color contrast to identify large flowers and green contrast to identify small flowers. While flower detection of small and large flowers both involve color contrast, honeybees and wild bee species appear to ignore flower brightness during foraging (Vorobyev and Brandt, 1997; Spaethe et al., 2001).

While flower color is vital in attracting pollinators from a distance, flower color and UV patterns are important to bee orientation and landing (Papiorek et al., 2015). Many plants have adapted to display floral guides, often referred to as nectar guides, that are color and UV absorbance and reflectance flower patterns that direct pollinators to nectar and pollen to increase pollination success (Orbán and Plowright, 2013). In radially symmetric flowers with UV floral guides, generally apical parts of flower petals have UV reflecting carotenoids and the central flower parts have UV absorbing flavonoids creating a “bullseye” pattern (Thompson et al., 1972; Harborne and Smith, 1978). In addition to UV absorbing flavonoids, another type of UV absorbing pigment, dearomatized isoprenylated phloroglucinols, have been identified in some species on the anthers and ovarian wall that perform both a visual and defensive function, such as being a deterrent to insect feeding (Gronquist et al., 2001). While most radially symmetric flowers have radially symmetric floral guides, most bilaterally symmetric flowers have bilaterally symmetric floral guides (Dafni and Kevan, 1996). Whether radially or bilaterally symmetric, these floral guides orient pollinators toward the center of the flower where nectar and pollen are produced.

Peanut (Arachis hypogaea L.) is an annual, self-pollinated crop that is grown worldwide on about 25.9 million ha as an oil, food and feed crop (FAOSTAT, 2019). Like most crops, the predominant goal for peanut breeding is increased yield either directly through selection of genes that increase yield or indirectly such as through improved yield retention via increased biotic and abiotic stress tolerance and/or resistance. Wild Arachis species are important as donors of strong, durable resistances to cultivated peanut (Stalker, 2017). Previous reports found high bee visitation to peanut flowers and that visitation by large bee species that trip flowers increased peanut yields in some U.S. varieties (Girardeau and Leuck, 1967; Rashad et al., 1979). A more recent report found bees did not increase yield in newer peanut varieties grown in north Queensland, Australia, likely due to low bee visitation, and suggested that selection for other desirable traits may have resulted in development of varieties less attractive to flower-tripping bees (Blanche et al., 2006). There is no recent study testing the effect of bee pollination on peanut yield of U.S. cultivars. Most Arachis species, including cultivated peanut, are self-pollinating species and are not exclusively dependent on bee pollination for seed set. However, peanut has been documented as attractive to honeybees, many wild bee species, and other insect pollinators with reported cross-pollination rates between 1 to 10% (Culp et al., 1968; Knauft et al., 1992), despite lacking a nectary, meaning insect visitation is likely for pollen foraging (Vogel, 1997; Hammons and Leuck, 1966; Leuck and Hammons, 1969; Blache et al., 2005; USDA, 2015). However, there have been no studies in Arachis evaluating the nutritional quality and quantity of pollen as well as no study documenting the presence or absence of nectar using a nectar extraction method. Without these studies, it is unknown whether pollinators are benefitting from Arachis flowers or being manipulated into exerting time and energy to foraging reward-less flowers by false nectar guides, which are floral guides that attract pollinators to flowers that have no nectar nor quality pollen resource (Slater and Calder, 1988; Lunau and Webster, 2017). If Arachis species are found to have nutritious pollen, high-density flowering species such as A. glabrata, A. repens, and A. pintoi that are used as summer legume forage and cover crops and as urban lawns, could be beneficial for increasing plant biodiversity to provide habitats for wild bees in urban landscapes (Anderson et al., 2015; Kröning et al., 2019).

Despite the documentation of bee-pollination in Arachis and the extensive morphological characterization of Arachis that has been performed, UV reflectance pattern diversity in cultivated peanut and other species in the Arachis genus has not been documented. Since Arachis species generally develop yellow and orange flowers, and yellow flowers are more likely than other colored flowers to contain UV floral guides (Horovitz and Cohen, 1972; Guldberg and Atsatt, 1975; Primack, 1982), it is likely that peanut species have UV reflectance patterns. Furthermore, given the wide pool of genetic and phenotypic variation in the Arachis genus (Kochert et al., 1996; Moretzsohn et al., 2005, 2013), different species may have distinct flower UV reflectance patterns. This study documented flower UV absorbance and UV reflectance levels and patterns as well as absorption levels at visible-light wavelengths, including at blue and green wavelengths, in cultivated peanut, diploid Arachis species, wild Arachis species-derived allotetraploid interspecific hybrids that are cross-compatible to peanut, and BC₁F₃ hybrids derived from crosses between cultivated peanut and allotetraploids. Our results indicate that UV floral guides of various sizes are present in wild Arachis species, allotetraploids, and A. hypogaea genotypes.

Field versus greenhouse environments had a significant effect on area of UV absorption on the banner (adaxial), area of UV absorption on the banner as a percentage of banner area (adaxial), area of UV absorption on the left-wing (adaxial), and area of UV absorption on the left-wing as a percentage of left-wing area (adaxial) (Table 2). No environmental effect was observed on banner area, left-wing area, area of UV absorption on the left-wing (abaxial), nor area of UV absorption on the left-wing as a percent of left-wing area (abaxial). In addition, no flowers collected from the field nor the greenhouse had UV absorption on the abaxial side of the banner. Flowers from the field had greater total area of UV absorption on the banner (adaxial) and area of UV absorption as a percent of banner area (adaxial) than flowers from the greenhouse (Fig. S2). Similarly, flowers from the field had greater total area of UV absorption on the left-wing (adaxial) and area of UV absorption as a percent of left-wing area (adaxial) than flowers from the greenhouse (Fig. S3). The UV absorption pattern on the adaxial side of the banner remained consistent within each genotype (Fig. S4). The A. hypogaea genotypes NC 3033 and Tifrunner had UV absorption areas on the banner that were shaped similarly to a vertical cross section of a mushroom while the allotetraploids had heart-shaped cross-sections whether grown in the field or greenhouse (Fig. S4).

Genotype had a significant effect on banner area, left-wing area, and all parameters of UV reflection on Arachis flowers collected from the greenhouse (Table 2). IpaCor2 and Cor had the largest banner area at 2.11 and 2.23 cm², respectively, which were numerically larger than the A. hypogaea genotypes but only statistically larger than Dur, Mag, and Sten10309, which had banner areas of 0.68, 0.96, and 0.97 cm², respectively (Fig. S5A). The only genotype with significantly different banner area from A. hypogaea was Dur. IpaCor2 and Cor had the largest left-wing area at 0.51 and 0.56 cm², respectively, and the latter was significantly greater than all four A. hypogaea genotypes (Fig. S5B). Only IpaCor2 and Cor were significantly different from at least one A. hypogaea genotype for left-wing area. All the genotypes had yellow wings, but only ValSten1, Sten410, and Sten10309 had completely yellow banners (Fig. S6). The remaining genotypes had banners with a yellow center and orange edge, though the intensity of the orange varied. All genotypes except Ipa, Mag, and two of the three Cor plants had thin red stripes originating from where the hypanthium and banner connect and spread outwards on their banners.

Numerically, the four A. hypogaea genotypes had the largest total area of UV absorption on the adaxial side of the banner as well as the largest UV absorption area as a percent of banner area (Fig. 1). 13-2113 had the largest total area of UV absorption at 0.56 cm² and Tifrunner had the largest UV absorption as a percent of banner area at 32.5%. The total area of UV absorption of the four A. hypogaea genotypes was not significantly larger than the allotetraploids with the exception of Tifrunner and 13-2113 being significantly larger than IpaDur3; however, they were significantly larger than all the diploid, wild Arachis species except Val. All the wild Arachis species had total UV absorption areas smaller than half of the total UV absorption areas of the A. hypogaea genotypes. The areas of UV absorption as a percentage of banner area of the four A. hypogaea genotypes were not significantly larger than the allotetraploids IpaCor2 and ValSten1, but most were significantly larger than IpaDur3. Similar to total UV absorption, the areas of UV absorption as a percent of total banner area of the A. hypogaea genotypes were significantly larger than most wild Arachis species. As previously mentioned, NC 3033 and Tifrunner have mushroom-shaped UV absorption areas with wings on the widest section of the UV absorption area that extended downwards towards the hypanthium (Fig. 2). These characteristic wings were absent in 13-1014 and 13-2113, which had a more intermediate UV pattern between the mushroom shape and the heart shape of the allotetraploids and some wild species (Fig. 2). Figure 1

Figure 1. (A) Area of UV absorption on the banner (adaxial) and (B) area of UV absorption as a percentage of banner area (adaxial) of A. hypogaea genotypes (orange, vertical stripes), allotetraploids (solid, yellow), and wild Arachis species (horizontal, pink stripes) collected from the greenhouse.

Arachis species, including peanut, may provide pollen for wild bee species; however, visual cues that bees can associate with pollen rewards, such as contrasting color and UV absorption and reflectance on flowers, have not been characterized in Arachis. Since yellow flowers are more likely than other colored flowers to contain UV floral guides (Horovitz and Cohen 1972; Guldberg and Atsatt 1975; Primack, 1982) and Arachis species generally have yellow or orange and yellow flowers, we hypothesized that peanut and Arachis species would have UV reflectance patterns. This study found that all genotypes, including four A. hypogaea genotypes, eight wild Arachis species, 19 unique allotetraploids, and four BC₁F₃ lines, had UV floral guides. These UV floral guides were bilaterally symmetrical consistent with previous findings that bilaterally symmetric flowers have bilaterally symmetric floral guides (Dafni and Kevan, 1996). All genotypes had UV absorption located centrally on the banner with UV absorption extending towards the junction of banner and hypanthium and had UV absorption on the distal tips of the abaxial and adaxial sides of the wings.

While all studied genotypes had UV floral guides, the size and shape of these floral guides varied. In general, A. hypogaea genotypes had greater total UV absorption and UV absorption as a percent of banner area than the wild Arachis species. This same trend occurred for UV absorption on the abaxial side of the wings. This complements the finding of Horth et al. (2014), who found that mean UV floral guide size was smaller in wild populations as compared to cultivated populations of Rudbeckia hirta. It is unknown if larger UV absorption on the adaxial side of the banner and abaxial side of the wings increases or decreases Arachis flower conspicuousness for detection by bees or if it affects bee landing and orientation after detection. In other species, when UV floral guides are completely obscured in both radially and bilaterally symmetric flowers, bee visitation decreased (Peter and Johnson, 2008; Klomberg et al., 2019). However, the effect of altering the size of UV absorption and reflectance without completely eliminating the UV pattern on bee visitation frequency has not been extensively studied, especially in bilaterally symmetrical flowers. In radially symmetrical flowers, increasing the area of UV absorbance of Hypoxis camerooniana flowers did not affect bee visitation, while increasing the area of UV absorbance of R. hirta, R. flugida, and Argentina anserina flowers did increase bee visitation (Horth et al., 2014; Rae et al., 2013; Klomberg et al., 2019).

Increased UV absorption on the banner of cultivated peanut flowers as compared to wild Arachis species flowers may in part be a consequence of polyploidy or an evolved trait. Allotetraploidization of Arachis species increases leaf photosynthetic pigment production (Leal-Bertioli et al., 2017), but its effect on flower pigment production has not been studied. In this study, IpaDur3, IpaCor2, and ValSten1 and the diploid, wild Arachis species from which they were derived were tested in the greenhouse environment. In all comparisons, the allotetraploids had greater mean total UV absorption and UV absorption as a percent of banner and greater mean total UV absorption and UV absorption as a percent of left-wing (adaxial and abaxial) than the wild Arachis species from which they were derived, except IpaDur3 which had a smaller UV absorption area as a percent of banner area than Dur. Though these increases in UV area were only significant for IpaCor2 as compared to Ipa and Cor for UV on the adaxial and abaxial side of the left-wing, this trend suggests polyploidization may increase flower UV pigment production. A follow-up study with greater genotypic replication may yield significant differences between additional allotetraploids and diploid, wild Arachis species to further test if polyploidization increases flower UV pigment production.

Larger UV absorption areas in A. hypogaea could be an evolved trait. All Arachis genotypes had “large” flowers, greater than 8 mm in diameter as defined by Spaethe et al. (2001), meaning bees likely use the color-contrast neuronal channel visualizing blue, green, and UV wavelengths to detect Arachis flowers. Therefore, flowers with a greater contrast to a foliage background color should be most conspicuous to bees. A small area of UV absorption surrounded by a large area of UV reflectance characteristic of wild Arachis species may stand out more in UV-absorbing foliage as compared to flowers with a large area of UV absorption surrounded by a small area of UV reflectance more characteristic of the observed A. hypogaea genotypes (Fig. S9). Further studies elucidating the relationship between UV absorption and UV reflection area with bee flower detection and visitation on bilaterally symmetric flowers are needed.

A study to determine the effect of UV absorption area on bee visitation in Arachis could use UV-absorbing chemicals such as butyl methoxydibenzoylmethane and ethylhexyl methoxycinnamate to manipulate the size of UV absorbance to various degrees in a few genotypes similar to the study performed by Klomberg et al. (2019). Using unscented, UV-absorbing chemicals would keep other flower cues consistent, such as flower shape and size, while only altering UV reflectance area. If increased UV absorption area correlated with increased bee visitation, larger UV area could have led to greater fitness in early cultivated peanut fields by promoting outcrossing and lineage recombination between landraces. In contrast, if decreased UV absorption area correlates with decreased bee visitation, it is possible that selection for other desirable traits may have inadvertently resulted in landraces less attractive to bees, similar to the suggestion of Blanche et al. (2006) that modern peanut varieties may have inadvertently been selected to be less attractive to flower tripping bees. However, further research is needed to elucidate the relationship between UV absorption area and bee attractiveness in Arachis.

For large flowers (> 8 mm), UV is not the only wavelength bees utilize to identify flowers. Blue and green color contrast with foliage in Arachis would increase flower conspicuousness as well. All genotypes tested for banner pigment had significant levels of absorption at wavelengths of 380, 415, 450, 473 and 495 nm, but not for wavelengths in the green interval. Levinson et al. (2021) reported a similar result for banner pigmentation of peanut breeding line 13-2113 and four allotetraploids, IpaCor2, IpaDur3, IpaSten10309, and ValSten1. UV absorption detection at 380 nm was consistent with the finding of UV absorption in the banners in the UV photographs. Flowers from peanut breeding lines, allotetraploids, and wild Arachis species absorb blue wavelengths (450, 473 and 495 nm) and reflect green wavelengths, which would contrast with Arachis foliage that absorbs primarily green wavelengths. In a future study, it would be interesting to test two additional camera filters, one that only allows blue and one that only allows green wavelengths to pass through, and to stack UV, blue, and green wavelength photographs of Arachis flowers still attached to the plant to better understand the contrast of all wavelengths between Arachis flowers and foliage.

For flower detection by bees, it is critical to consider flower visibility beyond just flower color. First, flower size affects flower detection time, in which flowers under 8 mm in diameter take wild bees a longer time to find (Spaethe et al., 2001). All germplasm in this study had flowers with diameters greater than 8 mm and their banner areas were generally similar. Therefore, banner size differences may not cause differences in bee detection of flowers of different Arachis species. Another aspect of flower visibility is the growth habit of the plant coupled with flower morphology. Wild Arachis species and allotetraploids derived from wild Arachis species have prostrate growth, in which branches grow flat along the ground (Stalker et al., 2016). This growth habit coupled with the long hypanthium area characteristic of wild Arachis species and allotetraploids allow flowers to rise above the foliage of the plant and is likely easier to detect by bees (Stalker et al., 2016; Levinson et al., 2021). In contrast, domesticated peanut has been selected for more compact growth, with shorter branches that curve upwards and flowers have a shorter hypanthium than wild Arachis species (Stalker et al., 2016; Levinson et al., 2021). Thus, flowers are more likely to be at least partially hidden from view by plant foliage.

Flower production intensity and duration of different genotypes must also be considered when determining their attractiveness to bees. Positive alleles for flowering precocity have been identified in some allotetraploids (Foncéka et al., 2012), and allotetraploids with significantly higher flower counts over the growing season as compared to cultivated peanut have been identified (Levinson et al., 2021). Furthermore, Levinson et al. (2021) found that allotetraploids made with perennial species such as A. correntina and A. stenosperma do not have a decline in flower production. However, cultivated peanut had a decline in flower production from its maximum of about 20 flowers/day from 52 to 66 days after transplanting to only 5 flowers/day at 87 days after transplanting. This is a large difference compared to the flower production of IpaCor allotetraploids that produced over 40 flowers/day from 52 to 108 days after transplanting (Levinson et al., 2021). In general, cultivated peanuts may have fewer flowers that may be harder to detect than some allotetraploid and diploid, wild Arachis species; however, studies including pollinator counts are needed to determine differences in bee visitation between wild Arachis species, allotetraploids, and cultivated peanut. Such studies may indicate some wild Arachis species and allotetraploids as good candidates for floral strips in agricultural fields and for ornamental use in urban landscapes if Arachis species are found to have abundant protein-, lipid-, and micronutrient-rich pollen.

Another finding of the study was a significant effect of the two different environments (field environment from 18 August to 18 September 2020 and greenhouse environment from 9 November 2020 to 26 February 2021) on the area of UV absorption on the banner, area of UV absorption on the banner as a percentage of banner area, area of UV absorption on the left-wing (adaxial), and area of UV absorption on the left-wing as a percentage of left-wing area (adaxial). However, no environmental effect was observed on banner area, left-wing area, area of UV absorption on the left-wing (abaxial), nor area of UV absorption on the left-wing as a percent of left-wing area. Therefore, while flower size remained similar between genotypes in the two different environments, variation of UV absorption on the banner and on the adaxial side of the wings can be displayed. This same trend of larger UV absorbing areas on flower petals grown in the field as compared to those grown in the greenhouse was also observed in Brassica rapa, in which Sasaki and Takahashi (2002) hypothesized this difference was due to enhancement of flavonoid biosynthesis caused by greater UV-B exposure in the field as compared to the greenhouse. Due to differences in both collection timing and environment in this study, the cause of UV absorbance variation in Arachis in these different environments could be due to UV-B exposure or other factors such as light quality and quantity, temperature, plant developmental stage, among others, that have been documented to affect flavonoid production in various plant tissues (Ghasemzadeh et al., 2010; Hemm et al., 2004). A study keeping all environmental factors constant while only varying UV-B exposure is needed to determine if increased UV-B light exposure increases production of UV-absorbing flavonoids in flowers.

Honeybees and wild bee species provide essential pollination services to improve yields for many fruit, nut, vegetable, and seed crops, yet poor nutrition from habitat loss, large monocultures, and insecticide exposure can result in honeybee colony declines. Pollen abundance and diversity are important to the growth and development of all bees by providing protein and essential amino acids while nectar provides carbohydrates, vitamins, minerals, and amino acids. Bridging the knowledge gap on crop nectar and pollen nutrition and on what contributes to flower attractiveness of various crops, including attractiveness caused by area and pattern of UV absorption, can lead to better decision-making regarding honeybee health and wild pollinator conservation near and within crop fields. Peanut has been documented to be attractive to honeybee and wild bee pollinators but little more of its relationship to bee pollinators is known. This study is one step towards narrowing this knowledge gap, and it found that UV floral guides of various sizes were present in wild, diploid Arachis species, synthetic allotetraploids, and A. hypogaea. The implications of the presence and size of these floral guides on bee visitation are dependent on the pollen quality and quantity in peanut and Arachis species. If future studies find Arachis species to be a source of high-quality pollen, then high-density flowering Arachis species and allotetraploids could be evaluated for promoting wild bee abundance and diversity in urban landscapes, especially those species already used as water-efficient lawns and as ornamental plants.