Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 27 2024 PRAIRIE NATURALIST 56:27–41 Floral Composition of Pollen Collected from two Rusty Patched Bumble Bee (Bombus affinis, Cresson) Nests in Southeastern Minnesota Michael P. Simanonok1, Elaine Evans2, Clint R.V. Otto1,*, Robert S. Cornman3, Deborah D. Iwanowicz4, and Tamara A. Smith5 Abstract – Understanding the forage diets of imperiled bumble bees can improve conservation planning and habitat restoration efforts. In this study, we describe the taxonomic composition of beecollected pollen from 2 Rusty Patched Bumble Bee (Bombus affinis, Cresson) nests located in southeastern Minnesota. This is the first published reporting of pollen collected from active B. affinis nests. We also compared pollen identification via traditional palynological light microscopy with genetic identification via ITS metabarcoding. Among the 49 pollen samples analyzed, we detected 41 and 56 distinct taxa via light microscopy and metabarcoding, respectively. Furthermore, 27 of 47 total genera overlapped between the 2 methods. Bittersweet Nightshade (Solanum dulcamara, Linnaeus) was the most detected species for both metabarcoding and microscopy identification for pollen prevalence. Pollen volume from the microscopy data showed that Lesser Burdock (Arctium minus, Bernhardi), Alfalfa (Medicago sativa, Linnaeus), Bittersweet Nightshade (Solanum dulcamara, Linnaeus), Plumeless Thistle (Carduus acanthoides, Linnaeus), and Red Clover (Trifolium pratense, Linnaeus) together comprised more than half of the total volume of pollen. Light microscopy and metabarcoding revealed compositionally distinct plant communities when analyzed at the species level. Methodological concordance improved when analyzing pollen data at genus level, but both methods still reveal marginally distinct groupings. Our study highlights specific plant taxa that are important components of B. affinis pollen diets and provides actionable research for conservation efforts in urban systems. Our study also supports that B. affinis is a generalist forager and will collect pollen from a variety of native and non-native host plants. Introduction Bombus affinis (Cresson) (Rusty Patched Bumble Bee) was granted federal protection under the Endangered Species Act in 2017 due to precipitous declines across its historical range (Colla and Packer 2008, Szymanski et al. 2016). The species used to be relatively common across the Midwest and eastern United States but is now confined to isolated populations in parts of the Upper Midwest and Appalachia (Szymanski et al. 2016). Loss of habitat and floral resources has been implicated in the decline of multiple pollinators, including B. affinis and closely related species, such as Bombus occidental Greene (Western Bumblebee) (Evans et al. 2008, Goulson et al. 2015, Graves et al. 2020). The rapid decline 1U.S. Geological Survey, Northern Prairie Wildlife Research Center, 8711 37th St. SE, Jamestown, ND 58401. 2University of Minnesota, Department of Entomology, 1980 Folwell Ave, Saint Paul, MN 55108, elainee@umn.edu. 3U.S. Geological Survey, Fort Collins Science Center, 2150 Centre Ave., Bldg C, Fort Collins CO 80526; rcornman@usgs.gov; https://orcid.org/0000-0001-9511- 2192. 4U.S. Geological Survey, Environmental Health Program Ecosystems Mission Area, 11649 Leetown Road, Kearneysville, WV 25430; diwanowicz@usgs.gov. 5U.S. Fish and Wildlife Service, Minnesota-Wisconsin Ecological Services Field Office, 3815 American Blvd., Bloomington, MN 55425; tamara_smith@fws.gov. *Corresponding Author: cotto@usgs.gov, 701-368-9028, https:// orcid.org/0000-0002-7582-3525. Associate Editor: Joshua Campbell, Northern Plains Agricultural Research Laboratory Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 28 of B. affinis and other related species has triggered scientific investigations that employ pollen microscopy, host-plant interaction records, and pollen metabarcoding to infer floral resource use of B. affinis across North America to aid in conservation efforts (Hepner et al. 2024, Otto et al. 2023, Simanonok et al. 2021, Wolf et al. 2022, Wood et al. 2019). Like all bumble bees, B. affinis collects both pollen and nectar from a variety of host plants. Nectar provides carbohydrates, whereas pollen provides protein, fatty acids, and micro-nutrients needed to raise offspring (Di Pasquale et al. 2013, Lau et al. 2022). Recent evidence has shown historical pollen diets of B. affinis have been surprisingly constant over the past 100 years, suggesting that factors other than shifts in the floral resource community are responsible for declines (Simanonok et al. 2021). Several studies have revealed that pollen diets of B. affinis over the past century consisted largely of common native and non-native forbs (Simanonok et al. 2021, Wood et al. 2019). Specifically, photographic records analyzed from community science data identified B. affinis host-plant interactions and determined the most visited taxa were in the genus Monarda, followed by Eutrochium purpureum Linnaeus (Joe Pye Weed), Eutrochium. Maculatum Linnaeus, and Veronicastrum virginicum Linnaeus (Culver’s Root) (Wolf et al. 2022). This investigation revealed intriguing insights into the forage diet of B. affinis, yet the authors highlight that inferring pollen diets from photographic evidence is challenging because bumble bees may collect pollen from multiple forb species during a single foraging event. Additionally, much of our scientific understanding of B. affinis pollen foraging stems from analysis of pollen from historical B. affinis museum specimens, meaning biologists lack information on presentday pollen diets. This is problematic considering many museum specimens of B. affinis were collected in rural areas, whereas the current documented distribution of the species is primarily confined to urban and suburban areas of the Upper Midwest (USFWS 2021). An analysis of contemporary forage patterns of B. affinis could help inform future conservation efforts, particularly in urban and suburban areas where the species may exhibit unique forb selection choice relative to more semi-natural landscapes. Furthermore, baseline data on pollen species that are being stored in B. affinis nests are lacking (Boone et al. 2022) and this information need is highlighted in the B. affinis Recovery Plan (USFWS 2021). Although B. affinis nests have only been discovered in a few cases over the past 30 years, scientists have documented natural history information when nests are discovered, including habitat use, colony activity, and nest architecture (Boone et al. 2022). Although no formalized protocols have been developed for documenting natural history of B. affinis at nesting sites, understanding the floral resources being brought back to the nest by workers is valuable information to obtain from these rare discoveries. In addition, discovery of Bombus nests gives researchers an opportunity to collect valuable colony-level information such as nest site selection data, colony size and survival estimates, behavioral traits, and biological samples to aid in conservation genetics and disease ecology (Boone et al. 2022, Lye et al. 2012). With this study we analyzed pollen collected from workers returning to 2 active B. affinis nests located in southeastern Minnesota, recently described by Boone et al. (2022). Our primary goals were: 1) describe the taxonomic composition of modern B. affinis pollen samples, and 2) compare the identification of pollen samples via traditional palynological light microscopy with genetic identification via ITS metabarcoding. To our knowledge, this is the first reporting of pollen collected from active B. affinis nests. Although our results and inference are limited due to the small number of B. affinis colonies and the limited timeframe of data collection, it represents a unique opportunity to understand contemporary forage patterns of this endangered species. Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 29 Methods Pollen samples originated from 2 nests described by Boone et al. (2022). Specifically, 1 nest was located between a concrete foundation and layers of insulation at a residential home in Red Wing, Minnesota, USA and another was located in a rodent burrow at a residence in Minneapolis, Minnesota, USA. Approximately 90% and 100% of the areas within 1 km of the nests were classified as ‘developed’ by the National Land Cover Database for the Red Wing and Minneapolis nests, respectively (Boone et al. 2022). Corbicular pollen samples were removed from returning workers at the Red Wing nests on 4 dates between 15 July and 10 August 2020, resulting in 47 samples. Meanwhile, pollen samples were collected from workers at the Minneapolis nest on 11 August 2020, resulting in 2 samples (Fig. 1, Table 1). Workers returning to the nest were netted and placed on ice until they were immobile. To avoid depleting colony resources, only a single pollen ball was removed from each individual worker. Pollen was removed from 1 leg using a pair of forceps and placed in a vial. Forceps were cleaned with 80% alcohol between bees. Occasionally, pollen fell off the bee during netting, so the pollen was collected and placed into the vial from the net, and the bee was released without chilling. After 10 August, activity at the Red Wing colony lessened and we stopped pollen collection to reduce potential impact on the colony’s ability to reproduce. Only 2 samples were collected from the Minneapolis nest on the first day of observation on 11 August, as it was assessed during these observations that colony activity was low. After collection, each individual pollen load was dried and weighed. The pollen was then split in half, with 1/2 analyzed using light microscopy and 1/2 analyzed using metabarcoding, thereby resulting in 49 paired samples. All activities in Minnesota were conducted under the authority of Endangered and Threatened Species Figure 1. A Bombus affinis worker with 1 corbicular pollen ball removed. Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 30 Permit FWS/AES-TE 16-07-3a. This permit allowed for the collection of all pollen from 1 pollen basket, per worker, from no more than 30 workers observed within a 0.5 km radius within a 15-day period. Light microscopy and metabarcoding Researchers typically rely on pollen microscopy palynology and metabarcoding to identify pollen collected by foraging bees (Cornman et al. 2015, Richardson et al. 2015, Wood et al. 2019). Historically, light microscopy palynology was used to identify pollen species against a reference library. More recently, researchers have developed molecular methods to determine species composition of bee-collected pollen, particularly ITS metabarcoding (Richardson et al. 2015). Generally, the results between microscopy and metabarcoding agree at a broad scale; however, both methods have their own strengths and weaknesses, and taxonomic resolution can vary across methods (Richardson et al. 2015, Smart et al. 2017). Although our main goal was to understand the forage ecology of B. affinis, we also were able to compare the results of pollen identification across thes e 2 widely used methods. The method used for identification and quantification of pollen via light microscopy was described in detail by Jones (2012). Briefly, each pollen load was homogenized in glycerine. One drop (approximately ~0.5mL) was removed and placed on a microscope slide and stained with 1 drop of Calberla’s solution (approximately 50 μL). The proportion of the total pollen load was estimated by counting all grains on 1 slide when less than 1,000 grains were present or 1,000 grains across sectors spaced out across the slide. Prior to counting, pollen grains were identified at 400x by comparing grains with pollen reference slides housed at the University of Minnesota Bee Lab as well as through comparison with images and descriptions available in online databases and references (Crompton and Wojtas 1993, Martin and Harvey 2017, PalDat 2023). Pollen grains that were not identifiable from reference collections or online databases were compared to images of species identified by metabarcoding (see below) to aid identification. Total estimated pollen volume across all samples was calculated by adding together the estimated volumes for each pollen type. Volume for each pollen type was estimated by using average grain sizes from references (Crompton and Wojtas 1993, Martin and Harvey 2017, PalDat 2023) to calculate the volume of a spheroid, calculating the proportional volume of pollen types within each pollen sample, and multiplying that proportional volume by the weight of that sample. Pollen metabarcoding was performed as described in Simanonok et al. (2021). Briefly, pollen DNA was extracted using a modification of methods outlines in Doyle (1991), followed by ITS2 ‘pre-amplification’ with the primers of Sickel et al. (2015). Sequencing Table 1. Number of Bombus affinis pollen samples collected from 2 nests in southeastern Minnesota in 2020. Date Number of Samples Nest location July 15 20 Red Wing July 22 10 Red Wing July 31 11 Red Wing August 10 6 Red Wing August 11 2 Minneapolis Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 31 libraries were generated from an initial pre-amplification PCR with unmodified primers, performed in triplicate and pooled to reduce stochastic variation. Pools were then processed according to Illumina’s amplicon protocol (Illumina 2023) with modified ‘fusion’ primers, as described in Simanonok et al. (2021), and then individually labeled (multiplexed) with Nextera dual indexes (Illumina). Libraries were sequenced on an Illumina MiSeq v.3 600-cycle cartridge to create 300-bp paired-end reads. Sequencing output was deposited in the Short Read Archive of the National Center for Biotechnology Information (NCBI) under PRJNA641863. Operational taxonomic units (OTUs), clusters of similar sequences that are assumed to derive from a single taxon and for which a single representative sequence is analyzed, were generated by clustering quality-filtered amplicon reads at 98% with VSEARCH (Rognes et al. 2016). OTU clusters were then processed with sequence-error correction algorithms in VSEARCH to ‘denoise’ and remove chimeras. Cluster representatives were aligned to the nucleotide (NT) database of the NCBI and a lowest common ancestor (LCA) taxonomic assignment approach (Huson et al. 2007) was applied to the list of high-scoring pairs (HSPs) for each Operational Taxonomic Unit (OTU). The assigned taxonomy was the LCA of all taxa matching within 3% of the highest bit score for that OTU, limited to standard taxonomic ranks, and with additional stringency required for species and genus level matches (see Simanonok et al. 2021 for details). We used Integrated Taxonomic Information System for plant and bee nomenclature (https://www.itis.gov/). Data analysis Due to the small sample size from the Minneapolis nest and the ecological similarity between the 2 nest sites, we grouped the Red Wing and Minneapolis pollen samples for analysis. We set a detection threshold of 2% of the total number of assigned values per sample (Simanonok et al. 2021) for both microscopy and metabarcoding datasets. The goal of this threshold was to reduce possible sources of error (e.g., non-target, wind-pollinated species) while treating both datasets equitably. In addition, we performed taxonomic comparison between the microscopy and genetic datasets in 2 ways. First, we compared the identities of detected pollen species as identified by each method. Our genetic methods, for example, are not inclusive across taxonomic levels, such that Asteraceae and Carduus acanthoides (Linnaeus) may be detected, while a genus-level Carduus spp. is not. Furthermore, taxonomic resolution may vary between microscopy and genetic methods (e.g., some Asteraceae species may be more difficult to discern under microscopy). Thus, we performed a second comparison where we removed all family-level identifications (these were overwhelmingly from the genetic data) and analyzed all identifications at the genus level. Such binning to a higher taxonomic resolution has been helpful in similar methodological comparisons (Simanonok et al. 2023). We performed permutational multivariate analysis of variance (Oksanen et al. 2019) comparing log-ratio transformed pollen taxa composition between our microscopy and genetic datasets for both the original data and then again with the genusbinned datasets. Results After applying our 2% threshold, we identified 41 distinct taxa with microscopy (Table 2). Of these taxa, 1 was unique but could not be identified and was considered “unknown” (only 100 total grains counted out of the initial 41,212; Table 2). The most prevalent families included Fabaceae, Asteraceae, and Solanaceae (Table 2). Some Lamiaceae could not be identified below family, while 21 taxa were identified to genus, and 17 taxa were identified Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 32 Table 2. Plant taxa, identified among 49 Bombus affinis pollen samples via light microscopy. Prevalence is the percent of 49 samples where a plant taxon was detected. Volume of each pollen type was estimated by using average grain sizes from references to calculate the volume of a spheroid, calculating the proportional volume of pollen types within each pollen sample, and multiplying that proportional volume by the weight of that sample. Taxa Family Number of Samples Present Prevalence (%) Volume (%) Solanum dulcamara Solanaceae 19 38 9 Trifolium pratense Fabaceae 11 22 8 Trifolium repens Fabaceae 10 20 2 Arctium minus Asteraceae 8 16 19 Daucus carota Apiaceae 8 16 2 Campanula spp. Campanulaceae 5 10 1 Eutrochium perfoliatum Asteraceae 5 10 0 Brassica spp. Brassicaceae 4 8 3 Hydrangea spp. Hydrangeaceae 4 8 2 Medicago sativa Fabaceae 4 8 14 Ageratina altissima Asteraceae 3 6 3 Liatris spp. Asteraceae 3 6 2 Melilotus spp. Fabaceae 3 6 2 Hosta spp. Liliaceae 2 4 3 Lotus corniculatus Fabaceae 2 4 0 Silphium perfoliatum Asteraceae 2 4 4 Allium spp. Liliaceae 1 2 1 Astilbe spp. Saxifragaceae 1 2 0 Brassica rapa Brassicaceae 1 2 4 Carduus acanthoides Asteraceae 1 2 8 Echinacea spp. Asteraceae 1 2 4 Eupatorium spp. Asteraceae 1 2 0 Eutrochium purpureum Asteraceae 1 2 0 Geranium spp. Geraniaceae 1 2 0 Helenium spp. Asteraceae 1 2 0 Hypericum perforatum Clusiaceae 1 2 0 Hypericum spp. Clusiaceae 1 2 0 Impatiens capensis Balsaminaceae 1 2 0 Lamiaceae Lamiaceae 1 2 0 Lotus spp. Fabaceae 1 2 0 Monarda fistulosa Lamiaceae 1 2 0 Nepeta cataria Lamiaceae 1 2 3 Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 33 to species (Table 2). Metabarcoding resolution was comparable, as we identified 56 taxa comprised of 4 family-level matches (Apiaceae, Asteraceae, Crassulaceae, and Solanaceae), 23 genera, and 29 species (Table 3). The number of taxa identified per sample were similar for both methods with 4.31 ± 0.28 taxa per sample from microscopy identification and 4.57 ± 0.32 taxa per sample with genetic identification. Among these genera and species, we identified 11 taxa that to our knowledge have not been previously reported as forage for B. affinis: Ageratina Adenophora (Sprengel), Allium thunbergii (Don), Astilbe chinensis (Maximowiez), Borago officinalis (Linnaeus), Campanula rapunculoides (Linnaeus), Helenium autumnale (Linnaeus), Mentha spp., Rhodiola spp., Silphium perfolatum (Linnaeus), Solanum lycopersicum (Linnaeus), and Sorbaria spp. When results were scaled to the genus-level within each method, there were 47 total genera detected, with 27 of those being shared by both methods (57.4%). Five genera were unique to microscopic identification (Geranium spp., Hosta spp., Monarda spp., Sambucus spp., and Vicia spp.) and 13 were unique to metabarcoding (Acer spp., Actaea spp., Borago spp., Cirsium spp., Laportea spp., Mentha spp., Parthenocissus spp., Plantago spp., Rudbeckia spp., Sagittaria spp., Sonchus spp., Urtica spp., and Veronicastrum spp.). Solanum dulcamara (Linnaeus) was the most detected species across both genetic and microscopy identification (Fig. 2). The genera identified as the next most abundant taxa were similar (Trifolium spp. and Arctium spp.); however, each method differed in which species it identified (Fig. 2). For example, genetic methods identified Trifolium repens (Linnaeus) as the second-most prevalent, while microscopy identified Trifolium pratense (Linnaeus) (Fig. 2). This lack of concordance at the species level is not unexpected for speciose genera that vary morphologically within species. Genetic and microscopic analyses yielded compositionally different samples (F1,97 = 3.51, R2 = 0.03, P < 0.001). Compositional agreement between genetic and microscopic methods improved when species-level identifications were standardized at the genus level (F1,97 = 1.46, R2 = 0.01, P = 0.04); however, both methods produced marginally different compositions. Taxa Family Number of Samples Present Prevalence (%) Volume (%) Rosa spp. Rosaceae 1 2 1 Sambucus spp. Caprifoliaceae 1 2 0 Sedum spp. Crassulaceae 1 2 0 Solidago spp. Asteraceae 1 2 0 Sorbaria spp. Rosaceae 1 2 1 Sparganium spp. Sparganiaceae 1 2 4 Unknown 2 NA 1 2 0 Unknown 1 NA 1 2 0 Vicia spp. Fabaceae 1 2 1 Table 2, continued. Plant taxa, identified among 49 Bombus affinis pollen samples via light microscopy. Prevalence is the percent of 49 samples where a plant taxon was detected. Volume of each pollen type was estimated by using average grain sizes from references to calculate the volume of a spheroid, calculating the proportional volume of pollen types within each pollen sample, and multiplying that proportional volume by the weight of that sample. Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 34 Table 3. Plant taxa, identified among 49 Bombus affinis pollen samples, via metabarcoding. Prevalence is the percent of 49 samples where a plant taxon was detected. Taxon Family Number of Samples Present Prevalence (%) Solanum dulcamara Solanaceae 21 42 Trifolium repens Fabaceae 19 38 Arctium lappa Asteraceae 15 30 Trifolium spp. Fabaceae 15 30 Hydrangea arborescens Hydrangeaceae 14 28 Hydrangea spp. Hydrangeaceae 14 28 Solanaceae Solanaceae 11 22 Medicago sativa Fabaceae 7 14 Ageratina adenophora Asteraceae 7 14 Hydrangea macrophylla Hydrangeaceae 6 12 Brassica spp. Brassicaceae 6 12 Liatris spp. Asteraceae 5 10 Melilotus spp. Fabaceae 5 10 Ageratina spp. Asteraceae 5 10 Campanula spp. Campanulaceae 5 10 Arctium spp. Asteraceae 5 10 Daucus carota Apiaceae 4 8 Campanula rapunculoides Campanulaceae 4 8 Silphium perfoliatum Asteraceae 3 6 Allium thunbergii Liliaceae 3 6 Eutrochium spp. Asteraceae 3 6 Eutrochium purpureum Asteraceae 2 4 Carduus acanthoides Asteraceae 2 4 Rosa rugosa Rosaceae 2 4 Sparganium eurycarpum Sparganiaceae 2 4 Actaea racemosa Ranunculaceae 2 4 Mentha spp. Lamiaceae 2 4 Sorbaria spp. Rosaceae 2 4 Rosa spp. Rosaceae 2 4 Impatiens spp. Balsaminaceae 2 4 Silphium spp. Asteraceae 2 4 Rhodiola spp. Crassulaceae 2 4 Crassulaceae Crassulaceae 2 4 Echinacea angustifolia Asteraceae 1 2 Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 35 Pollen volume based on light microscopy showed the potential nutritional importance of floral sources by taking the volume of the pollen grain and the dry weight of the original pollen load into account. Arctium minus (Bernhardi), Medicago sativa (Linnaeus), Solanum dulcamara, Carduus acanthoides, and Trifolium pratense together comprised more than half of the total volume of pollen (Fig. 3). Discussion Our study represents the first reporting of B. affinis pollen collected from active nests. We recommend caution interpreting or drawing inference from our results as they represent collections on 4 dates spanning 1 month from 1 colony, and 1 date from a second colony. These factors could result in higher prevalence of some plants in our study than would be found in typical B. affinis colonies in this region and do not include data from early colony development, a time period that may be particularly crucial to bumble bee colony success (Malfi et Taxon Family Number of Samples Present Prevalence (%) Helenium autumnale Asteraceae 1 2 Cirsium arvense Asteraceae 1 2 Veronicastrum virginicum Scrophulariaceae 1 2 Borago officinalis Boraginaceae 1 2 Solanum lycopersicum Solanaceae 1 2 Astilbe chinensis Saxifragaceae 1 2 Trifolium pratense Fabaceae 1 2 Parthenocissus quinquefolia Vitaceae 1 2 Sonchus arvensis Asteraceae 1 2 Rudbeckia hirta Asteraceae 1 2 Laportea canadensis Urticaceae 1 2 Nepeta cataria Lamiaceae 1 2 Hypericum spp. Clusiaceae 1 2 Echinacea spp. Asteraceae 1 2 Helenium spp. Asteraceae 1 2 Urtica spp. Urticaceae 1 2 Veronicastrum spp. Scrophulariaceae 1 2 Plantago spp. Plantaginaceae 1 2 Acer spp. Aceraceae 1 2 Allium spp. Liliaceae 1 2 Asteraceae Asteraceae 1 2 Apiaceae Apiaceae 1 2 Table 3, continued. Plant taxa, identified among 49 Bombus affinis pollen samples, via metabarcoding. Prevalence is the percent of 49 samples where a plant taxon was detected. Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 36 Figure 2. Percent prevalence (percent per number of samples) of the 20 most common pollen taxa detected in the genetic (A) and microscopy (B) datasets of 49 Bombus affinis pollen samples. Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 37 al. 2022). Nonetheless, our findings provide an important first step towards understanding the pollen diet of B. affinis, particularly in urban areas. In this study we identified 47 total plant genera, which included 11 taxa with no previous B. affinis foraging records. Solanum dulcamara, or Bittersweet Nightshade, was the most prevalent species in our samples and was among the top species by volume. This finding is further supported by Wood et al. (2019), who found Solanum spp. represented a significant component of B. affinis pollen diet in Michigan. This perennial vine was introduced from Europe as a garden ornamental and became widespread in the United States in the 1800s. It is possible that Solanum could be a preferred pollen source based on its exceptionally high protein content (Pamminger et al. 2019, Ruedenauer et al. 2019). Additional garden and ornamental plants detected in our analysis include Allium thunbergii, Hydrangea macrophylla (Thunberg), and H. arborescens (Linnaeus). Given the nests were in urban areas (Boone et al. 2022), it is not surprising to see an opportunistic reliance on garden and ornamental flowers by B. affinis. These results are consistent with historic B. affinis pollen samples, where agricultural and garden plants were commonly detected in corbicular pollen loads (Simanonok et al. 2021). This research points to specific plant species that can be important components of B. affinis pollen diets and provides actionable research for conservation efforts in urban systems (Baldock et al. 2019, Burr et al. 2018). There is a growing interest in creating pollinator habitat and “bee-friendly” lawns in urban areas throughout the United States (Larson et al. 2014, Ramer et al. 2019). Our study highlights the potential importance of plants often Figure 3. Percent volume (estimated percent of plant species volume to total pollen volume) of the 20 most common taxa detected via light microscopy identification of 49 Bombus affinis pollen samples. Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 38 included in “bee-friendly” lawn mixes, which include Trifolium spp. as well as common, inexpensive ornamental plants such as Hosta spp. and Hydrangea spp. Some plants that we found to be targeted pollen sources are more likely to be the subject of eradication efforts than be included in conservation plantings or gardens. Carduus acanthoides, Daucus carota (Linnaeus), Arctium minus, and Solanum dulcamara are examples of noxious and invasive plants that may the subject of eradication efforts but should be recognized as potential food sources for B. affinis. Ideally, efforts to remove these potential food sources from the landscape would operate in conjunction with efforts to establish non-invasive and native forbs that are also important food sources of B. affinis, such as those identified by Wolf et al. (2022). Furthermore, our pollen forage research may be valuable to programs that promote the planting of wildflowers and gardens in urban areas of the Midwest. Research has shown that planted and volunteer flowers in private lawns provide valuable forage habitat for native bees and other insects (Larson et al. 2014, Mody et al. 2020, Wolfin et al. 2023). As a result, programs such as Minnesota’s “Lawns to Legumes” (https://bwsr.state.mn.us/ l2l) and similar efforts are being championed across the Midwest to promote bee habitat in urban systems. Our study provides further evidence that volunteer Trifolium spp., which are often abundant in urban lawns, serves as an important component of B. affinis pollen diet. We note that our study describes use of forbs by B. affinis; however, we are unable to infer forb preference or whether the collected pollen is nutritious. Understanding pollen nutrition and floral preference of imperiled bumble bees is an important a rea of future research. Genetic and microscopic evaluation of pollen samples revealed compositionally dissimilar results at both the genus and species levels. However, concordance did improve when identified taxa were broadened to a taxonomic level above species. Our detected genera matching at 57.4% between our methods is in rough agreement with other studies using similar methods for bee-collected pollen samples (Richardson et al. 2015, Smart et al. 2017). Taxa identified within a single sample, via genetics or microscopy, were not hierarchically nested, which makes it difficult to align taxa between methods. For example, if we detected Carduus nutans in a genetics sample, and Carduus spp. in the sample microscopy sample, we would have reported a lack of concordance between these 2 taxa, even though the taxa are closely aligned. Binning species to genus, family, or other functional categories may be essential for fair treatment of results in studies which consider pollen identification from multiple methodologies (Simanonok et al. 2023). For example, we documented a high degree of concordance between both methods for Trifolium at the genus level, but lack of concordance at the species level. There are over 150 Trifolium species in USDA PLANTS (US Department of Agriculture 2023), and species such as T. repens and T. pratense are highly variable morphologically. It is unlikely that databases are balanced enough for the Trifolium genus for researchers to have strong confidence with assignments at the species level for either method. Thus, the 2 methods we used appear to be concordant about the Trifolium genus to the extent possible. This result contrasts with recent findings using B. affinis-collected pollen that showed significant disagreement between microscopic analysis on pollen and ITS metabarcoding, even when species were scaled up to higher taxonomic levels (Simanonok et al. 2023). The primary difference between these studies was sample age. Simanonok et al. (2023) used pollen dating from 1913–2013 while all pollen in this study were collected in 2020 and analyzed in 2021–2022. Here, we observed improved concordance with modern samples, and thus historical analyses of pollen may have introduced greater error. Future studies using historical pollen for analysis may consider using much stronger filtering thresholds in their methodologies to further reduce such error. Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 39 Taken together, our results reinforce other recent research regarding B. affinis and highlight forage plants in urban systems. Specifically, we found that B. affinis has a broad pollen foraging niche which includes many garden and ornamental cultivars and non-native species. Future work studying the foraging ecology of B. affinis would benefit from documenting pollen foraging in non-urban landscapes as well as throughout the foraging season and quantification of forb preferences as opposed to forb use (Pizante et al. 2023, Simanonok et al. 2021). The known range of B. affinis is largely confined to urban and suburban areas in the Upper Midwest (Boone et al. 2023), but it is unclear if this is due to particular habitat associations, an artifact of spatial sampling bias driven in recent years by community science in urban areas, or other driving factors (e.g., refuge from a deleterious pathogen). Certainly, the historical range of B. affinis included rural and agricultural landscapes. Understanding forage and habitat needs of B. affinis in these areas will be critical for achieving species recovery goals (USFWS 2021). In addition, understanding forage preference will help managers understand whether the myriad non-native forb species from which B. affinis collects pollen are important components of B. affinis diet, or simply collected because they are locally available in forb-limited landscapes. Acknowledgments We thank Nancy Kafka and Daniel Furuta for allowing access to the nests on their properties, Michelle Boone, Nicole Gerjets, and Jessica Petersen for assisting with colony observations, and Ian Roberts for helping with pollen counts. We thank Mark Hepner and 2 anonymous reviewers for providing thoughtful feedback and Jaxton Wiest for assisting with data cleaning. This work was supported through an Inter-Agency Agreement between the U.S. Fish and Wildlife Service and the U.S. Geological Survey. Partial funding was provided by the U.S. Fish and Wildlife Service Minnesota-Wisconsin Ecological Services Field Office contract number F0420P0264. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government. Genetics data generated during this study are publicly available as a U.S. Geological Survey data release (Cornman et al. 2024). Any opinion, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the U.S. Fish and Wildlife Service. Literature Cited Baldock, K.C.R., M.A. Goddard, D.M. Hicks, W.E. Kunin, N. Mitschunas, H. Morse, L.M. Osgathorpe, S.G. Potts, K.M. Robertson, and A.V. Scott. 2019. A systems approach reveals urban pollinator hotspots and conservation opportunities. Nature Ecology & Evolu tion 3:363–373. Boone, M.L., E. Evans, A. Wolf, H. Minser, J. Watson, and T.A. Smith. 2022. Notes from rusty patched bumble bee (Bombus affinis Cresson) nest observations. Insect Conservation and Diversity 15:380–384. Boone, M.L., Z.M. Portman, I. Lane, and S. Rao. 2023. Occupancy of Bombus affinis (Hymenoptera: Apidae) in Minnesota is highest in developed areas when standardized surveys are employed. Environmental Entomology 52:918–938. Burr, A., D.M. Hall, and N. Schaeg. 2018. The perfect lawn: Exploring neighborhood socio-cultural drivers for insect pollinator habitat. Urban Ecosystems 21:1123–1137. Colla, S.R., and L. Packer. 2008. Evidence for decline in eastern North American bumblebees (Hymenoptera: Apidae), with special focus on Bombus affinis Cresson. Biodiversity and Conservation 17:1379–1391. Cornman, R.S., C.R.V. Otto, D.D. Iwanowicz, and J.S. Pettis. 2015. Taxonomic characterization of honey bee (Apis mellifera) pollen foraging based on non-overlapping paired-end sequencing of nuclear ribosomal loci. PloS one 10:e0145365. Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 40 Cornman, R.S., D.D. Iwanowicz, and C.R.V. Otto. 2024. Amplicon sequencing of pollen foraged by Bombus affinis for compositional analysis. U.S. Geological Survey Data Release. Available online at https://doi.org/10.5066/P14UMYJM. Crompton, C.W., and W.A. Wojtas. 1993. Pollen grains of Canadian honey plants. Agriculture Canada:1892/E. Avaliable online at https://publications.gc.ca/site/eng/481262/publication.html. Di Pasquale, G., M. Salignon, Y. Le Conte, L.P. Belzunces, A. Decourtye, A. Kretzschmar, S. Suchail, J.-L. Brunet, and C. Alaux. 2013. Influence of pollen nutrition on honey bee health: Do pollen quality and diversity matter? PLoS ONE 8:e72016. Doyle, J. 1991. DNA protocols for plants. Pp. 283–293, In Molecular Techniques in Taxonomy. Springer, Berlin, Heidelberg, Germany. 410 pp. Evans, E., R.W. Thorp, S. Jespen, and S.H. Black. 2008. Status review of three formerly common species of bumble bee in the subgenus Bombus. Prepared for the Xerces Society of Invertebrate Conservation. Full technical report. 55 pp. Goulson, D., E. Nicholls, C. Botías, and E.L. Rotheray. 2015. Bee declines driven by combined stress from parasites, pesticides, and lack of flowers. Science 347:125 5957. Graves, T. A., W.M. Janousek, S.M. Gaulke, A.C. Nicholas, D.A. Keinath, C.M. Bell, S. Cannings, R.G. Hatfield, J.M. Heron, and J.B. Koch. 2020. Western bumble bee: Declines in the continental United States and range-wide information gaps. Ecosphere 11:e03141. Hepner, M.J., E. Orcutt, K. Price, K. Goodell, T. Roulston, R.P. Jean, and R.T. Richardson. 2024. Montane Central Appalachian forests provide refuge for the critically endangered rusty patched bumble bee (Bombus affinis). Forest Ecology and Management 556:121751. Huson, D.H., A.F. Auch, J. Qi, and S.C. Schuster. 2007. MEGAN analysis of metagenomic data. Genome research 17:377–386. Illumina. 2023. 16S Metagenomic Sequencing Library Preparation: Preparing 16S Ribosomal RNA Gene Amplicons for the Illumina MiSeq System. Available online at . Accessed 1 July 2023. Jones, G.D. 2012. Pollen analyses for pollination research, unacetolyzed pollen. Journal of Pollination Ecology 9:96–107. Larson, J.L., A.J. Kesheimer, and D.A. Potter. 2014. Pollinator assemblages on dandelions and white clover in urban and suburban lawns. Journal of Insect Conservat ion 18:863–873. Lau, P., P. Lesne, R.J. Grebenok, J. Rangel, and S.T. Behmer. 2022. Assessing pollen nutrient content: A unifying approach for the study of bee nutritional ecology. Philosophical Transactions of the Royal Society B 377:20210510. Lye, G.C., J.L. Osborne, K.J. Park, and D. Goulson. 2012. Using citizen science to monitor Bombus populations in the UK: Nesting ecology and relative abundance in the urban environment. Journal of Insect Conservation 16:697–707. Malfi, R.L., E. Crone, M. Rundlöf, and N.M. Williams. 2022. Early resources lead to persistent benefits for bumble bee colony dynamics. Ecology 103:e03560. Martin, A.C., and W.J. Harvey. 2017. The Global Pollen Project: A new tool for pollen identification and the dissemination of physical reference collections. Methods in Ecology and Evolution 8:892–897. Mody, K., D. Lerch, A.-K. Müller, N.K. Simons, N. Blüthgen, and M. Harnisch. 2020. Flower power in the city: Replacing roadside shrubs by wildflower meadows increases insect numbers and reduces maintenance costs. PloS one 15:e0234327. Oksanen, J., F.G. Blanchet, M. Friendly, R. Kindt, P. Legendre, D. McGlinn, P.R. Minchin, R. B. O’hara, G.L. Simpson, and P. Solymos. 2019. Package ‘vegan.’ Community ecology package, version 2. Otto, C.R.V., A.C. Schrage, L.L. Bailey, J.M. Mola, T.A. Smith, I. Pearse, S. Simanonok, and R. Grundel. 2023. Addressing detection uncertainty in Bombus affinis (Hymenoptera: Apidae) surveys can improve inferences made from monitoring. Environmental Entomology 52:108–118. PalDat -- A Palynological Database. 2023. Available online at http://www.paldat.org. Accessed 1 April 2021. Prairie Naturalist M.P. Simanonok, E. Evans, C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, and T.A. Smith 2024 No. 56 41 Pamminger, T., R. Becker, S. Himmelreich, C.W. Schneider, and M. Bergtold. 2019. Pollen report: Quantitative review of pollen crude protein concentrations offered by bee pollinated flowers in agricultural and non-agricultural landscapes. PeerJ 7:e7394. Pizante, R., J.H. Acorn, S.H. Worthy, and C.M. Frost. 2023. Existing flower preference metrics disagree on best plants for pollinators: Which metric to choose? Insect Conservation and Diversity 16:745–757. Ramer, H., K.C. Nelson, M. Spivak, E. Watkins, J. Wolfin, and M. Pulscher. 2019. Exploring park visitor perceptions of ‘flowering bee lawns’ in neighborhood parks in Minneapolis, MN, US. Landscape and Urban Planning 189:117–128. Richardson, R.T., C. Lin, D.B. Sponsler, J.O. Quijia, K. Goodell, and R.M. Johnson. 2015. Application of ITS2 metabarcoding to determine the provenance of pollen collected by honey bees in an agroecosystem. Applications in Plant Sciences 3:1400066. Rognes, T., T. Flouri, B. Nichols, C. Quince, and F. Mahe. 2016. VSEARCH: A versatile open source tool for metagenomics. Peer J 4: e2584. Ruedenauer, F.A., S.D. Leonhardt, K. Lunau, and J. Spaethe. 2019. Bumblebees are able to perceive amino acids via chemotactile antennal stimulation. Journal of Comparative Physiology A 205:321–331. Sickel, W., M.J. Ankenbrand, G. Grimmer, A. Holzschuh, S. Härtel, J. Lanzen, I. Steffan-Dewenter, and A. Keller. 2015. Increased efficiency in identifying mixed pollen samples by meta-barcoding with a dual-indexing approach. BMC Ecology 15:20. Simanonok, M.P., D.D. Iwanowicz, C.D. Raines, T.J. Wood, R. Isaacs, R.S. Cornman, and C.R.V. Otto. 2023. Comparison of microscopy and metabarcoding to identify pollen used by the critically endangered rusty patched bumble bee, Bombus affinis. Insect Conservation and Diversity 16:205–216. Simanonok, M.P., C.R.V. Otto, R.S. Cornman, D.D. Iwanowicz, J.P. Strange, and T.A. Smith. 2021. A century of pollen foraging by the endangered rusty patched bumble bee (Bombus affinis): Inferences from molecular sequencing of museum specimens. Biodiversity and Conservation 30:123–137. Simanonok, S.C., C.R.V. Otto, and D.A. Buhl. 2021. Floral resource selection by wild bees and honey bees in the Midwest USA: Implications for designing pollinator habitat. Restoration Ecology e13456. Smart, M.D., R.S. Cornman, D.D. Iwanowicz, M. McDermott-Kubeczko, J.S. Pettis, M.S. Spivak, and C.R.V. Otto. 2017. A comparison of honey bee-collected pollen from working agricultural lands using light microscopy and ITS metabarcoding. Environmental Ent omology 46:38–49. US Department of Agriculture, Natural Resource Conservation Service. 2023. The PLANTS Database, Version 3.1. Available online at http://plants.usda.gov. Accessed 1 April 2021. US Fish and Wildlife Service. 2021. Recovery Plan for Rusty Patched Bumble Bee (Bombus affinis). Midwest Regional Office, Bloomington, MN. Accessed online at https://ecos.fws.gov/docs/recovery_ plan/Final%20Recovery%20Plan%20_Rusty%20Patched%20Bumble%20Bee_2021.pdf. Wolf, A.T., J.C. Watson, T.J. Hyde, S.G. Carpenter, and R.P. Jean. 2022. Floral resources used by the endangered rusty patched bumble bee (Bombus affinis) in the Midwestern United States. Natural Areas Journal 42:301–312. Wolfin, J., E. Watkins, I. Lane, Z.M. Portman, and M. Spivak. 2023. Floral enhancement of turfgrass lawns benefits wild bees and honey bees ( Apis mellifera). Urban Ecosystems 26:361–375. Wood, T.J., J. Gibbs, K.K. Graham, and R. Isaacs. 2019. Narrow pollen diets are associated with declining Midwestern bumble bee species. Ecology 100:e02697. Zymanski, J., T. Smith, A. Horton, M. Parkin, L. Ragan, G. Masson, E. Olson, K. Gifford, and L. Hill. 2016. Rusty Patched Bumble Bee (Bombus affinis) Species Status Assessment: Final Report, Version 1. US Fish and Wildlife Service, Midwest Regional Office, Bloomington, MN. Accessed online at https://ecos.fws.gov/ServCat/DownloadFile/120109.