Diversity of Flower-Visiting Bees and Their Pollen Loads on a Wildflower Seed Farm in Montana April M

Entomology Publications Entomology

4-2012 Diversity of Flower-visiting Bees and their Pollen Loads on a Wildflower Seed Farm in Montana April M. Pearce Montana State University - Bozeman

K. M. O'Neill Montana State University - Bozeman

Richard S. Miller Montana State University - Bozeman

Sue L. Blodgett Iowa State University, [email protected]

Follow this and additional works at: http://lib.dr.iastate.edu/ent_pubs Part of the Biodiversity Commons, Entomology Commons, and the Systems Biology Commons The ompc lete bibliographic information for this item can be found at http://lib.dr.iastate.edu/ ent_pubs/271. For information on how to cite this item, please visit http://lib.dr.iastate.edu/ howtocite.html.

This Article is brought to you for free and open access by the Entomology at Iowa State University Digital Repository. It has been accepted for inclusion in Entomology Publications by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. Diversity of Flower-visiting Bees and their Pollen Loads on a Wildflower Seed Farm in Montana

Abstract During a two-year survey on a wildflower seed farm in southcentral Montana, we collected ∼50 species of bees from 18 genera in sweep samples on cultivated wildflowers and weeds. The two cultivated plant species most intensively sampled attracted different assemblages of bee visitors. Slender white prairie clover (Dalea candida) attracted 27 species, 94% of visitors being Apis mellifera (73%), Lasioglossum spp., Colletes phaceliae, and Bombus spp. Prairie coneflower (Ratibida columnifera) attracted 20 species, the majority being Halictus rubicundus and three Melissodes species; only 3% of visitors to this plant were A. mellifera, despite the fact that the coneflower field was closer to an apiary than were the prairie clover fields. Other apparently non-random plant-bee associations included A. mellifera onOnobrychis viciaefolia, Bombus spp. on Astragalus cicer, and Halictus ligatus and aMelissodes sp. on Symphyotrichum chilensis. Analysis of pollen loads suggests high flower constancy for A. mellifera, Bombus spp., and many of the native solitary bee species foraging on cultivated plants. The low numbers of honey bees on certain plants suggest that native, non-managed bees of such genera as Bombus, Melissodes, Halictus, and Lasioglossum may be critical for plant species for which honey bees show relatively low preference (especially when highly-preferred species such as D. candida are abundant).

Keywords Pollination, pollen loads, bee diversity, Apoidea, wildflower seed production

Disciplines Biodiversity | Entomology | Systems Biology

Comments This article is from Journal of the Kansas Entomological Society 85 (2012): 97, doi:10.2317/JKES111202.1. Posted with permission.

This article is available at Iowa State University Digital Repository: http://lib.dr.iastate.edu/ent_pubs/271 Diversity of Flower-visiting Bees and their Pollen Loads on a Wildflower Seed Farm in Montana Author(s): April M. Pearce, K. M. O'Neill, Richard S. Miller, and Sue Blodgett Source: Journal of the Kansas Entomological Society, 85(2):97-108. 2012. Published By: Kansas Entomological Society DOI: http://dx.doi.org/10.2317/JKES111202.1 URL: http://www.bioone.org/doi/full/10.2317/JKES111202.1

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/ terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research. JOURNAL OF THE KANSAS ENTOMOLOGICAL SOCIETY 85(2), 2012, pp. 97–108 Diversity of Flower-visiting Bees and their Pollen Loads on a Wildflower Seed Farm in Montana

1,2 1,3 1 4 APRIL M. PEARCE, K. M. O’NEILL, RICHARD S. MILLER, AND SUE BLODGETT

ABSTRACT: During a two-year survey on a wildflower seed farm in southcentral Montana, we collected ,50 species of bees from 18 genera in sweep samples on cultivated wildflowers and weeds. The two cultivated plant species most intensively sampled attracted different assemblages of bee visitors. Slender white prairie clover (Dalea candida) attracted 27 species, 94% of visitors being Apis mellifera (73%), Lasioglossum spp., Colletes phaceliae, and Bombus spp. Prairie coneflower (Ratibida columnifera) attracted 20 species, the majority being Halictus rubicundus and three Melissodes species; only 3% of visitors to this plant were A. mellifera, despite the fact that the coneflower field was closer to an apiary than were the prairie clover fields. Other apparently non-random plant-bee associations included A. mellifera on Onobrychis viciaefolia, Bombus spp. on Astragalus cicer, and Halictus ligatus and a Melissodes sp. on Symphyotrichum chilensis. Analysis of pollen loads suggests high flower constancy for A. mellifera, Bombus spp., and many of the native solitary bee species foraging on cultivated plants. The low numbers of honey bees on certain plants suggest that native, non-managed bees of such genera as Bombus, Melissodes, Halictus, and Lasioglossum may be critical for plant species for which honey bees show relatively low preference (especially when highly- preferred species such as D. candida are abundant). KEY WORDS: Pollination, pollen loads, bee diversity, Apoidea, wildflower seed production

Land rehabilitation projects in the western U.S. commonly involve reseeding disturbed lands with seed mixes that include wildflowers. Mass production of seed of most wildflower species used in rehabilitation programs requires services of insect pollinators (Cane, 2008). Such pollinators could include those from both managed and wild populations. Concerns about honey bees (Apis mellifera L.) include not only recent problems with the health of bees in commercial populations (vanEngelsdorp and Meixner, 2010), but their availability and effectiveness on all cultivated wildflower species (Cane, 2008). There is also the potential for managing bumble bees (Bombus spp.) and Megachile rotundata L., but these also come with attendant difficulties. Thus, communities of wild, native bees could play an important role in production of wildflower seed if their diversity and population sizes are sufficient on seed farms. Wildflower seed farms provide abundant nectar and pollen, but nesting sites could still be at a premium, especially in large, frequently-tilled fields with a high ratio of the area of fields to undisturbed edges that provide nesting substrate (Cane, 2008).

1 Department of Land Resources and Environmental Sciences, Montana State University, Bozeman, MT 59717. 2 1507 W. Comanche Ave. Tampa, FL 33603. 3 Corresponding author. [email protected] 4 Departments of Entomology and Natural Resource Ecology and Management, Iowa State University, Ames, IA 50011. Accepted 3 July 2012; Revised 17 July 2012 E 2012 Kansas Entomological Society 98 JOURNAL OF THE KANSAS ENTOMOLOGICAL SOCIETY

Multiple factors contribute to the value of a bee species as a pollinator of commercial crops, including its abundance, phenology relative to that of the plant species, behavior on the flower, flower constancy, and distances travelled during foraging bouts (Proctor et al., 1996; Mader et al., 2011). But one of the first steps in assessing the potential significance of wild bees to wildflower seed production is to document the existence of a diverse assemblage of bees. We undertook a survey at a wildflower seed production facility, the Bridger Plant Materials Center (BPMC), managed by the U.S. Department of Agriculture, Natural Resources and Conservation Service (USDA-NRCS) in Bridger, Montana. During two-year study, our objectives were to characterize 1) the plant-associations within an assemblage of bees visiting cultivated wildflowers at BPMC and 2) the contents of the pollen loads of bees. We were particularly interested in determining which species were visited by wild native bee species rather than, or in addition to, honey bees and alfalfa leafcutting bees (M. rotundata), both of which have been introduced at BPMC as managed populations. The two major plant species sampled were slender white prairie clover (Dalea candida Willd.; Fabaceae) and upright prairie coneflower (Ratibida columnifera [Nutt.] Woot. & Standl.; Asteraceae), but for comparison we also examined bee assemblages on five other cultivated species and two weed species.

Materials and Methods The study was conducted in July and August of 2006 and 2007 at the BPMC, 3.5 km southeast of Bridger, Carbon County, Montana (45u169N, 108u539W) at 3685–3775 m elevation. BPMC encompasses 56 ha devoted to seed production to provide plants for land rehabilitation. A honey bee (Apis mellifera L.) apiary containing 12 hives was situated in the northeast corner of BPMC. We sampled insects on seven species of flowering angiosperms cultivated for seed production. The two plant species sampled most intensively were Dalea candida and Ratibida columnifera. The two plantings of D. candida (Antelope germplasm; Majerus and Holzworth, 2003) at BPMC had areas of 0.6 ha (‘‘upper field’’, present both years) and 0.4 ha (‘‘lower field’’, present in 2006 only), whereas the single R. columnifera (Stillwater germplasm; Winslow et al., 2005) field was 0.4 ha, and situated adjacent to the upper D. candida plot (both years). We also sampled insects on Pacific aster (Symphyotrichum chilensis [Nees] G.L. Nesom; Asteraceae), cicer milkvetch, (Astragalus cicer L.; Fabaceae), sainfoin (Onobrychis viciaefolia Scop.; Fabaceae), common snowberry (Symphoricarpos albus [L.] Blake; Caprifoliaceae), and yarrow (Achillea millefolium L.; Asteraceae). Of the cultivated plants, all but cicer milkvetch and sainfoin are native to North America. The R. columnifera field was closest to the apiary (250 m), whereas the A. cicer plot was furthest (1000 m). For comparison to the cultivated species, we also sampled insects on two flowering weed species within the confines of BPMC: bindweed (Convolvulus arvensis L.; Convolvulaceae) and sow thistle (Sonchus sp.; Asteraceae). Other weed species that provided potential pollen sources included birdsfoot trefoil (Lotus corniculatus L.; Fabaceae), Canada thistle (Cirsium arvense L.; Asteraceae), and sweetclover (Melilotus officinalis [L.] Lam.; Fabaceae). Among the weed species, only Sonchus is native to North America. To characterize the flower-visiting bee assemblage BPMC, we conducted 167 sweep samples, each of which consisted of 50 sweeps with a 40-cm diameter sweep VOLUME 85, ISSUE 2 99 net. Plants were sampled on eight days in 2006 and 10 days in 2007. The cultivated plants were sampled along linear transects, but the weedy species were sampled along irregular transects that often had gaps. The number of sweep samples taken varied among plant species, with D. candida (40% of the samples), R. columnifera (25%), and S. chilensis (11%) being most intensively sampled. All sweep samples were frozen for later processing. We determined the size and composition of pollen loads of 346 female bees of 17 taxa and, for comparison, both sexes of several species of apoid wasps (Tachytes sayi Banks (Crabronidae) and Sphex ichneumoneus (L.) (Sphecidae). During both years, we collected bees and wasps of several species individually on flowers and placed them into 1.5 ml Eppendorf tubes that were frozen later the same day. To estimate the species composition and number of pollen grains in each pollen load we used methods described in O’Neill and O’Neill (2010). All means are presented 6 standard errors. Because different plant species were sampled different numbers of times, we compared the relative frequencies of selected bee taxa between different samples using chi-square contingency table analyses, with Yates correction for continuity (Everitt, 1977). Plant species diversity in pollen loads was characterized using Hill’s #2 diversity index, the effective number of very abundant species in a sample (Ludwig and Reynolds, 1988).

Results Overall Bee Diversity Of the 3048 bees in .50 species in 20 genera collected on the nine focal plant species (Table 1), 54.9% were honey bees (A. mellifera), 5.1% bumble bees (Bombus spp.), and 40.0% other genera. The most common other genera were Lasioglossum (12.8% of all bees), Halictus (12.1%), Melissodes (4.6%), Colletes (5.3%), and Agapostemon (2.1%). Apis mellifera was the only species and Halictus the only non- Apis genus observed on all nine plant species sampled. Most bees were either Apidae or Halictidae and, although at least ten species of Andrenidae and Megachilidae were collected on the cultivated plants, each family made up ,1% of all bees collected. The solitary bees included eight species within four genera of brood parasitic bees (Nomada, Triepeolus, Sphecodes, and Coelioxys), but these were relatively rare. The number of bee taxa collected on different plant species was positively correlated with the number of samples taken (Spearman correlation, r 5 0.72, P 5 0.02).

Bee-Plant Associations: Sweep Samples On D. candida, 72.8% of bees were honey bees (Table 1), most of which likely originated from the apiary 280 m from the upper D. candida field and 480 m from the lower field. Other relatively common bees on D. candida included Lasioglossum spp. (34.6% of the non-Apis bees, Colletes phaceliae (25.2%), Bombus huntii (8.6%), B. griseocollis (7.7%), and Halictus ligatus (7.2%). Halictus ligatus was also relatively common on A. millefolium and Sonchus sp., but C. phaceliae was rarely found elsewhere at BPMC. The only other plant species on which A. mellifera outnumbered other bees in sweep samples was O. viciaefolia, where it occurred in a similar ratio (2.51:1), relative to non-Apis bees, as they did on D. candida (2.67:1; x2 5 0.06, d.f. 5 1, P 5 0.81). Honey bees were also common, though in the minority of bees collected 0 ORA FTEKNA NOOOIA SOCIETY ENTOMOLOGICAL KANSAS THE OF JOURNAL 100 Table 1. Bee species collected in sweep samples and pan traps at BPMC. NC 5 not counted. For taxa identified to genus only, other than Melissodes, the number of unidentified species recognized in our samples is given in parentheses in column 1.

Plant species sampled (number of 50-sweep samples)1

Cultivated plants Weeds

Bee families and species AC (4) AM (10) DC (67) OV (6) RC (42) SC (19) SA (5) CA (6) SS (8)

Andrenidae Andrena p. prunorum Cockerell 1 1 9 - 2 - - - - Andrena spp. (3) - 2 1 - - - 1 - - Calliopsis andreniformis Smith - - - - 4 - - - - Calliopsis coloradensis Cresson - - 6 - 1 - - - - Perdita sp. - - - - 10 1 - - - Apidae Apis mellifera L. 3 1 1484 113 12 4 25 6 25 Bombus centralis Cresson 4 - 1 - - - 4 - - Bombus fervidus F. 1 ------1 Bombus griseocollis (DeGeer) 7 - 43 3 - - - - - Bombus huntii Greene 10 - 48 10 18 - - 1 5 Bombus rufocinctus Cresson - - 1 ------Bombus sp. --1------Melissodes sp. 1 - - 3 - 7 22 - - 1 Melissodes sp. 2 - - 1 - 74 1 - - - Melissodes sp. 3 - - 2 - 12 12 - - 4 Nomada spp. (3) - 2 - 1 4 - - - - Triepeolus spp. (2) - - 1 - 2 - - - - Colletidae Colletes fulgida Swenk - 2 1 - 6 - - - - Colletes petalostemonis Swenk - - - - 6 - - - - Colletes phaceliae Cockerell - 3 140 - 1 - - 1 - Colletes sp. --1------Hylaeus bisinuatus Foster - 2 - - 4 - 1 - - Hylaeus spp. (2) - 4 1 - - - 2 - - OUE8,ISE2101 2 ISSUE 85, VOLUME

Table 1. Continued.

Plant species sampled (number of 50-sweep samples)1

Cultivated plants Weeds

Bee families and species AC (4) AM (10) DC (67) OV (6) RC (42) SC (19) SA (5) CA (6) SS (8)

Halictidae Agapostemon angelicus/texanus2 -415232 -13 Agapostemon femoratus Crawford - 6 12 - 12 1 1 1 - Halictus confusus Smith - 4 6 1 2 - 1 - - Halictus ligatus Say - 33 40 - 1 53 - 3 20 Halictus rubicundus (Christ) 1 4 18 - 176 - 3 3 - Lasioglossum spp. 3 73 192 27 - 10 37 20 27 Sphecodes spp. (2) - - 1 - 1 - - - - Megachilidae Coelioxys mesae Cockerell - - 1 - 2 - - - - Heriades carinatus Cresson - 5 3 - - 1 2 - - Hoplitis spp. (2) - - 1 ------Megachile brevis Say --1------Megachile inimica Cresson - - - - - 1 - - - Megachile lippiae Cockerell - - 3 - 1 - - - - Megachile parallela Smith----1---- Megachile rotundata (F.) - - 2 1 4 - - - - Osmia spp. (2) 2 ------Total number of bees collected 32 146 2039 158 366 108 77 36 86 Total number/50-sweep sample 8.0 14.6 30.4 26.3 8.7 5.7 15.4 6.0 10.8 Number non-Apis/50 sweep sample 7.3 14.5 8.3 7.5 8.4 5.5 10.4 5.0 7.6 Ratio of Apis to non-Apis bees 0.10 0.01 2.67 2.51 0.03 0.04 0.48 0.20 0.41

1 AC 5 A. cicer;AM5 A. millefolium; Co. arvensis;DC5 D. candida;OV5 O. viciaefolia;RC5 R. columnifera;SA5 S. albus,SC5 S. chilensis;SS5 Sonchus sp.; 2 females indistinguishable, though we did definitively identify males of both species at the site. 102 JOURNAL OF THE KANSAS ENTOMOLOGICAL SOCIETY on S. albus (32.5%) and Sonchus sp. (29.1%). Overall, the ratio of Apis to non-Apis bees on D. candida + O. viciaefolia (2.66:1) was higher than on all five other cultivated species combined (0.07:1; x2 5 980.8, d.f. 5 1, P , 0.001). The predominant bees on R. columnifera were Halictus rubicundus (48.1% of all bees) and Melissodes sp. 2 (20.2%). Honey bees were relatively rare (3.3%), so that the ratio of Apis to non-Apis bees was much higher on D. candida than on R. columnifera (0.03:1; x2 5 634.6, d.f. 5 1, P , 0.0001). Although the total number of bees collected per sweep sample was much higher on D. candida than on R. columnifera, the number of non-Apis bees per sample was similar (Table 1). On S. chilensis, Halictus ligatus (49.1%) was predominant, along with Melissodes sp. 1 (20.4%). Thus, bees of the genera Halictus and Melissodes exhibited higher relative frequencies on R. columnifera and S. chilensis than on other cultivated plants. The ratio of Halictus to non-Halictus bees on these two plants combined (0.94:1) was greater than for all other cultivated plants combined (0.05:1; x2 5 749.8, d.f. 5 1, P , 0.001). A similar difference was seen in the comparing Melissodes to non-Melissodes bees (0.37:1 vs. 0.002:1; x2 5 651.4, d.f. 5 1, P , 0.0001). Bombus was the most common genus only on A. cicer (68.8%), whereas Lasioglossum made up the greatest percentage of bees of all genera on A. millefolium (50.0%)and Symphoricarpos (48.1%), as well as on the two weeds, Co. arvensis (55.6%)and Sonchus (31.4%).

Bee-Plant Associations: Pollen Load Analyses Mean pollen load sizes for the most common insects collected on the cultivated plants ranged from ,600 for the wasp S. ichneumoneus on D. candida to .450,000 for Melissodes sp. 2 collected on R. columnifera (Table 2). Pollen load size varied among the six most common species on D. candida listed in Table 2 (Kruskal-Wallis test, P , 0.001). Load sizes of B. griseocollis were greater than for all of other species (except B. huntii), and both B. huntii and A. mellifera carried more pollen grains than S. ichneumoneus (Dunn’s test, a 5 0.05). The 25 largest pollen loads observed (i.e., those estimated at .500,000 pollen grains were all from five bee species: 1) A. mellifera foraging on D. candida (N 5 1), 2) B. griseocollis on D. candida (N 5 11), 3) B. huntii on A. cicer (N 5 2), D. candida (N 5 4), and R. columnifera (N 5 1), 4) M. parallela on R. columnifera (N 5 3), and 5) Melissodes sp. 2 on R. columnifera (N 5 3). The largest pollen load was estimated at .1.74 million grains (B. griseocollis on D. candida, though most of the pollen on this individual was from L. corniculatus). Among the common bees, the smallest mean load sizes were carried by H. confusus and Lasioglossum spp. The size of pollen loads was related to the body size. Among 130 individuals in genera with small body size (i.e., Agapostemon, Colletes, Dianthidium, Halictus, and Lasioglossum), just one load exceeded 100,000 grains (N 5 126). But such large pollen loads were common in species with larger, more robust bodies: 21.5% of A. mellifera, 47.8% of B. huntii, 50.0% of Melissodes sp. 1, 60.0% of B. griseocollis, and 80.0% of M. parallela (total for large species 5 42.8%, N 5 187; chi-square contingency table analysis comparing small vs. large bees, x2 5 67.0, d.f. 5 1, P , 0.001). However, body size is obviously not the only factor influencing the size of pollen loads. Sphex ichneumoneus was one of the largest species collected and is quite hairy, at least compared to many other apoid wasps, but none of these wasps carried more than 2550 pollen grains. VOLUME 85, ISSUE 2 103

Among honey bees collected on D. candida, a mean of 98% of pollen grains came from D. candida itself, whereas only 0.8% derived from R. columnifera, the second most common type (Table 2). Nearly 47% of these honey bees carried only D. candida pollen, and none carried ,83% D. candida. Although six pollen types were detected on the honey bees from D. candida, no single female carried more than three types. High mean percentages of D. candida pollen were also found for Bombus, Colletes, and Lasioglossum collected on this plant, and these bees also carried a high number of pure pollen loads (e.g., 13 of 14 C. phaceliae and ,50% of the females of both B. griseocollis and B. huntii). The low diversity of pollen types was also observed for B. huntii on A. cicer and for H. ligatus and Melissodes sp. 1 on R. columnifera and S. chilensis. Among the H. ligatus collected on R. columnifera, none carried S. chilensis pollen; conversely, only 2 of 15 taken from S. chilensis carried R. columnifera pollen. The plantings of these two species were separated by 300 m. Weeds at BPMC tended to grow dispersed and interspersed among cultivated plantings, or in small patches, so we hypothesized that bees collected on weeds would carry a higher pollen diversity of pollen types. This turned out to be the case for A. mellifera collected on D. candida (mean Hill’s #2 diversity index 5 1.04 6 0.01, N 5 46) compared to those from Sonchus (mean 5 1.19 6 0.04, N 5 19; Mann-Whitney test, P , 0.001). However, the diversity of pollen types in all bees collected on D. candida (mean 5 1.09 6 0.02, N 5 165) did not differ significantly from that for all bees collected on Co. arvensis (1.28 6 0.061, N 5 57) (Mann-Whitney test, P 5 0.24).

Discussion The 56 ha wildflower seed farm at BPMC harbors a diverse bee fauna of at least 60 species. In two years, we found ,50 species of bees in general sweep net samples on the focal plant species, but we also collected others in yellow pan traps and trap nests, and individually on flowers (Pearce 2008, O’Neill et al., 2010). These other species included Bombus occidentalis (Greene), Bombus mixtus Cresson, Diadasia sp. (Apidae), Hylaeus episcopalis (Cockerell), Hylaeus stevensi Crawford (Colletidae), Halictus virgatellus Cockerell (Halictidae), Anthidium sp., Ashmeadiella bucconis (Cresson), Ashmeadiella cactorum (Cockerell), Ashmeadiella gillettei Titus, and Dianthidium sayi Cockerell (Megachilidae). Among the bees identified to species, only A. mellifera and M. rotundata (the alfalfa leafcutting bee) are not native to North America (Droege, 2008). Megachile rotundata was released at BPMC during 1998–2001 when alfalfa (Medicago sativa L.) seed trials were being conducted (S. Winslow, personal communication), but a feral population was well-established in 2006 (O’Neill et al., 2010). The importance of the bee assemblage for seed production at BPMC and similar facilities is corroborated by studies examining the pollination requirements of commercially-grown wildflowers. Among nine wildflower species used in restoration programs and examined by Cane (2006, 2008), all benefited from being given access to insect pollinators, and three of the nine experienced reproductive gains from ,40- fold (Dalea purpurea Vent.) to over 100-fold (Hedysarum boreale Nutt.). Robson (2010) showed that seed production in natural habitats by a rare species of aster, Symphyotrichum sericeum (Vent.) G.L. Nesom, was limited at times when visits to the plant by bumble bees was lower, apparently due to their being drawn away to a more abundant species of sunflower (Solidago nemoralis Ait.). Both A. cicer 0 ORA FTEKNA NOOOIA SOCIETY ENTOMOLOGICAL KANSAS THE OF JOURNAL 104

Table 2. Size and composition of pollens loads of bees and wasps; for ‘‘other pollen sources’’, only species that made up more than 1% of the mean are listed.

Pollen from plant where insect was collected Other pollen sources

Plant on which insects Mean 6 SE pollen Mean 6 SE percent of Plant species (mean 6 SE percent Total number were collected Insect species (N ) grains per bee total pollen load of total pollen load) pollen types

Cultivated plants A. cicer B. huntii (16) 219,181 6 60,061 94.5 6 5.1 Co. arvensis (5.2 6 5.1) 5 A. millefolium H. ligatus (6) 21,615 6 12,010 33.3 6 21.1 Ci. arvense (60.7 6 19.8) 4 Co. arvensis (5.3 6 5.3) D. candida A. mellifera (45) 99,620 6 23,839 98.0 6 0.6 - 6 Ag. angelica/texanus (11) 18,048 6 7155 84.0 6 6.9 Ci. arvense (10.2 6 6.5) 6 Co. arvensis (4.4 6 2.4) B. griseocollis (40) 272,748 6 51,450 93.3 6 3.3 L. corniculatus (5.4 6 3.2) 7 B. huntii (26) 229,798 6 79,991 92.1 6 3.5 L. corniculatus (4.6 6 2.9) 7 A. cicer (1.7 6 1.7) C. petalostemonis (7) 20,985 6 7928 99.5 6 0.4 - 4 C. phaceliae (14) 21,586 6 7810 98.2 6 1.8 S. chilensis (1.8 6 1.8) 2 H. ligatus (5) 23,834 6 11,455 79.5 6 0.20 R. columnifera (20.5 6 19.5) 3 H. rubicundus (3) 62,470 6 59,076 99.8 6 0.2 - 2 Lasioglossum spp. (10) 6126 6 3093 96.7 6 3.3 Ci. arvense (3.3 6 3.3) 2 M. lippiae (3) 67,762 6 60,019 66.6 6 32.7 L. corniculatus (32.6 6 32.0) 4 S. ichneumoneus (24) 449 6 111 82.5 6 6.3 R. columnifera (11.7 6 5.2) 6 A. millefolium (4.0 6 4.0) T. sayi (7) 12,948 6 8708 99.3 6 0.5 - 4 R. columnifera A. mellifera (4) 37,553 6 13,085 99.3 6 0.4 - 2 B. huntii (4) 45,686 6 45,674 46.7 6 26.6 D. candida (24.8 6 24.7) 4 S. chilensis (24.8 6 24.8) Co. arvensis (3.6 6 3.6) H. rubicundus (14) 104,781 6 23,218 99.1 6 0.4 - 4 M. parallela (5) 461,510 6 267,658 79.4 6 19.8 Ci. arvense (19.8 6 19.8) 4 Melissodes sp. 2 (10) 455,300 6 94,428 99.1 6 0.6 - 4 OUE8,ISE2105 2 ISSUE 85, VOLUME

Table 2. Continued.

Pollen from plant where insect was collected Other pollen sources

S. chilensis H. ligatus (15) 18,267 6 4067 96.1 6 2.0 Melilotus sp. (1.7 6 1.5) 6 Melissodes sp. 1 (16) 53,747 6 14,537 92.7 6 2.0 Co. arvensis (4.2 6 1.6) 4 D. candida (2.1 6 1.3) Melissodes sp. 3 (8) 97,503 6 34,040 87.8 6 9.3 Ci. arvense (9.1 6 9.1) 4 Weeds Co. arvensis Ag. angelica/texana (6) 1096 6 407 70.2 6 11.0 D. candida (29.8 6 11.0) 2 Dianthidium sp. 1 (5) 4518 6 2220 63.3 6 11.4 Ci. arvense (14.8 6 14.8) 4 R. columnifera (13.9 6 9.0) D. candida (8.0 6 5.0) Halictus ligatus (5) 1156 6 645 31.3 6 18.2 R. columnifera (54.5 6 14.8) 5 H. confusus (32) 410 6 123 85.0 6 5.8 R. columnifera (6.5 6 3.6) 6 D. candida (4.1 6 2.3) L. corniculatus (2.1 6 2.1) Co. arvense (1.9 6 1.4) Halictus virgatellus Cockerell (4) 1950 6 1530 97.9 6 2.1 - 2 Lasioglossum spp. (5) 205 6 74 80.2 6 22.1 D. candida (19.8 6 22.1) 2 Sonchus sp. A. mellifera (19) 52,251 6 22,651 84.7 6 6.0 Co. arvensis (5.2 6 4.6) 9 Symphoricarpos (4.0 6 4.0) D. candida (2.1 6 1.1) A. millefolium (1.7 6 0.5) 106 JOURNAL OF THE KANSAS ENTOMOLOGICAL SOCIETY

(Richards, 1986; Richards and Myers, 1997) and O. viciaefolia (Richards and Edwards, 1988) are dependent on insects for pollination, with honey bees, bumble bees, and leafcutting bees being the primary flower visitors. The presence of the apiary at BPMC means that honey bees likely provided major pollination services. Cane (2008) notes that managed bees ‘‘will initially be necessary, at least as a ‘bridge’ for growers in early production years’’ after farms are started for wildflower seed production. However, at BPMC honey bees were unevenly distributed among the cultivated species. They comprised nearly three-quarters of the bees on D. candida and O. viciaefolia and about one-third of those on S. albus, but #10% of those on A. cicer, A. millefolium, R. columnifera,andS. chilensis. Colletes phaceliae also apparently focused on D. candida, whereas Melissodes sp. 2 and H. rubicundus were more abundant on R. columnifera, and Bombus were most abundant on A. cicer, D. candida, and O. viciaefolia. Similar non-random plant associations were apparent in other bees. Others have noted plant associations similar to those we observed. Honey bees were rare visitors to Ratibida pinnata in Minnesota, even when an apiary was present at the site (Dickinson and McKone, 1992). At that site, native bees of both sexes visiting Ratibida pinnata (Vent.) Barnh. included Melissodes (37%), Halictus (23%), Andrena (19%), and Bombus (10%). The combined total for these for genera (89%) is similar to their total percentage among non-Apis bees visiting R. columnifera at BPMC (82%). The four Melissodes species observed on R. pinnata in Minnesota were said to be specialists on Asteraceae by LaBerge (1961). Similarly, Hilty (2011) states that ‘‘Colletes albescens and Colletes robertsonii are oligoleges’’ of Dalea purpurea. Although most pollen grains carried by bees likely end up in nest-provisions, rather than on the stigmas of conspecific flowers, the high level of purity of the pollen loads on many of the bees on A. cicer, D. candida, R. columnifera, and S. chilensis suggests a high degree of flower constancy among the bees found on those flowers. Relatively high fidelity also occurred, at least over the short term, for A. mellifera on Sonchus and H. ligatus on Co. arvensis. Although we found Halictus ligatus throughout BPMC and it is listed as a generalist by Moure and Hurd (1987), it carried mostly pure S. chilensis pollen when collected on that plant species. One curious aspect of our results is the relative scarcity in our sweep samples of M. rotundata, an important commercially-managed pollinator of alfalfa grown for seed (Pitts-Singer 2008) and one that may hold promise on wildflower seed farms (Cane, 2008). Even on alfalfa seed fields in Montana, M. rotundata females forage on a wide variety of plants, including species of Asteraceae, Brassicaceae, Chenopodaceae, Fabaceae, and Scrophulariaceae (Jensen et al., 2003; O’Neill et al., 2004; O’Neill and O’Neill, 2010). However, despite the presence of a feral population of M. rotundata at BPMC during our study (O’Neill et al., 2010), we found little evidence that it was an important pollinator of the cultivated species grown in 2006 and 2007 (which included Asteraceae and Fabaceae). Megachile rotundata comprised ,0.1% of bees observed on D. candida and ,1% of those on the six other cultivated species. Thus, M. rotundata must have been foraging on plants that we did not systematically sample for bee communities, such as Canada thistle, birdsfoot trefoil, sweetclover, and alfalfa, the latter which grew in an field along the northern edge of BPMC. BPMC not only provides bees with a diversity of cultivated and non-cultivated pollen and nectar sources, but also large areas of non-tilled soil of various slopes, aspects, and levels of compaction that may provide nesting substrate for genera such VOLUME 85, ISSUE 2 107 as Andrena, Melissodes, Colletes, Halictus, and Lasioglossum. In addition, a large expanse of uncultivated grassland just to the east of BPMC may provide a source of ground-nesting bees. Solitary bees of other genera, such as Hylaeus, Ashmeadiella, Hoplitis, Megachile, and Osmia (some of which we reared from trap-nests at BPMC; O’Neill et al., 2010), likely nest in cavities and crevices provided by cottonwood trees (Populus deltoides Marsh.), fence posts, and farm buildings at BPMC. If nesting habitat suitable for different non-managed bees was unevenly distributed at BPMC, some of the bee-plant associations we observed could be partially influenced by the proximity of nesting habitat to cultivated plots. Thus, it is not clear to what degree the uneven distribution of different bee species on flowers at BPMC was a manifestation of different floral preferences. However, floral preferences were likely important in some cases. For example, D. candida, which flowers through much of July and August, was much more attractive to honey bees than was R. columnifera, which bloomed at the same time. Even though the R. columnifera field was between the apiary and the upper D. candida field, most honey bees apparently ignored it in favor of D. candida and O. viciaefolia, which was even further to the south. At BPMC, the low numbers of honey bees on A. cicer, A. millefolium, R. columnifera, and S. chilensis suggest that native, non-managed bees of such genera as Bombus, Melissodes, Halictus,andLasioglossum may be critical for plant species for which honey bees show low preference. But whether non-Apis bees alone can provide sufficient pollination services for D. candida, needs to be assessed in situations where honey bee populations are not significantly enhanced by the presence of local apiaries. Finally, it would also be valuable to determine whether native bees are presently limited by the availability of nesting substrate and whether nest sites could be provided in an economical manner to enhance pollination services.

Acknowledgments We thank Bill Grey, Larry Holzworth, Katie Hopp, Mark Majerus, Ruth O’Neill, Joseph Scianna, and Susan Winslow for advice and assistance in various aspects of the study. This research was supported by the Montana Seed Foundation, the USDA-NRCS Bridger, and the Montana Agricultural Experiment Station.

Literature Cited Cane, J. H. 2006. An evaluation of pollination mechanisms for purple prairie-clover, Dalea purpurea (Fabaceae: Amorpheae). American Midland Naturalist 156:193–197. Cane, J. H. 2008. Pollinating bees crucial to farming wildflower seed of U.S. habitat restoration.In James, R. and T. L. Pitts-Singer (ed.). Bee Pollination in Agricultural Ecosystems, pp. 105–123. Oxford University Press, New York. xiv + 232 pp. Dickinson, J. A., and M. J. McKone. 1992. Insect floral visitors to four species of tall-grass prairie composite (Asteraceae: Helianthinae). Prairie Naturalist 24:159–174. Droege, S. 2008. North American (North of Mexico) introduced and alien bee species. URL: ,http:// pollinators.nbii.gov/documents/NAm_Introduced_and_Alien_Bee_Species_Jul2008.pdf. Last accessed 10 June 2011. Everitt, B. S. 1977. The Analysis of Contingency Tables. Chapman and Hall, New York. 128 pp. Hilty, J. 2011. Prairie wildflowers of Illinois. URL: ,http://www.illinoiswildflowers.info/prairie/plantx/ ppr_cloverx.htm. Last accessed 9 June 2011. Jensen, P. D., K. M. O’Neill, and M. Lavin. 2003. Pollen provision records for three species of solitary bees (Hymenoptera: Megachilidae; Megachile, Heriades) in southwestern Montana. Proceedings of the Entomological Society of Washington 105:196–203. 108 JOURNAL OF THE KANSAS ENTOMOLOGICAL SOCIETY

LaBerge, W. E. 1961. A revision of the bees of the genus Melissodes in North and Central America. Part III (Hymenoptera: Apidae). University of Kansas Science Bulletin 42:283–663. Ludwig, J. A., and Reynolds, J. F. 1988. Statistical Ecology. J. Wiley and Sons, New York. 337 pp. Mader, E., M. Shepherd, M. Vaughan, S. Hoffman Black, and G. Lebuhn. 2011. Attracting Native Pollinators: Protecting North America’s Bees and Butterflies. Storey Publishing, North Adams, MA. 371 pp. Majerus, M., and L. Holzworth. 2003. Antelope germplasm selected class slender white prairieclover. URL: ,http://www.plant-materials.nrcs.usda.gov/pubs/mtpmcbr5130.pdf. Last accessed 4 May 2011. Moure, J. S., and P. D. Hurd, Jr. 1987. An Annotated Catalog of the Halictid Bees of the Western Hemisphere (Hymenoptera: Halictidae). Smithsonian Institution Press, Washington, DC. vii + 405 pp. O’Neill, K. M., R. P. O’Neill, S. Blodgett, and J. E. Fultz. 2004. Variation in Megachile rotundata pollen load composition in relation to flower diversity (Hymenoptera: Megachilidae). Journal of the Kansas Entomological Society 77:619–625. O’Neill, K. M., A. M. Pearce, R. P. O’Neill, and R. S. Miller. 2010. Correlates of offspring size and sex ratio variation in a feral population of alfalfa leafcutting bees (Hymenoptera: Megachilidae; Megachile rotundata F.) Annals of the Entomological Society of America 103:775–784. O’Neill, R. P., and K. M. O’Neill. 2010. Assessment of pollen load composition and size in the leafcutting bee Megachile rotundata L. (Hymenoptera: Megachilidae). Apidologie 42:223–233. Pearce, A. 2008. Pollinators of Slender White Prairieclover. M.S. Thesis, Montana State University; Bozeman, Montana. xiii + 195 pp. Pitts-Singer, T. L. 2008. Past and present management of alfalfa bees. In James, R. and T. L. Pitts-Singer (eds.). Bee Pollination in Agricultural Ecosystems, pp. 105–123. Oxford University Press, New York. xiv + 232 pp. Proctor, M., P. Yeo, and A. Lack. 1996. The Natural History of Pollination. Timber Press Inc. Richards, K. W. 1986. Pollination requirements of cicer milkvetch, Astragalus cicer L. Journal of Range Management 39:457–459. Richards, K. W., and P. D. Edwards. 1988. Density, diversity, and efficiency of pollinators of sainfoin, Onobrychis viciaefolia Scop. Canadian Entomologist 120:1085–1100. Richards, K. W., and T. W. Myers. 1997. Commercially managed colonies of bumblebee for pollination of cicer milkvetch. Acta Horticulturae (ISHS) 437:293–298. Robson, D. B. 2010. A comparison of flower-visiting insects to rare Symphyotrichum sericeum and common Solidago nemoralis (Asteraceae). Botany 88:241–249. vanEngelsdorp, D., and M. D. Meixner. 2010. A historical review of managed honey bee populations in Europe and the United States and the factors that may affect them. Journal of Invertebrate Pathology 103:S80–S95. Winslow, S. R., M. Majerus, and L. Holzworth. 2005. Stillwater germplasm selected class prairie coneflower. URL:,http://www.plant-materials.nrcs.usda.gov/pubs/mtpmcbr6016.pdf. Last accessed 5 May 2011.