USBombus, a database of contemporary survey data for North American Bumble Bees (Hymenoptera, Apidae, Bombus) distributed in the United States

Abstract Background Bumble bees (Hymenoptera: Apidae, Bombus) are pollinators of wild and economically important flowering plants. However, at least four bumble bee species have declined significantly in population abundance and geographic range relative to historic estimates, and one species is possibly extinct. While a wealth of historic data is now available for many of the North American species found to be in decline in online databases, systematic survey data of stable species is still not publically available. The availability of contemporary survey data is critically important for the future monitoring of wild bumble bee populations. Without such data, the ability to ascertain the conservation status of bumble bees in the United States will remain challenging. New information This paper describes USBombus, a large database that represents the outcomes of one of the largest standardized surveys of bumble bee pollinators (Hymenoptera, Apidae, Bombus) globally. The motivation to collect live bumble bees across the United States was to examine the decline and conservation status of Bombus affinis, B. occidentalis, B. pensylvanicus, and B. terricola. Prior to our national survey of bumble bees in the United States from 2007 to 2010, there have only been regional accounts of bumble bee abundance and richness. In addition to surveying declining bumble bees, we also collected and documented a diversity of co-occuring bumble bees. However we have not yet completely reported their distribution and diversity onto a public online platform. Now, for the first time, we report the geographic distribution of bumble bees reported to be in decline (Cameron et al. 2011), as well as bumble bees that appeared to be stable on a large geographic scale in the United States (not in decline). In this database we report a total of 17,930 adult occurrence records across 397 locations and 39 species of Bombus detected in our national survey. We summarize their abundance and distribution across the United States and association to different ecoregions. The geospatial coverage of the dataset extends across 41 of the 50 US states, and from 0 to 3500 m a.s.l. Authors and respective field crews spent a total of 512 hours surveying bumble bees from 2007 to 2010. The dataset was developed using SQL server 2008 r2. For each specimen, the following information is generally provided: species, name, sex, caste, temporal and geospatial details, Cartesian coordinates, data collector(s), and when available, host plants. This database has already proven useful for a variety of studies on bumble bee ecology and conservation. However it is not publicly available. Considering the value of pollinators in agriculture and wild ecosystems, this large database of bumble bees will likely prove useful for investigations of the effects of anthropogenic activities on pollinator community composition and conservation status.

dataset represents a systematic survey that promises to be useful in future investigations of bumble bee ecology, conservation and policy.

Study area description:
This dataset covers a wide range of ecoregions found throughout the continental United States and Alaska, from 29° to 68° latitude and -150° to -68°l ongitude (Figs 1,2). Bumble bees reported in this dataset were surveyed in wild, urban, and agricultural landscapes across 41 states from 2007 to 2010. A special effort was made to document bumble bees distributed in US national parks and other federally protected areas, as these lands would likely have been less impacted by anthropogenic land-use change, agricultural intensification, and zoonotic diseases transmitted from commercially reared bumble bees. Nine states and Washington D.C. are not represented in our systematic survey primarily because they were relatively close to states and ecoregions that were intensively sampled (Figs 1, 3). The states not included in this survey and database are Delaware, Florida, Maryland, Michigan, New Hampshire, New Jersey, West Virginia, Rhode Island. Hawaii was not surveyed as bumble bees are not found on this archipelago.  Distribution of bumble bee species richness detected in surveys in the contiguous United States and Alaska. Warmer colors represent high species richness whereas cooler colors represent low richness. Species richness is simply defined as the number of different species detected at a study site. US states not included in the dataset are cross-hatched.  Percentage of specimen records per species detected in the western United States, including Alaska (Suppl. material 1). Western sites are defined as survey sites that are west of the Colorado Rockies (104th western longitude). Percentage of specimen records per species detected in the eastern United States. Eastern sites are defined as survey sites that are east of the Colorado Rockies (104th western longitude). Bumble bees that are found in both the western and eastern United States are grouped with the western bumble bee species in Fig. 4 (e.g., Bombus griseocollis) (Suppl. material 2).
We describe the distribution of bumble bees based on political boundaries and ecoregions that have been developed by the World Wildlife Fund for Nature (WWF) (Olson et al. 2001). A total of 55 ecoregions were surveyed in our national study of bumble bees (Fig.  3). In our survey B. sandersoni was detected only in the Appalachian-Blue Ridge Forests ecoregion, and is represented by a single specimen. However, it is likely that we did not survey at an optimal time for B. sandersoni as it has been recorded to be in high abundance in some parts of New York, New England, Tennessee, and North Carolina (Hatfield et al. 2015). Furthermore, given that multiple eastern North American bumble bees converge on similar color banding patterns, it is possible that we may have misidentified them in the field (Williams et al. 2014). In contrast, B. griseocollis was detected in 29 ecoregions across the conterminous United States, representing the species with the most ecoregion-diverse distribution in this dataset ( Table 1). The initial goal of our study was not to survey across all North American ecoregions equally, but rather investigate ecoregions and states where historic abundances of suspected declining North American bumble bee species were high . Based on WWF ecoregions, 62% and 18% of the bumble bees surveyed were collected in critically endangered and vulnerable ecoregions in the United States, respectively (Table 1) (Olson et al. 2001). Only 20% of the surveyed bumble bees were distributed in habitat that has been identified by the WWF as ecoregions that are relatively stable or intact (Table 1). In the western United States, most surveys took place in alpine environments (e.g., Cascade, Sierra-Nevada, and Rocky Mountains) and high elevation basins and plateaus (> 500 m). In the eastern United States, surveys were conducted across a variety of different habitats including prairies and deciduous forests. In Alaska, bumble bees were primarily surveyed in the tundra and taiga, specifically adjacent to large rivers ( Fig. 3 Design description: The purpose of the dataset is to make available data associated with a standardized survey of bees of the genus Bombus in the United States. That database was developed during the course of an assessment on the conservation status, disease ecology, genetic diversity, and decline of the following North American bumble bees: B. affinis, B. occidentalis, B. pensylvanicus and B. terricola , Cordes et al. 2012. The authors Jonathan Koch, James Strange, Terry Griswold, and their field crew primarily collected bumble bees in the western U.S.A. and Alaska while Sydney Cameron, Jeffrey Lozier, Nils Cordes, Leellen Solter, Isaac Stewart and their field crew collected bumble bees in the eastern U.S.A. (Fig. 1). Bumble bees collected by the western group were identified, labelled, pinned, and curated into the US National Pollinating Insect Collection housed at the USDA-ARS Pollinating Insects-Biology, Management, and Systematics Research Laboratory (PIBMSRL) in Logan, Utah. Bumble bees collected by the eastern group were identified in the field to species and released after the survey was completed. Specimens were retained in the western United States and Alaska surveys as several species are cryptic and notorious for misidentification . In the eastern United States survey, bumble bees were only retained if the specimens could not be identified to species with complete confidence. Eastern specimens were released as bumble bees could be confidently identified to species using field guides and taxonomic keys. Specifically, specimens of imperiled bumble bees identified in the  study, as well as B. vosnesenskii, B. bifarius, B. bimaculatus, and B. impatiens were retained for population genetic analysis and pathogen surveys. Species identifications were made by the authors with taxonomic keys (Stephen 1957, Thorp et al. 1983, LaBerge and Webb 1962, Mitchell 1962, Medler and Carney 1963, Chandler and McCoy 1965, Husband et al. 1980, Williams et al. 2008, Williams et al. 2014. Specimen data in the USBombus dataset has been digitized and entered into the US National Pollinating Insects Database (USNPID). Bumble bees collected by the western group have been affixed a six digit matrix barcode with the acronym BBSL. The acronym BBSL (Bee Biology and Systematics Laboratory) is in reference to a previous title of the PIBMSRU. Each physical specimen and associated data is represented by a single BBSL barcode. Bumble bee occurrence and abundance data collected by the eastern group have been incorporated into USBombus dataset in a manner different than the bumble bee specimens collected by the western group. For the eastern data each unique barcode represents the combination of one species with a single collection event (i.e., specific field site and date) with the abundance of each sex (male or female), and caste (queen or nonqueen) recorded. These survey events have a six digit matrix barcode with the prefix EBOD (Eastern Bombus). Both eastern (EBOD) and western (BBSL) specimen data have been entered into the USNPID using data entry forms with Microsoft Access 2008 r2. The USNPID represents one of the largest digital repositories of pollinating insects globally and has been used in numerous ecological, agricultural, and taxonomic investigations (e.g., Griswold et al. 2014).
All locations were georeferenced with a Garmin GPS unit in the field with the coordinate form of decimal latitude and longitude in the WGS84 datum. In this paper specimen records are represented geospatially using ArcGIS and WWF Biotic Regions (Figs 1, 2

Sampling methods
Study extent: This dataset was primarily developed to determine the extent of bumble bee decline in the United States. Thus, we did not survey in areas that have historically been under-sampled for bumble bees, nor did we survey well-sampled areas outside of the known ranges of the four focal species suspected to be in decline . Much of our survey efforts were guided by natural history specimen data that was digitized retroactively (Grixti et al. 2009, Koch 2011. The intent to survey in areas that were once populated with currently rare and declining bumble bee species was to determine changes in genetic structure over time, disease ecology, and population abundances (Lozier and Cameron 2009, Cordes et al. 2012, Cameron et al. 2007). Thus we sampled across both latitude and elevation gradients in a way that maximized our ability to detect and capture bumble bees when colony growth was at its maximum in the summer months of the northern hemisphere.
Sampling description: Specimens represented in the USBombus dataset are the result of systematic surveys conducted by researchers at the USDA-ARS-PIBMSRL, Utah State University, University of Illinois, and Illinois Natural History Survey. Surveys were conducted primarily using sweep nets to capture bumble bees on flowers and in flight. All surveys were timed and conducted for at least 0.5 hours (average of ∼1 ± 0.5 SD survey hours per site). Surveys were conducted by walking through floral patches and collecting all observed bumble bees without consideration of species identity. Site selection was based on locality data present on natural history collections and species distribution models. Specimens were collected with aerial nets while in flight or while foraging at flowers; then, they were placed in vials and chilled on ice until the end of the collection period. This dataset represents a total of 512 collector hours. Survey methods are further described in ). The number and name of surveyors can be queried from the USNPID by contacting the database manager associated with this data publication.
Quality control: All unrecognizable individuals collected in the field were carefully examined by the authors using taxonomic keys and field guides (Husband et al. 1980, LaBerge and Webb 1962, Mitchell 1962, Medler and Carney 1963, Thorp et al. 1983, Williams et al. 2008, Chandler and McCoy 1965, Stephen 1957. The authors are considered to be authorities in bumble bee identification in North America ) and globally (Williams et al. 2008).
Step description: All specimens described in this dataset have been batch entered into the USNPID following the flowchart in Fig. 6. With the exception of data collected by the eastern group (University of Illinois and Illinois Natural History Survey), specimen identification and subsequent update to the database occurred after record and event metadata had been entered into the USNPID. Bumble bee identification and associated metadata of bumble bees collected by the eastern group were retroactively captured from a spreadsheet and imported in the USNPID. In the USNPID dataset bumble bee queens are denoted by the Q identifier (0 = False, -1 = True). Workers and Queens are denoted by the F identifier as a quantity (0 -∞) and males are denoted by the M identifier as a quality (0 -∞). Values greater than one in these fields (M, F) indicate the total abundance of the specimens associated with that caste in the survey event and is specific to occurrence records associated with the EBOD prefix. Quantities were mapped to the Darwin Core DWC field "Individual Count" with cast and sex mapped to the DWC field "Sex" (Female, Female Queen, Male, Unknown Sex).

Geographic coverage
Description: This dataset includes occurrence records of bees in the genus Bombus across 41 states in the contiguous United States and Alaska. Surveys have taken place over a wide elevation gradient, starting at near-sea level sites including Galveston, Texas and San Juan Islands, Washington to 3500 m a.s.l. in Gothic, Colorado. Considerable effort was also made to survey multiple bumble bee communities north of the Arctic Circle (68° latitude) in Alaska. However, the majority of the field sites represented in this dataset are found throughout in grassland and alpine biomes of the contiguous United States (Figs  1, 2, 3).

Taxonomic coverage
Description: USBombus includes 39 species of the bee genus Bombus known to occur in the Nearctic region of the Western Hemisphere (Figs 4,5). Thus our survey efforts and this dataset document approximately 82% of the described Bombus species in North America north of Mexico (Williams et al. 2014). Bombus is the only extant genus of the tribe Bombini in the family Apidae. There are an estimated 250 described species across 15 subgenera of Bombus worldwide (Williams et al. 2008). Bumble bees are primitively eusocial insects and form colonies in which a division of labor exists among workers (females), drones (male), and queens (females). We differentiate between workers and queens in our dataset with unique identifiers (see dataset description).
In our dataset of North American Bombus, the subgenus Pyrobombus is the most abundant and most species-rich of the eight subgenera found in the Nearctic. In total, 12,780 bees representing 19 species in the subgenus were detected. In the western United States (including Alaska) the most widespread and abundant bumble bee is B. bifarius (Fig. 4), while in the eastern United States the species most abundant is B. impatiens (Fig.  5). In addition to being an abundant native bee, B. impatiens is commercially reared to pollinate a variety of crops including tomatoes and blueberries (Velthuis and Doorn 2006).
The least abundant and species-poor subgenus detected in our survey was Alpinobombus, represented by one species, B. balteatus. We also collected four species of bumble bees in the parasitic subgenus Psithyrus: B. insularis, B. fernaldae, B. suckleyi, and B. citrinus. We did not detect B. ashtoni in our survey. Psithyrus comprises a unique group of bumble bees in which the females usurp bumble bee colonies, bully or kill the subordinate queen, and use the queen's daughters to rear her own offspring.
The taxonomic status of three species in our dataset has been debated within the past decade, specifically cryptarum, and B. moderatus were found to be good species. In this dataset we maintain the species status as defined with molecular data by Cameron et al. 2007) and proposed taxonomy by Thorp et al. 1983). Finally, we did not detect B. cockerelli (= B. vagans) while surveying within its historic range ( Fig. 1) 4,5). The limited number of B. terricola may be due to low survey coverge in the Northeast (Fig. 1) where published species distribution models of B. terricola predict to be of high habitat suitability . Bombus franklini, which was not detected in our survey effort has the smallest known geographic distribution and only occurs in one ecoregion

Specimen preservation method: Dried and Pinned Specimens
Curatorial unit: Of the 17,930 bumble bee records, 9,380 records represent 9,363 dried and pinned adult individuals affixed with label data and matrix barcode. The specimens are housed in standard insect museum drawers and preserved from dermestid beetle damage by routine freezing of drawers at -20°C. All specimens are housed at the U.S. National Pollinating Insect Collection in Logan, Utah and are individually represented by the barcode prefix BBSL. The remaining 831 digital records represent 8,567 bumble bees that were caught and released in the field after identification in the eastern U.S.A. Thus, no pinned specimen or label data are associated with these data. These observation records are represented by the barcode prefix EBOD. All species determinations were made by authorities in bumble bee taxonomy, identification, and natural history.