Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2025 Nov 13;26(1):e70073. doi: 10.1111/1755-0998.70073

Sensitive Environmental DNA Methods for Low‐Risk Surveillance of At‐Risk Bumble Bees

Rodney T Richardson 1,✉, Grace Avalos 2, Cameron J Garland 2, Regina Trott 3, Olivia Hager 3, Mark J Hepner 4, Clayton Raines 5, Karen Goodell 6
PMCID: PMC12624520  PMID: 41229281

ABSTRACT

Terrestrial environmental DNA (eDNA) techniques have been proposed as a means of sensitive, non‐lethal pollinator monitoring. To date, however, no studies have provided evidence that eDNA methods can achieve detection sensitivity on par with traditional pollinator surveys. Using a large‐scale dataset of eDNA and corresponding net surveys, we show that eDNA methods enable sensitive, species‐level characterisation of whole bumble bee communities, including rare and critically endangered species such as the rusty patched bumble bee (RPBB; Bombus affinis ). All species present in netting surveys were detected within eDNA surveys, apart from two rare species in the socially parasitic subgenus Psithyrus (cuckoo bumble bees). Further, for rare non‐parasitic species, eDNA methods exhibited similar sensitivity relative to traditional netting. Compared with flower eDNA samples, sequenced leaf surface eDNA samples resulted in significantly lower rates of Bombus detection, and these detections were likely attributable to high rates of background eDNA on environmental surfaces, perhaps due to airborne eDNA or eDNA movement during rainfall events. Lastly, we found that eDNA‐based frequency of detection across replicate surveys was strongly associated with net‐based measures of abundance across site visits. We conclude that the COI‐based metabarcoding method we present is cost‐effective and highly scalable for quantitative characterisation of at‐risk bumble bee communities, providing a new approach for improving our understanding of species habitat associations.

Keywords: ecological restoration, endangered species act, occupancy modelling, pollinator conservation, presence absence, regulatory compliance, taxonomic bottleneck

1. Introduction

The global environment is constantly changing, and these fluctuations have important implications for the maintenance of habitable conditions and the availability of natural capital for future generations. The modern era has been marked by accelerated rates of environmental change (Elsen et al. 2022; Mottl et al. 2021; Potapov et al. 2022), making the monitoring of such change an important goal for taxa that provide critical ecosystem services. To better understand changing species distributions, as well as how populations respond to global environmental change, researchers need sensitive and scalable survey methods. This is particularly true for insects, which represent numerous critical taxa, including pollinators, and are notoriously difficult to monitor with traditional methods. With recent advances in the cost‐effectiveness and accuracy of DNA sequencing, environmental DNA (eDNA) methods have grown in popularity, with applications now spanning a variety of use cases in both aquatic and terrestrial settings (Allen et al. 2021; Avalos et al. 2024; Jerde et al. 2011; Krehenwinkel et al. 2022). Considering these advances, eDNA methods are expected to complement traditional survey methods by alleviating many of the bottlenecks associated with traditional means of population monitoring, though careful evaluation of performance is needed. Here, we demonstrate the strong potential of terrestrial eDNA methods for large‐scale monitoring of bumble bees (genus Bombus), a taxonomic group of valuable pollinating species of high conservation concern.

Bombus comprise an important component of robust pollinator communities, with over 250 species globally, approximately one‐third of which are believed to be in decline (Arbetman et al. 2017). Prior to the introduction of European honey bees ( Apis mellifera ), Bombus were the only eusocial pollinators in temperate regions of the North America. With relatively large body sizes and heightened visual capacities (Lunau et al. 2009), these species exhibit unique traits relative to most Anthophila (bees), including the ability to forage in low‐light conditions, at low temperatures and over long distances (Goulson 2003; Heinrich 2004; Tichit et al. 2024). Such traits make Bombus ideal for pollination in cold climates, forest understories and on plant species that require buzz pollination, including numerous agriculturally important crops (Kendall et al. 2019; Miller‐Struttmann et al. 2017; Oyen et al. 2016). Thus, the services provided by Bombus are often difficult to replace with other pollinator taxa (Brosi and Briggs 2013; Willmer et al. 1994). Unfortunately, Bombus have experienced range contractions and widespread turnover in community composition over recent decades, largely driven by dramatic declines in formerly widespread species (Arbetman et al. 2017; Grixti et al. 2009; Morales et al. 2013). These changes raise concern about the future of this functionally important pollinator group.

In North America, there is strong evidence of declines within the subgenera Bombus and Thoracobombus (Cameron et al. 2011; Colla and Packer 2008; Graves et al. 2020). In 2017, the rusty patched bumble bee (RPBB; Bombus affinis ) was listed under the US Endangered Species Act (Christopher 2017). At the time of listing, the species was widely believed to be restricted to areas of the Upper Midwest US until recent work revealed a persistent population within the Central Appalachian Mountains (Hepner et al. 2024). Accordingly, the Central Appalachian Mountains are an under‐surveyed region of high conservation potential that also harbors populations of other bumble bee species of concern, including B. terricola and B. pensylvanicus . To improve our knowledge of Bombus communities within this region, and test novel detection methods, we conducted a large‐scale survey of the region using both traditional netting and novel environmental DNA methods. Here, we evaluate the general efficacy of our eDNA methods and compare them to netting in terms of detection sensitivity, taxonomic breadth of detection, and quantitative reliability.

2. Methods

From June to September 2022, we conducted 84 site visits across 55 sites. Sites consisted of linear roadside rights‐of‐way, open areas or forest understories with abundant floral resources, typically 0.25–0.5 ha in area. The sampling region predominately included the Central Appalachian Mountains, USA, with one additional site near Easton, Maryland (MD). Survey sites were selected based on permitted access to public and some private lands based on local, state and federal permits. Within this region, specific sites for surveys were selected based on the availability of flowers attractive to bumble bees. We typically stayed close to navigable roads to maximize the area covered and number of survey sites. During each site visit, eDNA and netting surveys were conducted, with collection of 2–5 eDNA samples followed by collection of 1–6 net surveys. When flower patches were dense but only encompassed a small spatial area, multiple eDNA samples could be collected efficiently, while the feasibility of repeating net samples was limited. When surveying sites of perceived high value (i.e., dense patches of bumble bee‐attractive flowers) that encompassed large spatial areas, we increased net survey intensity relative to eDNA sampling. Notably, eDNA samples collected for this work have been reported on previously in an unrelated manuscript on the use of eDNA methods for broad‐scope surveillance of all Arthropoda (Richardson et al. 2025).

2.1. eDNA Sample Collection

During each site visit, flower eDNA and leaf surface eDNA samples were collected. For flower eDNA samples, flowers of a single plant species were clipped into a sterile quart‐sized, zip‐closable plastic bag until the bag was approximately ¼ to ¾ full of loose flower material. Samples were mixed with 25 mL of eDNA preservative (10% EtOH v/v, 40% propylene glycol v/v and 0.25% sodium dodecyl sulfate w/v), mixed gently and allowed to sit for 1–2 min before the eDNA‐containing preservative rinse was carefully poured into a labelled 50 mL conical vial for transport to the laboratory. Using the same methods, leaf surface eDNA samples were collected from trees or shrubs that were at least 3 m from prominent floral forage and approximately 1–2 m above ground. Leaf surface eDNA samples were collected to serve as comparison to flower eDNA, which would presumably have higher detection rates. We were careful to select leaves that showed no evidence of insect herbivory or visible debris. During eDNA collection, researchers used flame‐sterilised tools and wore disposable sterile nitrile gloves to avoid cross‐contamination between sites or between netting and eDNA surveys. To further reduce the risk of contamination between methods, sampling of eDNA was conducted prior to corresponding net surveys.

2.2. Traditional Netting Surveys

Since the RPBB is a protected species, the US Fish and Wildlife Agency encourages researchers to use extreme caution when surveying in areas known to be occupied by the species. Accordingly, we used two different types of 10‐min net surveys, hereafter referred to as ‘indiscriminate’ and ‘targeted visual’, depending on perceived RPBB occupancy within the local landscape. During indiscriminate 10‐min net surveys, an observer walked a meandering transect, independent of other observers, through areas with dense floral cover while searching for bees. The observer netted continuously and indiscriminately, targeting bumble bees while also opportunistically collecting other bee species. Following indiscriminate surveys, the contents of the net, which typically contained between 5 and 50 live bees, were transferred to a one‐gallon zip‐closable bag using the technique described in Hepner (2024). Any B. affinis present in the sample was transferred to a clean glass vial after carefully making an incision in the bag, chilled briefly on ice, then photo‐documented and released. Other species of conservation concern, including B. pensylvanicus and B. terricola , were also often removed from the sample and released to limit lethal take of these species to one worker or male per site visit. We sterilised nets between surveys by spraying them with a 5% bleach solution to reduce DNA contamination and potential for disease transmission across sites.

During site visits where the perceived risk of netting B. affinis was high, we modified our survey methods to reduce the risk of harm to individual bees. In these situations, observers conducted ‘targeted visual’ surveys, where bumble bees were netted and handled individually and observers focused exclusively on a suite of high conservation value species that are distinguishable in the field, including B. affinis , B. pensylvanicus , B. terricola , B. fervidus , and B. auricomus . Data from these targeted visual surveys were only used for certain analyses, as described below, since they were not sensitive for detection of B. impatiens, B. bimaculatus, B. griseocollis, B. perplexus, B. vagans , B. sandersoni , B. flavidus and B. citrinus . Bumble bee specimens that were lethally collected were pinned using standard entomological practices or frozen individually in 1.5 mL microcentrifuge tubes. We identified each specimen using microscopic examination of characters detailed in Colla et al. (2011), and Williams et al. (2014). Specimens of cryptic species, B. perplexus, B. vagans and B. sandersoni , were distinguished with a combination of characters from Milam et al. (2020) and inner hind tibial spur differences (Z. Portman, University of Minnesota, pers. comm.).

2.3. Laboratory eDNA Processing

A custom salt‐ethanol precipitation protocol was used to pellet DNA from each sample and follow‐up purification procedures, including use of the Qiagen PowerClean Pro Cleanup Kit, were used for further removal of PCR inhibitors (see Richardson et al. 2025). Following DNA isolation, Illumina MiSeq amplicon libraries were prepared using a 3‐step PCR‐based protocol established in previous work (Richardson et al. 2019). During library preparation, the bumble bee‐specific COI primers from Milam et al. (2020) were used for the initial amplification. To quantify critical‐mistagging rates during downstream sequencing (Esling et al. 2015), we incorporated 15 no‐library negative control dual‐index combinations during library preparation. Amplified libraries were purified with the SequalPrep Normalisation Plate Kit, pooled equimolarly and subjected to Illumina MiSeq sequencing using a 2 × 150 cycle v2 flow cell.

2.4. Taxonomic Annotation of eDNA Data

After sequencing, VSEARCH v2.8.1 (Rognes et al. 2016) was used to merge paired‐end COI sequences and remove priming sites from the 5′ and 3′ ends. Sequences were taxonomically annotated by semi‐global VSEARCH top‐hit alignment against a custom reference sequence database trimmed to the amplicon regions of interest using MetaCurator (Richardson et al. 2020), with dependencies HMMER v3.1b2 (Eddy 2011), MAFFT v7.270 (Katoh et al. 2002), VSEARCH and Taxonomizr v0.11.1 (Sherrill‐Mix 2019). During alignment, a minimum query cover of 0.8 was required. The reference database was constructed using sequences from Eastern North American Bombus available through NCBI, downloaded on February 2nd, 2023. Following reference data curation, the database consisted of 61 sequences representing all species in the sampling region except the parasitic B. variabilis and B. insularis . Notably, all species were represented by two or more reference sequences with the exception of B. fraternus . During alignment, 100% identity matches were considered confident species‐level detections, and this threshold was strongly supported by leave‐one‐out cross‐validation analysis of available reference data (Figure S1). Since all Illumina sequencing runs produce low‐frequency misidentifications when inferring dual‐index tags, leading to critical mistags (Esling et al. 2015), we used an established technique to remove identifications with greater than 0.005 probability of representing a critical mistag‐associated detection (Richardson 2022). Additionally, we removed any detections which were represented by two or fewer sequences in a sample. All computational analysis was performed on the Owens cluster (Ohio Supercomputer Center 1987).

2.5. Statistical Analysis

Statistical analysis of all resulting data was conducted using R (R Core Team 2021). Following taxonomic annotation of sequences, leaf surface eDNA samples were compared against flower eDNA samples using a X 2 test to evaluate differences in the frequency of eDNA detection per sample and a Wilcoxon Rank Sum test to evaluate differences in Bombus species richness per eDNA sample. To analyze species detectability and occupancy patterns, the spOccupancy package (Doser et al. 2022) was used to produce an integrated occupancy model for each species in the dataset. Integrated occupancy models allow for the specification of multiple detection processes, eDNA and netting in this case, each of which informs occupancy estimation. Across methods, significant differences in per species detection or occupancy were assessed based on Bayesian 95% credible interval overlap. Since netting and eDNA data were collected in a relatively paired fashion and our main goal was to compare these survey methods, null intercept‐only models were specified for both the occupancy and detectability components of each model. To broadly compare detection sensitivity across methods, we regressed eDNA detectability estimates against net survey detectability estimates for each species using ordinary least squares regression. Lastly, on a per site visit basis, we regressed the eDNA‐based frequency of detection of each species against the log‐transformed mean number of Bombus individuals per net survey to assess the degree to which eDNA detection predicts species abundance across sites. A binomial generalised linear mixed‐effects model (glmmTMB; Brooks et al. 2017) was used for this analysis, with species and sampling site specified as crossed random intercept terms to account for repeated measures. Regressions were performed separately for the flower and leaf surface eDNA samples and we only used data from indiscriminate surveys for this analysis since targeted visual surveys provided no sensitivity for detecting 5 of 13 species present in the study system.

3. Results

Similar numbers of samples were taken for both eDNA and netting surveys (Figure 1A), with a total of 251 flower eDNA samples, 22 leaf surface eDNA samples, 165 indiscriminate 10‐min net surveys and 103 targeted visual surveys. Across the study, we detected 13 Bombus species, and the frequencies of detection for each species were similar between the two survey methods (Figure 1B).

FIGURE 1.

FIGURE 1

Open in a new tab

Pairwise comparison of sampling effort per site visit for each survey method (A), where points were jittered using random draws from a U(−0.3, 0.3) distribution and the mean, 20th and 80th quantiles are shown with square points and lines. A bar plot (B) shows comparison of frequency of Bombus species detection, with numbers above bars indicating the count of positive surveys for each method. For eDNA bars, data from N = 251 flower eDNA samples were included and leaf surface samples were excluded. For netting bars, data from N = 268 10‐min net surveys (165 indiscriminate net surveys and 103 targeted visual surveys) were used to estimate frequency of detection for species of conservation concern, listed in the methods, while N = 165 indiscriminate net surveys were used for the remaining common species.

3.1. eDNA Survey Results

Among flower eDNA surveys, sampled plant species spanned 20 families, with the bulk of samples originating from species within Asteraceae (N = 109), Lamiaceae (N = 49) and Fabaceae (N = 43). Illumina sequencing yielded a total of 5.7 million successfully mate‐paired reads, with a median sequencing depth of 20,850 sequences produced per flower eDNA sample. Leaf surface eDNA samples yielded significantly fewer sequences, with a median of 528 sequences per sample (Figure 2A, Wilcoxon rank‐sum test: p = 0.006). Most sequences, 63% in total, belonged to bumble bees and there was minimal evidence of variance in this percentage across flower and leaf surface samples (Figure 2B, Wilcoxon rank‐sum test: p = 0.086). As expected, flower eDNA samples exhibited significantly greater frequency of Bombus detection (Figure 2C; X 2 = 8.89, p = 0.001) and significantly greater Bombus species richness (Figure 2D, Wilcoxon rank‐sum test: p < 0.001) relative to leaf surfaces. Out of the 13 Bombus species recorded across all surveys, eDNA methods detected 11, including multiple detections of the federally endangered B. affinis . Both species that went undetected using eDNA belonged to the socially parasitic subgenus Psithyrus. Among 22 leaf surface eDNA samples, a total of 15 Bombus detections were observed. Negative control detections were composed predominantly of highly abundant species including B. impatiens (N = 6), B. bimaculatus (N = 4) and B. sandersoni (N = 3), with one detection each for B. vagans and B. griseocollis.

FIGURE 2.

FIGURE 2

Open in a new tab

Comparison of flower (N = 251) and leaf surface (N = 22) eDNA samples, in terms of total sequences per sample (A), proportion of sequences assigned to Bombus per sample (B), frequency of Bombus detection (C) and Bombus species richness per sample (D).

3.2. Net Survey Results

Net surveys resulted in a total of 2769 lethally collected Bombus specimens from indiscriminate surveys, as well as non‐lethal observation of approximately 2074 additional individuals during targeted visual surveys. Notably, targeted visual surveys yielded 12 observations of species of conservation concern, including 1 observation of B. terricola and 11 observations of B. pensylvanicus (all at a single site near Easton, MD). Among lethally collected specimens, 2759 were identified to species. The remaining 10 specimens could not be identified due to the poor condition of diagnostic features.

3.3. Comparison of eDNA and Net Results

Integrated occupancy modelling of each species resulted in well‐converged MCMC models (maximum Gelman‐Rubin R^ = 1.04) with large effective sample sizes (minimum ESS = 1059). Model results revealed similar detection probabilities between eDNA and netting (Figure 3A), with some notable deviations. Overall, per species mean detection probabilities were highly correlated across the two survey methods (OLS Regression, p = 0.001, R 2  = 0.60, Figure S2). However, for several species within the subgenus Pyrobombus, eDNA displayed significantly lower detection probabilities relative to netting. Similarly, eDNA detection sensitivity was significantly less than that of netting for the only species of Cullumanobombus in the data, B. griseocollis . While eDNA exhibited relatively lower detection sensitivities for species within the subgenera Psithyrus and Bombias, sample sizes were minimal, obscuring any assessment of significance. For species within the subgenera Bombus and Thoracobombus, detection sensitivities were similar across the two methods, with no significant differences (i.e., credible intervals overlapped). Occupancy estimates across the sampled sites varied considerably among species (Figure 3B), with means ranging from 0.03 ( B. pensylvanicus ) to 0.98 ( B. impatiens ). Mean occupancy of B. affinis was 0.22, with lower and upper credible bounds of 0.07 and 0.59, respectively. For species of primary conservation concern, B. affinis and B. terricola, eDNA and net‐based observations occurred within heavily forested montane landscapes (Figure 4A,B). A third species of heightened concern, B. pensylvanicus , was observed at only a single site (Figure 4C), though this species is known to be associated with grasslands (Novotny et al. 2021), which were relatively rare in the landscapes we sampled. Lastly, investigation of the relationship between eDNA‐based frequency of detection and net‐based species abundance revealed highly significant associations for both flower eDNA (Figure 5A: p < 0.001) and leaf surface eDNA (Figure 5B: p = 0.008) samples.

FIGURE 3.

FIGURE 3

Open in a new tab

Comparison of species detection probabilities between eDNA and netting (A) and estimates of species occupancy across the survey sites (B). Occupancy estimates were inferred using both eDNA and net detections with integrated single‐species models (Doser et al. 2022). Points indicate means, with lines extending to 95% credible bounds. For eDNA, leaf surface sample results were excluded and inferences are from N = 251 flower eDNA samples. For species of conservation concern (list provided in Methods), netting inferences are from N = 165 indiscriminate net surveys and N = 103 targeted visual surveys. For common species, netting inferences are from N = 165 indiscriminate surveys since there was no sensitivity for common species within targeted visual surveys.

FIGURE 4.

FIGURE 4

Open in a new tab

Detection sites for primary species of conservation concern, B. affinis (A), B. terricola (B) and B. pensylvanicus (C). Spatial breadth of sampling is shown in green squares, with eDNA detections shown as orange diamonds and net detections represented by black crosses. To aid visualisation, the coastal Maryland site was excluded for cases in which the species of interest was not detected.

FIGURE 5.

FIGURE 5

Open in a new tab

Relationship between eDNA‐based frequency of detection and log‐transformed mean number of Bombus individuals per net survey for flower eDNA samples (A) and leaf surface eDNA samples (B). Each point represents a unique site visit and, for flower samples, point sizes represent the number of eDNA samples collected during the visit. Conditional Nakagawa and Schielzeth (2013) pseudo‐R 2 values are provided, with marginal estimates shown in parentheses.

4. Discussion

Widespread environmental change, coincident with mounting evidence of declining wildlife populations, necessitates increased monitoring to support conservation management. But limited scientific funding and cumbersome, often risky, survey methods present a considerable challenge for the monitoring of endangered species (Guénard et al. 2025; McCarthy et al. 2012). Within pollinator research, reliance on traditional survey techniques often limits sample size and spatial replication (e.g., Boone et al. 2023; Nunes et al. 2024; Otto et al. 2023). Increased sample sizes have been achieved through integration of diverse datasets, such as identifications from unstandardized historical sampling events or iNaturalist data (Ellis et al. 2025; Janousek et al. 2023; Hepner 2024), though this process comes with meaningful drawbacks. Integrating data from numerous sources that change over time inevitably introduces systematic bias regarding how, where, by whom, and for what purpose, samples were collected. For pollinators in particular, Portman (2022) provides a compelling demonstration of how such analyses can be strongly confounded. With these trade‐offs, researchers consistently express a need for improved monitoring techniques (Rousseau et al. 2024).

Environmental DNA methods are increasingly cost‐effective, highly scalable and easily distributed across spatially dispersed teams of scientists. Even minimally trained personnel can implement highly replicated survey designs, minimizing risk to target species of conservation concern, while maximizing the precision with which detection bias can be quantified. Obtaining permits to survey protected species using eDNA surveys is easier relative to more invasive and risky survey approaches that involve handling individuals. In some cases, including for surveys of reproductive queens of federally protected Bombus in spring and fall, eDNA is likely the only permissible survey approach available. eDNA methods are also highly amenable to spatially balanced probability sampling (Brown et al. 2015; Stevens Jr. and Olsen 2004), an approach that limits bias in population monitoring (Boyd et al. 2023), but has not been used in pollinator monitoring to our knowledge. Altogether, these features of eDNA sampling enable low‐cost, broadly dispersed collection of thousands of samples per season, which could be accomplished feasibly through community science initiatives. Implementation of such eDNA survey methods would greatly increase our power to track imperiled species without killing a single bee or overburdening the current capacity for taxonomic identification.

In comparing eDNA methods against traditional surveys, eDNA and net survey data revealed highly similar inferences of the bumble bee community at both regional and local scales. The similarity was true both qualitatively and quantitatively, suggesting that eDNA and traditional methods yield similar inferences regarding Bombus species presence and relative abundance at the site level. Regarding the highly imperilled B. affinis , our findings were consistent with the analysis presented in Hepner et al. (2024), with sightings enriched in heavily forested, high‐elevation landscapes and all but one sighting in close proximity to US National Forest land. To our knowledge, these results are unprecedented for a pollinator‐oriented study. Past pollinator eDNA studies, including our own, have generally struggled to demonstrate detection sensitivity and taxonomic breadth of detection on par with traditional surveys (Avalos et al. 2024; Gamonal Gomez et al. 2023; Kestel et al. 2023; Newton et al. 2023; Stothut et al. 2024).

In addition to establishing a new technique for Bombus surveillance, we provide baseline estimates of occupancy within the Central Appalachian region. We find that, while the federally endangered RPBB is present across much of the sampling region, with detections in Virginia, Maryland, and West Virginia, it is difficult to detect. RPBB detection probabilities were among the lowest of all 13 species detected for both eDNA and netting. In concert with a limited number of total detections, modelling suggested a high degree of uncertainty in regional RPBB occupancy. The closely related yellow‐banded bumble bee (YBBB; B. terricola ), which has suffered declines similarly to RPBB (Cameron et al. 2011; Grixti et al. 2009; Jacobson et al. 2018), exhibited similar estimates for detectability and occupancy in our study. Interestingly, five species exhibited mean occupancy estimates that were less than those of RPBB or YBBB. These included two parasitic species from the subgenus Psithyrus as well as three species considered to be grassland associates: B. auricomus , B. fervidus and B. pensylvanicus (Novotny et al. 2021). The low levels of occupancy observed for these species may reflect their life history traits more so than their current conservation status within the region. For grassland‐associated species, low occupancy estimates were expected because sampling sites were heavily forested, with a median forest cover of 87.8% at a 3‐km radius scale. Psithyrus species are social parasites that differ from other Bombus by producing only queens and males with no production of workers. Further, Psithyrus do not engage in foraging to support colony growth and reproduction, instead foraging only on nectar to meet individual metabolic needs (Bower et al. 2023; Lhomme and Hines 2019). Therefore, we would expect a lower detection probability of these species than for non‐parasitic species where sampling is based on flower surveys. Lacking any prior regional estimates of occupancy or abundance for these groups, it is difficult to assess population trends. Accordingly, our results provide baseline estimates for comparison with future Bombus monitoring outcomes.

The robustness of our eDNA results was supported by rigorous quality control measures, notably the inclusion of leaf surface eDNA samples during field collection and laboratory processing. We observed a significantly lower frequency and richness of Bombus detection within leaf samples compared to flower samples. Of the Bombus detections that occurred in leaf surface samples, some may represent cross‐contamination within the field or lab. However, increasing evidence suggests that eDNA disperses broadly within the environment (Allen et al. 2023; Thomsen and Sigsgaard 2019; Valentin et al. 2021), likely facilitated by airborne dispersal or movement during rainfall events (Bohmann and Lynggaard 2023; Garrett et al. 2023; Johnson, Barnes, et al. 2023). In a companion insect eDNA study of the samples analysed here, frequency and richness of detections were statistically indistinguishable across leaf and flower eDNA samples for Lepidoptera and all Arthropoda (Richardson et al. 2025), suggesting that Bombus detections from leaf samples plausibly represent genuine eDNA signal. Accordingly, a number of works have emphasised the potential of eDNA to document plant‐pollinator associations (Avalos et al. 2024; Johnson, Katz, et al. 2023), but our work suggests that researchers should be cautious about over‐interpreting the eDNA data. Such studies will likely require the inclusion of large numbers of carefully selected background eDNA samples to appropriately adjust for rates of genuine background detection.

In our work, only a single technical replicate was processed for each sample and sample sequencing depths were modest relative to past works (see table 1, Richardson et al. 2025). Current eDNA sensitivity for Bombus appears competitive with traditional approaches, and there is considerable potential to improve this sensitivity going forward. Achieving greater statistical power during eDNA surveys can be obtained in the field stage through additional biological replication, or in the laboratory analysis stage by increasing technical replication or sequencing depth. Approaches to improving the statistical power for traditional survey techniques are more limited and can only be accomplished during field work, where there is little opportunity for cost savings. Accordingly, we think our approach will be highly applicable to both near and long‐term pollinator conservation efforts, serving as a template for the development of terrestrial eDNA survey methods for other rare or cryptic taxa.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Interspecies genetic distances of all pairwise comparisons of available Eastern North American Bombus reference sequence data.

MEN-26-e70073-s002.pdf (4.4KB, pdf)

Figure S2: Relationship of mean detectability per species across the two detection methods, eDNA (y‐axis) and netting (x‐axis).

MEN-26-e70073-s001.pdf (5.3KB, pdf)

Acknowledgements

This work was primarily supported by a DoD ESTCP Grant to R.T.R. and K.G. (Project RC22‐B5‐7373). Additional support was provided by the Maryland Department of Natural Resources Power Plant Research Program and Western EcoSystems Technology Inc.'s Wildlife Research Initiative; the Appalachian Laboratory of the University of Maryland Center for Environmental Science; and Metamorphic Ecological Research and Consulting LLC. For land access, permitting and coordination, we thank USFWS, USFS, George Washington and Jefferson National Forest, VA DCR‐DNH, MD Forest Service, MD Park Service, MD DNR, WV DNR, WV State Parks, PA Game Commission and Easton Utilities for relevant permits and site permissions. This work required a federal recovery permit (ES26953C) for the rusty patched bumble bee, held by K.G. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Richardson, R. T. , Avalos G., Garland C. J., et al. 2026. “Sensitive Environmental DNA Methods for Low‐Risk Surveillance of At‐Risk Bumble Bees.” Molecular Ecology Resources 26, no. 1: e70073. 10.1111/1755-0998.70073.

Funding: This work was supported by the Environmental Security Technology Certification Program, RC22‐B5‐7373.

Handling Editor: Tatiana Giraud

Data Availability Statement

The data that support the findings of this study are openly available in Zenodo at https://doi.org/10.5281/zenodo.16985007. All sequence data and relevant summary tables needed to reproduce these analyses are publicly available at https://doi.org/10.5281/zenodo.16985008.

References

  1. Allen, M. C. , Lockwood J. L., Kwait R., et al. 2023. “Using Surface Environmental DNA to Assess Arthropod Biodiversity Within a Forested Ecosystem.” Environmental DNA 5, no. 6: 1652–1666. 10.1002/edn3.487. [DOI] [Google Scholar]
  2. Allen, M. C. , Nielsen A. L., Peterson D. L., and Lockwood J. L.. 2021. “Terrestrial eDNA Survey Outperforms Conventional Approach for Detecting an Invasive Pest Insect Within an Agricultural Ecosystem.” Environmental DNA 3, no. 6: 1102–1112. 10.1002/edn3.231. [DOI] [Google Scholar]
  3. Arbetman, M. P. , Gleiser G., Morales C. L., Williams P., and Aizen M. A.. 2017. “Global Decline of Bumblebees Is Phylogenetically Structured and Inversely Related to Species Range Size and Pathogen Incidence.” Proceedings of the Royal Society B: Biological Sciences 284, no. 1859: 20170204. 10.1098/rspb.2017.0204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Avalos, G. , Trott R., Ballas J., et al. 2024. “Prospects of Pollinator Community Surveillance Using Terrestrial Environmental DNA Metagenetics.” Environmental DNA 6, no. 1: e492. 10.1002/edn3.492. [DOI] [Google Scholar]
  5. Bohmann, K. , and Lynggaard C.. 2023. “Transforming Terrestrial Biodiversity Surveys Using Airborne eDNA.” Trends in Ecology & Evolution 38, no. 2: 119–121. 10.1016/j.tree.2022.11.006. [DOI] [PubMed] [Google Scholar]
  6. Boone, M. L. , Evans E., Arnold T., and Cariveau D. P.. 2023. “Increasing Sampling Efficiency of Bombus Communities With Rare and Endangered Species by Optimizing Detection Probabilities: A Multi‐Species Occupancy Modelling Approach Using Roadsides as a Case Study.” Biological Conservation 283: 110122. 10.1016/j.biocon.2023.110122. [DOI] [Google Scholar]
  7. Bower, R. , Bulling M., and Norton B.. 2023. “Concern for Cuckoo Bumblebees (Bombus subgenus Psithyrus): Addressing Our Lack of Knowledge.” Journal of Insect Conservation 27, no. 3: 517–522. 10.1007/s10841-023-00462-2. [DOI] [Google Scholar]
  8. Boyd, R. J. , Powney G. D., and Pescott O. L.. 2023. “We Need to Talk About Nonprobability Samples.” Trends in Ecology & Evolution 38, no. 6: 521–531. 10.1016/j.tree.2023.01.001. [DOI] [PubMed] [Google Scholar]
  9. Brooks, M. E. , Kristensen K., van Benthem K. J., et al. 2017. “glmmTMB Balances Speed and Flexibility Among Packages for Zero‐Inflated Generalized Linear Mixed Modeling.” R Journal 9, no. 2: 378–400. 10.32614/RJ-2017-066. [DOI] [Google Scholar]
  10. Brosi, B. J. , and Briggs H. M.. 2013. “Single Pollinator Species Losses Reduce Floral Fidelity and Plant Reproductive Function.” Proceedings of the National Academy of Sciences 110, no. 32: 13044–13048. 10.1073/pnas.1307438110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Brown, J. A. , Robertson B. L., and McDonald T.. 2015. “Spatially Balanced Sampling: Application to Environmental Surveys.” Procedia Environmental Sciences 27: 6–9. 10.1016/j.proenv.2015.07.108. [DOI] [Google Scholar]
  12. Cameron, S. A. , Lozier J. D., Strange J. P., et al. 2011. “Patterns of Widespread Decline in North American Bumble Bees.” Proceedings of the National Academy of Sciences of the United States of America 108, no. 2: 662–667. 10.1073/pnas.1014743108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Christopher, T. R. 2017. Endangered and Threatened Wildlife and Plants: Endangered Species Status for Rusty Patched Bumble Bee. US Fish and Wildlife Service. [Google Scholar]
  14. Colla, S. , Richardson L., and Williams P.. 2011. Bumblebees of the Eastern United States. U.S. Forest Service. [Google Scholar]
  15. Colla, S. R. , and Packer L.. 2008. “Evidence for Decline in Eastern North American Bumblebees (Hymenoptera: Apidae), With Special Focus on Bombus affinis Cresson.” Biodiversity and Conservation 17, no. 6: 1379–1391. 10.1007/s10531-008-9340-5. [DOI] [Google Scholar]
  16. Doser, J. W. , Finley A. O., Kéry M., and Zipkin E. F.. 2022. “spOccupancy: An R Package for Single‐Species, Multi‐Species, and Integrated Spatial Occupancy Models.” Methods in Ecology and Evolution 13, no. 8: 1670–1678. 10.1111/2041-210X.13897. [DOI] [Google Scholar]
  17. Eddy, S. R. 2011. “Accelerated Profile HMM Searches.” PLoS Computational Biology 7, no. 10: e1002195. 10.1371/journal.pcbi.1002195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ellis, K. S. , Otto C. R. V., Bailey L. L., Smith T. A., Choy S., and Hatch L.. 2025. “Integrating Data to Assess Occupancy Patterns of an Endangered Bumble Bee.” Conservation Biology 39, no. 4: e14458. 10.1111/cobi.14458. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Elsen, P. R. , Saxon E. C., Simmons B. A., et al. 2022. “Accelerated Shifts in Terrestrial Life Zones Under Rapid Climate Change.” Global Change Biology 28, no. 3: 918–935. 10.1111/gcb.15962. [DOI] [PubMed] [Google Scholar]
  20. Esling, P. , Lejzerowicz F., and Pawlowski J.. 2015. “Accurate Multiplexing and Filtering for High‐Throughput Amplicon‐Sequencing.” Nucleic Acids Research 43, no. 5: 2513–2524. 10.1093/nar/gkv107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Gamonal Gomez, N. , Sørensen D. H., Chua P. Y. S., and Sigsgaard L.. 2023. “Assessing Flower‐Visiting Arthropod Diversity in Apple Orchards Through Metabarcoding of Environmental DNA From Flowers and Visual Census.” Environmental DNA 5, no. 1: 117–131. 10.1002/edn3.362. [DOI] [Google Scholar]
  22. Garrett, N. R. , Watkins J., Simmons N. B., et al. 2023. “Airborne eDNA Documents a Diverse and Ecologically Complex Tropical Bat and Other Mammal Community.” Environmental DNA 5, no. 2: 350–362. 10.1002/edn3.385. [DOI] [Google Scholar]
  23. Goulson, D. 2003. Bumblebees: Their Behaviour and Ecology. Oxford University Press. https://books.google.com/books?id=doYgikR6ZjcC. [Google Scholar]
  24. Graves, T. A. , Janousek W. M., Gaulke S. M., et al. 2020. “Western Bumble Bee: Declines in the Continental United States and Range‐Wide Information Gaps.” Ecosphere 11, no. 6: e03141. 10.1002/ecs2.3141. [DOI] [Google Scholar]
  25. Grixti, J. C. , Wong L. T., Cameron S. A., and Favret C.. 2009. “Decline of Bumble Bees (Bombus) in the North American Midwest.” Biological Conservation 142, no. 1: 75–84. 10.1016/j.biocon.2008.09.027. [DOI] [Google Scholar]
  26. Guénard, B. , Hughes A. C., Lainé C., Cannicci S., Russell B. D., and Williams G. A.. 2025. “Limited and Biased Global Conservation Funding Means Most Threatened Species Remain Unsupported.” Proceedings of the National Academy of Sciences of the United States of America 122, no. 9: e2412479122. 10.1073/pnas.2412479122. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Heinrich, B. 2004. Bumblebee Economics. Harvard University Press. https://books.google.com/books?id=73yeOpls5qsC. [Google Scholar]
  28. Hepner, M. J. , Orcutt E., Price K., et al. 2024. “Montane Central Appalachian Forests Provide Refuge for the Critically Endangered Rusty Patched Bumble Bee (Bombus affinis).” Forest Ecology and Management 556: 121751. 10.1016/j.foreco.2024.121751. [DOI] [Google Scholar]
  29. Hepner, M. J. 2024. “Conserving Bumble Bee (Bombus) Biodiversity in the Appalachian Region: Techniques for Non‐Lethal Bumble Bee‐Monitoring Surveys in the Appalachian Range of Rusty Patched Bumble Bee (B. affinis).” Northeastern Naturalist 31, no. 1: 35–46. 10.1656/045.031.0103. [DOI] [Google Scholar]
  30. Jacobson, M. M. , Tucker E. M., Mathiasson M. E., and Rehan S. M.. 2018. “Decline of Bumble Bees in Northeastern North America, With Special Focus on Bombus terricola .” Biological Conservation 217: 437–445. 10.1016/j.biocon.2017.11.026. [DOI] [Google Scholar]
  31. Janousek, W. M. , Douglas M. R., Cannings S., et al. 2023. “Recent and Future Declines of a Historically Widespread Pollinator Linked to Climate, Land Cover, and Pesticides.” Proceedings of the National Academy of Sciences of the United States of America 120, no. 5: e2211223120. 10.1073/pnas.2211223120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Jerde, C. L. , Mahon A. R., Chadderton W. L., and Lodge D. M.. 2011. ““Sight‐Unseen” Detection of Rare Aquatic Species Using Environmental DNA.” Conservation Letters 4, no. 2: 150–157. 10.1111/j.1755-263X.2010.00158.x. [DOI] [Google Scholar]
  33. Johnson, M. D. , Barnes M. A., Garrett N. R., and Clare E. L.. 2023. “Answers Blowing in the Wind: Detection of Birds, Mammals, and Amphibians With Airborne Environmental DNA in a Natural Environment Over a Yearlong Survey.” Environmental DNA 5, no. 2: 375–387. 10.1002/edn3.388. [DOI] [Google Scholar]
  34. Johnson, M. D. , Katz A. D., Davis M. A., et al. 2023. “Environmental DNA Metabarcoding From Flowers Reveals Arthropod Pollinators, Plant Pests, Parasites, and Potential Predator–Prey Interactions While Revealing More Arthropod Diversity Than Camera Traps.” Environmental DNA 5, no. 3: 551–569. 10.1002/edn3.411. [DOI] [Google Scholar]
  35. Katoh, K. , Misawa K., Kuma K., and Miyata T.. 2002. “MAFFT: A Novel Method for Rapid Multiple Sequence Alignment Based on Fast Fourier Transform.” Nucleic Acids Research 30, no. 14: 3059–3066. 10.1093/nar/gkf436. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Kendall, L. K. , Rader R., Gagic V., et al. 2019. “Pollinator Size and Its Consequences: Robust Estimates of Body Size in Pollinating Insects.” Ecology and Evolution 9, no. 4: 1702–1714. 10.1002/ece3.4835. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Kestel, J. H. , Bateman P. W., Field D. L., White N. E., Lines R., and Nevill P.. 2023. “eDNA Metabarcoding of Avocado Flowers: ‘Hass’ It Got Potential to Survey Arthropods in Food Production Systems?” Molecular Ecology Resources 23, no. 7: 1540–1555. 10.1111/1755-0998.13814. [DOI] [PubMed] [Google Scholar]
  38. Krehenwinkel, H. , Weber S., Künzel S., and Kennedy S. R.. 2022. “The Bug in a Teacup—Monitoring Arthropod–Plant Associations With Environmental DNA From Dried Plant Material.” Biology Letters 18, no. 6: 20220091. 10.1098/rsbl.2022.0091. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Lhomme, P. , and Hines H. M.. 2019. “Ecology and Evolution of Cuckoo Bumble Bees.” Annals of the Entomological Society of America 112, no. 3: 122–140. 10.1093/aesa/say031. [DOI] [Google Scholar]
  40. Lunau, K. , Unseld K., and Wolter F.. 2009. “Visual Detection of Diminutive Floral Guides in the Bumblebee Bombus Terrestris and in the Honeybee Apis mellifera .” Journal of Comparative Physiology A 195, no. 12: 1121–1130. 10.1007/s00359-009-0484-x. [DOI] [PubMed] [Google Scholar]
  41. McCarthy, D. P. , Donald P. F., Scharlemann J. P. W., et al. 2012. “Financial Costs of Meeting Global Biodiversity Conservation Targets: Current Spending and Unmet Needs.” Science 338, no. 6109: 946–949. 10.1126/science.1229803. [DOI] [PubMed] [Google Scholar]
  42. Milam, J. , Johnson D. E., Andersen J. C., Fassler A. B., Narango D. L., and Elkinton J. S.. 2020. “Validating Morphometrics With DNA Barcoding to Reliably Separate Three Cryptic Species of Bombus Cresson (Hymenoptera: Apidae).” Insects 11, no. 10: 669. 10.3390/insects11100669. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Miller‐Struttmann, N. E. , Heise D., Schul J., Geib J. C., and Galen C.. 2017. “Flight of the Bumble Bee: Buzzes Predict Pollination Services.” PLoS One 12, no. 6: e0179273. 10.1371/journal.pone.0179273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Morales, C. L. , Arbetman M. P., Cameron S. A., and Aizen M. A.. 2013. “Rapid Ecological Replacement of a Native Bumble Bee by Invasive Species.” Frontiers in Ecology and the Environment 11, no. 10: 529–534. 10.1890/120321. [DOI] [Google Scholar]
  45. Mottl, O. , Flantua S. G. A., Bhatta K. P., et al. 2021. “Global Acceleration in Rates of Vegetation Change Over the Past 18,000 Years.” Science 372, no. 6544: 860–864. 10.1126/science.abg1685. [DOI] [PubMed] [Google Scholar]
  46. Nakagawa, S. , and Schielzeth H.. 2013. “A General and Simple Method for Obtaining R2 From Generalized Linear Mixed‐Effects Models.” Methods in Ecology and Evolution 4, no. 2: 133–142. 10.1111/j.2041-210x.2012.00261.x. [DOI] [Google Scholar]
  47. Newton, J. P. , Bateman P. W., Heydenrych M. J., et al. 2023. “Monitoring the Birds and the Bees: Environmental DNA Metabarcoding of Flowers Detects Plant–Animal Interactions.” Environmental DNA 5, no. 3: 488–502. 10.1002/edn3.399. [DOI] [Google Scholar]
  48. Novotny, J. L. , Reeher P., Varvaro M., et al. 2021. “Bumble Bee Species Distributions and Habitat Associations in the Midwestern USA, a Region of Declining Diversity.” Biodiversity and Conservation 30, no. 3: 865–887. 10.1007/s10531-021-02121-x. [DOI] [Google Scholar]
  49. Nunes, L. A. , Tai T., Zuckerberg B., et al. 2024. “Local Floral Abundance Influences Bumble Bee Occupancy More Than Urban‐Agricultural Landscape Context.” Insect Conservation and Diversity 17, no. 2: 215–228. 10.1111/icad.12719. [DOI] [Google Scholar]
  50. Ohio Supercomputer Center . 1987. “Ohio Supercomputer Center [Computer Software].” http://osc.edu/ark:/19495/f5s1ph73.
  51. Otto, C. R. V. , Schrage A. C., Bailey L. L., et al. 2023. “Addressing Detection Uncertainty in Bombus affinis (Hymenoptera: Apidae) Surveys Can Improve Inferences Made From Monitoring.” Environmental Entomology 52, no. 1: 108–118. 10.1093/ee/nvac090. [DOI] [PubMed] [Google Scholar]
  52. Oyen, K. J. , Giri S., and Dillon M. E.. 2016. “Altitudinal Variation in Bumble Bee (Bombus) Critical Thermal Limits.” Journal of Thermal Biology 59: 52–57. 10.1016/j.jtherbio.2016.04.015. [DOI] [PubMed] [Google Scholar]
  53. Portman, Z. 2022. “Exploring Trends in the Source of US Bee Data in GBIF.” Medium. https://medium.com/@zportman/exploring‐trends‐in‐us‐bee‐data‐in‐gbif‐3367d119b26a.
  54. Potapov, P. , Turubanova S., Hansen M. C., et al. 2022. “Global Maps of Cropland Extent and Change Show Accelerated Cropland Expansion in the Twenty‐First Century.” Nature Food 3, no. 1: 19–28. 10.1038/s43016-021-00429-z. [DOI] [PubMed] [Google Scholar]
  55. R Core Team . 2021. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. [Google Scholar]
  56. Richardson, R. , Avalos G., Garland C., et al. 2025. “Scalable Environmental DNA Methods Reveal Associations Between Landscape‐Scale Forest Habitat and Insect Richness.” BioRxiv. 10.1101/2025.04.01.645213. [DOI]
  57. Richardson, R. T. 2022. “Controlling Critical Mistag‐Associated False Discoveries in Metagenetic Data.” Methods in Ecology and Evolution 13, no. 5: 938–944. 10.1111/2041-210X.13838. [DOI] [Google Scholar]
  58. Richardson, R. T. , Curtis H. R., Matcham E. G., et al. 2019. “Quantitative Multi‐Locus Metabarcoding and Waggle Dance Interpretation Reveal Honey Bee Spring Foraging Patterns in Midwest Agroecosystems.” Molecular Ecology 28, no. 3: 686–697. 10.1111/mec.14975. [DOI] [PubMed] [Google Scholar]
  59. Richardson, R. T. , Sponsler D. B., McMinn‐Sauder H., and Johnson R. M.. 2020. “MetaCurator: A Hidden Markov Model‐Based Toolkit for Extracting and Curating Sequences From Taxonomically‐Informative Genetic Markers.” Methods in Ecology and Evolution 11, no. 1: 181–186. 10.1111/2041-210X.13314. [DOI] [Google Scholar]
  60. Rognes, T. , Flouri T., Nichols B., Quince C., and Mahé F.. 2016. “VSEARCH: A Versatile Open Source Tool for Metagenomics.” PeerJ 4: e2584. 10.7717/peerj.2584. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Rousseau, J. S. , Woodard S. H., Jepsen S., et al. 2024. “Advancing Bee Conservation in the US: Gaps and Opportunities in Data Collection and Reporting.” Frontiers in Ecology and Evolution 12: 1346795. [Google Scholar]
  62. Sherrill‐Mix, S. 2019. “taxonomizr: Functions to Work With NCBI Accessions and Taxonomy [Computer Software].” https://CRAN.R‐project.org/package=taxonomizr.
  63. Stevens, D. L., Jr. , and Olsen A. R.. 2004. “Spatially Balanced Sampling of Natural Resources.” Journal of the American Statistical Association 99, no. 465: 262–278. 10.1198/016214504000000250. [DOI] [Google Scholar]
  64. Stothut, M. , Kühne D., Ströbele V., Mahla L., Künzel S., and Krehenwinkel H.. 2024. “Environmental DNA Metabarcoding Reliably Recovers Arthropod Interactions Which Are Frequently Observed by Video Recordings of Flowers.” Environmental DNA 6, no. 3: e550. 10.1002/edn3.550. [DOI] [Google Scholar]
  65. Thomsen, P. F. , and Sigsgaard E. E.. 2019. “Environmental DNA Metabarcoding of Wild Flowers Reveals Diverse Communities of Terrestrial Arthropods.” Ecology and Evolution 9, no. 4: 1665–1679. 10.1002/ece3.4809. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Tichit, P. , Kendall L., Olsson P., et al. 2024. “The Interplay Between Visual Traits and Forest in Bumblebee Communities Across Sweden.” Ecology and Evolution 14, no. 12: e70635. 10.1002/ece3.70635. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Valentin, R. E. , Kyle K. E., Allen M. C., Welbourne D. J., and Lockwood J. L.. 2021. “The State, Transport, and Fate of Aboveground Terrestrial Arthropod eDNA.” Environmental DNA 3, no. 6: 1081–1092. 10.1002/edn3.229. [DOI] [Google Scholar]
  68. Williams, P. H. , Thorp R. W., Richardson L. L., and Colla S. R.. 2014. Bumble Bees of North America. Princeton University Press. http://www.jstor.org/stable/j.ctt6wpzr9. [Google Scholar]
  69. Willmer, P. G. , Bataw A. A. M., and Hughes J. P.. 1994. “The Superiority of Bumblebees to Honeybees as Pollinators: Insect Visits to Raspberry Flowers.” Ecological Entomology 19, no. 3: 271–284. 10.1111/j.1365-2311.1994.tb00419.x. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S1: Interspecies genetic distances of all pairwise comparisons of available Eastern North American Bombus reference sequence data.

MEN-26-e70073-s002.pdf (4.4KB, pdf)

Figure S2: Relationship of mean detectability per species across the two detection methods, eDNA (y‐axis) and netting (x‐axis).

MEN-26-e70073-s001.pdf (5.3KB, pdf)

Data Availability Statement

The data that support the findings of this study are openly available in Zenodo at https://doi.org/10.5281/zenodo.16985007. All sequence data and relevant summary tables needed to reproduce these analyses are publicly available at https://doi.org/10.5281/zenodo.16985008.

Articles from Molecular Ecology Resources are provided here courtesy of Wiley

RESOURCES