Validating Morphometrics with DNA Barcoding to Reliably Separate Three Cryptic Species of Bombus Cresson (Hymenoptera: Apidae)

Deep learning for identifying bee species from images of wings and pinned specimens

One of the most challenging aspects of bee ecology and conservation is species-level identification, which is costly, time consuming, and requires taxonomic expertise. Recent advances in the application of deep learning and computer vision have shown promise for identifying large bumble bee (Bombus) species. However, most bees, such as sweat bees in the genus Lasioglossum, are much smaller and can be difficult, even for trained taxonomists, to identify. For this reason, the great majority of bees are poorly represented in the crowdsourced image datasets often used to train computer vision models. But even larger bees, such as bumble bees from the B. vagans complex, can be difficult to separate morphologically. Using images of specimens from our research collections, we assessed how deep learning classification models perform on these more challenging taxa, qualitatively comparing models trained on images of whole pinned specimens or on images of bee forewings. The pinned specimen and wing image datasets represent 20 and 18 species from 6 and 4 genera, respectively, and were used to train the EfficientNetV2L convolutional neural network. Mean test precision was 94.9% and 98.1% for pinned and wing images respectively. Results show that computer vision holds great promise for classifying smaller, more difficult to identify bees that are poorly represented in crowdsourced datasets. Images from research and museum collections will be valuable for expanding classification models to include additional species, which will be essential for large scale conservation monitoring efforts.

A checklist of the bees (Hymenoptera: Apoidea) of Minnesota

Research studies and conservation actions aimed at improving conditions for bees require a basic understanding of which species are present in a given region. The US state of Minnesota occupies a unique geographic position at the confluence of eastern deciduous forests, northern boreal forests, and western tallgrass prairie, which has led to a diverse and unique bee fauna. In recent years there have been multiple ongoing bee-focused inventory and research projects in Minnesota. Combined with the historic specimens housed in the University of Minnesota Insect Collection and other regional collections, these furnished a wealth of specimens available to form the basis of a statewide checklist. Here, we present the first comprehensive checklist of Minnesota bee species, documenting a total of 508 species in 45 genera. County-level occurrence data is included for each species, and further information on distribution and rarity is included for species of regional or national interest. Some species have their taxonomy clarified, with Perdita citrinella Graenicher, 1910 syn. nov. recognized as a junior synonym of Perdita perpallida Cockerell, 1901, P. bequaerti syn. nov. recognized as a junior synonym of P. pallidipennis Graenicher, 1910 stat. nov., Anthidiellum boreale (Robertson, 1902) stat. nov. recognized as a full species, and Anthidiellium beijingense Portman & Ascher nom. nov. is proposed for A. boreale Wu to resolve the homonymy with A. boreale (Robertson). We further include a list of species that may occur in Minnesota and highlight 11 species occurring in the state that are considered non-native. Recent collecting efforts, as well as increased taxonomic attention paid to Minnesota bees, have resulted in 66 species that have only been documented in the last 10 years. As a first step in determining native bees of conservation concern, we document 38 species that have not been detected in the state during the last 50 years and discuss their conservation status, along with other species for which evidence of decline exists. The checklist of Minnesota bees will continue to grow and change with additional surveys and research studies. In particular, recent surveys have continued to detect new bee species, and many bee groups are in need of taxonomic revision, with the most recent revisions for many genera occurring decades ago. Overall, this checklist strengthens our understanding of the bees of Minnesota and the broader region, informs conservation assessments, and establishes a baseline for faunal change.

Local habitat type influences bumble bee pathogen loads and bee species distributions

Bumble bees (Hymenoptera: Apidae, Bombus Latreille) perform important ecological services in both managed and natural ecosystems. Anthropogenically induced change has altered floral resources, climate, and insecticide exposure, factors that impact health and disease levels in these bees. Habitat management presents a solution for improving bee health and biodiversity, but this requires better understanding of how different pathogens and bee species respond to habitat conditions. We take advantage of the washboard of repeated ridges (forested) and valleys (mostly developed) in central Pennsylvania to examine whether local variation in habitat type and other landscape factors influence bumble bee community composition and levels of 4 leading pathogens in the common eastern bumble bee, Bombus impatiens Cresson. Loads of viruses (DWV and BQCV) were found to be lowest in forest habitats, whereas loads of a gut parasite, Crithidia bombi, were highest in forests. Ridgetop forests hosted the most diverse bumble bee communities, including several habitat specialists. B. impatiens was most abundant in valleys, and showed higher incidence in areas of greater disturbance, including more developed, unforested, and lower floral resource sites, a pattern which mirrors its success in the face of anthropogenic change. Additionally, DNA barcoding revealed that B. sandersoni is much more common than is apparent from databases. Our results provide evidence that habitat type can play a large role in pathogen load dynamics, but in ways that differ by pathogen type, and point to a need for consideration of habitat at both macro-ecological and local spatial scales.

Integrative Taxonomy Approach Reveals Cryptic Diversity within the Phoretic Pseudoscorpion Genus Lamprochernes (Pseudoscorpiones: Chernetidae)

Pseudoscorpions represent an ancient, but homogeneous group of arachnids. The genus Lamprochernes comprises several morphologically similar species with wide and overlapping distributions. We implemented an integrative approach combining molecular barcoding (cox1), with cytogenetic and morphological analyses in order to assess species boundaries in European Lamprochernes populations. The results suggest ancient origins of Lamprochernes species accompanied by morphological stasis within the genus. Our integrative approach delimited three nominal Lamprochernes species and one cryptic lineage Lamprochernes abditus sp. nov. Despite its Oligocene origin, L. abditus sp. nov. can be distinguished from its closest relative only by molecular and cytogenetic differences, or alternatively, by a complex multivariate morphometric analysis involving other Lamprochernes species. The population structure and common haplotype sharing across geographically distant populations in most Lamprochernes species suggest that a phoretic manner of dispersal is efficient in this group.

The diversity, evolution, and development of setal morphologies in bumble bees (Hymenoptera: Apidae: Bombus spp.)

Bumble bees are characterized by their thick setal pile that imparts aposematic color patterns often used for species-level identification. Like all bees, the single-celled setae of bumble bees are branched, an innovation thought important for pollen collection. To date no studies have quantified the types of setal morphologies and their distribution on these bees, information that can facilitate understanding of their adaptive ecological function. This study defines several major setal morphotypes in the common eastern bumble bee Bombus impatiens Cresson, revealing these setal types differ by location across the body. The positions of these types of setae are similar across individuals, castes, and sexes within species. We analyzed the distribution of the two most common setal types (plumose and spinulate) across the body dorsum of half of the described bumble bee species. This revealed consistently high density of plumose (long-branched) setae across bumble bees on the head and mesosoma, but considerable variation in the amount of metasomal plumosity. Variation on the metasoma shows strong phylogenetic signal at subgeneric and smaller group levels, making it a useful trait for species delimitation research, and plumosity has increased from early Bombus ancestors. The distribution of these setal types suggests these setae may serve several functions, including pollen-collecting and thermoregulatory roles, and probable mechanosensory functions. This study further examines how and when setae of the pile develop, evidence for mechanosensory function, and the timing of pigmentation as a foundation for future genetic and developmental research in these bees.

Parasite Prevalence May Drive the Biotic Impoverishment of New England (USA) Bumble Bee Communities

Simple Summary

Here we discuss widespread changes in the community structure of bumble bees (Bombus spp.) found in the coastal-zone community of New England. One species in particular, Bombus impatiens Cresson, 1863, has increased in relative abundance nearly 45% since the 1990s to become the dominant species in the region, representing nearly 75% of all Bombus individuals collected in our studies. These changes in abundance may be, in part, due to differences in infection rates by microparasites, with B. impatiens having significantly fewer microparasites than several other less common and declining Bombus species. We discuss the possible role of microparasites in influencing the community composition of Bombus species in our region, and how these infections might be compounding declines in conjunction with habitat loss and climate change.

Abstract

Numerous studies have reported a diversity of stressors that may explain continental-scale declines in populations of native pollinators, particularly those in the genus Bombus. However, there has been little focus on the identification of the local-scale dynamics that may structure currently impoverished Bombus communities. For example, the historically diverse coastal-zone communities of New England (USA) now comprise only a few species and are primarily dominated by a single species, B. impatiens. To better understand the local-scale factors that might be influencing this change in community structure, we examined differences in the presence of parasites in different species of Bombus collected in coastal-zone communities. Our results indicate that Bombus species that are in decline in this region were more likely to harbor parasites than are B. impatiens populations, which were more likely to be parasite-free and to harbor fewer intense infections or co-infections. The contrasting parasite burden between co-occurring winners and losers in this community may impact the endgame of asymmetric contests among species competing for dwindling resources. We suggest that under changing climate and landscape conditions, increasing domination of communities by healthy, synanthropic Bombus species (such as B. impatiens) may be another factor hastening the further erosion of bumble bee diversity.

Assessing the potential for deep learning and computer vision to identify bumble bee species from images

Pollinators are undergoing a global decline. Although vital to pollinator conservation and ecological research, species-level identification is expensive, time consuming, and requires specialized taxonomic training. However, deep learning and computer vision are providing ways to open this methodological bottleneck through automated identification from images. Focusing on bumble bees, we compare four convolutional neural network classification models to evaluate prediction speed, accuracy, and the potential of this technology for automated bee identification. We gathered over 89,000 images of bumble bees, representing 36 species in North America, to train the ResNet, Wide ResNet, InceptionV3, and MnasNet models. Among these models, InceptionV3 presented a good balance of accuracy (91.6%) and average speed (3.34 ms). Species-level error rates were generally smaller for species represented by more training images. However, error rates also depended on the level of morphological variability among individuals within a species and similarity to other species. Continued development of this technology for automatic species identification and monitoring has the potential to be transformative for the fields of ecology and conservation. To this end, we present BeeMachine, a web application that allows anyone to use our classification model to identify bumble bees in their own images.