Overview: The Big-Bee project will advance research on bee ecology and taxonomy through the creation and standardization of open datasets for taxonomic and computational analysis. It will overcome challenges in bee identification and discovery by linking ecological and anatomical traits. Today, the vast majority of digital insect records include only taxon names, dates, and locations. This paucity of data greatly limits opportunities for research. To achieve fundamental changes in how insect specimens are digitized, the project will supply innovative methods for deep digitization of target taxa, including high-resolution imaging methods. These image data will enable the scoring of life-history traits and will facilitate identification from digitized specimens. Big-Bee will further revolutionize processes of insect specimen digitization by enabling global bee data to be integrated and linked. The project will produce important partnerships between researchers, industry, and government agencies. Specifically, Big-Bee will:

1) Digitize over 0.5 million new specimens via an image-to-data workflow that will capture both labels and specimens of 5,509 agriculturally and phylogenetically important global bee species.

2) Create over 1.3M specimen images from which the project will capture detailed functional trait information at the specimen level. It will create a widely accessible and citable image datasets. The data will include 109K high-resolution, focal-stacked exemplar and diagnostic images, and 10,300 image suites each consisting of 64 images capturing 360 degrees views of specimens, including over 659,000 high-resolution images of bees.  This will enable the scientific-grade 3D reconstruction of specimens from images and will supply rich data for computational discovery and computer vision applications.

 3) Create and steward a new Symbiota portal, Bee Library, that will index all digitized bee occurrence data. Through it, the project will furnish new tools for annotating and sharing ecological and anatomical information. The Bee Library will facilitate identification by producing curated, and openly reviewed datasets about critical bee functional traits, biotic interactions (parasites, floral associates), geographic distributions, and identification features. Project data curators will steward the Bee Library, evaluating and correcting georeference data for all digitized bee species, and providing reports back to collections.

Intellectual Merit: Animal pollination accounts for 35% of global food production. Insects pollinate 80% of wild plants. Most of these insect pollinators are bees. Although limited in scope, extant data show bee abundance is declining due to anthropogenic disturbances. A growing body of research demonstrates that pollinator responses to anthropogenic disturbances, such as land use and climate change, are mediated by functional or life-history traits that impact the fitness of individual species within the community. These effects impact plant-pollinator interactions in both natural and agricultural systems. While the US currently possesses a wealth of digitized historic bee specimen data, most records do not include associated data reflecting functional traits. This greatly limits the mechanistic insights they can provide about factors driving long-term community response to anthropogenic disturbances. In order to accurately track how and why pollinators are declining, better tools are urgently needed to support bee identification, functional trait annotation, and the exchange of historical species and trait information among researchers, industry, and policymakers.

Broader Impacts: Big-Bee will collaborate on the creation of the US National Native Bee Monitoring Plan as part of the Bee Monitoring Research Coordination Network currently funded through the USDA. It will create an infrastructure to support hundreds of bee researchers in specimen digitization. The project published datasets contribute to the Open Traits Network, Global Biotic Interactions, and other downstream applications. It will develop computer science and collections-based undergraduate curricula, increasing capacity in entomology for bee identification and research. Participating institutions will develop undergraduate course materials that involve examination of bee specimens and the use of data derived from specimens, and will contribute lesson plans to BLUE (Biodiversity Literacy for Undergraduate Education). Extensive workforce training in biodiversity informatics, museum preparation, bee identification, and the creation of citizen science monitoring programs will be offered to collections within and outside the Big-Bee network. Big-Bee museums will contribute to a new syndicated college radio science program, The Buzz, that will be produced by undergraduate students in science communication.

Bees are important pollinators responsible for creating and maintaining the ecosystems that many animals and humans rely on. With many of its species on the decline, the Cheadle Center for Biodiversity and Ecological Restoration is leading the Big-Bee Project Network (http://big-bee.net) in an effort to digitize bee traits for further research in bee ecology. Since researchers lack data on anatomical traits that may make bees either vulnerable or resilient to human-induced environmental changes, it is important to find new methods for efficiently collecting these data.

One bee trait identified to prioritize was a bee’s hairiness because they’re important for pollination, sensing nearby danger, and staying warm in cooler temperatures and cooler in hotter climates. Through the use of deep learning and computer vision, a new pipeline was created to measure the surface density of hair from bee images. By optimizing two pre-trained U-Nets on lateral-view images of bees, the model became capable of pixel-based classification of where the bee is in a photo and where the hair is. These models were trained and evaluated on hand-crafted datasets of 550 images, which were remasked overtime for better performance.

While the main challenges came from differentiating texture of hair and texture of eyes/wings, another deep learning model succeeded at removing this noise before being fed into the hair detection pipeline. This multi-step pipeline improves previous findings by further preprocessing input images and is able to target the important signals of the image (hair). It also opens up the opportunity of using traditional edge detection methods for optimizing this problem.

This research is sponsored by the McNair/Edison STEM Summer Research Program and the National Science Foundation (#2102006).

We are in the second year of a project to revitalize entomology at the University of California, Santa Barbara and have new updates of its impact on the campus to report. The UC Santa Barbara Natural History Collection (UCSB) at the Cheadle Center for Biodiversity and Ecological Restoration formed the UCSB Invertebrate Zoology collection from about 10K historical specimens that were found in a basement on the UCSB campus. The collection has continued to grow rapidly through Coastal California arthropod survey efforts, donated student collections, and faculty research projects. New results from the project include the formation of an outreach program through the UCSB Extension Department and discovery of several extirpated rare or endangered insects that once occurred on the UCSB campus. We will report on how the discovery of extirpated species on campus has received some press and the development of a collection on campus has raised the profile of entomology and insect conservation in the area.

This poster was presented at the 2018 Entomological Collections Network meeting held in Vancouver, Canada on November 10-11. 

The distribution of bees can be estimated, in part, by enviromental variables associated with geography. Using bee occurrence records, we quantified the estimated species richness of the EPA level 3 defined ecoregions using coverage-based rarefaction. Our study reveals that bee species richness is highest in ecoregions with warm dry summers. Furthermore, increased species richness was found to be correlated with maximum mean temperature.

Presented at the iDigBio 6th Annual Digital Data in Biodiversity Research Conference on 05/24/2022 and the University of California Santa Barbara EEMB Undergraduate Research Symposium on 04/23/2022

Background/Question/Methods

    A single species geographically separated into island and mainland populations often exhibits differences in phenotype between those populations. Size differences, such as dwarfism and gigantism, are well-studied particularly in mammals, but island-mainland population phenotype differences are not well-studied in insects.

Bees (Hymenoptera: Anthophila) have unique wing venation and imaged bee wings can be used to accurately identify bees to species. Wing venation patterns alone may be sufficient to classify variation between populations of the same species using geometric morphometrics. In this study, we applied these methods to determine if populations of island and mainland bees are significantly different. Populations are from Santa Cruz Island and Santa Barbara, California, a coastal mainland town. Santa Cruz Island is a 35-kilometer long island about 32 kilometers off the coast of Santa Barbara in the Pacific Ocean. Thus, there is a physical barrier between mainland and island populations. One species of sweat bee, Halictus tripartitus (Hymenoptera: Halictidae), was chosen. This species is commonly collected on both the island and the mainland, and specimens were obtained from both natural history collections and new collections. 

 To analyze wing venation patterns, both forewing and hindwing were removed, slide mounted, imaged, and annotated with digital landmarks using TPS morphometric software for approximately 500 specimens and 9 landmarks. In R, landmark coordinates were aligned with a generalized procrustes analysis (GPA) and simplified with a principal component analysis (PCA). The output of the PCA was plotted on biplots, and the output of the GPA was used in an analysis of similarities (ANOSIM) test.

Results/Conclusions

    Current ANOSIM results show that there is a statistically significant difference between island and mainland Halictus tripartitus populations (p-value < 0.05). Ongoing work includes adding additional specimens and running new tests, including non-metric multidimensional scaling (NMDS), and “envfit” regression (in R library vegan) to identify the strength and direction each of the 9 landmarks has on the group. This project shows that wing morphometrics can distinguish between populations of bees and may be a viable method of automated bee identification. Additionally, these found differences in bee wing morphology can lead to further work in island-mainland population variation, including evolutionary questions of gene flow between island and mainland populations.

Bees (Hymenoptera: Anthophila) have unique wing venation, which can be used to classify variation between populations of the same species using geometric morphometrics. Here, an application of this method is presented, showing successful distinction between populations of the sweat bee Halictus tripartitus (Hymenoptera: Halictidae) from either island or mainland collections. These two locations are Santa Cruz Island, a 35-kilometer long island about 32 kilometers off the coast of Santa Barbara in the Pacific Ocean, and Santa Barbara, California, a coastal mainland town. 

To analyze wing patterns, forewings were removed, imaged, and annotated with digital landmarks using TPS morphometric software for 9 landmarks. Significant difference between populations is shown with an ANOSIM test (p<0.05), and principal component analysis results are visualized on a biplot. Ongoing work includes adding additional specimens and running new tests, including non-metric multidimensional scaling (NMDS), and “envfit” regression (in R library vegan) to identify the strength and direction each of the 9 landmarks has on the group.

This poster was presented at the California Botanical Society 29th Graduate Student Symposium at Cal Poly Humboldt in Arcata, CA on April 1, 2023.

Summary: Observation, collection, curation, and archiving are central to ecological understanding and artistic process. Biodiversity data portals are repositories that can be used for a diversity of scientific and artistic applications. Digitization of historical collections increases accessibility and expands the range of potential hypothesis-driven research.