Bumble bee conservation across an urban to rural gradient

BUMBLE BEE CONSERVATION ACROSS AN URBAN TO RURAL GRADIENT

Sponsoring Institution

National Institute of Food and Agriculture

Project Status

COMPLETE

Funding Source

AFRI COMPETITIVE GRANT

Reporting Frequency

Annual

Accession No.

1018812

Grant No.

2019-67013-29298

Cumulative Award Amt.

$458,004.00

Proposal No.

2018-08617

Multistate No.

(N/A)

Project Start Date

Apr 1, 2019

Project End Date

Mar 31, 2023

Grant Year

2019

Program Code

[A1113]- Pollinator Health: Research and Application

Recipient Organization
UNIV OF WISCONSIN
21 N PARK ST STE 6401
MADISON,WI 53715-1218

Performing Department
ENTOMOLOGY

Non Technical Summary
Bumble bees (genus Bombus) are the second most economically productive pollinators in global agriculture, and are more efficient pollinators than honey bees on crops such as blueberries, cranberries, and tomatoes. However, there is evidence that their populations globally are in decline. In the US, at least seven of our 40 native species have experienced population declines or range contractions in the last 4 decades. Somewhat surprisingly, emerging research suggests that urban areas may be places where bees and other pollinators may find refuge and thus offer opportunities for conservation. In urban environments pockets of high plant diversity and longer growing seasons may give bees opportunities to survive and persist that are not present in more rural and agricultural areas of the landscape where bees have been shown to be in decline. Our project will examine the relationship between features of urban environments (degree of urban development, plant diversity, length of growing season, etc.) on bumble bee population size, abundance, and pathogen load.We will address this question by sampling bees across southern Wisconsin in locations with a wide variety of urban and rural characteristics (e.g., different flowering plants, human population densities, etc.). We will combine our bumble bee surveys with large citizen-science databases of bumble bee sightings and apply species distribution modeling to this data to determine how features of urbanization affect bumble bee abundance and species diversity. We will also use population genetics approaches to measure bumble bee colony survival, colony density across urban and rural areas. Finally, we will collect bees and measure their pathogen loads to understand if urban bees are healthier. Ultimately, this detailed study on bumble bees will help identify the aspects of cities and the countryside that may promote or limit bumble bee distribution, health, and survival across the landscape. This will give us better tools to potentially mitigate stresses on bees in both rural and urban environments, and offers opportunities for conservation of this important native group of key pollinators.

Animal Health Component

20%

Research Effort Categories

Basic

80%

Applied

20%

Developmental

(N/A)

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
136	3085	1070	50%
135	3085	1070	50%

Knowledge Area
135 - Aquatic and Terrestrial Wildlife; 136 - Conservation of Biological Diversity;

Subject Of Investigation
3085 - Other bees (non-Apis bees) - All bees except Apis cerana; includes Bombas, Alfalfa leaf-cutter;

Field Of Science
1070 - Ecology;

Keywords

urban ecology

bumble bees

pollinator conservation

species distribution modeling

Goals / Objectives
The primary focus of this study is a comparison of bumble bee populations between urban and surrounding rural areas in the Upper Midwest with a particular focus on southern Wisconsin.Objectives:1. At regional scales, evaluate the relative importance of urban and rural areas in constraining the distribution of a diversity of bumble bee species in the Midwest.2. At local scales, quantify the variability in bumble bee population size, abundance, lineage survivorship, and pathogens across an urban-rural gradient, and identify the important predictors (flower diversity, abundance, and phenology; microclimates; green spaces) of this variability.Wehypothesizethat, after accounting for observation biases and differences in detectability among bumble bee species, there will be a greater likelihood of bumble bee occurrence, higher relative abundance, larger colony number, higher lineage survival, and lower pathogen loads in urban compared to rural and agricultural areas.In particular, we postulate that features of urban landscapes such as (1) a higher diversity and abundance of floral resources, (2) greater accessibility and proximity to green spaces (e.g., parks, gardens, or vacant lots), (3) warmer and more stable microclimates, and (4) longer flowering phenologies (including earlier flowering in the spring) enhance bumble bee performance in urban areas, relative to rural areas.

Project Methods
For testing the hypotheses associated with Objective 1, we will investigate the geographic patterns of bumble bee distributions across the Upper Midwest (MN, IA, WI, MI, IL, IN, OH) and identify important environmental associations of their occurrence at regional scales. Our primary source of bee observations will be a publicly available database that is a compilation of records of Bombus species from citizen science projects, private and educational collections, and museum records. For the regional assessment, we will, for simplicity and to constrain the modeling effort initially focus on six species of bumble bees that are either stable or declining across the Eastern and Midwestern United States. We will create distribution models using maximum entropy modeling and the widely available software program Maxent. We will parameterize models using information on land cover and climate. The final model output will consist of spatial predictions of suitability representing the likelihood bee species will be present in an area at 5 x 5 km resolution, based on environmental inputs. To supplement our understanding of bumble bee patterns from our regional assessment, we will conduct a two-year sampling study of the occupancy (i.e., occurrence) and relative abundance of Bombus species along an urban-to-rural gradient across a geographic area that encompasses Madison, WI and the surrounding rural landscape. The rural landscape surrounding this city represents primarily agricultural uses including conventional, small to large-scale row crops (corn and soybean), alfalfa and hay, and pasture for beef and dairy cows. The intensive sampling study will encompass a 30 x 30 km area from the center of Madison. Within a 1 x 1 km sampling block, we will calculate the proportion of different land cover types. Blocks (n=18 total) will be assigned to three broad categories: urban (> 50% high-intensity development), peri-urban/suburban (> 50% low-intensity development), and rural (> 50% agriculture). Our approach will capture a full range of habitat availability ranging from urban to rural landscapes thus generating the urban-rural gradient of conditions expected to affect bumble bees. Over a 2-year period, we will collect information on the occurrence (detection/non-detection of foragers) and relative abundance (number of foragers present) of bumble bees by hand collection methods. Within each block, we will run transects allocated to areas within the block in proportion to the representation of the dominant land cover types. Bumble bee species will be netted or vacuumed, identified to species on location, sex/caste will be recorded, and released. A subset of captured individuals will be processed for population genetic and pathogen analysis (see below). The plant species on which bees were captured will be recorded. Between early April and late August, we will conduct 9 surveys at each transect with each visit spaced by 2 weeks, and a minimum of 3 visits in spring, early summer, and late summer. We will analyze both the impervious surface, land cover composition, and housing density to assess the environmental context of the sampling transects at multiple scales. In addition to land cover composition, we will characterize fine-scale variation in microclimate and floral resources. At each station, we will collect topographic and vegetation structure characteristics. We will analyze the count data using a dynamic Bayesian N-mixture model to make inferences about spatiotemporal changes in the distribution, occupancy and abundance of bumble bees, while accounting for imperfect detection. To test the hypotheses associated with Objective 2, we will use molecular genetic tools to identify related individuals and estimate the number of colonies that visit an area. We will non-lethally sample genetic material from bumble bees by catching foraging bees in a net, chilling the bees on ice, and then clipping a tarsus from one midleg of each bee. We will identify bees to species and record the sex/caste for each for use in intra- and inter-annual comparisons. We will focus on commonly occurring bees such as Bombus impatiens and B. vagans.Within each sampling year, we will generate multi-locus microsatellite genotypes of individuals, assign individuals to family using COLONY software, and estimate colony diversity and relative colony number among 50-forager samples. Specimens will be genotyped using 15 variable DNA microsatellite markers identified from the literature. Colony diversity and relative colony abundance in a block will be analyzed using a generalized linear mixed effects model (using Poisson count link functions for count data) with colony density as the dependent variable and landscape features (similar variables as used in occupancy modeling, e.g., percent impervious, percent agricultural land, plant species diversity, etc.), time (week of the season), as the independent variables, and hierarchical nested sampling terms (e.g., transect, block, and city) as random effects. This will allow us to test for colony diversity differences across the landscape gradient, to understand the effects of landscape on the population size of each species. To investigate the relationship between survival of bumble bee lineages and landscape traits, we will collect DNA from two common bumble bee species across the urban-rural gradient at three time points. Bees will be sampled in May of Year 1, July of Year 1, and July of Year 2. Population genetic analyses will then assess survivorship of colonies from the beginning of the season (May) to the peak abundance of bees in the summer in year 1 (July), as well as between the peaks of the summer in Years 1 and 2 (July samples). All individuals of B. impatiens and B. vagans encountered during these surveys will be netted, and a tarsal clipping will be removed and stored in 95% ethanol before the bee is released. To estimate a family lineage's probability of survivorship across urban to rural gradients, we will use the same genotyping method as above will be used to assign individuals sampled in July of Year 1 to families identified in May of Year 1 (intraannual survival), as well as to assign individuals sampled in July of year 2 to families identified in July of year 1 (interannual survival). Subsequently, the probability of intraannual and interannual lineage survival within a genetic sampling block will be analyzed as a function of landscape traits across the urban-rural gradient in order to identify key drivers of lineage persistence across generations. This will allow us to evaluate whether lineage survival is greater in urban than in rural areas, and which landscape traits drive this trend.Pathogen and parasite diagnosis will be conducted through microscopy and molecular detection. We will sample 50 individual bees of the two most common species (from prior sampling this will likely be B. impatiens and B. vagans) from four blocks in each habitat type (urban, suburban, or rural) for two consecutive years. These bees will be encountered through the transect sampling described above. The bees will be frozen and stored at -20C until dissections. The abdomen and thorax will be cut open and microscopically examined for dipteran parasites (Phoridae and Conopidae). These macroparasites will be assigned to the host matrix code, and each type will be given an additional unique identification number. Macroparasites will be frozen, identified to species, counted and added to the host data. DNA barcode analysis using our own database of reference organisms will be applied to immature dipteran specimens for which morphological species identification is impossible. Identification of these pathogens to species will be confirmed using species specific primers . An internal Bombus-positive control will also be included in molecular surveys to assess extraction success.

Progress 04/01/19 to 03/31/23

Outputs
Target Audience:The primary focus of this research is to inform bumble bee conservation and management in urban and agricultural environments, the design and analysis of citizen science programs, and academic studies of beeecology and theory in the context of anthropogenic change. As such, our primary audiences are academic colleagues, conservation practitioners (non-governmental and state and federal agencies), and local municipal planners interested in bee conservation broadly. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?Over course of this project, we have trained and completed three MS students (Tai, Moore, Whiteman), two post-docs (Nunes, not at Defenders of Wildlife as Sr. Conservation Specialist,Perry, now faculty at Ohio State University, and Clare, now at UC Berkeley). In addition, we hired four student hourly helpers during the fieldportion of the work who gained experience in ecological methods and bee biology. How have the results been disseminated to communities of interest?Over the course of the project, we have given 16conference or poster presentations, and three outreach talks where findings of this project were shared with a range of academic and public audiences. We have also met individually with partners at the Wisconsin Department of Natural Resources, who shared their Bumble bee brigade data to report on our findings and provide suggestions of how to improve their citizen-science program. What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? Executive Summary of Accomplishments- The overall consensus findings from this study from a diverse line of evidence found no consistent effects of urbanization on bumble bees relative to bumble bees in nearby agricultural areas. Occurrence of bumble bees across urban to agricultural landscapes as measured both through a systematic sampling study around Madison, WI, and by a novel analysis of citizen science bumble bee records for the state of Wisconsin, found no widespread difference in bees across the urban-agricultural gradient. However, at the fine scale, we found that the amount of flowering resources available as forage for bees at individual sites was a significant predictor on increased bumble bee occurrence. Nevertheless, bumble bee communities found in the anthropogenically modified environments examined in this study (urban and agricultural areas) were different-both in terms of traits and relative species composition-than species present in the regional pool, suggesting some level of biotic homogenization of communities. Analyses of the genetic structure of bumble bees across the urban-agricultural gradient found no evidence of population subdivision of isolation, suggesting widespread gene flow across the area, and pathogen loads were also not influenced by landscape characteristics. Rather, pathogen loads varied across years and across species, indicating other drivers of pathogen dynamics not associated with urbanization. In summary, the finding of no widespread and systematic effects of changes across urban to agricultural landscapes suggests that local characteristics of sites, primarily related to flower abundance, is a primary driver of species occurrences. Nevertheless, habitats that have been significantly influenced by humans, such as urban and agriculture areas, represent significant challenges for bumble bees at the community level. Citizen science and regional modeling of wild bee distributions- Our citizen science analysis focused on the integrated modeling of two programs: Bumble Bee Watch and Bumble Bee Brigade. Integrated species distribution model results suggest that climate (e.g., mean maximum July temperature) most strongly explains differences in bee communities. Urbanization also strongly drives volunteer sampling effort but appears to variably affect the sampled bumble bee species, with more southerly distributed species in the region exhibiting more positive associations with urbanization. The effects of agriculture on bumble bee occurrence was highly uncertain because agricultural areas are under sampled by citizen scientists. In practical terms these findings suggest that while volunteers exhibit a limited degree of preferential sampling (i.e., sampling where certain bee species are more likely to occur or be detected after controlling for environmental covariates), effort of sampling was environmentally unbalanced. Improving the balance of effort across habitats could enhance the precision of future monitoring inference (e.g., targeted sampling in agricultural landscapes). Our results also highlight the value of having multiple citizen science programs and protocols in place. The small area searches (or complete checklists) associated with Bumble Bee Brigade have particular value because absence is recorded, and the underlying sampling effort is transparent. Increased emphasis on "complete checklist" recording (recording all species detected) will be critical for inferring absence and ensuring reliable predictions. Intensive sampling of bumble bees across an urban-agricultural gradient- We designed a bumble bee sampling study that allowed us to examine in detail how variation in the landscape around collection sites that spanned a large urban to agricultural gradient, and variation in local habitat variation influence bumble bee occupancy. After accounting for species-specific detection probabilities, the effect of urbanization on bee occupancy was weak and no species were less likely to occupy urban than rural agricultural areas. On the other hand, our findings suggest that bumble bee occupancy is associated with a ''honeypot effect" where local resource availability, in the form of higher floral abundance, is most important in limiting the occupancy of bumble bees, and not, as previously hypothesized, the degree of urban or agricultural cover in the landscape. We also explored the potential for biotic homogenization occurring in communities of bumble bees observed across the environmental gradient. We used multiple dimensions of bumble bee diversity (species, phylogenetic, functional) to understand the species characteristics of bees that determine whether they are "winners" or "losers" in human-modified landscapes. We found that greenspaces sampled for bees that occurred in urban and agricultural landscapes supported a similar community of bumble bees. However, these communities were more similar to one another than expected by chance based on all diversity metrics. That is, compared to the regional pool of species, species successful in urban and agricultural landscapes were smaller, had shorter wings and larger eyes. Greenspaces surrounded by more forest habitat supported larger species. Human-modified landscapes, in this case both urban and agricultural spaces, acted as strong filters for bumble bees, selecting a subset of functionally and phylogenetically similar species that resulted in homogenization of communities relative to the pool of species occurring in the region. Population genetics to estimate colony abundance- Using population and landscape genetic techniques, we genotyped approximately 2,300 bumble bees at 15 microsatellite loci from Madison, WI to assess gene flow levels, heterozygosity, allelic diversity and colony numbers across 20 collection sites. Population structure did not vary across the landscape for any of the study species although we did observe an increase in abundance of the B. bimaculatus with increasing urbanization . However, the increase in abundance of bees did not correspond to an increase in colony numbers. These data suggest that at the scale of the study, greenspace is sufficient to maintain gene flow across both urban and agricultural land use variation. Pathogen Load variability across urban to agricultural habitats- Bumble bees collected across the urban-agricultural gradient (~2,000 individuals over the two years of the study) were dissected and screened for parasites. Electrophoresis-gel screening of PCR amplified sequences found 678 occurrences of Apicystis bombi, 13 of Crithidia spp., and 36 of Vairrimorpha spp. Apicystis bombi was the most common pathogen in the community (32.1% of individuals) and was overwhelmingly found in B. impatiens. Additionally, A. bombi was more prevalent in agricultural communities than urban sites one year, but the opposite was true in the second year, suggesting that interannual variation in pathogen prevalence, and not landscape, is driving pathogen dynamics. We discovered that conopid fly parasitism was higher at urban sites than in rural sites and that the two-spotted bumble bee was more frequently parasitized than other species. The higher rate of parasitism at urban sites may be related to the higher abundance of B. bimaculatus at those sites, again suggesting that host-parasite specific dynamics may actually be more important than landscape traits. Finally, we discovered eight new incidents of mermithid parasitism of bumble bees. These individuals are being sequenced to determine species. This rare phenomenon has only previously been documented 16 times in the published literature and our data represent the highest number of records in any single site study. Despite the rarity of parasitism, this further supports bumble bees as hosts for these nematodes.

Publications

Type: Conference Papers and Presentations Status: Accepted Year Published: 2022 Citation: Whiteman, L.B., K.I. Perry, and J.P. Strange. Poster. Bumble bee pathogens vary across an urban to rural gradient. Entomological Society of America Joint Annual Meeting, Vancouver, BC. November 13, 2022
Type: Conference Papers and Presentations Status: Accepted Year Published: 2022 Citation: Moore, I.S., Strange, J.P., Christman, M., Barkan, P.N., Gratton, C. and Tai, T. Effects of Urban Land Cover on Community Composition in Bombus spp. Entomological Society of America Joint Annual Meeting, Vancouver, BC. November 14, 2022
Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Whiteman, L.B., K.I. Perry, and J.P. Strange. Poster. Bumble bee pathogens vary across an urban to rural gradient. International Pollinator Conference, Penn State University. June 2023.
Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Clare J., Zuckerberg, B.,L. Nunes, Strange, J.P., Hatfield, R., Jepson, S. C. Gratton, Limited bias but loss of precision: impacts of preferential citizen science sampling on estimating bumble bee distributions across Wisconsin. Poster. International Pollinator Conference, Penn State University. June 2023.
Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Moore, I.S., Strange, J.P., Christman, M., Barkan, P.N., Gratton, C. and Tai, T. Effects of Urban Land Cover on Bumble Bee Gene Flow and Genetic Diversity. International Pollinator Conference, Penn State University. June 2023.
Type: Theses/Dissertations Status: Accepted Year Published: 2023 Citation: Whiteman, L.B. 2023. MS Thesis: Bumble bee pathogen prevalence determined by host species. Department of Entomology, The Ohio State University, Columbus, OH
Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Moore, I.S. 2023. MS Thesis: An Evaluation of Species Abundance, Gene Flow and Genetic Diversity in Bumble Bees (Hymenoptera: Apidae: Bombus) in Relation to Urban Land Cover. Department of Entomology, The Ohio State University, Columbus, OH
Type: Journal Articles Status: Under Review Year Published: 2023 Citation: Nunes, L., Tai, T., Clare ,JD., Zuckerberg, B., Jepsen, S., Strange, J. and Gratton, C.Local floral abundance influences bumble bee occupancy more than urban-agricultural landscape context. under review at Insect Conservation and Diversity
Type: Conference Papers and Presentations Status: Accepted Year Published: 2023 Citation: Gratton, C. Pollinator research at the University of Wisconsin. UW Arboretum Pollinator symposium

Progress 04/01/21 to 03/31/22

Outputs
Target Audience: Nothing Reported Changes/Problems:We have been approved for a 1-year no-cost extension to allow us to complete our work by 2023. With COVID-related delays and a renovation of a laboratory space, this has delayed our work. We now have all genetic data in hand and are in the analysis part of the project. What opportunities for training and professional development has the project provided?During this period, Taylor Tai (graduate student) completed her MS thesis and published one paper from her work.Dr. Laura Nunes (post-doc), Dr. Kayla Perry (post-doc, now post-doc at Kent State University), Liam Whiteman (MS student), and Iliana Moore (MS student) have continued on the project.Whiteman is currently processing the specimens for pathogen analysis, including conducting dissections, microscopy, and pathogen DNA extractions. How have the results been disseminated to communities of interest?Gratton, C. L. Nunes., T. Tai, B. Zuckerberg. Local and landscape-scale effects on bumble bee occupancy along a rural-urban gradient. Entomological Society of America. National Meeting, Denver CO. Nov. 2021. Zuckerberg, B., C. Gratton, J. Clare, L. Nunes. 2021 Multispecies integrated distribution models (iSDMs) to evaluate the role and efficacy of citizen science in monitoring bumblebees across the upper Midwest. Entomological Society of America. National Meeting, Denver CO. Nov. 2021. Perry, K., C. Gratton, T. Tai, J. Strange. Functional trait diversity of bumble bees along an urban-rural gradient. Entomological Society of America. National Meeting, Denver CO. Nov. 2021. Perry, KI. What drives bumble bee community composition? 2021. Interview with Andony Melathopoulos, PolliNation Podcast, Oregon State University Extension Service. https://extension.oregonstate.edu/podcast/pollination-podcast/199-kayla-perry-what-drives-bumble-bee-community-composition What do you plan to do during the next reporting period to accomplish the goals?In this final period of the project, now in a no-cost extension period, we plan to finish analyzing the genetic data collected from bumble bees across our urban-rural gradient, and to continue dissections and analyses of the of bumble bees for pathogen loads. From the occurence and occupancy work, we have three manuscripts in draft form, which we plan on finishing in this final period of the project.

Impacts
What was accomplished under these goals? Objective 1. At local scales (i.e., bee occupancy at a survey point) using the field data collected in Y1 (2019) and Y2(2020), we used static single species occupancy models treating each year/site as independent (while accounting for effect of year) to evaluate important drivers of bumblebee detectability and occupancy. Bumblebee detectability was strongly influenced by both the ordinal survey date (generally peaking at the height of summer) but also the temperature at the time of the survey (extreme heat or cold tending to reduce bumblebee activity). The fine-grained occurrence probability of most bee species was not affected by changes in land cover (rufocinctus more likely to locally occur as surrounding urban cover increased), but nearly all were more likely to occur as local floral abundance-but not local floral diversity-increased. Consequently, fine-grained patterns in bee richness were most strongly driven by contemporaneous floral abundance. Using citizen-science datasets, we found that at coarser (e.g., 2 km resolution) grains and larger (e.g., WI) extents, associations estimated within the integrated model suggest that climate (e.g., mean maximum July temperature) most strongly explains differences in bee communities. Urbanization strongly drives volunteer sampling effort, but appears to variably affect the constituent species, with more southerly distributed species in the region exhibiting more positive associations. The effects of agricultural area on bumblebee occurrence remain difficult to parse, largely becausethese areas are undersampled by citizen scientists. Ongoing work aimed at more accurately delineating bumblebee distributions will seek to more explicitly account for variation in volunteer sampling effort to better understand exogenous spatial variation in where different bee species are located. Operationally, the analysis suggests some potential need to better balance the allocation of sampling effort across habitat types. It also highlights the value of having multiple citizen science programs and protocols in place. Although bumblebee watch has extensive coverage regionally and beyond, the additional density of observations provided by bumblebee brigade was necessary to usefully estimate the habitat associations of less common bee species. The small area searches (or complete checklists) associated with bumblebee watch have particular value for understanding bee distributions, in that many species may be observed, absence is recorded, and the underlying sampling effort is transparent. In addition, we explored the relationship between urbanization and bumble bee functional traits. A conservative regional pool of 13 bumble bee species was determined for a 100 km area surrounding Madison, WI using records from Bumble Bee Watch and published literature. For each species in the regional pool, 27 functional traits were compiled either from the literature or via microscopic measurement of specimens. Taxonomic, functional, and phylogenetic homogenization of bumble bee communities was assessed using measures of beta-diversity and null models. We found evidence of homogenization of bumble bees in greenspaces embedded within landscapes dominated by urban and agricultural land cover, as communities were more taxonomically, functionally, and phylogenetically similar than expected from null models based on the regional species pool. These findings suggest that human-modified landscapes favor a subset of functionally similar and phylogenetically related bumble bee species. A manuscript highlighting these results is currently in preparation. We are examining the potential that pathogen loads in bumble bees vary across an urban-rural gradient. All (n = 1224) bees from 2019 have been dissected using 10x-40x magnification. The bees were visually inspected for external and internal mites, parasitoids/parasites, and mermithid worms. Their gut tissues were removed, placed in a sterile vial, and stored below freezing. There were 190 occurrences of conopid parasitoids in 188 bees (two cases of double conopid infections). 6 mermithid worms were found in 6 bees; there were no co-occurrences of mermithids and conopids. External mites were found on 4 bees. DNA has been extracted from all 1224 bee gut samples and 480 have been amplified using the multiplex PCR protocol found in Tripodi et al 2018. Electrophoresis-gel screening on 280 bees found 71 occurrences of Apicystis, 9 of Crithidia, and 2 varimorpha. All gels were photographed. Further evaluation on exact species has not been done. Of the 952 bees examined in 2020, 452 have been dissected and gut tissues preserved. There have been 18 occurrences of conopids parasitoids in 17 bees, 2 occurrences of mermithid worms, and no mites. Something of concern: many of the whole bee samples were spoiled thus making gut tissue extraction challenging. The project objective is to assess how urbanization-driven fragmentation impacts the population genetic health and colony abundance of bumble bees. Using population and landscape genetic techniques, we genotyped approximately 2300 bees (including species Bombus impatiens, Bombus griseocollis and Bombus bimaculatus) at 15 microsatellite loci from Madison, WI to assess gene flow levels, heterozygosity, allelic diversity and colony numbers across 20 collection sites. DNA extractions and PCR have been complete for all bees collected in Madison in 2019 and 2020 (approx. 2300), and allele scoring in Geneious Prime is complete for just over half of these (~1350 bees scored). Genotyping for the Madison bees is projected to be completed by May 15, 2022. Furthermore, approximately 1100 bees of the same three species (along with some of Bombus fervidus) have been collected from 17 sites in Columbus, OH, and impervious surface concentrations within 0.5 and 1.0 km radii around these sites have been calculated using ArcGIS and FRAGSTATS. Genotyping for the Columbus bees is projected to be completed by August 1, 2022. Data analysis remains to be completed for bees from both cities.

Publications

Type: Journal Articles Status: Published Year Published: 2022 Citation: Tai TM, Kaldor A, Urbina D, Gratton C. Within-Year Effects of Prescribed Fire on Bumble Bees (Hymenoptera: Apidae) and Floral Resources. Journal of Insect Science. 2022;22: 7. doi:10.1093/jisesa/ieab107
Type: Theses/Dissertations Status: Accepted Year Published: 2021 Citation: Tai, Taylor. 2021. Within-Year Effects of Prescribed Fire on Bumble Bees Commnities and Floral Resources. MS Thesis. University of Wisconsin, Madison, WI.
Type: Conference Papers and Presentations Status: Accepted Year Published: 2021 Citation: Gratton, C. L. Nunes., T. Tai, B. Zuckerberg. Local and landscape-scale effects on bumble bee occupancy along a rural-urban gradient. Entomological Society of America. National Meeting, Denver CO. Nov. 2021.
Type: Conference Papers and Presentations Status: Accepted Year Published: 2021 Citation: Zuckerberg, B., C. Gratton, J. Clare, L. Nunes. 2021 Multispecies integrated distribution models (iSDMs) to evaluate the role and efficacy of citizen science in monitoring bumblebees across the upper Midwest. Entomological Society of America. National Meeting, Denver CO. Nov. 2021.
Type: Conference Papers and Presentations Status: Accepted Year Published: 2021 Citation: Perry, K., C. Gratton, T. Tai, J. Strange. Functional trait diversity of bumble bees along an urban-rural gradient. Entomological Society of America. National Meeting, Denver CO. Nov. 2021.
Type: Conference Papers and Presentations Status: Accepted Year Published: 2021 Citation: Whiteman, L., T. Taylor, K. I. Perry, C. Gratton, and J. Strange. Distribution and prevalence of conopid parasitoids in bumble bees across an urban to rural gradient. November 1st, 2021. Denver, CO.
Type: Conference Papers and Presentations Status: Accepted Year Published: 2021 Citation: Moore, I.S. and Strange, J.P. 2021 Effects of Urbanization on Gene Flow between Common Eastern Bumble Bee (Bombus impatiens) Populations. Entomological Society of America Annual Meeting, Denver, CO, November 1, 2021.

Progress 04/01/20 to 03/31/21

Outputs
Target Audience: Nothing Reported Changes/Problems:Due to COVID and travel restrictions, we had to adapt our field methods and general timeline for the project. In the summer of 2020, due to our inability to hire undergraduate assistants, we had to scale back the amount of site-specific data that we could collect. As a result, the analyses between the two years of field work (2019 and 2020) were standardized in ways that allowed us to combine the datasets. Nevertheless, all bumble bees samples were collected as expected for Obj. 2. We had to pause genetic and pathogen analyses of samples in 2020 due to our inability to work in the lab. In addition, the Strange lab was in the process of moving from Utah to Ohio State. We are working with coPI Strange to have the genetic analyses processed at OSU, a process which is now underway. We were granted a no-cost extension to allow us to complete our work by 2022. Now that we have access to laboratory facilities, and all samples are in hand, we are confident that we can achieve the outlined objectives. What opportunities for training and professional development has the project provided?During this period, we have been working with Taylor Tai (graduate student), Dr. Laura Nunes (post-doc), John Clare (graduate student), Dr. Kayla Perry (post-doc), Liam Whiteman (MS student), and Iliana Moore (MS student). Tai was responsible for the field campaign in the last summary period and coordinated the training and sampling with the undergraduate team. Tai successfully completed and defended her thesis in Spring 2021 and will be finishing her MS in Summer 2021. Whiteman is currently processing the specimens for pathogen analysis, including conducting dissections, microscopy, and pathogen DNA extractions. Dr. Perry has compiled a functional trait database that includes 42 unique functional traits for each bumble bee species within the region. Using NLCD landcover data, Dr. Perry calculated 23 landscape indices at 500 m and 1500 m radii surrounding each site to assess the impacts of landscape context on bumble bee functional diversity. Dr. Nunes has been the lead on the spatial distribution modeling analysis coming from the Y2 summer field collections. Dr. Nunes presented the results from our Y1 analysis at the Annual Meeting of the Ecological Society of America in August 2020. How have the results been disseminated to communities of interest?Nunes, L, Tai, T., B, Zuckerberg and C. Gratton, "Occupancy and persistence of declining bumble bees across an urban-rural landscape", accepted contributed talk, Annual Meeting Ecological Society of America. Salt Lake City, UT, Aug 2020. What do you plan to do during the next reporting period to accomplish the goals?In the next reporting period, we will work to finalize the analysis from the field surveys of Y1 and Y2 and complete a scientific report to be submitted to peer-review. We will also conduct an analysis on bumble bee occupancy and richness across the Upper Midwest. To do so, we will use data from two citizen science efforts, the Bumble Bee Watch (national) and Bumble Bee Brigade (Wisconsin), for as many bumble bee species as the data allows. We will use species observations from 2017-2020. We will account for species detectability by including the date of the observations. We will estimate the probability of occupancy based on NLCD's land cover database (e.g., proportion of different land cover classes within 5x5km grid cells) and climate data for 2017-2020 from the PRISM database. In the next reporting period, we will finish work to process samples to build the dataset needed to test the hypotheses associated with Obj. 2. Specifically, we are starting to process collected bees to develop the genetic markers data and are now doing dissections to establish parasite diversity and load in bees collected across the urban-rural gradient. Isolation by resistance (IBR) modeling will be used to examine the relationship between genetic structuring and impervious surface density surrounding urban bee populations.

Impacts
What was accomplished under these goals? Activities: 2020 Field sampling: Added new sites to explore more urban sites for 2020 Occupancy Modeling With data at the local scale based on multiple-season sampling and occupancy modeling, we have developed models to estimat the effect of local and regional habitat features on the detection and occupancy probabilities of 9 bumble bee species combining the 2019 and 2020 data together. Using the model estimates from our local scale occupancy analysis and regional habitat attributes we predicted bumble bee occupancy within the greater Madison metropolitan region. Bumble bee sampling and genetics and pathogen analysis Collections of bees for genetics and pathogen work occurred in Y1 and Y2 and are now being processed. Accomplishments: Objective 1. We conducted an occupancy modelling analysis using the field data collected in Y1 (2019) and Y2 (2020). Due to the limitations of the sampling design of year 2, we were not able to conduct a dynamic occupancy model framework. Instead, we conducted a static occupancy modelling approach using both years of data and accounting for the effect of year in the models. Using this approach, we generated estimates on the detectability and probability of occupancy of individual bumble bee species using local (flower abundance and diversity) and habitat variables (urban-rural gradient) and evaluated the effect of these variables on bumble bee species richness. So far, we have found that, or all species studiedbumble bee detectability was strongly influenced by the day of the field season on which surveys were conducted f. The relationship between detectability and date of the survey varied between species, with some species having peaks of detectability earlier in the season and others later in the season, consistent with what is known about the phenology of different bumble bee species. In terms of probability of occupancy, we found that species were more likely to occur at sites with higher flower abundance regardless of large-scale habitat (i.e. landscape) features. Based on initial modeling and spatial predictions we have identified three types of patterns among bumble bee species in this area: Species with (1) High occupancy in both rural and urban environments (e.g., Bombus impatiens, Bombus vagans); 2) Low occupancy in both rural and urban environments (Bombus fervidus, Bombus auricomus); and 3) High occupancy in urban environments that declines in rural environments (Bombus rufocinctus). Thus, the influence of urbanization on occupancy appears to be species specific, with a few species tending to have higher occupancy in more urbanized landscapes relative to rural landscapes (e.g., Bombus rufocinctus), and several species having no associations with a rural-urban gradient (Bombus vagans, Bombus bimaculatus). No species were found to have higher occupancy in rural landscapes relative to urban landscapes. Species richness was also more influenced by local flower abundance.Overall, these findings highlight that species detectability is not constant across the field season or between species (some peaked early while other species peaked later in the season) and thus accounting for temporal patterns in detectability is critical when modelling occupancy and richness of bumble bees. These results also support the hypothesis of a 'honeypot effect', whereby flower-rich sites are able to attract pollinators despite being located in unsuitable habitats. These findings can inform local management efforts for pollinators in inhospitable landscapes, where initiatives to increase floral resource availability have the potential to attract pollinators. Objective 2. Thus far, 403 bumble bees have been dissected, of which parasitic Conopidae larvae (Diptera) have been found in 83 bees. Forty bees have been processed for PCR. Microsatellites representing Apicystis spp., Nosema spp., and Crithidia spp. will be used to detect the genetic markers of parasites and pathogens in the gut tissues. Moore is conducting the DNA extractions of bees and conducting the PCR and genotyping of specimens from 2019 and 2020. Thus far, DNA extractions have been completed for all bees (~2300 individuals), and PCR has been completed for nearly 40% of these bees (~ 875 individuals).

Publications

Type: Conference Papers and Presentations Status: Accepted Year Published: 2020 Citation: Nunes, L, Tai, T., B, Zuckerberg and C. Gratton, Occupancy and persistence of declining bumble bees across an urban-rural landscape, accepted contributed talk, Annual Meeting Ecological Society of America. Salt Lake City, UT, Aug 2020

Progress 04/01/19 to 03/31/20

Outputs
Target Audience:During this reporting period, the target audience of this work primarily scientific, regulatory, and practitioner community interested in the biology and conservation of pollinators. We presented results at general conferences (Entomological Society of America), as well as more specialized venues such as pollinator conservation conferences (International Pollinator Conference). In addition, state agencies (WI DNR) were informed of the basic findings of the project through sharing of bumble bee data. Changes/Problems:There were no major changes in the expected project activities during this reporting period. However, we were unable to start our genetics analyses as planned this spring (February-March) due to COVID-related disruptions in travel. We had to pause this activity until we can determine how this part of the project can be accomplished, hopefully in the Fall 2020-Spring 2021. In addition, we are working on contingency plans for a significantly altered second year of field work. Due to the labor-intensive (needing several undergraduate assistants) and field-based aspects of the work (no social distancing possible and interaction with the public) which are currently disallowed as per University policy, we are altering the intensive sampling protocols that were used in 2019. This is being done such that we can still accomplish the primary goals of objective 2 (distribution modeling and the genetics work), but will require that we modify some of our analytical approaches. What opportunities for training and professional development has the project provided?During this period, we have been working with Taylor Tai (graduate student) and Dr. Laura Nunes (post-doc). Tai has been responsible for the field campaign in the last summary period and coordinated the training and sampling with the undergraduate team. Tai trained three undergraduate students and two recent college graduates to identify bumble bees and local wildflowers to species. In addition, mentees learned how to conduct nonlethal bee surveys in a wide range of habitats and to collect bee specimens for pathogen and genetics work. When not in the field, this team also participated in close readings of relevant scientific literature, guided discussions on field methods and bumble bee ecology, and end of summer presentations to scientific audiences. Dr. Nunes has been the lead on the spatial distribution modeling analysis coming from the Y1 summer field collections. Dr. Nunes will be presenting the results from our Y1 analysis at the Annual Meeting of the Ecological Society of America in August 2020. How have the results been disseminated to communities of interest?C. Gratton, B. Zuckerberg, "Citizen Science Data for Mapping Bumble Bee populations" International Pollinator Conference, Davis, CA, Jul. 2019. C. Gratton, "Big Data and Ecoinformatics in Entomological Research", keynote talk, Big Data in Pollinator research. Symposium Entomological Society of America, St. Louis, MO, Nov. 2019. What do you plan to do during the next reporting period to accomplish the goals?In the next reporting period, we will collect data and samples to continue to build the dataset needed to test the hypotheses associated with Obj. 2. In this next year, we are on track to obtain genetics and parasite data. In addition, we will continue our analyses of datasets obtained for Obj. 1. Despite some COVID-related slowdowns and adjustments of protocols, we expect to be able to fulfill the primary objectives of this project, albeit a slower rate than expected.

Impacts
What was accomplished under these goals? This project successfully completed its first year of data collection in this reporting period. Our major accomplishments this year were to bring on board to the project two key personnel (graduate student Tai and post-doc Nunes). We established a process of selecting study locations, we developed a sampling protocol for bees, and have successfully collected observations and samples for the first year of data. To date we are on track to fulfill most of our milestones for Y1 and well positioned to meet our milestones for Y2. Although COVID-related travel restrictions and social distancing requirements have delayed some of our progress (especially on the genetics front), we have plans in place to meet our objective in this next year of the project. Preliminary results to date have shown two things. First, citizen science data from the newly developed WI Bumble Bee Brigade project and associated expert validation of contributed photos offers important opportunities for data validation. By comparing volunteer and expert-reviewed observations, we found that bumble bees can have high classification accuracy, especially for common species. Moreover, inaccurate (i.e., false negative or false positives) identifications of bumble bee species are highest for species that are especially difficult to identify (B. sandersoni-vagans are challenging to identify from photos, and B. rufocinctus is highly polymorphic). However, even some rare species, that are seldom encountered, such as the federally listed B. affinis (rusty patch bumble bee) are rarely misidentified by non-specialists (False positives = 25%, False negatives = 3%). The consequences of such misidentifications can be important for subsequent analysis of citizen science data. Our framework can now be applied to better understand the factors associated with species misclassification (e.g., seasonality, location, species prevalence of and inform future efforts of modifying protocols or adjusting model-based predictions.) In sum, as the use of citizen science data sources continues to increase, understanding how best to use these data for accurate representations of species distributions will be essential. Second, our initial analyses of bumble bee occupancy patterns along our urban-rural gradient show contrasting patterns between common species (e.g., B. impatiens), moderately common species (B. bimaculatus and B. griseocollis) and rare species (e.g., B. fervidus). The occupancy of common species is estimated to be equally high across our urban-rural gradient. Moderately common bumble bee species are estimated to have higher probability of occupancy in urban areas, with occupancy rates declining in rural areas. Finally, we find the opposite pattern in rare bumble bee species, with higher probability of occupancy in rural areas than urban areas. Overall, we found local habitat features such as the number of flowers at a site to increase both detection and occupancy rates of bumble bees. These are interesting, and somewhat surprising findings since they go counter to some of our initial expectations (e.g., rare species found in urban environments), that will need to be corroborated with a second year of data. Objectives: 1. At regional scales, evaluate the relative importance of urban and rural areas in constraining the distribution of a diversity of bumble bee species in the Midwest. Obtain citizen science and other bumble bee databases Ongoing: we have obtained the Xerces bumble bee watch dataset, and the 2018 WI bumble bee brigade dataset; we are in the process of updating these datasets with 2019 data We have performed preliminary analyses of 2018 WI Bumble Bee Brigade datasets to explore challenges with using citizen science for bee identification data and the use of expert review for assessing rates of false negatives and false positives. 2. At local scales, quantify the variability in bumble bee population size, abundance, lineage survivorship, and pathogens across an urban-rural gradient, and identify the important predictors (flower diversity, abundance, and phenology; microclimates; green spaces) of this variability. Analysis of landscape, establishment of sampling design; specific site identification Completed: we designed a process by which we rigorously analyzed and selected sites within the greater Madison metropolitan area that represented an unbiased selection of sampling locations that spanned a rural-urban gradient Occupancy Modeling With data at the local scale we have estimated the effect of local and regional habitat features on the detection and occupancy probabilities of 9 bumble bee species. Using the model estimates from our local scale occupancy analysis and regional habitat attributes we predicted bumble bee occupancy within the greater Madison metropolitan region. Based on these preliminary spatial predictions, we have identified three patterns: 1) High occupancy in both rural and urban environments; 2) High occupancy in urban environments that declines in rural environments; and 3) Low occupancy in urban environments that increases in rural environments. We assessed the degree of confidence in our spatial models based on the suitability of the sampling design (combinations of number of transects and number of repeat visits). We found that for the 5 most commonly detected species, the sampling design is appropriate thus generating confident estimates for occupancy and detections. The opposite was found for the less commonly detected species. Our assessment suggested insufficient number of repeat visits to confidently infer presence or absence for rare species, which could result in less confident estimates for detection and occupancy. We expect that the additional Y2 data could improve model estimates for these species.... Bumble bee sampling (genetics and pathogens) Ongoing: we designed a sampling design protocol for surveying bumble bees along transects along with measurements of important local covariates. Sampling occurred in Y1 and abundance data for the occupancy modeling is already being analyzed. Collections of bees for genetics and pathogen work occurred in Y1 and will be repeated in Y2. Population genetics and lab work; Pathogen analysis Forthcoming. The data for this part of the objective requires a second year of collection and processing will begin in the Fall-Winter of the next reporting period. Samples for year 1 have been collected and frozen and protocols for analysis (including analytical pipelines) have been adapted from previous studies and microscopic and molecular analyses are ready to begin when workers are allowed to return to campus labs after COVID-19 restrictions are lifted.

Publications

Type: Conference Papers and Presentations Status: Accepted Year Published: 2019 Citation: C. Gratton, B. Zuckerberg, Citizen Science Data for Mapping Bumble Bee populations International Pollinator Conference, Davis, CA, Jul. 2019
Type: Conference Papers and Presentations Status: Accepted Year Published: 2019 Citation: C. Gratton, Big Data and Ecoinformatics in Entomological Research, keynote talk, Big Data in Pollinator research. Symposium Entomological Society of America, St. Louis, MO, Nov. 2019.