Community science participants gain environmental awareness and contribute high quality data but improvements are needed: insights from Bumble Bee Watch

Introduction

Community science, which is also commonly known as citizen science, is a fast-growing field where participants commonly collect and/or analyse data as part of a scientific study, often overseen by professional scientists (Silvertown, 2009; Griffin Burns & Harasimowicz, 2012; Miller-Rushing, Primack & Bonney, 2012; Birkin & Goulson, 2015). These projects can deliver baseline or monitoring data, answer research questions, provide skills training and specific knowledge, increase stewardship and awareness, and influence conservation actions and policies (Bonney et al., 2009b; Wiggins & Crowston, 2011; Conrad & Hilchey, 2011; Griffin Burns & Harasimowicz, 2012; Newman et al., 2012; Danielson et al., 2014; Follett & Strezov, 2015; Jordan et al., 2016; Thomas, 2016; Le Féon et al., 2016; Acorn, 2017). While the majority of community science projects involve a biological topic, other examples exist ranging from other sciences (e.g., astronomy, atmospheric sciences) to traditional ecological knowledge, model building, monitoring traffic, medical fields, and informational sciences (Wiggins & Crowston, 2011; Shirk et al., 2012; Follett & Strezov, 2015). Yet the need to create community science frameworks that are successful and sustainable is consistent across all types of projects.

An important part of any successful community science project is an understanding of who the participants are, their reasons for participating, and how their experiences have been to date (Bonney et al., 2009a; Bonney et al., 2009b; Silvertown, 2009; Berg, Dann & Dirkx, 2009). Community science volunteers, particularly those in nature-focused programs, participate for many reasons, such as to learn more about the world around them (Bonter et al., 2012; Griffin Burns & Harasimowicz, 2012; Shirk et al., 2012; Trautmann et al., 2012; Russell, 2014; Van der Wal et al., 2016), to contribute to our overall body of knowledge (Dickinson & Bonney, 2012; Griffin Burns & Harasimowicz, 2012; Shirk et al., 2012; Trautmann et al., 2012; Russell, 2014), to aid in conservation efforts (Shirk et al., 2012; Van der Wal et al., 2016), to increase personal skills and abilities (Bell et al., 2008; Bonter et al., 2012; Dickinson & Bonney, 2012; Shirk et al., 2012; Van der Wal et al., 2016), and/or to simply spend time in nature or with like-minded people (Bell et al., 2008). Individuals who have participated in nature-related projects often participate in other nature-related activities (Federal Provincial and Territorial Governments of Canada, 2014), and many community scientists have knowledge or skills in the area they volunteer (Bell et al., 2008; Bonter et al., 2012; Cooper, Hochachka & Dhondt, 2012; Hames, Lowe & Rosenberg, 2012; Miller-Rushing, Primack & Bonney, 2012; Shirk et al., 2012). While anyone can be a community scientist, participants tend to be highly educated, white, female, and late middle aged through the early senior years (Bonter et al., 2012; Purcell, Garibay & Dickinson, 2012; Toomey & Domroese, 2013). They tend to participate infrequently, submitting few records per person (Worthington et al., 2012; Silvertown et al., 2013; Silvertown et al., 2015; Kelling et al., 2015; Domroese & Johnson, 2017).

There are often spatial biases in community science programs with more data coming from heavily human populated areas than less populated areas (Cooper, Hochachka & Dhondt, 2012; Van der Wal et al., 2016) as there are more potential participants there and it may be easier for them to participate close to their homes (Davies et al., 2011; Van der Wal et al., 2016, although there are exceptions (Cooper, Hochachka & Dhondt, 2012). As well, while individuals may have strong conservation views regardless of where they live (Lutz, Simpson-Housley & De Man, 1999), some studies have found that urban dwellers may have stronger environmental and conservation views (Berenguer, Corraliza & Martin, 2005; Olive, 2014; Olive, 2015). Different regions and audiences also may need different approaches to outreach and education (Bubela et al., 2009; Nisbet & Scheufele, 2009; Chu, Leonard & Stevenson, 2012).

Bumble Bee Watch is a community science program run across Canada and the United States in collaboration by the Xerces Society for Invertebrate Conservation, York University, Wildlife Preservation Canada, and other organizations. It collects information on (Hymenoptera: Apidae: Bombus) species phenology, distribution, and more, with the goal to use these data to help conserve at-risk bumblebees. It was launched in 2014 with a web-based platform (http://www.bumblebeewatch.org), with mobile applications for iOS and Android launched in 2017 and 2018 respectively. Based on the idea that bumble bees may, in many cases, be identifiable to species from photographs (Lye et al., 2011; Suzuki-Ohno et al., 2017; Falk et al., 2019), participants take photos of bumble bees and submit through the web or app, along with a date observed and location information. They have the option to use an interactive key or smart filter to assign species name (pre-limited by location), and regional experts verify the identifications. The website also includes resources for the general public and conservationists, including an increasing photo gallery of verified species images, and allows individuals to view data for species, months, areas, and/or to request data for specific project needs.

We wanted to compare metrics on the Bumble Bee Watch program participants, including their knowledge about bumble bees, experience with bumble bee identification, and general feedback, in order to compare it to other community science programs. While Bumble Bee Watch data is currently being used by researchers, such as for species conservation assessments (Szymanski et al., 2016; Committee on the Status of Endangered Wildlife in Canada, 2018; MacPhail, Richardson & Colla, 2019) and regional checklists (Gibbs et al., 2017), we also wanted to know how experts in the field of bumble bee biology and conservation perceive the program, including their experiences with verifying data for the program and actual or potential uses of the data. With over a quarter of our North American bumble bees in decline (IUCN, 2019), and continuing stressors identified (Cameron & Sadd, 2020; Soroye, Newbold & Kerr, 2020), this program has the potential to help contribute data for conservation efforts. However, not all community science programs are successful, failing or being shut down due to a variety of reasons, from poor study design to a lack of funding (Conrad & Hilchey, 2011; Dickinson & Bonney, 2012; Silvertown et al., 2013; Hannibal, 2016; Acorn, 2017). One very important area that can be neglected is the need to ensure that volunteers have a good experience and have their expectations for participating met (Silvertown, 2009; Conrad & Hilchey, 2011; Chu, Leonard & Stevenson, 2012; Fitzpatrick, 2012; Greenwood, 2012; Purcell, Garibay & Dickinson, 2012; Silvertown et al., 2013).

Our main research questions were:

1. What motivates Bumble Bee Watch participants to participate in the program?

2. Are the demographics of participants and participation rates similar to other community science programs?

3. Are there more participants and submissions from urban areas than rural or suburban ones?

4. Do individuals from urban areas have stronger views related to conservation than suburban or rural ones?

5. Do participants increase their knowledge about bumble bees after participating in the program?

6. Do participants increase their skill in bumble bee identification after participating in the program?

7. Have experts used Bumble Bee Watch data in their own research to date?

8. What aspects of the Bumble Bee Watch program are working well, and which ones can be improved?

Methods

Data from Bumble Bee Watch Program

An export of the main Bumble Bee Watch program database, including both pending, tentative, and verified nest and bee records, was obtained on February 28, 2018. Summaries were calculated in Excel (Microsoft Office 365 ProPlus, Version 1909) to compare responses obtained in the user survey to metrics from the database, such as the relative number of contributors and the number of submissions per person and per province/state/territory (“jurisdiction”). The number of registered users of Bumble Bee Watch was provided by Rich Hatfield of the Xerces Society for Invertebrate Conservation to VJM on February 28, 2018; this number differs from that of the number of contributors to the Bumble Bee Watch database as not everyone who had created an account had submitted records.

Results

Discussion

Bumble Bee Watch participants are highly motivated by the desire to contribute to scientific data collection and to help save bumble bees, as well as to further their own education, such as learning about local species, and for social reasons, such as an activity to participate in with family and friends and/or as part of a special event, in addition to general enjoyment. These motivations were consistent across ages of participants, years of participation, and areas they live in. Most of the participants reported becoming more aware about bumble bee diversity and improving their identification skills after participating, and the top feature of the program they liked were the resources on the website. To keep the program current and sustainable, it is important to understand and continue to motivate participants.

These are all common motivations in community science projects (Van Den Berg, Dann & Dirkx, 2009; Bonney et al., 2009a; Bonney et al., 2009b; Bonney & Dickinson, 2012; Chu, Leonard & Stevenson, 2012; Newman et al., 2012; Shirk et al., 2012; Silvertown et al., 2013). Our results also show that the desire for bumble bee conservation is as high as the general contribution of scientific data; participants could be given more concrete ways that they can help and/or they could be recruited by other projects to help move conservation efforts forward in their areas. Bees are popular animals right now, and The Great Pollinator Project (Domroese & Johnson, 2017), found that volunteers ranked the desire for education and learning about pollinators to be the number one motivation followed by the participation in a community science project.

These results are consistent with other research which show programs with a nature-based theme often recruit individuals who care about the environment and wish to take action to help, unlike broader public education programs where participants are motivated by other reasons, such as career advancement (Guiney & Oberhauser, 2009; Van Den Berg, Dann & Dirkx, 2009; Newman et al., 2012; Lewandowski & Oberhauser, 2017). Regardless, it is important for each project to recognize the motivations driving their volunteers and to ensure these desires are met in order to keep interest and thus retention high (Bonney et al., 2009b; Bonney et al., 2009a; Silvertown, 2009; Conrad & Hilchey, 2011; Chu, Leonard & Stevenson, 2012; Fitzpatrick, 2012; Greenwood, 2012; Purcell, Garibay & Dickinson, 2012; Silvertown et al., 2013; Domroese & Johnson, 2017). In the case of Bumble Bee Watch, this can be accomplished by showing the participants the scientific value of their data (e.g., frequent summaries of the data and examples of how it has been used, such as through selected publications) and how it is helping to meet the project goals (e.g., where it has been used in species at risk policy designations, conservation decisions), in addition to providing more educational resources to the participants (e.g., tips on creating nesting habitat, best flower species to plant, guidance on how to reach out to decision makers). This has already been occurring through the use of an e-newsletter and blog hosted by The Xerces Society for Invertebrate Conservation (2020) but more frequent and regional communications would be beneficial to keep interest high.

It is also important to improve the user experience with the program to encourage continued participation, from the initial recruitment and registration through reporting and retention stages. Community science projects traditionally have low participation versus registration rates, and low contributions on a per person basis (Worthington et al., 2012; Silvertown et al., 2013; Silvertown et al., 2015; Kelling et al., 2015; Domroese & Johnson, 2017); Bumble Bee Watch is no exception. About 39% of registered Bumble Bee Watch users submitted at least one record, which is higher than other programs, such the 25% of registrants in the Great Pollinator Project (Domroese & Johnson, 2017) and 30% of those who contacted organizers in the Protea Atlas Project (Silvertown et al., 2013).

Virtually all (94%) of the Bumble Bee Watch database contributors contributed only a few (<10) records, with only a few (0.3%) of participants submitting >100 records (17.1% of the data). This was also seen among our survey respondents. Similarly, the Protea Atlas Project found that only 20% of their participants submitted more than 50 site records, and 2.5% more than 1,000 (Silvertown et al., 2013); with the eBird program, 10% of the participants submitted 90% of the records (Wood et al., 2011). These are low numbers of active users but quantity of participants and quality of the resulting dataset are not always equal. These super volunteers or power users often contribute high amounts of high quality data to a program that can help to offset the low participation; these individuals should be especially targeted for retention (Wood et al., 2011; Hames, Lowe & Rosenberg, 2012; Silvertown et al., 2013; Kelling et al., 2015; Hannibal, 2016).

Programs generally want to increase the number of new participants each year to both grow the program and offset participants who do not return. For example, FeederWatch has a 70% annual retention rate, but needs 3,000 to 4,000 new participants annually to keep participant numbers and geographic coverage consistent (Bonter et al., 2012). It could be expected that distribution of years of participation would be greatest numbers in the most recent cohort and lower numbers each year. This was seen in the Great Pollinator Project where 41%, 36%, 17%, and 7% of survey respondents participated in one to four years, respectively, of the project (Domroese & Johnson, 2017), but the greatest number of our Bumble Bee Watch survey respondents were not from under one year or first year participants but for second years (33%), although the fewest numbers of respondents (6.4%) had been the early adopters, participating for four years. This may be a feature of program roll-out and advertisement being stronger at the time these second-year participants started; further investigations into the actual numbers of new and returning users as compared to promotion efforts would be needed to better identify any correlation. However, it is clear that volunteer retention is key; indeed, Sheppard et al. (2017) believe retention is a measurement that can account for both data quality and volunteer engagement.

Our user survey respondents had a number of similar responses in terms of what they liked about the program and what could be improved. Commonly stated areas for improvement related to their motivation to learn and to see that they are helping science. Users wanted to get faster feedback on their submissions, more tools and resources to identify and attract pollinators, and increased communication to learn what others have been finding and how the data has been used. In regards to the communications with others, in-person workshops and direct contact with experts are not always possible, but there may be a role for virtual training and online community discussion forums in sharing information and experiences and getting feedback and advice (Bonney & Dickinson, 2012; Bonter et al., 2012; Triezenberg et al., 2012; Toomey & Domroese, 2013; Austen et al., 2018).

The Great Pollinator Project also struggled with meeting the desired speed of feedback and sharing of data to users, particularly for higher level summaries and reports that can take a long time before analyses can be completed, and encouraged program leads to find solutions for this common problem (Domroese & Johnson, 2017). While providing feedback can be time intensive, the Bee Watch program developed an automated natural language generated feedback program that immediately gave generalized feedback and information to users after expert review based on the species being (mis)identified (Blake et al., 2012; Van der Wal et al., 2016). The Bee Watch program found that this automated feedback greatly increased participant learning, skill, and retention (Blake et al., 2012; Van der Wal et al., 2016). The Evolution MegaLab project gave immediate automated feedback to users upon record submission, including text and figures comparing their submission to historic records in the same area, which also helped to motivate their participants (Worthington et al., 2012). Several community science programs, such as iNaturalist, Pl@ntNet, and Merlin Bird ID, use an artificial intelligence and deep learning computer programs to suggest identifications based on comparison to other photos in their database and nearby records , which provides instant feedback to users (Wäldchen & Mäder, 2018; Ceccaroni et al., 2019). The use of new technologies, including new software programs, will help to remedy some currently existing challenges in both Bumble Bee Watch and community science programs in general to improve data collection and the user experience going forward (Newman et al., 2012; Wäldchen & Mäder, 2018; Ceccaroni et al., 2019).

The areas of improvement noted by Bumble Bee Watch users were also commonly the main areas users enjoy with the program, such as the resources available on the website and their ability to help save the bees through the sharing of their data. This suggests that although more can be done, such as customized regional bee identification workshops or pollinator-friendly gardening guides, program organizers are on the right track with the provided features. We do not discuss the accuracy or perception of user and expert identification in this paper as has been done for other bumble bee community science programs (Lye et al., 2011; Suzuki-Ohno et al., 2017; Falk et al., 2019), but this information (including commonly mis-identified species) would also help to improve the program and data quality (but see MacPhail et al. (unpublished data) for this).

Our survey was conducted in the winter months when bumble bees are not active across most of North America, and submissions to the program are thus low. It is possible that users might have had slightly different opinions, or different rankings of opinions, than if we had done the survey during the peak of the active season. For instance, respondents may have forgotten difficulties such as potential issues with the submission process but been more aware of the delays in getting verification of their photos. However, as some respondents had multiple years of participation to reflect on and some submit their photos throughout the fall and winter, and as a wide variety of responses were recorded across the participants, any temporal-related biases in our results are likely low.

The experts who responded to our survey were generally pleased with the Bumble Bee Watch program, although they also had suggestions on how to improve the program interface and quality of the data. The main areas that they believed the data could be used for included filling knowledge gaps, supporting provincial and federal assessments, effecting policy and legislation, encouraging habitat restoration and management efforts, and guiding further research. However, despite the many areas of potential use, only about half of the experts have actually used Bumble Bee Watch data in their research. The advertisement of the dataset availability and example questions may be worthwhile for Bumble Bee Watch to share with researchers and conservationists through formal academic means and social media. As the program and dataset grow, more avenues to use the data will also evolve. Indeed, Bumble Bee Watch data is routinely accepted into the Bumble Bees of North America database managed by Dr. Leif Richardson and the USGS BISON (Biodiversity Information Serving Our Nation) database, and has also been added to other national and regional databases, so the data is increasingly becoming available to researchers.

Over half of our user survey respondents were middle-aged, between 45 and 64 years of age, which is common of many programs as they include the group who often have more time to volunteer as they have older children and more stable employment or have retired (Guiney & Oberhauser, 2009; Purcell, Garibay & Dickinson, 2012; Toomey & Domroese, 2013; Domroese & Johnson, 2017; Sheppard et al., 2017). Interestingly, we had more individuals that were 75 years of age and older respond to our survey than those under 24 years of age. While we had few respondents in the latter group, this may be because most youth under 18 would only participate with an adult and/or under an adult’s account. Yet there are benefits for youth participation, not only to get them started on a life-long love of nature and contribution to community science and conservation projects but also because they can often be single-focused and energetic, and be more skilled than adults in some areas (Wells & Lekies, 2006; Guiney & Oberhauser, 2009; Griffin Burns & Harasimowicz, 2012; Silvertown et al., 2013). Our results show that there was no difference in the number of records submitted per age group. Our program encourages people to get outside and be active while enjoying their local gardens or trails, which is a necessary part of having a healthy physical, mental, and social life, especially for older adults who tend to become inactive over time (Buchner et al., 1992; Wagner et al., 1992; Phillips, Schneider & Mercer, 2004; Bushway et al., 2011).

While every province, state and territory that have bumble bees were represented in the Bumble Bee Watch database, some of these regions have low participation rates as compared to the relative proportion of the population and to the entire program as a whole. Increased promotion of the program in these areas, including through targeted presentations, ”bee walks”, and dedicated survey programs could help to increase data coverage. For example, one of us (VJM), gave a public presentation, bee walk, and CBC media interview on Prince Edward Island in August 2019, and in the following three months the number of submissions had almost doubled (1.8 times) what had been submitted from the province in the previous five and a half years combined; in fact, 25.5 times more records were submitted in the three-month August through October 2019 period as compared to the average of the same three months in the preceding five years. After a single CBC media interview in New Brunswick in August 2019, about 5.8 times the number of records were submitted in August through October time period compared to the average of those months over the previous 5 years. General advertising should account for the motivations of residents from different types of areas of physical and demographic communities, such as advertising the ability to find and share rare species to those in rural and suburban areas or to participate in special events to those in urban areas. Although there was no difference in the number of records submitted based on the area in which people live, urban residents had the least number of respondents in our survey; it is possible that this is due to the erroneous belief that cities do not contain bumble bees or that they are already well surveyed: advertising could help alleviate these concerns. As we did not provide definitions as to what constituted urban, suburban, or rural, it is also possible that some respondents identified with the suburban category instead of the urban category, explaining the lower number in the latter group.

While there are challenges with community science programs, there is an increased understanding of the need for good practices, from data collection and validation to participant communications and retention (Bell et al., 2008; Silvertown, 2009; Conrad & Hilchey, 2011; Cooper, Hochachka & Dhondt, 2012; Phillips, Bonney & Shirk, 2012; Worthington et al., 2012). The development of appropriate protocols and supports (administrative, technological, and financial), can overcome the challenges and ensure the positive, long-lasting outcomes of these programs (Conrad & Hilchey, 2011; Cooper, Hochachka & Dhondt, 2012; Dickinson & Bonney, 2012; Pocock et al., 2015).

Conclusion

Bumble Bee Watch has the potential to engage the public, answer research questions, and increase the conservation of bumble bees across the United States and Canada through avenues resulting from having thousands of participants of various ages from rural through urban locations increasing their understanding of bumble bee diversity and their desire to conserve them while generating the data for research and conservation purposes via photo submissions of the bumble bees they see across the landscape. Improvements are needed in order to maximize volunteer retention and data quality, which could include the adoption of new technology such as automated feedback and the provision of more targeted resources such as identification guides and species lists. Further promotion of community science programs amongst academia, conservationists, policy makers, politicians, and the general public should occur in order to increase the number of participants and submissions (both geographically and temporally), the use of the data amongst a broader group of stakeholders and finally, our knowledge of species and our interest in conserving them.

Supplemental Information