Buzzing into Innovation: A Conversation with BeeMachine’s Creator (Ep.39)

World-wide, native bees are facing significant threats from habitat loss, pesticides, and climate change. With over 16,000 known species of bees, pinpointing which populations are most at risk can be daunting. This challenge underscores the necessity for innovative solutions to monitor and support these crucial pollinators.

Enter BeeMachine, a smartphone app designed to assist scientists and bee enthusiasts in identifying and documenting a wide array of bee species. We explore the app’s development, its impact on bee conservation, and practical steps you can take to support native bee populations.

Brian Spiesman is an Assistant Professor at Kansas State University in the US. You can learn more about his work here, or BeeMachine here. Download it from the App Store or Google Play today.

Good to know

Historically, researchers like Brian relied on the work of taxonomists to identify bee species. Taxonomists are scientists who discover, name, classify, and document species. Being a taxonomist and identifying bees requires a lot of education and is hands-on work, just like being an entomologist and studying bees in the field. Enter tech and AI which allows anyone to support building a comprehensive dataset that informs conservation strategies and raises awareness about bee populations.

Transcript

[00:00:00] Welcome to The Bee’s Knees. I’m your host, Jacy Meyer. The Bee’s Knees is a podcast wild about native bees. Wild and native bees are under threat worldwide. In each episode, we look at actionable things we can do to support these adorable little guys whose pollination work is crucial for maintaining biodiversity.
Thanks for being here.

There are over 16, 000 known species of bees found on every continent except Antarctica. We know their populations are declining due to habitat loss. pesticides, and climate change. But exactly which bees over time and space are struggling is harder to pinpoint when you’ve got so many bees to check in on.

Enter Bee Machine, a cutting edge smartphone app. This innovative tool leverages artificial intelligence to help both scientists and bee lovers, like you and I, identify and document [00:01:00] various bee species. My name is Brian Spiesman. I’m an assistant professor at Kansas State University in the Department of Entomology.

Where did the idea for Bee Machine come from? Well, being an ecologist, which is kind of how I was trained, I realized that it’s very difficult to identify bees, especially down to the species level. So there’s so many bees around the world and they all look so similar. We have to look under a microscope to be able to identify them.

We need to contact experts to help us identify them. And it’s just a very difficult process. So I figured, you know, why not try, see how we can do with artificial intelligence, um, computer vision. And it kind of went from there. And so started off with a small little pilot project and, uh, just decided to keep going with it.

So how exactly did you build it? We actually started off with just some images of bee wings. And so wings of bees are. You know, they have all of the venation and [00:02:00] that pattern of venation, you know, we’ve discovered it’s, it’s pretty unique between species. So taxonomists often use the different patterns in the venations to help identify different species, but they never rely fully on that.

And so we wanted to see if, if that pattern can sort of provide a unique marker so that we can identify these that way. So we, you know, got this data set of bee wings and, uh, basically trained it using an artificial neural network. And so that’s kind of how these, these AI models work is with artificial neural networks, which kind of, you know, conceptually sort of mimics, they call it an artificial neural network because it kind of mimics the structure of neurons in, you know, organisms, uh, brains.

So we use that and it turned out it worked pretty well. And so we thought, okay, you know, people aren’t really going to be. Pulling wings off of bees a lot. We want to make a [00:03:00] more usable product, especially for the, you know, everyday person out in their backyard garden or anybody. So we decided to start working with full bees and yeah, so that’s kind of where we are now is just developing large data sets so that we can train our neural networks to be able to identify bees of any type around the world.

So exactly how does it work? The neural networks that we use take an image and it kind of breaks it down into component parts that it can learn from. And so kind of like a person will, you know, view an image and sort of subconsciously break it down into smaller pieces so that we can, you know, process it and figure out what’s actually there.

So kind of like a human face, we’ll look at the eyes, we’ll look at the nose, we’ll look at the mouth. We’ll look at just edges. So what’s the shape of the face? Um, and we’ll sort of take all of those pieces and then assemble it, you know, very quickly in our own minds [00:04:00] so that we can identify it. So do we know this person?

Um, if we know this person, you know, who is it? Does it resemble somebody? That’s kind of how it works conceptually for bees. And so yeah, it’ll take that image. It’ll sort of divide it up sort of hierarchically into these little features that it learns on its own and. It’ll use that to make the identification.

So we, in the modeling process, we don’t tell it anything about the bee. The only thing we do is we tell it, so here’s an image of a bee. Here’s the species that we know that it is, or that we think it is. And the algorithm through by seeing thousands or even millions of these images will. Break it down into these parts and learn on its own what are the key parts of the image.

And so we never tell it, okay, look for this, uh, Venetian pattern or look for the striping pattern. Look for this, you know, the shape of the face on the bees. It learns on its own that the key features that are required to make the identification. And so that’s how it [00:05:00] learns. And then we take that process and we sort of harness it to make, um a usable Product., So the app is available for download and people can upload photos of bees they’ve seen.

Do you have many users and how do these or will these bee photos support your work? Yeah, so we have about 3, 500 users and, uh, we’re, uh, Just about 10, 000 observations around the world. And so that’s great. So right now we only have about 355 kinds of bees. Some are species, some are at the genus level. We can also differentiate between flies and wasps and butterflies and moths and beetles and bees. We’re adding more flower visiting insects. So we’ll be able to say, okay, well, this kind of looks like a bee, but it’s actually a fly. It’s a common thing to confuse. So yeah, we’re constantly adding new species, new taxa so that we can eventually, hopefully be able to cover all of the [00:06:00] bees around the world.

And what do you hope you’ll be able to do with these images? How will they support your work? Yeah, so it’s the images and maybe even more importantly for conserve- for bee conservation is the locations and the species names. So bees are, you know, depending on the bee, some are large, some are small, some have really wide ranges.

Some have more narrow ranges, and so it’s really important to be able to know where the bees are located, how their populations are doing, and are, you know, are they, are they trending up? Are they trending down? Are they remaining stable? And so some of these species vary so widely that it’s very difficult for any single ecologist or conservationist to get a really good handle on how their populations are doing.

So we need lots of eyes out there collecting data all the time. And helping us develop these large data sets that span both space and time. So we need lots of observations throughout the year in, in [00:07:00] the same place, in multiple places so that we can get a handle on how these different species are doing.

So yes, some are increasing. We have some bee species that are doing great. Some are just sort of maintaining and other ones are on the decline. And so it’s really important to be able to know which ones are doing what. I’m having the images, having the locations, the dates and times, having to just having people’s comments about the bees.

And so we can start to develop new kinds of algorithms that take into consideration both the image and the physical description of a bee. So if we have pairs of an image paired with a description of. You know, okay, so this bee was forging on, you know, a Columbine and I saw it in a meadow at this location.

And, you know, it has sort of a description of what’s going on with the bee, what it looks like. So it’s a bumblebee and it’s, you know, T1 is yellow and T2 is black. If we can have these descriptions of like what the bees looks like in what we can see in the image, then we can pair the [00:08:00] descriptions with the images and make more powerful algorithms that way.

So thinking about all the stressors that are affecting native bees right now. Which one do you think is the worst, if we can say the worst, or Potentially the most dangerous so they kind of act together. And so if you had to pick out one key stressor for, you know, that’s causing bee decline right now, I think most people agree that it’s habitat loss.

So, bees needs, uh, they need flowers. They need food to eat. Um, they also need. Places to nest and, you know, these things together were, you know, they have to have both of those things. So you have to have food to eat. You have to have shelter. That’s those are two key fundamental things that, you know, any living thing needs.

So habitat loss can reduce both of those things. So we’re reducing. Both, uh, the amounts and the diversity of different flower types on the landscape. So we need a lot of [00:09:00] native flowers, you know, in any location we need the native flower species that occur in those locations. So if we don’t have those, then, you know, they’re going to be competing more and more for those limiting resources, but we also need places for them to nest.

Most bee species are ground nesters. So a lot of people, when they think of bees nesting, they think of hives. Most species are not living that way. . They live underground in solitary nests, so you have one female, she acts as the queen, and the forager, and the nest builder, she basically does it all, and she goes out, collects pollen and nectar that she uses to feed her brood, her baby bees in her nest that she constructs herself, and so we need places like that, sort of undisturbed places where bees can nest in the ground, And have the resources nearby to support those nests.

Um, so, yeah, so habitat loss is a key factor, you know, it’s one of those things where a single person can actually do a lot. So, if you, you know, you [00:10:00] plant a few more flowers in your yards and your gardens, you keep, you know, devote instead of having so much lawn, take a little bit aside and plant some native flowers.

That’s great. Just things like leaving. Undisturbed areas in your backyard. So we want to, you know, a lot of people like to keep their yards nice and tidy and make sure everything is pristinely manicured and that looks great. But for bees, that’s not necessarily what’s best. And you don’t have to make your whole yard, you know, look like a wild prairie or anything, but having a small section set aside where maybe you don’t rake or pick up leaves and things like that.

Then that’s fine. All of those places, or all of that together can help bee populations, both by providing floral resources and by providing nesting habitat. So why do you do what you do? Why bees? I think they’re interesting. When I started as a grad student, I started off as a community ecologist. So I look at sort of natural communities of organisms.

How they [00:11:00] live together, how they compete, you know, what makes some species do better than others. You know, you have predation, you have competition for resources and you have mutualism. So bees and flowers are mutualists. You know, from a scientific standpoint, it’s kind of interesting to see how all of these factors work together.

So you have these positive interactions like bees visiting flowers and pollinating. Flowers provide resources for bees and sort of sacrifice some of the energy they take up to make. Nice flowers for bees that are very attractive. And so, you know, all of these, these sort of interactions between plants and animals.

And it’s very interesting to me to see how, you know, there’s also many limiting resources out there. We need to be able to understand why devote key energy to making showy flowers or why devote key energy to, you know, only visiting certain kinds of flowers. And that to me is really interesting. And so bees and other pollinators make a great study organisms.

[00:12:00] And so it’s just sort of interesting for me to figure out what, you know, how it all works together and what creates a stable ecosystem, what creates that ecosystem that’s diverse and can sort of weather the storm. So we find that oftentimes More diverse ecosystems, so more different kinds of species, uh, living and working together, trying to survive together can be more resilient to different kinds of disturbances.

So, you know, anything from storms to fires to, you know, human caused disturbances, the more diverse the ecosystem, the more resilient it is oftentimes. And so. I try to understand why that is, is very interesting to me, at least conservation is something that’s really important to me. Everybody can contribute in some way, whether it’s, you know, actually planting flowers in their yard or just, you know, if you download the mobile app, next time you see an interesting insect on a flower, take a picture of it and hopefully we’ll be able to identify it for you.

And. Yeah, I’m hoping that it turns into, you [00:13:00] know, one of these sort of citizen science projects that anybody can take part in. You don’t have to do it consistently, but, you know, you’re just, you happen to be out at a garden or, you know, on a nature walk or something like that. You see something interesting, take a picture of it and see what it is.

And hopefully it will spark a little bit more interest in pollinators and bees and, uh, you know, get people a little bit more interested in nature. You know, a lot of people are really interested in birdwatching. So that’s a really. popular activity. Um, we’re hoping that bee watching becomes sort of the next big thing.

The challenges facing our native bees all over the world highlights the urgent need for immediate conservation efforts. Your involvement can make a significant difference. Download the Bee Machine app and start documenting the bees you encounter. By snapping photos and submitting your sightings, you contribute valuable data that helps researchers monitor bee populations and identify trends that could steer future conservation efforts.

Or as Brian suggested, leave a [00:14:00] little bit of your garden or yard a bit messy. Bees don’t need much. My thanks to Brian for joining us today, and thanks as always to you for tuning in to The Bee’s Knees. I hope you enjoyed our conversation. Please follow and tell your friends about the podcast. Your kind words really help the show grow and spotlight the bees.
Until next time, become a bee watcher.