[Automated Transcript]

Deep: Hello, I'm Deep Dhillon, your host, and today on your AI Injection, we're joined by Dr. Chris Cosma, a community ecologist and pollination biologist at the Conservation Biology Institute, with a PhD from UC Riverside. Chris has worked extensively on the effects of climate change on plant insect interactions, and he's developed innovative conservation tools, such as the butterfly net, now widely used in California.

Chris, thanks so much for joining. So Chris, maybe start us off by telling us like, what's the problem that you're trying to solve? Uh, what happens when somebody doesn't use your approach or solutions, like, what's the typical thing going on today? And, what's different, you know, when they use your approach or solutions?

Chris: , I, I work in conservation. I'm, I'm specifically an applied ecologist. And so, you know, we, we have a lot of tools in our tool belt for addressing things like biodiversity loss, climate change mitigation and all of these sort of, uh, what can, what can come off as very existential issues that we're facing today as a planet.

and so, AI is just, you know, It's just, to me, one tool in the toolbox for addressing these sorts of problems. and, and I really view it as sort of just, again, another tool that specifically allows us to, address these problems in a more powerful approach. So, analyzing larger amounts of data quicker.

And gaining more, useful insights from that data. but that's not to say that, you know, AI is the only approach that I use, and, in fact, far from it. So, , most of my actual background is in, , field ecology and actually going out there, into the field and collecting data on, these species and these organisms, and their traits and how they're interacting with each other and their environment and, and stuff like that.

Deep: So maybe let's jump up a couple levels like what's what's going on in the world today with respect to insect populations and what are your areas of concern maybe start there. 

Chris: The main thing that we're seeing today with biodiversity in general, , and this may come off as a little bit obvious given the news headlines, , but yeah, biodiversity is declining.

So,, with insects in particular, we're seeing somewhere around a 1 2 percent loss of insect populations per year. This magnitude and this rate of loss is a really major problem.

And it's a problem for humans because we actually rely on these organisms, including insects, for, things like our food production. So what I specifically focus on are plant pollinator interactions. And we know that about, uh, 90 percent of all plant species, and that includes around 70 percent of our crop plants, rely on insects to pollinate them and to produce The fruits and the seeds and the nuts that we eat, and so when we're talking about this rate and this magnitude of insect loss and biodiversity loss in general, it's not just that those species are disappearing and that is inherently bad because of their intrinsic value.

 But it's also, really bad news for us as, other organisms on this planet, and also bad for ecosystems in general, um, and the services that, that they supply. 

Deep: So, are we, we're talking about bees in particular, or are we also, in the decline of the bee population, or are we also talking about butterflies?

Like, what exactly are we talking about when you say pollinators? 

Chris: Yeah, so pollinators is a pretty broad term and it, it does encompass bees. That is probably the first thing that comes to mind for most people when they hear the word pollination or pollinators. And bees are very important pollinators and they've gotten a lot of attention for that.

But, part of my, my research agenda and my work that I'm doing now at CBI, is trying to also shift the focus to our more, underappreciated pollinator groups, and that includes moths. So, I actually did my PhD in moth ecology, , and moths, it turns out, are extremely important pollinators.

They're usually active at night, so we don't see them as much out there visiting flowers, but they are pollinating just like bees are during the day. You know, it's not just insects, so pollinators are all sorts of different groups of insects, bees, butterflies, moths, flies, but we also have hummingbirds are important pollinators, and then even mammals can pollinate, so it's a pretty, pretty, broad array of different organisms, but insects are the primary pollinators.

Deep: And tell us when pollinators decline, like, what do they normally do, specifically with respect to, our food supply, and what happens when their numbers are dropping.

Chris: Yeah. So that's a great question. Um, and probably the reason that insect decline is getting so much attention is because like you said, we rely on insects for our food supply and getting a little bit more technical. What that means from a pollination perspective. These organisms, these insects and other animals are actually physically transporting pollen, which are basically plant sperm, , from one individual to another individual, um, and thereby allowing those plants to reproduce.

And so this is essentially how plants have sex. For plants, they, they can't just get up and walk and meet their neighbors, and reproduce that way. Um they actually do rely on other organisms to transfer that pollen for them and to allow reproduction and kind of scaling up to how this all affects humans is that, uh, that active pollination of transferring, , pollen to the female reproductive parts of flowers.

 Actually results in a lot of the food that we eat. So many of the fruits and vegetables that we eat, , nuts and seeds, in fact, about one in three bites of food that we eat, come from the interaction between a plant and a pollinator. And for the most part, those pollinators are insects. 

Deep: So walk us through that in more detail.

So I think most humans are not eating wild food. They're eating food that's going to grown in some kind of agricultural operation. and most of us think of a farm as being, somebody planting seeds and something growing, impact specifically in that? Commercial food production chain.

Chris: It's essentially at every stage like you mentioned wild plants, non agricultural plants, pollinators are, of course, important to our native wild plant communities. And that's another major concern. About pollinator decline is that this is going to affect our natural ecosystems and those natural ecosystems supply, , tons of ecosystem services like air and water purification.

So that's another way that pollination can affect humans. I mean, these crop plants, are also relying on insect pollinators. So, even though they're, bred and cultivated varieties of plant species, , pretty far removed from their native ancestors, they still, many of them, require, , biotic pollination by insects.

 And, in fact, it's about 70 percent of all of our major food crops still either completely depend on or benefit from insect pollination. 

Deep: Is that because the seeds that get harvested from The plant will not bear fruit unless that plant was pollinated. 

Chris: That can be the case. Yeah. So, I mean, a lot of times if a plant is not outcrossed, it either just will not produce fruit at all or it will produce lower quality fruit.

So a lot of the benefit that we see, , coming from pollination is, is not just that it helps produce the fruit in the first place, but that it leads to larger and higher quality fruit. Uh, fruit and mostly the type that we desire as consumers. So those big juicy strawberries and blueberries that you're eating, they get that way mostly from biotic pollination.

Deep: Okay. So now some kind of dumb questions like, why are these pollinators dying off? Is it just loss of. loss of acreage of wild lands. Is it roundup everywhere? Like what's actually killing them off? 

Chris: Yeah. I mean, that's not a dumb question at all because it's, it's actually a very hard question to answer.

So, the phrase death by a thousand cuts has been applied to this problem, It's basically this idea that there, there are just so many threats being thrown at insects from all sides. And, and every one of them is contributing in part to their decline. , and so some of the major ones we're seeing here are, , climate change.

So that's, a lot of my, my past research is focused on the effects of climate change on plant pollinator interactions and insects in general. Um, and you know, just, just like when it gets too hot for humans to be comfortable and sometimes to survive, the same is true for, other animals. So sometimes we're seeing, you know, direct mortality caused by climate change, but it also causes all sorts of indirect effects by, for example, reducing their food supply, , in the plants that they're, , gathering pollen and nectar from.

, some other major threats to pollinators are habitat loss. , in fact, that's probably, the data tells us that that is still the number one cause of biodiversity loss across the board, and it certainly is a major driver of pollinator loss. when we mow down. native natural ecosystems and replace them with agricultural fields, or, parking lots or houses, you are getting rid of the resources that native insects and other wildlife relies on.

, and that, that is a major problem for, for insects and the rest of biodiversity. , and then, , to add onto this list, uh, some other threats of invasive species. So, you know, when you have. , non native species moving into those native ecosystems and some, some cases replacing the native plants, for example, that those pollinators have evolved to rely on.

You're also reducing the quantity and quality of their food supply. And then you have other invasive species that are direct, pests or pathogens of native pollinators. And then another major one is, pollution in various forms and, and light pollution is actually a pretty under recognized threat to insects, partly because, a large proportion of insects are nocturnal, around 60%.

 Or so of all insects are nocturnal and so light pollution is actually a major disruptor to , their natural life cycle and , their natural activities that they should be performing out there like feeding and reproducing, and light pollution can actually disrupt all of those things and also exacerbate insect declines.

Deep: so, enumerated a bunch of stuff, but some of them seem, easier for me to wrap my head around, remediation, or like, ways to kind of deal with it, so, habitat loss, for example, it seems like the clear thing to do is to restore habitat, or prevent loss, but, some of these other ones, I don't, I don't even know what, where you would start with something like invasive species, I mean, I guess you can jack up laws for, Whatever agricultural inspections or something, but it seems like a hard one.

How are you prioritizing your guys efforts at the biology Institute at the conservation biology Institute? 

Chris: Well, yeah, I mean, at CBI, we address a lot of different problems from a lot of different angles. And, and like you mentioned things like invasive species are very difficult questions and problems to tackle, it really, at the end of the day, we're a great team.

It requires coordination from a lot of different entities, , scientists like myself, policy makers. We, we really need strong policies in place to, uh, first of all, help us do our work better as conservationists, and restorationists, but also, to just, , put the conditions in place there to prioritize biodiversity protection. It can be extremely challenging and, sometimes, uh, too challenging. Sometimes, what we tried doesn't work, but, I think that A key point is that we just need to be putting more and more resources, more time, more ideas and more tools, including AI into this fight for biodiversity.

 Or else, we're gonna start to see some major impacts here. 

Deep: okay, let's, let's switch a little bit to the AI thing, because you brought it up, like, so. where is the bulk your energy going right now? Is it in the monitoring and trying to figure out and measure population loss?

Because I can imagine, you can, maybe leverage machine learning like computer vision or to sample and compute insect density levels or bee density levels based on cameras or traps or a combo. That seems like a use case. And like, what are some other use cases?

I imagine you're also looking at certain types of ecosystems and then maybe looking at satellite or. Drone or other aerial imagery to figure out habitat loss and you know, which areas are more vulnerable. Like, maybe walk us through how do you stack rank the AI usage scenarios in terms of their potential benefit and impact to this kind of overwhelming problem?

Chris: I'll start by saying I think it's important to keep in mind that the fields of ecology and conservation biology have actually been using AI for decades, at least since the 1990s. Species distribution models, which are a really common tool in ecology and conservation, mapping where species occur.

 In relation to environmental predictors, has actually used machine learning again since the 1990s. And that, that's something that, When I took a step back, a few years ago with the advent of chat GPT, when AI really came into the spotlight, and for a lot of people, including myself, it was sort of like Oh, wow.

This new, maybe even scary tool that we have here. , but when I took a step back, I realized that, , AI has been a part of my field, at least for a very long time. , and again, it's, it's just another tool in the toolbox and, me personally, I've, used AI in various aspects throughout, my, scientific and professional career, including species distribution models.

And specifically right now in my position at CBI, um, I'm Like you mentioned, I am using AI to enhance biodiversity monitoring, which is a really important part of conservation, knowing where species are occurring, when, how many of them are there. That's key data behind things like species conservation assessments.

For example, you know, designating which species are threatened or endangered. For the endangered species act. , that's a really important field right now in biodiversity monitoring where we've seen in the past, five years or so, these really exciting new advancements and joining AI with, uh, this technology like camera traps and acoustic monitors.

and so just sort of getting into the details of how that works and how I've been using it. please 

Deep: do dig in there because that the acoustic monitoring just suddenly like triggered a bunch of Oh, yeah, I guess you could listen to crickets and I guess you could, you know, Yeah. 

Chris: So, I mean, I think one of the key points here is that there's, there's a lot of untapped data out there, whether it be images or sound, that we, as, manual data recorders as humans, can't really tap into very effectively, and AI coupled with, Camera technology and acoustic monitoring technology can really give us the ability to tap into those, vast reservoirs of data.

, with these camera traps, which is one thing that I've been thinking a lot about recently, we have the ability to go out there and set up these, cameras that are just constantly taking pictures of biodiversity. So, some of them are motion activated.

Some of them are just time lapse cameras that are, , taking pictures of, things at, , predetermined intervals and later on going through those images and counting what's in there. And that process of camera trap imaging. has been a part of conservation for a very long time, but historically it, required humans to then go through every single image out of thousands of images and manually, first of all, just determine if, there was something in the image at all, sometimes there's not, sometimes there's false triggers.

And in the case of time lapse imagery, you may only have 1 percent of all of those images that actually have an organism in it. , so that's part of it, and then after you, I'd determine if there's anything in that image to actually identify it, usually manually. And so what AI, , computer vision, technologies have, have given us the ability to do is automate that process and, increase its speed and efficiency by orders of magnitude.

So now what used to take humans, weeks to months, takes an AI algorithm, going through camera trap images, seconds to minutes. And so that's really, really expedited a lot of, progress and insights and all of that data is very useful for conservation. , we just 

Deep: double click on that for a second.

So maybe walk us through that in a little more detail. Like, where do you put the cameras? How do you determine where to put the cameras? Are you always looking in wildlands? Are you also looking in private farmlands? Once you put the cameras down, 

do you have a process for, like, labeling, the imagery? There must be like bounds on it looking for itty bitty tiny, tiny, tiny aphids or something, , like you have camera resolution challenges. I mean, looking for a grizzly bears a lot, probably easier than looking for, something super small.

Yeah, maybe just like walk us through a little bit. Like, how does the typical project roll out? Is it. Um, you know, like one species that you're trying to, like, develop across one kind of landscape. , and does that whole thing kind of generally go down? 

Chris: Yeah, that's a great question.

And I mean, Honestly, my typical project is not, , what most, conservation scientists typical project is because I focus on insects , and insects, we haven't really applied camera trapping technology to insects like we have for large mammals. And so , the out of the box.

Camera traps that you can buy, on Amazon, uh, they're not designed for her monitoring insects. They're not so good at taking this, up close type of, macro photography of insects. And then taking a step further, the recent advances in AI powered camera traps, whether it's, edge processing where these cameras are.

in real time identifying what's in these images and pinging that data up to satellite or even just post processing of those images using AI. We don't have those models as well established for insects, that we do for large mammals, for example. So there are, open source, Uh, , software and online, web applications that people can use to, automate the, detection and classification of mammals from camera trap images, but we don't have that same thing for insects.

And so what I've actually been working on recently is taking this technology that we have working very well for mammals now in part because of the advances that the AI has allowed here and trying to, tweak that to make it work for monitoring insects with camera traps and monitoring plant pollinator interactions, for example, and I think, Where I see a lot of opportunity, and I'm partly awaiting a lot of progress here because we have a lot of different groups around the world now that have identified this as a knowledge gap. But where I see a lot of opportunity is joining the, hardware with the software intelligently. And so we have, , a lot of software improvements that need to happen to enable insects and pollinator monitoring to be more effective. But we also, like I said, need the hardware to work better for that as well.

 We need the hardware and the software to interact well, 

Deep: and that's because you're trying to like push the cost of analysis of imagery down to the device. 

Chris: I mean, it is partly a cost thing. So right now, for example, a lot of the. Progress that's been made in automated monitoring of insects.

So we do have groups around the world that have developed these devices basically camera traps that, you know, you have insects come in, you take photos of them, and then you use usually a post processing AI. to, to weed through those images and detect and classify the insects.

The problem right now is a lot of those units are extremely expensive. What I would like to see happen is that we make insect monitoring with camera traps as. Cheap and efficient as it is with monitoring large mammals where you have these, these camera traps that are not too expensive that you can buy to monitor mammals.

 Even those equipped with AI, uh, we do, we do not have that for insects right now. So 

Deep: I guess I'm trying to understand the specific challenges with respect to insects, because it seems like The hardware question is sort of orthogonal, like you get, you can grab the image, throw it up in the cloud and deal with it, you know, on whatever you can get off of one of the cloud providers, whatever GPUs you've got, but like the imagery itself.

 All I can think of is like, resolution and frame rate seem like the two things that would matter more. Is that what you mean by hardware? , I don't know, bees move faster, , certain insects just like, move a lot faster and might be really tiny, like a pixel or something in a large space.

Chris: Exactly. So yeah, I mean, that's the major issue right now is with the camera hardware , we do have cameras like if you use your smartphone to go take an up close photograph of an insect, it does pretty good job, and you can put that photo on iNaturalist and their AI.

Computer vision algorithms does a pretty good job at, identifying those insects. But these sort of out of the box camera traps that we have right now are not really good at doing that, for insects because of those issues that you mentioned, the resolution, 

Deep: lower, lower quality cameras, 

Chris: lower quality images, and also related to insect behavior.

 and the fact that, you know, when you're taking up close videos or images of tiny little insects. On flowers, for example, any sort of natural movement in the environment, like the swaying of those plants in the wind is going to be, you know, magnified, uh, in that up close image. So there's a lot of these sorts of like technological logistical challenges to overcome to make this work as well for insects as we have it working for large mammals.

Deep: Oh, I see. Yeah. And are these, conservation biologists that are purchasing these traps, are they, like, is all this stuff open sourced, this hardware, , or is it there's, there's enough of a market that people are producing them at scale that the costs are low or is that not the case?

And they're like props, just kind of having their grad students tinker with hardware and make their own traps. Like what, what's the state of it today? 

Chris: Yeah. No, I think it's sort of runs a gamut there. Like you, you do have some organizations that are. at least attempting to, build out like commercial scale models of these things, , and, and producing them at commercial scales.

 But then, a lot of the effort has been by, you know, university researchers who are making things open source for the most part. So publishing their methods. A lot of it's sort of like DIY and that's, those are the things that I've played around with the most is like, we have phone cameras and like we can do a lot of this stuff just by rigging it up ourselves in our backyards and, , I find that stuff fun.

So I've done that a lot on the side, I think a big goal right now is to make this sort of automated monitoring of insects scalable. And I think part of what that requires is making it more affordable if there are commercial versions of these things, to allow, for example, a scientist like myself who wants hundreds of units to deploy out into the field for, for a field study, um, rather than just one or two.

Deep: , it seems like If you're designing, billing and selling solely to scientists, the scale of the market is just not big enough, and the cash supply is not there, to drive the prices super low, but if you can, it seems like private farms, have, my guess, have an incentive to figure out whether insects there for a different purpose maybe, but if you can leverage that to drive the cost , and scale, then you could take the resulting hardware, and then scientists could benefit from it too.

Chris: Yeah, absolutely. And actually, a lot of the work that's been done around this is an agricultural pest control perspective. Yeah. I know several companies that sell these units that are specifically designed to automate the monitoring of pest species in agricultural areas.

They probably work great. But again, for like a researcher like myself, I can't buy hundreds of those units because there are thousands of dollars a piece. Most of 

Deep: their costs haven't scaled down yet. 

Chris: , not that I've seen. And another part of the scalability is, the fact that a lot of these units that require for insects sometimes specialized attractants and viewing surfaces and, you're not, now you're not talking about a little camera trap.

You're talking about a huge box basically that you have to transport. Across maybe wild lands or agricultural fields or wherever you're deploying them. So again, I think for insects, we as a field have been focusing more on like, what's possible, to maximize like, I want to detect the most number of species as efficiently as possible, and that sort of side of things and not so much on the scalability, making this more, cheap, smaller, more, transportable, able to run in the field on batteries or solar power.

Yeah, you 

Deep: got, you got power issues. You might have satellite comms or cellular service comms. You got all the problems that run with running a gidget. Out in the middle of nowhere. 

Chris: Yeah. And so what, in my opinion, at least if we want to see sort of this envisioned future where we, we have a lot of automated biodiversity monitoring occurring across large areas, which requires a lot of units, these monitoring units, whether they're countertops or acoustic monitors, we need , to allow this technology to be scalable, and we're there with With, camera traps for larger organisms were there with, acoustic monitors.

 We have acoustic monitors now, that costs like 50 apiece. And so you can really, buy a lot of those and deploy them at very large scales and collect a ton of biodiversity data, very efficiently. 

Deep: Is there anything going on in the crowdsourcing arena? Like, people just installing an app on their phone and then just wandering around the wilds and audio gets sampled, captured, archived, you got a GPS coordinate, it seems like a kind of nice app for it.

I mean, you'd have to clean up the shuffled pockets and, other peripheral noise, but that seems achievable. Is anything like that going on? 

Chris: I mean, I think essentially what you're talking about is, you know, community science right now. Yeah. I'm 

Deep: talking about like you got 7 billion sensors in people's pockets and we, use them.

Chris: Yeah. So, I mean, absolutely. In my research and, increasingly in more ecology and conservation related research, we are using, community science, citizen science data. And that can be images from iNaturalist. iNaturalist is probably the most widely used, citizen science platform, at least in the U.

S. Um, and it's mostly image based, They do allow, audio recordings as well. You're getting millions and millions of images, each one of those images being a point of data, a species at a place in time. And that is the type of data that we as ecologists, as conservation biologists rely on, , actually to me, Probably one of the most important and exciting innovations in my field in ecology and conservation biology of the past decade is the marriage of community science with artificial intelligence.

and that's because on one side with community science, you have the ability to generate orders of magnitude more data than we ever could as scientists, as professional scientists. And then on the other side with AI, you have the ability to rapidly make sense of that data, including things like computer vision models that are able to, with pretty high accuracy now, classify that image, um, at least to a broad group of species.

And amazingly, increasingly to species level, including for pretty cryptic groups and unknown groups of organisms like moths, which I spend a lot of time on. And so it's, a really powerful, cross pollination of technology that we've seen here, plus 

Deep: you get free analysis.

If you can do it on device, right? I mean, like analyzing millions, billions or trillions of images or audio samples is expensive, especially if the money's going to Amazon or Google and their cloud. But if you're doing it on device, then somebody else is paying for the electricity. A citizen's paying for the, compute cycles and you just have to pay for storage, which is of the results.

I probably you want to keep on the originals too, but.

Chris: Yeah. , I, I don't really understand as much as I should about that side of things, compute power and who's paying for what. And I really hope that it's not like falling on, underprivileged communities, those sorts of costs.

But yeah, I mean, that's a good point is that, we're sort of sharing the cost of generating this biodiversity data and storing it and and not just relying on, conservation nonprofits like the one that I work for that often don't have the sort of money.

To buy a huge data Well, that's what i'm 

Deep: wondering right like what you're describing is sort of a global effort to have maybe near real time monitoring of insect populations Someone's got to pay for that. How how does that stuff get funded today? Are we talking about? scientific grant money kind of and conservation grant money and environmental grant money Or is there some private market where it's super valuable to actually have this data?

I don't know like hedge funds come to mind hedge funds that are doing commodities trading might Want to know when a band of locusts is about to destroy the entire crop from kenya or something 

Chris: to be honest, I don't know much about that. But I do know that there are people biodiversity, conservation, biology, nonprofits or organizations, that have pretty much developed their entire model, around collecting and storing that type of data.

Presumably, eventually to sell or to use in some other way. so I, I think that, a lot of people are thinking about that. Like, that's a big issue in general with biodiversity data is like data sovereignty and like, who owns this data? Does anyone own this data, biodiversity data in general, and then when and where it's collected, does that influence who should own, decision making power over it.

, in my perspective, like I get kind of frustrated over, when people are unwilling to share their data, at least from certain sources. And I know that data sovereignty, again, is a major issue in there. There are communities out there that won't share their data for very good reasons, but I also don't think that the privatization of this data, is a very good solution either.

Deep: what you're describing feels to me like a fractured landscape of private or public entities. Thank you. Gathering some data, probably research funded, probably on particular plots of land or locales. , but what you're trying to get to is a more kind of global, holistic kind of view, something akin to maybe what NOAA does with, atmospheric sensing and, ocean data.

 It feels like the realm of a really well funded big. Government entity to like step in. And then I guess the 2nd, the 2nd part of the question is like, let's say we get all this data. Does that just mean that we have a really great television to watch the demise of, the pollinator population or.

Does something actually happen with it? And do we do something with it and stop the decline?

Chris: For the first part of your question, I think that like in general, in a lot of different scientific fields, including in ecology and conservation, there's been a general movement towards sort of open data.

, and this idea that, probably because of the severity of the global issues that we're facing today. We really need to foster collaboration. And part of that is sharing data. And that's because any data can be extremely valuable. , and we in biodiversity science, we have a major data problem.

We did not have a lot of data. , for the vast majority of the species that are out there. And it becomes especially a problem when you talk about things like species interactions. I think only about one in ten species on earth potentially have been described even.

Less than 10 percent of all of Earth's biodiversity have even been, has even been discovered or described. , and then for the vast majority of those species even, we know very, very little about, their biology, their ecological interactions, their traits, all of this different stuff that, that we, from an applied perspective, getting into your second part of the question here.

We need this data to. to feed into, conservation decision making, knowing in my sort of niche field of research, focusing on plant pollinator interactions, for which we have very, very little data for overall. , we can use that data and we need that data for a lot of important potential conservation applications.

So knowing, for example, which plants to plant where, based on the patterns of usage of those plants by insects. That is critical ecological information, and actually basic ecological information, that That we simply just don't have. And so from my perspective, when, especially as a scientist, in the scientific community, whether it's in academia or nonprofit or government, when you collect that data to not make it open source is, potentially disadvantaging, everyone, including yourself.

, because that is data that could be very, very valuable to conservation efforts. And so there's a lot of efforts right now, including ones that I'm involved in to, first of all, synthesize this type of ecological data, because even when people do share it, it's usually not in a very accessible format.

It's spread out around a lot of different sources. And so, a major task, in my field is to, consolidate that information in one place where we as a scientific community can access it and use it for various applications, whether it's basic or applied research. We've made a lot of progress on that for things like, Species distribution data.

So where species are occurring. We have, massive global data sets and databases for this now. And I'm in my work. Working towards that same situation for other ecological data, like species interactions.

Deep: So, given what you have looked at and what you can see, what is the analysis of the data told you?

Like, what are some high level insights that maybe policy decisions could, shake out of? Or just collective actions? 

Chris: Yeah, and yeah, so I forgot to answer the second part of your question, which is like how do we make sense of this data? Is it even useful or is it just telling us all the same story that, things are bad?

And yeah, I mean unfortunately like that is a big part of it, the fact is that Species are declining, and at an ever accelerating rate. So what that means is for the most part, when we collect data out there on species, population sizes, and, assess their conservation status, it's rarely in a positive direction, for most species.

 There are exceptions there. And actually I want to. point out, the efforts of the conservation community really do and have made a difference. There was a recent paper that came out, a meta analysis showing that, , globally conservation efforts make a difference. They have led to positive impact.

And so what that's, tells me and should tell other people is that there is absolutely hope. We can, turn the tide here on things like climate change and biodiversity loss. It requires more cooperation. Again, like I mentioned among scientists and policymakers and all these different entities, it requires more resources being poured into this.

It requires more public understanding and awareness of these issues as well. And so I think for me as a scientist working in conservation, um, sometimes it can be, very, you know, pessimistic, the situation that we find ourselves in. But I like to focus on the hopeful things and that is that, again, our efforts are making a difference and it's not just our efforts as, , professional conservation.

Can 

Deep: you give us like some specific examples? If you want to maybe talk about the Butterfly net, does that make a difference if you have something else that you think are specific cases? I probably know 40 people that, went out and made beehives, , and are like harvesting honey at some point.

Like, does any of that matter? Where are the pockets of positivity? 

Chris: For me, a big pocket of positivity is, what anyone can do in their own backyards and gardens. So a lot of what the butterfly net, , was aimed at is helping people home gardeners and landscapers make better planting decisions to, support biodiversity.

So we, we choose plants in our backyards and gardens for a lot of different reasons. Mostly it's because they're pretty or maybe because that was what was available at a local nursery. But increasingly, we need to be making planting decisions with biodiversity in mind because, as I mentioned earlier, insects and a lot of other wildlife rely on plants and usually native plants, plants that are, have evolved and are native to those specific regions.

So the butterfly net was really, um, geared for And just for our 

Deep: audience's benefit, this is a tool that you built? 

Chris: Yeah, the butterfly net is an app. Maybe 

Deep: just describe it really quick, yeah. 

Chris: Yeah, so the butterfly net is an app that I built during my Ph. D. And it's essentially a plant selection tool. So it helps people, whether they're home gardeners or even restoration professionals, make, better plant selections to support insects.

And so it, it specifically guides people towards what we call Keystone native plant species. And those are. Plants that, are disproportionately important to the insect community there. So maybe those plants that support the highest number of insects. And it does this in a spatially explicit approach.

So it, It looks at where species are occurring and who they're interacting with where. And it, uses that information to identify these regionally appropriate native plant species. And so. 

Deep: Unpack that. How does it do that? How does it know where who's interacting with what? I mean, are you using the camera?

Like what? What are you doing there? 

Chris: Yeah. So we, well, I say we, because this project is now blossomed into, , a working group that I'm leading through the National Center for Ecological Analysis and Synthesis. But we are basically, looking at, , those types of data sources that I've been mentioning hearing.

Okay. including from citizen science, where people are documenting, which species are occurring where. And we're using that information to develop, species distribution models. We're also looking at temporal patterns of activities. So when are the insects occurring? It's active and when are the plants blooming and we're using all of that information together to identify the best plant species, if you will.

 So those plant species that are optimizing the benefit for the insect community. Was that 

Deep: based on trap, trap imagery or based on the apps, like a phone, I'm envisioning a phone app imagery or something else. 

Chris: Yeah, so most of the. The data that we're using actually is coming from iNaturalist and this is mostly from people's phone cameras taking photos and iNaturalist imagery has, just contributed exponentially to the amount of data available to produce things like species distribution models.

So if our 

Deep: listeners want to do something here to help out, they should download the iNaturalist app and go shoot pictures of. 

Chris: Oh, absolutely. That's, that's to me, the number one way that anyone can contribute to conservation is by using community science platforms. iNaturalist is just the largest one, but there's others as well, including PlantNet and eBird.

 But that type of data is being used by, conservation scientists, including myself, and it's, it's super, super valuable data. It's also been assisted by AI. So part of the reason we have all of this data and all the people interested in using community science platforms is because the app will tell you right there what you're looking at because of their computer vision model.

Deep: Once you figure out, what plants for a particular region need to be planted more there, is there room for, you know, legislation or some kind of action to, like, push the nurseries to, like, include a lot more of those plants? Maybe that's just going out and talking to them.

 Maybe it's more direct where the government just gives them more money so they can drop the prices of native species or something. Is any of that happening? Should it happen? 

Chris: It absolutely needs to happen. In fact, one of the main reasons that people do not plant native plants is because they're simply not available like ornamental, often invasive plant species are that you can buy, for example, at Home Depot and other big box stores.

So for the most part, native plant gardening, supply is restricted to these sort of smaller nurseries, and online seed sellers. And there's, number one, there's not a lot of them. And number two, it's sometimes hard to find them and there's no really like centralized framework of how to access that information of where do I buy this native plant that I'm supposed to be planting here.

So that is a major thing that. I and others have been thinking about, and it's, a really essential component of this wider picture of, how do we get everyone contributing to restoration that helps biodiversity, and again, a big part is just having the supply there, having the right plant species available for people to actually implement them in their gardens.

Deep: Huh. so that kind of like urban suburban nursery driven kind of intelligence driven strategic planting actually makes a difference at the on the numbers to the point where it would show up in your data and it matters to you. 

Chris: If you're talking about from , does this have an impact on species populations in terms of like, 

Deep: yeah, pollinators, , does that absolutely, 

Chris: I mean, a great example is the monarch butterfly.

I, I try not to talk about the monarch butterfly as much because it's a great example of our bias towards individual species. As humans, but also in conservation, but the monarch really is a great example to highlight these ideas. A lot of the, uh, success around, helping to support monarch populations is attributed to native plant gardening specifically.

Milkweed planting efforts in private land. So in people's backyards and gardens, and there, there have been numerous studies showing that type of even small scale restoration of, really crucial Keystone plant species like milkweed really can have a positive impact on species populations, , and, Help their potential conservation status.

 And you know, the fact is that in order to address the problem that we have right now, we need everyone doing this. we need to get rid of, lawns. Lawns are probably the worst things for biodiversity right now. 

Deep: and it's not just the grass, it's also the fact that once you have a lawn, people get obsessed with getting rid of the weed.

So they start spraying stuff all over it. What's bigger, that physical fact that you got the lawn, the fact that it displaced native species, or the fact that you're slamming it with chemicals? 

Chris: Oh, it's all of those things. Absolutely. So, I mean, first of all, when you have a turfgrass lawn, it's one species of often non native grass.

And what that means from a biodiversity perspective, As if you envision, each plant species supports a given number of native insects. When you have one plant there, you're maybe supporting a few. , if you have two, you're supporting a little bit more. If you have 10. 50, 100, you're supporting more and more and more.

So the more we sort of downscale the plant diversity that we put in our. manage outdoor spaces like our yards and gardens, the less insects we are supporting. And then the second part of that is absolutely the, the amount of chemicals that, the average person is using in their yard and garden to control the weeds, to get rid of insects.

, and that's, , a major problem. Part of like wildlife gardening and gardening for pollinators. is number one, increasing native plant diversity, but also number two, , decreasing chemical usage, also getting okay with the fact that we share this planet with other organisms, including insects, , and that for the, the vast majority of those insects are completely harmless to us.

And actually, um, Uh, are very, very beneficial to us. And so we don't want to get rid of them. We want to actually help support them. 

Deep: And what about , is there any efforts to like, we have a ton of public lands here and we're in Washington state. I think you're in Oregon. But, anywhere, Rockies or West has just a ton of public lands.

 I'm imagining there's a lot of native species that have just reduced due to the last century of logging and other activities. Like, are there efforts to, leverage your kind of really insect focused approach to pulling those numbers back through the planting of milkweed or other specific species so that it's not just. Urban and maybe suburban landscapes, but it's this like just vast tracts of land that we have. 

Chris: Yeah, absolutely. I mean, actually, I would say that, major hurdle that we have right now is making our urban and agricultural landscapes more friendly to biodiversity because that's where we see most of the degradation occurring.

 Not to say that there's not issues in, , wild lands, protected lands as well. And I think the same general idea that we need to be restoring ecosystems, and managing ecosystems in very targeted ways that, use the available data. Including the data that we're generating with AI approaches , to sort of guide those plant species selections and those restoration strategies.

And that, that is very relevant to agriculture. It's very relevant to urban areas, but it's, you know, equally relevant to the restoration that we're doing in protected lands as well. 

Deep: That's interesting. So I'm reading into what I heard you say and correct me if I'm wrong, was That the insect populations that are relatively better in those, public lands, whether it's DNR, I mean, obviously a national park would probably be fine, but like, there's all this other lands that are wide open spaces, but maybe, they're not as heavily impacted as the private and suburban.

I would think they would also be impacted just with, all the ranching and, like, logging in our area in the Northwest. Plus with the fires and, there's just a lot going on. 

Chris: I mean, it's very, very context dependent.

So it, it depends on where geographically you're talking about. It also depends on, what insect groups or what insect species you're, talking about there. , there are some, uh, disturbance loving insect species, a lot of agricultural pests. Uh, which will have, very high abundance in agricultural areas.

Does that mean that that agricultural area, is supporting, healthy natural community of insects? Probably not. Um, I would say that for the most part, protected land, is an essential component. To protecting, you know, biodiversity in general, including insects. Some of those lands are not managed very well.

 They could be managed better, but I would say in general, the fact is that we are increasing our rate of agricultural conversion of, urbanization. And along with that, that means that we need to decrease the impact of those land use changes on biodiversity and on the planet.

And that's where, at the Conservation Biology Institute, I've been focusing most of my efforts because I really see that as one of the most important challenges that we face today. Agriculture is actually the number one contributor to biodiversity loss. If you look at the numbers in terms of, you know, habitat destruction, what is causing most of habitat destruction, that's agricultural conversion, and so trying to make agriculture more environmentally friendly, including reducing its impact on insect populations and other, organisms is very, very essential for, our shared future on this planet.

We as humans need the food, obviously, but we also rely on those organisms that we share this planet with, those native insects and wildlife. I, I'm excited in all the opportunities in this area and making agriculture more sustainable. 

Deep: thanks so much for coming on the show.

It's been awesome. I'll end with, one final question. Um, maybe let's fast forward 10 years out, imagine all of the AI efforts, you know, helping and whatever else. Give us two scenarios, one where we're all screwed and everything's hell, and the other one where something actual positive happens and it's not as bad, how can the, the AI and other technologies like if they play out really well, help us out and how can they not matter and us fail as humans to like do the right thing and how bad does that world look?

 My biggest concerns around AI, and I really don't have a lot of concerns around AI. Again, I view AI as another tool, in the toolbox. And I, I think it all depends on how we as humans decide to use that tool. And hopefully we decide to use it for the better. Some of my concerns around AI, specifically in the context of conservation, uh, are perpetuating these sort of entrenched biases that we have, both, you know, social biases and the data that we're using to train these models, um, a lot of the biodiversity data that we have is from North America and Europe and China.

Chris: Um, which is paradoxical because most of Earth's biodiversity is housed in the global South. And so we need more data collection efforts in the global South and tropical countries. And we need to make sure that , our use of AI is not perpetuating these sorts of biases and that the potential benefits of AI are being shared.

 Among, all of these diverse communities. I would say my real dystopian future view, which I don't believe in because I think we will get our act together.

 Is that, as E. O. Wilson said, the collapse of insects will be the collapse of humanity. We cannot survive on this planet without insects. , if we don't do something to curb their decline, we're gonna lose out in the end as well. My optimistic future on the other side of that, a very common, , perspective in ecology is the idea that 

the whole is greater than the sum of its parts. you know, That's a really important idea in ecosystem sciences where we know that each individual species is playing an essential role and together. Including through the interactions that we all have with other species on the planet, we form, what is the biosphere, all of life on the planet, a, , self sustaining system.

We need to perpetuate that model, including how we deal with AI. So I really see that, if we use AI intelligently, it will be another essential entity in those parts that form a greater whole. And I really think that, the combination of. Human ingenuity, the capabilities of AI and the, intrinsic value that biodiversity has on this planet, as well as all of the services that biodiversity supplies, can lead to a very, very exciting, future where we can all thrive.

Deep: Awesome. Thanks so much. And thanks for coming on. . 

Chris: Thank you so much for having me on. It's, it's been a pleasure.