Episode Transcript
[00:00:00] HOST: Welcome back to our podcast for the Wild Ones. I'm your host, Josh Adler. This podcast is brought to you by the International Wildlife Coexistence Network.
[00:00:10] HOST: Thanks for joining us.
[00:00:37] HOST: On this episode. Well explore the fascinating world of bioacoustics with our special guest, Carly Batiste from rainforest Connection. Carly shares her journey from aspiring veterinarian to a leading researcher in the field of bioacoustics and primatology. We delve into her fieldwork in Madagascar, the importance of vocal communication in animal behavior, and how bioacoustic monitoring is revolutionizing wildlife conservation.
[00:01:24] GUEST: The animal thread was always there. I thought I wanted to be a veterinarian and went to undergrad to do that, but then realized I kind of didn't really like the medicine part of it. I did an internship at a wildlife rehabilitation center in Belize in Undergrad, and one of the taxa that they had that were often rescued from the illegal wildlife trade were spider monkeys and howler monkeys.
And that kind of got me hooked on primates.
So I ended up pursuing a masters in primatology to continue that thread and did my master's research on lemurs in a captive setting at the Duke Lemur center, but then wanted to get experience studying them in the wild. So I did my PhD at the City University of New York with a lab that had experienced, my advisor had done a lot of fieldwork for lemur research. The research itself shifted a bit from my master's to PhD. My master's was more on basic science, social learning and cognition, and communication. I was looking at vocal communication a bit for my PhD, and then kind of found this more applied version of vocal communication work, which is passive acoustic monitoring and using acoustics in a much more applied sense to survey animals. So that was kind of the thread. And so my PhD dissertation work focused on that. And I was within a biological anthropology programs, which I think was really helpful in giving me a diverse way of looking at things beyond just straight ecological or scientific lenses. So did my PhD doing field work in the southeast of Madagascar and using acoustic monitoring to survey a particular species of lemur, the black and white ruffed lemur, which is critically endangered in a forest corridor thats very fragmented. And we didnt really have a good grasp on where the lemurs existed within these fragments. So I put up a bunch of recorders throughout all the fragments and got to see how their range and their distribution was influenced. Sneak peek of research thats currently in review now is that we found this kind of abrupt cutoff halfway through the corridor, a very narrow north south corridor, and halfway through, there's an abrupt cutoff, and they're not found anywhere south of that cutoff point, which also coincides to this thick band of deforestation that almost cuts the corridor in two. So we think that that's preventing them from crossing and going south, potentially, yeah. So that was my foray into the passive acoustic monitoring world and how we can use acoustics in conservation.
And now I'm lucky enough to have continued that work from my dissertation. In the real world and outside of academia, I now work at the nonprofit Rainforest Connection, which develops the RbMon platform, which is a free, no code acoustic analysis platform that I myself used for my PhD before I was even working here. And now get to be on the other side supporting folks who are doing these types of acoustic monitoring projects around the world.
[00:05:23] HOST: Thats a really rich background. And its exciting how youve crossed over now into supporting the bioacoustics monitoring world, which is very much emergent, but also has a lot of field testing and has been field proven in terms of the value that it brings for biodiversity monitoring for a lot of different species.
In what you were saying about your background, you mentioned that your social lens helped you, and I was wondering if you could just go a little bit more into how that social lens in your studies helped you see things differently and maybe arrive at different conclusions or understanding as well as just could you just mention, what were the habitat pressures on the lemurs that you were studying? Why was that belt of deforestation breaking.
[00:06:11] GUEST: Up their corridor in Madagascar, and particularly the area that I was in? I was working closely with particular ethnic groups, the tanala and the Betzaleo people. And Tanala in Malagasy literally means of the forest. So that group of people have been living in and around the forest ecosystem there since humans first got to Madagascar a couple thousand years ago. And so they have a rich, intimate knowledge of the forest and its inhabitants and were absolutely critical to the work that I was doing because there hadnt really been a lot of research there, let alone by non malagasy people. And so I was really lucky that they kind of welcomed me and the team in and we got to talk to them and understand what the situation was and how they felt. And it really is just kind of an issue of needing sustainable development and integrated community and conservation work because they are logging the forest for firewood, for building materials. Charcoal is really big to support their families. There's a lot of splash and burn agriculture to plant rice paddies. Rice is the cornerstone diet in Madagascar. And so it's a lot of subsistence farming. A lot of people have zabu, which are a cattle species there, that also clear land for to grow crops for the zebu as well, or have the zebu graze in the forests. So theres a lot of different pressures. Theres a lot of mining that goes on too, which ranges from quite small scale to kind of full industrial scale. So mining is also an issue. And theres various legalities of where this can, should or shouldnt be allowed because theres a suite of protected areas, but its difficult to enforce because the parks department there is, as with many environmental agencies and park services, chronically underfunded and understaffed. So there's just not a lot of investment in the malagasy conservation space and sustainable development space. I think Madagascar gets forgotten about a bit compared to the mainland, particularly when we think of rainforests in Africa, you think of the Congo basin. But there are also rainforests in Madagascar, as well as a suite of other ecosystems that are incredibly unique, like the spiny forest in the south, and there's dry forests in the west. There's mangroves, there's grassland. There's this huge variety of ecosystems with one of the highest endemism rates in the world, plants. It's about 90% endemic animals ranges from 70% to 80% endemic. So it's an incredible place full of species, the majority of which are not found anywhere else. So it's really kind of, how can we diversify income streams and try to increase educational awareness? When you talk about it, they know that it's a problem. They talk about how there's a lot more erosion, there's a lot more landslide now that fill in all their rice paddies, because the root structure is usually what's keeping soil erosion from taking place. They know that they're not getting great crop yields because the soil doesn't have high nutrient contents anymore. They know that the water isn't clean. They're aware of all of these things, but it's a short term mentality of getting through today and the next day. And those are all kind of more long term problems. So it's a difficult situation and balance that we're seeing across the conservation world in many spaces of really the way that you do conservation well is to uplift and empower the communities around and work on alternative income streams, better agroforestry practices or regenerative agriculture practices, alternative protein sources. There are these bugs called secundry, which taste like bacon like eerily similar. And it's actually, there's projects now that are like farming secundry as an alternative protein source so that folks aren't going and hunting bush meat. So those kinds of things that are very community focused that then have those downstream effects on biodiversity in the ecosystem is really what needs to happen.
[00:11:26] HOST: It sounds like some of those alternative sustainable practices and business models are gaining traction there with the malassian Madagascar region. Of course, the scale, as you said, in terms of funding, is always below the threshold of generating the impact thats really required. Im wondering about the endemic species rates that you mentioned. Why is that important and how is it assessed, if you have that information?
[00:11:56] GUEST: Yeah. So if a species is endemic, it means it's only found in that particular area. It can be endemic to a country like Madagascar, or it can be endemic to a particular region, because Madagascar is an island and kind of following with island theory and what we see in other islands around the world, islands tend to have high rates of endemism and have lots of species that are only found in that island and are not found on the nearest mainland or neighboring islands. Once that landmass historically has shifted away, the species have become isolated, and thats when natural selection and evolution really runs rampant and created species that we don't see anywhere else because they're evolving. So we can assess whether a species is endemic or not by seeing if it is only found on the island and not found on neighboring islands or the mainland. That would make it kind of endemic to the island or at a more local scale. You can also say that something is endemic to a particular province or habitat. So there are species of lemurs that are endemic, in fact, at really a local scale, where there's only a couple hundred left and they're endemic to one reserve. So that's what we think of in terms of endemic.
[00:13:29] HOST: If you had to guess in terms of the number of unknown species or undiscovered species in Madagascar, what percentage do you think it would be approximately?
[00:13:41] GUEST: If you consider invertebrates like insects, even plants? I mean, it's a small percentage.
The deforestation rate in Madagascar is so high that we don't even know what we're losing. For all we know, we could have already lost species before we even found them.
[00:14:02] HOST: You're saying it's a small percentage that we actually have identified and do know about. Yeah, it's a huge percentage, potentially, that is there to be discovered or unknown or may never be known.
[00:14:13] GUEST: Yes, for sure. Particularly with plants and invertebrates like insects, where there's huge species rich groups, Madagascar is a generally lower species richness or like number of species than other tropical forests, like the Congo basin or southeast Asia or the Amazon, which there's some theories behind that. But overall, it tends to be a lower kind of species richness compared to other tropical areas. But within Madagascar itself, the tropical rainforests are where we have the highest biodiversity levels in the country.
[00:14:52] HOST: Now, I want to turn towards bioacoustics, particularly in the context of what we're discussing about identifying species richness or complexity of an area. What is bioacoustics tech for those folks who are not up on it, and how does it work?
[00:15:09] GUEST: So, the field of bioacoustics is the study of biological sound. And a large part of that is the vocalization that animals make. A lot of animals use vocalizations as an important communication mode. And there are calls that are species specifics. A jaguar has a different call than a lion. The ruff lemurs have a different call than the shafakas. So we're able to identify species by their calls. And in some cases, even go a step lower than that and identify the sex of that animal that's calling. Or even down to the individual in some cases. But largely, we use acoustic monitoring to monitor species. So we're able to take specialized sound recorders, microphones like you would see in a movie set, and recorders, but that are kind of specialized for field use. So they're quite small, they're getting very cheap, very power efficient, and we're able to them out into the habitat of interest, whether it be a forest, a grassland, even below the water. There are hydrophones which can detect sound in marine or robotic spaces. And where the passive comes in from, passive acoustic monitoring is that the recorders are running on their own. So you set them up with a particular parameter and a schedule. Maybe if you only want to study back, you just have them run at night or things like that that you can adjust. And you put them up and they can run for weeks, even months, sometimes on their own. So it's a passive way of surveying. You just need to go in and put them up and then get the data on the SD card that it has inside. That's how the data is saved to it. It's less invasive than needing to have people on the ground all the time. It covers a fairly decent spatial scale. Because microphones are omnidirectional, they can take in sound from 360 degrees. So you just have to put a recorder regular eye level, and you're still able to record the animals that are calling way up at the tops of the trees and in the canopies. So it's pretty good spatial and temporal scale coverage. They last for a long time. They take in sound from a decent area.
It's passive. And basically, those recorders just record every sound that emanates in that landscape, which doesn't just have to be animal sounds, right? It will also record if there's thunder and there's a storm or wind or rain. It'll record potentially human activity, like chainsaws or vehicles, boat noise. Sometimes we'll get anthropogenic adjacent activity, like dog barking or chickens. So you get this entire soundscape, which is kind of what we refer to as all of these sounds emanating from a landscape. And so we're able to then extract insights from those soundscapes in those recordings to see what animals are actually there.
[00:18:39] HOST: Sounds like a lot of data to sift through. How does that work effectively?
[00:18:43] GUEST: Luckily, artificial intelligence and machine learning have become a really powerful tool for doing this because the bottleneck used to be at the data collection phase. The recorders used to be really expensive, really big and bulky, and so it was difficult to get a lot of data in the first place. But now, with the advent of cheaper, smaller, more efficient recorders and hardware, that bottleneck has shifted to the analysis phase, because now we're able to put out 100 recorders that are all running continuously. And so you're ending up with hundreds of thousands of minutes of recordings that would take literal lifetimes to get through manually. But we've taken advantage of this big data revolution, and we now use machine learning tools to help us analyze this. How that works is there's a computer model that you input a bunch of labeled data. There still is a manual aspect to this, because for the model to learn how to identify these species from their calls, it needs examples. So you say, okay, this is a hauler monkey. This is this type of woodpecker. This is a tree frog. And you put those labels into the recording itself so that you have a training data set of labeled calls that are assigned to a species. So this is the call for this tree frog, this is the call for a jaguar. And those are the examples that then get fed into the computer model that uses those examples to learn and then is able to predict on new data what that species is based on the inputs. And then usually there's some sort of kind of post model verification process where a human comes back in and verifies the model's predictions so that we still know that we're getting good data, keeping a human in the loop because the models, for better or worse, aren't completely trustworthy, so we still want to check.
[00:21:03] HOST: Yeah, it's still early days and tricky times for our AI collaborators and colleagues, albeit I'm surprised we're still relying on SD cards at this point. We're not just uploading to Starlink every particular intervals and things like that. I'm sure it's coming eventually or soon. But for folks who are like, yeah, this tech is great, it's fancy, it sounds cool, but how is it helping biodiversity? How is it helping communities of people, really?
[00:21:30] GUEST: This whole methodology and the AI that kind of comes with it to expedite it, is usually just a way to get data faster for a particular research question. A lot of our projects, as with any conservation project, there's kind of an initial question that we're assessing, and it could be as simple as what species are in this reserve, or it could be much more specific, like where is this critically endangered species found in this larger landscape? And it can be used for kind of quasi experimental studies. We do a lot of work at restoration areas and monitoring how well restoration areas support species coming back to them that have left the area. And so you can put recorders in the restoration sites and then a good and a bad baseline, if you will, where the good baseline is a nearby primary forest, and the quote unquote bad baseline is a traditional cultivated farmland area. And then we can see what animals are in the different habitats. And ideally, you want your restoration site and the species richness to get further from the plantation or cultivated land and closer to the primary forest. And by having acoustic monitors in all of these places, we're able to detect what species are there and see how the restoration sites are doing beyond. Just, we planted this many trees. Okay, that's wonderful. But what's in the trees? Right. So this is just a tool with which we answer those questions of, is the restoration area supporting native species? What is the distribution of this endangered animal? What is the biodiversity baseline in this preserve? And we want to track it over time. So you do annual acoustic surveys or something like that. And because the acoustic monitoring generate static data, it lives as an audio archive. So the same audio can be used for different questions. Right. I was just focusing on this one lemur species for my PhD research, but I have all of this data that has birds and frogs and insects in it. So I now have collaborators who are using the same data set to look at seed dispersing birds. So you're able to reanalyze the data again and again, because it's this permanent record, so there's a myriad of different ways and use cases that acoustic monitoring can be used for, but it really just should be thought of as a tool in which to answer the question you have.
[00:24:32] HOST: And so what's an example of a case, an application that you found really exciting?
[00:24:39] GUEST: What I was thinking of when I was talking about the restoration use case was a project we did in the atlantic forest of Brazil, which is in the southeast part of Brazil. And it gets less attention than the Amazon, but it's in worse shape, it's more fragmented, there's higher land conversion rate. So we were working with IPE, which is a brazilian institute, and a number of other partners. A lot of these are quite collaborative to monitor how well restoration sites were supporting native species, because they're particularly restoring areas in the hopes of making wildlife corridors between the main primary area and fragments that have been isolated. We found that the restoration site actually supported almost as many species as the primary forests, which were much higher than the farmland. So there was many more species, but the restoration sites didn't match the primary forest exactly. There were still some species that are missing. And so, just goes to show, right, restoration is a long term process and that this particular project seems to be heading in the right direction, but to kind of continue that long term monitoring, to see how long it takes to fully resemble the species assemblage of a primary, undisturbed area there.
[00:26:11] HOST: It is always exciting to me when I hear that restoration projects work, maybe not completely 100%, but are better than not conducting the restoration process. What about threat detection?
[00:26:22] GUEST: As I had mentioned, these recorders can pick up any sound in the environment, which includes human activities, that can be indicative of illegal logging or mining or poaching. So picking up sounds like gunshots or chainsaws, that might be indicative of poaching or logging. So we're also able to kind of detect potential threats which can be used to better understand where there are potential hotspots of illegal activity in an area. Based on how we're detecting those proxy sounds, we can also develop AI models that are trained to identify gunshots or chainsaws or vehicle noise. And so that's another kind of aspect of acoustic monitoring that can be used in conservation.
[00:27:17] HOST: Now, can that kind of acoustical identification lead to alerts and intervention?
[00:27:23] GUEST: Reinforced connection developed a real time system called the Guardian, which is a type of recorder that is able to run models on the edge, meaning the models run on the device itself in real time. So as it's listening, it's actively processing all of the sounds that's coming in. And when it hears whatever it's been trained to hear, usually chainsaws, it will send an alert via satellite or cell service, if it's available.
And so it can send an alert down to folks on the ground. Coming to your point of what you said, with uploading data to Starlink, what's probably more likely? Just because audio data is very large, difficult to package up sound files into a small enough package to be sent over a satellite. But what we can do is load the AI model onto the recorders themselves, so that the models are running in real time. And the only data that's being saved are potential detections. So it's onboard processing on the recorder itself, rather than processing afterwards on your computer. So that's a very active area of research that's happening today for both threat detection and for biodiversity monitoring, to enable more real time insights if needed.
[00:28:56] HOST: Are there any other areas in experimentation for applications that you're excited about?
[00:29:02] GUEST: I'm really excited to see how acoustics can be combined with other sensor data, as well as remote sensing data, like satellite imagery. There's some projects where we've combined acoustics with EDNA acoustics, with camera traps. And so I'm really excited about how we now bring all of those various different sensing technologies that we have in our conservation tech toolkit now and bring them together to get really comprehensive insights. Because obviously, with acoustics, not every animal makes noise, right? So you are missing out on some. But those animals might get detected via EDMA or via a camera trap. And then thinking about what the on ground sensing will tell you that this species is located in this site and this site, and we detected it at this time. And this time, by bringing in environmental variables and satellite imagery and geospatial layers, we're able to see. Okay, well, why is that species only found in this site? In the example of my PhD work with the lemurs, where I said that there was that big kind of band of deforestation, that was a cutoff point. It still begot the fact of why weren't they down in the south to begin with? And when I was able to incorporate some geospatial imagery, I was able to find that it was only in these kind of high quality forest sites that the lemurs were found in. So it added another kind of contextualizing layer to the findings that we had. So all of the ways that we combine all of this data that is being generated now is something I'm really excited about. And I think is going to be kind of the future moving forward.
Integrated insight and as I think we.
[00:31:06] HOST: Were talking earlier, there's still a lot.
[00:31:08] HOST: Of room to grow in terms of.
[00:31:09] HOST: That kind of comprehensive analytical and insight compositing, or even just the metrics that we're using. Maybe the best metrics haven't even emerged yet in terms of how we're assessing ecosystem health, functionality, integrity. It's very much an emergent field. Even though we have this ever growing technical toolkit, putting the pieces together is still this huge task. Also, like in what you just said about the lemurs preferring the high quality forest, you can't blame them for wanting a good neighborhood. In terms of help of bioacoustics and maybe some of these other sensory tools to monitor biodiversity and ecosystems, what do you feel can help anybody to shift their understanding or relationship, or connect more with nature?
[00:31:59] GUEST: I think that acoustics has really untapped potential in the art science interface space and how we think about our environments around us. Sound is very emotive, it's very evocative. And humans, a lot of times we tend to be very visual, right? And so I often recommend when people are going out into any space, close your eyes, shut down that primary sense, and open your ears and listen to what's around you, because you're tapping into a whole other kind of sensory world that I think we aren't instinctively doing. I really like how the kind of sound itself can relate us to particular places. And I think there's a lot of untapped potential in conservation awareness and education. We've had presentations where we've played this is what a healthy forest sounds like, and it's a very rich soundscape. You can hear lots of animals calling, and then this is what a palm oil plantation sounds like, and there's only a couple chirps. People hear the difference. They can hear, okay, there's lots of vocal activity in this one and not this one. So there's a lot that I think can be done with incorporating it into education campaigns, awareness campaigns, as well as just our own internal connection to the world around us.
[00:33:30] HOST: Is there a moment that you can think of in your own experience where you were acoustically shifted by the soundscape you were in?
[00:33:41] GUEST: I associate different sounds with different things. I grew up near the ocean on Long island in New York, and so the sound of waves always brings me to memories of home in Madagascar. It's funny because forests are quite noisy. We think of a noisy space as human noise sounds. But forests themselves are actually quite noisy. And so when I'd be camping in the woods in Madagascar, there's cicadas going, the frogs, and everything is quite noisy, especially at night, too. I love to hear what is noisy there and what people think of typically when they think of a noisy soundscape that's usually dominated by human sounds. But give me the forest noise any day of the week, and I sometimes play the recordings I have from Madagascar to myself while I'm working, just as background noise, because it brings me back to the good times that I had in the field there.
[00:34:40] HOST: I love the personal application of bioacoustical information as almost a therapy that you can bring into your life, into the built environment or the human environment, and populate it with the ambiance or the life of the forest.
[00:34:54] GUEST: Thanks, Josh.
[00:35:02] HOST: Thanks for tuning in to this investigation of the powerful role of sound in understanding and protecting our natural world.
As always, we've got lots more exciting and diverse stories from the world of wildlife conservation and coexistence coming up on for the wild ones. So subscribe now and keep listening.
