How rat poison endangers wildlife, and using sound to track animal populations

A
This podcast is supported by the Icon School of Medicine at Mount Sinai, the academic arm of the Mount Sinai Health System in New York City and one of America's leading research medical schools.

What are immunology researchers working on to transform patient care and prolong lives?

To find out, we invite you to read a special supplement to science magazine prepared by Icon Mount Sinai in partnership with Science. Just visit our website at www.science.org and search for FrontiersofMedical Research Immunology the Icahn School of Medicine at Mount Sinai we.

B
Find a way Japan's noster specializes in postbiotic gut microbiota metabolite based pharmaceuticals research to treat metabolic and immune related diseases. Noster's products include biosynthesized GMP bacteria preparations and Qmec, the world's first HYA 50 metabolite postbiotics healthcare supplement.

Their analytical services include liquid chromatography, mass spectroscopy, metabolome analysis, and state of the art genetic sequencing. For gut microbiota analysis, visit Noster, Inc. To discover how NLSTR can help you.

A
This is a science podcast for July 12, 2024. I'm Sarah Crespi. First up this week, producer Kevin McLean talks with freelance science writer Dena Fein Marin about the dangers of rat poisons making their way into the food web. Next, I speak with researcher Yep, have rasmussen about bringing bioacoustics and artificial intelligence together for ecological studies. We talk about how training and AI on animal sounds can help with tracking populations and their health and behavior.

C
Hi.

A few years ago, I visited a veterinary clinic in the Bay Area that had a bunch of sick parrots in San Francisco. There are these flocks of wild parrots that are descendants of pets that have been released. This vet clinic had managed to figure out that the parrots had been exposed to rodenticides, toxic chemicals meant to control pests like rats and mice.

Seeing these animals that were struggling to move, mostly unable to fly, brought up all kinds of questions for me at the time, like what are these chemicals? And how did the parrots find them? And I realized how little I actually know about the war humans have been waging against rodents for thousands of years. So when I saw that freelance writer Dena Feinmaren was writing about the long reach of rodenticides, I was definitely intrigued. She's here today to talk about the current state of rat poisons and what scientists are learning about how these chemicals make their way into the food web. Hi Dena. Welcome to the science podcast.

D
Hi. Great to be here.

C
So I know you mentioned in your story that humans have been trying to control rodents for millennia, basically. Did you come across in your research for this article some of the historical or maybe ancient ways that they've tried to manage rodents?

D
What's interesting is that rodents, it's a balancing act, right? They're important parts of our food web, but, of course, they carry a lot of diseases that can create public health threats, and they're a nuisance. They can eat crops. They can destroy our homes or aspects of our homes, rather. And so people have been trying to control them for a really, really long time, and that has happened in a lot of different ways. I was surprised to find that some of the most ancient art in Egypt has depictions of cats showing them facing down field rats. And people, of course, later on, they turn to various chemicals, one of which has been arsenic. But obviously, that has a lot of risk for humans as well. So scientists continue to try to find a better way to control rodents. In the 1950s, chemists perfected what's called anticoagulant rodenticides. As you might think from the word anticoagulant. It destroys their ability to clot blood, so the animals literally bleed to death. Those carried less risks for humans still, of course, risks, but less risky than arsenic. But then, evolution being what it is, rodents started to evolve resistance to many of these compounds over time. So scientists back in their labs came up with some other alternatives, and in the 1970s, they introduced what are called second generation anticoagulant rodenticides.

C
Are they just more potent, or do they act in the same way?

D
These are different chemicals that carry more lethality with them, meaning that animals would die after eating just a single dose. Although they don't die right away, the process can take up to a week, and during that time, the rodents will potentially continue to go back and eat more of that poison. But they're also easy pickings for animals that would eat them, like falcons, like hawks, bobcats, mountain lions, and so forth. The reason they're of greater concern for animals and the ecosystem is that they have a much longer half life in the animal. So it sticks around in the animal for a much, much longer time. And that means that for animals that eat them and maybe the animals that eat those animals, that raises questions for everyone's well being throughout that chain. Some studies have shown that they can stick around in certain organs and rats for almost a year.

C
Oh, wow.

D
Yeah. And that doesn't mean that all the animals will die. Right. But it does raise questions, even if they're not lethal. About how that might be affecting animal behavior and this question of bioaccumulation. And that's why I was really interested in tackling this story.

C
Yeah, I was really surprised at the wide variety of different types of species that they can get into. I mean, not just in a specific animal, but in the ecosystem as a whole. Can you talk a little bit more about that?

D
Yeah. I was really surprised to find, as I was diving into this story, that animals, amphibians, crustaceans, even insects, they've been detected in all of these animals. We had frogs. There were hedgehogs, for example. There's a study this year in Australia that looked at rodenticides and dead frogs, and they concluded, well, some animals likely ingest the poisons when they're eating contaminated prey, but contaminated soil could also pose a threat. And also rodent carcasses, unprotected bait, the feces of contaminated animals, all of these things could play a role as well.

C
Yeah. Once you get those chemicals into the system, they sort of make their way in a lot of different places, it sounds like.

D
Yeah, exactly.

B
So is there any way to treat animals once they've been exposed to any of these chemicals?

D
The answer is a bit complicated. What you want to do to help an animal gain its ability to clot blood again is you. And you rehydrate the animal because it's been sick and probably not getting as much water, but you also inject it with a dose of something called vitamin K, and that helps the body amp up its ability to clot. But unfortunately, because many of these animals are wildlife, they're not typically making it in time to a facility where they might receive help. One of the tough things with this particular exposure is it can look like so many other things. The symptoms for avian flu, for a bird of prey, are very similar to the symptoms of what you'd be seeing here with this blood clotting inability. Unfortunately, you're just kind of going to look sick. But what's interesting is one of the sort of back of the envelope ways of testing for this. They literally take a sample of blood, they put it in a test tube, they slowly shake the test tube back and forth, just turning it over, turning it over, turning it over, and see how long it takes for blood to clot. Now, if it clots in 510 minutes, that's pretty standard. That's good news. But if it's taking much longer than that, that increases the chances that indeed there's a problem with blood clotting. And that's when you'd want to turn to vitamin K to help.

C
I see. And so you said this is sort of like a field test kind of thing. Are there formal tests to detect all of these poisons as well?

D
There are not. That can be quickly done or inexpensively done when an animal dies. If there are funds provided, let's say, because of an ongoing research project to go and test a liver sample, you can indeed identify what materials were in the liver and say, aha, I'm a toxicologist, and I see that there is this poison, but that's not very useful, obviously, to the animal that was sick at the time. And that's one of the real problems here, is that it is really difficult to get these answers and get them quickly. One of the things that came up on was there was a study in PLOS ONe that published in 2021, and it looked at dead bald and golden eagles and had found, okay, more than 80% of more than 130 dead birds. They carried rodenticides, but then they were like, okay, so there's rodenticides inside them. And they could only definitively say that poisoning killed 4% of those animals. And one of the lead researchers in that work, wildlife pathologist Nicole Nemeth, she told me, okay, well, this might have led to a lot of animals being disoriented, and then they were hit by cars or had other accidents, and then when they show up at her lab, they're so bloody and messed up that it's a bit hard to make sense of exactly what happened and why. Scientists haven't yet said, aha. At this percentage, this is lethal, or if you have 100 ML, lethal, if you have two sick rodents, lethal. Obviously, scientists know a lot about that. This is hurting animals and killing animals, but they don't have that kind of granularity, and that can make it difficult sometimes to have policy interventions.

C
Right. Yeah. I mean, are there any regulations in terms of, like, who can buy and use these types of rodenticides? Is it like, if I just go to the store or go online, can I get all of these? What's sort of the state of that right now?

D
A lot of countries have tried to take action at various levels to limit these substances. In the United States States in 2008, EPA issued regulations on ten rodenticides, including second generation rodenticides, like we're talking about here. And they said, you can't sell them in small quantities. You can't sell them in settings like grocery stores. But unfortunately, that left a lot of loopholes for the context of this story, I looked on Amazon, for example, and I was like, okay, let's just type in the name of one of these second generation rodenticides and see. And the first thing that popped up, I could get around 8 them. That was, you know, marketed as saying something like, this is the strongest anticoagulant on the market today. And it was like $130. And, you know, so that's underscoring. Indeed. It is very easy for someone to get this.

C
Yeah. Oh, man. We've talked quite a bit already about how problematic some of these chemicals are, and you certainly mentioned controlling rodents has benefits, disease and damage that they can cause. But what sort of are then our best options when it comes to rodent control?

D
Honestly, it's kind of a complicated question because there is that balance and people are trying to strike it. There's certainly a lot of options that any individual person can do. For example, making sure your trash containers are closed and secure, cleaning up spilled food outside. If you feed feral animals, you know, feral cats and dogs in your neighborhood, not leaving those food trays out all the time, you're just supposed to put them out the same time every day for a short time so the animals know to come, but hopefully the rats aren't also coming. One of the other things I found interesting as a city dweller here is that, of course, people don't like it when there's dog poop out in the alley or on the street or you step in it in the sidewalk. But it also can be real, really an attractant for rodents. It's nutrient filled, so the animals might come and eat it. And so that's another, another reason to pick up after your dog is that that can exacerbate rodent problems. One thing that's being explored in California is trying to actually release raptors to help them take care of the rodents, which, you know, sounds really great in theory, because you're like, okay, rodents are eaten by raptors. That makes sense. But you can only trial that kind of project in an area where there aren't the poisons because you don't want to accidentally start exacerbating the raptor death problem. Something that people have been excited about are these basically birth control for rodents, where you take these same kind of bait boxes that you put the poison in now, and instead you put anti fertility drugs in them.

So basically birth control, but it's also kind of expensive. You know, you have to keep doing it over time. You also need to have technology in place to alert you when the fertility drugs are getting low. So you go and replace them. So there are a lot of moving parts you have to think about there. It's obviously not an easy fixed. The real public health intervention messaging would be, all right, prevention is the best cure here. So you need to be doing those things upstream about making sure that there aren't rodent problems in your area, by taking care of your trash, by making sure that you're thinking ahead of time about how to help prevent rodents. But we all know that rodents are a fact of life and they won't just disappear overnight.

C
Absolutely. Yeah. You also mentioned in the story that there are some ecologically based rodent management strategies that some folks are employing. What do those entail?

D
Yeah, I was reading some interesting research from Africa that was looking at how farming communities were trying to manipulate the environment to avoid rodent problems. So these were situations like, oh, well, the rodents might like certain areas, so make sure you clear grass and brush from those areas because rodents might be using them as cover. Deploy cats or birds to help control rodent populations. For example, one thing that was being explored in Ethiopia was taking local plants and making them into rodent poisons that would have a relatively short half life as compared to these rodenticide chemicals we've been talking about. That particular intervention isn't perfect. A biologist who was studying the issue told me that rodents need to consume it multiple times. So that's not ideal, obviously, but these kinds of ecologically based rodent management systems have been in pretty widespread use in Southeast Asia and rice farms. And so they're thinking about, well, how could we apply lessons learned from there to other settings in Africa or elsewhere?

C
Yeah, absolutely. I mean, I know you've spoken with veterinarians and researchers and folks in the conservation world.

How did the sources you talk to see the future? When it comes to looking at rodent control, how are things looking into the future?

D
You know, I don't feel like any of the researchers I spoke with had a clear vision of what it was going to be going forward. There was definitely a greater message for most folks that they wanted more policy interventions, that the existing EPA regulations have not done enough, and that these rodenticides are used much more widely than they should be used. But unfortunately, no one had like, okay, if we just do this intervention or that intervention alone, it's going to fix the problem. California has really restricted the use of these rodenticides, but even with the restrictions they've put in place that have been further than what we've done at a federal level, there still continue to be widespread cases of animals being exposed to these rodenticides. So that just really says to me, okay, this is a problem that's going to need continued solutions, and those solutions will need to be tweaked over time.

C
Well, thank you so much, Dina.

D
Thank you. Great to talk about this with you.

C
Absolutely. Dena Feinmaren is a freelance science writer based in Washington, DC. You can find a link to the story we discussed@science.org. podcasts.

A
A deeper understanding of genetic diversity is essential to ensuring robust, equitable healthcare for all of the all of us. Project, led by the National Institutes of Health, is making great strides towards this goal by sequencing the genomes of over a million diverse individuals, including previously underrepresented populations. To ensure high quality sequencing data, the NIH used Roche's Kappa Hyperprep library prep kits. Learn more about this study and Kappa hyperprep kits by visiting go dot roche.com hyperprep.

E
Before we get to the next part.

F
Of the show, I'd like you to.

E
Consider subscribing to news from Science.

Every week we share stories from our news site, news from Science. Science journalists and editors kindly come on here and tell a story for our ears that they've been spending sometimes weeks or even months reporting and writing. If we were counting, our award winning journalists publish as many as 20 stories a week, from tracking policy to investigations, international science news. And yes, when we find new secrets about mummies, we report on that, too. It's an unbelievably valuable service. If you were here with us during early Covid days, you must have heard how plugged in and devoted our news team truly is. Please consider supporting nonprofit science journalism by becoming a subscriber for around fifty cents a week. To subscribe, go to Science.org news, scroll down a little bit, and click subscribe on the right side. That's Science.org news. Scroll down a little bit. Click subscribe on the right side.

A
Stay tuned for my chat with Yepe Rasmussen about training AIH to listen for seals, frogs, and whales.

F
One of my most treasured memories is going out in the middle of the night into the peruvian Amazon and being completely enveloped in this overwhelming chorus of frog calls. Squeaks, screams, croaks there's one that sounds like rattling dice. It's just this amazing, beautiful amphibian cacophony.

Soundscapes like this are actually tough for our human ears to analyze unless you have some serious expertise. But these days, machine learning and artificial intelligence are helping analyze scenes like these in order to keep track of animal populations. This week in science, Jepe Heve Rasmussen wrote a commentary piece on using AI and bioacoustics to monitor animals in the wild. We talk about where the technology is now and what it might be able to do in the future. Hi, Yepe.

G
Nice to talk to you.

F
Yeah, I'm really glad to have you. I hope we're going to be able to get some sound in here so people can really appreciate both the actual pleasure of listening to animals, and then, you know, what technology and science can do with that information.

So let's get into the basics of what we can learn about an ecosystem from its sound. How can a frog call or a whale song help us understand what's going on with a species or an ecosystem?

G
Many times it can be really, really hard to monitor populations, either because the animals are really hard to spot in the natural environment. The frog in the jungle is pretty hard to find. But also really big animals like the mighty blue whale can be really hard to monitor visually because the environment that it's living in is not really beneficial for visual confirmation of if they are there and how many of them are there. And thereby acoustics is a really, really good proxy in many cases.

F
So there's some advantages over, say, a camera trap. It's not going to see frogs that are practically invisible on purpose. Right. They don't want to be seen, but they do want to be heard for various reasons.

G
Absolutely.

F
So how can AI or machine learning, I'm going to just use them pretty interchangeably here. How can that help with understanding what's going on in the ecosystem in terms of bioacoustics? What tools are researchers deploying now using this technology to learn more about these animal populations?

G
There's a couple of reasons why AI is a huge benefit compared to ordinary human analysis. First of all, us humans, we have limited abilities to listen and discern which sound is connected to which animal.

Their AI got much better abilities, but also our concentration tend to lack. Once in a while, if you're sitting in front of your computer and you have to listen to animal calls, and once again, I can go back to the marine environment and say, there's a lot of people in the whale community that's sitting listening to these recordings for months, if not years. And if you've got a project that runs through several years, then you might have different people that got different levels of skill and different levels of attention. So you cannot be certain that your analysis is consistent across the data set. Their AI is much more consistent.

F
Yeah. And the AI can kind of build off of what those expert listeners were doing. They can annotate data that will help.

G
Exactly. The vast majority of the AI that is being used now is based on supervised learning, so that we have a human ground truth that we then use to train this neural network, that this is this kind of frog and this is the other kind of frog. But there's also unsupervised learning that's gaining territory all the time.

F
So to get a little bit technical here, this machine learning isn't actually listening to audio per se. It's turning that into an image, into a visual. What happens from there?

G
Our neural networks are really, really good at laserfung images. They're really good at looking at an image and saying, this is a cat or this is a mouse, or this is a coastline. In bioacoustics, we quite often transform the sounds into image like two dimensional structures called spectrograms, where we, on the y axis we've got the frequency bands, and on the x axis we've got the timeline. And then it's a color map of the energy of the sound. So that if you imagine a sound that goes like then on a spectrogram, there'll be a descending line going from higher frequencies and down to lower frequencies as the time progresses. And that's how a lot of the bioacoustics and AI that we are working on now, that's how they're doing it. But in the future, it will be that we are using the AI directly on the sounds itself.

F
Yeah. What kind of discrimination can these neural networks do? Can they pick out a bunch of different frogs from a noisy soundscape like that?

G
Oh, yes. There's been very nice study in the region of Saba in Malaysia. They had 39 species of birds and reptiles and amphibians that they wanted to see. Can we find them using AI and bioacoustics? And out of those 39 species, they found 34 of them in the recordings.

F
Wow. So, as you mentioned before, this is especially helpful for inaccessible places like deep, dense vegetation or in the ocean. And you have some sounds for us from. What is it? Seals in Hawaii?

G
Yeah. A couple of years ago, Gillian sills and other authors actually classified monk seal calls underwater into six different categories.

It's connected closely to the paper we just wrote because in that way, we can moniza this highly threatened species. There's only 1400 monk seals. So monitoring by listening for their underwater calls is a very good way of looking into spatial distribution and trends. If they are moving somewhere, if there's been less vocalizations. Also the classification that we are talking about brings another aspect into this study. Not only are we able to monitor the populations, but it can also teach us more about the structure of the vocalizations.

F
Yeah, absolutely. Okay, so let's listen to some of these sounds that you brought for us. Give us an example here of monk seals calling underwater.

G
Yes. This is one of the categories of the monk seal underwater vocalization, and it's called a whoop.

F
How do you spell that word?

G
W h o o p. Whoop.

F
Oh, like a whoop. Oh, that's great. Okay.

G
Yeah.

D
A whoop is that.

F
That's great. What else do we have to listen to?

G
Yeah. Then there's also the bout with the different types. Of course, monk seals make 90% of their sounds in so called bouts. And no one really looked into how is the structure? Is there some kind of semantic in these sentences we can call them?

F
Yeah, and this is such an interesting question because, you know, so far we've been talking about, can we detect the presence of a specific species or, you know, a type of animal in an ecosystem? And that's really important information for conservation, for understanding animal population sizes and maybe some details about which kind of male, female, juvenile, that kind of thing. But now we're getting into what are the meanings of these calls, which is kind of a different job for artificial intelligence, but also super interesting.

G
Oh, it's super fascinating. The way that we are kind of grouping the different vocalizations are based on fairly primitive measures that we extract from these calls. But these trends to using laughs, language models like chat, DBT, and all the fancy networks that we are starting to use more and more, those can also be used for analyzing animal vocalizations. And that opens up a whole new set of possibilities.

F
Do you need to have some kind of observation and behavioral component, though, to really, truly understand the meaning of those vocalizations? I mean, if you just are, like, listening only, can you really extract meaning? Or I guess the calls themselves are a behavior.

G
Us humans, we can quite easily communicate just using our voices.

F
That is true.

G
Animals, it's not quite the same that a lot of animals, they need some kind of content together with the vocalizations to actually have it make sense. Of course, in the oceans. Again, sorry for going back to the oceans all the time.

F
That's okay.

G
But it's, you know, a lot of people are talking about the dolphins are the smartest animals on the planet and so on, but they spend up to, like 70% of their time emitting their own signature coin. So they're basically like, that's right.

F
It's me. It's me.

G
Exactly. It's like Yeber yeber yabber all the time.

And on the surface you could say that that does not really sound that smart. But combined with the connection in which it's been vocalized and the behavior, I'm certain that's a deeper meaning to saying it's signature called all the time.

F
Very cool. So what about whales? I mean we talked about dolphins, we talked about monk seals. We might as well finish out our ocean journey. What can machine learning tell us about whale songs?

G
Oh, I'm so glad that you asked that question because that's one of my previous projects was listening for blue whales and fin whales. The males of these whales got very stereotypical calls to impress the females. But both sex of blue whales and fin whales also produce social calls that are much more variable and that makes it much harder for traditional algorithms to automatically detect them in the passive acoustic monitoring recordings. I got hired at Texas A and M to see if we could find a way of finding these circle calls. And I tried with doing a two step neural network detector. So for the first time we are actually able to automatically detect both sexes of the whales on the passive acoustic monitoring recordings. Whereas until then it's only been the males that we've been able to track and keep count on.

F
Yeah, that's super interesting. And I mean obviously so much better than sitting and listening to something that happened two years ago that you had to get an expert to listen to.

G
That's another huge advantage we've got with the whales and the AI is that we can have recording devices out in the shipping channels, as we actually do have outside California, to listen for the whales. And the AI can give an immediate response on that. Theres a good chance that theres a blue whale around here right now. So we can communicate that to the shipping companies and then ask them, could you lower the speed of your container ships a bit to avoid collisions between the whales and the big container ships?

F
That's amazing. That's great. So we've talked about these kind of various paths that this technology can take.

What do you see as some limitations to the approach that's going on right now?

G
Yeah, there's actually two things that, first of all, I think that the scientific community ought to get better at sharing the ground truth for training these neural networks and have them connected with relevant ecological and behavioral data. There's so many projects now that's more or less standing alone and starting from scratch and building up on the data that they need to train the neural networks. Fortunately, people that are publishing huge datasets should be used by other researchers. And there's also pre trained networks out there that is easy to access and use for your own investigations.

F
So, like an off the shelf starter kit. If you want to do bioacoustics, here's step one. And you need to put more specifics into it.

G
But also, this tend to be a little bit of like a separation of where the ecological studies are being done and where we need to protect the nature and the environment and where the analysis is happening. It's quite often in the global south that we are looking for the frogs in the jungle or coral reefs. Yeah, yeah, exactly. Whereas then it's quite often institutions in the global north that neural networks are being trained, and it could be highly beneficial both for the environment, but also for the research.

And luckily, there's a shift happening that cloud computing and more easily accessible software might be able to connect these two separate regions.

F
Yeah, that's great. All right. Yeah, but I feel like I should ask some sound questions because this is a podcast. So what are some of the more interesting sounds that you've gotten to listen to because of the work that you do?

G
There's so many things. I did my PhD on sound production in birds. And birds are amazing. Regarding sound production. Songbirds got this super specialized song organ called the syrinx with just hat with muscles, and we basically know like two or four of them what they do. And the rest of them, we still don't have no idea. But the sounds that they can make, I could suggest the listeners to go on YouTube and maybe search for Lyrebird. I also worked on a project with cod, the fish. Cod. Very few people know that cotton are making sounds, but about 50% of fish are making sounds.

F
Really?

G
Yeah, and cod are basically acting like songbirds.

F
Are they vocalizing or clicking or what are they doing?

G
They are using specialized muscles that are attached to their swim bladder. Theyve got a repertoire of different sounds that they can make. And male caught are basically acting like songbirds. In the mating season, they go down to the bottom of the ocean and then they establish a territory, and they defend this territory against the other male caught by vocalizing to them saying, like, this is my territory. Exactly like zombirds. And if a cute female caught is coming into his territory, he will try to woo her with a different kind of vocalizations. And if they find each other attractive, you know, what then can happen? And there is even different kinds of vocalization happening to synchronize the relief of sperm and egg.

F
All right, now I'm going to have to try it really hard to find some cod music online. Thank you so much, Jepe.

G
It was a pleasure.

F
Jepe Hawe Rasmussen is a postdoctoral fellow in the behavioral ecology group at the University of Copenhagen. You can find a link to the commentary piece that we discussed, also links to cod music and the sounds of the lyrebird@science.org.

A
Podcast and that concludes this edition of the Science podcast. If you have any comments or suggestions, write to us at Sciencepodcast.

You can find us on podcasting apps by searching for Science magazine, or you can listen on our website, science.org podcast. This show was edited by me, Sarah Crespi and Kevin McLean. We also had production help from Megan Tuck at Podigy. Our music is by Jeffrey Cook and Wen Koi wen on behalf of science and its publisher, AAA's. Thanks for joining us.

How rat poison endangers wildlife, and using sound to track animal populations

Primary Topic

Episode Summary

Main Takeaways

Episode Chapters

1: The Dangers of Rodenticides

2: Tracking Wildlife with Sound

Actionable Advice

About This Episode

People

Companies

Books

Guest Name(s):

Content Warnings:

Transcript