 There we go. Next month, I just want to let everyone know we're going to have a wonderful program of an extremely talented author and illustrator of birds and other wildlife. Her name is Julie Sikafus and she is spectacular. And her program, it's going to be a little early in the month compared to our typical timing. It's going to be Tuesday, March 12th, the second Tuesday in March compared to our usual slot of the third Tuesday. And the title of the program is Baby Birds An Artist Looks Into the Nest. And she is a spectacular artist, has produced wonderful books on a variety of bird related subjects. If you go to our website, you'll be able to see some samples of her work and a little more description of what she does. So we hope you can join us for that on March 12th. That one will be also only by Zoom. It will not be in person, it will be only by Zoom. And so we hope you can join us for that. But tonight, we're delighted to have something a little out of the ordinary for us. We were more focused on birds typically or local wildlife. And tonight's topic is neither of those. Although some birds may make an appearance or two, either visually or in an auditory manner, but auditory manner. But we have with us and we're delighted to have with us Bobby Estabrook, who works for the Cornell Lab of Ornithology, which most of us are quite familiar with. But when we think of it, we don't think of elephants. And so it's something, an awakening for us. I'm very interested to see how that fits in with other aspects of what Cornell does. And I have some questions for Bobby at the end. We will be having a Q&A at the end. Feel free to type any questions into the, into the, do we have a Q&A box? Yep. Into that during the, anytime during the program. And we'll see how many we can get answered at the end of the program. The goal of the Elephant Listening Project is to help conserve the second largest block of rainforest on earth. And the biodiversity that it harbors by focusing on forest elephants as key architects of these forests. And before working on audit, auditory work related to elephants, Bobby also had spent 10 years working on the acoustics of Baleen whales. So that's another fascinating area. I don't know if that will come up at all in the course of tonight's program. But with that, I'd like to welcome you all. My name is Bill Graben. And on behalf of the board of directors of York County Audubon, happy to welcome you to tonight's program. And Bobby, turn it over to you. Thank you. Thank you very much, Bill. I'm going to share my screen right now. And hopefully this will work okay. Okay. Yeah, thank you so much, Bill. And thank you, everyone, for having me tonight. I'm pretty excited to speak with you about the Elephant Listening Project and share a bit about what we do. I know it's a little outside of the typical topics of your seminar series. So I appreciate this opportunity. My name is Bobby Asterbrook and I'm a researcher and a data analyst with the Elephant Listening Project. So a lot of my work involves analyzing and managing very large acoustic data sets that focused on forest elephants ecology and the impacts of anthropogenic activities on elephant behavior. And just to give you a little bit of background, so the Elephant Listening Project, as Bill mentioned, is part of the Cornell Lab of Ornithology, whose mission is to interpret and conserve biological diversity through research, education, and citizen science. And it's focused mainly on birds and nature. But within the lab is the Elephant Listening Project is part of the K. Lisa Yang Center for Conservation Bioacoustics, which uses sound in nature to inspire and inform conservation of wildlife and habitats. So as part of that mission, the Elephant Listening Project aims to conserve tropical forests in Africa using acoustic monitoring, sound science, and education, which focuses on forest elephants. So today I'll just I'll describe a bit about how and why we strive to meet this mission. But first, I'd like to give a broad overview of our program. The Elephant Listening Project was founded in 1999 by Katie Payne, who's in the upper left corner of these photos. And over the years, our team has grown and currently comprises researchers in Ithaca, New York, which is where the lab is based, as well as in the Republic of Congo and in the Central African Republic. So we're one of the first groups to use passive acoustic, sorry, passive acoustic monitoring in terrestrial systems. And we've continued to develop and implement the passive acoustic monitoring tools to help meet our mission. And central to our mission is a really strong applied component. We work with other conservation organizations to provide them with information on forest elephant populations and anthropogenic disturbances so that they can evaluate and adapt their conservation strategies as necessary. We're also conducting research that's based on acoustic data to enhance our understanding of elephant behavior and ecology. And we strive to advance technology and employ technology, such as machine learning detectors and real-time monitoring systems. But the foundation of everything that we do is really capacity building. We train and we empower Central African and West African conservationists and researchers to conduct their own passive acoustic monitoring projects independently. Our work mainly focuses in the Congo basin, which is pretty far from York, Maine. It's in Central Africa and it spans across six countries, Cameroon, the Central African Republic, the Democratic Republic of Congo, the Republic of Congo, Equatorial Guinea, and Geban. And the Congo basin is the second largest rainforest in the world and it's the largest rainforest in Africa. And we also conduct studies in the Ghanaian forest of West Africa, which unfortunately is experiencing really high or it's very highly fragmented forests there so it's not nearly as contiguous a block as the Congo basin. So first one thing I'd like to do is just take us deep into the forest of Central Africa, in the Central African Republic, which is where some of our work takes place. And I'd like to just play a sound for you to listen to just for a little bit and just see what you hear. There's a nice cuckoo in there that I like listening to. But just from listening to that forest, the sounds of that forest, it's pretty clear that there's a lot of animals there. And healthy forests are very rich in biodiversity. And in fact, in this recording you couldn't hear it but there was also a couple elephants vocalizing and I'll explain that a little bit later. But I heard a lot of birds there and the Congo basin rainforest is a global hotspot for biodiversity. It's home to 10,000 species of tropical plants, over 400 species of mammal, 700 fish species and over a thousand bird species. And the tropical rainforest in Western Central Africa play a crucial role in global health. They serve as biodiversity hotspots, which are homes to some of these species that I mentioned, but some of them are very unique to these biomes and don't exist anywhere else. And in addition, millions of people depend on the resources of the tropical rainforests for food, water and shelter. And the forests also serve as an important buffer against climate change by acting as carbon sinks, where the rainforest stores large volumes of carbon in the vegetation and in the tree trunks. In fact, the Congo basin is the world's largest carbon sink and it's considered the lungs of Africa. Among Africa's wildlife are the world's largest living terrestrial mammals, elephants. And elephants form very complex fish and fusion societies and they have core social units that consist of adult females and their offspring. They're also, they're a keystone species, meaning that they play a very crucial role in maintaining the ecosystem in which they live. Most people are familiar with savanna elephants, which live in open savannas. And we often see savanna elephants in films and in photographs. But there's another species of African elephant that live in dense, remote rainforests of western central Africa that are called the African forest elephant. And only recently were these two species recognized as genetically distinct from each other, which warrant special attention for each of their own conservation needs. Most of what we know about elephant social and vocal behavior derives from savanna elephants, largely because they're easier to observe in their open habitats. Savanna elephants are the largest of the three living elephant species with heights around 10 to 13 feet. They form really pretty large social groups, family groups that can comprise of maybe 10 females and their offspring, their calves. And sometimes multiple family groups can join together to form clans, which can comprise many, many individuals. Their diets primarily consist of grasses, but they'll also eat a variety of plants and fruits depending on the season. And females reach sexual maturity around 11 to 14 years of age. And for most males, it takes a little longer before they they reach that. And females also conceive typically every three to four years. And currently they're classified by the International Union for the Conservation of Nature, the IUCN, as endangered. In comparison, very little is known about the social structure of forest elephants, because they're really, really difficult to visually observe in the dense vegetation. The forest elephants are smaller in body size than savanna elephants due to the densely vegetated surroundings. And they also have much more narrow, more straight tusks than savanna elephants do. Their family group sizes are much smaller and don't really exceed more than 20 individuals, often much smaller than that. They predominantly feed on fruits, whereas the savanna elephants primarily fed on grass. But forest elephants will also eat leaves and tree bark, grasses and seeds. Now one big distinction is that female African forest elephants typically first give birth around 23 years of age. Now that's compared to savanna elephants that could give birth around 14 years of age. So they're much older by the time they first give birth. And their intervals, their birth, interbirth intervals are about five to six years, whereas compared to savanna elephants, it was three to four. And the IUCN classifies forest elephants as critically endangered. Now forest elephants are a keystone species for the African rainforests. And they're considered the arch, the engineers of the forest, which makes them an important species to protect and to conserve. So by focusing our efforts on studying forest elephants, we aim to help conserve the lungs of Africa. Forest elephants are fruit eaters, and they play a crucial role in seed dispersal of many of the tree species throughout West and Central African forests. Their uniquely large body sizes enable them to ingest really large seeds, which they can disperse across large ranges and cover vast areas of forest. In fact, forest elephants were capable of moving seeds three miles away from their parent tree and allows the seeds to colonize in areas where they otherwise wouldn't have been able to do on their own. This ultimately helps to shape the structure and composition of the forest ecosystem. And some tree species are entirely dependent on the elephants. They will not germinate unless they've actually passed through the elephant digestive system. So without elephants, those trees would not be able to successfully reproduce. And these trees, importantly, they're globally important because they trap a lot of carbon that would otherwise be released in the environment as CO2 and would otherwise accelerate global warming. Over many, many years, elephants have carved out large networks of pathways in the forest that connect resource locations, such as mineral deposits and fruit sources. And studies have found that fruit tree abundance, the fruit tree area and the tree size, tend to be greater along trail intersections than at random sites in the forest. And many of those large trails actually end up leading to forest clearings called bys. And bys are open areas of the forest that have mineral rich streams and watering holes like Zangabye, which is pictured here in the Central African Republic. And Zangabye was likely created by forest elephants over many, many years as they trampled and ate much of the vegetation that was in this area. And they're attracted to this area because it provides rich minerals and salts that supplement their diets, giving them nutrients that they can't obtain from forest vegetation. Also, these bys act as important social gathering places for elephants. And these clearings tend to be used by other species as well. But forest elephants are threatened right now by habitat loss and by poaching just between 2002 and 2011. Forest elephants declined in population size by about 62%. And they've also lost about 30% of their geographic range. So the IUCN recognized that elephants were facing an extremely high likelihood of extinction in the wild. And that's when they classified them as critically endangered on the red list. Now worsening the situation is the fact that forest elephants have a much slower reproductive rate compared to savanna elephants, as I mentioned earlier, which will make it really difficult for them to recover from ongoing poaching unless there's some immediate action taken. There's a little bit of good news though. So while poaching continues to be a significant problem, there's some hope because we have been seeing some declines in annual poaching mortality in recent years. So I think that there are some successful efforts in place to mitigate poaching. Since forest elephants inhabit very densely vegetated rainforests, they're really difficult to visually observe. So we don't know nearly as much about forest elephants as we do about savanna elephants. And most of what we do know about forest elephant demography and behavior comes from many, many years of visually monitoring forest clearings like Zonga Bay, which I mentioned earlier. You saw the aerial photo for it. And now this is another photo of Zonga up close. In collaboration with the World Wildlife Fund, researchers at Zonga have been able to identify individuals and learn a lot about population demography, social interactions, family group sizes, mortality and birth rates, as well as kinships and individual life histories. I'm going to just play a short video of my colleague, Daniela Hedwig, describing how Zonga Bay has played a very crucial role in helping us to understand forest elephants. This video can be pretty quiet. So during the video, you might want to I think during the video, you might want to turn your volume up and then when it's over turn it back down so I don't hurt your ears when I speak. Okay. I'm a researcher with the Elephant Listening Project at the Cornell Lab of Onatology. I am here at one of the last remaining pristine wonders of nature, Zonga Bay. Zonga Bay is a vast forest clearing in the heart of the rainforest of the Zonga Nuki National Park in the Central African Republic. Zonga Bay is an exceptional place because it is home to the largest stone aggregation of forest elephants. Anytime you come here, you can expect to see between 50 and 150 forest elephants. The elephants have come here over many generations and have shaped and maintained this clearing. They visit the clearing to access the mineral rich water through small water holes in the ground. But Zonga doesn't provide them only with minerals and water. It also provides them with the opportunity to socialize. Here the females meet their mothers and sisters and the males establish the hierarchies and the youngsters play and learn elephant social etiquette. Little is known about forest elephant behavior. This is mainly because they are so extremely difficult to observe in the rainforest itself. But here at Zonga we get the rare chance to observe them and study them when they come out into the open. Elephants use no frequency rumble vocalizations to communicate with each other. Those rumbles play very important roles in coordinating the social interactions between the elephants. We will now continue monitoring this amazing elephant population to better understand the social relationships between the elephants and also their vocal communication. So hopefully you could hear that video okay? I'm sorry that it was so low. I'll just reiterate something that Daniela mentioned is that Zonga Bay is an incredible place. It has the largest known aggregation of forest elephants in that location. And data that have been collected here by the elephant listening project and our colleagues has really provided us with crucial information about the conservation needs of forest elephants. So in addition to monitoring during daytime with video cameras, we've also been able to use thermal imaging technology to monitor the by at night. So these cameras have allowed us to count the number of individuals present at a given time, as well as observe nocturnal behavior and interactions between individuals, which gives us some great insights in how they utilize the clearing throughout the full day. So I'm not sure if you noticed because the audio was pretty low, but in the videos that I just played you might have heard some elephant vocalizations. The the buys kind of provide this really unique opportunity for us to conduct direct visual and acoustic observations of elephants, which gives which allows us to validate acoustic recording methods and data sets and improve what information we can extract from sound recordings. In fact, most of the elephant listening projects early acoustic work has focused largely on these single location monitoring at these clearings with simultaneous visual observations. So here is a spectrogram of elephant vocalizations that were recorded in a clearing in Gabon in a by there along the x-axis. Let's see if I can get my pointer. Along the x-axis here is time. And along the y-axis is frequency. And then areas where it's really dark indicate that there's a lot more energy. There's higher amplitude. So most people are familiar with trumpets and roars that elephants produce. We hear those sounds a lot. But these vocalizations that you see on the screen are called rumbles, and they're a little bit different. They consist of a fundamental frequency, which is this low part down here. And then these harmonic stacks above it. So fundamental frequency and harmonic stacks. I'm going to play this sound. This sound itself is actually quite loud. So it's not too much louder than my voice, I think. But I'll play it now. It's louder than the video that I did just play. And it starts here. There's a bit of a gap here. And then you'll hear these. There's a trumpet. It's walking through water. So these rumbles were produced really close to the microphone. So they have a nice clear and loud sound. Sometimes when elephant vocalizations are pretty far away, we might not actually hear them. Or sometimes elephants can produce infrasonic vocalizations, which are below the threshold of human hearing. And we wouldn't be able to hear them anyway. But we can see them pretty well in a spectrogram. And in this spectrogram, at the beginning, you might have very clearly heard two elephant rumbles. But there's actually a third one in here. And I'll play it one more time. It's right here. This one's pretty low. And it's going to just sound like a vibration. But I'll play this part one more time in the beginning. And we'll see if you hear anything there right here. So again, this rumble is from a separate individual who's much further away from the microphone than this one. Rumbles are a great elephant vocalization to monitor. They're produced by all age and sex classes and in a variety of different behavioral contexts. Rumbles are low frequency signals. They have a dominant frequency that's typically around 50 hertz, which allows the sound to propagate much further through the environment than higher frequency signals, since low frequencies are less impeded by the environment. And so by monitoring these low frequency signals, we are able to better understand forest elephant ecology, their call context and behavior. And we can apply what we learn about. We can apply what we learn from by studies to forests when elephants are in a place where elephants are nearly impossible to see. Since 1999, the elephant listening project has been utilizing passive acoustic technologies to record the forest elephants at clearings like Zonga Bay. And since the beginning of our research, technology has very rapidly advanced, thankfully. In the early days, we had to use very large batteries and bulky equipment and deploy them up in the trees to collect long term acoustic recordings. And these units would be placed at sites where they'd be left for several months at a time. And at that time, we were really largely limited by memory card storage capacity and battery power, as well as weight. These were pretty hefty units. So we couldn't really conduct extensive large scale surveys deep in the remote forest. But nowadays, passive acoustic recording equipment is much more compact and it's much lighter than it used to be. It has longer recording capacities. It allows us to, which allows us to deploy equipment much deeper into the remote forests and at more locations simultaneously. So we could monitor beyond the vise now with this equipment on a landscape scale with much more spatial and temporal coverage than was previously possible in remote areas. Now, there are two recording devices that I'm showing in these photos, but they're essentially the same device. These were developed by engineers in the K. Lisa Yang Center for Conservation Bioacoustics and they are called SWIFTS. The one on the left in that picture on the left is a standard SWIFT. And that's that's typically how the SWIFTS are made. But we had to have them specifically, especially designed to handle the harsh conditions in tropical rainforest. And so that's what the photo in the middle shows is a SWIFT in a rugged case. Unfortunately, we have learned a lot over the years about the various ways in which animals can destroy recording equipment. And we've had several devices that elephants manage to get their trunks on and step on them or stab them with their tusks. So our current design is not perfect, but it is much, much harder than its predecessors and it's doing a great job. It's also really exciting to think about that technology is only going to continue to get better. So right now we don't have this yet, but we're working with engineers to implement real time monitoring technology in tropical rainforest. So currently the dense forest canopy limits solar power and satellite transmission, but we're pretty confident that things are going to change in the future. So it's pretty exciting that we might be able to use this technology. I'm going to play another video, which gives us a glimpse of deployment for one of our forest study sites in the Republic of Congo. That is part of our landscape scale monitoring efforts. And first I want to make sure the volume is quite low. So many of our sites are in pretty remote areas that require a combination of a small boat ride through the different river systems and quite a lot of trekking through the dense vegetation to reach our sites. The vegetation can be really tricky to walk through. It could take a while to go a short distance. It's also usually quite hot and humid in the tropical rainforest here and that adds some challenge to it. And the insects, the insects don't make it easy at all. This is a colleague of mine Peter Regg with the deployment team. And they're setting up one of our older recording equipment, recording units. And we have to hoist the recording devices about 10 meters above the ground just to keep them out of reach of elephants because they pretty quickly spot anything that doesn't belong and if it's in their reach they're going to investigate it. Once the recording devices are deployed, we leave them out there for several months to record. And then when they're done, we'll change out the batteries and the memory cards and place them back at their site again typically. We take careful notes during the deployment which allows us to easily find where the unit was several months later even when it blends in pretty well with the vegetation. After the field team collects the memory cards, they typically will load them on hard drives and then they'll send it back to Ithaca. In Ithaca, we'll upload the sound files to our local server and our cloud storage and then we just check the files for any potential errors and we'll run some automated detectors. Because of the massive volume of data that we have, we depend very heavily on automated detector algorithms to help us more efficiently process the data and find elephant rumbles as well as other sound sources. Otherwise, it would take us years to go through all of our files. Just to give you a sense for the volume that we have of data, just one of our study areas in the Republic of Congo has over 2 million sound hours and counting. It's an ongoing project. It's over 300 years of data, so it's a lot. There are primarily two detectors that we run on our data sets. One is a rumble detector and the other one is a gunshot detector. And we also can run detectors to find other sound sources, other species of interest, like different bird species. For example, the African gray parrot, different primate sounds and anthropogenic sounds like illegal chainsaws or chainsaws. And while our current detectors substantially improve the efficiency of our analyses, they are imperfect. For example, I'll get my pointer again. So for example, for the elephant rumble detector, here we have an elephant rumble up on the left and we also have a vehicle sound, which easily confuses the detector as a rumble. So it looks similar enough where our detector will often find a vehicle sound. Similarly, raindrops or tree branches that tree branches cracking or breaking can be very easily mistaken by our gunshot detector. So these are gunshots here on the left and this is branches on the right. Now in the past couple years, we've been testing different machine learning approaches for rumble and gunshot detection, which will likely be vastly superior to our current detectors. However, these signals, the elephant rumble and the gunshots are quite challenging even for machine learning right now. But we've been seeing some really encouraging results. And given how rapidly the technology has been advancing, particularly in machine learning, we anticipate that it's really not going to be too long before we have a much better detector. For now, however, we do need to validate our detector output to ensure data accuracy. And to do that, we use a software program called Raven Pro, which was developed by our colleagues at the K. Lisa Yang Center for Conservation Bioacoustics. So here we're looking at a screenshot of Raven Pro. And you can see a spectrogram, which you've seen earlier. This is a different spectrogram. And then there's a table, I think I still, yeah, so there's this table underneath, which is our detector output. So we ran a rumble detector. And in this table are things that the rumble detector thinks are rumbles. So what we'll do is we'll work our way through the table and annotate if it's a false positive, like you're seeing here in the screen, that's the false positive detection, or if it's an elephant, true positive detection. And our detector gives us some score values. So it gives us a little bit of level of confidence in what it thinks is a rumble. There's also this feature in Raven that we are able to use, where we can actually look at maybe 48 or more mini spectrograms at a time and annotate each detection as a false positive or a true positive. So is it a not a rumble or a rumble? And through this feature in Raven, we can really quickly go through a very, very large volume of data and rapidly annotate. The outcome of this is essentially we'd have a data set where we could then run models and run statistics. So using those kind of data, we basically want to transform from the sound data and those tables into conservation. So what have we learned so far from the work that we've been doing? Well, from very early data collected at Zangabai, Maya Thompson, Katie Payne and colleagues found that acoustic monitoring could indeed be useful, a useful method for passively, remotely and continuously monitoring forest elephants. They also determined that elephant call rates can be an index for elephant abundance, which could then mean that it might be possible to estimate population abundance from acoustic data alone. Now I say this cautiously because right now we can't yet estimate absolute abundance of forest elephants in the forest setting, but these results give us insights into its potential use and using these methods for abundance estimation for elephants. We've also seen how forest elephant rumbles can give us insights into forest elephant demography. So through visual and acoustic observations at the clearing, we've been able to observe the duration and the frequency of elephant rumbles can be indicative of age class of the elephants. For example, if we look at the plot on the left side here, rumbles that last more than four seconds tend to be produced by adult males, while rumbles that last about two seconds or less in duration are usually by juveniles and infants. And similarly, the plot on the right shows that the frequency of the rumble can indicate an age class of an elephant, where the rumbles that are much lower in frequency are more likely to be produced by an adult, and higher frequency rumbles are typically produced by juveniles and infants, which makes sense because much larger-bodied individuals are capable of producing much lower frequencies, frequency sounds. Observations at Zangaba have also allowed us to make comparisons in elephant vocal behavior between the two African species. Daniella Hedwig, Joyce Poole, and colleagues found similarities and differences in how forest elephant and savanna elephant colorates differ by behavioral contexts. They also found that the behavioral context of a rumble had a significant effect on the duration of the rumble, and that they were different between the two species. Round the clock monitoring of clearings and forest settings, so 24-hour-long recordings like we typically do, have allowed us to look at elephant vocal activity between daytime and nighttime. For example, one of our forest recording sites in Gabon here showed that there was no significant tendency for elephants to vocalize more during the day or during the night, and so here in this plot we're looking at hour across the x-axis, and then date is on the y-axis, and the grayed-out area is nighttime hours. So the dots, there's a histogram here of the total number of rumbles that were recorded for each date and hour. So there's no significant difference between day and night in the forest, but in contrast when we look at this plot here on the right, it shows that there are hourly, this is hourly vocalization, hourly behavior in a forest clearing, and we see that there's a really clear shift in vocal activity towards nighttime hours, where they're really quiet during the daytime. Now the reason that the elephants here shifted their hourly calling behavior tonight hours was because of significant human disturbance at that location, there was a lot of logging around the area. So from studies such as this, we learned that change in the proportion of elephant activity that occurs at night could be an index of anthropogenic pressure and something to monitor. One of our largest and longest acoustic monitoring surveys is in the Noble and Doki National Park in the Republic of Congo. In close collaboration with the Wildlife Conservation Society, we've been continuously monitoring elephants across a 1,250 square kilometer area since 2017. This is a 50-unit acoustic grid and it's the first of its kind for landscape scale acoustic monitoring. So from our validated elephant rumble detections, we can create maps like this where we get to look at the average number of elephant rumbles or vocalizations that are detected at each recording site per month, with the green colors reflecting the lower number of rumbles and the white areas represent higher number of rumbles. Data like this allows us to better understand elephant movement and site use across seasons and in response to anthropogenic disturbances such as logging and gun hunting. Using those same data from that 50-unit grid, we were able to compare elephant activity between three different strata. The protected national park, a logging area, and an inactive logging area. So this is the active logging area, this is the protected national park, and this is the inactive now abandoned logging area. And each red dot represents the site location of a recording device. We observed that elephant presence was lowest in the active logging area, which was not surprising and it would make sense that elephants would avoid areas of high human disturbance. But we were surprised to see that there was more presence in the former logging area than in the protected national park. However, their high presence in the abandoned logging area actually makes sense as well. So due to the abundance of young herbaceous vegetation that grows after the trees were harvested, elephants have been really attracted to this area for food. These results highlight the importance of logging concessions for elephant conservation. And while we review the audio files for elephants, we can also look at gunshots, which represent illegal gun hunting within a protected park or around it. So acoustic monitoring here allows us to provide unbiased systematic data on where and when gun hunting took place every four months. So we deploy the equipment for four month time periods and then swap out the battery and memory cards. So for example in this map, we've overlaid the average elephant vocal activity with the number of gunshots that recorded during two specific time periods, just to get a glimpse of how the elephants are responding. Our partners who manage and patrol this area consider this type of information to be invaluable. It's an invaluable addition to their intel and their ongoing anti-poaching efforts, for which they largely used to only rely on evidence that was found during their patrols and informants from the villages. We were able to look at the effects of gunfire events at a site on elephant call rates before and after gun events on two different time scales, five hours before and after and 24 hours before and after. And we found that there was a decrease in the calling rates before gun hunting, which actually suggests that elephants are aware of poachers or people in the area. And there was an increase in the call rates during the five hours after the gun event, which suggests that elephants are coordinating with each other. When we look at the 24 hours after the gun event, we can see that elephant call activity remains pretty low, which indicates that they're either quiet or that they've left the area. We've also been able to look at the gun hunting trends in the park. And one of our main findings from monitoring this is that illegal gun hunting has actually been reducing in and around the survey area since 2019. This could indicate that the anti-poaching patrols have shifted their strategies and implemented or improved the effectiveness of their mission, or that poachers maybe have moved away from this area. But in fact, park managers can and have used these data to evaluate the effectiveness of their anti-poaching patrol efforts. In this plot here, we show the average number of detected gun events in the blue bars and along with the number of kilometers that were walked by the patrol team in the park during that same time period, which is represented by the red line. So our takeaway from this is that when poaching patrol increased, gun hunting decreased. So far, I've talked about our background, the technology we use and the type of research that we do. Lastly, but most importantly, I just want to touch on the foundation of our mission, which is capacity building. We've been initiating intense training efforts to form independent local research teams that can help take the lead on both the Zonga by monitoring project and the Central African Republic, in the Central African Republic, and the Nwabliandoki National Park monitoring project in the Republic of Congo. We've hosted a few people to come to Ithaca to study and train with us, and we'll also go out to their region. We've also hosted different sound analysis workshops, elephant monitoring workshops, and we've created a center for biocoustics training and analysis, which I'll talk about a little bit. And in terms of outreach, there's a really exciting new initiative that started at Zonga called Ita Itdoli, where local teams actually joined the Zonga research team for two weeks, and they stay there and work with them and learn about the work and conservation. So for several years, we've been working very closely with colleagues at the Wildlife Conservation Society in the Republic of Congo to train them on acoustic analyses and field methods to monitor forest elephants and gunshots. And since 2017, they've started to take the lead of our largest project in Nwabliandoki National Park, and they're beginning to manage the acoustic monitoring efforts there. They now deploy and retrieve the acoustic equipment, they process and manage the sound data, they run the gunshot detectors, and they review the detections as well. And we're currently training them in the final stages of rumble detection and analysis. And they're also working on their own research questions that they might be able to answer with the data that they've been collecting. And in the Central African Republic, we've also been employing and training local people to manage and implement our ongoing acoustic monitoring and bi-monitoring efforts there as well. So as part of this initiative, two of our colleagues, Onnesi Samba and Frelsia Bambi, joined us in Ithaca from the Republic of Congo. They came for a couple months last year to continue learning with us and to teach us more about their own work. And they're actually going to join us again this year for a few more months where we're going to continue to work closely and ensure a smooth transition of the project to their management. We're also really excited about establishing a bioacoustic training and analysis center, which I mentioned just a little bit ago, with our colleagues at the Wildlife Conservation Society at the Nwabliandoki headquarters, which Onnesi, Frelsia, and one of our colleagues, Fael, have been leading. In 2023, we hosted two workshops for over 20 participants from various protected areas and conservation organizations in Congo, Geban, Cameroon, the Central African Republic, and Equatorial Guinea. This center serves as a training hub for practitioners throughout West and Central Africa to learn about all aspects that are necessary to start and expand acoustic monitoring programs and to improve conservation strategies for forest elephant populations in their respective areas. And we've also been running an ongoing seminar series where participants, past participants can join and we have discussions on their current research and offer support and open office hours to answer any questions or help out however we can. And we maintain an active WhatsApp group where folks chat and message each other with questions and discussions, which is pretty exciting. If you're interested in any of the work that we do or curious, please feel free to check out our website or send me an email at bobby.esterbroke at cornell.edu. And I'd just like to thank and acknowledge our partners and our funders for continuing support in our work and the governments of the countries that we work in, the Republic of Geban, Congo, and Central African Republic for permits in allowing us to be able to conduct the work that we do. And also, we have such an amazing team that we get to work with. I mentioned earlier we have engineers, software developers, and it's such a fantastic group with amazing support. So thank you very much for your attention. That was great, Bobby. Thank you. Joining you online again. If this reminder to people listening, if you have any additional questions, just please type them into the Q&A. But I have a few for you. You mentioned the ages at which they become reproductively active. But do you have any information about the life expectancy? I think the life expectancy is not dramatically different between Savannah elephants and Asian elephants. Around 65 or 70 years old is the typical life expectancy. Yeah. And how, well, a related question there, do you know how old they can be before they are no longer reproducing? That I actually do not know. I believe there have been some females who surprisingly were able to produce very healthy offspring at ages much older than what they anticipated. But I don't know the answer to that. Yeah. And, you know, what is your sense or what is the sense of the population numbers, I mean, you've talked about percentage decreases, but no, I don't think you mentioned absolute numbers. And I know that may not be clear information that's available. But it's, I don't remember what the latest estimate was. So actually a lot of estimates for abundance have been a result of dung count surveys in which people will go out into the forest and manually look for dung piles. And they rely heavily on different rates in the forest, defecation rate, and rates at which the dung piles will decompose. So a lot of the estimates have come from those studies. And in this forest, I think there, well, I'm probably not going to say the right number, so I won't say. That's fine. I just got a question, which is one that I was wondering about also, which is how stable are the countries, you're working in four different countries, how stable are the governments, how safe is it to be there? It can vary. We've had fantastic experiences with our work and no problems with the governments. In terms of safety to be there, some of them are listed as unsafe for their only travel, if absolutely necessary. But most of them are totally safe to go to and not bad, yeah. I assume you're not encountering poachers yourself? No, we're not. No, for the anti-poaching patrol teams, though, that can be a really dangerous job. They do encounter poachers, and I know that some rangers have lost their lives for this work. So that part can be dangerous. We've been very fortunate and we haven't had any negative encounters like that, and nor have some of our closest colleagues. You talked about machine learning. Is that what we refer to as AI? Yeah, yeah. It's essentially, you give it a bunch of examples of something that you want it to find or to look for, and it trains itself. So the more examples you give it of something to look for or to not look for, the deeper the learning gets and the more advanced it gets and capable of finding something pretty complex. So, yeah, it's pretty similar. They're very similar things. Here's a question that you may have addressed. Oh, no. You mentioned in the savanna versus forest elephants that there were difference in terms of the frequency of gestation. Is there any understanding of why that is the case? That I don't know the answer to, actually. Yeah, I won't speculate on that. I think there are papers out on that that have some thoughts. Okay. We have a question from one participant as to whether you could use any photographers. Oh, gosh. Yeah, I think absolutely. It's, you know, the photographing forest elephants can be very hard in the forest, but like at Zonga Bay, it's just the most incredible thing. And I think we're always looking for great photographers. So if somebody's interested, that'd be pretty cool. Right. I'm sure they could email you. Yeah, absolutely. And one last question I have is, I mean, I was certainly surprised to find out that there was an elephant listening project as part of the Cornell lab of ornithology. So my question is, what else is or might be going on at Cornell that we might be surprised to find out is going on there? Anything on that? Yeah, absolutely. It's a really good question. In the beginning of the presentation, you had mentioned that I had done some work with marine mammals and ocean noise. And that actually was through the K. Lisa Yang Center for Conservation Bioacoustics as well. So we, at the center, we study a lot of different taxa, including marine mammals, primates, insects. And the reason for that is because our mission is to, it's conservation and biodiversity. And so that fits in with the lab's mission as well. And so we're very lucky to have that support. But if you're interested in what else we work on, I would encourage you to visit the website for the K. Lisa Yang Center for Conservation Bioacoustics, which is on the Cornell Lab of Ornithology's website. And you'll get a little glimpse of the variety of different kinds of research going on. And I mentioned earlier too that we are so lucky because we also have software developers and engineers. The engineers design our underwater marine acoustic recording equipment as well for certain projects. And they're working on real-time monitoring devices as well. So it's pretty interesting. Well, that's great. That was a wonderful program. Thank you so much for joining us tonight. Thank you very much for having me and thank you for the great questions and for your attention. And best of luck with all of your work there. I'd like to remind everyone that our program next month will be on Tuesday, March 12th with an exceptional book author and illustrator, Julie Sikafus, on Baby Bird Nests. And with that, we wish you all a good night and see you next month. Thank you. Bye.