In the conservation field, accredited zoos have long played major roles in studying and protecting endangered species beyond just engaging visitors with wildlife, with an ongoing goal of using knowledge obtained through captive animal research to benefit those in the wild. Dayle Taylor works at The Stone Zoo with the Zoo New England's two jaguars, Chessie and Seymour, and she knows these big cats exceptionally well; she was even there when Chessie was born in 2008! Now these two jaguars are part of a growing database of footprints supporting Wildtrack's FIT, and Dayle is helping to improve this unique system for automatically identifying not only species, but individual animals, as well as their sex and age.
In the wild, FIT's machine learning identification capabilities can serve as a powerful tool for monitoring populations in a non-invasive way that does not impact the behavior of those animals who are being tracked. The concept of FIT is built from long-standing indigenous expertise of tracking wildlife, and importantly, this technique is low-cost and accessible, requiring no expensive and specialized technological gear, so the people who live locally near these species can contribute to scientific data on the environments and animals that they know best. All you need to get started is a phone with a camera and connection to the internet.
But in order for FIT's ML algorithm to be as effective as possible, large amounts of data are needed so it can learn to identify individuals and provide an accurate picture on how many animals live within specific populations. In some research scenarios, other tech tools like camera traps and tracking devices can help to confirm FIT's identifications. But even so, in the unpredictable lab of the wild, it can be difficult to get many high-quality footprints from the same individuals over and over, a necessary step in strengthening the algorithm. This is precisely where zookeepers like Dayle enter the picture.
Last year, Dayle began fine-tuning her strategies for collecting the perfect footprints from Chessie and Seymour. For the best results, the goal is to get as many photographic samples as possible from all four of each big cat's paws. Dayle uses her sessions cleaning the cats' enclosure to FIT's advantage, scanning the ground for clear footprints whilst the jaguars are safely out of the way. Although the process may sound simple - after all, these cats are constantly walking about the enclosure, so footprints should be abundant! - Dayle explains that gathering this data is a fair bit trickier than snapping a photo.
"It's a lot of trial and error, a lot more than people think."Dayle Taylor
"It's a lot of trial and error, a lot more than people think. Once you get it going, once you have a concept of what you're trying to do, it gets easier. But in the beginning, it took awhile to know what works best when you're working under pressure with a short amount of time, and you're always in a hurry." Because the jaguar holding pen, used when zookeepers enter the enclosure, is up against the exhibit rather than set back and accessed through a tunnel or other enclosed passageway, one of the trickiest aspects at Stone Zoo is not being able to control where the jaguars will step, and thus being unable to accurately prepare a surface for clear footprints in advance. As it turns out, not all surfaces are equally effective at holding the perfect print, and just because the jaguars stroll around in the enclosures' dirt all day, it doesn't mean they'll leave behind clear enough prints for accurate data collection.
To remedy this, Dayle can put down layers of sand or other substrates, but whether or not the jaguars will walk through it that day comes down to luck. "You set up the yard in the morning, put the animals out, and when they come in and you go to service the yard, you check to see how it's gone and hope you got something this time. You bring your supplies, you take a look, and when you find one, you get the picture and information as fast as you can."
Proper preparation also helps to speed along the process when a good footprint does turn up. "In the beginning, I didn't have the right rulers, I didn't have the little papers to label the print for the photo. So I was trying to do everything all at once, and take the picture on my phone, and it was stressful. Especially in the beginning, when you feel pressured to do it every day, and you're going through wheelbarrow after wheelbarrow of substrate trying to get the prints. When I was pressuring myself to keep it up at that rate, after awhile I was like, okay, this is too much."
Of course, sometimes nature itself lands a hand in the process, particularly when it snows. "Snow is definitely the easiest surface to work with. When it snowed here last winter, we were getting great prints - really clear and deep, with good details, which is what you need to pick up on the tiny differences that'll teach teach the system to know which individual this is, if this is a male or female, which paw you've picked up These are small, small variations that tell FIT a lot."
As winter helped the process along, Dayle allowed herself to relax and realize that using FIT could be opportunistic rather than a stressful routine. "Once I backed off, I was like, oh look! It's right there! We're getting quality prints! There's no stress or pressure, you don't have to dedicate a lot of time to it, and you're still getting this really valuable data. And the new FIT app has also smoothed out the process a lot. Downloading the app was a gamechanger." During last year's snowy season, Dayle says, "I was getting maybe 10 to 15 prints [a day] to photograph. But I don't know how many of those worked or were acceptable. But it was still a lot more data than usual."
By "acceptable," Dayle is referring to the other challenges of photographing footprints that may not seem obvious or particularly difficult until you experience them during a time crunch. As explained earlier, there is no way to direct these jaguars to a specific spot to leave footprints, and while an evenly spread surface like snow throughout the enclosure increases the chances of getting a clear set of prints, it is still impossible to control whether or not these prints will be in an area with good, even lighting.
Natural lighting is a very significant factor for collecting this data, as shadows or a lack of bright light can obscure the prints' finer details. Time of day and season also have a huge impact on the lighting and the ability for keepers to collect prints. "In the winter, you might get all this snow, and great prints left behind, but if the sun is setting early, you don't have as many daylight hours to get in there and take the pictures before you're dealing with sunset and shadows. When the days get longer, you've got more sunny hours to work with, but you're back to hoping they step in certain spots, or maybe you're dealing with shadows from having the sun really bright in the morning and at a bad angle. And that's hard to fix when you've got limited time to be in the enclosure.... Something like a phone flashlight isn't going to do much to brighten up the shadows outside.... And flash might be too much up close, at the angle you need for data. You still lose those details. So it's tricky, and it's partially luck. But you figure it out."
Even when nailing the perfect print photo is challenging, it's extremely rewarding for Dayle to contribute to this database, knowing that jaguars Chessie and Seymour are representing their species in a way that will help their wild counterparts. The data from captive animals is crucial to filling in the gaps on FIT, and Dayle hopes other zookeepers will learn from her experiences and begin working with their own zoos' cats to photograph prints.
"Don't look at the animal, look at the set-up."Dayle Taylor
For those looking to try collecting FIT photos in captive settings, Dayle recommends strategizing based on the enclosure layouts available to keepers. "Don't look at the animal, look at the set-up. You want a set-up where the animals have to take a specific path, like a narrow chute that goes to their holding area." By spreading an even substrate in such a confined and routinely used space, keepers can easily increase the amount of prints they collect. "At first, maybe the animals will sniff it and mess it up, or they'll use it like a big litterbox, or mark it. But eventually, if you're always putting the same thing in there, it'll become the norm to them, and they'll just walk through it. And that'll become a consistent source of data."
Aside from making the best of an enclosure's spaces, her biggest advice for other zookeepers is to lean into FIT's convenience and flexibility. "Other zookeepers should know that this program is ongoing. You don't have to get everything all at once. It doesn't have to be this big stressful part of your routine. It's important to get what you can, when you can, and just be flexible. Maybe one strategy won't work with your animals, and that's fine. You just pause and say, okay, what else can work here? It's really about being opportunistic, and being ready to get those prints when you run across them."
But what about those of us who don't work with captive animals, don't live in environments with species like jaguars or cheetahs, but still want to contribute using FIT? Dayle thinks one of the greatest things about the FIT system is its accessibility for citizen scientists. No matter what environment you live in or what species share your local natural spaces, you have the opportunity to gather meaningful data. For example, many North Americans will never see big mammals like cougars, coyotes, or wolves roaming around their local trail spaces, and may not even be able to spot smaller animals like raccoons easily, but their footprints will let you know that they're nearby.
"What I collect at the zoo is just as important as the data my neighbors are collecting. It's all valuable." Dayle Taylor
"You are most likely not going to see those really cool animals like cougars, because they heard you coming from a mile away, and they've taken off or hidden. Instead, you're looking for evidence. You know they're around you, maybe they even saw you! And you're the one who's going to get data on that animal." Not only does this strengthen the FIT algorithm, but it also helps to fill in population data on wildlife of all kinds. "You could even find evidence of a species that was possibly thought to be gone from your area, maybe extinct in that region. And through a footprint identification, you'll get proof that these animals are still there. You could contribute that data. That could be you! And that's what makes FIT really exciting!"
For those interested in trying FIT for themselves as citizen scientists, Dayle offers this inspirational message: "This system isn't just about one kind of data. It's incorporating professionals and citizen scientists. It can come from anywhere. And what I collect at the zoo is just as important as the data my neighbors are collecting. It's all valuable."
Want to learn more about getting started with FIT in the field, in a zoo, or as a citizen scientist? Visit our Get to Know FIT feature here, and join their group forum here on WILDLABS to get to know others in the FIT community!
*All photos courtesy of Dayle Taylor.
Thanks to Dayle Taylor for sharing her experiences working with FIT, and thanks to Zoe Jewell, WildTrack co-founderand FIT forum manager.