Candid Camera

At first there is nothing—windblown leaves maybe, or the quicksilver skitter of a squirrel. I can’t identify the source of the movement, and settle back expectantly because soon, I know, there will be more chances.

Huddled in the twilit hour I am hunting, expecting the common whitetail deer—but hopeful for more elusive game. Where there are deer there could be a wolf, right? A bear? Either would make the wait worthwhile. Or perhaps something I’ve never seen, like the elusive fisher?

Some time passes before I see the princely buck, so hale and burnished brown that my gaze lingers long in pure appreciation. His neck and shoulders are heftier than even the regal eight-point crown suggests. I’ve seen a lot of deer already, but he has presented broadside, at perfect range. My finger hesitates as I savor the action. And finally I decide, yes, this is a keeper.

I shift in my perch and refocus. Yes, there is the heart. My finger flexes. And I click on the heart icon. Subject 4988060, a Dane County buck snapped last November, is now in my favorites folder.

My hunting perch, you may now realize, is my customary recliner, and I’m using my laptop to spy on the wildlife of Wisconsin while dinner warms. In 20 minutes I’ll go through a few hundred of the millions of photos already collected by Snapshot Wisconsin, a growing net- work of trail cameras.

By now everybody’s seen trail cam photos. Maybe you or someone you know already uses them to scout deer, or just to see what’s on your land when you’re not looking.

Certainly someone’s emailed you a photo or short video, or they’ve shown up in your social media feeds. Those are the special shots, curated, viral. Snapshot Wisconsin is the raw feed, and therein lies the fun. Because here you can get your wildlife fix and be a scientist, too. Identifying these animals contributes to a cutting-edge effort that may fundamentally change the way we study wildlife.

“It’s like having 350 people out there in the woods day and night recording everything they see,” says Jennifer Stenglein MS’13 PhD’14, a research scientist with the Wisconsin Department of Natural Resources (DNR) who directs Snapshot Wisconsin. “That’s amazing data that we’ve never really had before.”

And 350 is just for starters. The goal is four cameras in every township in Wisconsin. Stenglein will be happy if they can reach at least 3,000 cameras. “We are, I believe, going to have one of the best data sets in the world,” she says.

At 10:40 every morning a NASA satellite flies over Wisconsin and snaps a series of pictures. The photographs measure many things, including a day-by-day record of how green the landscape is, which in turn gives us an idea of how well the plants are doing. The data has been collected for years—one of the satellites, Terra, has been in orbit since 1999—and offers an ever-lengthening perspective on the American landscape.

Satellite photos are now commonplace, but for most people remote sensing data is an abstraction. Woody Turner, program manager for NASA’s Ecological Forecasting, is always working to make that data matter to as many Americans as possible. “It’s really important to be able not only to understand what’s happening in your backyard or your woodlot but also to put it in the broader context,” he says. “The satellite brings in the broader context.”

In 2012 NASA announced it wanted to fund a project connecting its data with state agencies and university researchers. These are regular customers, but now there was a twist: NASA wanted a project that also used trail cameras and citizen scientists.

Phil Townsend, a professor of forest and wildlife ecology at CALS, had wanted to connect trail cams and remote sensing data for years, and he quickly called his professor colleague Ben Zuckerberg to brainstorm the citizen science angle. Then they reached out to Karl Martin BS’91, then the DNR’s forestry and wildlife research chief,
who knew camera prices were dropping and was also thinking about how to use them to improve research techniques. Martin also had access to a rich store of potential volunteers.

With all the ingredients NASA was looking for, the Wisconsin team won a pilot grant to install 80 cameras. It was an opportunity to improve wildlife research and put big data to work in the natural world. It even seemed like a promising tool for youth engagement—a partial antidote to nature deficit disorder. “It’s a very good example of cross-disciplinary, cross-agency teamwork,” says Martin, now the interim dean and director for UW–Extension Cooperative Extension. “This is how you leverage the Wisconsin Idea.”

Almost as soon as it began, state budget woes put the project on ice. In a curious twist, a raging national debate over gun control led to record sales of guns and ammunition. These sales are federally taxed, and a portion is returned to the states via the Pittman–Robertson Act for natural resource projects. With a secure funding stream, Snapshot Wisconsin began in earnest.

While the technology has been available for years, the ambitious scale remains a challenge. Educators and tribes can install cameras throughout the state, but cameras for private land are being rolled out gradually. Racine, Vernon and Dodge counties recently joined Iowa, Iron, Jackson, Manitowoc, Sawyer and Waupaca. At last count 417 volunteers were operating 607 cameras that have taken more than 8 million photos.

“The logistics are a big part of it,” says Townsend. “The scale that we’re doing this at has never been done before.” But scale is also the payback. Townsend is interested in phenology—the cycling of the landscape from brown to green and back again. Factors ranging from climate change to land use change can influence phenology. The Snapshot cameras are programmed to take an image at 10:40 a.m. every day, in sync with the satellite, providing a much richer data profile for that precise location.

Meanwhile the motion trap captures the phenological patterns of the animals. “Animals respond differently to their environment,” says Townsend. When they give birth, when and where they feed, when they’re out and about and when they’re in hiding all change, and we understand only a fraction of the whys. Bringing landscape data together with animal data may answer a lot of outstanding questions.

“Wildlife research every now and then gets transformed by technology,” notes Tim Van Deelen, a professor of forest and wildlife ecology. Radio telemetry revolutionized wildlife study in the ’70s, but it also took a while before researchers were able to put that information to use.

“That’s where we are with camera data,” Van Deelen says. “We’re in that lag phase where we are figuring out how to be efficient with the use of that data. I’m betting that as cool as things are right now, they’re going to get cooler as analytic techniques develop. I think there is a lot of basic biology that is going to come clear because underlying Snapshot Wisconsin is a very robust sampling scheme.”

There are two kinds of Snapshot Wisconsin volunteers. One group maintains cameras—either on their own land or special project cameras on public lands. Sited away from human activity and preferably on a game trail, the cameras operate day and night, snapping three photos in quick succession via a motion trigger. Memory cards and batteries need to be changed at least every three months, and the card uploaded back to Snapshot Wisconsin. Here technology takes over. To avoid any possibility of surveillance, the images on the card are encrypted. After decoding they are uploaded to Microsoft Cognitive Services, where special software removes images that contain humans. Then the image batches are sent back to each camera volunteer, who removes any people pictures the software may have missed.

After this double-check, the images move to me in my armchair via Zooniverse, a citizen science web platform designed by the Adler Planetarium in Chicago. Its goal is to harness our digital enthusiasm for something more than selfies and cat videos. On Zooniverse you can help with research projects that range from finding evidence of water on Mars to transcribing Civil War telegrams.

Why not just let a computer do it? Even in this age of the Watson cognitive computing platform and pervasive facial recognition, the human mind is still the most agile tool available for subtle pattern recognition. “There is no machine that’s as good as the human brain when it comes to being able to capture these kinds of images and classify them appropriately,” explains Zuckerberg.

Log on to Zooniverse and you’ll soon begin to appreciate both the challenge and your gift. The three-photo sequence captures movement. Some images are empty, and if the frame sways, you can tell that wind triggered the snap. But then you find an empty image where just a tiny bit of vegetation moves, and you realize that something has just passed by. Sometimes there’s just a blur of color, or—at night—eye gleam. After a while, you begin to recognize places and patterns, to appreciate the different ways that animals use and move across the landscape. Even the boring photos can surprise you. There is one squirrel in Sawyer County who loves to run a steeplechase along a few fallen birch logs. Occasionally this camera catches a deer. But just as I was getting frustrated with what felt like the 99th photo of the same squirrel, I realized the field beyond was crowded with 14 young turkeys.

Citizen science dates back at least as far as the then-nascent Audubon Society’s first Christmas bird count in 1900. (Plain folk have been collecting astronomical and meteorological observations for far longer.) In Wisconsin, thousands participate in all kinds of projects, monitoring everything from water quality to bat populations.

Zuckerberg hopes that through Snapshot Wisconsin, biology can join the ranks of such disciplines as meteo- rology that collect data continuously. “Collecting biological data tends to be very difficult,” he explains. State-of-the- art radio tracking can follow only a few individuals. Ecologists want to see how species respond across broad stretches of space and time.

“To me the real value of this is being able to think about animal communities over the course of an entire year,” Zuckerberg says. “It’s thinking about big-pattern ecology.”

Snapshot Wisconsin is in what you might call its giddy start- up phase. There isn’t an end product yet, but as the project ramps up, the anecdotal excitement grows. Director Jennifer Stenglein can tell you that there are quite a few porcupines, not so many striped skunks and a fair number of fly- ing squirrels. Also, that we don’t capture as many wolves as you might think, and that it can be very hard to tell coyotes from wolves. And, to no one’s surprise, there are lots and lots of deer. In fact, 60 percent of the animal photos from Sawyer and Iowa counties have deer. Which leads to an obvious question: Can Snapshot Wisconsin close the persistent (and politically sticky) gap between hunters and the DNR about deer populations? Nobody is taking bets on that, but the project should upgrade research techniques overall. “The way that the DNR tallies wildlife is highly sporadic,” says Townsend. “It’s not systematic, it’s different among different wildlife species, it’s difficult to do and it’s expensive to do well.”

Stenglein’s other major DNR responsibility is care and feeding of the state deer population model, and she sees Snapshot Wisconsin as a dual-use tool. On the one hand, it can contribute to the modeling currently in place, providing an index for population size, some idea of overwinter survival, and the fawn-doe ratio. “Cameras can be the best way to get a couple of those deer metrics, we think,” she says.

“It might also lead to an entirely different way of understanding the deer population,” Stenglein notes. The current model uses data from two observation windows: an August/September survey conducted by the DNR and the public, and the nine-day gun season harvest data. Snapshot would provide many more data points in time.

Two important research projects will help determine the ultimate value of the cameras. Elk reintroduction in Sawyer, Ashland, Bayfield and Jackson counties includes a much higher density of cameras. This will allow scientists to check the validity of the lower-density Snapshot data. And because many of the elk are also collared, traditional telemetry data can also be compared with the camera data. Similar comparisons can be made on another project in Dane, Iowa and Grant counties studying the survival impact of chronic wasting disease. Deer and their predators (coyote and bobcat) are both being collared, and cameras are also planned.

Current deer population models have a strong grasp of general population dynamics, but they are missing crucial landscape factors that we know influence deer. That, says Townsend, is where Snapshot Wisconsin will make the difference. “You are not going to get any one township perfectly, but by sampling enough townships you are going to sample the diversity of land cover and land uses,” he explains.

When all of those cameras meet all of that diversity, patterns will emerge. Find a relationship between deer density and vegetation and you can begin to make predictions. “The strength is in numbers,” Townsend says. “The remote-sensing data is everywhere. Can we harvest all that information to help make the models better?”

Charged with predicting deer populations, Stenglein usually thinks about lots of deer all at once. But as she’s built up Snapshot Wisconsin, a different window on wildlife has opened.

It began when she saw the work of an artist who was using her own trail cam photos for inspiration. Stenglein realized the artist was not painting a generic raccoon, but a very particular raccoon. The artist didn’t “know” the raccoon, and was just looking at photos. Yet there was a kind of individual relationship on view. “I realized that so much of this project is actually about the individuals in these photos,” Stenglein says. “That’s what draws people to this project.”

It was easy to imagine the connection landowners might feel for a camera they install and maintain on their property, or even one on public lands that they use. Stenglein gets lots of email from volunteers thrilled the first time they get a fisher or black bear they didn’t know they had on their property. Sue Steinmann MS’83 volunteered to place a camera on her scrub oak barrens near Arena “to see if we have bear or bobcats,” she says. “I really think we had a wolf come through last winter.” Now she’ll have more than footprints for proof.

Steinmann and her husband are active in ecological restoration, so they are probably more engaged in natural resource issues than most people in Wisconsin. But one of the things being studied by Snapshot Wisconsin is how citizen science can lead to better communication between scientists, resource managers and the public—and how this might lead to better resource management overall.

“When you have folks who are engaged in the process in more depth, and maybe helping to drive some of the questions, or helping to partici- pate in the interpretation of the data, that’s where you’re starting to see some of these community-level outcomes,” says Christine Anhalt-Depies, who is currently pursuing a PhD in wildlife ecology.

Anhalt-Depies is watching the online dynamic among the volunteers— some of whom come from all over the world—and how that evolves. Members of the research team are identified in Zooniverse, and the project also includes a few moderators (you can think of them almost as docents)—volunteers who help new users navigate the learning curve. The chatter is informed and supportive, and while the task might seem rote, it quickly becomes fun.

“I get addicted to doing that and have to stop after a while,” admits Sue Johansen BS’94. As a naturalist at Devil’s Lake State Park, she monitors three cameras for the park and one Snapshot Wisconsin camera in the West Bluff area. While the cameras began as a new way to engage visitors, they’ve also found animals—flying squirrels and short-tailed weasels—that no one knew were in the park. “What happens when you’re not around?” she says. “It’s a different way to connect to the outdoors.”

Then there are the “super users.” Zooniverse projects tend to develop their own core volunteers, people who process fantastically more images than most people. Some of these people are fully vested in the community aspect, engaging in conversation through message boards. Others remain silent. What are they getting from it, Anhalt-Depies wants to know. Will it translate to engagement in the real world?

“These are not cyborgs out there,” Zuckerberg says. “These are people very invested in the research.”

It’s these modern times that make Snapshot Wisconsin so fascinating.

We are becoming so acclimated to screens, to surveillance, to the omnipresence of cameras. Social networks have always mattered, but they are more visible than ever as we attempt to reap their bumper crops and avoid their vicious undertow. Selfies may be changing our very sense of our place in the world. Science and business are being rapidly remade by our ability to collect big data, and by our struggle to understand it.

Snapshot Wisconsin rides the rebounding ripple effects of all of these phenomena. And yet somehow nature remains at the center of the experience.

I admit: I had my doubts. But I threw both hands up in delight when I scored my first black bear. I was tickled to learn the blob that I had thought might be a wounded turkey turned out to be, literally, a happy family pileup of otters. I laughed longer than I should have when the camera caught a coyote leaving a fecal sample. (Photo bomb.)

In nature there is no substitute for observation. And while the parade of images in Snapshot Wisconsin should not be mistaken for being out there, it’s a legitimate supplement, a booster shot against nature deficit disorder.

“If you are going to maintain nature or wild places on this earth as our own numbers grow, I think it’s going to be because we care about it,” says NASA’s Woody Turner. “And to care about something you have to be at least somewhat familiar with it.”

Zuckerberg worries that we are increasingly detached from nature— that some children actually view nature as something to fear. Sometimes he listens to his children, ages 9 and 14, on Zooniverse in the next room. They love all the deer pictures but get totally jazzed by the occasional bear.

“I think using technology to allow another experience is what makes this project fun,” he says. “This offers a window for kids to become interested and engaged in natural history. I think any way you can do that is going to be a positive experience.”

From Space to the Field

Images captured from satellites orbiting 440 miles above the Earth tell a powerful, richly detailed story about crop yields—revealing the lushness or deficits of fields with surprising precision.

With the help of about 100 farmers so far as part of a citizen science project, researcher Phil Townsend and his team hope to coax even more valuable information from the satellite photos and change how farmers’ yields are reported and analyzed.

“The reporting of crop yields is now done at the county level with information confidentially reported by farmers to the USDA,” says Townsend, a professor of forest and wildlife ecology. “Counties can be very large. We now have the ability to analyze yields at the field level with these images, giving us much more accurate and granular data.”

By tracking the greenness of fields from the satellites and analyzing climate data, researchers hope to identify impacts of insect pests, crop diseases and weather events like frost, tornados or hail.

First, however, they needed to test their yield estimates against farmers’ actual yields. So Townsend’s team developed a website—yieldsurvey. wisc.edu—that allows researchers to confidentially crowdsource crop yield information. Townsend is encouraging even more submissions.

Farmers can enter their field’s location by dropping pins on a Google map, along with the type of crop and the actual crop yield for as many seasons as possible. The information is then analyzed and compared to estimates developed in Townsend’s lab.

“Our yield estimates are within about 15 percent of what the farmers report,” Townsend says. “Our target is to get that down to 10 percent. If you’re a farmer, the closer you can get to that could be the difference between making money and breaking even.”

Citizen science is a two-way street—it helps researchers tweak their estimates based on real data, and those more accurate numbers can help farmers be more productive and better managers.

Ultimately, Townsend says, the satellite technology and climate data, refined by knowing the actual yields from participating farmers, have the potential to predict crop yields well before harvest time.

Farmers see promise in the new approach, says Kevin Erb, a UW–Extension agronomist based in Green Bay.

“Farmers benefit from using remote sensing technology,” Erb says. “If we know early in the season that we have the potential for above-average yields, that can affect the types of pesticides and fertilizers that you use during the season.”

Being able to make decisions during the season based on this sort of predictive data could increase profits $50 or more per acre, Erb says.

Townsend’s team is cooperating with the USDA and hopes to snag funding to broaden the project. The effort is an example of the Wisconsin Idea at work, Townsend says.

“We have to connect with our constituents, and that’s where crowdsourcing and citizen science comes into play,” he says. “Farmers are participating in the science, and they see the benefits. It’s building trust.”

IMAGE: This map looking at soybean crops in the Upper Midwest shows how yields are predicted to vary even within the same county. Researchers are trying to verify their estimates by working with farmers to determine their actual yields.
Courtesy of Aditya Singh/UW–Madison CALS