Catching Animals...on Camera in the Ellsworth Creek Preserve

Using 31 motion-detecting wildlife cameras in the study—tied or sometimes nailed to trees depending on their size—Murphy-Williams and the team are able to observe animals in their natural habitat in a non-invasive, low-impact way, giving TNC a better understanding of how animals use the landscape and the impacts of forest management and restoration efforts on the ground.

“The overall goal of this project is to inform how we're making holistic forest-management decisions, considering all facets of the forest system,” Murphy-Williams said. “Sometimes things like wildlife diversity and distribution are just assumed, rather than actually measured. The great benefit of working at the Ellsworth Preserve is that we can test assumptions on the ground.”

The Experiment

The camera locations were randomly selected in four of the eight experimental watershed basins of the preserve. Each basin is categorized based on the forest restoration method in use: control, passive restoration and active restoration. Control sites receive no restoration intervention, passive restoration sites have their roads removed, and active restoration sites receive tree thinning and additional planting of native plant species. The goal within active restoration sites is to accelerate the restoration of previously logged industrial timberland to old-growth characteristics, which can lead to thriving fish, wildlife and clean water.

The data were gathered through motion-activated cameras that would take a burst of images if movement was detected. Getting the proper camera angle was essential to minimize false motion detection from foliage moving in the wind. Murphy-Williams found that the easiest way to accomplish this positioning is to get resourceful. She placed sticks behind the cameras to angle them upwards or downwards before attaching them firmly to the tree.

The cameras collected data for six weeks in the fall of 2021 before the crew trekked back out to the camera spots to collect both the cameras and the memory cards. From there, the images were uploaded to Wildlife Insights, a program that identifies different animal species using artificial intelligence and machine learning. University of Washington research assistants Sophia Hara and Samantha Michelsen went through each image, double-checking the identifications and correcting them when necessary.

Initial results from nearly 4,000 images taken by the cameras identified about 23 unique species, including elk, mule deer, black bear, rabbit, opossum, racoon, bobcat, coyote and various small mammals and birds, but what interested the researchers most is where those animals were found.

The team’s hypothesis suggested that an actively restored forest area (thinned treatment type), will have higher biodiversity. Initial findings support this hypothesis with more detections and higher species richness from the camera data in areas that were thinned. However, they also found that the age of groupings of trees had the strongest significant correlation to animal detections, meaning that the older the forest, the more animal detections, regardless of restoration efforts. More analysis and data collection need to happen before results can be conclusive.

The Experience and the Team

The field team, composed of Murphy-Williams, three undergrad students from the University of Washington and other research scientists at TNC, spent three days in the preserve setting up wildlife cameras across Ellsworth’s experimental basins spanning a gradient of tree ages from young tree groupings logged fewer than 20 years ago to the original old-growth areas with trees more than 100 years old. They camped in tents around a base-camp cabin, enduring what two members of the field team described as a “torrential downpour.”

The team utilized compasses and GPS to navigate off-trail to the camera site locations, enduring unfamiliar terrain along the way. Murphy-Williams said that at some points it was like crawling up a hill on your hands and knees because of how steep some of the terrain was. According to her, often 80% of the work of field biology is getting to the study locations.

“We would drive to a spot that looked like it was the closest to the point that we needed to get to, and then we really had no idea what was in between the road and the camera point,” field technician Sophia Hara said. “Sometimes we’d hike up to I’d say a mile, maybe a mile and a half out to the camera points. It was very rigorous, very steep and it was very muddy.”

Looking Forward

Murphy-Williams sees the potential for the expansion and replication of this study to other preserves given that it’s a non-invasive monitoring technique. With the development of Wildlife Insights, a cloud-based platform that uses machine learning to identify animals in camera-trap images, tagging images could also be completed relatively quickly to inform holistic land management decisions on other preserves. As the climate warms and ecosystems change, our ability to conduct research and gather data in the field to inform management decisions becomes more vital.

The wildlife project is just one piece of the puzzle for effective whole system forest restoration. It is part of a suite of amazing projects and research happening at Ellsworth creek today.

The team will return to their monitoring locations to set the cameras up again in May for another six weeks of data collection to gain a stronger understanding of how animals might be using the space differently in the fall and spring.

Stay tuned for updates on this blog for what the team discovers!