Monitoring Loon Populations via Non-invasive Digital Image Analysis

Loons are a charismatic, black and white spotted and striped species that, much like a living bar code, appear ripe for digital image analysis. Photo by Andrew Reding.By Jessica Bridgers & Mark Pokras; a Christine Stevens Wildlife Award study

As many wildlife populations decline, the ability to monitor population sizes and changes is critical to conservation efforts. To determine population trends, researchers often must capture animals and apply unique bands or tags that can be used to identify individuals in the future. While these techniques reward researchers with fascinating and irreplaceable information, the process of capturing and handling is unavoidably intrusive and stressful to the animals.

Common loons are large, aquatic birds that inhabit northern lakes of North America. These birds possess extremely streamlined legs that make them highly efficient at pursuing the fast swimming fish that comprise most of their diet. To identify loons for studies, large bands are applied to the lower leg, and most individuals receive two to four bands. It is speculated that these bands may disrupt the streamlined nature of the legs, making it more difficult for loons to obtain prey. In fact, studies in penguins, another species of aquatic fish-eaters, have shown that bands negatively affect survival and reproductive success because of the extra energy needed to swim with bands.1 Research is currently underway to determine if bands disrupt the flow of water around loons’ legs. However, studying the ecological effects of bands on loons will require the ability to compare foraging ability, mortality, and reproductive success in banded and un-banded wild birds. This poses a challenge, as there are currently no alternative methods to identify un-banded loons.

But this may be changing. Digital image analysis is emerging as an alternative to traditional identification methods in several distinctly patterned species such as whale sharks, manta rays, zebras, cheetahs, and African penguins. Much like facial recognition software for humans, this technique uses a computer algorithm to analyze images and determine which individuals have been identified previously and which ones are new. Different programs require varying levels of input from biologists, but one unifying theme is the reliance of these technologies on citizen scientists to obtain photographs and location information for individual animals. This approach has been extremely successful for species monitoring projects such as ECOCEAN’s whale shark database, which has received 41,000 images from citizen scientists around the world.2

Loons are a charismatic, black and white spotted and striped species that, much like a living bar code, appear ripe for digital image analysis. To assess the utility of this technique in loons, we first identified three body regions of interest: large spots on the animals’ backs, “necklaces” and “chinstraps” on the birds’ necks, and bill and facial shape and dimensions. To determine the stability and variation of the spot and stripe patterns, we used existing software optimized for manta ray spots and zebra stripes, respectively. Unfortunately, feathers move around as birds change their positions, and our testing yielded poor results. However, bill and facial shape are more stable and will be the focus of our next studies.

Because the analysis of bill and facial measurements presents a novel problem for which no existing programs appear immediately useful, we are obtaining facial measurements from a sample of birds with the goal of assessing which, if any, may distinguish one bird from another. Ideally, we hope to find a series of ratios (for example, bill depth versus length) that will allow us to identify individual birds, since ratios can be standardized for photos taken 10, 20, or 30 feet away, or even from a bird in the hand. If successful, this technique could revolutionize the current methods of loon monitoring, involve the public in conservation efforts, and, most importantly, prevent unnecessary stress and anxiety to individual loons.

Jessica Bridgers is working toward an M.S. in Animals and Public Policy at the Center for Animals and Public Policy at Tufts University. Mark Pokras, DVM, is an associate professor at Tufts’ Cummings School of Veterinary Medicine and co-founder of the Tufts Center for Conservation Medicine.

1. Saraux, C., et al. 2011. Reliability of flipper-banded penguins as indicators of climate change. Nature. 469: 203-206.
2. http://www.whaleshark.org/