Cornell deer sterilization anomaly

INFORMATION ACQUISITION AND DECISION SUPPORT FOR SUBURBAN DEER MANAGEMENT   Issues in Information Systems Volume 18, Issue 2, pp. 158-170, 2017 166

The Cornell Anomaly.  Below is an excerpt from a paper [link above]  providing analysis of a deer sterilization project at Cornell, New York, using ovariectomy and tubal ligation on does.  Contradictory results were achieved in a similar project in San Jose, California.

The result of a sterilization project at Cornell in New York received considerable public attention in a Washington Post article titled: “Trying to limit the number of deer, with surprising results” (Landers, 2014). The lead biologist on the project summarizes that: “I thought that sterilization in an open population where things can move in and out won’t work. Maybe it was worth doing it in a sophisticated way to say we tried in the best possible way and it didn’t make a difference.” According to the article, although the number of does and fawns was reduced, an offsetting increase in the buck population resulted in no reduction of the total population. Bucks were drawn there by the sterilized does that continued to go into heat, attracting bucks as part of the breeding process. 

The published study reported “a 38% and 79% decrease of total adult females and fawns visible in sampled photographs, respectively, and an 873% increase in adult male visitation to camera traps” (Boulanger & Curtis, 2016, p. 727). The data for sampled photographs include repeat visits of the same deer. Male visitations in the first year were only 4% of the total, for a buck to doe ratio of about one to eight. The sex of fawns and many deer in the photographs could not be determined. In a suburban herd the buck to doe ratio is typically about two to three. The 873% increase was just enough to bring the buck to doe ratio back to its normal level. There is no explanation in the paper for the incredibly low starting number of bucks. Although the paper concludes that the experiment failed because the population remained approximately the same, the number of deer seen in sampled photographs actually decreased by 32% (Boulanger & Curtis, 2016, p 732). 

The conclusion that the deer population was not reduced by the experiment is based on analysis of the photographic data using the software NOREMARK (White, 1996). While no population estimates were provided for adult males, the population estimates for adult females show an increase over the study (Boulanger & Curtis, 2016, p 732), in contrast to the 38% decrease in total females from sampled photographs. Total population estimates later presented by the researchers appear to show that the total population trend was about the same but a little higher than the female population based on the software estimate, suggesting little effect from adult males on the total population trend estimate (Boulanger & Curtis, 2017). There is a dramatic difference in the population trends based on the photographic data and the software interpretation, yet both results are used to selectively draw conclusions.

 An important technical issue left out of the Washington Post article is that two sterilization methods were used, ovariectomy and tubal ligation, not just tubal ligation. Does having their ovaries removed, ovareictomy, typically cannot go into heat, so are not likely to attract bucks. Two other sterilization projects with publically accessible websites both report reduced deer populations: in Ohio at CliftonDeer.org (Year 2 Field Operations Report, 2016-17) and in Maryland at www.WildlifeRescueInc.org (First ever Non-Lethal White-tailed Deer Birth Control Sterilization, n.d.). The San Jose case also used both sterilization methods, although a much smaller proportion of tubal ligation, but as reported in Table 3 resulted in an 89 percent decrease in the number of bucks. These counts were made by direct observation. The authors of the Cornell study acknowledge “it is unclear why male numbers increased,” “an inability to reliably estimate the male portion of the campus deer population,” (Boulanger & Curtis, 2016, p. 732) and that that their result of a dramatically increased number of bucks is “currently undocumented in the literature” (Boulanger & Curtis, p. 733). The increase in the number of bucks photographed may also have been caused by the food piles used as bait to attract does for sterilization, by the fawns born in the area during the study, by a change in the areas frequented by a group of bucks.

The experience of a one year camera survey in San Jose makes very clear the difficulty of counting unmarked deer. The cameras are good for counting marked deer and were used in San Jose to verify the physical count done by volunteers. Among the marked deer we found that the same deer would return frequently to the same location, but this varied over time. Getting accurate population estimates continues to be a challenge for wildlife management and is the subject of much research devoted to improving available technology and methods in an environment of tight funding. The contractor on the San Jose Project, White Buffalo, continues to experiment with population control methods, providing new data and possible alternative solutions. 

Comments