Do you know species are going extinct at an alarming rate? Do you know bees are in peril?
What if you found out those are all based on surveys? As we saw in the US election in 2016, surveys are not science, no matter how much groups claim they weight them. Unless the results are easily predictable (like the 2012 election) there is no way statistical significance will mean an accurate result.
Wildlife surveys are not always better, but some are better than others. A new study in The Condor: Ornithological Applications uses the results of more than thirty years of surveys of the Rocky Mountain population of Sandhill Cranes to try and infer how accurate surveys are. Because these cranes don't involve as much fundraising as bees, they use a three-year "moving average" to smooth out year-to-year irregularities in survey results, not one season in one year. Yet is it more accurate? Brian Gerber of Colorado State University and William Kendall of the U. S. Geological Survey assessed whether the annual population changes reported by these moving averages were realistic, based on what is known about crane demographics, and how they compared to the results of a more sophisticated statistical approach called a hierarchical Bayesian time series model. They found that while the moving average population estimates were reasonable, the more complex method performed better over a large number of scenarios.
Sandhill Crane. Credit: T. Cacek
Bayesian approaches offer a structured way to incorporate new information as it becomes available. "The model-based approach we looked at is very flexible and has some major advantages over other methods," says Gerber. "By taking a Bayesian approach, we can include additional information about both the observation process and the true population to obtain more realistic estimates and predictions. Also, the model-based approach includes measures of uncertainty about our population estimates, which are not usually provided by more common approaches and are crucial for understanding the level of confidence we have about our estimates."
Evidence suggests that management practices over the last twenty years have largely met the annual population objectives for the Rocky Mountain Sandhill Crane population. "Looking forward," adds Gerber, "managers may still be interested in adopting our more robust modeling approach due to its flexible framework, which makes implementing any changes relevant to the survey easier." The investment in collecting these long-term data may pay off not just for crane management, but for an advance in methods that can be applied to other species as well.