Factors affecting efficiency of field sampling for wildlife

Uncertainty about “detectability” complicates the interpretation of wildlife population surveys. Two approaches are available to reduce this uncertainty: (i) estimating detectability at all survey locations by double sampling, and (ii) “partial double sampling” in which a detectability estimate is developed by double sampling a subset of survey locations and applying it to other survey locations that are sampled only once. A key question is: Which approach is more efficient for different conditions?

This question is addressed in a recent paper titled “Population Estimation Using Partial Double Sampling” (Forest Science 56:417-420) by Craig Loehle of NCASI.  A constant sampling effort was considered because resources are never unlimited for field sampling. Results indicate that partial double sampling is more efficient (i.e., provides lower error estimates) when detectability is high and/or there is high variation in population between survey locations. In contrast, double sampling all locations is more efficient when detectability is low and/or there is substantial variation in population within survey locations. 

Contact Information