Searching for American chestnut: the estimation of rare species attributes in a national forest inventory
American chestnut, once a dominant tree species in forests of the Northeastern United States, has become extremely rare. It is so rare, in fact, that on completion of 80 percent of the plot measurements of the U.S. Department of Agriculture Forest Service's most recent inventory in Pennsylvania, only 33 American chestnut trees with a diameter at breast height 2: 1.0 in were found, out of 72,416 sampled trees. This paper discusses auxiliary sampling strategies that allow Forest Inventory and Analysis (FIA) units to estimate rare species in general as a first step in considering the especially difficult problems that American chestnut poses. The strategies involve (l) an increase of the initial plot size, (2) the use of adaptive cluster samping, and (3) a combination of the first two. Adaptive cluster sampling was developed for the estimation of rare clustered events and is considered here because American chestnut is not only rare but also known to occur almost exclusively in clusters.