Estimating tree diversity in forest ecosystems by two-phase inventories

What is it about?

A two-phase strategy, suitably compatible with the most adopted sampling designs in large-scale forest inventories, is proposed. In the first phase, tessellation stratified sampling is performed by partitioning the study area into a grid of quadrats and by randomly selecting a point in each quadrat. The first-phase points are classified as forest or nonforest using remotely sensed imagery. In the second phase, a sample of points is selected from those classified as forest by means of simple random sampling without replacement. The second-phase points constitute the centers of circular plots that are visited in the field to record plant species (usually trees) and their abundance. Estimators of abundance and diversity and estimators of their variances are presented.

Featured Image

Why is it important?

The two-phase strategy adopted in this paper is quite in contrast with the traditional literature on diversity index estimation, in which it is customarily assumed that individuals (trees in this framework) are selected from the community by means of simple random sampling with replacement. Under this assumption, the frequencies with which species occurred in the sample nicely follow a multinomial distribution with probabilities coinciding with relative species abundance so that the inference on diversity indices is straightforward. At least in our opinion, these works are of little utility in our framework, because trees cannot be selected from forests by simple random sampling with replacement froma list sampling frame. On the other hand, the two-stage strategy adopted in this paper allows the estimation of species abundance and diversity indices in a very realistic framework and can be readily embedded in the two-phase sampling schemes adopted by most large-area forest surveys (e.g., National Forest Inventories).

Perspectives