ADAPTIVE CLUSTER SAMPLING FOR ESTIMATION OF ABUNDANCES WITHIN LOCAL POPULATIONS OF LOW-ABUNDANCE PLANTS

Abstract

Quantitative estimates of abundance for rare plants can be difficult, as the most widely used sampling techniques are ill-suited for rarity. Adaptive cluster sampling (ACS) can take advantage of the spatial clustering common in rare plant populations to provide more efficient unbiased estimates of population sizes than simple random sampling. When plants are found in a quadrat, all adjacent quadrats are adaptively added to the sample. Despite this biased sampling, the Horvitz-Thompson estimator for adaptive cluster sampling provides unbiased estimates of population means or totals, and variances of those estimates. Because ACS disproportionately samples quadrats with plants, it can provide additional efficiency whenever further attributes of rare individuals need to be assessed, such as demographic parameters or genotypes. Adaptive cluster sampling was performed on a population of Aletris bracteata, a moderately visible herbaceous species, in a savanna near Chekika, Everglades National Park, Florida. Both 1-m2 and 4-m2 quadrats provided reasonable estimates of the population size. The 1-m2 sampling included 30–36% of the estimated total plants while sampling only 5% of the total area. The 4-m2 sampling captured 78% of the estimated total population while sampling only 21% of the area.