REPEATABILITY OF COMMUNITY DATA - SPECIES RICHNESS VERSUS GRADIENT SCORES IN LARGE-SCALE LICHEN STUDIES

Repeated ecological assessments based on permanent plot data require s ufficient data quality to detect a signal of change against a backgrou nd of noise (sampling error of various kinds). We analysed several com ponents of error in the time-constrained method for sampling lichen co mmunities used by the Forest Health Monitoring program: between-crew ( Technicians), crew-to-expert, between-expert, and seasonal variation. Data were from the southeastern United States and Oregon. Two types of dependent variables were used: species richness and scores an lichen community gradients (responses to climatic and air quality gradients). Gradient scores were repeatable to within 2-10% for experts and techn icians alike and did not differ between those groups. Species richness is much more difficult to estimate reliably. Despite relatively low s pecies capture by technicians, the high repeatability in gradient scor es demonstrates the statistical redundancy in information provided by various lichen species. These results imply that repeated assessments of species richness will contains considerable observer error, but tha t shifts in community composition may nevertheless be detected reliabl y.