Is sampling subjectivity a distorting factor in surveys for vegetation diversity?

Sampling design has long been a key methodology issue of vegetation science. There are still unsolved problems, amongst which the influence of the positioning (spatial arrangement) of plots remains a basic one. Random sampling is considered the only statistically valid option, along with its amendment by stratifying. However, huge amounts of plot-sample data originated from using subjective plot positioning, involving a great deal of the author's judgement, experience and bias. Field studies examining the role of subjectivity within a larger area (a landscape) are lacking. In the presented case-study, datasets sampled using computer-based geographically stratified random (R) and field-based subjective (S) designs are compared. In 1999-2000, 120 R and 244 S plots were spread across some 270 km(2). Their properties are compared regarding the differences in vegetation variability by ordination, and diversity by rarefaction. Although the differences are not conspicuous, both datasets partly contain different vegetation types and the S dataset contains at least some species-richer plots. The number of R plots had been clearly underrated while the expected species diversity values were nearly reached by the S. More R than S plots would probably be needed to reach the expected diversity maximum.