Effects of plot size and their spatial arrangement on estimates of alpha, beta and gamma diversity of plants in alpine grassland

Little is known about the statistical consequences of plot size and their spatial arrangement when addressing biodiversity questions. Thus, we compared three vegetation plot sizes and two sampling designs with different spatial arrangements in alpine grasslands by analyzing their effects on common indices of alpha, beta and gamma plant species diversity. Data were collected in three 1-km(2) squares in summer pastures in the subalpine and alpine zones in Switzerland. In each square, 45 nested plots were distributed according to systematic random sampling (i.e., when the focus is on the abundances of the vegetation types and species) and maximum variation sampling (i.e., when the focus is on the set of vegetation types and species occurring). To compare the effects of the plot sizes and sampling designs on estimates of gamma diversity, we also estimated gamma diversity from an exhaustive sampling of the squares. Compared with systematic random sampling, maximum variation sampling hardly affected the indices of alpha diversity, but resulted in higher indices of beta and gamma diversity. More importantly, alpha and gamma diversity were found to be higher as plot size increased, whereas beta diversity was lower in the larger plots. Gamma diversity from the exhaustive sample was most similar to gamma diversity from maximum variation sampling on the largest plot sizes. Moreover, results from a variation partitioning analysis showed that landscape variability and plot size had a much larger effect on species diversity estimates than the sampling design. In conclusion, studies focusing on the occurring set of vegetation types and species within a landscape may apply maximum variation sampling if the most important (environmental) gradients for the stratification are known. Studies also focusing on the abundance of vegetation types or species may apply systematic random sampling.