摘 要: |
Vegetation ecologists sample vegetation by plots of size depending on the vegetation type being studied; e.g., forest vegetation is usually sampled by plots 100-400 m2, while grassland vegetation by plots 16-25 m2. Traditionally held assumption states that if the data should be comparable, especially for the purpose of diversity estimation, the size should be kept fixed (e.g. sampling all forest vegetation by plots of 400 m2). Alternative opinions, however, state that if the diversity comparison between community types should be fair, the sampling should consider the fact that with the same sampling intensity (plot area in case of vegetation studies) the sampling of species-rich vegetation types will be less complete than a sampling of species-poor ones. Solution to this is to standardize sampling to completeness instead of sampling intensity, which would as a consequence mean that species richer vegetation types should be sampled by larger sample plots than species-poor vegetation types. Interesting is that this concept is actually not new in vegetation ecology. Until 70's, vegetation ecologists sampled different vegetation types by samples with areas derived from "minimal area concept", which was calculated from species-area curve (constructed from a nested set of vegetation plots of increasing area) as an area for which slope of the species-area curve is the same as for other communities. This criterion is actually remarkably similar to standardizing sampling by coverage, which in the case of incidence-based data is one complement of the slope of the rarefaction curve for given number of samples. Minimum area concept has been heavily criticized for being subjective and difficult to apply, and it became a new standard to sample vegetation using samples of fixed size. I suggest that it may be time to revisit this decision again and start to sample vegetation by plots of variable size, depending on vegetation type being sampled. I will review the theoretical and practical considerations behind, and illustrate this approach using two real datasets, one of forest vegetation in Taiwan and the second using the herb understory vegetation sampled in the Czech Republic.
|