Premium
Quantification of plant cover estimates
Author(s) -
Bonham Charles D.,
Clark David L.
Publication year - 2005
Publication title -
grassland science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.388
H-Index - 19
eISSN - 1744-697X
pISSN - 1744-6961
DOI - 10.1111/j.1744-697x.2005.00018.x
Subject(s) - quadrat , vegetation (pathology) , sample (material) , bernoulli trial , matching (statistics) , statistics , bernoulli's principle , cover (algebra) , sample size determination , mathematics , econometrics , plant community , ecology , point (geometry) , species richness , biology , medicine , mechanical engineering , chemistry , geometry , pathology , shrub , chromatography , engineering , aerospace engineering
Despite it's popularity and usefulness as a vegetation community descriptor, plant cover is often considered by ecologists to be a qualitative or semiquantitative variable. The relative merit of visual estimation within quadrats versus the use of multiple points to measure cover has been debated. This apparent conflict has been of concern for decades to ecologists in regulatory agencies and in private industry. Coincidently, regulatory agencies often rely on plant cover estimates to form the basis of sound decisions regarding reclamation success. We offer a resolution to these issues. If each sample site within a vegetation community is regarded as a repeated Bernoulli trial of sufficient size, say 100 trials, probability of the number of successes can be closely approximated by the standard normal distribution. There is little to be gained in decreasing or increasing the number of Bernoulli trials away from n = 100. Either quadrats or points may be used to equal effect only if the area formed by the quadrat and placement of points within a point‐frame or along a line are the same. Further, even with a moderate sample size of 30 repeated sets of Bernoulli trials ( n = 100) randomly located within a vegetation community, there is generally no need to transform the data to approximate a normal distribution prior to conducting hypothesis tests.