There is evidence that the deposition of botanical collections in at least one of the world’s major herbaria (Missouri Botanical Gardens) has declined sharply in the last decade.
(Click on thumbnails to open the figure)
This may be due to a range of factors that we are currently investigating. It has been suggested that monitoring the dynamics of populations within permanent plots may be a more effective way of identifying tree species response to climate change than the rather arbitrary practice of botanical collection. This would concentrate effort on a few, hopefully representative, areas of forest from which changes in the broader distribution of species might be inferred.
How do the frequencies of species represented in botanical collections compare with the frequencies in permanent plots? Both are biased samples, but is there any correlation between them?
The following R code first reads data from a database under construction. The data includes all the georeferenced collections of tree species available from MOBOT. In order to run this code requires the user to have login details to this database that are not included here. The R package vegan includes a matrix of plots and species taken from the famous Barro Colorado permanent plots. By merging two tables of species frequencies, one derived from MOBOT and the other from the BCI data we can test whether species rank abundances are correlated.
Both the relative abundance distributions are approximately log-normal (note that actually fitting the log-normal is rather more complicated). However there is clearly no correlation beween the rank abundances of the species in the two data sets as can be confirmed by a simple statistical test.
Analysis of Variance Table
Df Sum Sq Mean Sq F value Pr(>F)
rank(a$freqBCI) 1 5346 5346 1.5232 0.2186
Residuals 203 712511 3510