Do endemic taxa correlate?

One of the main assumptions of many biodiversity assessments and conservation efforts is that biodiversity correlates across taxa. In other words, an ecosystem with high plant diversity might be expected to also harbor high lichen diversity, high arthropod diversity, and a great many birds, bees, and bloomin' confusion. 

If this assumption is true, than scientists could study one taxa, say lichen, and use the results as a surrogate for studying all of the other possible taxa. But a new study by Dr. Che-Castaldo questions this "surrogacy" assumption:

"Testing Surrogacy Assumptions: Can Threatened and Endangered Plants Be 
Grouped by Biological Similarity and Abundances?"

Abstract:

"There is renewed interest in implementing surrogate species approaches in 
conservation planning due to the large number of species in need of 
management but limited resources and data. One type of surrogate approach 
involves selection of one or a few species to represent a larger group of 
species requiring similar management actions, so that protection and 
persistence of the selected species would result in conservation of the 
group of species. However, among the criticisms of surrogate approaches is 
the need to test underlying assumptions, which remain rarely examined. In 
this study, we tested one of the fundamental assumptions underlying use of 
surrogate species in recovery planning: that there exist groups of 
threatened and endangered species that are sufficiently similar to warrant 
similar management or recovery criteria. Using a comprehensive database of 
all plant species listed under the U.S. Endangered Species Act and 
tree-based random forest analysis, we found no evidence of species groups 
based on a set of distributional and biological traits or by abundances 
and patterns of decline. Our results suggested that application of 
surrogate approaches for endangered species recovery would be unjustified. 
Thus, conservation planning focused on individual species and their 
patterns of decline will likely be required to recover listed species."

Similar conclusions have been reached by a other studies. For example, Erhlich et. al. 2002 found that in subalpine meadows in Colorado, indicator taxa show no skill in predicting diversity of other taxa, even among phylogenetically related species (in this case, butterflies and moths).