A keystone species is one that "influences the ecological composition, structure, or functioning of its community far more than its abundance would suggest." (See http://glossary.eea.europa.eu/terminology/concept_html?term=keystone%20species.) Although the term is often thrown about, demonstrating that a keystone species really exists by a controlled experiment was only first shown in a paper published in 2012.
Researchers in Australia for two years surveyed the bird population of every patch of woodland to be used in an experiment. They then sorted woodland into two groups: those with and those without mistletoe. They further divided the former in two; half the patches had the mistletoe removed and half had it left alone. Then three years following the removal of the mistletoe from some of the patches of woodland they undertook another bird survey [in those patches with mistletoe removed, ongoing activities were undertaken to cull any mistletoe regrowth].
As one would expect, in patches in which the mistletoe was removed/culled, species that feed on mistletoe berries or nested in mistletoe plants disappeared. No surprise. But, the survey found that around 1/4 of other bird species disappeared from the removed/culled patches of woodland. In patches in which the mistletoe was left in place, the number of bird species actually increased. At least part of the increase was attributed to the end of a period of drought. So, mistletoe was acting as a keystone species, but by what mechanism?
The researchers concluded that mistletoe was enhancing the food supply, but not merely by providing berries. First, mistletoe weakens or kills the host trees, opening them to infestation by various insects etc. Second, mistletoe sheds a lot of leaves, which added to the depth of leaf litter in the woods. Deep litter favors insects. In either case, an enhanced food supply was provided for insectivorous birds.
The study can be found at: http://rspb.royalsocietypublishing.org/content/early/2012/07/03/rspb.2012.0856.