We present a framework for biodiversity metrics that organizes the growing panoply of metrics. Our framework distinguishes metrics based on the type of information-abundance, phylogeny, function-and two common properties-magnitude and variability. Our new metrics of phylogenetic diversity are based on a partition of the total branch lengths of a cladogram into the proportional share of each species, including: a measure of divergence which standardizes the amount of evolutionary divergence by species richness and time depth of the cladogram; a measure of regularity which is maximal when the tree is perfectly symmetrical so that all species have the same proportional branch lengths; a measure that combines information on the magnitude and variability of abundance with phylogenetic variability, and a measure of phylogenetically weighted effective mean abundance; and indicate how those metrics can be decomposed into α and β components. We illustrate the utility of these new metrics using empirical data on the bat fauna of Manu, Peru. Divergence was greatest in lowland rainforest and at the transition between cloud and elfin forests, and least in upper elfin forests and in cloud forests. In contrast, regularity was greatest in lowland rainforest, dipping to its smallest values in mid-elevation cloud forests, and then increasing in high elevation elfin forests. These patterns indicate that the first species to drop out with increasing elevation are ones that are closely related to other species in the metacommunity. Measures of the effective number of phylogenetically independent or distinct species decreased very rapidly with elevation, and β-diversity was larger. In contrast, a comparison of feeding guilds shows a different effect of phylogenetic patterning. Along the elevational gradient, each guild generally loses some species from each clade-rather than entire clades-explaining the maintenance of functional diversity as phylogenetic diversity decreases.