This dataset contains data on vegetation biomass removed from treatment plots during the establishment of a biodiversity experiment at the Jornada Basin LTER site in southern New Mexico, USA. In fall of 1995, various combinations of plant functional groups or species were experimentally removed from 25 x 25 meter plots with the objective to distinguish the differential effects of plant community biomass, functional groups, and biodiversity within functional groups on ecosystem and plant community function. Eight different treatments were established with selective removal of species or functional groups: control (C, no removals); four functional group removal treatments (PG, perennial grass removed; S, shrubs removed; SSh, subshrubs removed; Succ, succulents removed), and three species richness manipulation treatments. Richness manipulations included a simplified treatment (Simp), where only the single most abundant species of each growth form is preserved and all other species in the growth form are removed, a reduced‐Larrea treatment (rL), where the Larrea is assumed to be the dominant and is removed while minority components remain, and a reduced-Prosopsis treatment (rP), where Prosopis rather than Larrea is removed as the shrub dominant. The amount of plant material removed during the establishment of these treatments was recorded for later use as covariate or measure of disturbance. Removed fresh material was weighed in the field by species, then converted to dry mass using a subset of removed plot vegetation that was oven-dried and weighed in the lab. Variables in this file summarize the dry mass of plants removed by growth form (functional group, i.e. shrub, subshrub, perennial grass, succulent), and by total live dry mass, for each plot in the biodiversity experiment. Also provided are masses of dead material collected from plots (same groups as live material removed for each treatment) and total dry mass, live plus dead. Species-level data are available upon request. This study is complete.