True restoration of highly disrupted native ecosystems is universally difficult, and has not been achieved on waste rock dumps at a rock phosphate mine in the resource-limited rangeland environment of semiarid Queensland, Australia. Thirteen years after rehabilitation, there is general correspondence, but some temporal variation, in tree and shrub species richness between rehabilitated and analogue native vegetation sites. Large differences in the composition and extent of ground cover vegetation were associated with dominance of most sites by an introduced perennial pasture grass species and by periodic drought. Complete reinstatement of native ecosystems is shown to be inconsistent with the physical constraints of waste rock dumps, the erratic climate, the previous history of extensive grazing and a lack of effective rehabilitation planning. Revised goals of strict landscape stability and broad biodiversity attributes, based on those of relevant native ecosystem analogues, are proposed to enable the identification of appropriate native and novel ecosystem targets for the modified landscapes. Novel ecosystems combining tree and shrub components of the native vegetation, and ground cover (including introduced grasses) could provide the most effective targets for mined land rehabilitation in a semi-arid environment. Principles are suggested for the selection of vegetation targets for the rehabilitation of mine waste rock dumps, considering the new lithologies and landforms and the occurrence of a dominant introduced grass species. These principles should be applicable widely, but especially in resource-limited environments. (C) 2015 Elsevier B.V. All rights reserved.
