Reestablishing native plant communities is an important focus of ecosystem
restoration. In complex landscapes containing a diversity of ecosystem type
s, restoration requires a set of reference vegetation conditions for the ec
osystems of concern, and a predictive model to relate plant community compo
sition to physical variables. Restoration also requires an approach for pri
oritizing efforts, to facilitate allocation of limited institutional resour
ces. Hierarchy theory provides a conceptual approach for predicting plant c
ommunities of disturbed ecosystems and, ultimately, for prioritizing restor
ation efforts.
We demonstrate this approach using a landscape in southwestern Georgia, USA
. Specifically, we used an existing hierarchical ecosystem classification,
based on geomorphology, soil, and vegetation, to identify reference plant c
ommunities for each type of ecosystem in the landscape. We demonstrate that
ecosystem identity is highly predictable using only geomorphic and soil va
riables, because these upper hierarchical levels control the development of
vegetation, a lower hierarchical level. We mapped the potential distributi
on of reference ecosystems in the landscape and used GIS (geographic inform
ation systems) to determine relative abundance of each ecosystem, as a meas
ure of its historical rarity. We joined the reference ecosystem map with a
current cover map to determine current abundance of each reference ecosyste
m, and percentage conversion to different disturbance classes. We show that
over half of the landscape supports something other than reference plant c
ommunities, but degree of rarity varies widely among ecosystems. Finally, w
e present an index that integrates information on historical and current ra
rity of ecosystems, and disturbance levels of individual polygons, to prior
itize restoration efforts. The premise of the index is that highest priorit
y be given to restoring (1) currently rare ecosystems that were also histor
ically rare and (2) the least disturbed examples of these ecosystems, as th
ese will require the least effort to restore. We found that 80% of high-pri
ority sites occur within just three (of 21) ecosystems, Moreover, the high-
priority ecosystems all occur within stream valleys. Our approach provides
managers with a straightforward methodology for determining potential distr
ibution of reference ecosystems and for allocating efforts and resources fo
r restoration in complex landscapes. Development of a priority index for a
specific landscape requires an understanding of the hierarchical relationsh
ips among geomorphology, soil characteristics, and plant communities, in ad
dition to well-defined restoration objectives.