Measuring success: evaluating the restoration of a grassy eucalypt woodland on the Cumberland Plain, Sydney, Australia

  • Published source details Wilkins S., Keith D.A. & Adam P. (2003) Measuring success: evaluating the restoration of a grassy eucalypt woodland on the Cumberland Plain, Sydney, Australia. Restoration Ecology, 11, 489-503.


The Cumberland Plain is a shale valley which contains the western portion of Sydney, Australia's largest city. The plain supported distinctive grassy eucalypt woodlands, which since European settlement, have been reduced to 9% of their original extent by extensive clearance for agriculture and urban development. The woodlands are recognized as an endangered ecological community under state and national legislation. A 10-year project was undertaken to try and restore some former eucalypt woodlands in this region. This study assessed the floristic composition and structure of restoration areas with that of an unrestored pasture (acting as a control site) and remnant native vegetation (reference site) in western Sydney.

Study sites: The study sites were located on farmland in western Sydney, New South Wales, southeast Australia. The sites were selected as they included restored vegetation of varying age amongst patches of remnant vegetation and abandoned pasture. All sites were probably originally wooded and dominated by spotted gum Corymbia maculate, grey box Eucalyptus moluccana, narrow-leaved ironbark E.crebra, and forest red gum E. tereticornis with a patchy shrub understory of Bursaria spinosa, Dillwynia sieberi and Indigofera australis and a partial ground cover of grasses e.g. Themeda australis, Echinopogon caespitosus and Entolasia marginata and herbs including Brunoniella australis, Dichondra repens and Pratia purpurascens.

Vegetation restoration: Vegetation restoration commenced in the area in 1992 with planting taking place over 10 years (managed by 'Greening Australia'). The restored areas comprised over 1,000 ha of abandoned pasture. Historically, it had been was sown with non-native grasses, particularly Phalaris spp., cattle-grazed and had fertilizer applied. Plantings were undertaken to connect remnant patches of woodland. To evaluate success, suitable reference sites of remnant vegetation were compared with the restored sites over time. An abandoned pasture with no planting was used as a control from which restored sites were expected to become increasingly dissimilar in vegetation composition and structure with time.

In 1992 weed control was undertaken at restoration sites by slash and spray application with glyphosate before planting. Twenty-six indigenous tree and shrub species, propagated from local seed, were planted mechanically in rows. The mix of planted species aimed to match species with soils and topographic positions as would be occupied by wild plants. All plants were weed matted with a recycled paper disk and surrounded by a protective plastic sleeve. Glyphosate was again applied in spring and autumn for 2 to 3 years after planting to reduce weed competition. Mechanical slashing cut to reduce fire risk. in 1998 fencing was constructed to exclude livestock from restoration and native remnant areas

Data collection: Sites selected for sampling were close together and had similar topography to minimize environmental variation that might potentially confound management effects. All three treatments (untreated pasture, restored and remnant vegetation) and four different ages of restored vegetation across four sites were sampled.

Vascular plant species composition was recorded. Differences in frequency scores between pasture and restoration treatment were expected to indicate recruitment of new individuals rather than changes in cover alone. Cover-abundance data was available from a previous survey of remnant vegetation over a larger area of the Cumberland Plain, thus data was recorded in the same format to allow for comparison with remnant vegetation over a broader geographic area than that collected in this study.

Average height and cover of each vegetation stratum were visually estimated to assess vegetation structure. Percentage cover of bare ground and leaf litter (estimated visually) and aspect, slope, soil texture and location were recorded.

Species composition: Remnant habitat patches were clearly different from untreated pasture and restored vegetation. Changes in composition of the restored sites over time did not appear to be in the desired direction of increasing resemblance to native vegetation communities. There was a significant difference in species composition between all three management treatments when all species were included. However, differences between pasture and restored sites were slight (25 species contributed to 50% of the floristic dissimilarity between these two treatments, but only two, Acacia parramattensis and Eucalyptus tereticornis, had been planted). In contrast, differences between planted areas and remnant vegetation remained highly significant.

Eight introduced species (five grasses: Briza subaristata, Cynodon dactylon, Paspalum dilatatum, Phalaris minor and Setaria gracilis, and three forbs: Plantago lanceolata, Senecio madagascariensis and Sida rhombifolia) and five natives (Carex inversa, Geranium solanderi, Glycine clandestina, Microlaena stipoides and Oxalis perennans) had high abundance across both abandoned pasture and restored areas.

No significant differences in species composition were found between sites of different restoration age, and none differed from pasture., but sites of all restoration ages differed significantly from remnant vegetation.

Community structure: The pasture and restored treatments both had a significantly higher proportion of introduced species than remnant vegetation, but pasture and restored treatments did not differ. These differences were driven mainly by the larger number of native species in remnant vegetation. There was no significant trend in the number of introduced species with age among the restored sites.

Conclusions: The species composition of restored vegetation and the untreated pasture remained similar and both were significantly different from that of remnant native vegetation. There was some evidence for convergence in plant structure of restored with remnant vegetation partly attributed to plant growth. The results of this study suggest either failure of restoration treatments, or monitoring over 10 years of establishment is too short a time period to detect changes.

Note: If using or referring to this published study, please read and quote the original paper, this can be viewed at:

Output references
What Works 2021 cover

What Works in Conservation

What Works in Conservation provides expert assessments of the effectiveness of actions, based on summarised evidence, in synopses. Subjects covered so far include amphibians, birds, mammals, forests, peatland and control of freshwater invasive species. More are in progress.

More about What Works in Conservation

Download free PDF or purchase
The Conservation Evidence Journal

The Conservation Evidence Journal

An online, free to publish in, open-access journal publishing results from research and projects that test the effectiveness of conservation actions.

Read the latest volume: Volume 19

Go to the CE Journal

Discover more on our blog

Our blog contains the latest news and updates from the Conservation Evidence team, the Conservation Evidence Journal, and our global partners in evidence-based conservation.

Who uses Conservation Evidence?

Meet some of the evidence champions

Endangered Landscape Programme Red List Champion - Arc Kent Wildlife Trust The Rufford Foundation Save the Frogs - Ghana Bern wood Supporting Conservation Leaders National Biodiversity Network Sustainability Dashboard Frog Life The international journey of Conservation - Oryx British trust for ornithology Cool Farm Alliance UNEP AWFA Butterfly Conservation People trust for endangered species Vincet Wildlife Trust