Restoring Australia's temperate grasslands and grassy woodlands: integrating function and diversity


  • Suzanne M. Prober,

  • Kevin R. Thiele

  • Suzanne Prober is a research associate with The Johnstone Centre, Charles Sturt University (PO Box 789, Albury, NSW 2640, Australia, Email:, and Kevin Thiele works as a freelance ecologist for Ecological Interactions (5165 Bonang Road, Martins Creek, VIC 3888). This article was prepared to promote a broad-based approach to restoring grassy ecosystems, providing relevant contexts for goal-setting and a functional basis for identifying and overcoming barriers to restoration.

  • Box 1. Grassy White Box woodlands: what were they like before European settlement?

    Grassy woodlands with an overstorey formed by mosaics of White Box (Eucalyptus albens), Yellow Box (Eucalyptus melliodora) and Blakely's Red Gum (Eucalyptus blakelyi) were once common in the eastern half of the vast New South Wales western slopes regions, extending also into northern Victoria and southern Queensland. Because they occur on productive country, most of these woodlands have been cleared or modified for agriculture. Less than 0.05% remain in little-modified condition, and they are listed as threatened at State and Federal levels. Conservation of these woodlands relies on conserving and restoring the woodlands that still remain, and even reconstructing more of them. A first step to achieving this is simply knowing what they were like, and perhaps even how they ‘worked’.

     Because there is so little intact woodland remaining, it's hard to know exactly what the woodlands were like before European settlement. We've built a picture of woodland understoreys (Fig. 3) based on early records and what we can still see in some of the least disturbed places across the NSW western slopes. These are typically country cemeteries, rail easements, travelling stock reserves and other areas that have escaped cultivation and regular grazing by livestock (Prober & Thiele 1995; Prober 1996).

    • image(3)

    [ Diagram of little-disturbed grassy White Box woodland understorey indicating species composition and soil attributes inferred from a range of reference sites. (Photo: Courtesy S. Prober and K. Thiele.) ]


    Soils were generally soft, deep and moderate to rich in most nutrients, except that levels of available nitrogen and phosphorus were low. Nitrogen was tied up in the thick roots and other parts of the dense perennial grasses, and little became available in the topsoil in any season. Topsoils beneath trees were often higher in nutrients (total nitrogen, total carbon, available phosphorus, available potassium and cations) and pH than topsoils in open areas, probably due to factors such as nutrient concentration by extensive tree root systems (Prober et al. 2002a).

    Scattered shrubs

    On the deeper, more fertile soils, true grassy woodland shrubs such as Acacia decora and Indigofera adesmiifolia were naturally scattered or patchy in the woodland understorey. Other shrubs such as Dodonea viscosa, Cassinia arcuata and Maireana microphylla increased on shallower or drier soils or with disturbance.

    A diverse ground layer

    Over much of the NSW western slopes region, the natural dominants in the grassy White Box woodland understorey were Kangaroo Grass (Themeda australis), especially in more open or frequently burnt areas, and Snow Tussock (Poa sieberiana), which was often dominant beneath trees (Prober 1996; Prober et al. 2002a). These tussock grasses played a significant role in the functioning of this ecological community, and distinguish it from other grassy woodlands on the plains to the west. They influenced levels of important soil nutrients, protected the soil surface, and provided habitat for birds, reptiles and invertebrates. A wide range of other native plants relied on the gaps between the tussocks, and their diversity was particularly high where the thick thatch of Kangaroo Grass was occasionally removed by fire, or where the dominant grasses were less competitive for other reasons (e.g. beneath trees or on less fertile soils).

    Many of the woodland forbs and grasses were naturally widespread across the region, but some were more prominent in southern winter rainfall areas, and others in northern, summer rainfall areas. On heavy black clays in some northern areas, grasses like Plains Grass (Austrostipa aristiglumis), Blue Grass (Dichanthium sericeum) and Native Sorghum (Sorghum leiocladum) became more prominent (Prober 1996).

  • Box 2. Grassy White Box woodlands: how has the understorey changed?

    With European settlement, the once abundant grassy White Box woodlands (Box 1) were transformed into a productive landscape of crops and pastures. Even in areas that weren't cleared or cultivated, livestock grazing and other landuses caused many changes in the woodland understoreys and the soils they grew in. Native diversity was reduced to varying extents under different grazing or cultivation regimes, soils were depleted or enriched depending on management and other disturbance, shrub cover was lost, and different grass species, both native and introduced, became dominant under different conditions (Prober & Thiele 1995; Prober et al. 2002b). One key change across all types of degraded remnants has been an increase in soil nitrate levels, which in turn encourages the growth of annual exotic plants (Prober et al. 2002b). Figure 4 describes some of the different ways that the woodlands have changed and identifies some of the degrading processes. Other types of changes, such as replacement by perennial pasture species or invasion by perennial exotics such as Coolatai Grass (Hyparrhenia hirta), have also occurred.

    • image(4)

    [ Five typical degraded states of grassy White Box woodland understorey, indicating some associated changes in topsoil properties. Note that these changes don't necessarily occur in a linear sequence, rather, they can result from differing management influences as indicated. (Photo: Courtesy S. Prober and K. Thiele.) ]

  • Box 3. Approaches for restoring Kangaroo Grass/Snow Tussock understoreys in grassy White Box woodlands

    Understanding changes that have occurred in grassy White Box woodlands (Box 2) provides us with clear directions for determining ecological methods for their restoration. Common goals in restoring woodland understorey are to increase native plant diversity and cover, to decrease exotic abundance, to re-establish scattered shrubs, and to re-establish the original perennial dominant grasses Kangaroo Grass and Snow Tussock.

     Modifying conditions to favour desirable native species and disadvantage exotics is a key to long-term success. We can use tools such as fire, carbon addition and seed reintroduction to restore more natural soil nutrient levels, soil seed banks, and competitive environments. These techniques can be used in conjunction with agronomic techniques such as cultivation, herbicides and direct drilling, or can be applied directly to remnants to help tip the balance to favour native species over exotics. Below, we introduce some restoration techniques that we have begun to experiment with, focusing on reducing annual exotics. Annual exotics are important competitors with native species in many degraded remnants. They are ecologically very different from most native plants, which are mostly perennial species with lower nitrogen requirements.

    Soil nitrate depletion

    In most degraded remnants, soil nitrate rises to high levels over the summer and autumn, encouraging lush growth of annual exotics as they germinate in autumn. Suppressing this nitrate peak is particularly effective for reducing the vigour of annual exotics, and hence enhancing establishment and competitiveness of desirable natives (Fig. 5). We can reduce soil nitrate temporarily by adding sugar or other carbon sources (e.g. sawdust) to the soil. This increases the carbon: nitrogen ratio, causing soil microbes to flourish and use up any available nitrogen. In preliminary field trials we added 0.5 kg sugar per m2 at 3-monthly intervals, but further studies are needed to determine effective minimum rates of carbon application for different types of remnants (S. Prober, K. Thiele & I. Lunt, unpub. data, 2004).

    • image(5)

    [ Addition of sugar or other carbon-rich materials to the soil temporarily reduces available soil nitrogen by encouraging growth of soil microbes. Plots on the left (a, b) were not treated with sugar and support robust annual exotics. Plots on the right were initially similar to those on the left but regular sugar applications led to (c) dramatically reduced weed growth after 1 year, and (d) successful establishment and growth of re-seeded native Kangaroo Grass within 2 years. (Photo: Courtesy S. Prober and K. Thiele.) ]

    Seed bank manipulation – Exotics

    Degraded remnants usually support a large seed bank of annual exotics (Lunt 1990). The seed bank of one important group of exotics, the cool-season annual grasses (e.g. Wild Oats and Bromes), can be reduced by burning in spring before established plants set seed (Fig. 6). Pulse grazing, herbicides or repeated slashing in spring may be similarly effective. Techniques involving spring biomass removal are best for sites with few broadleaf annual exotics, as many broadleaf annuals have longer-lived seed banks and can increase on bare soil (Prober et al. 2004). Scalping (removal of a complete thin layer of topsoil from a remnant) is another technique that has been tried for removing weed seed banks. This technique may also influence soil nutrient levels, but effects on soils or vegetation in grassy woodlands have not yet been well-documented.

    • image(6)

    [ Spring burns can be used to minimize seeding and discourage re-establishment of exotic annual grasses in weedy remnants. They are relatively safe to undertake because the surrounding countryside is still green but can be difficult to achieve. For small areas, a gas-powered weed burner is effective, as in this degraded remnant near Young, NSW. For larger areas, weeds can be dried-off before burning using knock-down herbicides or steam. Burns should be undertaken in patches to minimize soil erosion and impact on native fauna. (Photo: Courtesy S. Prober and K. Thiele.) ]

     Manipulation of weed seed banks is best accompanied by augmentation with seed of native perennials (especially Kangaroo Grass), to ensure the exotics are replaced by natives rather than other weeds.

    Seed bank manipulation – Natives

    Native species that no longer persist on a site are unlikely to return quickly without assistance – their seed banks are short-lived (Lunt 1997), and their distribution throughout the landscape is too sparse to rely on natural seed dispersal. Thus, reintroduction of seed of desired species is usually a necessary part of the restoration process. Native seed is best sourced locally. If seed is limited, restoration can begin with small patches and further seed harvested as patches mature.

    Perennial native sward establishment

    Re-establishment of a dense sward of native perennial grasses, particularly the original dominants Kangaroo Grass and Snow Tussock, is important for restoring the natural functioning of the grassy woodland ecosystem. In particular, dense native grasses are likely to play an ongoing role in out-competing weeds, through above-ground competition and by maintaining low levels of available nitrogen and phosphorus. Preliminary data suggest, for example, that Kangaroo Grass is superior to other native or exotic grasses for reducing soil nitrate levels. Techniques for re-establishing Kangaroo Grass are increasingly well known. It has large, awned seeds that drill themselves into the soil, making cultivation or direct-drilling less critical for their establishment. Spring burns can enhance germination by releasing seed dormancy, as well as controlling annual grasses. Carbon amendment (e.g. sugar application) or herbicides can enhance establishment and growth by reducing competition with weeds (Cole & Lunt 2005; S. Prober, K. Thiele & I. Lunt, unpub. data, 2004). Establishment needs of Snow Tussock are currently poorly understood.

    Perennial exotic control

    No ecological methods are yet known for dealing with serious invasive perennials such as St. John's Wort (Hypericum perforatum) or Coolatai Grass (Hyparrhenia hirta), as these species are similar to native plants in their ecology. All invasive perennial exotics should be dealt with promptly with the aid of herbicides or hand removal, otherwise they may heavily infest the site and can build up long-lived seed banks. Avoiding disturbances that encourage their invasion is likely to be one important preventative measure (e.g. Coolatai Grass often begins invading after roadside scalping). Development of biological control methods for invasive exotics is another unexplored option for addressing these threats.


Long-term grassy woodland researchers explain how restoration of grassy ecosystems can be improved through an understanding of their key natural patterns and processes, and the ways these change during degradation.


Sub-humid temperate grasslands and grassy woodlands have the unfortunate honour of occurring on some of the most productive agricultural lands in south-eastern Australia. From coastal rainshadow woodlands, through treeless basalt plains, to inland floodplain fringes, they have been extensively cleared and subject to degradation through agricultural and urban development. As a result, these are now some of the most threatened ecological communities in Australia (Benson 1991), and ecological restoration is an essential component of efforts towards their conservation.

Many landholders and land managers are aware of the dire ecological consequences of 200 years of European settlement on the condition and biodiversity of these ecological communities and the landscapes in which they occur. For a range of reasons (economic, social or biological), many individuals and organizations are also prepared to do something about this widespread degradation, offering their time, their land and their effort, sometimes supported by government funding schemes and programs.

However, ecologists, planners and practitioners alike are all too often at a loss as to how we might best enhance and restore these grassy ecosystems. In this article, we suggest an approach for addressing the restoration challenge in Australia's temperate grasslands and grassy woodlands, emphasizing the importance of both ecological function and ecological diversity at a range of scales. We incorporate current progress in this approach, and illustrate it with our experiences towards understorey restoration in grassy White Box woodlands of NSW (Boxes 1–3).

Temperate grasslands and grassy woodlands in Australia

Australian vegetation is usually classified on the basis of structure or floristic composition of overstorey species (Groves 1981). However, the range of ecological communities most devastated by agricultural development in southern Australia are increasingly distinguished by their prominent and diverse ground layer dominated by perennial tussock grasses. These are the sub-humid temperate grasslands and grassy woodlands (hereafter grasslands and grassy woodlands) in which the shrub layer is sparse to discontinuous, and trees (to 60% cover) may or may not be present.

Grasslands and grassy woodlands comprise a wide range of distinct ecological communities, predominating in agricultural regions of South Australia, Victoria, Tasmania, NSW and southern Queensland, with possible historic occurrence in south-western Australia (Groves 1981; Hobbs & Yates 1999). From an ecological perspective, these ecosystems can be considered in broad groups reflecting the functional significance of the natural dominant perennial grasses.

The most well known of the grasslands and grassy woodlands are those that were naturally dominated by Kangaroo Grass (Themeda australis, also known as T. triandra) and Snow Tussock (Poa sieberiana). These species predominated in the higher rainfall grasslands and grassy woodlands of temperate south-eastern Australia. The prominence of Kangaroo Grass and Snow Tussock decreased in drier grasslands and grassy woodlands (< 550 mm rainfall), giving way to communities dominated by other native grasses, particularly Wallaby Grasses (Austrodanthonia spp.), Spear Grasses (Austrostipa spp.) and Windmill Grasses (Enteropogon spp.), probably with a denser shrub component (e.g. Figure 1). Other distinctive grasslands and grassy woodlands are thought to have occurred patchily in areas with heavy cracking-clay soils, for example, the Liverpool Plains of central NSW (Hobbs & Yates 1999; Keith 2004).

Figure 1.

A remnant of Grey Box (Eucalyptus microcarpa) woodland in a travelling stock reserve near Temora, NSW. The understorey is dominated by Wallaby and Spear Grasses, interspersed with a range of other native grasses and forbs. Large trees, scattered shrubs, tussock grasses and woody debris provide habitat for a range of fauna. Remnants like these help us understand natural patterns and processes in grassy woodlands. (Photo courtesy S. Prober and K. Thiele.)

Integrating function and diversity in restoration

Restoring grassy ecosystems across south-eastern Australia is important for a range of purposes, particularly for maintaining productive landscapes and conserving biodiversity (Hobbs & Norton 1996). This article focuses on restoring grasslands and grassy woodlands for biodiversity conservation. A logical goal in this regard is to shift species composition in sites or landscapes towards appropriate ‘reference’ states, often reflecting observed or inferred predisturbance states with few weeds and a high diversity and cover of native species (Aronsen et al. 1993). While a complete return to reference systems may not be achievable or desirable, augmenting fauna habitat, enhancing native species diversity, and controlling exotic species are often realistic goals at site scales, while increasing the area, connectivity and diversity of vegetation across catchments are common goals at landscape scales.

Beyond these general statements though, a broad-based approach is needed to maximize restoration outcomes across ecological communities and landscapes. Such an approach needs to establish appropriate ecological contexts for decision-making, and to encompass ecological function for ensuring on-ground success.

Past and present patterns of species composition and diversity in grassy ecosystems provide the ecological context against which we make restoration decisions, including setting regional priorities and identifying appropriate site targets. These patterns include natural variation in species composition within target ecological communities, relationships among different ecological communities and their environments within regions, and the ways and extent to which these ecosystems have changed in response to different degrading influences. Understanding patterns in composition and diversity is, therefore, especially important at the scales of the ecological community and the landscape, and can also be relevant at the site scale (Table 1).

Table 1.  Both ecological patterns and processes influence restoration decisions and outcomes. Patterns in species composition and diversity provide appropriate contexts for decision-making, particularly at the scale of the ecological community and the landscape. Processes direct our choice of restoration techniques and are most relevant at site and landscape scales. A bioregional view can act as a compromise between landscape and ecological community considerations
 Ecological contexts – patterns of diversity and changeEcological function – processes driving species composition
Ecological CommunityOutcomes across whole ecological communities are maximized through consideration of patterns of variation and change, for example, for identifying priorities, predicting local species composition, and interpreting the significance of sites. 
LandscapePriority setting within landscape or regional contexts directs restoration to the ecological communities most in need, and aims to optimize landscape diversity and configurations and associated functional outcomes.Addressing landscape processes is critical for maintaining ecosystem viability in the long term, and may be necessary before site-scale goals can be achieved, for example, to ameliorate hydrological processes, nutrient transfers or tree decline.
SiteLocal heterogeneity can reflect ecological function so can be an important goal in restoration, for example, patches of shrubs within a matrix of tussock grasses may be a necessary combination of shelter and food sources for some woodland birds.Manipulating internal ecological processes (e.g. nutrient cycling, plant–animal interactions) is important for promoting conditions that favour the desired species composition and increasing resistance to unfavourable external influences.

Ecological function is the way an ecosystem works, and provides the ecological basis for its repair (Table 1). Restoration of grasslands and grassy woodlands to date has focused on removing degrading influences (e.g. through fencing) and planting trees and shrubs. This approach contributes to restoring landscape processes, and limits further decline in remnants. However, return to a diverse native species composition rarely occurs through these interventions alone, and exotic species often persist in the long term (e.g. Spooner et al. 2002). Thus, we must also address key underlying ecological processes that are associated with degradation, aiming to re-establish ecosystem functions that favour the target species composition.

Ecological function is critical at both site and landscape scales (Table 1). Clearly, restoration of sites will be ineffective if whole landscapes are dysfunctional, justifying the emphasis on restoring landscape function in agricultural landscapes in the past decade (Hobbs 1993a; Saunders & Hobbs 1995). Nevertheless, landscape restoration is ultimately dependent on coordinated restoration and reconstruction of sites, and the effectiveness of these efforts for supporting native biodiversity will often depend on site-scale issues such as habitat quality and local soil conditions.

Developing a broad understanding of ecosystem patterns and processes in grasslands and grassy woodlands is beyond the scope of any one individual or community group. Rather, diversity and function form the two main ecological pillars of a collective approach towards maximizing restoration outcomes in these ecosystems. As we detail further below, we suggest four steps towards building these pillars, and consequently identifying optimal restoration goals and techniques for achieving on-ground success: (i) identifying what the target ecosystems were like before degradation; (ii) understanding how and why the target ecosystems have changed; (iii) restoring diversity through process; and (iv) adaptive restoration.

The experiences of on-ground workers, planners and ecologists over the past two decades have already led to considerable progress relevant to this approach. The importance of landscape function is increasingly recognized, and patterns of species composition in some grassy ecological communities have been studied in detail. Nevertheless, landscape restoration will always be a mammoth task, and many other aspects of grassy ecosystem restoration are still poorly understood. Thus, ecologists and planners must continue to lead in building the relevant ecological context and functional frameworks that underpin on-ground works, and practitioners must contribute through the general knowledge and local detail that inevitably accumulates with experience.

Steps towards maximizing restoration outcomes at local to national scales

What were the target ecosystems like before degradation?

Knowing where to begin in any restoration program can be difficult. A good starting point is getting to know the ecological communities and landscapes of interest and how they varied before European settlement. This provides the detail about patterns of species composition and diversity needed for effective and coordinated decision-making at various scales (Hobbs & Norton 1996). This step is often taken for granted, but in reality, inadequate supporting information often leads to less than optimal, or even misguided, goals (e.g. dense shrub plantings to enhance grassy woodlands). As well, this information provides the basis from which to begin the more difficult task of identifying key ecological processes associated with degradation and restoration.

The value of predisturbance reference ecosystems for restoration is sometimes questioned because determining what the ‘natural’ ecosystems were like can be difficult and recreation of past conditions may be impossible or futile (Hobbs & Norton 1996). However, European settlement has influenced ecosystems over much shorter timeframes in Australia than in the Northern Hemisphere, so restorationists in Australia can be more optimistic about inferring a picture of predisturbance ecosystems. Where reasonable inferences are possible, this information is well worth seeking for the insights it can give us, even if it is unrealistic to restore degraded remnants to such condition.

Understanding natural patterns of diversity in grasslands and grassy woodlands

Remnant patches that have escaped major influences of agricultural development have proven invaluable for understanding the natural diversity of grasslands and grassy woodlands. Sites such as rail easements and cemeteries that were fenced from livestock grazing early in the history of settlement have yielded considerable information regarding indigenous floristic composition and soil characteristics, particularly in Kangaroo Grass/Snow Tussock ecosystems (e.g. Box 1; Moore 1953a; Stuwe & Parsons 1977; Kirkpatrick et al. 1988; Benson 1994; Prober 1996), and to a lesser extent in Wallaby Grass/Spear Grass ecosystems (e.g. Figure 1; Moore 1953a; Davies 1999; Prober & Thiele 2004). Larger, intermittently grazed remnants such as travelling stock reserves have also been important, particularly for groups such as birds and mammals that need larger remnants to survive.

Viewed together with other elements in the landscape and augmented by historical information, this collection of remnants also provides a broader picture of the way grassy ecosystems varied across environments, seasons and ecological communities, as well as indicating natural patterns of genetic diversity in woodland and grassland species (e.g. Prober & Brown 1994; Young et al. 1999).

Understanding natural processes in grasslands and grassy woodlands

Knowledge of the predisturbance ecological community and its landscape context also provides useful benchmarks for understanding the way ecosystems function. Dominant structural elements such as trees and perennial grasses can dramatically influence ecological processes at both site and landscape scales, so knowing what these species were, their distribution in the landscape, and their influence on soils, hydrology and other species is a first step towards understanding these processes (e.g. Box 1). Similarly, historical records indicate a once abundant and diverse fauna of medium-sized ground-dwelling mammals (Robinson & Traill 1996; Lunt & Bennett 1999), which must surely have influenced ecological function. Relatively unmodified remnants provide reference points that indicate site-scale processes important for maintaining these key functional groups, as well as processes that support natural diversity in these systems (e.g. seed banks and seedling recruitment; Lunt 1997; Morgan 1998, 200; Clarke & Davison 2004).

There remains much scope for understanding diversity and function of less obvious biological groups in grassy ecosystems, particularly invertebrates, reptiles and soil organisms (Fig. 2), using these kinds of reference systems. Basic descriptive information is also lacking for some grassy communities, and our knowledge of natural site and landscape processes is still limited. Understanding predisturbance ecosystems, therefore, remains an integral part of building a broad ecological context for restoration.

Figure 2.

The rare Striped Legless Lizard (Delma impar) lives beneath rocks or debris in Kangaroo Grass grasslands, feeding on spiders and insects. The needs and functions of these less conspicuous types of organisms in grassy ecosystems are generally poorly understood. (Photo: Courtesy Andrew Tatnell/Big Island Photographics.)

How and why have the target ecosystems changed?

For the purposes of restoration, it is impractical to understand all ecological processes in reference ecosystems. Rather, the association of particular patterns and processes with change and degradation, inferred from comparisons between reference and degraded ecosystems, provides the basis for setting restoration goals and key clues for designing restoration techniques.

Change between different states of an ecological community can be viewed as gradual and successional (Luken 1990), or rapid and perhaps associated with an ecological threshold (Westoby et al. 1989). The former more traditional view often sees restoration as adjusting species and environments to set a site back onto a trajectory of natural successional processes. In the latter view, identifying ecological thresholds is a key to determining restoration methods. Whichever view is relevant, understanding how and why ecosystems have changed is a logical next step towards successful restoration.

Patterns of change in grasslands and grassy woodlands

Clearing, livestock grazing and cultivation have been the main drivers of change in agricultural landscapes. This has led to severe fragmentation of many once continuous grassy ecosystems, leaving only scattered remnants within a matrix of crops and introduced pastures. In regions with less extensive cropping and pasture improvement, landscape changes are more diffuse, leading to variegated landscapes that support more native species within the general landscape matrix (McIvor & McIntrye 2002). While much of the clearing and landuse change in these regions occurred over 50 years ago, continued clearing, loss of paddock trees, and increasing landuse intensity continue to contribute to landscape change today (Robinson & Traill 1996; Reid & Landsberg 2000; Gibbons & Boak 2002).

Faunal change in most grassy ecosystems has been particularly dramatic. Medium-sized ground-dwelling mammals have all but disappeared, with 14 of 19 species in Victoria now extinct. Similarly, at least 25% of woodland birds are in serious decline and woodland-specific arboreal mammals have become threatened, while introduced mammals such as foxes and rabbits have become widespread. Consequently, native faunal assemblages have become grossly simplified and impoverished at the site scale (Robinson & Traill 1996; Lunt & Bennett 1999).

Floristic changes in Kangaroo Grass/Snow Tussock ecosystems due to livestock grazing are relatively well understood. Early work described key changes in the dominant grasses, with rapid breakdown of the Kangaroo Grass/Snow Tussock sward, its replacement by secondary native perennial grasses, and increasing abundance of exotic annuals (Moore 1953b; Moore 1970). Later work has built on these observations, indicating a range of modified states (e.g. Box 2; Lodge & Whalley 1989; Prober et al. 2002b) and focusing on changes in subsidiary species that contribute the greatest diversity to the grassy sward (e.g. Kirkpatrick et al. 1988; McIntyre & Lavorel 1994; Lunt 1995; Prober & Thiele 1995). Changes in Wallaby Grass/Spear Grass communities are less well understood but include reduced diversity of native grasses, decline in many native perennial forbs, and an increase in native and exotic annuals (e.g. Moore 1953b; Prober & Thiele 2004). Floristic changes in other grassy communities and effects of other management changes (e.g. fire or fertilizer regimes) are still poorly understood. Invasion by introduced perennial plants such as Coolatai Grass (Hyparrhenia hirta) and African Lovegrass (Eragrostis curvula) are also of serious concern in many grassy ecosystems.

Underlying processes associated with change in grasslands and grassy woodlands

While clearing, cultivation and livestock grazing are the main direct causes of degradation in grasslands and grassy woodlands, these processes lead to various, less obvious changes that can threaten the survival of remaining natural communities and hinder restoration efforts even after initial disturbances have ceased. Often, it is these underlying changes that must be addressed before superficially restored ecosystems can become self-sustaining.

Widespread tree clearing and loss of deep-rooted perennial grasses leads to disruption of a host of physical processes within the landscape, affecting microclimates, soil nutrient flows and hydrology, leading to devastation through salinization, and threatening remaining flora and fauna (Hobbs 1993a). Habitat fragmentation resulting from clearing and changed landuse is central to ongoing faunal decline, e.g. through disrupted dispersal, inadequate habitat size or discontinuity of seasonal food resources (Lunt & Bennett 1999). As well, population fragmentation affects gene flow, and can lead to reduced genetic diversity, inbreeding and local extinctions (e.g. Prober & Brown 1994; Young et al. 1999; Burrows 2000).

Livestock grazing causes floristic change at the site scale through many different processes. These include direct effects related to palatability and defoliation tolerance, and indirect effects resulting from soil compaction and disturbance, weed invasion and modified nutrient cycling (Greenwood & McKenzie 2001; Prober et al. 2002b). Reduced habitat quality due to grazing management, particularly loss of structural diversity, also causes further faunal decline (Robinson & Traill 1996; Lunt & Bennett 1999; MacNally et al. 2001).

Crucially, changes in species composition resulting directly or indirectly from clearing and grazing can themselves modify ecological processes. Indeed, feedback loops initiated by changed management can lead to relatively stable ecosystems that favour a different species composition from that of the original ecological community, and it is often these alternate stable states and processes that require attention during restoration.

Soil nitrogen, an important limit to plant growth in grassy ecosystems worldwide, is a key example in this regard. Increases in available nitrogen can be initiated by soil disturbance or nutrient inputs. This encourages growth of nitrogen-loving annual exotics, which in turn promote seasonally high available soil nitrogen concentrations through the breakdown of plant material after they die each year. Seasonally high nitrogen then favours the ongoing persistence of annuals over native species (Prober et al. 2002b and references there-in). Similarly, disturbance and subsequent loss of many native plant species from the vegetation leads to their rapid disappearance from soil seed banks, as seed banks of most grassland and grassy woodland plants are short-lived (Lunt 1990, 1997; Morgan 1998; Clarke & Davison 2004). Seed banks become dominated by exotics (Lunt 1990), favouring the ongoing persistence of these species.

Building a picture of the way ecosystems change and what drives these changes can be facilitated by the use of state and transition models, which depict a range of ecosystem states within an ecological community and the mechanisms that cause changes from one state to another (e.g. Box 2; Lodge & Whalley 1989; Westoby et al. 1989; Prober et al. 2002b). These descriptive models allow accumulation of information across projects and experiences and offer a framework for interpreting the significance and needs of remnants.

Restoring diversity through process

Understanding predisturbance grasslands and grassy woodlands and how they have changed provides the context for setting restoration goals and the functional basis for identifying and overcoming barriers to restoration.

Goals and priorities will be influenced by the status of different elements of an ecological community and the degree of landscape degradation. Thus, for an ecological community, a broad goal might be to restore a range of remnants representing its original patterns of variation and diversity. Site targets will be based on this knowledge of patterns in species composition, ideally also allowing for incipient problems such as climate change. In fragmented landscapes, broadscale loss of perennial vegetation has particularly severe consequences for landscape processes, requiring extensive revegetation. In variegated systems, more native perennial species survive in the landscape matrix, and augmentation of existing habitat may be a greater priority (Hobbs 1993a; McIvor & McIntyre 2002).

For achieving these goals, a long road lies ahead in identifying and restoring critical ecological processes but some solutions can already be inferred and acted upon.

Landscape restoration is a huge task that will take decades to take effect. The past 20 years has seen progress towards this goal but rates and scales of revegetation and tree regeneration need to be substantially increased (Saunders & Hobbs 1995; Robinson & Traill 1996; Reid & Landsberg 2000), augmented by enhanced utilization and management of deep-rooted perennial native pastures and tree crops (e.g. Dorrough et al. 2004). Information regarding effective landscape design for supporting different species and ecological processes is increasing, and needs to be integrated into revegetation and restoration designs to simultaneously address multiple issues, such as salinity, habitat connectivity and tree decline (Hobbs 1993b).

At the site scale, local projects have targeted fauna habitat through provision or retention of nesting hollows, protective shrubs and woody debris (e.g. MacNally et al. 2001), and a coordinated approach to fauna enhancement across landscapes would be valuable. For enhancing floristic values, modification of grazing regimes can promote native species that are still locally present (Dorrough et al. 2004), while augmentation of native propagules is needed for restoring species that have disappeared from soil seed banks and surrounding areas. However these actions alone will often be inadequate for restoring plant diversity, as evidenced by the failure of many forbs and grasses to spread significantly in restored Kangaroo Grass grasslands (McDougall & Morgan 2005).

Likely reasons for such failures are competition with exotics and shifts in underlying ecological conditions such as soil nutrients, weed seed banks, and establishment conditions. Reducing the prevalence of annual exotic grasses through spring biomass reduction, and reducing available soil nitrogen to disadvantage nitrogen-loving exotics, are promising approaches to enhancing grasslands and grassy understoreys through a focus on underlying ecological processes (Box 3; Davies 1999; Prober et al. 2004).

Preliminary evidence also illustrates the importance of restoring key functional species for achieving necessary shifts in ecosystem function to favour native plant species (Aronsen et al. 1993). For example, long-term restoration of soil nitrogen cycles may be dependent on restoring a dense sward of perennial native tussock grasses, particularly Kangaroo Grass, as this species may be superior to other native grasses for reducing soil nitrate levels (S. Prober, K. Thiele & I. Lunt, unpub. data, 2004). Agronomic techniques using cultivation, direct drilling and herbicides also have a potential role in this regard, by facilitating establishment of key species that subsequently ameliorate ecosystem processes. In this context, identifying the most effective species for achieving desired ecological change is an important aspect of restoration.

Adaptive restoration

Adaptive management approaches advocated for refining management strategies in grassy ecosystems are equally relevant to restoration. Appropriate coordinating frameworks are needed to encourage monitoring and feedback so that each restoration program or project can become part of a more general learning process. Where suitable restoration techniques are not well known, practitioners might compare alternatives. Ecologists can help by designing simple experimental and monitoring components for on-ground works, and building broader frameworks from which local projects can draw. Failures may be disappointing but can act as important flags for mismatches between desired species composition and ecological conditions or processes, which in turn provide direction for alternative techniques or adjustment of restoration goals.

Concluding remarks

Coordinated priority and goal setting within broad ecological contexts, and addressing ecological function through attention to key ecological processes, are essential for increasing the effectiveness of the restoration dollar across sites, landscapes and ecological communities. Nevertheless, restoration planning must be undertaken with the knowledge that some ecological changes are irreversible, and that addressing others will be expensive. For example, it is more cost-effective to avoid conditions that promote invasion by serious environmental weeds than to remove these weeds from sites they have already infested.

Even when restoration goals are achieved, restored sites and landscapes generally lack the entire composite of species and processes of the target ecological community. Restoration must, therefore, be viewed as just one component of the conservation tool-kit for grasslands and grassy woodlands. It remains critical to maximize conservation of existing remnants, and to limit ongoing advancement of degrading processes. A green light to swapping good quality remnants for ones that might be restorable someday is not a desirable outcome of restoration efforts. Rather, restoration must aim to increase the overall extent, representativeness and quality of these threatened ecological communities.