Question: Can vegetation changes that occur following cessation of cultivation for cereal crop production in semi-arid native grasslands be described using a conceptual model that explains plant community dynamics following disturbance? Location: Eighteen native grasslands with varying time-since-last cultivation across northern Victoria, Australia. Methods: We examined recovery of native grasslands after cessation of cultivation along a space fortime chronosequence. By documenting floristic composition and soil properties of grasslands with known cultivation histories, we established a conceptual model of the vegetation states that occur following cessation of cultivation and inferred transition pathways for community recovery. Results: Succession from an exotic-dominated grassland to native grassland followed a linear trajectory. These changes represent an increase in richness and cover of native forbs, a decrease in cover of exotic annual species and little change in native perennial graminoids and exotic perennial forbs. Using a state-and-transition model, two distinct vegetation states were evident: (1) an unstable, recently cultivated state, dominated by exotic annuals, and (2) a more diverse, stable state. The last-mentioned state can be divided into two further states based on species composition: (1) a never-cultivated state dominated by native perennial shrubs and grasses, and (2) a long-uncultivated state dominated by a small number of native perennial and native and exotic annual species that is best described as a subset of the never-cultivated state. Transitions between these states are hypothesized to be dependent upon landscape context, seed availability and soil recovery. Conclusions:Legacies of past land use on soils and vegetation of semi-arid grasslands are not as persistent as in other Australian communities. Recovery appears to follow a linear, directional model of postdisturbance regeneration which may be advanced by overcoming dispersal barriers hypothesised to restrict recovery.

The effects of stock grazing on native grassy ecosystems in temperate southern Australia are well documented. However, less is known about the potential of ecosystems to recover after a long history of stock grazing and, in particular, whether the removal of stock will have positive, negative or neutral impacts on biodiversity. We examined the response of understorey vegetation to the removal of sheep grazing in a herb‐rich Eucalyptus camaldulensis (red gum) woodland in western Victoria. Using a space‐for‐time chronosequence, woodlands were stratified into groups based on their time‐since‐grazing removal; these were long‐ungrazed (>20 years), intermediate‐time‐since‐grazing (9–14 years), recently ungrazed (5 years) and continuously grazed. We found significantly higher species density in long‐ungrazed sites relative to sites with a more recent grazing history. No differences were found in species density between continuously grazed sites and those ungrazed in the previous 14 years. Species composition differed with time‐since‐grazing removal and indicator species analysis detected several native species (including tall native geophytes and herbs) associated with long‐ungrazed sites that were absent or in low abundance in the more recently grazed sites. Seven of the eight species significantly associated with continuously grazed sites were exotic. Removal of sheep grazing in red gum woodlands can have positive benefits for understorey diversity but it is likely that recovery of key indicators such as native species will be slow.

Summary  Since the mid 1990s, there has been a significant increase in the area of semi‐arid grasslands included in the National Reserve Systems in the Victorian Riverine Plain. This expansion has not been matched by an improved understanding of the alternate disturbance regimes that might produce better outcomes for native ecosystem conservation. Over the past 150 years, stock grazing has completely replaced fire in these grasslands. As a result, the impact of fire on native (and exotic) plant biodiversity is little understood. This study compared the current grazing regime (i.e. ‘status quo’) with burning and the removal of grazing (‘deferred’ management) across three grasslands in the Victorian Riverine Plain to determine the effects of short‐term exposure to alternate disturbances on community structure. Our results showed little change in species density, composition or abundance under the three disturbance treatments. A long exposure to stock grazing may have reduced the abundance of species likely to respond positively to burning. The cover of the biological soil crust responded positively to fire; such changes are known to significantly influence establishment and the functional composition of communities. As such, further investigation of the functional attributes of these communities may broaden our understanding of short‐term responses to alternate disturbance events. To better understand the utility of fire as a management tool, a long‐term commitment to expanding the implementation of this regime from its current extent will greatly increase the understanding of alternate disturbances in this landscape.

Summary A key task for native grassland managers is to assess when biomass reduction is necessary to maintain plant and animal diversity. This requires managers to monitor grassland structure. Parks Victoria and La Trobe University developed a method for rapid assessment of grassland structure using golf balls. Baker‐Gabb et al. (Ecological Management & Restoration, 17, 2016, p235) provide an example of where the method has been used to manage grassland structure to favour an endangered bird, the Plains‐wanderer (Pedionomus torquatus). In this study, we provide further critical analysis of the method using three data sets collected across different parts of Victoria that relate golf ball scores to various habitat attributes. We demonstrate how the golf ball score provides a good surrogate for key aspects of grassland structure. We show that the method does not provide a reliable surrogate for above‐ground biomass or vegetation cover, although we discuss how biomass and cover are not particularly good indicators of grassland structure. We argue that elements of grassland structure may be better correlated with desired conservation outcomes (e.g. plant species diversity or the presence of a particular species) than biomass or cover alone. We discuss examples of how the golf ball method has been used, and how it can be improved. The method will be particularly useful where a link can be demonstrated between golf ball scores and desired conservation outcomes, such as in the case of the Plains‐wanderer.