Summary1. Characterization of the seed bank is one of the most important demographic assessments that can be undertaken for a plant community. Overlapping generations, evidence of past above-ground vegetation and histories of invasion and disturbance are recorded in the seed bank. 2. Two broad approaches have been used to elucidate seed bank components -sifting-sorting techniques and germinability assays. The utility of these approaches varies with community type and habitat although a common theme among studies has been the quest for an efficacious method. Here, we compare the two approaches for semi-arid ephemeral wetlands: seed extraction through flotation and seedling emergence. 3. Species composition of the soil seed bank differed dramatically depending on the technique, with only 19 species common to both methods and a total of 66 species detected using both procedures. 4. Both techniques provided similar estimates of seed density and species richness of the seed bank in the top 5 cm of soil. However, samples collected from 5 cm to 20 cm had lower seed densities using the flotation technique than with the seedling emergence technique. 5. Differences in seed detectability between the two approaches may be related to seed size, seed dormancy and specific germination requirements. 6. The community composition of soil seed banks for ephemeral wetlands depends on the choice of technique.

Between 2000 and 2002, central Australia experienced the largest fire season in three decades when ~500 000 km2 burned. The effects of these and preceding wildfires in the 1980s on spinifex (Triodia spp.) sand-ridge plant communities were examined at 38 sites in central Australia. We used both multivariate and univariate techniques to assess floristic differences among sites of contrasting time-since-fire, fire season and fire interval. Time-since-fire had a consistent floristic influence across the landscape, with increased abundances of ephemeral grasses and forbs and Triodia seedlings, and species richness soon after fire but decreasing long after fire. Fire season had little effect on most functional groups of plants, although seedlings of woody species were significantly more abundant following summer than winter fires. Likewise, recent short fire intervals appeared to have little impact on the population dynamics of most functional groups, although some transient effects were observed on abundances of ephemeral forbs, Triodia seedlings and herbaceous clonal species. Long-term woody species abundances appeared to be affected by short fire intervals in the 1980s when repeated fires seemed to stimulate recruitment of some resprouting species. The present study highlighted the relative stability of spinifex vegetation types in the face of landscape-scale pyric perturbation, but emphasised that localised shifts in the composition and structure of the plant community may occur under certain fire regimes.

The hummock grasslands of arid Australia are fire-prone ecosystems in which the perennial woody plants mostly resprout after fire. The resprouting ability among these species is poorly understood in relation to environmental variation; consequently, little is known about the impacts that contemporary fire regimes are having on vegetation within these systems. We examined the resprouting ability of adults and juveniles of four widespread Acacia species (A. aneura, A. kempeana, A. maitlandii, A. melleodora) by experimentally testing the effects of fire severity, interval and season. We found that fire severity and season strongly affected survival, but the magnitude of the effects was variable among the species. Unexpectedly, a short fire interval of 2 years did not have a strong negative effect on resprouting of any species. Fire severity had variable effects among the four species, with those species with more deeply buried buds being more resilient to high-severity soil heating than those with shallow buds. Season of fire also strongly affected survival of some species, and we propose that seasonal variation in soil heating and soil moisture mediated these effects. The species by environment interactions we observed within one functional group (resprouters with a soil-stored seed bank) and in one genus suggest that modelling landscape response to fire regimes will be complex in these arid ecosystems. We predict, however, that the dominant resprouting acacias in hummock grasslands of central Australia are highly resilient to a range of fire regimes.

Questions: The relationship between fire, aridity and seed banks is poorly understood in plant community ecology. We tested whether there was a close correspondence between the seed bank and standing vegetation composition with time‐since‐fire in a desert. We also examined whether longer‐lived species showed seed limitation relative to more ephemeral species, as this could influence grass‐woody ratios in a major biome. Location: Dune hummock grasslands/shrublands of central Australia. Methods: The effects of time‐since‐fire on floristic and functional group composition were examined by comparing plots unburned since 1984 against plots that had been burned in 2002. Three methods were used to quantify seed abundances: a germination trial using heat and smoke application, a flotation method, and a sieving method. Results: Seed bank densities were very low (<3000 m−2). Species similarity between the seed bank and standing vegetation was high at sites recently burned (0.86) and low in sites long‐since burned (0.52). The relative abundance of ephemeral species in the seed bank peaked in recently burned plots, but the relative abundance of seeds of woody species did not match the pattern of abundance in the standing vegetation. Remarkably, the dominant perennial grasses and woody species were either absent from the seed bank or present at extremely low abundances. Discussion: Differences in the relative abundance of ephemeral species between standing vegetation and seed bank relate to the post‐fire succession process. The small soil pool of seed from woody species may be explained by allocation to belowground carbohydrate storage over seed production. Field observations suggest, however, that production of strongly dormant seed can be prolific and that high levels of seed predation make this system strongly seed‐limited. The discovery of this seed bank syndrome indicates that shifts in grass‐woody ratios can be driven by the juxtaposition of unpredictable seed rain and fire events in these desert dunes. However, estimates of grass‐woody ratios due to changing fire regimes will be difficult to predict.

Questions: Has the species-rich vegetation of upland hay meadows been maintained under low intensity management imposed by an agri-environment scheme? Is the target plant community re-establishing where it has been modified previously by intensive agricultural practices? What combinations of management practices and soil properties are associated with changes towards or away from the target community? Location: The Pennines, northern England, UK. Methods: A survey of 116 hay meadows in 1987 was repeated in 2002 by recording plant species in permanent quadrats. Changes in community variables (species richness, Ellenberg values, upland hay meadow community coefficients) were analysed in species-rich, modified species-rich and degraded grassland types. Redundancy Analysis and Generalised Linear Models were used to show the relationship between management practices and soil properties and change in species composition and community variables. Results: Few sites contained the species-rich grassland type, and here forb richness declined. In the modified species-rich type, total and grass species richness increased but Ellenberg N-values also increased. Total and grass species richness increased in the degraded type and the community coefficient increased. Management was weakly related to change in species composition but showed clear relationships with the community variables. Re-establishment of the target species-rich community was more likely with late cutting, in the absence of cattle or prolonged spring grazing, and at lower soil nutrient status. Conclusion: The species-rich community was not maintained but some reversion occurred in degraded grassland. Inorganic fertiliser application and intensive spring grazing should be avoided and cutting delayed until late July.Nomenclature: Stace (1997).Abbreviations: ESA = Environmentally Sensitive Area; GLM = Generalised Linear Model; MG3 = Upland hay meadow plant community code; RDA = Redundancy Analysis.