Grazing of wooded grasslands by domestic stock has led to dominance by woody plants, in many parts of the world. In semi—arid grasslands in Australia Dodonaea attenuata is a common, invasive shrub of perennial grasslands on sandy soils. There is evidence that cohorts of D. attenuata establish infrequently in vigorous grasslands and more frequently where the perennial grass has been damaged by overgrazing. This study examined the influence of herbaceous growth on mortality of D. attenuata seedlings, by growing them in plots with controlled moisture input. The herbaceous layer in the plots was either natural, clipped, or killed. Without summer irrigation all the seedlings in the natural grassland died. Seedling survival was inversely related to the amount of herbaceous growth on the other treatments. Summer irrigation maintained D. attenuata seedlings and produced a large biomass of perennial grass, which left the seedlings susceptible to grass fires. A water—balance model was run for the experimental site for the 97 yr of climatic records. The soil—moisture patterns associated with the establishment of three known cohorts of D. attenuata were identified. The model output was interrogated for similar soil moisture patterns in other years. It was estimated that only six widespread D. attenuata establishment events have occurred in 97 yr. In intact grasslands D. attenuata rarely establishes densely, and when it does so it is susceptible to grass fires. Modern management has increased establishment frequency by increasing the availability of summer soil moisture by overgrazing the grasslands and has suppressed grass fires. This has changed open grassland to dense shrubland.

Vegetation maps were prepared from aerial photographs taken in 1943?45 and 1991?92 of three, widely separated areas of sclerophyll forest adjacent to the western edge of rainforest on granitic soils in north Queensland. Nine types of sclerophyll communities could be discerned from aerial photos and characterized by field measurement. Two types of Wet Sclerophyll Forest (WSFa and b) were separated on the species of tree composing the tallest stratum and these were subdivided according to whether the ground layer was dominated by grass or young rainforest. A related type showed large, residual Eucalyptus grandis emergent from mature rainforest. Closed canopy sclerophyll forest with no emergents (SF), sclerophyll woodland and Acaciaforest were also discerned. WSF was defined as having more than 30 per cent of the closed crown cover contributed by trees more than 35 m tall. During the 50-year study period rainforest invaded 70 per cent of WSFa (tallest stratum dominated by E. grandis), which principally occurs as a narrow strip along the rainforest margin, and 57 per cent of the adjacent WSFb (tallest stratum composed of mixed species). Grass would be quickly excluded from invaded areas and thereafter they would only burn under extreme atmospheric conditions. Because sclerophyll trees are unable to regenerate in shade and usually require fire to provide the appropriate conditions, a long-term transition to rainforest may ensue. The final stages of this transition were observed in areas that exhibited full-stature rainforest with large, relictual E. grandis emergents in 1943, but had disappeared by 1992. The initial cause of this vegetation transition is a fire-free period of sufficient length for rainforest tree seedlings to establish and suppress the grass layer. It is not known whether these vegetation changes represent a trend, possibly caused by a change a century ago from fire management by Aboriginal people to management for the cattle industry, or whether it is a temporary phase in the fire-induced, dynamic relationship between rainforest and sclerophyll vegetation. The current loss of WSF probably endangers the survival of a range of genetically endemic biota. Most groups are poorly known but the marsupial Yellowbellied Glider Petaurus australis reginae is totally dependent upon WSF and a number of vertebrates would probably go locally extinct if WSF is replaced by rainforest. WSF is the wettest part of the sclerophyll communities and probably acts as a refuge in times of unusual aridity. To maintain the WSF habitat, fire management is clearly indicated, but the intensity of fire required to reverse the advance of rainforest may be socially unacceptable to instigate or impossible to control if it occurs by accident.

An estimate of the above-ground biomass of leaf and wood in the shrubs and trees of Eucalyptus populnea woodland near Cobar, N.S.W., was made, by regression of leaf and wood weight on height or trunk diameter of the most common species. A new double regression technique was used for E. populnea to avoid destructively sampling whole trees. The mass of branches was regressed on primary branch diameter and these equations were used to estimate the total mass of crowns of trees. The estimated crown mass was then regressed on trunk diameters. Regression equations for estimation of mass of wood and leaf are provided for Geijera parvifora, Cassia nemophila, Dodonaea viscosa, Eremophila mitchellii, E. sturtii, E. bowmanii, Myoporum deserti and young Acacia aneura. The error of the estimates varied between 3 and 18% for different species. Log transformation of the data usually improved the correlation coefficients obtained but sometimes increased the standard error of the estimate. Estimates based on canopy measurements were rarely more accurate than those from height or trunk diameter. The overall estimate of biomass was 3.4 t of leaf and 51.4 t of wood per ha with a standard error of < 15%.

In the tropical rainforests of northern Australia, we investigated the effects of habitat fragmentation and ecological parameters on the prevalence of blood-borne parasites (Plasmodium and Haemoproteus) in bird communities. Using mist-nets on forest edges and interiors, we sampled bird communities across six study sites: 3 large fragments (20–85 ha) and 3 continuous-forest sites. From 335 mist-net captures, we recorded 28 bird species and screened 299 bird samples with PCR to amplify and detect target DNA. Of the 28 bird species sampled, 19 were infected with Plasmodium and/or Haemoproteus and 9 species were without infection. Over one third of screened birds (99 individuals) were positive for Haemoproteus and/or Plasmodium. In forest fragments, bird capture rates were significantly higher than in continuous forests, but bird species richness did not differ. Unexpectedly, we found that the prevalence of the dominant haemosporidian infection, Haemoproteus, was significantly higher in continuous forest than in habitat fragments. Further, we found that ecological traits such as diet, foraging height, habitat specialisation and distributional ranges were significantly associated with blood-borne infections.