A suite of woody plants inhabiting the seasonally dry tropics flower while leafless during the dry season, raising intriguing questions about the role of moisture limitation in shaping their phenology. Brachychiton megaphyllus is one such species, a shrub of open forests and savannas in northern Australia. We documented leaf and reproductive phenology of 14 shrubs, and irrigated a further 15, to determine if soil moisture affected leafiness and reproductive activity. Brachychiton megaphyllus showed first flower buds shortly after the cessation of wet-season rains, and budded and flowered throughout the dry season. In some plants, leaf flush occurred prior to the first rains. Rates of fruit set and maturity were very low. Irrigation did not significantly influence leaf shoot or subsequent canopy development. Contrary to expectation, irrigation decreased the production of buds and flowers though it had no impact on the production of fruit, a response for which we suggest a number of hypotheses. Phenological responses to irrigation may have been limited because B. megaphyllus responds primarily to cues other than soil moisture and is buffered against seasonal drought by a large tap root. This suggests mechanisms by which flowering while leafless may occur in a range of species.

A fire-mediated recruitment bottleneck provides a possible explanation for the coexistence of trees and grasses in mesic savannas. The key element of this hypothesis is that saplings are particularly vulnerable to fire because they are small enough to be top-killed by grass fires, but unlike juveniles, they take several years to recover their original size. This limits the number of recruits into the adult size classes. Thus savanna vegetation may be maintained by a feedback whereby fire restricts the density of adult trees and allows a grass layer to develop, which provides fuel for subsequent fires. Here, we use results from a landscape-scale fire experiment in tropical Australia, to explore the possible existence of a recruitment bottleneck. This experiment compared tree recruitment and survival over 4 y under regimes of no fire, annual early and annual late dry-season fire. Stem mortality decreased with increasing stem height in the fire treatments but not in the unburnt treatment. Tree recruitment was 76–84% lower in the fire treatments than the unburnt treatment. Such fire-induced stem loss of saplings and reduced recruitment to the canopy layer in this eucalypt savanna are consistent with the predictions of the fire-mediated recruitment bottleneck hypothesis.

The abundance and distribution of reptiles and amphibians inhabiting tropical monsoonal wetlands in Kakadu National Park, northern Australia, were monitored between late 1979 and 1982. Seasonal trends in species richness, and the influence of various environmental attributes (including those caused by feral ungulates) on species distribution patterns were also examined.

These wetlands support a high proportion of the amphibian species known from the Park, but are of less importance to reptiles. Within the wetlands, the structurally more complex forested ‘margins’ support many more species than the treeless floodplains. Amphibian species distributions seem to be influenced primarily by elevation, through its effects on soil moisture and flooding levels, while reptile distribution patterns reflect a more complex set of environmental factors, with structural attributes (e.g. vegetation height structure and cover, refuge abundance, leaf litter cover and depth, extent of flooding) being of prime importance.

Feral buffalo and pigs, through their grazing, trampling and wallowing, may considerably influence such structural attributes and change the duration and extent of water lie, and thus indirectly affect species distribution patterns. Current buffalo culling programmes provide an opportunity to monitor vegetation and faunal changes, and quantify feral animal impact post facto.