The plant components and chemical composition of litter fall and the litter layer in a forest growing on deep, nutrient-poor sands were examined on North Stradbroke Island, south-eastern Queensland. The seasonal distribution of litter fall was examined over a 26-month period. While the total litter fall was greatest during summer months, the dominant tree species differed in their individual patterns of litter fall. Eucalyptus signata showed a single summer peak for leaf fall while E. umbra exhibited one peak in early summer and another in autumn. The possibility is discussed that these and other temporal differences are evolutionary expressions of niche differentiatibn to reduce competition between species in the ecosystem. The total litter fall averaged 640 g m-1 yr-1 and the accumulated forest floor mass totalled 2700 g m-2. Total nutrient pools and nutrient inputs in litter fall are presented. A litter half-life of 2.9 years is estimated, a figure close to the half-life of most of the nutrients in the litter. Manganese appears to be markedly concentrated in eucalypt leaves before they fall. Only sodium, potassium, copper and chloride appear to be leached easily from leaves slashed from trees and left on the forest floor. Patterns of litter production and decay in this subtropical forest fit within trends extrapolated from temperate Eucalyptus-dominated communities studied to date.

The plant biomass of a Eucalyptus signata-dominated forest 15 m tall growing on infertile sands off the Queensland coast is characterized in detail. The forest has a biomass of 180 t/ha, 90% of which is found in the nine species achieving > 2.5 m height. Of the total biomass, 42.5 % is below ground. Pteridium esculentum occupies 41 % of the understorey biomass, with 50 shrub and herb species partitioning the remainder. Dimension analysis of 10-11 individuals of each of three tree species- Eucalyptus signata, E. umbra subsp. umbra and Banksia aemula-has served to characterize the above- and below-ground growth forms of each species, and provide regressions of the mass of tree components on easily measured plant parts. The size distribution of tree and shrub stems on the site suggests that the major species have evolved quite different reproductive strategies for main- taining a steady-state population in the face of recurrent fires.

The pattern of plant species distribution, the richness of species composition, the distribution of resources among species and the manner of change in these parameters across landscapes are examined in three habitats on North Stradbroke Island. The closed-heath vegetation of a freshwater swamp, the grassland and low closed-forest of a frontal low dune sequence and the closed sclerophyllous forests of inland high dunes are studied by a variety of statistical techniques, including a new method for determining the degree and scale of microcommunity pattern. Of the three vege- tation types, the floristic diversity and complexity of spatial patterning within a small area is highest in the freshwater swamp vegetation. Not only is the scale of spatial variation smallest in this habitat, but the scale of patterning and co-association also becomes smaller as one proceeds into the swamp. The complexity of the vegetation is enhanced by a remarkably high species richness, a very low concentration of dominance and a high degree of intermingling of species. The floristic diversity and complexity of spatial patterning within such a small area of vegetation bears comparison with that of complex tropical rain-forest. The swamp vegetation appears to have undergone a high degree of speciation; at the same time, intermixture of species and the small scale of patterning have served as a means of minimizing intraspecific competition. If, like the tropical rain-forest, the ecosystem has evolved characteristics of rigid stability the system is likely to recover only slowly from major perturbation.

Pattern analysis data are presented for a number of tree species populations sampled from two sites located in dry sclerophyll forest within the Ku-ring-gai Chase park of New South Wales. The distributions proved to be predominantly contagious or random, with regularity occurring only occasionally. Observed variations in the degree of aggregation exhibited by a species were taken into account in interpreting pattern analysis curves. The relation of pattern analysis data to sample quadrat data fitted to known mathematical models is extremely variable, and it is shown that pattern at block sizes other than the one under consideration may suppress the appearance of deviations from randomness at block sizes which do show contagion when sampled with randomly placed quadrats. The possible origins of contagious distributions in eucalypt forest are briefly discussed.

Pattern diversity is a measure of the degree of spatial intermingling of species in a community of sessile organisms, combining the effects of species richness and equitability with those due to spatial patterning. The pattern diversity index was computed for an Australian Eucalyptus forest and the results interpreted in the light of an earlier analysis of the pattern of component species. The pattern diversity index was found to be sensitive to the number of individuals sampled at each point, reflecting a dependence of the index on the scale of species pattern, and hence on the scale of sampling. Limitations of the method in relation to low sample sizes and to particularly diverse samples are also discussed.