Tropical forest vegetation occurs within a narrow corridor along the semi-arid lower floodplain of the Tana River, Kenya. An inventory of the woody flora and ecological study of 12 forest areas in the Tana River National Primate Reserve (TRNPR) describe the composition and structure of this riverine forest ecosystem and identify patterns of regional and local diversity. Only 175 species in 49 families are recorded in the woody flora, but the geographic affinities of 98 species are from four major floristic regions in Africa: Zanzibar-Inhambane (31%), Somalia-Masai (16%), Guinea-Congolian (12%), and Zambezian (1%). Ten species are rare and/or disjunct. Forests have a disturbed physiognomy characterized by a low mean height (14.4 m), high species importance attributable to a few trees, low density (409 ha−1) and coverage (23.1 m2 ha−1) of trees > 10 cm dbh, and an abundance of palms and lianes. Changes along the river in sediment deposition and hydrology explain the high diversity of landforms and corresponding forest types in the TRNPR vicinity, but the absence of trees which are most important near Bura, Acacia elalior, and near Wema, Barringtonia racemosa, suggests that the regional diversity of the ecosystem is inadequately protected. Preservation of key resources, such as the endangered primates, should be coupled with protection of the forest heterogeneity that characterizes this dynamic landscape.

The endangered Gouldian Finch Erythrura gouldiae is the only Australian finch to nest exclusively in tree hollows or, more rarely, in termite mounds. It is sympatric with the abundant Long-tailed Finch Poephila aculicauda which nests frequently in tree hollows. The aim of this study was to define the characteristics of nest sites and breeding areas of Gouldian and Long-tailed Finches to determine whether nest hollows were in short supply, and where finches fed relative to their nest sites. The Gouldian Finch chose hollows with smaller, more northerly entrances than Long-tailed, or randomly chosen hollows, nested deeper down the hollow and on steeper hill-slopes. Both species preferred single to multi-trunked trees of larger diameter than trees with randomly chosen hollows. Discriminant analysis classified 22–25% of the randomly chosen hollows and 28–38% of Long-tailed nesting hollows as Gouldian hollows. Gouldian Finches were more specific in their choice of nest sites than Long-tailed Finches. The vegetation at the two sites differed floristically, but within each site there was no floristic distinction between feeding or breeding sites or sites chosen at random. Gouldian Finches chose feeding habitat where trees were more spaced, on less rocky, barer ground than around breeding sites.

On the basis of the criteria measured, there was no shortage of suitable hollows available to the Gouldian Finch for nesting. The overlap between feeding and breeding sites indicates the importance of managing breeding habitat for conservation.

The habitat distribution and relative abundance of 113 species of Gramineae were documented by releves in 82 stands near Conception, Santa Cruz, Bolivia. A factor analysis was used to compare the floristic similarity of stands situated in semideciduous forest, cerradao, cerrado, campo rupestre, valley-side campo, pantanal complex or on granite outcrops. Individual grass species usually had a preferred habitat and occurred with decreasing abundance in stands judged to be transitional by physiognomic, edaphic, and floristic criteria. Stands situated in cerrado vegetation were most similar to one another in grass species composition. Cerradao was transitional to semideciduous forest and cerrado but certain grasses were characteristic of this vegetation type. The floristic composition of the single campo rupestre locality was somewhat similar to cerrado; however, several of the more abundant grass species of this vegetation type did not occur in any nearby cerrado stand. Granite outcrops had a distinct grass flora and showed little similarity to other vegetation types. In savanna wetland communities, grass species distribution was influenced by water regime. Differences between stands on valley-side campo corresponded to topographic position on a gradient of increasing water surplus. Seasonally humid/dry stands on valley-side campo and pantanal complexes had a high degree of similarity. Stands lower on the catenary sequence of pantanal complexes and valley-side campos were increasingly dissimilar, a result of the different edaphic conditions of the seasonally flooded soils of pantanal complexes when compared with the permanently saturated (but never flooded) soils of valley-side campos. Pantanal complexes had the richest grass flora of all vegetation types because their microtopographic variability creates numerous micro-habitats with distinct water regimes, each supporting different grass species.

A life-form classification from Neotropical Gramineae is proposed for 113 native species from eastern lowland Bolivia. A factor analysis was performed on a correlation matrix of 51 vegetative attributes to identify ‘character suites,’ which were then used as the basis for the classification. The principal types are arboreal, liana, shrub-like, forb-like, bunch grass with basal foliage, bunch grass with basal and caulescent foliage, caespitose annuals, elongate rhizomes, turf grass, stoloniferous, and decumbent (the last category is further divided into subgroups: -perennial, -indefinite, -floating aquatic, and -vine-like). A key component of the life-form classification is the morphology of leaf blades, with five major patterns recognized: elongate and inrolled, elongate and conduplicate, elongate and flat with an attenuate base, short and broad with an amplexicaule base, and short and broad with a pseudopetiole.

The distribution and relative abundance of three major photosynthetic categories were compared with life-form and habitat preference. Photosynthetic type of individual species was identified based on leaf anatomy. C4-(XyMS −) species predominated in cerradao, cerrado, campo rupestre, valleyside campos, and seasonally inundated savanna, as well as on granite outcrops. C4-(XyMS + ) species were relatively rare in all habitats, but were significantly more abundant in the xeric microenvironments of granite outcrops. C3 species predominated in semi-deciduous forest and increased in relative abundance in wetland habitats. Caespitose annuals and bunch grass species were predominantly C4 type, while arboreal and liana species were uniformly C3 type. Grasses classified as decumbent or shrub-like were predominantly C3 type, while rhizomatous, stoloniferous, and forblike grasses were a mixture of C3 and C4 type.

The intent of the study was to group morphological characteristics objectively in order to derive a life-form classification specific for the Gramineae. Comparisons of the relative abundance of the various life-forms in a broad range of habitat types provide a descriptive assessment of the adaptive value of these characteristics.

The volcanic island of Anak Krakatau emerged from the sea in 1930. The antlion Myrmeleon frontalis has only become established on Anak Krakatau within the last five years, despite having been present on the other Krakatau Islands for at least 50 years.

The appearance of the antlions on Anak Krakatau seems to have coincided with the construction of a shelter hut in 1986 which provides a suitable under-floor habitat of dry dusty soil for the larvae to construct their pits. From the hut site they have extended into sub-optimal locations around the bases of clumps of wild sugarcane, Saccharum spontaneum, on the adjacent lower slopes of the outer cone ash fields.

The soil around the Saccharum chimps is variable. Fine deposits are suitable for antlion larvae, more gravelly soils are not. The antlion larvae inhabiting the Saccharum clumps are disturbed by rain which destroys their pits.

These two habitats, the hut and Saccharum sites, are the only places on the island that are occupied by the antlion larvae. They have quite different micro-climates so that the two populations are out of synchrony with each other.

The seasonal patterns of surface activity and behaviour were recorded for three species of juliform millipede in a savanna habitat in south-east Botswana. Initially, surface activity followed rainfall events but this association weakened as the season progressed and, although significant, the correlation between the magnitude of rainfall events and activity was weak. Activity also occurred on moist, relatively cool and cloudy days, particularly in Alloporus uncinatus which was much larger than the other species. The onset and duration of surface activity varied between species. Observations of feeding behaviour suggested that millipedes in semi-arid savannas did not subsist exclusively on tree leaf litter but appeared to select a range of food types. Mobility is necessary to obtain these patchy resources but differences between the sexes in patterns of behaviour are more likely related to the mating patterns of these species.

The ‘campos de murundus’ (carthmound fields) are typical landscapes of the cerrado (savanna) region of Central Brazil and are characterized by many rounded earthmounds called ‘murundus’ which are scattered over a grassland surface. As these areas are normally liable to seasonal water-logging, the earthmounds are true habitat islands capable of bearing woody plants typical of the cerrado, which are normally very intolerant of root inundation. The vegetation of an area of murundus field at Cuiaba, in the State of Mato Grosso, Central Brazil, was surveyed for the purpose of evaluating the island-effect on the plant community growing on the mounds/islands. Two main aspects were considered: the influence of the size of the mound and the differences between the types of cerrado occurring on the mounds and on the areas surrounding the murundus field. Most plant species occurring on the mounds were also present in the flora of the surrounding cerrado. However, only a fraction of these species was able to colonize the mounds successfully. The number of species on the mounds was found to be significantly correlated with mound area. Speciesarea relationships were better described with the power model S = C(Az) than with the exponential model S = C + z (logA) (fitted with linear regression). The mound size also had a strong influence on the spatial distribution of the species on the surface of the mounds. The species were organized in an increasing number of concentric topographic belts corresponding to a gradient of soil depth above the Hooding level.

This paper presents the quantities of macro-nutrients (N, P, K, Ca, Mg and S) released from the ground layer vegetation to the atmosphere during six experimental burnings, in a Brazilian open savanna, named ‘campo-cerrado’. The values were obtained by comparing the amounts of nutrients in the fuel and in the ash resulting after the burn, in three different seasons. On average, 20.6 kg of N, 1.6kg of P, 7.1 kg of K, 12.1kg of Ca, 3.0 kg of Mg, and 3.2kg of S ha−1 were transferred from the plant biomass to the atmosphere, representing mean percentages of 95% for N, 51% for P, 44% for K, 52% for Ca, 42% for Mg, and 59% for S. The burn season showed no significant influence on the amounts of nutrients transferred to the atmosphere. Some relationships involving the quantities of macro-nutrients released, fuel moisture, the combustion efficiency during the burns, and the initial standing biomass in the area were also tested. Comparing the average output of macro-nutrients to the atmosphere obtained in this study with their inputs via rainfall, in the same area, it was estimated that three years would be an adequate interval between prescribed burnings for this campo-cerrado, to speed up nutrient recycling without impoverishing the system.