Abstract. Through interpretations of remote-sensing data and/or theoretical propositions, the idea that forest and savanna represent "alternative stable states" is gaining increasing acceptance. Filling an observational gap, we present detailed stratified floristic and structural analyses for forest and savanna stands located mostly within zones of transition (where both vegetation types occur in close proximity) in Africa, South America and Australia. Woody plant leaf area index variation was related to tree canopy cover in a similar way for both savanna and forest with substantial overlap between the two vegetation types. As total woody plant canopy cover increased, so did the relative contribution of middle and lower strata of woody vegetation. Herbaceous layer cover declined as woody cover increased. This pattern of understorey grasses and herbs progressively replaced by shrubs as the canopy closes over was found for both savanna and forests and on all continents. Thus, once subordinate woody canopy layers are taken into account, a less marked transition in woody plant cover across the savanna–forest-species discontinuum is observed compared to that inferred when trees of a basal diameter > 0.1 m are considered in isolation. This is especially the case for shrub-dominated savannas and in taller savannas approaching canopy closure. An increased contribution of forest species to the total subordinate cover is also observed as savanna stand canopy closure occurs. Despite similarities in canopy-cover characteristics, woody vegetation in Africa and Australia attained greater heights and stored a greater amount of above-ground biomass than in South America. Up to three times as much above-ground biomass is stored in forests compared to savannas under equivalent climatic conditions. Savanna–forest transition zones were also found to typically occur at higher precipitation regimes for South America than for Africa. Nevertheless, consistent across all three continents coexistence was found to be confined to a well-defined edaphic–climate envelope with soil and climate the key determinants of the relative location of forest and savanna stands. Moreover, when considered in conjunction with the appropriate water availability metrics, it emerges that soil exchangeable cations exert considerable control on woody canopy-cover extent as measured in our pan-continental (forest + savanna) data set. Taken together these observations do not lend support to the notion of alternate stable states mediated through fire feedbacks as the prime force shaping the distribution of the two dominant vegetation types of the tropical lands.

Abstract. Variations in leaf mass per unit area (M a ) and foliar concentrations of N, P, C, K, Mg and Ca were determined for 365 trees growing in 23 plots along a West African precipitation gradient ranging from 0.29 to 1.62 m a -1 . Contrary to previous studies, no marked increase in M a with declining precipitation was observed, but savanna tree foliar [N] tended to be higher at the drier sites (mass basis). Generally, M a was slightly higher and [N] slightly lower for forest vs savanna trees with most of this difference attributable to differences in soil chemistry. No systematic variations in [P], [Mg] and [Ca] with precipitation or between trees of forest vs savanna stands were observed. We did, however, find a marked increase in foliar [K] of savanna trees as precipitation declined, with savanna trees also having a significantly lower [K] than those of nearby forest. These differences were not related to differences in soil nutrient status and were accompanied by systematic changes in [C] of opposite sign. We suggest an important but as yet unidentified role for K in the adaption of savanna species to periods of limited water availability; with foliar [K] being also an important factor differentiating tree species adapted to forest vs savanna soils within the 'zone of transition' of Western Africa.

No abstract

Abstract. Through interpretations of remote sensing data and/or theoretical propositions, the idea that forest and savanna represent "alternative stable states" is gaining increasing acceptance. Filling an observational gap, we present detailed stratified floristic and structural analyses for forest and savanna stands mostly located within zones of transition (where both vegetation types occur in close proximity) in Africa, South America and Australia. Woody plant leaf area index variation was related in a similar way to tree canopy cover for both savanna and forest with substantial overlap between the two vegetation types. As total woody plant canopy cover increased, so did the contribution of middle and lower strata of woody vegetation to this total. Herbaceous layer cover also declined as woody cover increased. This pattern of understorey grasses and herbs being progressively replaced by shrubs as canopy closure occurs was found for both savanna and forests and on all continents. Thus, once subordinate woody canopy layers are taken into account, a less marked transition in woody plant cover across the savanna-forest species discontinuum is observed compared to that implied when trees of a basal diameter > 0.1m are considered in isolation. This is especially the case for shrub-dominated savannas and in taller savannas approaching canopy closure. An increased contribution of forest species to the total subordinate cover is also observed as savanna stand canopy closure occurs. Despite similarities in canopy cover characteristics, woody vegetation in Africa and Australia attained greater heights and stored a greater concentration of above ground biomass than in South America. Up to three times as much aboveground biomass is stored in forests compared to savannas under equivalent climatic conditions. Savanna/forest transition zones were also found to typically occur at higher precipitation regimes for South America than for Africa. Nevertheless, coexistence was found to be confined to a well-defined edaphic/climate envelope consistent across all three continents with both soil and climate playing a role as the key determinants of the relative location of forest and savanna. Taken together these observations do not lend support the notion of alternate stable states mediated through fire-feedbacks as the prime force shaping the distribution of the two dominant vegetation types of the tropical lands.

The study was carried out in three selected plantations. Sampling