A conceptual model that represents the importance of large arborescent palms in the dynamics of seasonally dry Amazonian forests was conceived to guide this thesis. Specifically, the model is aimed at guiding the investigation of the role of large arborescent palms on forest regeneration and succession. Aspects of the model were examined with an ecosystem experiment that took advantage of a palm-rich secondary forest patch (=palmeiral) close to the Pinkaiti research station, south-eastern Amazon, immersed in a matrix of notably well protected seasonally dry forests, preserved for the purpose of scientific investigation and preservation. We focused on the palms Attalea maripa and Astrocaryurn aculeatum, the first naturally abundant and the second rare at the Pinkaiti, attempting to establish comparisons between the two species, whenever possible. The Pinkaiti, like other seasonally dry forests, have great heterogeneity in forest structure and composition, associated with biotic characteristics of the most important tree species, natural disturbance and history of land-use. The palm grove, moderately dominated by the arborescent palm Attalea maripa (Aubl. ) Mart., presented high tree species diversity and was floristically similar to undisturbed forests at the study site. Despite being superficially similar, the palms A. maripa and A. aculeatum are distinct in the structure and development of their stem. The samples of A. maripa stems were more homogeneous in density than those of A. aculeatum, both from the internal to the external, and from the lower to the upper parts of the stem. Field estimates of stem growth rates revealed that A. maripa growth is also more constant through development in height. Short A. aculeatum palms had faster growth rates than A. maripa, as they get taller, stem growth rates are reduced and approach A. maripa rates. The study of A. maripa distribution in relation to forest disturbance suggests that, both natural and anthropic disturbance affects favorably their occurrence. However, due to the long time scale of palm development in relation to ordinary scientific research, the study of the importance of forest disturbance to the development of arborescent palms must consider the historical perspective and explicit variations of palms distribution across the landscape. The seeds of A. aculeatum are substantially larger that those of A. maripa and carry more reserves for the embryo. The study of the seed rain of both species, based on the seedling distribution patterns in relation to reproductive adults, contradicts general patterns among Angiosperms as, at the scale considered, the larger seeds of A. aculeatum were more homogeneously dispersed than the seeds of A. maripa. The result can be explained by the seed dispersal services paid by scatterhoarding caviomorph rodents (agoutis, Dasyprocta aguti). The rodents apparently prefer the more nutrient rich seeds of A. aculeatum, being more efficient dispersers of this palm, as compared with their behaviour towards A. maripa, which "pays" smaller rewards for the seed dispersal service. Hypotheses for the maintenance of tropical forest tree diversity were tested with a system that includes the palm Attalea maripa, the beetle Pachymerus cardo and vertebrate predators of large seeds in a palm patch and its adjacent forest in Southeastern Amazon. Seed predation was affected by the degree of host-specificity of the animals involved. Seed survivorship depended positively (when the predator was a specialist) or negatively (when the predator was a generalist) on the distance from the palm-patch but was unrelated to distance from individual fruiting palms. The results presented in the data chapters of this thesis suggest that the proposed model is appropriate and that its study is rewarding for understand the importance of large arborescent palms on the dynamics of seasonally dry Amazonian forests. The results highlight their role as tools for environmental restoration in a likely future of larger scale deforestation in the Amazon. We also investigated and compared the determinants of palm diversity at different spatial scales: along a moisture gradient in a seasonally dry Amazonian forest and using a climatic model comprising the entire Brazilian territory. Climatic hazards seem to strongly determine the distribution of palms, with the relative importance of moisture being highly consistent across scales.

Satellite radar imaging is a promising technique for biomass mapping and the monitoring of deforestation in tropical forests and reducing the uncertainty in the quantification of forest biomass in tropical regions. The present paradigm in radar imaging is the fitting of empirical relationships between the radar signal and biomass for diverse forest ecosystems, especially in the humid tropics. Therefore, there is a great need to generate knowledge about how to monitor and characterise the biomass of intact and disturbed tropical forest biomass. This research presents the analysis of data from four years of L-band radar imagery from ALOS PALSAR within a carbon dense, tropical peat swamp forest ecosystem in Central Kalimantan (Indonesia). The results showed that the temporal behaviour of the radar signal varied across a gradient of forest biomass, being highly variable at low biomass levels. Critically a large amount of signal change was unrelated to biomass change. Changes in the radar signal were related in a complex non-linear manner to changes in the peatland water table. This allowed, for the first time, the estimation of water table depth at high spatial resolutions from radar images. It was found that the radar signal related to loss of primary forest biomass after fire were in the opposite direction to that expected according to fitted radarbiomass equations. Burnt areas showed highly variable temporal radar with variability linked to rainfall indicating a possible interaction between the water table and remaining (dead) aboveground biomass. The implications of these results are that, at least in tropical peatlands, estimates of biomass based on single date radar images are likely to be highly misleading; multitemporal radar data sets are required to both interpret disturbance histories and to produce accurate classifications of above ground biomass.

Amazonian rainforests are important systems that vigorously cycle carbon and water at the global and regional scale. However, despite their importance, dynamics of organic matter cycling in tropical rainforests within the global carbon cycle is poorly understood. New field-based hydrology and geochemistry data is presented from a three-year study (2009-2011) in the pristine rainforest of central Guyana at the northern rim of Amazonia. The study shows that two commonly used satellite- and interpolated-based models to estimate rainfall greatly overestimate and underestimate in the wet and dry season, respectively. This misrepresentation of hydrology at local and regional scales greatly affects our ability to understand and predict the connections and feedbacks between the hydrological and geochemical cycles. River water δ2H and δ18O isotopes from peak wet and dry seasons suggest that the majority of rainfall has a residence time of at least one month, which provides regions of water saturated zones for organic matter (OM) to be preserved in deeper rainforest soils (Δ14C ages of 360-1200 years). In contrast, Δ14C values of surface soils and dissolved organic matter (DOM) in river water recycle within ~60 years or less. Carbon normalised yields of lignin phenols, used as tracers of vascular plant material, are abundant across the terrestrial-aquatic interface. However, lignin appears to accumulate in river bed sediments. Soil leachate experiments confirm that during mobilisation from soil particulates into the dissolved phase, desorption processes change the composition of lignin biomarker ratios. Surface soil δ13C signatures show that the majority of carbon is fixed through the C3 pathway (-26.4 to -32.0‰). However, a strong variability of up to 10‰ in riverine δ13C of dissolved organic carbon suggests that changing δ13C of DOC reflects changing contributions of degraded and fresh organic compounds within the total OM pool. Superimposed on seasonal cycling, short intense rain events cause rapid mobilisation of large amounts of DOC (up to 114 mg/L) that is divided into two main fractions, humic substances and ‘invisible’ DOM, or ‘iDOM’. The latter group is characterised by non UV-absorbing organic compounds of mono- and oligosaccharides, alcohols, aldehydes, ketones and amino sugars. Importantly, iDOM contributes up to 89% to the total river OM pool during peak DOC supply.

No description available.

No description available.

No description available.

No description available.

No description available.

We explore the connection between the observed spatial patterns of trees and shrubs in the rainforest and the ecological processes that shape these patterns. In particular we study the information that is captured by within-species aggregation and between species co-association. In the first part of this thesis, we introduce an individual-based model of the reproduction and death of trees of a single species. We use this as a null-model to compare within-species aggregation under idealised conditions with those found in the field. We show that the within-species aggregation of a species is expected to be strongly dependent on its local abundance. Based on this result we examine the effect of dioecy and recent changes in local abundance. We find that within-species aggregation maintains information on recent changes in local abundance. In the second part of this thesis, we examine the pair-wise cross-species spatial co-associations. In this work we do not focus on the spatial co-variation of individual species pairs, but on all pairwise combinations of a large group of species. We introduce a novel technique to normalise the cross-species co-association values that helps to make them comparable between different pairs of species, taking into account the different local abundances and within-species spatial patterns of the species. We use these normalised co-association values to find sub-communities of species that are co-located within the same spatial regions. Based on those sub-communities of species, we then investigate the effect of habitat and shade-tolerance in structuring the ecosystem.

Tropical rainforests comprise the world's oldest and most biodiverse biome, covering over 10% of Earth surface and providing livelihoods for over 300 million people. The upcoming consequences of anthropogenic climate change on tropical rainforests remain poorly understood. This is particularly true of the lowland tropical rainforests of Southeast Asia, aka Dipterocarp rainforests, in relation to the projected changes in El Niño/Southern Oscillation dynamics (ENSO). To address this important knowledge gap, this thesis aims to elucidate the ecological responses of Dipterocarp rainforests to past ENSO changes using a sediment-based multiproxy palaeoecological approach. A suite of fossil proxies (e.g. lipid biomarkers, pollen, charcoal) from a Philippine lacustrine sequence spanning the last 1,400 years were used to generate records of precipitation, fire, nutrient availability and plant dynamics. These records were used to explore the effects of major environmental drivers on the structure and functioning of Dipterocarp rainforests, specifically their dominant tree family, Dipterocarpaceae, and their pioneer plant taxa. The results suggest that Dipterocarp rainforests have been exposed to a prolonged period of enhanced ENSO regime during the second half of the Little Ice Age (~AD 1650-1900), yet there was no apparent effect of this climatic variability on the biomass of Dipterocarp trees. In contrast, the findings indicate that major fluctuations in phosphorus availability at the study site have had a significant influence on the biomass of Dipterocarp trees, though it remains uncertain which aspect of their biomass was primarily affected. Finally, my results propose that changes in ENSO conditions can significantly affect the abundance of pioneer taxa, with the nature and magnitude of that effect depending heavily on the level of fire activity. This thesis examined for the first time the vegetation responses of Dipterocarp rainforests to large-scale climatic changes in the past. The findings presented here provide new insights into the likely structure and functioning of this globally important ecosystem in a warming world, and provide a context for understanding its resilience to current and future environmental change.