Introduction

1
Introduction We are characterizing temporal and
spatial variation in carbon storage and fluxes in
central Amazon forest. Our study sites are
primarily two 20 x 2,500 m transect plots within
INPAs ZF-2 Forest Management Reserve located 60
km north of Manaus (Fig. 1). Vegetation is
old-growth closed-canopy forest. Precipitation
varies seasonally (Fig. 2). 1176 tree species
have been identified in nearby Ducke Reserve1.
The landscape is undulating, with soils
comprising Oxisols on plateaus (70 clay, 15
sand), Utilsols on slopes (transition textures),
and Spodosols associated with small valleys
(baxios) and streams (70 clay, 15 sand)2.
Dendrometer studies We are studying monthly
radial growth increment (i.e. wood production)
using tree dendrometer bands distributed among
four projects community, population (3 species),
wood density, and phenology (evergreen,
deciduous). Preliminary results from the
community level project shows large variability
among individuals, and strong seasonality in wood
production rates (Fig. 3). Comparing population
and community studies shows that wood production
variation within species (n 3) is almost as
great as variation within the entire community.
Soil respiration We are measuring monthly soil
respiration rates with 57 chambers placed along a
1500 m transect that includes plateau, slope,
and baxio. Preliminary results for June show a
significant decline in respiration with soil
moisture (Fig. 4). At many sites along the
transect plots, soils are saturated from heavy
1999/2000 precipitation (Fig. 5). Stem
Respiration We have measured stem respiration
rates for trees stratified by growth rate from
the population dendrometer project. Preliminary
results shows a weak correlation between
respiration and wood production (Fig. 6). In
some cases, trees with very slow growth rates
have relatively high stem CO2 efflux.
Log(growth rate cm yr-1)
Figure 12. Distribution of tree mortality rates
measured since the early 1980s from the Biomass
and Nutrient Experiment (Bionte) and the
Biological Dynamics of Forest Fragments Project
(BDFFP).
Figure 13. Tree diameter increment rates were
log-normally distributed. The poor fit for slow
growth rates is probably indicative of errors
using standard DBH measuring tapes.
wood density g cm-3
Figure 1. Much of the central Amazon is
characterized by this undulating topography.
INPAs north-south (NS) and east-west (EW)
permanent forest inventory plots in red.
Figure 7. Most leaves achieved maximum carbon
fixation rate at low light levels, although there
was significant height effect.
Figure 8. Coarse litter decomposition rates
measured from boles of dead trees were dependent
on wood density (from Chambers et al. 2000).
Figure 2. Monthly precipitation for Manaus
(1910-1985). There is more variability in the
wet season than the dry season.
Figure 15. There was a slight reduction in
diameter increment rates for smaller trees (BDFFP
data). This relationship, and 14 pseudo-species
representing maximum stem diameter classes, was
important for modeling forest size structure.
Figure 14. Wood density among 268 Amazon tree
species7 was normally distributed.
Photosynthesis We are characterizing
photosynthetic traits (Amax, Adiel, Nleaf, SLA,
etc.) for trees from the dendrometer projects.
Preliminary results show that most leaves achieve
maximum photosynthetic rates at low light levels
(Fig. 7). We will explore estimating whole tree
gross primary production using these measurements
in conjunction with local allometric
relationships3. Coarse litter Decomposition and
respiration studies for coarse litter have been
carried out4. Decomposition has been
characterized as a function of wood density and
bole diameter (Figs. 8 and 9). Coarse litter
respiration was highly correlated with wood
moisture content (Fig. 10). We are continuing
these studies toward understanding how seasonal
changes in moisture effect respiration
rates. Modeling Data and results from these
projects are being linked using an
individual-based empirical-statistical model to
explore temporal and spatial variability in
carbon cycling dynamics5. Presently the model
has been parameterized to explore the carbon
balance of total large wood (TLW boles, branches
and coarse litter 10 cm diameter) (Fig. 11).
Distributions of tree mortality (Fig. 12), growth
(Fig. 13), wood density (Fig. 14), growth rate
suppression (Fig. 15), and other characteristics
of individual trees were used to simulate forest
dynamics. The model predicts forest age
structure (Fig. 16), coarse litter standing
stocks (Fig. 17), and that 10 ha is the minimum
plot size required for assessing forest carbon
balance6. The model also demonstrates that for a
relatively short time period following a
catastrophic mortality event, large wood acts as
a strong carbon source, followed by many years as
a weaker carbon sink, while maintaining overall
carbon balance (Fig. 18). Acknowledgements This
work supported by INPE/NASAs Large-Scale
Biosphere-Atmosphere Experiment in Amazonia
(LBA-ecology), INPAs Jacaranda Project funded by
the Japanese International Cooperation Agency
(JICA), and INPA/Smithsonian Institutions
Biological Dynamics of Forest Fragments Project.
a
Figure 3. ANOVAs for monthly wood production
rates from the community dendrometer project (n
309 trees). The number above the bars compare
months significantly different, the number above
the line is the mean (original units), and below
the line the number of trees with rate 0.
Figure 10. Coarse litter respiration responded
strongly to changes in wood moisture content for
dead trees in forest (green) and pastures (blue).
Figure 9. Coarse litter decomposition rates were
also dependent on bole diameter from a sample of
155 dead trees (Chambers et. 2000).
Figure 4. June soil respiration rates declined
significantly with volumetric H2O content (TDR).
In general, soils above 30-40 vol. were
completely saturated.
Figure 16. Tree age structure simulated by the
model. Model results for the largest and oldest
trees compared well with radiocarbon estimates8.
b
c
Figure 11. The individual-based model (coded in
Java) currently simulates the carbon balance of
total large wood (TLW). The relationships used
to parameter the model are based on numerous
intensive field studies.
Figure 5. Surface soil moisture in the wet
season from the NS transect plot (Fig. 1). High
moisture content is associated with plateaus and
small valleys, and the lowest values are
associated with the lower part of slopes.
Figure 6. Average woody tissue respiration rate
was only weakly correlated with average wood
production rate for trees from the population
dendrometer study. There was no significant
differences among species (n 3).
Figure 18. Model predicted effects of background
mortality (a), a single 50 catastrophic (b), and
two 20 catastrophic (c) mortality events on
total large wood carbon balance.
Figure 17