B33A-01 INVITED
What is the future of Amazon forests under climate change?
Large changes in Amazon carbon and water cycles, as predicted to occur with climate change by some coupled carbon/climate models, are expected to have global impact. Some models predict that these forests may be vulnerable to catastrophic collapse due to global warming-induced increases in drought, whilst others do not. Hence it is critically important to understand the mechanisms of forest-climate interactions in Amazonia. In order to address this question on observable timescales, we propose investigations at annual to decadal timescales which can give insight into the mechanisms that, in the model simulations, cause forest collapse. We first investigated large-scale seasonal and spatial patterns of Amazonian carbon fluxes, and the effects of land-use conversion thereon, by combining spatially continuous satellite data from the Terra- Moderate Resolution Imaging Spectroradiometer (MODIS) with data from a network of ground-based eddy flux towers. This work showed that primary forest photosynthetic activity begins its seasonal increase in the dry season, well before onset of the wet-season, a broad pattern opposite to that predicted by ecosystem models which simulate dry-season declines in primary forest photosynthesis due to water-limitation. Extending these observations to decadal timescales, ideally through another large ENSO-scale drought, should give strong insight into the longer term mechanisms implicated in predictions of Amazon forest collapse.
B33A-02
Combined effects of climate change and forest clearing on the Amazon vegetation: Projections for 2080-2100
A regional climate model with resolution of 60 km coupled with a potential vegetation model is used to simulate future vegetation distributions over South America. The coupled model, which produces an accurate representation of today's climate and vegetation, is forced with increasing atmospheric CO2 concentrations, sea surface temperature from a global model, and scenarios of future land use practices to predict climate and vegetation distributions for the last 2 decades of the 21st century. When only climate change is considered, under a business-as-usual scenario for global emissions, the extent of the Amazon rainforest is reduced by about 70 per cent by the end of this century, and the shrubland (caatinga) vegetation of Brazil's Nordeste region spreads westward and southward. Reductions in annual mean precipitation are widespread and rainfall becomes insufficient to support the rainforest in these regions, but some areas receive more precipitation. The length of the dry season increases in the central and southern Amazon in association with changes in the large-scale tropical circulation. Without this change in seasonality, local refugia of Amazon vegetation would be preserved and the retreat of the rainforest would be somewhat less extensive. Including various projections of future land use practices in addition to climate change may accelerate the unrecoverable demise of the rainforest and feedback to modify climate on regional space scales. The portions of the rainforest that are most vulnerable to climate change are the same as those that are under the most pressure from human activity, presenting a remarkable competition.
B33A-03
Remote Sensing of Miombo Woodland's Aboveground Biomass and LAI using RADARSAT and Landsat ETM+ Data
Estimations of biomass are critical in Miombo Woodlands because they represent a primary source of food, fiber, and fuel for 340 million rural peoples and another 15 million urban dwellers in southern Africa. The purpose of this study is to estimate woody aboveground biomass and Leaf Area Index (LAI) in Niassa Reserve, northern Mozambique. The objective of this study is to use optical and microwave satellite data with contemporaneous field data to estimate biomass and LAI. Fifty field plots were surveyed across the Niassa Reserve for biomass and LAI in July and December 2004, respectively. Remote sensing data consisting of RADARSAT backscatter (C- band, ë=5.6 cm) and a June 2004 Landsat ETM+ were acquired. Normalized Difference Vegetation Index (NDVI), Simple Ratio (SR), and a land-cover map (72% total accuracy) were derived from the Landsat scene. Field measurements of biomass and LAI correlated with Radarsat backscatter (Rsqbiomass=0.45, RsqLAI = 0.35, P<0.0001 ), NDVI (Rsqbiomass =0.15, RsqLAI=0.14-, p <0.0001 ) and SR (Rsqbiomass=-0.14, RsqLAI= 0.17, p <0.0001). A jackknife stepwise regression technique was used to develop the best predictive models for biomass (biomass = -5.19 +0.074*radarsat+1.56*SR, Rsq=0.53) and LAI (LAI= -0.66+0.01*radarsat+0.22*SR, Rsq=0.45). The addition of NDVI did not improve the model. Forest biomass and LAI maps were then produced for Niassa Reserve with an estimated peak total biomass of 18 kg/hm2 and a mean LAI of 2.8 m2/m2. In the east both biomass and LAI are lower than the western Niassa Reserve.
B33A-04
Response of leaf litter decomposition to rises in atmospheric CO2 and temperature
Atmospheric concentrations of CO2 have considerably increased in the last century and are expected to rise further. Elevated CO2 concentrations not only increase global temperature but also have potential to change plant litter quality, for example by increasing lignin content, changing C:N ratios and altering tannin contents. These chemical changes may interact with increased temperature to alter litter decomposition. To test whether changes in litter quality and warming affect decomposition, we conducted a field experiment with leaf litter collected from six species of mature deciduous trees exposed to either ambient or elevated CO2 levels. We used a set of 16 enclosures installed in four blocks in a freshwater marsh in a prealpine lake to test for the effects of CO2-mediated litter quality and temperature and the interaction of both factors. We measured leaf mass loss of the twelve litter types in control and heated enclosures (4 °C above ambient) and also in the open marsh. In contrast to expectations, species decomposing at low (oak and beech) and medium (hornbeam and maple) rates showed faster mass loss when leaves were grown under elevated CO2 conditions, whereas fast-decomposing species (cherry and basswood) showed no clear response. The accelerated decomposition of CO2-enriched litter could be due to higher amounts of nonstructural carbohydrates, which may have been either leached or readily degraded. Warming had a surprisingly small influence on mass loss of the tested litter species, and interactive effects were weak. These results suggest that direct and indirect effects of elevated CO2 levels on litter decomposition may not be readily predictable from first principles.
B33A-05
Atmospheric CO2 and N cycling: An ecosystem scale 15N tracer experiment
A major question in C sequestration science is whether N availability will limit long-term biological sequestration of atmospheric CO2. At the Duke Forest Free Air CO2 Enrichment (FACE) site, increased net primary productivity was sustained over a decade of experimental CO2 fumigation. Additional N was assimilated to support increased growth under elevated CO2, but the source of this additional N remains unknown. We used an ecosystem scale 15N tracer experiment to identify additional N resources. Isotopically labeled NH4NO3 was applied to the forest floor in May 2003. The forest plots were then sampled for isotope recovery in plant and soil pools for three subsequent growing seasons (September 2003, 2004, 2005). Results indicate that despite higher N uptake under elevated CO2, the total recovery of 15N in trees was similar under ambient and elevated CO2. At the same time, the concentration of 15N in the pine canopy was significantly lower under elevated compared to ambient CO2. This suggests that trees growing under elevated CO2 assimilate an additional, unlabelled source of N, causing the canopy 15N signature to be depleted relative to leaves growing under ambient conditions. Recovery of the 15N tracer decreased with soil depth, indicating that additional N uptake may be derived from roots exploiting unlabelled mineral resources deeper in the soil profile.
B33A-06
Tracing Water and Nitrogen Sources to Identify Controls on Stream Nitrogen Variation.
Our work explores ecological and hydrological processes that control the variation of key nitrogen species in streams of an upland catchment in northeastern Vermont, USA. Using high-frequency sampling over event and seasonal timescales along with chemical and isotopic tracers of water and nitrogen sources, we show how anthropogenic nutrient enrichment directly affected the forms and concentrations of nitrogen in streams draining a northern hardwood forest. Although flushing of nitrate from nitrified sources in soils is an important control on stream nitrogen variation during storm flow, atmospheric sources contributed up to 50 percent of the stream nitrate during specific runoff events. This atmospheric nitrate was delivered to the stream along preferential flowpaths effectively bypassing biogeochemical processes that typically retain nitrogen in the landscape. Furthermore, we investigated coupled biogeochemical transformations and hydrological transport processes to shed light on seasonal controls on the timing and magnitude of stream nitrogen fluxes. For example, stream nitrogen responses to autumn leaf fall showed that interactions of the water, carbon, and nitrogen cycles affected fluxes of nitrate and dissolved organic nitrogen in stream water. Overall, our high-frequency hydrochemical data provided key insights to open the blackbox of landscape processes that influence stream nutrient variation over time.
B33A-07
RP-RainNet: The Rio de la Plata Atmospheric Deposition Network. Set up and Preliminary Results
Atmospheric deposition is a key flux for understanding nutrient cycling in ecosystems. Southern South America
almost completely lacks atmospheric deposition data. We began rain collections at three (9/2005) and then
seven (9/2006) sites in Argentina and Uruguay across a west (dry) to east (wet) transect at latitude 34°S.
Preliminary data showed a low and uniform NO3 deposition across the region in the first three study sites. Nitrate
deposition was similar in populated (Buenos Aires, 13 million people) and unpopulated regions (Flores, less
than 5 persons per km2), in spite of the relative high industrial activity of Buenos Aires, suggesting relative well
mixed conditions. Ammonia deposition (6.0 kg/ha.year) doubled that of NO3 (3.2 kg/ha.year) with a high
correlation between them across rain events (r=0.91), supporting a similar origin of both N forms, likely found in
agricultural fields and rangelands. Chloride, Mg and Na concentrations were highly correlated suggesting a
marine origin. Ca and K had lower correlations with other ions, but correlated better with land derived ions (NH4
and NO3). Correlations between ions were stronger in Montevideo (seaside site), weaker in Buenos Aires
(estuary site) and medium in Flores (inland site). All ions concentrations in rain events decreased exponentially
with precipitation amount. A Principal Component Analysis (PCA) showed a high variability in ions concentrations
of rain events within each site but separated well rain events from different sites with the first component being
driven by Cl concentration vs. other ions and the second by marine (Cl, Mg and Na) vs. land ions (NH4, NO3, Ca
and K). Ongoing collections will expand or results in space and time including more continental situations.
http:www.agro.uba.ar/users/pineiro/
B33A-08
Quantifying the multiscale environmental controls on wildfire from species distribution models
Despite its widespread occurrence globally, wildfire preferentially occupies an environmental middle-ground and is significantly less prevalent in biomes characterized by environmental extremes (e.g., tundra, rainforests, deserts). We evaluated the biophysical "environmental space" of wildfire from regional to continental extents, using methods developed for modeling species distributions ("niche models"). This approach is particularly suitable for the biogeographical study of wildfire, because it simultaneously considers patterns in multiple factors controlling wildfire suitability over large areas. We used the Maxent algorithm to asses relationships between wildfire and environmental predictors for three levels of complexity in variable inclusion at three spatial scales, the conterminous United States, the state of California, and five wildfire-prone ecoregions of California. The resulting models were projected geographically to obtain spatial predictions of wildfire suitability and also projected to other regions to assess their generality and spatial "transferability." The models identified several important variables that were previously unsuspected in the large-scale control of wildfires and successfully predicted the potential range of wildfire among study areas. Models projected to different areas were useful only when they overlapped appreciably with the target area's environmental space, which has implications for creating future models. Application of this approach should allow us to explore the global range of wildfire in a changing climate, the potential for wildfire restoration where it has been "extirpated," and, conversely, the "invasiveness" of wildfire following changes in plant species composition. To our knowledge, it is also the first application of niche models to characterize environmental controls on a process.