PhD Research
Background
The global climate system is changing. There is a large, and growing, body of evidence which supports this conclusion (IPCC 2001). These changes are not wholly natural (See Figure 1, IPCC 2001), and are instead largely a result of human production of carbon dioxide (CO2) and other chemicals. CO2 allows incoming radiation to pass through the atmosphere, but prevents longer-wave thermal radiation from escaping. Since the atmospheric CO2 concentration has risen sharply from approximately 280 ppm during the 18th century to the current level of 370 ppm, there has been a concomitant rise in global temperatures. This relatively simple change in temperature has caused a complex series of knock-on effects. These effects may be broadly grouped into two categories: direct effects upon other physical aspects of the climate, and indirect effects upon carbon (C) cycling.
The second category is important because it may introduce feedback mechanisms into the climate change process; accelerating or reducing the rate of change. For example, the increase in CO2 concentration during the 20th century is thought to have caused an increase in the rate of C uptake in plant biomass (Phillips et al. 1998, DeLucia et al. 1999). This is a negative feedback upon CO2 concentration caused mainly by the effect of CO2 upon plant growth. However, over time, associated changes in the climate such as temperature, drought and frequency of El Nino events are likely to over-ride the effect of CO2 and initiate a net efflux of CO2 from forests (Cox et al. 2000, Dufresne et al. 2002). This could generate a strong positive feedback, whereby increased CO2 concentration causes hotter, dryer conditions which in turn stimulates further release of CO2 into the atmosphere.
Although some experimental research has been carried out on the effect of the environment upon C allocation in plants (Snyder & Carlson 1984, Wilson 1988) it is unclear how well these simplified systems mimic complex reality, and what the effects are upon C transfer into the atmosphere as CO2. Evidence is also accumulating for large-scale ecosystem shifts in the Amazon due to climate change (Phillips et al. 2002) but these modelling results remain controversial. Consequently, there is an urgent demand for a more detailed understanding of the potential ecological responses to climate change. Specifically, one of the principal modelling uncertainties is the C allocation response of plants to warming and drying (Dufresne et al. 2002) Minor changes in the pattern of allocation could potentially have relatively large effects upon the overall residence time of C in the terrestrial biosphere, and the rate of release of CO2 to the atmosphere.
Experimental DesignMy research was conducted at Caxiuanã National Forest Research Station, in Pará State, NE Brazil. A large-scale drought experiment was established in 2001 where a hectare of forest was covered in plastic panels to intercept incident rainfall (Dry plot). This plot was compared to a nearby control plot (Sand plot) where precipitation patterns remain unaltered. Additional plots (Clay and Fertile plots) was established, in nearby forests on different soils, in order to examine the additional influence of soil structure and fertility. Above- and below-ground forest growth, together with soil respiration, on all plots was monitored over a full seasonal cycle.
Key Results
Methodological
1) (see article in press in Forest Ecology & Management) There was considerable spatial variation particularly in root biomass and growth. This meant that a very large number of samples was required to quantify these variables with a reasonable degree of accuracy. A review of sample sizes used in the literature indicated that roots and soil are usually seriously undersampled, which then limits the degree of confidence which may be placed upon estimates of biomass and growth.
2) (see article in New Phytologist) In response to this problem, i designed a method for extracting roots from soil which is relatively quick and accurate. The method relies upon splitting the extraction period into time steps, and then using the observed pattern of cumulative extraction over time to predict the amount that you would have obtained if you had continuing beyond the actual manual extraction period.
3) (article in review for Plant & Soil) On a different note, rhizotrons are root observation boxes that have become a popular means to monitor root dynamics, but they yield measurements in units (root length per two dimensional area of observation screen) that are not easily comparable to most other ecosystem processes (usually mass per unit ground area). Various unit conversion methods have been proposed. I found that applying different conversion methods to the same rhizotron data yielded worryingly different estimates of root growth. Based upon a critical review of the various sources of uncertainty and error inherent in each method, i recommend that one particular conversion method is likely to be the most reliable.
Scientific
4) (article in review for Functional Ecology) There was enormous variation in growth of root mass, length and surface area both within and between plots. Growth appeared to be highest in the wet season, and inhibited in the dry season. This does not fit with the general paradigm that plants should allocate resources to increase uptake of the most limiting resource (ie.: increase root growth to maximise water uptake under drought conditions). Instead, it suggests that root growth may be influenced by local soil density- during dry conditions the soil is more difficult to penetrate and so root growth is impeded. There were interesting variations in the length and surface area of roots per unit mass, which suggest that the trees at the plots may be modifiying their root morphology in response to changes in soil moisture. The sheer quantity of root length and surface area deserves mention too: i estimate that annual root growth is 2-4 km or 8-12 m2 per m2 of ground surface. This means that for every kilogram of dry root tissue you get 8-10 km or 24-34 m2 of roots!!
5) (see article in press in Journal of Geophysical Research-Biogeosciences) Overall, litter, roots and soil organic matter accounted for about 7, 55 and 33% of total respiration at the study site. Though, the amount of CO2 produced by different soil components varied substantially within and between plots. This variation was not clearly attributable to soil moisture or temperature. Instead, the amount of root mass, and the specific respiration rate of surface organic litter explained the majority of root and litter respiration respectively, and were therefore more useful predictors of soil respiration as a whole. We suggest that ecosystem models could incorporate this information to better simulate the impacts of future climate change upon soil respiration, and ecosystem carbon cycling in the Amazon region.
6) (article in preparation, to be submitted to Journal of Geophysical Research-Biogeosciences) Leaf dark respiration was significantly higher on the Drought plot (7.3 t C/ha/yr) compared to the control Sand plot (5.9 tC/ha/yr). This effect was most pronounced in the uppermost canopy, while leaves nearer the ground showed no clear difference in respiration between plots. This apparent drought effect of about 1.4 t C/ha/yr may appear minor but could significantly alter our picture of the effect of drought upon the net efflux of CO2 into the atmosphere from the forest, given that the current sink estimate for the region (based upon tree stem growth) is only about 0.5 t C/ha/yr.
7) (article in preparation, to be submitted to Global Change Biology) The drought effect on leaf respiration was a bit of a suprise, because otherwise my measurements indicated that the forest was fairly indifferent to the drought treatment. Canopy area, total litterfall, and litter chemistry changed little. Though interestingly, reproductive structures (fruits, flowers, seeds) were clearly lower on the Drought plot. This may have little direct short-term on the ecosystem but if maintained over time could have grave consequences for seedling recruitment, tree-age structure and forest carbon storage capacity. On both plots, stem growth was a minor component of plant production (16% of total) with most growth occuring in foliage and roots. Integrating tree growth and soil respiration measurements, i was able to estimate the net efflux of CO2 from the forest into the atmosphere, for both plots. The control Sand plot appeared to be a small carbon sink of about 0.8 t C/ha/yr while the Drought plot was a net carbon source of 0.5 t C/ha/yr. There is considerable uncertainty around these estimates, but they do highlight one key issue: several atmosphere-biosphere models have predicted that the Amazon may become a net source of CO2 over the next century by simulating a drop in photosythesis, but a rise in soil respiration. While photosynthesis may well fall under drought conditions, neither theory nor experimental results from this, and other studies, supports the idea that soil respiration will necessarily rise. This is for two reasons: microbial activity will tend to be inhibited under drier conditions, and if photosynthesis declines so will the flux of carbon below-ground in the form of root growth, respiration and exudates. A more realistic representation of soil respiration, together with inclusion of other potentially important components of the ecosystem carbon cycle (e.g.: leaf dark respiration, see above) could result in more reliable predictions of the effects of future climate change on the Amazon rainforest.
