Download files
Abstract
Experiments show that forest net primary productivity (NPP) increases as atmospheric [CO2] increases. The aim of this thesis is to use process-based models to analyse the mechanisms underlying this response to atmospheric [CO2].
I used two models named MAESTRA, which is a complex model of canopy photosynthesis and MATE, a simpler model of forest carbon and water balance that is useful for long-term simulations. These models are applied to a Free-Air CO2 Enrichment (FACE) experiment on a sweetgum plantation at Oak Ridge, Tennessee, USA. In this experiment, trees growing at elevated CO2 (eCO2, 550 ppm) have a higher net primary production (NPP) than trees at ambient CO2 (aCO2, 375 ppm). The aim of my research is to understand the reasons for differences in production at aCO2 and eCO2 rings. Simulations of both models are analysed to determine whether differences between NPP at aCO2 and eCO2 are the result of changes in light use efficiency (LUE) or absorbed photosynthetically-active radiation (APAR). A second question is whether the models can explain inter-year variation in NPP. The model focuses on a two-year period, 1999-2000 when extensive data are available on NPP, photosynthesis, leaf area index (LAI), and tree transpiration.
Using MAESTRA I found that:
1. Because LAI was not affected by eCO2, simulated APAR is insensitive to eCO2. Therefore enhanced total canopy photosynthesis at eCO2 is caused by increased LUE.
2. Jmax and Vcmax were slightly higher at aCO2 than in eCO2 (except for May and September 1999), indicating that photosynthesis was not down-regulated at eCO2. Values of LUE were higher in 2000 compared with 1999 at both aCO2 and eCO2. LUE was higher on days when incident PAR (IPAR) was low.
3. APAR was reduced from 1999 to 2000 due to changes in IPAR and LAI. The counter-action between reduced APAR and increased LUE cancelled each other so that NPP simulated by MAESTRA did not differ greatly from 1999 to 2000. This was in contrast with NPP data, which showed considerably higher values in 2000 than in 1999.
4. Modelled transpiration agreed well with sapflux data in mid-season at both aCO2 and eCO2 but transpiration was underestimated at start and end of year.
5. The response of simulated NPP to eCO2 was 18 and 15% in 1999 and 2000, respectively. These results compare with measured NPP responses to eCO2 of 16 and 27%, respectively.
Using MATE model I found that:
1. Simulated APAR is insensitive to eCO2, thus increased GPP at eCO2 was associated with enhanced LUE rather than enhanced APAR.
2. Simulated GPP in 2000 was higher than in 1999 due to increased LUE in 2000, despite reduced annual APAR.
3. Modelled transpiration agreed well with sapflux data at aCO2 in mid-season but underestimated transpiration at eCO2. The model underestimated transpiration at start and end of the year.
4. Simulations of plant available water showed wetter soils at eCO2.
5. Modelled NPP showed less response to eCO2 in 1999 (6% increase) than in 2000 (12% increase). These CO2 responses compare with measured NPP increases of 16 and 27% in 1999 and 2000, respectively.