1. Influence of tree crown parameters on the seasonal CO2-exchange of a pine forest in Brasschaat, Belgium.
Jelle Hofman
Promotor: Dr. Sebastiaan Luyssaert
Co-promotoren: Bert Gielen
Prof. Dr. Ivan Janssens

2. Introduction Carbon cycle (soils-oceans-biosphere-atmosphere)
CO2-exchange vegetation:
Photosynthesis  Respiration
Depends on meteorological (temp, VPD, radiation,…) and ecological (LAI, N, chlorophyl,…) parameters
Ecosystems affect atmospheric CO2-concentrations

3. Introduction
Insight in the influence of climate and ecology on the underlying processes that determine CO2-exchange:
Short term: climate parameters are suitable to predict C02-exchange.
Longer term: ecological parameters become increasingly important.
The response of ecosystems (ecological parameters) on both slow and abrupt climate changes will therefore dominate the CO2-flux on the long term.

4. Goal of master thesis
Determine the influence of ecological tree crown parameters on the GPP-course for a tree plot (‘De Inslag’) in Brasschaat.
Influence on different time scales:
Seasonal (2009): LAI, SLA, N, chlorophyl, FAPAR
Interannual ( ): FAPAR

5. Materials and methods
Fluxnet sampling tower in experimental plot ‘De Inslag’:
NEE: “netto ecosystem exchange” by eddy covariance measurements  GPP (µmol/m²s) is derived
Meteorological parameters: Temperature (°C), soil temperature (°C), radiation (W/m²), precipitation (mm), soil water (%vol), vapour pressure (kPa).
FAPAR: “fraction of absorbed photosynthetically active radiation”
Radiation, absorption (chlorophyl)  combination meteo/ecology

6. Materials and methods Own measurements (March 2009 - March 2010):
LAI (m²/m²)  hemispherical photography
SLA (m²/g)  surface/weight ratio needle samples
N-content (w%)  C/N analysis needle samples
Chlorophyl content (µg/ml)  spectrophotometer
34 sampling days:
34 LAI-values (average from 23 photos)
24 needle samples: SLA, N en chlorophyl
All data series were interpolated to daily values in MATLAB.

7. Materials and methods Neural networks: Used parameters:
Arithmetic network of different processing units (“nodes”).
Output (GPP-course) can through nodes (functions) be explained by input variables (LAI, SLA, FAPAR, …).
Network “trained” in MATLAB to improve relation between input-output
Technique is used to represent complex, non-linear relations between variables  natural ecosystems.
Used parameters:
Non linear influence on GPP
Coupled:
vb. temp-humidity, radiation-FAPAR,...

8. Materials en methods
Network was created in which GPP was explained by meteo parameters of  trained/optimalised.
With this network, GPP was estimated based on meteo parameters for 2009 (seasonal) and (interannual).
The estimated GPP (network) and the measured GPP (tower) were plotted next to eachother:
Difference: Residual GPP-course
Difference can’t be explained by the meteo parameters  caused by biological responses or noise
Residual GPP-course is explained in neural networks by the different ecological tree crown parameters.
 How much of the residual variation is explained by ecological parameters?

9. Results
Influence of ecological parameters on residual seasonal GPP-variation (2009):
74% of GPP-course was explained by meteo
All ecological parameters individualy explained a part of the residual GPP-variation:
LAI (51%)>chlorophyl (48%)>SLA (33%)>FAPAR (32%)>N (27%)
Random (25%)
All parameters together explained 40% (Random: 20%)

10. Results
Influence of FAPAR on residual interannual GPP-variation ( ):
72% of GPP-course was explained by meteo parameters
FAPAR explained a part of the residual GPP-variation:
FAPAR: 37%
Random: 14%

11. Conclusion
Ecological parameters correlated both mutually as with meteo parameters as with GPP-course  proves their influence on GPP and the complex relations between the different parameters!
All ecological tree crown parameters explained residual GPP-variation that couldn’t be explained by the meteo parameters.
Residual GPP-variation 2009 (seasonal):
LAI 26%, chlorofyl 23%, SLA 8%, FAPAR 7%, N 2% more than the random data set.
Residuele GPP-variatie (interannual):
FAPAR 23% more than the random data series.
Influence FAPAR increases on longer timescales: 7  23%
Influence meteo decreases on longer timescales: 74  72%

12. Conclusion
Ecological parameters lead both on seasonal and interannual timescales to better GPP-estimates by explaining a part of the residual variation.
Climate and ecology hereby have a combined influence on the CO2-exchange between vegetation and atmosphere through complex mutual relations.
Physical/financial costs of a permanent sampling of ecological parameters must be considered:
 remote sensed: FAPAR!
Worldwide
Different ecosystems
Permanent sampling
More reliable results by evolution in resolution/sensor technology

13. Questions?