This document discusses a study examining the spatial variability of leaf nitrogen (N) concentration in a Eucalyptus woodland ecosystem exposed to elevated atmospheric carbon dioxide (CO2) levels through a free-air CO2 enrichment (FACE) experiment. It finds that the largest source of leaf N variability is between individual trees rather than between plots. Power analyses show the study is adequately powered to statistically detect a 15% decrease in leaf N or a 20% increase in leaf photosynthesis with the experimental design. The conclusions are that the FACE experiment can detect the predicted effects of elevated CO2 on plant physiology and help answer how increased CO2 may drive changes in forest canopy photosynthesis.

1. Spatial variability in leaf N:
detecting the elevated CO2 response in a
Eucalyptus woodland ecosystem in the
Cumberland Plain (‘EucFACE’)
or: How much sampling may be needed?
D. Ellsworth1, K. Crous1, B. Moore1, T. Gimeno1,
J. Powell1, P. Reich1,2
David Ellsworth
Hawkesbury Institute for the
Environment
University of Western Sydney, NSW
AUSTRALIA
D.Ellsworth@uws.edu.au

2. Forest canopies are tall and variable

3. Leaf N is a key ecosystem variable
Eucalyptus
Confers
Net photosynthesis
greenness
(nmol g s )
-2 -1
20
via N-
10 containing
chlorophyll
0
10 20 30 40
-1
Leaf Narea (mg g )
N relates to leaf photosynthetic
protein content
N reflects nutritional content C/N ratio affects biogeochemical
for herbivore feeding cycling

4. Leaf N in plant canopies is relatively
stable & can be remotely sensed
New Hampshire, USA
Smith et al. (2002) Ecol. Applic.
12: 1286

5. There is evidence that much of local-scale
canopy variation in leaf N concentration is tree-
based
E. microcorys: region,
What kind of sampling is site and tree variance in N
needed to resolve a plot-
plot difference within a
native Eucalypt woodland
for leaf N and Anet?
If we conduct an
ecosystem manipulation,
are we able to detect the
outcome?
Moore et al. (2004) Ecol. Monogr. 74: 553

6. Study tract of native Cumberland Plain woodland
located in W. Sydney

7. Increasing global CO2 concentration
400
Direct measure
Air [CO2] (ppm-v)
360
Icecore air The atmospheric gas
CO2 is increasing in
320 the atmosphere
Year 2010
in spite of
389 ppm
280
39% above pre-industrial international
1850 1900 1950 2000
agreements to control
YEAR CO2 emissions
Rate of increase
[CO2] at 520 to 555
1970 – 1979: 1.3 ppm y-1 ppm is expected by
1980 – 1989: 1.6 ppm y1 2050
1990 – 1999: 1.5 ppm y-1
2000 - 2009: 1.9 ppm y-1 Data Source: P. Tans & T. Conway, NOAA/ESRL
http://www.esrl.noaa.gov/gmd/ccgg/iadv

8. EucFACE
involves circular
plots for free-air CO2
enrichment
Plots are 25m diameter
and release CO2 in a
computer-controlled
fashion.
There are 6 plots, 3
ambient and 3 ambient
+150ppm CO2

9. EucFACE
A study of elevated CO2 effects on a mature grassy woodland
ecosystem
Free-Air CO2 Enrichment
Computer-controlled emission of CO2 from vent pipes
CO2
Wind Calm
CO2
Valve emitting CO2 Valves must be modulated rapidly:
Valve closed < 10 sec response
EucFACE controls plot [CO2] at 540 ppm
after 6-month ramp-up period

10. Stepped increase in [CO2] in EucFACE
540
+150ppm
+120ppm
Treatment [CO2]
+90ppm
+60ppm
+30ppm
Step
2 Jan.
Current [CO2]
390
level A/S S/O O/N N/D D/J J/F Years …
Timeframe (months)

11. Stepped increase in [CO2] in EucFACE
125
Daytime treatment [CO2]
in ppm above ambient
100
540
75
50
Timeframe (months)
25
Month

12. Simple hypotheses for elevated CO2 effects
on plant processes
Do the effects we know at the small scale
propagate to affect ecosystem processes
for a mature forest?

13. Leaf net photosynthesis (Anet) and leaf chemistry
is safely measured at ~20m height in the canopy

14. Canopy access to top of
mature E. tereticornis at
22m height

15. Pretreatment leaf %N among 66 mature Eucalyptus
trees (6 month old leaves sampled in autumn)
Grand mean 1.73 ± 0.06%
1.69%
1.69% 1.76%
1.81%
1.77%
1.64%

16. Power curves - likelihood we don’t detect a
CO2 effect on leaf N that is real
Effect size = 0.15
High probability of rejecting
the null and finding a
treatment difference when
Power (1-β) for leaf %N
there actually is a difference
Low probability of rejecting
the null when it is actually
false
# trees sampled per plot

17. How much sampling is needed to detect an
CO2 effect on leaf N?
Effect size = 0.05 Effect size = 0.10 Effect size = 0.20
Power (1-β) for leaf %N
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
# trees sampled per plot

18. How much sampling? Power curves
Effect size = 0.05 Effect size = 0.10 Effect size = 0.20
Power (1-β) for leaf %N
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
# trees sampled per plot

19. Power - likelihood we don’t detect a CO2
effect on leaf N that is real
Effect size = 0.15
High probability of rejecting
the null and finding a
treatment difference when
there actually is a difference
Power (1-β) for leaf %N
With 2 leaves per tree, the possible minimum sample size to be reasonably likely
to detect a -15% effect on leaf N may be about 3 trees per plot
Low probability of rejecting
the null when it is actually
false
# trees sampled per plot

20. Pretreatment leaf net photosynthetic capacity at
390 ppm CO2 among 18 mature Eucalyptus trees (6
month old leaves sampled in autumn)
Grand mean Anet = 16.4 ± 0.7 mmol CO2 m-2 s-1
or 72 nmol CO2 g-1 d.w. s-1
16.8
17.9 18.1
17.0
14.9
13.7

21. Power - likelihood we don’t detect a
CO2 effect on photosynthesis that exists
Effect size = 0.2
High probability of rejecting
the null and finding a
treatment difference when
there actually is a difference
Power (1-β) for Anet
With 2 leaves per tree, the possible minimum sample size to be reasonably likely
to detect a 20% effect on net photosynthesis may be about 3 trees per ring
Low probability of rejecting
the null when it is actually
false
# trees sampled per ring

22. Conclusions – can we detect effects of
elevated CO2 at +150ppm on plants?
 This study evaluated variability in leaf N and Anet
before the CO2 treatment in FACE started.
 The largest source of variability in N is tree-tree, not
plot-plot.
 The experiment is robust enough that we can find a
statistical difference when it really exists, with our
minimal replication.

23. Conclusions – continued
 We can statistically detect a CO2 effect decreasing
leaf N concentration by 15% with 3 rings and
adequate tree subsamples, but not less.
 We can detect a 20% increase in Anet with eCO2 with
our replication.
 FACE has now (2 wks ago) reached full treatment for
[CO2].

24. Acknowledgements
See: www.uws.edu.au/hie/eucface
EucFACE is an initiative supported by:
Commonwealth through DIISRT
and the Australian Research
Council

25. How much did ↑CO2 drive increased
photosynthesis in the canopy?
K. Crous, T. Gimeno data
Ambient
Elevated
+37%
25 Pretreatmt 25 +17% 25
Net photosynthesis
20 20 20
(mol m s )
-2 -1
15 15 15
10 10 10
5 5 5
0 0 0
May 2012 Oct 2012 Feb 2013 est.
(+15% in CO2) (+39% in CO2)
Data from ‘old’ leaves at the canopy top of 18 Eucalyptus trees across the 6
plots. The expected enhancement is estimated from short-term [CO2] increases
at leaf-level and this is an upper bound to what is possible.

26. What does is a free-air experiment?
FACE = free-air CO2 enrichment
[CO2] is computer-controlled
within a large volume