Environmental and Experimental Botany 73 (2011) 116–119
Contents lists available at ScienceDirect
Environmental and Experimental Botany
journal homepage: www.elsevier.com/locate/envexpbot
Conclusive remarks. Reliability and comparability of chlorophyll fluorescence
data from several field teams
Filippo Bussotti a,∗ , Martina Pollastrini a , Chiara Cascio a , Rosanna Desotgiu a , Giacomo Gerosa b ,
Riccardo Marzuoli b , Cristina Nali c , Giacomo Lorenzini c , Elisa Pellegrini c , Maria Giovanna Carucci c ,
Elisabetta Salvatori d , Lina Fusaro d , Massimo Piccotto e , Paola Malaspina f , Alice Manfredi f ,
Enrica Roccotello f , Stefania Toscano g , Elena Gottardini h , Antonella Cristofori h , Alessio Fini i ,
Daniel Weber j , Valentina Baldassarre k , Lorenzo Barbanti l , Andrea Monti l , Reto J. Strasser m
a
University of Firenze, Dept. of Agricultural Biotechnology, Piazzale delle Cascine 28, 50144 Firenze, Italy
Catholic University of Brescia, Dept. of Mathematics and Physics, Brescia, Italy
University of Pisa, Dept. of Tree Science, Entomology and Plant Pathology “Giovanni Scaramuzzi”, Pisa, Italy
d
Sapienza University of Rome, Department of Plant Biology, Roma, Italy
e
University of Trieste, Dept. of Life Science, Trieste, Italy
f
University of Genova, DIP.TE.RIS., Polo Botanico Hanbury, Italy
g
University of Catania, Faculty of Agriculture, Italy
h
IASMA Research and Innovation Centre, Fondazione Edmund Mach, San Michele all’Adige, Trento, Italy
i
University of Firenze, Dept. of Horticulture, Italy
j
Biodiversity & Climate Research Centre, Frankfurt am Main, Germany
k
University of Milano, Dept. of Plant Production, Italy
l
University of Bologna, Dept. of Agroenvironmental Science and Technology, Italy
m
University of Geneva, Dept. of Plant Biology, Switzerland
b
c
a r t i c l e
i n f o
Keywords:
Chlorophyll fluorescence
Field exercises
FV /FM
Harmonization
Intercalibration
a b s t r a c t
Two field exercises were carried out to compare chlorophyll a fluorescence measurements taken in the
field by field teams working on the same project. In the first exercise (2007, Passo Pura, Ampezzo, Udine,
Northern Italy) the operators took measurements on the same leaf areas (maintaining fixed leaf clips);
in the second (2009, Monterotondo Marittimo, Grosseto, Central Italy) the teams worked independently,
but addressing a common research question. The results of the first exercise showed that: (a) FV /FM was
stable and had little variation among teams and instruments; (b) the results from the different teams
correlated well; (c) the most suitable parameters of fast kinetics analysis are those measured on the
normalized fluorescence transients. In the second exercise, when the teams worked independently, the
results were much more variable and the correlations between measurements of different operators
were weak. These results suggest that field chlorophyll a fluorescence measurements taken by different
teams/operators can be comparable only if particular care is taken to the internal variability of the samples
and a standardized sampling strategy is applied. A statistically sound representation of a population can
be then reached.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
In the literature, chlorophyll fluorescence data can be compared
only in relative terms (i.e., significance of differences; relative differences between treatment and control), and the comparison of
their absolute values might be problematic because of the large
variability in instrumental and working conditions. Fluorescence
data are usually expressed as ratios (for example the maximum
∗ Corresponding author.
E-mail address: filippo.bussotti@unifi.it (F. Bussotti).
0098-8472/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.envexpbot.2010.10.023
quantum yield of primary photochemistry in the dark adapted
state, FV /FM , Paillotin, 1976) and/or in arbitrary units, and it is not
possible to refer measurements to a specific units. On the contrary,
other eco-physiological measurements, such as gas exchange measurements, can be quantitatively compared in an absolute scale. For
this reason, datasets from different experiments cannot be merged,
and when more instruments and operators work within the same
project, great care must be taken in comparing the raw data.
In 2007 we started a project aimed at determining the variations
of measurements taken by different instruments under the same
working conditions and, hence, the limits and possibilities of comparison. Assuming that the fluorimeters employed were in good
117
F. Bussotti et al. / Environmental and Experimental Botany 73 (2011) 116–119
A
1.2
P
4000
Relave variable fluorescence
Fluorescence intensity (arbitrary units)
5000
I
J
3000
A
B
C
D
E
2000
O
1000
B
1.0
0.8
0.4
0.2
0.0
0.01
0
0.01
0.1
1
10
100
A
B
C
D
E
0.6
1000
0.1
1
Time [ms]
1000
D
C
0.8
0.40
0.7
0.35
Linkage Distance
FV /FM (B-F)
100
Eucledian Distance
0.45
0.9
0.6
0.5
0.4
A-B:
A-C:
A-D:
A-E:
A-F:
0.3
0.2
0.2
10
Time [ms]
0.3
0.4
0.5
0.6
r = 0.92, p = 0.00
r = 0.90, p = 0.00
r = 0.81, p = 0.00
r = 0.93, p = 0.00
r = 0.91, p = 0.00
0.7
0.8
0.9
B
C
D
E
F
0.30
0.25
0.20
0.15
0.10
D
C
E
F
B
A
FV/FM (A)
45
45
35
35
30
30
25
20
15
25
20
15
10
10
A-B:
A-C:
A-D:
A-E:
5
0
-5
-5
F
40
PIABS (B-E)
PIABS (B-E)
40
E
0
5
10
15
20
25
30
A-B:
A-C:
A-D:
A-E:
5
r = 0.90, p = 0.00
r = 0.94, p = 0.00
r = 0.73, p = 0.00
r = 0.89, p = 0.00
35
PIABS (A)
B
C
D
E
0
-5
-5
0
5
10
15
PIABS (A)
20
r = 0.91, p = 0.00
r = 0.94, p = 0.00
r = 0.73, p = 0.00
r = 0.89, p = 0.00
25
30
35
B
C
D
E
Fig. 1. Results of the exercise at Passo Pura (2007). (A) Example of original fluorescence transients measured at the same leaf sample (same leaf clips) by different surveyors.
(B) The transients reported in (A) were normalized for F0 and FM . (C) FV /FM measured by the six surveyors. Correlation between the reference team (A) and the others (B–F).
Teams A–E used HandyPEA; team F used MiniPAM. (D) Cluster analysis of the different teams, compared according to the variable FV /FM . (E) PIABS measured by the different
surveyors. Correlation between the reference team (A) and the others (C–F). The equations of the straight lines are: A–B = 1.73 + 1.12x; A–C = 1.33 + 0.79x; A–D = 1.93 + 0.48x;
A–E = 2.23 + 0.83x. (F) The same correlation of (E), after transformation with the formula [(1/mx) − a]. The equations were now: A–B: −0.48 + 1.01x; A–C: 0.35 + 0.99x; A–D:
2.02 + 0.99x; A–E: 0.44 + 0.99x.
conditions, differences in the results can derive from their intrinsic
characteristics (different adjustments among similar devices), from
setting (intensity of the lamp, gain, timing of the data acquisition)
and from the dark adaptation (Bussotti, 2007). The activities aimed
at ensuring the same results from different operators are called
Intercalibration. This activity is needed when different operators
and/or research teams are working on the same project (Ferretti
et al., 2009; Bussotti et al., 2009).
A further aim was to verify the possible influence of the sampling
strategy (choice and number of samples) on the results. The question is to verify if, given the same case-study, different surveyors,
with different instruments, reach the same conclusions (Harmonization).
The exercises were carried out in two different pristine areas
during summer 2007 and 2009. We assume that it is possible to
implement a real Quality Assurance (QA, Ferretti, 2009) programme
118
F. Bussotti et al. / Environmental and Experimental Botany 73 (2011) 116–119
F0
FM
M0
VJ
VI
FV /FM
PIABS
Mean
St. dev.
CV
761
2477
0.92
0.57
0.77
0.67
11.37
179
492
0.11
0.03
0.02
0.03
3.08
23.60
20.32
13.37
5.25
3.49
4.69
29.08
by monitoring the systematic errors deriving from the instruments
and sampling strategies employed.
2. The exercise at Passo Pura (2007)
The first exercise was held at Passo Pura (Udine, Northern Italy,
46◦ 25′ 21′′ N–12◦ 44′ 38′′ E, 1400 m asl) in summer 2007. Forests of
beech (Fagus sylvatica L.) and Norway spruce (Picea abies Karst.)
represent the most important plant communities. Six working
groups, from the scientific institutions to which the Authors of
this paper belong, took chlorophyll a fluorescence measurements
with 5 HandyPEA (Hansatech Ins., Norfolk, England) and 1 MiniPAM
(Walz, Effeltrich, Germany) fluorimeters. The groups were anonymous and were listed alphabetically (A–F). Forty leaves (belonging
to 10 vascular plants chosen at the forest edge, i.e., 4 leaves per
plant) were dark adapted, using leaf clips provided for instruments,
20 min before each measurement. All the HandyPEA were fixed to
the same settings (maximum lamp intensity: 3000 mol m−2 s−1 ;
duration: 1′ ; gain: 1). The results were then elaborated as mean
values for each plant, are reported in Fig. 1 and Table 1. Team A,
working with a recently calibrated instrument, was considered as
reference.
An example of fluorescence transients, measured by different
HandyPEA on the same leaf sample, is reported in Fig. 1A. The differences between the teams are obvious but, when the transients
themselves were normalized between F0 and FM (Fig. 1B), all curves
overlapped and, with the exception of team “D”, the differences
disappeared.
The parameters of the original transients, such as F0 and FM ,
show a large coefficient of variation (%CV = [St. dev./mean] × 100)
(Table 1), whereas the normalized signals VJ and VI (for explanation
and formulae see the list in Table 1, Bussotti et al., in this issue), and
FV /FM (in this latter case the MiniPAM, team “F”, was included in the
elaboration) were comparable, with a fairly low %CV. The relatively
high %CV values of M0 reflected the variability of the measurements
in the initial slopes. Lastly, the Performance Index, PIABS was the
most variable parameter being the combination of three different
parameters.
Once the variability of the measurements among the instruments (Intercalibrations) was considered, the further question was
to verify if the results of the surveyors correlated, although intrinsically subject to a certain degree of variability. Fig. 1C, referring to
the most commonly used parameter FV /FM , shows the correlation
of each individual surveyor with reference team “A”. The r-values
were in most cases higher than 0.9, but this coefficient was a bit
lower for the correlation A–D (Fig. 1C).
The Pearson coefficients of correlation (r) were high also for
PIABS (Fig. 1E), but the linear regressions have very different patter
(Fig. 1E). The linear regressions are expressed by y = a + mx equations where “a” is the intercept point on the y-axis, “x” is the
measured PIABS value, and “m” a coefficient which expresses the
inclination of the line (1 = 45◦ , there are no differences between
the two series of measures). Each individual measurement point
was aligned by transformation: [(1/mx) − a]. The result is shown in
Fig. 1F. By correcting the PIABS in this way, the CV for this parameter
decreased from 29% to 9%.
The anomaly of “D” in relation to the other teams is confirmed
in Fig. 1D. Cluster analysis, carried out considering the parameter
FV /FM , shows the distance of “D” team’s results from the others.
3. The exercise at Monterotondo Marittimo (2009)
The second exercise was organized in the nearby of Monterotondo Marittimo (Grosseto, Central Italy, 43◦ 09′ 13′′ N–10◦ 51′ 17′′
E, 600 m asl). The main vegetation types are represented by mesophylous broadleaves forests (Castanea sativa Mill., Quercus cerris L.),
with the presence of Mediterranean evergreen elements (Quercus
suber L., Quercus ilex L.).
That exercise was performed at the end of the workshop
“Chlorophyll fluorescence: from theory to (good) practice” (May 2009)
and was aimed at evaluating the influence of different sampling
strategies. The exercise was carried out at the edge of an active
geothermal area (Parco delle Biancane). Two sites were chosen (A:
directly exposed to the geothermal vapours, and B: control). Three
trees were selected in each site: one individual of Q. suber, one Q.
cerris and one of their hybrid Quercus crenata. The measurements
A
0.9
A
0.8
B
0.7
C
0.6
E
0.5
F
G
0.4
H
0.3
I
QCE
QCR
QSU
QCE
QCR
QSU
D
A
B
0.4
Linkage Distance
Table 1
Results of the exercise at Passo Pura (2007). Mean, standard variation and coefficient
of variation (CV, expressed as % of the mean), among the field teams measuring
the same leaf samples, for some of measured parameters. FV /FM included a MiniPAM. Explication of the parameters and relative formulae can be found in Strasser
et al. (2004) and Bussotti et al. (in this issue). F0 = Minimal fluorescence from a dark
adapted leaf; FM = Maximal fluorescence from a dark adapted leaf; M0 = Slope of the
curve at the origin of the fluorescence rise. It is a measure of the rate of the primary
photochemistry; VJ = Relative variable fluorescence at 2 ms; VI = Relative variable
fluorescence at 30 ms; FV /FM = [FM − F0 ]/FM = Maximum quantum yield of primary
photochemistry; PIABS = Performance Index on absorption basis.
0.3
B
Eucleidean distance
0.2
0.1
0.0
I
F
E
C
G
D
B
H
A
Fig. 2. Results of the exercise at Monterotondo Marittimo (2009). Behaviour of
FV /FM, measured by the nine field teams (A–I) in the two sites (A = geothermal
area; B = control) and tree species (QCE = Quercus cerris; QCR = Quercus crenata;
QSU = Quercus suber). (B) Cluster analysis of the teams, compared according to the
variable FV /FM .
F. Bussotti et al. / Environmental and Experimental Botany 73 (2011) 116–119
were made on the current year leaves. The criteria to choose the
sample leaves (exposure, light or shade leaves, etc.) were defined
autonomously by each group. Nine different groups (anonymous,
alphabetically listed as A through I) worked independently with
5 HandyPEA (Hansatech Ins., Norfolk, England), 1 PEA (Hansatech
Ins., Norfolk, England); 1 PAM-2000 (Walz, Effeltrich, Germany),
1 MiniPAM (Walz, Effeltrich, Germany) and 1 OS-1 (Optiscience
Corporation, Tyngsboro, MA). Each group replicated 5–10 measurements on each tree.
The data analysis took into consideration the behaviour of FV /FM
in the six trees sampled, according to the different surveyors. Fig. 2A
shows a certain degree of heterogeneity in the results. Unlike the
exercise at Passo Pura, here the correlations between surveyors
were weak (data not reported). The results of the clusters are
shown in Fig. 2B. Two “homogeneous” clusters of surveyors were
detectable: one of A together with H, and the other including G,
D and B. The others behaved differently. Team “I” was completely
different from all the others.
A further elaboration concerned the analysis of variance (twoway ANOVA) taking into account the factors “Site” and “Plant”, and
their interaction, for each different surveyors. The aim of this analysis was to check if, even if with different results at tree level, the
general results were similar. Only the surveyors providing a complete dataset were considered. Based on the “Site” factor, significant
differences were found by A, C, D, F, but not by B and G. Based on
the “Plant” factor, significant differences were found only by B, C
and D. Lastly, the interaction between the two factors was found
significant by B, C, D, F and G, but not by A.
4. Conclusive considerations and future directions
Intercalibration between teams can be verified easily by measuring the same leaf sample with different instruments. The extreme
values F0 and FM proved to be quite variable, but FV /FM , as well as the
parameters derived from the normalized curves (for example VJ , VI
and those related to them), have proved to be very robust and comparable across different instruments and measurement conditions.
The good correlation between instruments, however, suggests the
possibility of calculating conversion factors to adjust the results
(see example Fig. 1E and F). Lastly, the presence of an instrument
that gives strongly divergent results (team “D”, see Fig. 1D) highlights the need for an accurate check, before any measurement
campaign, to align or eliminate divergent instruments.
Harmonization requires a more complicated approach. The very
large differences among surveyors observed in the exercise at Monterotondo Marittimo (Fig. 2) can be explained by the different
strategies adopted in choosing the leaf samples within a tree crown
(sun vs. shaded leaves; bottom or top of the branches; top or bottom
of the crown, cardinal direction of the crown section and so on, see
for e.g. Gielen et al., 2007; Sarijeva et al., 2007). The problem of the
sampling strategy across a tree crown is discussed also within the
pan-european programme for forests health monitoring (Luyssaert
et al., 2002). The issues that need to be addressed are: the variability of the leaf responses within the crown, the scientific question
119
to be investigated (assessment of the response in the whole tree,
or just in a specific population of leaves assumed as target), as well
as the sampling strategy taking in account the spatial heterogeneity of the photosynthesis (Sakai and Akiyama, 2005; Strain et al.,
2006). This consideration may appear trivial, but the study of the
variability and the sampling strategy are rarely stated in this kind
of research.
We suggest some (obvious) precautions before starting with
a survey, such as (i) to explicitly identify the target population;
(ii) to adopt common protocols; (iii) to perform common exercises to prevent bias and misinterpretations. The most important
point, however, is likely to be the study of the variability within
and between the population(s) considered, in order to estimate the
sampling errors and produce robust and statistically sound results.
Acknowledgements
We would like to deeply thank Prof. Mauro Tretiach (Trieste) for
the organization of the exercise held at Passo del Pura. We thank
also the Municipality of Monterotondo Marittimo for the logistic
support.
References
Bussotti, F., 2007. Misure ecofisiologiche su piante arboree. Comparabilità e fonti di
errore. Sherwood 133, 45–49.
Bussotti, F., Cozzi, A., Cenni, E., Bettini, D., Sarti, C., Ferretti, M., 2009. Measurement
errors in monitoring tree crown conditions. Journal of Environmental Monitoring 11, 769–773.
Bussotti, F., Desotgiu, R., Cascio, C., Pollastrini, M., Gravano, E., Gerosa, G., Marzuoli,
R., Nali, C., Lorenzini, G., Salvatori, E., Manes, F., Schaub, M., Strasser, R.J., in this
issue. Ozone stress in woody plants assessed with chlorophyll a fluorescence. A
critical reassessment of existing data.
Ferretti, M., 2009. Quality assurance in ecological monitoring—towards a unifying
perspective. Journal of Environmental Monitoring 11, 726–729.
Ferretti, M., König, N., Rautio, P., Sase, H., 2009. Quality assurance (QA) in international forest monitoring programmes: activity, problems and perspectives from
East Asia and Europe. Annals of Forest Science 66, 403.
Gielen, B., Löw, M., Deckmyn, G., Metzger, U., Franck, F., Heerdt, C., Matyssek, R., Valcke, R., Ceulemans, R., 2007. Chronic ozone exposure affects leaf senescence of
adult beech trees: a chlorophyll fluorescence approach. Journal of Experimental
Botany 58, 785–795.
Luyssaert, S., Raitio, H., Vervaeke, P., Mertensb, J., Lustb, N., 2002. Sampling procedure
for the foliar analysis of deciduous trees. Journal of Environmental Monitoring
4, 858–864.
Paillotin, G., 1976. Capture frequency of excitations and energy transfer between
photosynthetic units in the photo system II. Journal of Theoretical Biology 58,
219–235.
Sakai, T., Akiyama, T., 2005. Quantifying the spatio-temporal variability of net
primary production of the understory species. Sasa senanensis, using multipoint measuring techniques. Agricultural and Forest Meteorology 134, 60–
69.
Sarijeva, G., Knapp, M., Lichtenthaler, H.K., 2007. Differences in photosynthetic
activity, chlorophyll and carotenoid levels, and in chlorophyll fluorescence
parameters in green sun and shade leaves of Ginkgo and Fagus. Journal of Plant
Physiology 164, 950–955.
Strain, E, Beardall, J., Thomson, R., Roberts, S., Light, B., 2006. Spatio-temporal variability in the photosynthetic characteristics of Zostera tasmanica measured by
PAM. Acquatic Botany 85, 21–28.
Strasser, A., Tsimilli-Michael, M., Srivastava, A., 2004. Analysis of the fluorescence
transient. In: Papageorgiou, G.C., Govindjee (Eds.), Advances in Photosynthesis
and Respiration Series. Chlorophyll Fluorescence: A Signature of Photosynthesis.
Springer, Dordrecht, NL, pp. 321–362.