Directory UMM :Data Elmu:jurnal:T:Tree Physiology:Vol15.1995:
Tree Physiology 15, 71--83
© 1995 Heron Publishing----Victoria, Canada
Leaf CO2 exchange of Erythrina poeppigiana (Leguminosae:
Phaseolae) in humid tropical field conditions
PEKKA NYGREN
University of Helsinki, Forestry Field Station, FIN-35500 Korkeakoski, Finland
Received December 3, 1993
Summary An idealized model was developed to describe
leaf CO2 exchange in the leguminous tree Erythrina poeppigiana (Walpers) O.F. Cook under well-watered field conditions. Photosynthetic rate in mature leaves (p) was modeled as
a rectangular hyperbolic function of photon flux density (q)
and ambient CO2 concentration (ca), relative photosynthetic
capacity (π) was modeled as a logistic s-function of leaf age
(la), metabolic dark respiration rate (rm) was modeled as an
exponential function of leaf temperature (Tl), and photorespiration rate (rp) was modeled as a hyperbolic function of ca.
Assimilation rate (ac) was modeled as the difference between
the product of p and π and the sum of rm and rp :
ac = p(q,ca )π(la ) − [rm (Tl) + rp (ca )].
The model parameters were estimated separately for five
sources of E. poeppigiana (Clones 2660, 2662, 2687 and 2693
and half-sib Family 2431) from field data measured with a
portable closed-loop gas exchange system at a humid tropical
site in Costa Rica. The between-source differences in leaf CO2
exchange characteristics were small, but statistically significant. Aboveground biomass production was highest in sources
that maintained high relative photosynthetic capacity throughout the leaf life span. Quantum yield varied between 0.046 and
0.067, and light-saturated assimilation rate (q = 2000 µmol
m −2 s −1 and Tl = 28 °C) at natural atmospheric ca (350 µmol
mol −1) was 16.8--19.9 µmol m −2 s −1. Increasing ca to 1000
µmol mol −1 resulted in an approximate doubling of the lightsaturated assimilation rate. Foliole nitrogen concentration,
which was 45.3--51.2 mg g −1 in mature leaves, was positively
correlated with relative photosynthetic capacity. Foliole nitrogen concentration, quantum yield and maximum assimilation
rate of E. poeppigiana are among the highest values observed
in tropical woody legumes.
Keywords: assimilation rate, leaf age, leaf temperature, modeling, nitrogen concentration, photon flux density, photosynthetic capacity, respiration.
Introduction
Erythrina poeppigiana (Walpers) O.F. Cook, which is a nitrogen fixer (Escalante et al. 1984, Lindblad and Russo 1986), is
a commonly used agroforestry tree species in tropical America. It is native to northwestern South America, from Bolivia
to Panama and Venezuela, and has been introduced to Central
America and several Caribbean islands (Krukoff 1969). Its
natural distribution extends from humid tropical lowlands to
1500 m a.s.l. (Holdridge and Poveda 1975). Erythrina poeppigiana, which reaches a height of over 25 m and a diameter
of more than 1 m (Holdridge and Poveda 1975), is the most
popular shade tree for coffee plantations in Costa Rica (Russo
and Budowski 1986). It is also used for shading cacao (Escalante et al. 1984), and more recently for green manuring in
alley cropping (Kass et al. 1989, Nygren and Jiménez 1993)
and as a protein supplement for dairy cattle (Romero et al.
1993).
Agroforestry trees are generally completely pruned twice or
more annually to give a pole with a height of 0.75 to 1.5 m in
alley cropping (Kass et al. 1989, Nygren and Jiménez 1993)
and 3 m in coffee plantations (Russo and Budowski 1986). The
species resprouts easily and the amount of foliage biomass
pruned every 6 months varies from 1.52 to 2.70 kg per tree in
alley cropping (Nygren and Jiménez 1993) to 6.96 kg per tree
(about 2000 kg ha −1) in coffee plantations (Russo and
Budowski 1986). Considerable clonal variation in foliage
biomass production (ranging from 0.72 to 3.03 kg per tree in 6
months) has been observed in E. poeppigiana (Pérez Castellón
1990).
Dynamic models that describe plant growth on the basis of
the response of ecophysiological processes to environmental
factors integrate ecological and physiological information
about plant productivity in a way that permits prediction of
plant productivity in a changing environment. Ecological
growth models may be especially useful in tropical agroforestry, where resource partitioning between the tree and crop
components is the key factor in understanding the functioning
of the system and the effect of management practices on it.
Construction of an agroforestry system model is a hierarchical
process that first requires the formulation of ecological growth
models of the component species. The tree component is
usually less well understood than the other component species.
This is the case for coffee (Russo and Budowski 1986) and
cacao (Escalante et al. 1984) plantations with E. poeppigiana
shade trees, and for alley cropping of E. poeppigiana with
maize and beans (Kass et al. 1989, Nygren and Jiménez 1993).
72
NYGREN
Development of a leaf CO2 assimilation model is the first
step in the construction of an ecological plant growth model
(Thornley and Johnson 1990). To be applicable as part of an
ecological agroforestry production model, the leaf CO2 exchange model should be idealized such that it accurately describes the response of assimilation rate to the most important
environmental factors, but does not require detailed plant
physiological measurements, which may be difficult to obtain
under tropical field conditions. Further, because strong correlations between environmental variables make it difficult to
separate the effects of too many variables (Berninger and Hari
1993), it is necessary to use simple leaf CO2 exchange models
in field research.
I have studied, in the context of the development of an
ecological biomass production model, the effects of environmental factors, foliole nitrogen concentration and leaf age on
the assimilation rate and night respiration of four clones and a
half-sib family of E. poeppigiana under humid tropical conditions. The results are presented in the form of an idealized leaf
CO2 exchange model.
Leaf CO2 exchange model
If we assume that a homogeneous leaf is uniformly irradiated
by photon flux density q and the internal CO2 concentration at
the photosynthetic sites, ci, is uniform, then photosynthetic rate
p can be described by the semiempirical relationship
(Landsberg 1986, cf. Thornley and Johnson 1990):
kq q kc ci
,
p=
k q q + kc ci
(1)
where kq is the quantum yield for incident photon flux density,
and kc is carboxylation efficiency. The CO2 concentration at the
photosynthetic site cannot be measured directly in the field. In
the absence of water stress, C3 leaves tend to maintain a
constant ratio of ci to the ambient CO2 concentration, ca (Long
1985, 1991):
c i = g l c a,
(2)
where gl is the conductance equivalent to the whole diffusion
pathway from the air through the leaf boundary layer, stomata
and mesophyll sap to the chloroplasts. Substitution of ci from
Equation 1 in Equation 2 yields:
p=
kq q kc gl ca
.
k q q + k c gl c a
(3)
Because of the strong correlation between gl and kc, they
cannot be estimated separately from field data, but their product, κc, can be used:
p=
kq q κc ca
,
kq q + κc ca
(4)
The assimilation rate, ac, is the difference between photosyn-
thetic rate and daytime respiration rate, rd :
a c = p − rd .
(5)
Daytime respiration integrates metabolic respiration independent of photosynthesis and photorespiration. The metabolic respiratory rate, rm, was assumed to depend exponentially
on leaf temperature, Tl (Larcher 1975):
r m = rm (20 ) exp[kr(Tl − 20 )] ,
(6)
where rm(20) is the respiration rate at a leaf temperature of
20 °C and kr is the rate of change of rm.
Biochemically, photorespiration is based on the dual affinity
of Rubisco for O2 and CO2: in light, photorespiration competes
with photosynthesis for Rubisco. If we assume that, at a constant ambient CO2/O2 concentration ratio, the photorespiration
rate is proportional to the photosynthetic rate, then the absolute
photorespiration rate cannot be estimated from field data.
Changes in ambient CO2 concentration alter the CO2/O2 concentration ratio and, subsequently, the photorespiration rate;
low ambient CO2 concentrations favor photorespiration,
whereas high ambient CO2 concentrations suppress it (Edwards and Walker 1983). This dependency can be modeled as
a hyperbolic function:
r d = rm + αr/ca ,
(7)
where αr is a parameter. The complete function for the assimilation rate is:
ac =
(8)
kq q κc ca
− (rm(20 ) exp[kr(Tl − 20 )] + αr/ca ).
k q q + κc ca
Physiological changes that affect photosynthetic capacity
occur during leaf development. Subsequently, photosynthetic
capacity is affected by leaf age (Larcher 1975, Farquhar and
von Caemmerer 1982). Nitrogen is needed for Rubisco regeneration (von Caemmerer and Farquhar 1981), and leaf nitrogen
concentration is also an important determinant of photosynthetic capacity (Mooney et al. 1984). These effects can be
taken into account by introducing the relative photosynthetic
capacity π into Equation 5:
ac = π p − rd .
(9)
Leaf age, la, was used as an indicator of physiological leaf
development. A modification of the logistic s-curve was applied to describe the relationship between la and π:
π=
α π − β π la
,
1 + exp (γ π − kπ la)
(10)
where απ and βπ are parameters that define the maximum value
of π (the asymptote), kπ is the initial rate of change of π, and γπ
is a parameter that measures the initial photosynthetic rate
(Mead and Curnow 1983).
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
73
An asymptotic function was applied to describe the relationship between foliole nitrogen concentration, nl, and relative
photosynthetic capacity (cf. Cromer et al. 1993):
π = απ( 1 − exp[−kn(nl − nmin )]) ,
(11)
where nmin is the minimum foliole nitrogen concentration required, kn is the rate of change of the relative photosynthetic
capacity, and απ is the maximum photosynthetic capacity
(asymptote).
Field data
Figure 1. Monthly averages of daily photon flux, day and night
temperatures and monthly precipitation at the study site in Turrialba,
Costa Rica, from December 1991 through November 1992.
Experimental site
Gas exchange measurements
The experiment, which was designed to collect data to model
the growth of E. poeppigiana, was carried out at the experimental farm of the Centro Agronómico Tropical de Investigación y Enseñanza (CATIE), situated in Turrialba, Costa Rica
(9°53′ N, 83°39′ W, 600 m a.s.l.). The experiment consisted of
two groups of trees. In one group of 30 trees, the root systems
were separated by a plastic barrier down to a depth of 1 m and
the ground was kept completely weed-free by manual weeding
to facilitate root and nodule sampling. Another group of 30
trees was grown without root enclosures and weeding, but the
ground vegetation was cut every 2 months.
The experiment was established in March 1991 at a planting
density of 4 × 4 m. The tree material consisted of 1.5-m long
rooted cuttings of four clones selected by the Nitrogen Fixing
Tree Project of CATIE (2660, 2662, 2687 and 2693), and
4-month-old greenhouse-grown seedlings of a half-sib family
(2431 of the Latin American Forest Seed Bank, CATIE). Six
trees of each source were planted according to a completely
randomized design in each group.
The trees were pruned and pollarded to 1.5 m on December
12, 1991. A second pruning was carried out on June 12, 1992
in accordance with normal management practices for E. poeppigiana on Costa Rican coffee farms and CATIE’s alley cropping experiments.
Climatic data for the study site, collected by an automatic
weather station (Delta-T Devices, Cambridge, U.K.) are presented in Figure 1. Average daytime relative humidity varied
from 77.1% in March to 82.3% in October; nighttime humidity
was saturating throughout the year. The maximum daytime
temperature was 28--30 °C throughout the year. Nighttime
minimum temperatures dropped to 13--15 °C from December
through March, and were above 17 °C in June and July.
Comparison with long-term climatic data (since 1943) from
the CATIE weather station indicated that rainfall from December 1991 until the end of February 1992 was considerably
lower than the long-term mean (long-term means of 328, 171
and 141 mm, respectively). In May, more than half (119 mm)
of the monthly precipitation was received during a rain storm.
The atmospheric pressure in Turrialba is stable throughout the
year (94.2 ± 0.6 kPa).
The gas exchange measurements were made between March 3
and June 9, 1992, with a battery-operated Li-Cor LI-6200
portable photosynthesis system (Li-Cor Inc., Lincoln, NE,
USA), which contains an infrared gas analyzer for determination of assimilation rate and functions as a steady-state
porometer for the determination of transpiration rate. The
system was used in closed mode in all measurements. The
measuring apparatus included a 0.25-l Plexiglas assimilation
chamber lined with Teflon to reduce water adsorption, a quantum sensor (calibration accuracy ± 5%) attached outside the
chamber to monitor photon flux density during the gas exchange measurements, a linearized thermistor (accuracy ±
0.5 °C) to record chamber temperature, a chromel-constantan
thermocouple (accuracy ± 1 °C) to measure leaf temperature,
and a Vaisala Humicap thin film capacitance sensor (Vaisala
Ltd., Helsinki, Finland) to record the relative humidity inside
the chamber.
The measuring system was controlled by a datalogger,
which also calculated assimilation and transpiration rates and
stomatal conductance for H2O according to the formulae presented by von Caemmerer and Farquhar (1981). For every
sample, the sensor and gas analyzer output were read approximately every second, and the mean and range of the instantaneous values were stored. The calculation of CO2 exchange
rate was based on the mean slope of CO2 depletion in the
chamber (Li-Cor Inc. 1990a, 1990b).
Most of the gas exchange measurements were carried out on
the 30 trees with root enclosures, because more information
was available about these ramets and half-sib family individuals than the trees without enclosures. Because the trifoliate
leaves of E. poeppigiana are large (surface area of a mature
foliole is about 1.9--2.7 dm2), only part of a foliole was enclosed in the assimilation chamber. The chamber temperature
was about 4--5 °C higher than the ambient air temperature in
full sunlight, and about 2 °C higher during the nighttime
respiration measurements. The warm air flowing from the
infrared gas analyzer was presumably responsible for the temperature increase at night, and the greater warming in sunny
conditions was due to radiation absorption by the chamber
walls. However, with a normally transpiring leaf in the cham-
74
NYGREN
ber the temperature remained constant (range ± 1 °C) for more
than 20 min even in full sun.
Chamber temperature and the water vapor pressure deficit
were correlated with photon flux density (r = 0.87 and r = 0.88,
respectively). Leaf temperature was strongly correlated with
both chamber temperature and photon flux density according
to the linear regression (R2 = 0.99):
ity during the life span of the leaves were monitored every 2
weeks at saturating photon flux density on one leaf in every
tree. If the selected leaf became shaded at an older age, the
shading foliage was carefully bent to expose the leaf to full
sun. If this was not possible the leaf was rejected and no
measurements were made.
Measurement of aboveground biomass production
Tl = 0.647 + 0.942Ta + 0.00137q,
(12)
where Ta is chamber temperature.
The responses of assimilation and transpiration to changes
in photon flux density were measured on the terminal foliole
of a selected 5th to 8th fully open trifoliate (2--3 weeks old)
leaf from early morning to midday on a sunny day. Only sun
leaves were selected, because the leaves of E. poeppigiana
emerge in full sun and the older leaves are shaded by the
growing foliage. Preliminary tests indicated that there were no
differences in assimilation rates among the folioles of the same
leaf. A total of five leaves could be monitored during one day,
and the measurements were repeated on one leaf from each of
the trees growing in the root enclosure section. Records in
which the range of instantaneous values of photon flux density
was more than 10% of the average or 100 µmol m −2 s −1 during
the recording were eliminated from the analysis.
The response of assimilation rate to ambient CO2 concentration was determined by injecting CO2-enriched air into the
closed system and following the decrease in CO2 concentration
from about 1000 µmol mol −1 to the CO2 compensation point
of the assimilation rate. This took about 20 min. The results of
McDermitt et al. (1989) indicate that the CO2 response curves
measured by this method are comparable with the results from
steady-state laboratory measurements. Because the leaves
were transpiring at a high rate, the increase in leaf temperature
did not cause any problems. The CO2 response curve measurements were repeated on each tree within the root enclosure
section at saturating photon flux density (1600--2000 µmol
m −2 s −1) and on an overcast day (photon flux density of
300--1000 µmol m −2 s −1).
The metabolic dark respiration rate was determined from
nighttime measurements carried out shortly after sunset and
just before dawn on a selected leaf on each tree with a root
enclosure. Dew formation disturbed the measurements before
dawn and so the transpiration records were not analyzed.
Stabilization of the measuring apparatus with the ambient
humidity before dawn took 30 to 90 min compared with 10 to
15 min during the daytime.
The effect of leaf age on relative photosynthetic capacity
was determined from two sets of measurements. Rapid
changes in young leaves were followed on one marked leaf per
tree in three trees per source within the root enclosure section.
The measurements were carried out every 2 days between
1 and 14 days after leaf emergence, and a control measurement
was made 22 days after leaf emergence. The measurements
were performed at ambient CO2 concentration and at saturating photon flux density (1600--2000 µmol m −2 s −1) whenever
possible. Long-term changes in relative photosynthetic capac-
Leaf biomass and area were determined at the pruning on June
12, 1992 by means of the relationship between branch crosssectional area below the first leaf or bifurcation and leaf
biomass or area supported by the branch as described previously (Nygren et al. 1993). Leaf litter was collected weekly
from the root enclosure section. The litter lying inside the
plastic barrier around each tree was assumed to have fallen
from the same tree. The biomass of pruned green twigs and
woody branches was determined by weighing all the pruned
material for each tree separately, and drying the subsamples for
24 h at 105 °C to determine the water content.
On December 12, 1991 and June 12, 1992, stem volume was
determined by measuring the pole diameter at 20 cm from the
base and 20 cm from the top. The volume of the pole was
estimated as a cylinder. The mean of the pole cross-sectional
areas at the measuring points was used as the basal area of the
cylinder (Kilkki 1982). Core samples of known volume were
taken from the pole and dried for 24 h at 105 °C to determine
the wood bulk density. The stem volume increment between
December 12, 1991 and June 12, 1992 was multiplied by the
average bulk density (0.25 kg dm −3) to calculate stem biomass
increment.
Determination of leaf life span and nitrogen concentration
Average leaf life span was determined by marking 10 newly
emerged leaves on each tree in the root enclosure section.
Initially, the leaves were checked twice a week and severe
insect or pathogen damage was noted. After the first appearance of senescent leaves, the leaves were checked daily and the
age at shedding due to senescence was noted. The number of
leaves shed as a result of herbivory or pathogen damage before
senescence was used to compute an average herbivory/pathogen loss percentage by source.
To determine changes in foliole nitrogen concentration during the leaf life span, 25 newly emerged leaves were marked
after pruning in each tree in the root enclosure section. Folioles
of five leaves were sampled 2 and 4 weeks after marking, and
thereafter at monthly intervals for the determination of total
nitrogen content by the micro-Kjeldahl method (Müller 1961).
An additional sample of yellow leaves, removed just before
shedding, was also analyzed.
Specific leaf mass was determined from the foliole nitrogen
concentration samples. The area of each foliole was determined by means of Chacón’s equation (Chacón Espinoza
1990):
Af = lm wm ff,
(13)
where lm is the maximum length of the foliole, wm is the
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
maximum width of the foliole, ff is the form factor and Af is the
area of the foliole. The folioles were measured in the field to
the nearest millimeter with a ruler. The form factor applied was
0.58 (Chacón Espinoza 1990). Because the samples for nitrogen determination were dried for 48 h at 70 °C to avoid
volatilization losses of N, their mass was corrected to give a
value equivalent to the samples dried for 24 h at 105 °C by
multiplying the sample mass by 0.95 (Nygren et al. 1993).
Parameter estimation and model evaluation
Parameter values for Equations 8 and 10 were estimated from
the data by applying the multivariate secant method for least
squares curve fitting (SAS Institute Inc., Cary, NC, USA). The
overall fit of the models was evaluated by means of the modeling efficiency, EF (Mayer and Butler 1993):
EF = 1 −
Σ(yi − y^i)2
_ ,
Σ(yi − y)2
(14)
where yi is the value of the ith observation of the dependent
_
variable, y^i its estimated value for the ith observation, and y is
the mean of the observed values of the dependent variable. In
the case of linear regression with intercept, EF is equal to the
determination coefficient, R2. In the case of nonlinear regression, EF equally measures the proportion of variance explained
by the model, but its lower limit is negative infinity because the
data are compared against a fixed line, not against the best fit
line. The upper limit of EF is unity also in the case of nonlinear
regression (Mayer and Butler 1993).
The parameter estimation was carried out separately for
each source. To detect the significance of the between-source
differences, analyses of residual variance of the fitted models
for each source and for the combined data set were carried out
(Mead and Curnow 1983).
The parameters of the metabolic respiration model were
estimated from the night respiration measurements. The value
of rm(20) was determined as the average night respiration rate
measured at 20 °C (measurements before dawn). The measurements after sunset were mainly carried out at a leaf temperature
of 24 °C. Their average was computed and kr determined
analytically from Equation 6.
75
Results
Metabolic respiration
The night respiration rate at 20 °C was 0.60--0.70 µmol m −2
s −1, except in Clone 2693 (Table 1). More variability among
sources was observed at 24 °C than at 20 °C. Accordingly, the
response of the metabolic respiration rate to leaf temperature
varied among sources and was strongest in Clone 2662 and
weakest in Clone 2693. The metabolic respiration rate approximately doubled for every 10 °C increase in leaf temperature in
Clones 2660, 2662 and 2687, but increased only 1.17-fold in
Clone 2693 and about 1.75-fold in the half-sib Family 2431.
Effects of photon flux density and ambient CO2
concentration on assimilation rate
A typical response of assimilation rate of mature (2- to 3-weekold) leaves of E. poeppigiana to changes in photon flux density
is presented in Figure 2 for Clone 2660. Most of the observed
variation was caused by depletion of the ambient CO2 concentration in the closed-loop system during the measurements, but
a minor part of the variation was due to variations in leaf
temperature among measurements. Differences in assimilation
rates between leaves were small (Figure 2). Maximum assimilation rate was about 20 µmol m −2 s −1 in all sources studied.
The response curves to ambient CO2 concentration showed
little variation (Figure 2). The curves at different photon flux
densities had the same form, but differed in maximum assimilation rate. The curves measured at photon flux densities of
1560 and 2060 µmol m −2 s −1 had similar maximum rates,
about 40 µmol m −2 s −1. The observed maximum was slightly
above 40 µmol m −2 s −1 in all sources, indicating a strong
response to the elevated CO2 concentration. The apparent ‘‘lack
of fit’’ of the estimated assimilation rate at 900 µmol m −2 s −1
was due to differences between leaves; the curve was fitted to
all observations on Clone 2687, not to a specific leaf.
No apparent systematic deviation was observed in the plots
of model residuals against photon flux density, ambient CO2
concentration and leaf temperature, except for a slight underestimation at low ambient CO2 concentration in Clone 2662
(Figure 3). The range of residuals was large, but they were
systematically distributed on both sides of the zero (perfect fit)
line. The same was also true for the plot of the residuals against
leaf-to-air water vapor pressure difference (Figure 3), indicating that the exclusion of this environmental factor from the
Table 1. Average metabolic respiration rate ± SD at leaf temperatures of 20 and 24 °C in mature leaves of Erythrina poeppigiana, calculated from
the night respiration measurements. Means followed by the same letter within a column are not significantly different (Duncan’s multiple range
test at 5%).
Tree source
n
rm(20)
(µmol m 2 s −1)
n
rm(24)
(µmol m −2 s −1)
Rate of change
(per °C)
2431
2660
2662
2687
2693
Combined
18
9
18
21
15
81
0.68 ± 0.11 b
0.72 ± 0.24 b
0.64 ± 0.17 b
0.61 ± 0.14 b
0.89 ± 0.07 a
0.69 ± 0.17
44
44
33
49
45
251
0.85 ± 0.18 bc
0.97 ± 0.23 a
0.87 ± 0.40 abc
0.80 ± 0.18 c
0.95 ± 0.18 ab
0.89 ± 0.18
0.056
0.074
0.077
0.068
0.016
0.064
76
NYGREN
photosynthetic rate, analysis of residual variance of the complete assimilation rate models (Equation 8) fitted for each
source and for the combined data set was carried out. The
results indicated highly significant differences (F12,1034 =
14.05, P < 0.001).
Model estimates of the assimilation rates at a photon flux
density of 2000 µmol m −2 s −1, a leaf temperature of 28 °C and
an ambient CO2 concentration of 350 µmol mol −1 were 17.0,
18.9, 18.1, 19.9 and 16.8 µmol m −2 s −1 in the half-sib Family
2431 and Clones 2660, 2662, 2687 and 2693, respectively. At
1000 µmol mol −1, the respective assimilation rates were estimated to be 40.1, 43.5, 40.6, 44.7 and 41.5 µmol m −2 s −1.
Effect of leaf development on relative photosynthetic capacity
Figure 2. Top: Assimilation rate as a function of photon flux density
in mature leaves of Clone 2660 of Erythrina poeppigiana. Symbols
represent observations on different leaves (n = 6) and the curve
represents the estimate by Equation 8 at ambient CO2 concentration
(300 µmol mol − 1) and a leaf temperature of 28 °C. Bottom: Assimilation rate as a function of ambient CO2 concentration at photon flux
densities of 310 (j), 900 (h), 1560 (d) and 2060 (s) µmol m −2 s − 1
in mature leaves of Clone 2687. Symbols represent observations on
individual leaves and the curves represent the average fit of Equation 8
for the clone.
assimilation rate model was justified. The modeling efficiency
varied between 0.88 and 0.96 (Table 2).
The quantum yield for incident radiation, kq, was similar in
the half-sib Family 2431 and Clones 2660 and 2687, but was
significantly lower (significance determined according to the
overlap of 95% confidence intervals) in Clone 2662 and higher
in Clone 2693 (Table 2). Slightly more variation was observed
in the value of κc, which combines the conductance of the CO2
pathway from the air to the chloroplasts and the carboxylation
efficiency. It was highest in Clones 2660, 2662 and 2687. The
95% confidence intervals for the quantum yield and κc estimates were relatively narrow, indicating a reliable fit of these
parameters.
The value of the photorespiration parameter, αr, was highest
in Clone 2662, followed by Clone 2687, and was lowest in
Clone 2693; however, only the difference between Clones
2662 and 2693 was significant (Table 2). The confidence
intervals of αr were wider than those of kq and κc, indicating
that αr parameter estimation was less accurate than the parameter estimation of photosynthetic rate. Because significant
differences were observed in the values of the parameters of
Leaf age satisfactorily explained changes in the relative photosynthetic capacity of the leaves. The modeling efficiency varied between 0.81 and 0.92 (Table 3), and the residuals plotted
against leaf age, although scattered, did not show any systematic deviation (Figure 4). The slope of the model was significantly steeper in the half-sib Family 2431 and Clone 2693 than
in Clones 2660 and 2662. The 95% confidence interval of the
slope parameter βπ of Clone 2660 included zero, indicating
that the parameter was not important for the model. The confidence intervals of the values of the parameters γπ and kp,
which measure initial rate of change of the relative photosynthetic capacity, overlapped between all sources. According to
the analysis of residual variance, the differences between the
relative photosynthetic capacity models fitted by source were
significant (F16,829 = 13.27, P < 0.001).
Equation 10 did not hold for senescent leaves from about
6 days before shedding, but the relative photosynthetic capacity declined with a steep slope (Figure 5). This senescence
slope was determined analytically as a straight line, which
intercepted the curve of Equation 10 at a leaf age corresponding to 6 days before the average shedding age, and was zero at
the average shedding age of each source (Table 4).
The average leaf life span was short and varied among
sources; leaves of the half-sib Family 2431 and Clone 2662
had a life span almost 2 weeks longer than leaves of Clones
2660 and 2687 (Table 4). The premature loss of leaves as a
result of herbivory or pathogen damage was lowest in Clone
2662 at 8.3%, and highest in Clone 2687 at 50.0%. Herbivory
or pathogen loss was 28.3, 35.0 and 48.3% for sources 2660,
2431 and 2693, respectively.
Three patterns of leaf development were detected (Figure 5):
long life span with a considerable decrease in photosynthetic
capacity as a function of leaf age (half-sib Family 2431), short
life span with almost constant photosynthetic capacity (Clone
2660) and short life span with a considerable decrease in
photosynthetic capacity (Clone 2687). Development of leaves
in Clones 2662 and 2693 followed intermediate patterns (Tables 3 and 4): long life span with a moderate decrease in
photosynthetic capacity (Clone 2662) and intermediate life
span with a considerable decrease in photosynthetic capacity
(Clone 2693).
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
77
Figure 3. Residuals of the assimilation rate model for mature leaves of Erythrina poeppigiana (Equation 8) as a function of photon flux density,
ambient CO2 concentration, leaf temperature and leaf-to-air water vapor pressure difference.
Table 2. Parameter values with 95% confidence interval for the rectangular hyperbolae describing the dependence of assimilation rate on ambient
CO2 concentration and incident photon flux density in mature leaves of Erythrina poeppigiana; Equation 8 adjusted to data on the effects of photon
flux density and CO2 concentration.
Tree source
kq
κc
αr
SSE
2431
0.055
0.049--0.061
0.056
0.052--0.060
0.046
0.043--0.048
0.054
0.049--0.058
0.067
0.059--0.075
0.067
0.062--0.072
0.076
0.072--0.079
0.079
0.076--0.083
0.081
0.076--0.086
0.063
0.059--0.066
455
323--586
449
360--538
677
582--771
541
418--663
391
296--485
1438
209
0.88
555
190
0.96
568
202
0.96
1565
224
0.91
969
209
0.95
0.053
0.051--0.056
0.073
0.071--0.075
497
445--549
5926
1046
0.92
2660
2662
2687
2693
Combined
1
df
EF1
Abbreviations: kq = quantum yield (mol CO2 assimilated/mol photons incident on leaf), κc = product of total conductance of the CO2 pathway
from air to chloroplast and carboxylation efficiency (mol m − 2 s−1), αr = parameter of photorespiration rate, SSE = sum of squared errors, df =
degrees of freedom for error, and EF = modeling efficiency.
Foliole nitrogen concentration and leaf mass/leaf area ratio
The foliole nitrogen concentration was high and declined with
leaf age (Figure 6). However, even senescent leaves had a
nitrogen concentration of 23.0--26.4 mg g −1, which was about
half the value observed in vigorous 28- to 56-day-old leaves.
The leaf mass/leaf area ratio was low in young leaves (38--46
g m −2) and increased with leaf age stabilizing at 60--70 g m −2
(Figure 7).
Equation 11 provided a satisfactory description of the rela-
78
NYGREN
Table 3. Parameter values with 95% confidence interval for the logistic s-curves describing the relationship between leaf age and relative
photosynthetic capacity in Erythrina poeppigiana; Equation 10.
Tree source
απ
βπ
γπ
kπ
SSE
df
EF1
2431
1.26
1.20--1.32
1.10
1.03--1.17
1.05
1.01--1.10
1.12
1.06--1.18
1.21
1.15--1.28
0.0067
0.0057--0.0077
0.0021
−0.00065--0.0048
0.0039
0.0026--0.0052
0.0064
0.0043--0.0086
0.0072
0.0056--0.0088
4.78
3.77--5.79
6.25
4.67--7.83
5.71
4.63--6.78
4.10
3.30--4.90
5.66
4.58--6.74
0.53
0.41--0.65
0.81
0.60--1.03
0.79
0.63--0.94
0.68
0.53--0.83
0.66
0.52--0.80
1860
246
0.81
1170
153
0.88
708
162
0.92
1024
140
0.85
510
128
0.90
1.16
1.13--1.18
0.0059
0.0053--0.0065
5.12
4.57--5.67
0.67
0.59--0.74
6622
845
0.84
2660
2662
2687
2693
Combined
1
Abbreviations: απ, βπ = parameters of the maximum photosynthetic capacity (asymptote), kπ = initial rate of change of the relative photosynthetic
capacity (day − 1), γπ = parameter, SSE = sum of squared errors, df = degrees of freedom for error, and EF = modeling efficiency.
Figure 4. Residuals of the relative photosynthetic capacity model for
Erythrina poeppigiana (Equation 10) as a function of leaf age.
Figure 5. Relative photosynthetic capacity as a function of leaf age in
the half-sib Family 2431 and Clones 2660 and 2687 of Erythrina
poeppigiana. Calculated according to Equation 10.
Table 4. Average leaf shedding age ± SD and the equations describing
the ‘‘senescence slope’’ of relative photosynthetic capacity as a function of leaf age in Erythrina poeppigiana.
Tree source
Leaf shedding
age (days)
Senescence slope
equation
Application
range (days)
2431
2660
2662
2687
2693
87 ± 16
74 ± 10
86 ± 15
73 ± 9
80 ± 17
π = 10.4 − 0.120la1
π = 11.8 − 0.160la
π = 10.6 − 0.123la
π = 8.41 − 0.115la
π = 9.03 − 0.113la
82--87
69--74
81--86
68--73
75--80
Combined
81 ± 15
π = 9.57 − 0.120la
76--81
1
Abbreviations: π = relative photosynthetic capacity (unitless), and la
= leaf age (days).
tionship between foliole nitrogen concentration and relative
photosynthetic capacity; the modeling efficiency was 0.98-0.99 in all sources (Table 5). The value of parameter απ (maximum photosynthetic capacity) was fixed beforehand to the
value estimated from the fit of Equation 10 to the data on the
effect of leaf age on relative photosynthetic capacity (Table 3),
Figure 6. Foliole nitrogen concentration as a function of leaf age in
four clones and a half-sib family (2431) of Erythrina poeppigiana.
because there were too few foliole nitrogen concentration data
to estimate the three parameter values reliably. According to
the analysis of residual variance, the models fitted by source
differed significantly (F8,11 = 6.72, P < 0.001).
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
Figure 7. Leaf mass/leaf area ratio as a function of leaf age in four
clones and a half-sib family (2431) of Erythrina poeppigiana.
Stomatal conductance
Stomatal conductance for CO2 varied between 158 and 238
mmol m −2 s −1 at photon flux densities below 1000 µmol m −2
s −1 and between 190 and 366 mmol m −2 s −1 at higher photon
flux densities (Table 6). A typical response of assimilation rate
to stomatal conductance is presented in Figure 8. Stomatal
conductance was strongly affected by photon flux density and
the dependence of assimilation rate on stomatal conductance
followed the same pattern as its dependence on photon flux
density (cf. Figure 2).
Aboveground biomass production
Aboveground biomass production between December 12,
1991 and June 12, 1992 was highest in Clone 2660, followed
by Clone 2662, and was lowest in Clone 2693 (Table 7). The
proportion of harvestable biomass (foliage, green twigs and
woody branches) was highest in sources with high total aboveground biomass production (76 and 70% in Clones 2662 and
2660, respectively). The within-source variation in biomass
production was lowest in Clone 2660 and highest in the halfsib Family 2431 and Clone 2662. Litterfall consisted of foliage
79
litter only and presented the highest within-source variation of
the biomass compartments, probably because its measurement
was less accurate than that of the other compartments.
Correlations were calculated between the parameters of the
CO2 exchange model (Tables 1--3) and the production of total
aboveground and harvestable biomass. The only significant
correlations were found between both total and harvestable
biomass production and the parameter βπ (r = − 0.99, P =
0.0019 and r = −0.99, P = 0.0003, respectively), which measures the rate of decline of the relative photosynthetic capacity
as a function of leaf age. This indicates that sources that
maintain a high photosynthetic capacity during the entire leaf
life span produce more biomass than sources that only maintain a high photosynthetic capacity for part of the leaf life span.
Also the leaf area on June 12, 1992 was correlated with βπ
(r = − 0.88, P = 0.0494). Weaker correlations were detected
between rate of change of metabolic respiration rate, kr, and the
total aboveground (r = 0.73, P = 0.1667) and harvestable
biomass production (r = 0.73, P = 0.1596) and leaf area on
June 12, 1992 (r = 0.76, P = 0.1327).
Discussion
Field data were used to determine the parameter values for an
idealized model describing the effect of environmental factors
and leaf development on leaf CO2 exchange in E. poeppigiana.
The assimilation model was divided into four components, the
photosynthetic rate in mature leaves, relative photosynthetic
capacity, metabolic respiration and photorespiration. The photosynthetic rate in mature leaves was described as a function
of the photon flux density and ambient CO2 concentration, and
the relative photosynthetic capacity as a function of leaf age. It
was introduced into the model as the coefficient of the basic
photosynthesis model described by Equation 4. This is consistent with the observations of Long et al. (1993) that the quantum yields of 11 taxa were not affected by the age of the
photosynthetic organs, but that the maximum assimilation
Table 5. Parameter values with 95% confidence interval for the asymptotic curves describing the relationship between foliole nitrogen
concentration and relative photosynthetic capacity in Erythrina poeppigiana; Equation 11.
Tree source
απ
kn
nmin
SSE
2431
1.26
2660
1.10
2662
1.05
2687
1.12
2693
1.21
0.085
0.064--0.106
0.109
0.057--0.160
0.091
0.054--0.129
0.072
0.026--0.118
0.059
0.039--0.078
26.5
25.0--28.0
23.3
22.0--24.6
22.9
20.9--25.0
26.2
21.6--30.8
26.3
23.3--29.3
Combined
1.16
0.073
0.061--0.084
24.6
23.5--25.8
1
df
EF1
0.0090
3
0.99
0.0028
2
0.99
0.0038
2
0.99
0.0142
2
0.98
0.0051
2
0.99
0.2057
19
0.99
Abbreviations: απ = maximum photosynthetic capacity (asymptote), kn = rate of change of relative photosynthetic capacity (g mg −1), nmin =
minimum foliole nitrogen concentration (mg g −1), SSE = sum of squared errors, df = degrees of freedom for error, and EF = modeling efficiency.
80
NYGREN
Table 6. Average stomatal conductance for CO2 ± SD (mmol m − 2 s − 1)
in mature leaves of Erythrina poeppigiana.
Tree source
2431
2660
2662
2687
2693
Photon flux density
≤ 1000
> 1000
183 ± 122
238 ± 125
158 ± 66
164 ± 83
200 ± 136
300 ± 104
326 ± 122
190 ± 70
366 ± 126
225 ± 94
Figure 8. Assimilation rate as a function of stomatal conductance for
CO2 in mature leaves of Clone 2687 of Erythrina poeppigiana at
different photon flux densities.
rates tended to be lower in young and old organs than in mature
photosynthetic organs. A temperature dependence of quantum
yield (Farquhar and von Caemmerer 1982) was not observed;
it may have been obscured by the strong correlation between
photon flux density and leaf temperature.
Von Caemmerer and Farquhar (1981) observed a decrease in
both the maximum assimilation rate and carboxylation efficiency in Phaseolus vulgaris grown at low nitrogen availability and concluded that nitrogen is needed for Rubisco
regeneration. In the present study, the observed decrease in
photosynthetic capacity as a function of leaf age may have
been caused by the simultaneous decrease in foliole nitrogen
concentration. This possibility is supported by the close rela-
tionship between foliole nitrogen concentration and relative
photosynthetic capacity, and suggests that foliole nitrogen
concentration could be used instead of leaf age to model
relative photosynthetic capacity. The relationship between foliole nitrogen concentration and relative photosynthetic capacity was curvilinear as has been observed in Gmelina arborea
Roxb. leaves (Cromer et al. 1993).
A constant ratio was assumed between ambient and chloroplast CO2 concentration. The conventional method of estimating internal leaf CO2 concentration by applying stomatal
conductance or resistance estimated from transpiration data
(von Caemmerer and Farquhar 1981, Landsberg 1986, Thornley and Johnson 1990, Long and Hällgren 1993) was avoided
for three reasons. First, a constant ratio between ambient and
intercellular CO2 concentration has been observed in leaves of
well-watered C3 plants (reviewed by Long 1985). Second,
according to the optimization theory (Cowan 1977), gas exchange is optimal when the maximal amount of carbohydrates
is produced per unit of water transpired. In non-water-limited
leaves this implies that stomatal opening is determined by the
capacity of mesophyll tissue to fix carbon (Wong et al. 1979),
or that stomatal conductance is affected by the same factors as
assimilation rate (cf. Wong et al. 1985). Third, the determination of stomatal conductance is based on a series of measurements, the accuracy of which is often poor under humid
tropical conditions. For example, an error of 1 °C in leaf
temperature between 30 and 31 °C means a 0.25 kPa error in
the estimate of water vapor pressure in the substomatal cavity.
Further, a 2% error in the humidity sensor output causes a 15%
error in conductance computed according to von Caemmerer
and Farquhar (1981), when the air humidity is about 80%.
However, the error increases to 70% at an air humidity of 90%
(Field and Mooney 1984); 80--90% was a common humidity
range at the experimental site.
Daytime respiration was divided into metabolic respiration,
which was modeled as a function of leaf temperature, and
photorespiration, which was modeled as a function of ambient
CO2 concentration. The metabolic respiration rate, which was
determined from the night respiration rate measurements and
was assumed to continue unaltered during the daytime (cf.
Edwards and Walker 1983), approximately doubled with a
10 °C increase in leaf temperature. At typical leaf temperature
Table 7. Average aboveground biomass production ± SD (kg per tree) of Erythrina poeppigiana from December 12, 1991 to June 12, 1992. Means
followed by the same letter within a row are not significantly different (Duncan’s multiple range test at 5%).
Biomass compartment
Tree source
2431
2660
2662
2687
2693
Foliage
Green twigs
Woody branches
Total harvestable
1.48 ± 0.52 b
0.74 ± 0.39 b
0.90 ± 0.82 c
3.12 ± 1.70 c
2.44 ± 0.25 a
1.55 ± 0.24 a
2.21 ± 0.25 a
6.19 ± 0.58 a
2.54 ± 0.74 a
0.99 ± 0.53 b
1.63 ± 0.81 ab
5.16 ± 2.00 ab
1.75 ± 0.44 b
0.73 ± 0.25 b
1.13 ± 0.40 bc
3.61 ± 1.00 bc
1.47 ± 0.42 b
0.60 ± 0.19 b
0.72 ± 0.35 c
2.79 ± 0.96 c
Litterfall
Stem increment
Total aboveground
0.66 ± 0.42 a
1.28 ± 1.02 a
5.06 ± 2.91 b
1.34 ± 0.80 a
1.37 ± 0.42 a
8.90 ± 0.92 a
0.93 ± 0.43 a
0.75 ± 0.41 ab
6.86 ± 2.66 ab
1.21 ± 0.35 a
0.72 ± 0.38 ab
5.55 ± 1.28 b
1.19 ± 0.56 a
0.38 ± 0.30 b
4.36 ± 1.35 b
Leaf area (m2)
26.3 ± 9.2 bc
43.9 ± 4.4 a
32.5 ± 9.4 bc
34.0 ± 8.4 b
22.8 ± 6.5 c
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
values at the experimental site, metabolic respiration rates
were 1--2 µmol m −2 s −1.
At ambient CO2 concentration, the photorespiration rate was
comparable to the metabolic respiration rate. In the model, it
was assumed that, at a constant atmospheric CO2/O2 concentration ratio, the photorespiration rate was proportional to the
photosynthetic rate, and that the photorespiration term in the
model only measures the deviation of photorespiration rate
from photosynthetic rate when the CO2/O2 concentration ratio
is altered as a result of variations in atmospheric CO2 concentration. Inclusion of the photorespiration term in the model was
valid because a model with only a temperature-dependent
respiration term overestimated the assimilation rate at low
ambient CO2 concentrations. At high atmospheric CO2 concentration, however, the photorespiration term approached
zero and could be omitted from the model.
Both the solubility of CO2 in mesophyll cell sap relative to
O2 solubility and the specificity of Rubisco for CO2 relative to
O2 change in favor of O2 as a function of increasing leaf
temperature, which should result in an increased photorespiration rate relative to the photosynthetic rate (Long 1991). However, such dependence of photorespiration rate on temperature
was not observed in this study. Attempts to model the photorespiration term by applying the changes in CO2/O2 solubility ratio and CO2/O2 specificity factor suggested by Long
(1991), or by a simple ‘‘black box’’ type dependence of photorespiration rate on leaf temperature and ambient CO2 concentration, resulted in a poorer fit of the assimilation rate
model to the data than was achieved with Equation 8.
The failure to detect any dependence of the photorespiration
term in Equation 8 on temperature may be caused by several
interacting factors. Both the solubility ratio and Rubisco specificity factor are less affected by temperature at the high temperatures typical of tropical environments. According to the
formulae presented by Long (1991), the average rate of change
of the internal CO2/O2 concentration ratio is 1% per 1 °C
between 25 and 35 °C. Most of the assimilation rate measurements were carried out within this temperature range. The
81
average rate of change of the Rubisco specificity factor is
somewhat higher, 4% per 1 °C within the same temperature
range. These changes are small enough to be obscured by the
high correlation between photon flux density and leaf temperature inside the assimilation chamber. They are also small
compared to the doubling of the internal CO2/O2 concentration
ratio, calculated according to Long (1991), in favor of CO2
when ambient CO2 concentration increases from 350 to 700
µmol mol −1.
When the parameters of the leaf CO2 exchange model were
compared with the aboveground biomass production, only the
parameter measuring changes in photosynthetic capacity as a
function of leaf age, βπ, showed a significant correlation; the
photosynthetic capacity of the most productive tree sources did
not decrease much during the leaf life span. In the idealized
assimilation model, βπ was the only parameter that was indicative of the plant’s ability to integrate photosynthetic production
over time. The correlation between βπ and biomass production
emphasizes the importance of following leaf CO2 exchange
characteristics for a long time period, if the aim is to relate
these characteristics to biomass production. Comparisons between leaf assimilation and biomass production have usually
been made based on the instantaneous assimilation rate measured at light-saturated conditions, and the results vary from
positive correlation in 10 tropical tree species (Muthuchelian
1992) to no correlation in provenances of Acacia mangium
Willd. (Atipanumpai 1989) and negative correlation in half-sib
families of Robinia pseudoacacia L. (Mebrahtu et al. 1991).
The maximum assimilation rate of E. poeppigiana is among
the highest observed in tropical woody legumes (Table 8), and
is comparable with the values observed in Prosopis chilensis
(Mol.) Stuntz. (Pinto 1989). The high photosynthetic capacity
of E. poeppigiana may be a result of high foliole nitrogen
concentrations (Table 8 and cf. Mooney et al. 1984).
The quantum yield of E. poeppigiana was comparable with
the quantum yield observed in the neotropical rain forest
Caesalpinoids, Copaifera venezuelana Pittier and Harms
(0.065) and Hymenaea courbaril L. (0.056), in laboratory
Table 8. Maximum assimilation rate and nitrogen concentration in mature leaves of some tropical leguminous trees.
Species
Maximum assimilation rate
(µmol m −2 s −1)
Leaf nitrogen concentration
(mg g −1)
Erythrina
© 1995 Heron Publishing----Victoria, Canada
Leaf CO2 exchange of Erythrina poeppigiana (Leguminosae:
Phaseolae) in humid tropical field conditions
PEKKA NYGREN
University of Helsinki, Forestry Field Station, FIN-35500 Korkeakoski, Finland
Received December 3, 1993
Summary An idealized model was developed to describe
leaf CO2 exchange in the leguminous tree Erythrina poeppigiana (Walpers) O.F. Cook under well-watered field conditions. Photosynthetic rate in mature leaves (p) was modeled as
a rectangular hyperbolic function of photon flux density (q)
and ambient CO2 concentration (ca), relative photosynthetic
capacity (π) was modeled as a logistic s-function of leaf age
(la), metabolic dark respiration rate (rm) was modeled as an
exponential function of leaf temperature (Tl), and photorespiration rate (rp) was modeled as a hyperbolic function of ca.
Assimilation rate (ac) was modeled as the difference between
the product of p and π and the sum of rm and rp :
ac = p(q,ca )π(la ) − [rm (Tl) + rp (ca )].
The model parameters were estimated separately for five
sources of E. poeppigiana (Clones 2660, 2662, 2687 and 2693
and half-sib Family 2431) from field data measured with a
portable closed-loop gas exchange system at a humid tropical
site in Costa Rica. The between-source differences in leaf CO2
exchange characteristics were small, but statistically significant. Aboveground biomass production was highest in sources
that maintained high relative photosynthetic capacity throughout the leaf life span. Quantum yield varied between 0.046 and
0.067, and light-saturated assimilation rate (q = 2000 µmol
m −2 s −1 and Tl = 28 °C) at natural atmospheric ca (350 µmol
mol −1) was 16.8--19.9 µmol m −2 s −1. Increasing ca to 1000
µmol mol −1 resulted in an approximate doubling of the lightsaturated assimilation rate. Foliole nitrogen concentration,
which was 45.3--51.2 mg g −1 in mature leaves, was positively
correlated with relative photosynthetic capacity. Foliole nitrogen concentration, quantum yield and maximum assimilation
rate of E. poeppigiana are among the highest values observed
in tropical woody legumes.
Keywords: assimilation rate, leaf age, leaf temperature, modeling, nitrogen concentration, photon flux density, photosynthetic capacity, respiration.
Introduction
Erythrina poeppigiana (Walpers) O.F. Cook, which is a nitrogen fixer (Escalante et al. 1984, Lindblad and Russo 1986), is
a commonly used agroforestry tree species in tropical America. It is native to northwestern South America, from Bolivia
to Panama and Venezuela, and has been introduced to Central
America and several Caribbean islands (Krukoff 1969). Its
natural distribution extends from humid tropical lowlands to
1500 m a.s.l. (Holdridge and Poveda 1975). Erythrina poeppigiana, which reaches a height of over 25 m and a diameter
of more than 1 m (Holdridge and Poveda 1975), is the most
popular shade tree for coffee plantations in Costa Rica (Russo
and Budowski 1986). It is also used for shading cacao (Escalante et al. 1984), and more recently for green manuring in
alley cropping (Kass et al. 1989, Nygren and Jiménez 1993)
and as a protein supplement for dairy cattle (Romero et al.
1993).
Agroforestry trees are generally completely pruned twice or
more annually to give a pole with a height of 0.75 to 1.5 m in
alley cropping (Kass et al. 1989, Nygren and Jiménez 1993)
and 3 m in coffee plantations (Russo and Budowski 1986). The
species resprouts easily and the amount of foliage biomass
pruned every 6 months varies from 1.52 to 2.70 kg per tree in
alley cropping (Nygren and Jiménez 1993) to 6.96 kg per tree
(about 2000 kg ha −1) in coffee plantations (Russo and
Budowski 1986). Considerable clonal variation in foliage
biomass production (ranging from 0.72 to 3.03 kg per tree in 6
months) has been observed in E. poeppigiana (Pérez Castellón
1990).
Dynamic models that describe plant growth on the basis of
the response of ecophysiological processes to environmental
factors integrate ecological and physiological information
about plant productivity in a way that permits prediction of
plant productivity in a changing environment. Ecological
growth models may be especially useful in tropical agroforestry, where resource partitioning between the tree and crop
components is the key factor in understanding the functioning
of the system and the effect of management practices on it.
Construction of an agroforestry system model is a hierarchical
process that first requires the formulation of ecological growth
models of the component species. The tree component is
usually less well understood than the other component species.
This is the case for coffee (Russo and Budowski 1986) and
cacao (Escalante et al. 1984) plantations with E. poeppigiana
shade trees, and for alley cropping of E. poeppigiana with
maize and beans (Kass et al. 1989, Nygren and Jiménez 1993).
72
NYGREN
Development of a leaf CO2 assimilation model is the first
step in the construction of an ecological plant growth model
(Thornley and Johnson 1990). To be applicable as part of an
ecological agroforestry production model, the leaf CO2 exchange model should be idealized such that it accurately describes the response of assimilation rate to the most important
environmental factors, but does not require detailed plant
physiological measurements, which may be difficult to obtain
under tropical field conditions. Further, because strong correlations between environmental variables make it difficult to
separate the effects of too many variables (Berninger and Hari
1993), it is necessary to use simple leaf CO2 exchange models
in field research.
I have studied, in the context of the development of an
ecological biomass production model, the effects of environmental factors, foliole nitrogen concentration and leaf age on
the assimilation rate and night respiration of four clones and a
half-sib family of E. poeppigiana under humid tropical conditions. The results are presented in the form of an idealized leaf
CO2 exchange model.
Leaf CO2 exchange model
If we assume that a homogeneous leaf is uniformly irradiated
by photon flux density q and the internal CO2 concentration at
the photosynthetic sites, ci, is uniform, then photosynthetic rate
p can be described by the semiempirical relationship
(Landsberg 1986, cf. Thornley and Johnson 1990):
kq q kc ci
,
p=
k q q + kc ci
(1)
where kq is the quantum yield for incident photon flux density,
and kc is carboxylation efficiency. The CO2 concentration at the
photosynthetic site cannot be measured directly in the field. In
the absence of water stress, C3 leaves tend to maintain a
constant ratio of ci to the ambient CO2 concentration, ca (Long
1985, 1991):
c i = g l c a,
(2)
where gl is the conductance equivalent to the whole diffusion
pathway from the air through the leaf boundary layer, stomata
and mesophyll sap to the chloroplasts. Substitution of ci from
Equation 1 in Equation 2 yields:
p=
kq q kc gl ca
.
k q q + k c gl c a
(3)
Because of the strong correlation between gl and kc, they
cannot be estimated separately from field data, but their product, κc, can be used:
p=
kq q κc ca
,
kq q + κc ca
(4)
The assimilation rate, ac, is the difference between photosyn-
thetic rate and daytime respiration rate, rd :
a c = p − rd .
(5)
Daytime respiration integrates metabolic respiration independent of photosynthesis and photorespiration. The metabolic respiratory rate, rm, was assumed to depend exponentially
on leaf temperature, Tl (Larcher 1975):
r m = rm (20 ) exp[kr(Tl − 20 )] ,
(6)
where rm(20) is the respiration rate at a leaf temperature of
20 °C and kr is the rate of change of rm.
Biochemically, photorespiration is based on the dual affinity
of Rubisco for O2 and CO2: in light, photorespiration competes
with photosynthesis for Rubisco. If we assume that, at a constant ambient CO2/O2 concentration ratio, the photorespiration
rate is proportional to the photosynthetic rate, then the absolute
photorespiration rate cannot be estimated from field data.
Changes in ambient CO2 concentration alter the CO2/O2 concentration ratio and, subsequently, the photorespiration rate;
low ambient CO2 concentrations favor photorespiration,
whereas high ambient CO2 concentrations suppress it (Edwards and Walker 1983). This dependency can be modeled as
a hyperbolic function:
r d = rm + αr/ca ,
(7)
where αr is a parameter. The complete function for the assimilation rate is:
ac =
(8)
kq q κc ca
− (rm(20 ) exp[kr(Tl − 20 )] + αr/ca ).
k q q + κc ca
Physiological changes that affect photosynthetic capacity
occur during leaf development. Subsequently, photosynthetic
capacity is affected by leaf age (Larcher 1975, Farquhar and
von Caemmerer 1982). Nitrogen is needed for Rubisco regeneration (von Caemmerer and Farquhar 1981), and leaf nitrogen
concentration is also an important determinant of photosynthetic capacity (Mooney et al. 1984). These effects can be
taken into account by introducing the relative photosynthetic
capacity π into Equation 5:
ac = π p − rd .
(9)
Leaf age, la, was used as an indicator of physiological leaf
development. A modification of the logistic s-curve was applied to describe the relationship between la and π:
π=
α π − β π la
,
1 + exp (γ π − kπ la)
(10)
where απ and βπ are parameters that define the maximum value
of π (the asymptote), kπ is the initial rate of change of π, and γπ
is a parameter that measures the initial photosynthetic rate
(Mead and Curnow 1983).
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
73
An asymptotic function was applied to describe the relationship between foliole nitrogen concentration, nl, and relative
photosynthetic capacity (cf. Cromer et al. 1993):
π = απ( 1 − exp[−kn(nl − nmin )]) ,
(11)
where nmin is the minimum foliole nitrogen concentration required, kn is the rate of change of the relative photosynthetic
capacity, and απ is the maximum photosynthetic capacity
(asymptote).
Field data
Figure 1. Monthly averages of daily photon flux, day and night
temperatures and monthly precipitation at the study site in Turrialba,
Costa Rica, from December 1991 through November 1992.
Experimental site
Gas exchange measurements
The experiment, which was designed to collect data to model
the growth of E. poeppigiana, was carried out at the experimental farm of the Centro Agronómico Tropical de Investigación y Enseñanza (CATIE), situated in Turrialba, Costa Rica
(9°53′ N, 83°39′ W, 600 m a.s.l.). The experiment consisted of
two groups of trees. In one group of 30 trees, the root systems
were separated by a plastic barrier down to a depth of 1 m and
the ground was kept completely weed-free by manual weeding
to facilitate root and nodule sampling. Another group of 30
trees was grown without root enclosures and weeding, but the
ground vegetation was cut every 2 months.
The experiment was established in March 1991 at a planting
density of 4 × 4 m. The tree material consisted of 1.5-m long
rooted cuttings of four clones selected by the Nitrogen Fixing
Tree Project of CATIE (2660, 2662, 2687 and 2693), and
4-month-old greenhouse-grown seedlings of a half-sib family
(2431 of the Latin American Forest Seed Bank, CATIE). Six
trees of each source were planted according to a completely
randomized design in each group.
The trees were pruned and pollarded to 1.5 m on December
12, 1991. A second pruning was carried out on June 12, 1992
in accordance with normal management practices for E. poeppigiana on Costa Rican coffee farms and CATIE’s alley cropping experiments.
Climatic data for the study site, collected by an automatic
weather station (Delta-T Devices, Cambridge, U.K.) are presented in Figure 1. Average daytime relative humidity varied
from 77.1% in March to 82.3% in October; nighttime humidity
was saturating throughout the year. The maximum daytime
temperature was 28--30 °C throughout the year. Nighttime
minimum temperatures dropped to 13--15 °C from December
through March, and were above 17 °C in June and July.
Comparison with long-term climatic data (since 1943) from
the CATIE weather station indicated that rainfall from December 1991 until the end of February 1992 was considerably
lower than the long-term mean (long-term means of 328, 171
and 141 mm, respectively). In May, more than half (119 mm)
of the monthly precipitation was received during a rain storm.
The atmospheric pressure in Turrialba is stable throughout the
year (94.2 ± 0.6 kPa).
The gas exchange measurements were made between March 3
and June 9, 1992, with a battery-operated Li-Cor LI-6200
portable photosynthesis system (Li-Cor Inc., Lincoln, NE,
USA), which contains an infrared gas analyzer for determination of assimilation rate and functions as a steady-state
porometer for the determination of transpiration rate. The
system was used in closed mode in all measurements. The
measuring apparatus included a 0.25-l Plexiglas assimilation
chamber lined with Teflon to reduce water adsorption, a quantum sensor (calibration accuracy ± 5%) attached outside the
chamber to monitor photon flux density during the gas exchange measurements, a linearized thermistor (accuracy ±
0.5 °C) to record chamber temperature, a chromel-constantan
thermocouple (accuracy ± 1 °C) to measure leaf temperature,
and a Vaisala Humicap thin film capacitance sensor (Vaisala
Ltd., Helsinki, Finland) to record the relative humidity inside
the chamber.
The measuring system was controlled by a datalogger,
which also calculated assimilation and transpiration rates and
stomatal conductance for H2O according to the formulae presented by von Caemmerer and Farquhar (1981). For every
sample, the sensor and gas analyzer output were read approximately every second, and the mean and range of the instantaneous values were stored. The calculation of CO2 exchange
rate was based on the mean slope of CO2 depletion in the
chamber (Li-Cor Inc. 1990a, 1990b).
Most of the gas exchange measurements were carried out on
the 30 trees with root enclosures, because more information
was available about these ramets and half-sib family individuals than the trees without enclosures. Because the trifoliate
leaves of E. poeppigiana are large (surface area of a mature
foliole is about 1.9--2.7 dm2), only part of a foliole was enclosed in the assimilation chamber. The chamber temperature
was about 4--5 °C higher than the ambient air temperature in
full sunlight, and about 2 °C higher during the nighttime
respiration measurements. The warm air flowing from the
infrared gas analyzer was presumably responsible for the temperature increase at night, and the greater warming in sunny
conditions was due to radiation absorption by the chamber
walls. However, with a normally transpiring leaf in the cham-
74
NYGREN
ber the temperature remained constant (range ± 1 °C) for more
than 20 min even in full sun.
Chamber temperature and the water vapor pressure deficit
were correlated with photon flux density (r = 0.87 and r = 0.88,
respectively). Leaf temperature was strongly correlated with
both chamber temperature and photon flux density according
to the linear regression (R2 = 0.99):
ity during the life span of the leaves were monitored every 2
weeks at saturating photon flux density on one leaf in every
tree. If the selected leaf became shaded at an older age, the
shading foliage was carefully bent to expose the leaf to full
sun. If this was not possible the leaf was rejected and no
measurements were made.
Measurement of aboveground biomass production
Tl = 0.647 + 0.942Ta + 0.00137q,
(12)
where Ta is chamber temperature.
The responses of assimilation and transpiration to changes
in photon flux density were measured on the terminal foliole
of a selected 5th to 8th fully open trifoliate (2--3 weeks old)
leaf from early morning to midday on a sunny day. Only sun
leaves were selected, because the leaves of E. poeppigiana
emerge in full sun and the older leaves are shaded by the
growing foliage. Preliminary tests indicated that there were no
differences in assimilation rates among the folioles of the same
leaf. A total of five leaves could be monitored during one day,
and the measurements were repeated on one leaf from each of
the trees growing in the root enclosure section. Records in
which the range of instantaneous values of photon flux density
was more than 10% of the average or 100 µmol m −2 s −1 during
the recording were eliminated from the analysis.
The response of assimilation rate to ambient CO2 concentration was determined by injecting CO2-enriched air into the
closed system and following the decrease in CO2 concentration
from about 1000 µmol mol −1 to the CO2 compensation point
of the assimilation rate. This took about 20 min. The results of
McDermitt et al. (1989) indicate that the CO2 response curves
measured by this method are comparable with the results from
steady-state laboratory measurements. Because the leaves
were transpiring at a high rate, the increase in leaf temperature
did not cause any problems. The CO2 response curve measurements were repeated on each tree within the root enclosure
section at saturating photon flux density (1600--2000 µmol
m −2 s −1) and on an overcast day (photon flux density of
300--1000 µmol m −2 s −1).
The metabolic dark respiration rate was determined from
nighttime measurements carried out shortly after sunset and
just before dawn on a selected leaf on each tree with a root
enclosure. Dew formation disturbed the measurements before
dawn and so the transpiration records were not analyzed.
Stabilization of the measuring apparatus with the ambient
humidity before dawn took 30 to 90 min compared with 10 to
15 min during the daytime.
The effect of leaf age on relative photosynthetic capacity
was determined from two sets of measurements. Rapid
changes in young leaves were followed on one marked leaf per
tree in three trees per source within the root enclosure section.
The measurements were carried out every 2 days between
1 and 14 days after leaf emergence, and a control measurement
was made 22 days after leaf emergence. The measurements
were performed at ambient CO2 concentration and at saturating photon flux density (1600--2000 µmol m −2 s −1) whenever
possible. Long-term changes in relative photosynthetic capac-
Leaf biomass and area were determined at the pruning on June
12, 1992 by means of the relationship between branch crosssectional area below the first leaf or bifurcation and leaf
biomass or area supported by the branch as described previously (Nygren et al. 1993). Leaf litter was collected weekly
from the root enclosure section. The litter lying inside the
plastic barrier around each tree was assumed to have fallen
from the same tree. The biomass of pruned green twigs and
woody branches was determined by weighing all the pruned
material for each tree separately, and drying the subsamples for
24 h at 105 °C to determine the water content.
On December 12, 1991 and June 12, 1992, stem volume was
determined by measuring the pole diameter at 20 cm from the
base and 20 cm from the top. The volume of the pole was
estimated as a cylinder. The mean of the pole cross-sectional
areas at the measuring points was used as the basal area of the
cylinder (Kilkki 1982). Core samples of known volume were
taken from the pole and dried for 24 h at 105 °C to determine
the wood bulk density. The stem volume increment between
December 12, 1991 and June 12, 1992 was multiplied by the
average bulk density (0.25 kg dm −3) to calculate stem biomass
increment.
Determination of leaf life span and nitrogen concentration
Average leaf life span was determined by marking 10 newly
emerged leaves on each tree in the root enclosure section.
Initially, the leaves were checked twice a week and severe
insect or pathogen damage was noted. After the first appearance of senescent leaves, the leaves were checked daily and the
age at shedding due to senescence was noted. The number of
leaves shed as a result of herbivory or pathogen damage before
senescence was used to compute an average herbivory/pathogen loss percentage by source.
To determine changes in foliole nitrogen concentration during the leaf life span, 25 newly emerged leaves were marked
after pruning in each tree in the root enclosure section. Folioles
of five leaves were sampled 2 and 4 weeks after marking, and
thereafter at monthly intervals for the determination of total
nitrogen content by the micro-Kjeldahl method (Müller 1961).
An additional sample of yellow leaves, removed just before
shedding, was also analyzed.
Specific leaf mass was determined from the foliole nitrogen
concentration samples. The area of each foliole was determined by means of Chacón’s equation (Chacón Espinoza
1990):
Af = lm wm ff,
(13)
where lm is the maximum length of the foliole, wm is the
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
maximum width of the foliole, ff is the form factor and Af is the
area of the foliole. The folioles were measured in the field to
the nearest millimeter with a ruler. The form factor applied was
0.58 (Chacón Espinoza 1990). Because the samples for nitrogen determination were dried for 48 h at 70 °C to avoid
volatilization losses of N, their mass was corrected to give a
value equivalent to the samples dried for 24 h at 105 °C by
multiplying the sample mass by 0.95 (Nygren et al. 1993).
Parameter estimation and model evaluation
Parameter values for Equations 8 and 10 were estimated from
the data by applying the multivariate secant method for least
squares curve fitting (SAS Institute Inc., Cary, NC, USA). The
overall fit of the models was evaluated by means of the modeling efficiency, EF (Mayer and Butler 1993):
EF = 1 −
Σ(yi − y^i)2
_ ,
Σ(yi − y)2
(14)
where yi is the value of the ith observation of the dependent
_
variable, y^i its estimated value for the ith observation, and y is
the mean of the observed values of the dependent variable. In
the case of linear regression with intercept, EF is equal to the
determination coefficient, R2. In the case of nonlinear regression, EF equally measures the proportion of variance explained
by the model, but its lower limit is negative infinity because the
data are compared against a fixed line, not against the best fit
line. The upper limit of EF is unity also in the case of nonlinear
regression (Mayer and Butler 1993).
The parameter estimation was carried out separately for
each source. To detect the significance of the between-source
differences, analyses of residual variance of the fitted models
for each source and for the combined data set were carried out
(Mead and Curnow 1983).
The parameters of the metabolic respiration model were
estimated from the night respiration measurements. The value
of rm(20) was determined as the average night respiration rate
measured at 20 °C (measurements before dawn). The measurements after sunset were mainly carried out at a leaf temperature
of 24 °C. Their average was computed and kr determined
analytically from Equation 6.
75
Results
Metabolic respiration
The night respiration rate at 20 °C was 0.60--0.70 µmol m −2
s −1, except in Clone 2693 (Table 1). More variability among
sources was observed at 24 °C than at 20 °C. Accordingly, the
response of the metabolic respiration rate to leaf temperature
varied among sources and was strongest in Clone 2662 and
weakest in Clone 2693. The metabolic respiration rate approximately doubled for every 10 °C increase in leaf temperature in
Clones 2660, 2662 and 2687, but increased only 1.17-fold in
Clone 2693 and about 1.75-fold in the half-sib Family 2431.
Effects of photon flux density and ambient CO2
concentration on assimilation rate
A typical response of assimilation rate of mature (2- to 3-weekold) leaves of E. poeppigiana to changes in photon flux density
is presented in Figure 2 for Clone 2660. Most of the observed
variation was caused by depletion of the ambient CO2 concentration in the closed-loop system during the measurements, but
a minor part of the variation was due to variations in leaf
temperature among measurements. Differences in assimilation
rates between leaves were small (Figure 2). Maximum assimilation rate was about 20 µmol m −2 s −1 in all sources studied.
The response curves to ambient CO2 concentration showed
little variation (Figure 2). The curves at different photon flux
densities had the same form, but differed in maximum assimilation rate. The curves measured at photon flux densities of
1560 and 2060 µmol m −2 s −1 had similar maximum rates,
about 40 µmol m −2 s −1. The observed maximum was slightly
above 40 µmol m −2 s −1 in all sources, indicating a strong
response to the elevated CO2 concentration. The apparent ‘‘lack
of fit’’ of the estimated assimilation rate at 900 µmol m −2 s −1
was due to differences between leaves; the curve was fitted to
all observations on Clone 2687, not to a specific leaf.
No apparent systematic deviation was observed in the plots
of model residuals against photon flux density, ambient CO2
concentration and leaf temperature, except for a slight underestimation at low ambient CO2 concentration in Clone 2662
(Figure 3). The range of residuals was large, but they were
systematically distributed on both sides of the zero (perfect fit)
line. The same was also true for the plot of the residuals against
leaf-to-air water vapor pressure difference (Figure 3), indicating that the exclusion of this environmental factor from the
Table 1. Average metabolic respiration rate ± SD at leaf temperatures of 20 and 24 °C in mature leaves of Erythrina poeppigiana, calculated from
the night respiration measurements. Means followed by the same letter within a column are not significantly different (Duncan’s multiple range
test at 5%).
Tree source
n
rm(20)
(µmol m 2 s −1)
n
rm(24)
(µmol m −2 s −1)
Rate of change
(per °C)
2431
2660
2662
2687
2693
Combined
18
9
18
21
15
81
0.68 ± 0.11 b
0.72 ± 0.24 b
0.64 ± 0.17 b
0.61 ± 0.14 b
0.89 ± 0.07 a
0.69 ± 0.17
44
44
33
49
45
251
0.85 ± 0.18 bc
0.97 ± 0.23 a
0.87 ± 0.40 abc
0.80 ± 0.18 c
0.95 ± 0.18 ab
0.89 ± 0.18
0.056
0.074
0.077
0.068
0.016
0.064
76
NYGREN
photosynthetic rate, analysis of residual variance of the complete assimilation rate models (Equation 8) fitted for each
source and for the combined data set was carried out. The
results indicated highly significant differences (F12,1034 =
14.05, P < 0.001).
Model estimates of the assimilation rates at a photon flux
density of 2000 µmol m −2 s −1, a leaf temperature of 28 °C and
an ambient CO2 concentration of 350 µmol mol −1 were 17.0,
18.9, 18.1, 19.9 and 16.8 µmol m −2 s −1 in the half-sib Family
2431 and Clones 2660, 2662, 2687 and 2693, respectively. At
1000 µmol mol −1, the respective assimilation rates were estimated to be 40.1, 43.5, 40.6, 44.7 and 41.5 µmol m −2 s −1.
Effect of leaf development on relative photosynthetic capacity
Figure 2. Top: Assimilation rate as a function of photon flux density
in mature leaves of Clone 2660 of Erythrina poeppigiana. Symbols
represent observations on different leaves (n = 6) and the curve
represents the estimate by Equation 8 at ambient CO2 concentration
(300 µmol mol − 1) and a leaf temperature of 28 °C. Bottom: Assimilation rate as a function of ambient CO2 concentration at photon flux
densities of 310 (j), 900 (h), 1560 (d) and 2060 (s) µmol m −2 s − 1
in mature leaves of Clone 2687. Symbols represent observations on
individual leaves and the curves represent the average fit of Equation 8
for the clone.
assimilation rate model was justified. The modeling efficiency
varied between 0.88 and 0.96 (Table 2).
The quantum yield for incident radiation, kq, was similar in
the half-sib Family 2431 and Clones 2660 and 2687, but was
significantly lower (significance determined according to the
overlap of 95% confidence intervals) in Clone 2662 and higher
in Clone 2693 (Table 2). Slightly more variation was observed
in the value of κc, which combines the conductance of the CO2
pathway from the air to the chloroplasts and the carboxylation
efficiency. It was highest in Clones 2660, 2662 and 2687. The
95% confidence intervals for the quantum yield and κc estimates were relatively narrow, indicating a reliable fit of these
parameters.
The value of the photorespiration parameter, αr, was highest
in Clone 2662, followed by Clone 2687, and was lowest in
Clone 2693; however, only the difference between Clones
2662 and 2693 was significant (Table 2). The confidence
intervals of αr were wider than those of kq and κc, indicating
that αr parameter estimation was less accurate than the parameter estimation of photosynthetic rate. Because significant
differences were observed in the values of the parameters of
Leaf age satisfactorily explained changes in the relative photosynthetic capacity of the leaves. The modeling efficiency varied between 0.81 and 0.92 (Table 3), and the residuals plotted
against leaf age, although scattered, did not show any systematic deviation (Figure 4). The slope of the model was significantly steeper in the half-sib Family 2431 and Clone 2693 than
in Clones 2660 and 2662. The 95% confidence interval of the
slope parameter βπ of Clone 2660 included zero, indicating
that the parameter was not important for the model. The confidence intervals of the values of the parameters γπ and kp,
which measure initial rate of change of the relative photosynthetic capacity, overlapped between all sources. According to
the analysis of residual variance, the differences between the
relative photosynthetic capacity models fitted by source were
significant (F16,829 = 13.27, P < 0.001).
Equation 10 did not hold for senescent leaves from about
6 days before shedding, but the relative photosynthetic capacity declined with a steep slope (Figure 5). This senescence
slope was determined analytically as a straight line, which
intercepted the curve of Equation 10 at a leaf age corresponding to 6 days before the average shedding age, and was zero at
the average shedding age of each source (Table 4).
The average leaf life span was short and varied among
sources; leaves of the half-sib Family 2431 and Clone 2662
had a life span almost 2 weeks longer than leaves of Clones
2660 and 2687 (Table 4). The premature loss of leaves as a
result of herbivory or pathogen damage was lowest in Clone
2662 at 8.3%, and highest in Clone 2687 at 50.0%. Herbivory
or pathogen loss was 28.3, 35.0 and 48.3% for sources 2660,
2431 and 2693, respectively.
Three patterns of leaf development were detected (Figure 5):
long life span with a considerable decrease in photosynthetic
capacity as a function of leaf age (half-sib Family 2431), short
life span with almost constant photosynthetic capacity (Clone
2660) and short life span with a considerable decrease in
photosynthetic capacity (Clone 2687). Development of leaves
in Clones 2662 and 2693 followed intermediate patterns (Tables 3 and 4): long life span with a moderate decrease in
photosynthetic capacity (Clone 2662) and intermediate life
span with a considerable decrease in photosynthetic capacity
(Clone 2693).
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
77
Figure 3. Residuals of the assimilation rate model for mature leaves of Erythrina poeppigiana (Equation 8) as a function of photon flux density,
ambient CO2 concentration, leaf temperature and leaf-to-air water vapor pressure difference.
Table 2. Parameter values with 95% confidence interval for the rectangular hyperbolae describing the dependence of assimilation rate on ambient
CO2 concentration and incident photon flux density in mature leaves of Erythrina poeppigiana; Equation 8 adjusted to data on the effects of photon
flux density and CO2 concentration.
Tree source
kq
κc
αr
SSE
2431
0.055
0.049--0.061
0.056
0.052--0.060
0.046
0.043--0.048
0.054
0.049--0.058
0.067
0.059--0.075
0.067
0.062--0.072
0.076
0.072--0.079
0.079
0.076--0.083
0.081
0.076--0.086
0.063
0.059--0.066
455
323--586
449
360--538
677
582--771
541
418--663
391
296--485
1438
209
0.88
555
190
0.96
568
202
0.96
1565
224
0.91
969
209
0.95
0.053
0.051--0.056
0.073
0.071--0.075
497
445--549
5926
1046
0.92
2660
2662
2687
2693
Combined
1
df
EF1
Abbreviations: kq = quantum yield (mol CO2 assimilated/mol photons incident on leaf), κc = product of total conductance of the CO2 pathway
from air to chloroplast and carboxylation efficiency (mol m − 2 s−1), αr = parameter of photorespiration rate, SSE = sum of squared errors, df =
degrees of freedom for error, and EF = modeling efficiency.
Foliole nitrogen concentration and leaf mass/leaf area ratio
The foliole nitrogen concentration was high and declined with
leaf age (Figure 6). However, even senescent leaves had a
nitrogen concentration of 23.0--26.4 mg g −1, which was about
half the value observed in vigorous 28- to 56-day-old leaves.
The leaf mass/leaf area ratio was low in young leaves (38--46
g m −2) and increased with leaf age stabilizing at 60--70 g m −2
(Figure 7).
Equation 11 provided a satisfactory description of the rela-
78
NYGREN
Table 3. Parameter values with 95% confidence interval for the logistic s-curves describing the relationship between leaf age and relative
photosynthetic capacity in Erythrina poeppigiana; Equation 10.
Tree source
απ
βπ
γπ
kπ
SSE
df
EF1
2431
1.26
1.20--1.32
1.10
1.03--1.17
1.05
1.01--1.10
1.12
1.06--1.18
1.21
1.15--1.28
0.0067
0.0057--0.0077
0.0021
−0.00065--0.0048
0.0039
0.0026--0.0052
0.0064
0.0043--0.0086
0.0072
0.0056--0.0088
4.78
3.77--5.79
6.25
4.67--7.83
5.71
4.63--6.78
4.10
3.30--4.90
5.66
4.58--6.74
0.53
0.41--0.65
0.81
0.60--1.03
0.79
0.63--0.94
0.68
0.53--0.83
0.66
0.52--0.80
1860
246
0.81
1170
153
0.88
708
162
0.92
1024
140
0.85
510
128
0.90
1.16
1.13--1.18
0.0059
0.0053--0.0065
5.12
4.57--5.67
0.67
0.59--0.74
6622
845
0.84
2660
2662
2687
2693
Combined
1
Abbreviations: απ, βπ = parameters of the maximum photosynthetic capacity (asymptote), kπ = initial rate of change of the relative photosynthetic
capacity (day − 1), γπ = parameter, SSE = sum of squared errors, df = degrees of freedom for error, and EF = modeling efficiency.
Figure 4. Residuals of the relative photosynthetic capacity model for
Erythrina poeppigiana (Equation 10) as a function of leaf age.
Figure 5. Relative photosynthetic capacity as a function of leaf age in
the half-sib Family 2431 and Clones 2660 and 2687 of Erythrina
poeppigiana. Calculated according to Equation 10.
Table 4. Average leaf shedding age ± SD and the equations describing
the ‘‘senescence slope’’ of relative photosynthetic capacity as a function of leaf age in Erythrina poeppigiana.
Tree source
Leaf shedding
age (days)
Senescence slope
equation
Application
range (days)
2431
2660
2662
2687
2693
87 ± 16
74 ± 10
86 ± 15
73 ± 9
80 ± 17
π = 10.4 − 0.120la1
π = 11.8 − 0.160la
π = 10.6 − 0.123la
π = 8.41 − 0.115la
π = 9.03 − 0.113la
82--87
69--74
81--86
68--73
75--80
Combined
81 ± 15
π = 9.57 − 0.120la
76--81
1
Abbreviations: π = relative photosynthetic capacity (unitless), and la
= leaf age (days).
tionship between foliole nitrogen concentration and relative
photosynthetic capacity; the modeling efficiency was 0.98-0.99 in all sources (Table 5). The value of parameter απ (maximum photosynthetic capacity) was fixed beforehand to the
value estimated from the fit of Equation 10 to the data on the
effect of leaf age on relative photosynthetic capacity (Table 3),
Figure 6. Foliole nitrogen concentration as a function of leaf age in
four clones and a half-sib family (2431) of Erythrina poeppigiana.
because there were too few foliole nitrogen concentration data
to estimate the three parameter values reliably. According to
the analysis of residual variance, the models fitted by source
differed significantly (F8,11 = 6.72, P < 0.001).
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
Figure 7. Leaf mass/leaf area ratio as a function of leaf age in four
clones and a half-sib family (2431) of Erythrina poeppigiana.
Stomatal conductance
Stomatal conductance for CO2 varied between 158 and 238
mmol m −2 s −1 at photon flux densities below 1000 µmol m −2
s −1 and between 190 and 366 mmol m −2 s −1 at higher photon
flux densities (Table 6). A typical response of assimilation rate
to stomatal conductance is presented in Figure 8. Stomatal
conductance was strongly affected by photon flux density and
the dependence of assimilation rate on stomatal conductance
followed the same pattern as its dependence on photon flux
density (cf. Figure 2).
Aboveground biomass production
Aboveground biomass production between December 12,
1991 and June 12, 1992 was highest in Clone 2660, followed
by Clone 2662, and was lowest in Clone 2693 (Table 7). The
proportion of harvestable biomass (foliage, green twigs and
woody branches) was highest in sources with high total aboveground biomass production (76 and 70% in Clones 2662 and
2660, respectively). The within-source variation in biomass
production was lowest in Clone 2660 and highest in the halfsib Family 2431 and Clone 2662. Litterfall consisted of foliage
79
litter only and presented the highest within-source variation of
the biomass compartments, probably because its measurement
was less accurate than that of the other compartments.
Correlations were calculated between the parameters of the
CO2 exchange model (Tables 1--3) and the production of total
aboveground and harvestable biomass. The only significant
correlations were found between both total and harvestable
biomass production and the parameter βπ (r = − 0.99, P =
0.0019 and r = −0.99, P = 0.0003, respectively), which measures the rate of decline of the relative photosynthetic capacity
as a function of leaf age. This indicates that sources that
maintain a high photosynthetic capacity during the entire leaf
life span produce more biomass than sources that only maintain a high photosynthetic capacity for part of the leaf life span.
Also the leaf area on June 12, 1992 was correlated with βπ
(r = − 0.88, P = 0.0494). Weaker correlations were detected
between rate of change of metabolic respiration rate, kr, and the
total aboveground (r = 0.73, P = 0.1667) and harvestable
biomass production (r = 0.73, P = 0.1596) and leaf area on
June 12, 1992 (r = 0.76, P = 0.1327).
Discussion
Field data were used to determine the parameter values for an
idealized model describing the effect of environmental factors
and leaf development on leaf CO2 exchange in E. poeppigiana.
The assimilation model was divided into four components, the
photosynthetic rate in mature leaves, relative photosynthetic
capacity, metabolic respiration and photorespiration. The photosynthetic rate in mature leaves was described as a function
of the photon flux density and ambient CO2 concentration, and
the relative photosynthetic capacity as a function of leaf age. It
was introduced into the model as the coefficient of the basic
photosynthesis model described by Equation 4. This is consistent with the observations of Long et al. (1993) that the quantum yields of 11 taxa were not affected by the age of the
photosynthetic organs, but that the maximum assimilation
Table 5. Parameter values with 95% confidence interval for the asymptotic curves describing the relationship between foliole nitrogen
concentration and relative photosynthetic capacity in Erythrina poeppigiana; Equation 11.
Tree source
απ
kn
nmin
SSE
2431
1.26
2660
1.10
2662
1.05
2687
1.12
2693
1.21
0.085
0.064--0.106
0.109
0.057--0.160
0.091
0.054--0.129
0.072
0.026--0.118
0.059
0.039--0.078
26.5
25.0--28.0
23.3
22.0--24.6
22.9
20.9--25.0
26.2
21.6--30.8
26.3
23.3--29.3
Combined
1.16
0.073
0.061--0.084
24.6
23.5--25.8
1
df
EF1
0.0090
3
0.99
0.0028
2
0.99
0.0038
2
0.99
0.0142
2
0.98
0.0051
2
0.99
0.2057
19
0.99
Abbreviations: απ = maximum photosynthetic capacity (asymptote), kn = rate of change of relative photosynthetic capacity (g mg −1), nmin =
minimum foliole nitrogen concentration (mg g −1), SSE = sum of squared errors, df = degrees of freedom for error, and EF = modeling efficiency.
80
NYGREN
Table 6. Average stomatal conductance for CO2 ± SD (mmol m − 2 s − 1)
in mature leaves of Erythrina poeppigiana.
Tree source
2431
2660
2662
2687
2693
Photon flux density
≤ 1000
> 1000
183 ± 122
238 ± 125
158 ± 66
164 ± 83
200 ± 136
300 ± 104
326 ± 122
190 ± 70
366 ± 126
225 ± 94
Figure 8. Assimilation rate as a function of stomatal conductance for
CO2 in mature leaves of Clone 2687 of Erythrina poeppigiana at
different photon flux densities.
rates tended to be lower in young and old organs than in mature
photosynthetic organs. A temperature dependence of quantum
yield (Farquhar and von Caemmerer 1982) was not observed;
it may have been obscured by the strong correlation between
photon flux density and leaf temperature.
Von Caemmerer and Farquhar (1981) observed a decrease in
both the maximum assimilation rate and carboxylation efficiency in Phaseolus vulgaris grown at low nitrogen availability and concluded that nitrogen is needed for Rubisco
regeneration. In the present study, the observed decrease in
photosynthetic capacity as a function of leaf age may have
been caused by the simultaneous decrease in foliole nitrogen
concentration. This possibility is supported by the close rela-
tionship between foliole nitrogen concentration and relative
photosynthetic capacity, and suggests that foliole nitrogen
concentration could be used instead of leaf age to model
relative photosynthetic capacity. The relationship between foliole nitrogen concentration and relative photosynthetic capacity was curvilinear as has been observed in Gmelina arborea
Roxb. leaves (Cromer et al. 1993).
A constant ratio was assumed between ambient and chloroplast CO2 concentration. The conventional method of estimating internal leaf CO2 concentration by applying stomatal
conductance or resistance estimated from transpiration data
(von Caemmerer and Farquhar 1981, Landsberg 1986, Thornley and Johnson 1990, Long and Hällgren 1993) was avoided
for three reasons. First, a constant ratio between ambient and
intercellular CO2 concentration has been observed in leaves of
well-watered C3 plants (reviewed by Long 1985). Second,
according to the optimization theory (Cowan 1977), gas exchange is optimal when the maximal amount of carbohydrates
is produced per unit of water transpired. In non-water-limited
leaves this implies that stomatal opening is determined by the
capacity of mesophyll tissue to fix carbon (Wong et al. 1979),
or that stomatal conductance is affected by the same factors as
assimilation rate (cf. Wong et al. 1985). Third, the determination of stomatal conductance is based on a series of measurements, the accuracy of which is often poor under humid
tropical conditions. For example, an error of 1 °C in leaf
temperature between 30 and 31 °C means a 0.25 kPa error in
the estimate of water vapor pressure in the substomatal cavity.
Further, a 2% error in the humidity sensor output causes a 15%
error in conductance computed according to von Caemmerer
and Farquhar (1981), when the air humidity is about 80%.
However, the error increases to 70% at an air humidity of 90%
(Field and Mooney 1984); 80--90% was a common humidity
range at the experimental site.
Daytime respiration was divided into metabolic respiration,
which was modeled as a function of leaf temperature, and
photorespiration, which was modeled as a function of ambient
CO2 concentration. The metabolic respiration rate, which was
determined from the night respiration rate measurements and
was assumed to continue unaltered during the daytime (cf.
Edwards and Walker 1983), approximately doubled with a
10 °C increase in leaf temperature. At typical leaf temperature
Table 7. Average aboveground biomass production ± SD (kg per tree) of Erythrina poeppigiana from December 12, 1991 to June 12, 1992. Means
followed by the same letter within a row are not significantly different (Duncan’s multiple range test at 5%).
Biomass compartment
Tree source
2431
2660
2662
2687
2693
Foliage
Green twigs
Woody branches
Total harvestable
1.48 ± 0.52 b
0.74 ± 0.39 b
0.90 ± 0.82 c
3.12 ± 1.70 c
2.44 ± 0.25 a
1.55 ± 0.24 a
2.21 ± 0.25 a
6.19 ± 0.58 a
2.54 ± 0.74 a
0.99 ± 0.53 b
1.63 ± 0.81 ab
5.16 ± 2.00 ab
1.75 ± 0.44 b
0.73 ± 0.25 b
1.13 ± 0.40 bc
3.61 ± 1.00 bc
1.47 ± 0.42 b
0.60 ± 0.19 b
0.72 ± 0.35 c
2.79 ± 0.96 c
Litterfall
Stem increment
Total aboveground
0.66 ± 0.42 a
1.28 ± 1.02 a
5.06 ± 2.91 b
1.34 ± 0.80 a
1.37 ± 0.42 a
8.90 ± 0.92 a
0.93 ± 0.43 a
0.75 ± 0.41 ab
6.86 ± 2.66 ab
1.21 ± 0.35 a
0.72 ± 0.38 ab
5.55 ± 1.28 b
1.19 ± 0.56 a
0.38 ± 0.30 b
4.36 ± 1.35 b
Leaf area (m2)
26.3 ± 9.2 bc
43.9 ± 4.4 a
32.5 ± 9.4 bc
34.0 ± 8.4 b
22.8 ± 6.5 c
ERYTHRINA CO2 EXCHANGE IN FIELD CONDITIONS
values at the experimental site, metabolic respiration rates
were 1--2 µmol m −2 s −1.
At ambient CO2 concentration, the photorespiration rate was
comparable to the metabolic respiration rate. In the model, it
was assumed that, at a constant atmospheric CO2/O2 concentration ratio, the photorespiration rate was proportional to the
photosynthetic rate, and that the photorespiration term in the
model only measures the deviation of photorespiration rate
from photosynthetic rate when the CO2/O2 concentration ratio
is altered as a result of variations in atmospheric CO2 concentration. Inclusion of the photorespiration term in the model was
valid because a model with only a temperature-dependent
respiration term overestimated the assimilation rate at low
ambient CO2 concentrations. At high atmospheric CO2 concentration, however, the photorespiration term approached
zero and could be omitted from the model.
Both the solubility of CO2 in mesophyll cell sap relative to
O2 solubility and the specificity of Rubisco for CO2 relative to
O2 change in favor of O2 as a function of increasing leaf
temperature, which should result in an increased photorespiration rate relative to the photosynthetic rate (Long 1991). However, such dependence of photorespiration rate on temperature
was not observed in this study. Attempts to model the photorespiration term by applying the changes in CO2/O2 solubility ratio and CO2/O2 specificity factor suggested by Long
(1991), or by a simple ‘‘black box’’ type dependence of photorespiration rate on leaf temperature and ambient CO2 concentration, resulted in a poorer fit of the assimilation rate
model to the data than was achieved with Equation 8.
The failure to detect any dependence of the photorespiration
term in Equation 8 on temperature may be caused by several
interacting factors. Both the solubility ratio and Rubisco specificity factor are less affected by temperature at the high temperatures typical of tropical environments. According to the
formulae presented by Long (1991), the average rate of change
of the internal CO2/O2 concentration ratio is 1% per 1 °C
between 25 and 35 °C. Most of the assimilation rate measurements were carried out within this temperature range. The
81
average rate of change of the Rubisco specificity factor is
somewhat higher, 4% per 1 °C within the same temperature
range. These changes are small enough to be obscured by the
high correlation between photon flux density and leaf temperature inside the assimilation chamber. They are also small
compared to the doubling of the internal CO2/O2 concentration
ratio, calculated according to Long (1991), in favor of CO2
when ambient CO2 concentration increases from 350 to 700
µmol mol −1.
When the parameters of the leaf CO2 exchange model were
compared with the aboveground biomass production, only the
parameter measuring changes in photosynthetic capacity as a
function of leaf age, βπ, showed a significant correlation; the
photosynthetic capacity of the most productive tree sources did
not decrease much during the leaf life span. In the idealized
assimilation model, βπ was the only parameter that was indicative of the plant’s ability to integrate photosynthetic production
over time. The correlation between βπ and biomass production
emphasizes the importance of following leaf CO2 exchange
characteristics for a long time period, if the aim is to relate
these characteristics to biomass production. Comparisons between leaf assimilation and biomass production have usually
been made based on the instantaneous assimilation rate measured at light-saturated conditions, and the results vary from
positive correlation in 10 tropical tree species (Muthuchelian
1992) to no correlation in provenances of Acacia mangium
Willd. (Atipanumpai 1989) and negative correlation in half-sib
families of Robinia pseudoacacia L. (Mebrahtu et al. 1991).
The maximum assimilation rate of E. poeppigiana is among
the highest observed in tropical woody legumes (Table 8), and
is comparable with the values observed in Prosopis chilensis
(Mol.) Stuntz. (Pinto 1989). The high photosynthetic capacity
of E. poeppigiana may be a result of high foliole nitrogen
concentrations (Table 8 and cf. Mooney et al. 1984).
The quantum yield of E. poeppigiana was comparable with
the quantum yield observed in the neotropical rain forest
Caesalpinoids, Copaifera venezuelana Pittier and Harms
(0.065) and Hymenaea courbaril L. (0.056), in laboratory
Table 8. Maximum assimilation rate and nitrogen concentration in mature leaves of some tropical leguminous trees.
Species
Maximum assimilation rate
(µmol m −2 s −1)
Leaf nitrogen concentration
(mg g −1)
Erythrina