Short‐term variation in leaf‐level water use efficiency in a tropical forest

Abstract

The representation of stomatal regulation of transpiration and CO2 assimilation is key to forecasting terrestrial ecosystem responses to global change. Given its importance in determining the relationship between forest productivity and climate, accurate and mechanistic model representation of the relationship between stomatal conductance (gs) and assimilation is crucial.
We assess possible physiological and mechanistic controls on the estimation of the g1 (stomatal slope, inversely proportional to water use efficiency) and g0 (stomatal intercept) parameters, using diurnal gas exchange surveys and leaf‐level response curves of six tropical broadleaf evergreen tree species.
g1 estimated from ex situ response curves averaged 50% less than g1 estimated from survey data. While g0 and g1 varied between leaves of different phenological stages, the trend was not consistent among species. We identified a diurnal trend associated with g1 and g0 that significantly improved model projections of diurnal trends in transpiration.
The accuracy of modeled gs can be improved by accounting for variation in stomatal behavior across diurnal periods, and between measurement approaches, rather than focusing on phenological variation in stomatal behavior. Additional investigation into the primary mechanisms responsible for diurnal variation in g1 will be required to account for this phenomenon in land‐surface models.

Full text

Short-term variation in leaf-level water use efficiency in a
tropical forest
Kenneth J. Davidson
1,2
, Julien Lamour
1
, Alistair Rogers
1
, Kim S. Ely
1
, Qianyu Li
1
,
Nate G. McDowell
3,4
, Alexandria L. Pivovaroff
5
, Brett T. Wolfe
6,7
, S. Joseph Wright
7
,
Alfonso Zambrano
7
and Shawn P. Serbin
1
1
Department of Environmental and Climate Sciences, Brookhaven National Laboratory, Building 490A, Upton, NY 11973, USA;
2
Department of Ecology and Evolution, Stony Brook
University, 650 Life Sciences Building, Stony Brook, NY 11794, USA;
3
Atmospheric Sciences and Global Change Division, Pacific Northwest National Laboratory, PO Box 999, Richland,
WA 99352, USA;
4
School of Biological Sciences, Washington State University, PO Box 644236, Pullman, WA 99164-4236, USA;
5
Biology Division, Glendale Community College, 1500
N Verdugo Rd, Glendale, CA 91208, USA;
6
School of Renewable Natural Resources, Louisiana State University, Room 227, Renewable Natural Resources Bldg, Baton Rouge, LA 70803,
USA;
7
Smithsonian Tropical Research Institute, Apartado, 0843–03092 Balboa, Panama
Author for correspondence:
Kenneth J. Davidson
Email: kdavidson.sci@gmail.com
Received: 28 September 2022
Accepted: 13 December 2022
New Phytologist (2023) 237: 2069–2087
doi: 10.1111/nph.18684
Key words: climate model, diurnal response,
leaf phenology, sap flux, stomatal
conductance, stomatal optimization,
transpiration, water use efficiency (WUE).
Summary
The representation of stomatal regulation of transpiration and CO
2
assimilation is key to
forecasting terrestrial ecosystem responses to global change. Given its importance in deter-
mining the relationship between forest productivity and climate, accurate and mechanistic
model representation of the relationship between stomatal conductance (g
s
) and assimilation
is crucial.
We assess possible physiological and mechanistic controls on the estimation of the g
1
(stom-
atal slope, inversely proportional to water use efficiency) and g
0
(stomatal intercept) parame-
ters, using diurnal gas exchange surveys and leaf-level response curves of six tropical
broadleaf evergreen tree species.
g
1
estimated from ex situ response curves averaged 50% less than g
1
estimated from survey
data. While g
0
and g
1
varied between leaves of different phenological stages, the trend was
not consistent among species. We identified a diurnal trend associated with g
1
and g
0
that sig-
nificantly improved model projections of diurnal trends in transpiration.
The accuracy of modeled g
s
can be improved by accounting for variation in stomatal behav-
ior across diurnal periods, and between measurement approaches, rather than focusing on
phenological variation in stomatal behavior. Additional investigation into the primary mecha-
nisms responsible for diurnal variation in g
1
will be required to account for this phenomenon
in land-surface models.
Introduction
Properly parameterizing Earth system models (ESMs) is of vital
importance to forecasting terrestrial ecosystems’ responses to
novel future climates (Rogers et al., 2017). In particular, the
ESM representation of the terrestrial biosphere in land-surface
models (LSMs) used to predict the dynamics, responses, and
feedbacks of vegetation to the climate system is a critical require-
ment for climate modeling (Bonan, 2008). This includes the rep-
resentation of photosynthetic carbon assimilation (A), which
directly determines ecosystem net carbon uptake (Ainsworth &
Rogers, 2007). Transpiration (E), which is a dominant flux in the
hydrologic cycle (Schlesinger & Jasechko, 2014), is fundamen-
tally governed by stomatal conductance (g
s
), which regulates both
Aand E(Meidner & Mansfield, 1968; Hetherington & Wood-
ward, 2003; Lawson & Vialet-Chabrand, 2019). Despite the
large influence of g
s
on global carbon and water cycles,
parametrization of g
s
in models remains one of the largest
uncertainties in current predictions of net ecosystem productivity
(Bauerle et al., 2014; Dietze et al., 2014; Jefferson et al., 2017;
Franks et al., 2018; Ricciuto et al., 2018).
The LSMs that are commonly used in ESMs tend to represent
g
s
using an assumed linear relationship between g
s
and Awhere
the slope parameter governs the ratio of g
s
:A(Damour
et al., 2010). The majority of these formulations are phenomeno-
logical (e.g. Wong et al., 1979; Farquhar & Wong, 1984; Ball
et al., 1987; Leuning, 1995; Damour et al., 2010); however,
Medlyn et al.(2011) argued that the relationship could be mod-
eled in terms of stomatal optimality theory. The principal of
stomatal optimization is that the evolution of vascular plants has
prioritized maximizing A, per unit of water lost via E(Cowan &
Fraquhar, 1977; Cowan, 1978). Stated in terms of water, stom-
atal regulation should minimize g
s
so that Eis minimized for a
given amount of A, at least during periods of water limitation.
This is mathematically represented by ∂E/∂A=λ, where 1/λrep-
resents the whole plant water use efficiency (WUE). This
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087 2069
www.newphytologist.com
Research

relationship is conceptualized in the ‘Unified Stomatal Optimiza-
tion model’ (USO; Medlyn et al., 2011; Eqn 1).
gs¼g0þ1:61þg1
ffiffiffiffi
D
p

A
Cs
:Eqn 1
Unlike phenomenological models, the g
1
term in the USO
model relates directly to optimality theory (Cowan & Fraqu-
har, 1977; Cowan, 1978). The stomatal slope term (g
1
)is
approximately inversely proportional to WUE. The stomatal
intercept (g
0
) is the expected g
s
when net assimilation (A
net
)=0
and occurs at an irradiance level equivalent to the light compen-
sation point. The g
0
parameter was an empirical addition and
intended to help constrain the model and ensure predicted g
s
is
always positive, rather than a feature predicted by optimality
(Duursma et al., 2019). The other terms in the USO model relate
to environmental conditions, including the leaf-to-air vapor pres-
sure deficit (D) and the CO
2
concentration at the leaf surface
(C
s
). Previously, Wu et al.(2019b) tested four different stomatal
model formulations in the wet tropics and found that the USO
model was best suited to represent g
s
in this system.
Data used to fit the USO model can be collected either by tak-
ing survey measurements at ambient conditions (Bernacchi
et al., 2006; Lin et al., 2015;Wuet al., 2019b) or by the use of a
response curve method (Ball et al., 1987; Leuning, 1995; Leakey
et al., 2006; Domingues et al., 2014; Wolz et al., 2017; Davidson
et al., 2022; Lamour et al., 2022b). In the survey method, in situ
measurements of instantaneous gas exchange from many individ-
ual leaves are collectively used to fit a regression from which g
1
and g
0
are obtained. In the curve method, slow irradiance, vapor
pressure deficit (VPD), or CO
2
response curves are used to illicit
stomatal response, usually on a cut branch segment. For this
approach, it is typical to calculate g
0
and g
1
for an individual leaf.
Much of the work to date on assessing variation of stomatal
parameters has focused on specific biotic or abiotic drivers.
Research into the effects of drought, soil moisture (H´
eroult
et al., 2013), and temperature (Lin et al., 2015; Wolz
et al., 2017) has shown that, in some systems, the abiotic control
on g
1
may be stronger than any biotic controls (Zhou
et al., 2014,2016). Other work has linked leaf phenology and
photosynthetic properties in tropical evergreen forests, where the
rate of leaf production and leaf longevity both influence leaf func-
tion (van Schaik et al., 1993). Leaves with longer average lifes-
pans have reduced maximum carboxylation rate of the enzyme
Rubisco (V
cmax
) as compared to shorter lived leaves (Kitajima
et al., 1997,2002). Similarly, Wu et al.(2016) found that V
cmax
changed over the lifetime of tropical leaves, and this leaf-age vari-
ation is an important factor to consider when modeling tropical
forest carbon assimilation (Wu et al., 2017). Yet, it remains to be
seen what impact leaf phenology has on stomatal parameters in
tropical species. A strong phenological partitioning of leaf-level
stomatal response, if it occurs, would most likely be an important
development for ESMs given the large impact phenology has on
terrestrial carbon dynamics (De Weirdt et al., 2012). This is espe-
cially true for evergreen tropical forests, where leaves are found
across a particularly wide range of phenological stages and where
phenological changes drive a large variation in seasonal carbon
uptake (Wu et al., 2016,2017).
Other research has focused on the temporal consistency of
stomatal parameters (Resco de Dios et al., 2020), or the condi-
tions in which the g
s
response should be optimized (Katul
et al., 2010). Plants may optimize their stomatal operation across
a variety of time scales, from near instantaneous response to stim-
uli such as sunflecks (Way & Pearcy, 2012; Campany
et al., 2016), diurnal responses such as those to drying soil
(Knapp, 1993; Zhang et al., 2018), and long-term responses such
as those which optimize water and nutrient uptake by roots
(Franklin, 2007). Thus, whole plant WUE is the integration of
all these factors, with leaf-level WUE lying at the end of a hierar-
chy of responses (Katul et al., 2010). Wu et al.(2019b) addressed
the seasonal consistency of tropical WUE by analyzing dynamics
of g
1
during a drought event and found that species-specific g
1
did not significantly vary by month, suggesting that at the whole
canopy-scale WUE is stable through time. However, the diurnal
consistency of leaf-level WUE in this system remains untested. If
they occur, diurnal adjustments in WUE may be in response to
rapidly changing external stimuli such as changes in VPD during
the day (Katul et al., 2009), or may be the consequence of shift-
ing water supply and carbon demand dynamics within the leaf
(Bonan et al., 2014; Dewar et al., 2018). Understanding the
degree to which WUE varies diurnally, and the main mechanisms
which regulate these diurnal changes, may be helpful in develop-
ing more representative and robust process-based models of g
s
and Ein tropical broadleaf evergreen forests.
In this study, we measured leaf gas exchange on six evergreen
tropical tree species to identify key biotic and abiotic controls on
stomatal parameters and WUE. We had four main objectives: (1)
compare survey and ex situ response curve-derived g
0
and g
1
to
assess whether these two commonly used approaches provide
comparable estimates of the key parameters; (2) determine
whether g
0
and g
1
vary with leaf phenological stage; (3) assess
whether g
0
and g
1
vary over a diurnal period; and (4) conduct sim-
ulations within a simplified canopy-scale model to understand the
impacts of parameterization on E, comparing our estimated flux
to a model benchmark derived from sap flux measurements. By
addressing these four objectives, our goal was to develop a more
detailed process-based understanding of stomatal function in wet
tropical forests and identify several key axes of biotic and abiotic
variation, which could be targeted for further investigation.
Materials and Methods
Site and species description
Data collection occurred from 6 January to 23 March 2020, in the
San Lorenzo Protected Forest (9°16ʹ1.71ʺN, 79°58ʹ28.27ʺW), on
the Col´
on Province on the Republic of Panama’s Caribbean coast
(Fig. 1). The site is characterized as a wet lowland (130 m asl)
tropical forest, with a mean annual temperature of 23.4°C and
an average of 3330 mm of rainfall (Paton, 2020). We used a
crane operated by the Smithsonian Tropical Research Institute
(STRI) for canopy access. Six tree species were selected for study,
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
Research
New
Phytologist
2070

chosen to span early, mid, and late successional growth strategies
(Table 1). For each species, we sampled from two trees that were
exposed to full sun throughout the day. Four of the trees had pre-
viously been instrumented for measurement of sap flux.
Gas exchange and leaf trait data collection
Gas exchange measurements were carried out using five LI-
6400XT portable photosynthesis systems using a 6400-02(B)
LED Light Source and an LI-6800 Portable Photosynthesis Sys-
tem using a Multiphase Flash Fluorometer leaf chamber (6800-
01A; Li-Cor Biosciences, Lincoln NE, USA). For all measure-
ments, the color spectrum of the irradiance was set to a fixed ratio
of 90% red : 10% blue.
We used two different methods of gas exchange: an ex situ
response curve approach and an in situ survey approach. Stomatal
response curves on cut branch segments were carried out as
described previously (Leakey et al., 2006; Davidson et al., 2022;
Lamour et al., 2022b). Upper canopy, fully sun-exposed branches
were cut predawn and immediately recut underwater to ease
xylem tension and prevent embolism (Sperry, 2013). We recut a
total of at least 50 cm back from the first cut to remove open ves-
sels (Wolf et al., 2016) with final branch lengths of c. 1.5 m. Cut
branches were transferred in buckets to a shaded (c. 150 μmol
PAR) location where all response curves were carried out. Each
species was measured on 10–15 different days and spread across
the duration of the field campaign. For each curve, irradiance level
was reduced from saturating down to 0 μmol m
−2
s
−1
photosyn-
thetic photon flux density (PPFD) in a stepwise fashion every 20–
40 min (Fig. 2a; Lamour et al., 2022b). The irradiance value at
saturation was derived for each species using photosynthetic
response curves collected at multiple locations from across the top
of each canopy (Lamour et al., 2021b; Rogers et al., 2022).
Survey measurements followed previous approaches (Rogers
et al., 2004;Wuet al., 2019b). Fully sunlight top-of-canopy
leaves were measured with the LI-6800. Before each
Fig. 1 Geographic location of the crane site (b) and the San Lorenzo Protected Forest (c) within the Isthmus of Panama, and (a) monthly climate averages
collected at the site between 1997 and 2019. The period between January and April is considered the dry season, with less than 20% of annual rainfall
(black bars). During the period of study, mean temperature was 25.5°C and mean daily precipitation was 1.3 mm.
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
New
Phytologist Research 2071

Table 1 Values for species-specific photosynthetic and conductance parameters and canopy scaling factors used in simulations of leaf-level Aand g
s
.
Species
V
cmax.25
(μmol m
−2
s
−1
)
J
max.25
(μmol m
−2
s
−1
)
R
dark.25
(μmol m
−2
s
−1
)g
1
(kPa
0.5
)
g
0
(mol m
−2
s
−1
)
DBH
(cm)
LAI
(m
2
m
−2
)
Crown
Area (m
2
)
Sapwood
area (m
2
)
Successional
status
Leaf
lifespan (d)
Brosimum utile (Kunth)
Oken
Young: 68.84
Mature: 51.28
Old: 35.22
84.61
60.44
49.17
1.36
1.17
0.89
Response
Curve: 2.40
Survey: 3.04
0.013
0.029
–– – – Late 324
Cecropia insignis Liebm. Young: 77.48
Mature: 78.70
Old: 60.50
109.48
108.48
89.33
1.50
1.81
1.68
Response
Curve: 1.75
Survey: 2.87
0.031
0.107
–– – – Early 231
Guatteria dumetorum R.
E. Fr.
Young: 38.66
Mature: 48.18
Old: 35.79
65.71
67.53
64.71
1.24
0.82
0.97
Response
Curve: 1.96
Survey: 3.32
0.028
0.067
59.0 1.5 155.7 0.210 Mid 212
Miconia borealis (Bonpl.)
DC.
Young: 64.59
Mature: 71.60
Old: 54.06
106.29
108.38
94.55
1.81
0.92
0.79
Response
Curve: 2.20
Survey: 4.15
0.034
0.003
34.0 1.15 74.42 0.080 Mid 189
Terminalia amazonia (J.F.
Gmel.) Exell
Young: 70.28
Mature: 77.85
Old: 51.34
106.07
116.16
87.52
1.42
1.14
1.24
Response
Curve: 0.78
Survey: 2.83
0.022
0.032
52.9 1.5 134.57 0.174 Mid 160
Vochysia ferruginea Mart. Young: 46.57
Mature: 57.46
Old: 43.66
76.14
78.47
78.62
1.67
1.18
0.78
Response
Curve: 0.67
Survey: 2.57
0.036
0.040
58.0 2.0 153.24 0.204 Mid 215
For details regarding parameter estimates and tree functional traits, see section ‘Canopy Esimulations’. Data on leaf lifespan were previously reported in Osnas et al.(2018).
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
Research
New
Phytologist
2072

measurement, the conditions within the leaf chamber were set to
match ambient air temperature (22.8–32.6°C), CO
2
concentra-
tion (c. 400 ppm), and irradiance (3–2400 μmol PAR). The rela-
tive humidity in the leaf chamber was maintained at 70% for all
measurements, a level which allowed for VPD
air
to be close to
mean daytime conditions (Paton, 2020; Fig. S1), while minimiz-
ing condensation risk in the instrument. Each survey measure-
ment consisted of a 30–90 s stabilization period, followed by five
measurements of Aand g
s
at 5-s intervals, which were then aver-
aged to produce a single data point.
Surveys were carried out from 06:30 h (approximate sunrise)
to 18:30 h (approximate sunset) across eight different days. Two
survey schemes were used, one in which we measured as many
leaves as possible at three positions on a single tree crown for
40 min every hour for 4 h, and a second where each of the 12
tree crowns (two per species and three positions per tree) was
measured as many times as possible for 12 min, replicating this
pattern five times in a 12-h period. The aim of the first scheme
was to assess the diurnal dynamics of a single species on a given
day, while the aim of the second scheme was to broadly survey
the forest. We used the data from the first scheme to parameterize
our models (see Notes S1) and data from the second scheme,
along with Wu et al.(2019b), to validate our models.
Immediately following the single species gas exchange survey
measurements, the leaf was cut at the base of the petiole and
stored in a humid, cool, dark box for further processing (Leach
et al., 1982; Rodriguez-Dominguez et al., 2022). Processing
included using a Scholander-type pressure chamber to measure
leaf water potential (Ψ
leaf
; Scholander et al., 1964). Following
measurement of Ψ
leaf
, samples were dried to constant mass and
weighed to determine leaf mass area (LMA; dry leaf mass per unit
leaf area). Measured leaves were classified into one of three phe-
nological stages, young, mature, or old (Table 1;Fig.3).
Reflectance spectroscopy
We used a full-range spectroradiometer (PSR 3500+; Spectral
Evolution Inc., Lawrence, MA, USA) together with an LC-RP-
Pro leaf clip foreoptic (Spectra Vista Corp., Poughkeepsie, NY,
USA) containing an internal, full-spectrum calibrated light
source to measure leaf reflectance on all leaves following comple-
tion of stomatal response curves. The reflectance data were used
to estimate V
cmax.25
and J
max.25
using the spectral methods and
model presented in Lamour et al.(2021c).
Sap flux measurement
Sap flux velocity (cm s
−1
) data were collected for four of the 12
study trees. Sap flux velocity was measured with Granier-type
sensors (Model PS-TDP8; PlantSensors, Nakara, Australia),
which use heat dissipation to empirically infer sap flux velocity
(Granier, 1987). Each tree was instrumented with two sensors,
Fig. 2 Schematic representation of the response curve method. (a) Raw data from a response curve, with each black point representing an observation of
stomatal conductance (g
s
). At each irradiance level (horizontal grey bars), g
s
drops then stabilizes, at which point irradiance is reduced to the next level
(vertical grey bars). Across all irradiance levels, leaf temperature and vapor pressure at the leaf surface were held constant and close to ambient conditions.
The CO
2
inside the leaf chamber was maintained at 400 μmol mol
−1
. (b) a typical regression fit of the points extracted from the seven irradiance levels,
and the stomatal slope (g
1
) and stomatal intercept (g
0
) estimates produced from this curve. In this example, the nonlinearity of the response curve data,
explored in detail by Lamour et al.(2022b), is readily apparent. For data analysis, many curves are combined to obtain one overall regression, with each
curve acting as a random effect on the overall g
1
and g
0
estimates.
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
New
Phytologist Research 2073

located at breast height, and positioned perpendicular to each
other at a depth of 2 cm into the sapwood of the trees. Raw tem-
perature data were processed in the R package AQUAFLUX
(v.1.0.0, Speckman, 2019) to derive sap flux velocity. Vapor pres-
sure deficit data obtained on site from the STRI meteorological
station were used to normalize minimum flux. Sap flux velocity
data were multiplied by sapwood area (Meinzer et al., 2001)to
derive sap flow (g s
−1
) at 15-min intervals, with the assumption
that velocity was consistent across the sapwood radial profile. In
total, there were 42 diurnal (05:30–18:30 h) periods of complete
sap flux coverage for each of the four study trees during our per-
iod of study (Pivovaroff et al., 2020).
Estimation of conductance parameters
Stomatal response was evaluated using the USO model (Medlyn
et al., 2011; Fig. 2b; Eqn 1; Notes S1). g
1
represents the slope in
the relationship shown in Eqn 2, a reconfiguration of Eqn 1
where g
1
directly acts as the slope between g
s
and A, atmospheric
CO
2
concentration (C
a
), and VPD. The parameters g
1
and g
0
are
mathematically identical in Eqns 1,2(Lamour et al., 2022b);
however, Eqn 2provides for simpler graphical depiction of the
data. Eqn 2also uses C
a
in place of C
s
, as when using gas exchange
systems, the terms are very similar (Lamour et al., 2022c).
gs1:6A
Ca¼g0þg11:6A
ffiffiffiffi
D
pCa

:Eqn 2
As with any linear model, a limited range of measured g
s
can
affect the model fit by introducing uncertainty in the regression
(g
1
). The wider the range of the data, the lower the uncertainty in
the fit of the model slope, assuming that the variation around the
mean slope is consistent (e.g. no heteroscedasticity). For response
curve data, we captured a full range of conductance values, as we
are theoretically measuring g
s
from its maximum (light saturated)
value to its minimum (dark-adapted) value. A limited range of g
s
can also lead to uncertainty in estimating a value for g
0
, especially
if most measurements occurred far above the light compensation
point. To keep analysis consistent between datasets, we chose to
fit g
0
for the survey dataset rather than prescribe a value
(Duursma et al., 2019).
R
dark
was calculated from dark-adapted measurements of A
net
(the final response curve point) and was scaled to 25°C(R
dark.25
)
using an inverse Arrhenius equation (Eqn 3; Bernacchi
et al., 2001; Von Caemmerer, 2013):
Rdark:25 ¼Rdark
exp Ha
RT25
ðÞ
Ha
RTleaf
ðÞ

:Eqn 3
where H
a
is the activation energy of mitochondrial respiration
(H
a
=46 390 J mol
−1
), and Ris the ideal gas constant
(R=8.314 J mol
−1
K
−1
).
Canopy Esimulations
As a method of evaluating the performance of different stomatal
parameterizations against a benchmark data set, we modeled
canopy-scale Eresponse to ambient environmental conditions.
We chose not to evaluate the model performance in simulating
canopy scale A, as we have no benchmark which to compare
model estimates. We used the R package LEAFGASEXCHANGE
(v.1.0.1, Lamour & Serbin, 2021), which includes a coupled leaf
scale steady state assimilation, conductance, energy balance, and
radiation interception model. Details of model parameters and
equations used can be found in Notes S2.
The model simulates environmental conditions around the leaf
using local meteorological data (Paton, 2020). To simulate the
radiation environment at the leaf surface, the model uses the radi-
ation interception model developed by Norman (1979), which
partitions incoming radiation into direct and diffuse streams. In
each simulation, we assumed six canopy layers with a uniform
distribution of leaf area index (LAI) between layers (B´
eland &
Baldocchi, 2021). To simulate wind speed within the canopy we
assumed an exponential decrease with LAI within the canopy fol-
lowing the model presented in equation 14 in Buckley
et al.(2014). Values for V
cmax.25
,J
max.25
,R
dark.25
,g
0
, and g
1
, esti-
mated as described previously, were averaged by species and by
phenological stage (Table 1). In each model run, uncertainty was
simulated by rerunning the model using g
1
and g
0
values corre-
sponding to 1 SE of the fitted value.
Leaf-level estimates of Ewere scaled to the canopy by first
summing across canopy layers and then multiplying by total
Fig. 3 Examples of the three phenological
stages for four of the six species in this study.
Young leaves were characterized as lighter in
color, with minimal damage, softer more
flexible texture, and a distal position on the
branch. Mature leaves comprised the
majority of the leaf material, characterized as
fully expanded, having an intermediate color
and texture as well as an intermediate
position. Old leaves were very scarce for
most species and tended to have minor
surface damage, were dark in color, and
were in extreme proximal positions on
branches.
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
Research
New
Phytologist
2074

canopy area as estimated using the allometric function presented
in Martı´nez Cano et al.(2019). Leaf area index estimates were
made for each species (Condit et al., 2013; Sirri et al., 2019),
which allow for an appropriate accounting of leaf area per ground
area.
We then conducted a model simulation sensitivity analysis to
explore the potential impacts of biotic variation on emergent,
canopy scale E. To do this, we simulated different canopies with
different combinations of leaf phenological classes, with leaf
physiological properties based on leaf phenological stage. For this
analysis, we assumed ‘natural’ conditions were a 15-80-5% parti-
tioning between young, mature, and old leaves, an estimate that
is in line with Wu et al.(2017). We compared our simulation
results for canopy scale Ewith measured sap flow as a benchmark.
Here, we expected that there was a delay between flux measured
at the base of the tree and Ein the canopy, so we averaged flux
across a 15-min interval to minimize this discrepancy (Meinzer
et al., 2003).
Statistical analysis
To compare g
1
and g
0
estimates between tree species and pheno-
logical stages within tree species, we used mixed effects models
constructed and analyzed in the R package NLME (v.3.1, Pinheiro
et al., 2020). For response curve data, the leaf served as the ran-
dom effect, while for survey data, the location within the canopy
of the tree served as the random effect. For all data, we considered
the species and phenological stage as fixed effects. The equations
and coefficients for the mixed models can be found in Notes S1
and Tables S1–S3.
To examine which of our three statistical models (response
curve, survey, survey +time-of-day; Notes S1) predicts leaf-level
g
s
most accurately, we performed a validation of the fixed effects
components of the models using two additional survey-derived
datasets, one which was collected on four separate days during
our 2020 field campaign (described above as scheme 2), and a
second data set collected by Wu et al.(2019b) at the same site in
2016 (Rogers et al., 2022). We then used validation statistics (r
2
,
RMSE, likelihood ratio test) to determine which model predicted
g
s
most accurately.
We assessed the effect of species and species : leaf phenological
stage on V
cmax.25
,R
dark.25
, and the effect of LMA, Ψ
leaf
, and suc-
cessional stage on g
1
and g
0
using ANOVA and post hoc Tukey
tests. Finally, we compared canopy level transpiration simulations
using ANOVA at the species level, and a post hoc Tukey’s honest
significant difference test to distinguish significance between fac-
tor levels. All analysis was performed using the R open-source
software environment (v.3.6.2, R Core Team, 2013).
Results
Gas exchange dynamics of diurnal surveys and response
curves
Here, we present the results from an analysis of 80 stomatal
response curves (Fig. S2a) and 665 survey data points (Fig. 4;
Lamour et al., 2021b). Both Aand g
s
respond strongly to the
diurnal cycle of irradiance during the survey measurements
(Fig. 4). VPD
leaf
also exhibits a diurnal trend, peaking around
solar noon, due to higher leaf temperatures, and a small increase
in VPD
air
(Fig. S1). Stomatal response curves can be used to
examine potential maximum photosynthetic and stomatal fluxes.
Here, we see that among species, there is significant variation in
maximum g
s
, with Brosimum utile,Cecropia insignis, and M. bore-
alis having maximum g
s
values up to three times that of the other
three species (Fig. S2b).
We can compare the relationship between g
s
and environmen-
tal conditions for both datasets (Fig. S2). We see that not only do
the survey data have a larger overall range of conductance, there
is also more within-species variation in conductance for a given
set of environmental conditions. We can also see the nonlinearity
of the response curve data, explored in detail by Lamour
et al.(2022b), which is much less pronounced for the survey
data set.
Model comparison and validation
We evaluated the accuracy of the models built on the different
datasets (response curve, survey, and survey +time-of-day), first
assessing their calibration scores, and then by predicting g
s
mea-
sured by the survey method at the same site in 2016 (Rogers
et al., 2022) and using our 2020 validation dataset. In the calibra-
tion phase, the survey and survey +time-of-day models can be
compared directly as they are derived using the same data set.
The model including a time-of-day effect on g
1
and g
0
has an
improved calibration r
2
(0.80 vs 0.75), a lower RMSE (0.0564 vs
0.0627 mol m
−2
s
−1
), a lower AIC score (−1844 vs −1741), and
a likelihood ratio test reveals the models were significantly differ-
ent (P<0.001). However, during the validation phase, the
model calibrated on the survey dataset without a time-of-day
effect performed the best, with the lowest RMSE, and the highest
validation r
2
(Fig. 5).
g
1
values estimated using the response curve dataset were sig-
nificantly (P<0.001) lower than when they were estimated
using the survey data set (Fig. 6). On average, the response curves
produced estimates of g
1
which were 50% lower than estimates
obtained from survey data. Only in the species B. utile, there was
no significant difference in estimated g
1
(Fig. 6).
Finally, we simulated canopy scale Efor all six species to
compare how the different methods of deriving stomatal param-
eters perform in a modeling context at the larger whole-tree
scale. Four of the six species also had scaled sap flux velocity
measurements (Pivovaroff et al., 2020) to compare the modeled
Eallowing us to benchmark results (Fig. 7). We found that
when comparing estimated and observed E, the survey +time-
of-day parameterization had the lowest RMSE (0.013 g H
2
O
m
−2
s
−1
) followed by survey (0.014 g H
2
Om
−2
s
−1
), and the
response curve method (0.022 g H
2
Om
−2
s
−1
;Fig.S3). All
three methods closely tracked changes in irradiance. However,
the two survey methods captured the magnitude of the change
in canopy Emore closely than the response curve method
(Fig. 7).
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
New
Phytologist Research 2075

Fig. 4 Overview of leaf-level conditions observed during diurnal survey measurements. Species-specific averages of hourly (a) photosynthesis (A), (b)
stomatal conductance (g
s
), (c) vapor pressure deficit at the leaf surface (VPD
leaf
), and (d) photosynthetic photon flux density (PPFD). Error bars represent
2 SE of the mean.
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
Research
New
Phytologist
2076

Effect of leaf phenological stage on estimated stomatal
parameters
For the response curve data set, both g
1
and g
0
varied by tree spe-
cies (P<0.001) and phenological stage (P=0.009) with a sig-
nificant stage : species interaction (P<0.001, Figs 8a,S4;
Table 2). For the survey data set, g
0
varied by tree species
(P<0.001), phenological stage (P<0.001) with a significant
stage : species interaction (P<0.001, Fig. S4; Table 2), but g
1
only varied among tree species (P<0.001; Fig. 8b; Table 2).
When assessing the differences between the interactive effects
of phenological stage and species on g
1
for response curves, we
only found four significant effects: B. utile young and mature
(P=0.003, 66.9% decrease), M. borealis young and mature
(P=0.006, 61.2% decrease), M. borealis young and old
(P=0.041, 49.7% decrease), and C. insignis young and old
(P=0.007, 111% increase). We also used response curve data to
investigate continuous relationships between g
0
and g
1
with other
related leaf physiological traits, but did not observe a significant
effect of LMA, Ψ
leaf
, or successional stage on g
1
or g
0
(Figs S5,
S6).
Finally, we analyzed the interactive effects of phenological
stage and species on V
cmax.25
and R
dark.25
, derived from leaf spec-
troscopy. The relationships between stage and R
dark.25
(P<0.001) and stage and V
cmax.25
(P<0.001) were significant,
c. 40% decrease in values between the mature and old phenologi-
cal stages (Fig. S7; Table 1).
Investigation of a time-of-day effect
Using the survey dataset (Fig. 4), we tested whether the time-of-
day influenced g
0
and g
1
. We found a significant decline in g
1
from morning to evening for the six species (P<0.05, Table 2).
g
1
was on average 2.5 times higher in the early morning than in
the late afternoon; however, the magnitude of the decline differed
between species (Fig. 9). Unlike g
1
,g
0
did not differ by the time-
of-day for the six species (P=0.432).
Canopy level transpiration simulations
To assess the impact of data collection methodology on model
performance, we compared estimates of mean daytime Efor the
six species, three methods, and sap flux benchmark (when avail-
able, Fig. 10). We assumed a constant flux of Efor each 15-min
interval and averaged Efor each 12-h diurnal period. For five of
the six species, the survey data produce significantly higher
(P<0.001) estimates of Ethan for the response curve data (B.
utile =26.0%, C. insignis =87.1%, Guatteria dumeto-
rum =124.9%, Terminalia amazonia =107.4%, and Vochysia
ferruginea =44.3% increase). In M. borealis, there was no signifi-
cant difference between Eas estimated from curve-derived values,
and Eas estimated from survey-derived values (P=0.152), even
though in M. borealis the g
1
value estimated from response curves
was nearly half that of the g
1
value estimated from survey mea-
surements (Fig. 6). However, g
0
values for M. borealis were espe-
cially low with no significant difference between curve and
survey-based estimates (Fig. S4), and M. borealis has a saturating
irradiance, which was 50% higher than the other five species in
this study Rogers et al.(2022).
We also found no significant differences between the survey
approaches for any of the six species. In all four species for which
we have a benchmark, the benchmark and survey estimates
showed strong correspondence with estimates being statistically
indistinguishable (P>0.05).
Finally, we examined the impact of leaf phenological stage dis-
tribution assumptions on the modeling of canopy E(Fig. S8).
Fig. 5 Validation plots of observed values of leaf-level stomatal conduc-
tance (g
s
) vs model predicted g
s
. Data come from (a) the model based on
response curve-derived stomatal intercept (g
0
) and slope (g
1
), (b) the
model based on survey-derived g
0
and g
1
, and (c) the survey model with
an inclusion of a time-of-day effect on g
1
.r
2
and root mean squared error
(RMSE) values in each panel show that the survey model improves predic-
tions of g
s
over the response curve model; however, the inclusion of the
time-of-day effect slightly worsens model predictive accuracy. Open circles
were from the 2016 dataset collected by Wu et al.(2019b), while closed
circles correspond to the 2020 validation data. Solid black, grey, and blue
trend lines correspond to the lines of best fit for the 2016 data, 2020 data,
and all the data, respectively. Dashed black line is a 1 : 1 reference line.
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
New
Phytologist Research 2077

For this analysis, we reran the model prescribing different distri-
butions of leaf phenology, including even (33.3% young, 33.3%
mature, and 33.3% old), all-young, all-mature, and all-old. Inter-
estingly, we found no significant differences in predicted daily
canopy transpiration between the even phenology distribution,
the ‘natural’ (15% young, 80% mature, 5% old) phenology dis-
tribution, or the all-mature phenology distribution for any spe-
cies when using any of the three models of stomatal parameters.
However, when modeled using a single phenological stage, we
see strong variation in estimates of daily E(Fig. S8) with daily
estimates of Evarying by up to 150% in some species.
Discussion
A key step toward reducing ESM uncertainty is to account for
the abiotic and biotic effects on the stomatal parameters g
0
and
g
1
. In this study, we demonstrate that ex situ stomatal response
curves and in situ survey style measurements produce statistically
different estimations of stomatal parameters in tropical broadleaf
evergreen forests (Fig. 6). These differences in parameterization
resulted in large (between 26% and 125%) differences in simu-
lated fluxes of water (Fig. 10). Furthermore, we found that leaf
phenology plays a role in regulating stomatal traits; however, this
role was not consistent across species (Figs 8,S4), was not
observed consistently using the different measurement
approaches (Fig. 6), and, for a given methodology, did not drive
variation in modeled E(Fig. S8). We also observed a consistent
decline in g
1
during the photoperiod (Fig. 9), with g
1
declining
by an average of 250%. When these results are taken together
and used to model forest function, we found that models using
stomatal parameters derived from ex situ response curves signifi-
cantly underestimated canopy level Eboth at instantaneous time
scales (Fig. S3) and across a full dry season (Fig. 7), and that
while leaf traits do vary among leaf phenological stage, models,
which only include mature vegetation, perform similarly to those
that explicitly simulate three leaf stages (Fig. S8).
Derivation of stomatal model parameters is highly sensitive
to the method of data collection
One of the clearest observations from this study is the significant
difference in fitted values of g
1
between response curves and sur-
vey style measurements. In five of the six species studied, g
1
was
significantly lower (50% reduction) for response curves than for
survey measurements (Fig. 6). We explore several possible mecha-
nistic explanations for the observed difference in g
1
estimates.
All response curves were measured on excised branch seg-
ments. While previous work on tropical tree species has shown
that excision does not impair photosynthetic function (Verryckt
et al., 2020), stomata will be affected more acutely by hydraulic
damage than chloroplasts if branches are not allowed to suffi-
ciently rehydrate before measurement (Wolf et al., 2016; Lawson
& Vialet-Chabrand, 2019). This decline in g
s
, without a com-
mensurate decline in A(Davidson et al., 2022), may result in a
decrease in estimated g
1
. An excision effect was observed previ-
ously in plants collected from the same site as this study, where
levels of Aand g
s
diminished from pre-excision values within
60 min of branch removal (Santiago & Mulkey, 2003). Here,
the researchers attributed excision-induced embolism as the likely
mechanism underlaying their result. Missik et al.(2021) found
that immediately following excision, levels of g
s
were reduced by
up to 62% in woody deciduous trees, again attributing this
decline to embolism. However, in both studies, branches were
cut during the daytime when transpiration fluxes are higher, and
the risk of embolism is much greater (Sperry, 2013). Further-
more, Santiago & Mulkey (2003) limited their experiment to just
Fig. 6 Fitted values of stomatal slope (g
1
)by
species from the response and survey
methods. t-Tests indicate that for all species
other than Brosimum utile g
1
from response
curves is significantly lower than g
1
estimated
from the survey data (P<0.001). Error bars
represent 2 SE of the mean.
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
Research
New
Phytologist
2078

60 minutes, which may not be sufficient time for branches to
rehydrate and acclimate to a new leaf-level environment. Other
studies (e.g. Meng & Arp, 1993; Miyazawa et al., 2011) have
found mixed, and sometimes contrasting effects of excision on
gas exchange measurements.
A more relevant comparison may be drawn between this study
and Davidson et al.(2022) who, in a comparison of cut and
uncut branch segments, demonstrated that in a woody deciduous
species when excision is carried out predawn, stomatal behavior is
not significantly altered in leaves measured within 8 h of exci-
sion. We observed no significant difference in g
0
or g
1
estimated
from curves conducted in the morning (07:00–12:00 h) and the
afternoon (12:01–18:00 h) local time (Fig. S9) suggesting that
excision stress did not alter stomatal function over the course of
measurements. In addition, longer duration response curves were
correlated with higher estimates of g
1
(Fig. S10), again suggesting
that a stress response is not accumulating during measurement.
It is also possible that the direction of the response curve (in-
creasing irradiance vs decreasing irradiance), combined with an
asymmetrical response of stomata to long-term changes in irradi-
ance (Lawson & Blatt, 2014), may have led to the negative bias
observed for g
1
. To test this, we collected response curves in both
directions on a similar set of tropical species (Fig. S11); however,
we did not observe any evidence for a negative bias associated
with curves in which irradiance is reduced.
There could also be error associated with the survey measure-
ments. Accurate survey assessment of the A:g
s
relationship relies
upon the assumption that levels of Aand g
s
are at a steady state
(H¨
oltt¨
aet al., 2017), and are acclimated to the environment sur-
rounding the leaf, however errors when selecting an appropriate
level of PPFD, VPD
leaf
,orC
a
will cause alteration of the leaf-
level environment. Taking irradiance as an example; if an incor-
rect level of irradiance is chosen, the level of assimilation will
change rapidly in response to the new irradiance; however, the
stomatal response will lag, which will alter the estimate of g
1
. Sto-
mata are known to react more slowly to increases in irradiance
than decreases in irradiance (Lawson & Vialet-Chabrand, 2019).
This means that if, on average, we misestimated irradiance in the
same proportion (equal under and over-estimation), we would
expect a slightly lower WUE (higher g
1
) for survey data, which is
consistent with our findings (Fig. 6). This issue of improperly
estimated irradiance is not insignificant, as irradiance can fluctu-
ate dramatically, particularly under cloudy conditions (Fig. 7),
which may lead to errors matching the natural environment to
the leaf chamber conditions.
A final factor to consider is the leaf-to-leaf variability implicit
in each measurement type, with survey measurements consisting
of roughly 10 times the number of individual leaves as response
curves (Fig. S2). Survey measurements showed a slightly poorer
performance and a significantly higher variance, in an internal
model calibration than curve-derived measurements (Fig. S12;
Tables S1,S2). This may be due to the larger leaf-to-leaf variabil-
ity captured using the survey method, or it could be due to unsta-
ble ambient conditions at the time of measurement which can
create noise in survey data (Figs S2b,S13c,d). When the two
models were validated using an external dataset comprised of c.
Fig. 7 Simulated canopy-scale transpiration (E,gH
2
Os
−1
)at15 mintime
points. Data presented are for (a) Guatteria dumetorum,(b)Miconia bore-
alis,(c)Terminalia amazonia, and (d) Vochysia ferruginea observed on 20
February 2020. Black points and line represent the sapflux benchmark,
green points and line represent response curve-derived simulation, gold
points and line represent survey-derived simulation, and blue points and line
represent survey-derived simulation with stomatal slope (g
1
)changingbased
on time-of-day. Shading around the simulated data represents error propa-
gated into the simulation by using g
1
1 SE of the estimated value. Grey
dashed line represents the solar profile (PPFD), vertical grey lines show the
transition between g
1
values for the survey time-of-day model, and horizon-
tal red line shows the level of saturating irradiance for each species.
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
New
Phytologist Research 2079

120 new individual leaf estimates per species, both models had
similar residual variation, while the response curve model exhib-
ited a clear bias, underestimating levels of g
s
(Fig. 5).
Clearly, both methods have benefits and drawbacks, and nei-
ther is free from error. Survey style measurements can integrate
the diurnal dynamics of stomatal response, which may be an
appealing feature for those who model stomatal response to a
range of multivariate environmental conditions. Response curves
capture stomatal response at truly stabilized conditions, which,
while stability of environment rarely occurs in nature, is useful
for developing generalizable parameter estimates, which can be
used in ESMs.
The effect of leaf phenology and traits on g
0
and g
1
Previous work documenting phenological changes in leaf proper-
ties in wet evergreen tropical forest found a strong effect of leaf
phenology on V
cmax
(Kitajima et al., 1997;Wuet al., 2017). This
change in V
cmax
leads to a change in potential A, and an assumed
commensurate change in the potential g
s
, driven by changes in
leaf-level metabolism (Tobin & Rogers, 1992;Wingleret al., 2004).
However, if the relationship between Aand g
s
is not coordinated
across leaf development, g
1
will vary between leaf phenological
stages. Theoretically, a phenology-related increase in g
1
,as
observed in C. insignis (Fig. 8a), may be adaptative as a means of
maximizing the photosynthetic efficiency of older vegetation. If
V
cmax
declines in older vegetation, as we observe in this study
(Fig. S7), we may also expect an increase in g
1
as a higher level of
intercellular carbon (C
i
) is required for the same rate of A. How-
ever, in this study, we did not find evidence across all species for
a consistent effect of leaf phenological stage or relative leaf age on
g
1
or g
0
(Fig. 8). Here, we explore two possible explanations for
the lack of an effect observed in our study.
In the USO model, the relationship between Aand g
s
is the
chief control on the g
1
parameter. Coordination between Aand
g
s
means that for a given level of A, there is a given level of g
s
and
thus a given level of C
i
. Conversely, if Aand g
s
lack coordination,
at a given level of A,g
s
, and thus C
i
, can vary. Therefore, tight
coordination between Aand g
s
will result in leaves maintaining a
fixed ratio of C
i
:C
a
across changes in photosynthetic capacity
Fig. 8 Fitted values of stomatal slope (g
1
) estimated using data from the two measurement methods. Estimates are from (a) response curve data, and (b)
survey data. Analysis of Variance and post hoc Tukey tests indicates that data in (a) are significant to the species and phenological stage levels, while data
in (b) were only significant at the species level. Error bars represent 2 SE of the mean.
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
Research
New
Phytologist
2080

(Wong et al., 1979,1985; McDowell et al., 2006). We did not
find significant differences in C
i
:C
a
across any of the phenologi-
cal stages across species (Fig. S14), suggesting that although V
cmax
may shift with leaf phenology (Fig. S7a), g
s
will change in pro-
portion to supply and maintain a stable level of intercellular CO
2
(Long & H¨
allgren, 1993). This coordination in V
cmax
and g
smax
(g
s
at saturating irradiance) is indicative of stable WUE or g
1
across phenological stages (Fig. 8). However, this does not sug-
gest a lack of optimization of carbon capture through ontogeny.
Even though the ratio of Aand g
s
does not change with leaf phe-
nology, V
cmax
changes with leaf phenology (Fig. S7a).
Time-of-day effect on g
1
While g
1
is often thought to be fixed over daily to seasonal time-
scales (Medlyn et al., 2011), others suggest that at the leaf level,
g
1
may vary over shorter, diurnal time scales (Katul et al., 2010).
Daytime adjustments in g
1
may be driven by rapidly changing
external stimuli such as irradiance (Way & Pearcy, 2012; Cam-
pany et al., 2016), or VPD and C
a
(Katul et al., 2009), to which
individual leaves may alter their stomatal response in order to
maximize assimilation, or minimize water loss. Supply and
demand dynamics may also effect stomatal behavior, with a
Table 2 Fixed effects components from the mixed effects models tested.
Model Effect Numerator df Denominator df F-value P-value
Response curve g
0
~species 6 607 94.30 <0.0001
g
0
~stage 2 607 10.30 <0.001
g
0
~stage : species 10 607 3.13 <0.001
g
1
~species 6 607 110.67 <0.001
g
1
~stage 2 607 4.73 0.009
g
1
~stage : species 10 607 5.31 <0.001
Survey g
0
~species 6 629 630.08 <0.001
g
0
~stage 2 629 25.61 <0.001
g
0
~stage : species 10 629 9.10 <0.001
g
1
~species 6 629 118.21 <0.001
g
1
~stage 2 629 0.08 0.922
g
1
~stage : species 10 629 0.50 0.889
Survey +time-of-day g
0
~species 6 623 764.05 <0.001
g
0
~stage 2 623 31.06 <0.001
g
0
~stage : species 10 623 11.04 <0.001
g
1
~species 6 623 143.35 <0.001
g
1
~time 3 623 124.29 <0.001
g
1
~time : species 18 623 8.11 <0.001
A full description of each model, and the random effects components are provided in Notes S1.
Fig. 9 Trends in fitted values of stomatal
slope (g
1
) by time-of-day and species. g
1
decreases significantly with increasing time-
of-day for all six species investigated.
Brosimum utile,Cecropia insignis,Guatteria
dumetorum, and Vochysia ferruginea, all
exhibit a very strong and consistent decline in
g
1
, whereas Miconia borealis and Terminalia
amazonia exhibit a more moderate decline.
Error bars represent 2 SE of the mean.
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
New
Phytologist Research 2081

reduction in the water supply to the substomatal cavity (Bonan
et al., 2014), leading to a more conservative response. Here, we
observed declining g
1
over the diurnal periods for all six species
(Fig. 9).
A decreased supply of water to the leaf late in the day (Ven-
turas et al., 2017) may be the most likely explanation for the
observed decrease in g
1
. In this case, lower g
s
at the same level of
stomatal index (a term representing Aat a combination of leaf-
level environmental conditions) would result in a decrease in the
fitted value of g
1
(Fig. S15). These findings are consistent with
other studies of stomatal response (Manzoni et al., 2011), which
suggest that WUE responds to short-term changes in the supply
of water. Longer term adjustments in WUE in response to sea-
sonal or yearly drought do not appear to be occurring at this site
(Wu et al., 2019b).
The finding that g
1
declines throughout the day is also consistent
with models of stomatal optimality, which apply a penalty on
stomatal opening, either in the form of vascular damage (Wolf
et al., 2016; Sperry et al., 2017;Anderegget al., 2018)orinnon-
stomatal limitations on A(H¨
oltt¨
aet al., 2017; Dewar et al., 2018;
Kennedy et al., 2019). Many of these models apply a function
where stomata reduce their aperture in response to Ψ
leaf
(Klein,
2014). To investigate whether Ψ
leaf
can help explain the diurnal
dynamics of g
1
, we implemented the optimality model described in
Anderegg et al.(2018). This model applies a reduction in estimated
g
s
, which is proportional to the assumed decline in stem xylem con-
ductivity with increasing Ψ
leaf
(see also Notes S3;TableS4). How-
ever, we found that the Ψ
leaf
penalty from Anderegg et al.(2018)
poorly explained observed patterns of g
s
(Fig. S16). Examination of
stomatal model residuals vs measured Ψ
leaf
(Fig. S6) demonstrates
that Ψ
leaf
, as an explanatory factor alone, is not capable of explain-
ing the variation in g
1
observed, consistent with previous observa-
tions (Wu et al., 2019b). As an alternative to Ψ
leaf
based models,
many ESMs use a βfactor, dictated by soil moisture, to downregu-
late V
cmax
and g
s
. Recently, Li et al.(2022) tested the applicability
of a βfactor at the same site in this study and found that, like
Anderegg et al.(2018), it overpenalized drought-related loss of pro-
ductivity. Several other hydraulic parameters such as hydraulic
capacitance and leaf hydraulic conductance may be more pertinent
and may play a larger role in stomatal regulation; however, we lack
those data in this current study.
The low morning WUE observed across all species may be the
result of circadian resonance (Green et al., 2002), whereby sun-
rise anticipation linked to a plant’s endogenous rhythm may
prime stomata to achieve their largest opening early in the day.
This behavior, in systems with a fixed photoperiod, is thought to
be adaptive as a means of maximizing carbon assimilation (Resco
de Dios et al., 2016). Higher than optimal morning g
s
would lead
to elevated estimates of g
1
, which is consistent with previous find-
ings that suggest circadian regulation fails to help to optimize
diurnal variations in stomatal conductance (Resco de Dios
et al., 2020).
The failure of a time-of-day effect to significantly improve pre-
dictions of g
s
(Fig. 5) suggests that while an overall pattern of
declining g
1
is expected, the magnitude of the decline each day is
the by-product of abiotic processes that we may not have fully
captured in this study, such as dynamics in the soil–plant-
atmosphere continuum (Zhang et al., 2018), or an effect of ABA
on regulating stomata (McAdam & Brodribb, 2015). Without
these data, and a more detailed understanding of the response of
stomata to small changes in VPD (Buckley, 2016), it may not be
possible to model diurnal fluctuations in WUE. Thus, we recom-
mend future research explore diurnal shifts in g
1
in greater detail,
as an understanding of the mechanisms behind these dynamics
has the potential to significantly advance our model representa-
tion of canopy E.
Fig. 10 Comparison of estimated mean total
daily canopy transpiration (E,kgH
2
Om
−2
ground area) using the three models
(response curve, survey, and survey +time-
of-day). Sapflux benchmark values are
presented for four of the six species in this
study. Mean values are the average
integrated total of Eover the 42 diurnal
(05:30–18:30 h) periods of complete sapflux
coverage. Error bars represent 2 SE of the
mean.
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
Research
New
Phytologist
2082

Impact of measurement method on models of canopy
function
Our simulations of canopy Erevealed that using parameters
derived from response curves leads to a systematically underesti-
mated (26–125%) diurnal flux (Figs 7, 10), consistent with our
external model validation (Figs 5,S3). This underestimation is in
part attributable to lower g
1
values from response curves (Fig. 6)
which limit daytime E, due to the assumption that at an irradi-
ance level greater than saturating, g
s
=g
smax
.
All three models underestimate the level of Eat low irradiance
as the simulations assume that any time irradiance is below the
light compensation point, g
s
=g
0
. However, this assumption may
be flawed due to the combination of cuticular conductance and
stomatal leakage (Machado et al., 2021), measurement artifacts
(Duursma et al., 2019), or unaccounted for transpiration from
stored crown water supplies (Meinzer et al., 2003). The inte-
grated diurnal flux of water (Fig. 10) also reveals a large impact
of parameterization, with the survey method producing species
level values of Ebetween 9% and 77% higher than model output
using response curve-derived parameters, underscoring the
impact that g
1
can have on model estimation of ecosystem func-
tion (Dietze et al., 2014) and reinforcing the need for careful
choices with regard to the method used to estimate g
1
.
Finally, in our assessment of the impact of selected pheno-
logical distribution on simulations of canopy E(Fig. S8), we
observed no significant difference in daily Ebetween models
run using a natural leaf distribution an even phenological distri-
bution or an all-mature distribution. Previous work in the
Brazilian Amazon has suggested that appropriate estimation of
leaf phenological distribution is essential for accurate estimation
of gross primary productivity (GPP) and canopy A
net
(Wu
et al., 2016,2017). The fact that our models of Eare not sensi-
tive to leaf age distribution may be due in part to the limited
role leaf phenology plays in stomatal traits (Fig. 8), with the
majority of between phenological stage variation being modeled
as variation in V
cmax
(Fig. S7;Wuet al., 2019a). While V
cmax
plays a role in determining g
s
and therefore E, variation in the
g
1
and g
0
parameters would be expected to manifest in larger
variation in E(Rogers et al., 2017). Our findings underscore
the need to closely consider the method used, and assumptions
implicit to that method, when parameterizing generic PFTs, or
when applying trait-based approaches to ESMs (Van Bodegom
et al., 2012).
Acknowledgements
We would like to thank the staff of the STRI for assistance with
field logistics and support during data collection, particularly
Edwin Andrades for his skillful crane driving. This work was sup-
ported by the Next-Generation Ecosystem Experiments (NGEE)
–Tropics project, which is funded by the Biological and Envi-
ronmental Research (BER) Program within the US Department
of Energy’s (DOE) Office of Science and through DOE contract
no. DE-SC0012704 to Brookhaven National Laboratory. BTW
was supported by the National Institute of Food and Agriculture,
US Department of Agriculture, McIntire Stennis project under
LAB94493.
Author contributions
KJD, JL, AR, SJW and SPS conceived the study. KJD, JL, AR,
KSE and SPS contributed to the leaf-level data collection and
quality checking. NM, ALP, BTW and AZ contributed to the
sap flux data. KJD, JL, AR, QL, ALP, BTW and SPS analyzed
the data. KJD drafted the manuscript, and all authors con-
tributed to the final version.
Competing interests
None declared.
ORCID
Kenneth J. Davidson https://orcid.org/0000-0001-5745-
9689
Kim S. Ely https://orcid.org/0000-0002-3915-001X
Julien Lamour https://orcid.org/0000-0002-4410-507X
Qianyu Li https://orcid.org/0000-0002-0627-039X
Nate G. McDowell https://orcid.org/0000-0002-2178-2254
Alexandria L. Pivovaroff https://orcid.org/0000-0002-3104-
1900
Alistair Rogers https://orcid.org/0000-0001-9262-7430
Shawn P. Serbin https://orcid.org/0000-0003-4136-8971
Brett T. Wolfe https://orcid.org/0000-0001-7535-045X
S. Joseph Wright https://orcid.org/0000-0003-4260-5676
Alfonso Zambrano https://orcid.org/0000-0001-5442-5897
Data availability
Gas exchange data that support the findings of this study are pub-
licly available in: Lamour et al.(2021b) and Rogers et al.(2022).
Leaf trait data that support the findings of this study are publicly
available in: Lamour et al.(2022a). Spectroscopic data that sup-
port the findings of this study are publicly available in: Lamour
et al.(2021a). Sap flux data that support the findings of this study
are publicly available in: Pivovaroff et al.(2020).
References
Ainsworth EA, Rogers A. 2007. The response of photosynthesis and stomatal
conductance to rising [CO
2
]: mechanisms and environmental interactions:
photosynthesis and stomatal conductance responses to rising [CO
2
]. Plant, Cell
& Environment 30: 258–270.
Anderegg WRL, Wolf A, Arango-Velez A, Choat B, Chmura DJ, Jansen S, Kolb
T, Li S, Meinzer F, Pita P et al. 2017. Plant water potential improves
prediction of empirical stomatal models. PLoS ONE 12: e0185481.
Anderegg WRL, Wolf A, Arango-Velez A, Choat B, Chmura DJ, Jansen S, Kolb
T, Li S, Meinzer FC, Pita P et al. 2018. Woody plants optimise stomatal
behaviour relative to hydraulic risk. Ecology Letters 21: 968–977.
Ball TJ, Woodrow IE, Berry JA. 1987. A model predicting stomatal conductance
and its contribution to the control of photosynthesis under different
environmental conditions. In: Biggins J, ed. Progress in photosynthesis research.
Dordrecht, the Netherlands: Springer, 221–224.
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
New
Phytologist Research 2083

Bauerle WL, Daniels AB, Barnard DM. 2014. Carbon and water flux responses
to physiology by environment interactions: a sensitivity analysis of variation in
climate on photosynthetic and stomatal parameters. Climate Dynamics 42:
2539–2554.
B´
eland M, Baldocchi DD. 2021. Vertical structure heterogeneity in broadleaf
forests: effects on light interception and canopy photosynthesis. Agricultural
and Forest Meteorology 307: 108525.
Bernacchi CJ, Leakey ADB, Heady LE, Morgan PB, Dohleman FG, Mcgrath
JM, Gillespie KM, Wittig VE, Rogers A, Long SP et al. 2006. Hourly and
seasonal variation in photosynthesis and stomatal conductance of soybean
grown at future CO
2
and ozone concentrations for 3 yr under fully open-air
field conditions. Plant, Cell & Environment 29: 2077–2090.
Bernacchi CJ, Singsaas EL, Pimentel C, Portis AR Jr, Long SP. 2001. Improved
temperature response functions for models of Rubisco-limited photosynthesis:
in vivo Rubisco enzyme kinetics. Plant, Cell & Environment 24: 253–259.
Bonan G. 2008. Ecological climatology: concepts and applications. Cambridge, UK:
Cambridge University Press.
Bonan GB, Williams M, Fisher RA, Oleson KW. 2014. Modeling stomatal
conductance in the earth system: linking leaf water-use efficiency and water
transport along the soil–plant–atmosphere continuum. Geoscientific Model
Development 7: 2193–2222.
Buckley TN. 2016. Stomatal responses to humidity: has the ‘black box’ finally
been opened?: stomatal responses to humidity. Plant, Cell & Environment 39:
482–484.
Buckley TN, Martorell S, Diaz-Espejo A, Tom`
as M, Medrano H. 2014. Is
stomatal conductance optimized over both time and space in plant crowns? A
field test in grapevine (Vitis vinifera). Plant, Cell & Environment 37: 2707–2721.
Campany CE, Tjoelker MG, von Caemmerer S, Duursma RA. 2016. Coupled
response of stomatal and mesophyll conductance to light enhances
photosynthesis of shade leaves under sunflecks: mesophyll conductance
response to light. Plant, Cell & Environment 39: 2762–2773.
Condit R, Windsor DM, Hubbell SP. 2013. NPP tropical forest: Barro Colorado,
Panama, 1969–1990, R1. Oak Ridge, TN, USA: ORNL DAAC. doi: 10.3334/
ORNLDAAC/157.
Cowan IR. 1978. Stomatal behaviour and environment. In: Preston RD,
Woolhouse HW, eds. Advances in botanical research. London, UK: Academic
Press, 117–228.
Cowan IR, Fraquhar GD. 1977. Stomatal function in relation to leaf metabolism
and environment. In: Jennings D, ed. Integration of activity in the higher plant.
Cambridge, UK: Cambridge University Press, 471–505.
Damour G, Simonneau T, Cochard H, Urban L. 2010. An overview of models
of stomatal conductance at the leaf level: models of stomatal conductance.
Plant, Cell & Environment 39: 1419–1438.
Davidson KJ, Lamour J, Serbin SP, Rogers A. 2022. Late day measurement of
excised branches results in uncertainty in the estimation of two stomatal
parameters derived from response curves in Populus deltoides Bartr. ×Populus
nigra L. Tree Physiology 42: 1377–1395.
De Weirdt M, Verbeeck H, Maignan F, Peylin P, Poulter B, Bonal D, Ciais P,
Steppe K. 2012. Seasonal leaf dynamics for tropical evergreen forests in a
process-based global ecosystem model. Geoscientific Model Development 5:
1091–1108.
Dewar R, Mauranen A, M¨
akel¨
aA,H
¨
oltt¨
a T, Medlyn B, Vesala T. 2018. New
insights into the covariation of stomatal, mesophyll and hydraulic conductances
from optimization models incorporating nonstomatal limitations to
photosynthesis. New Phytologist 217: 571–585.
Dietze MC, Serbin SP, Davidson C, Desai AR, Feng X, Kelly R, Kooper R,
LeBauer D, Mantooth J, McHenry K et al. 2014. A quantitative assessment of
a terrestrial biosphere model’s data needs across North American biomes:
PEcAn/ED model-data uncertainty analysis. Journal of Geophysical Research:
Biogeosciences 119: 286–300.
Domingues TF, Martinelli LA, Ehleringer JR. 2014. Seasonal patterns of leaf-
level photosynthetic gas exchange in an eastern Amazonian rain forest. Plant
Ecology & Diversity 7: 189–203.
Duursma RA, Blackman CJ, Lop´
ez R, Martin-StPaul NK, Cochard H, Medlyn
BE. 2019. On the minimum leaf conductance: its role in models of plant
water use, and ecological and environmental controls. New Phytologist 221:
693–705.
Farquhar GD, Wong SC. 1984. An empirical model of stomatal conductance.
Functional Plant Biology 11: 191–210.
Franklin O. 2007. Optimal nitrogen allocation controls tree responses to elevated
CO
2
.New Phytologist 174: 811–822.
Franks PJ, Lombardozzi L, Bonan GB, Berry JA, Holbrook NM, Herold N,
Oleson KW. 2018. Comparing optimal and empirical stomatal conductance
models for application in earth system models. Global Change Biology 24:
5708–5723.
Granier A. 1987. Evaluation of transpiration in a Douglas-fir stand by means of
sap flow measurements. Tree Physiology 3: 309–320.
Green RM, Tingay S, Wang Z-Y, Tobin EM. 2002. Circadian rhythms
confer a higher level of fitness to Arabidopsis plants. Plant Physiology 129:
576–584.
H´
eroult A, Lin Y-S, Bourne A, Medlyn BE, Ellsworth DS. 2013. Optimal
stomatal conductance in relation to photosynthesis in climatically contrasting
Eucalyptus species under drought: stomatal responses of eucalyptus under
drought. Plant, Cell & Environment 36: 262–274.
Hetherington AM, Woodward FI. 2003. The role of stomata in sensing and
driving environmental change. Nature 424: 901–908.
H¨
oltt¨
a T, Lintunen A, Chan T, M¨
akel¨
a A, Nikinmaa E. 2017. A steady-state
stomatal model of balanced leaf gas exchange, hydraulics and maximal source–
sink flux. Tree Physiology 37: 851–868.
Jefferson JL, Maxwell RM, Constantine PG. 2017. Exploring the sensitivity of
photosynthesis and stomatal resistance parameters in a land surface model.
Journal of Hydrometeorology 18: 879–915.
Katul G, Manzoni S, Palmroth S, Oren R. 2010. A stomatal optimization theory
to describe the effects of atmospheric CO
2
on leaf photosynthesis and
transpiration. Annals of Botany 105: 431–442.
Katul GG, Palmroth S, Oren R. 2009. Leaf stomatal responses to vapour
pressure deficit under current and CO
2
-enriched atmosphere explained by the
economics of gas exchange. Plant, Cell & Environment 32: 968–979.
Kennedy D, Swenson S, Oleson KW, Lawrence DM, Fisher R, Lola da Costa
AC, Gentine P. 2019. Implementing plant hydraulics in the community land
model, v.5. Journal of Advances in Modeling Earth Systems 11: 485–513.
Kitajima K, Mulkey SS, Samaniego M, Joseph WS. 2002. Decline of
photosynthetic capacity with leaf age and position in two tropical pioneer tree
species. American Journal of Botany 89: 1925–1932.
Kitajima K, Mulkey SS, Wright SJ. 1997. Decline of photosynthetic capacity
with leaf age in relation to leaf longevities for five tropical canopy tree species.
American Journal of Botany 84: 702–708.
Klein T. 2014. The variability of stomatal sensitivity to leaf water potential across
tree species indicates a continuum between isohydric and anisohydric
behaviours. Functional Ecology 28: 1313–1320.
Knapp AK. 1993. Gas exchange dynamics in C
3
and C
4
grasses: consequence of
differences in stomatal conductance. Ecology 74: 113–123.
Lamour J, Davidson K, Ely K, Anderson J, Rogers A, Serbin S. 2021a. Leaf
spectral reflectance and transmittance, San Lorenzo, Panama, 2020. 1.0. NGEE
Tropics Data Collection (dataset). doi: 10.15486/ngt/1781003.
Lamour J, Davidson K, Ely K, Anderson J, Rogers A, Serbin S. 2022a. Leaf
structural and chemical traits, and BNL field campaign sample details, San
Lorenzo, Panama, 2020. 1.0. NGEE Tropics Data Collection (dataset). doi: 10.
15486/ngt/1783737.
Lamour J, Davidson KJ, Ely KS, Anderson J, Serbin SP, Rogers A. 2021b. Leaf
gas exchange and fitted parameters, San Lorenzo, Panama, 2020. 1.0. NGEE
Tropics Data Collection (dataset). doi: 10.15486/ngt/1781004.
Lamour J, Davidson KJ, Ely KS, Anderson JA, Rogers A, Wu J, Serbin SP.
2021c. Rapid estimation of photosynthetic leaf traits of tropical plants in
diverse environmental conditions using reflectance spectroscopy. PLoS ONE
16: e0258791.
Lamour J, Davidson KJ, Ely KS, Le Mogu´
edec G, Leakey ADB, Li Q, Serbin
SP, Rogers A. 2022b. An improved representation of the relationship between
photosynthesis and stomatal conductance leads to more stable estimation of
conductance parameters and improves the goodness-of-fit across diverse data
sets. Global Change Biology 28: 3537–3556.
Lamour J, Davidson KJ, Ely KS, Li Q, Serbin SP, Rogers A. 2022c. New
calculations for photosynthesis measurement systems: what’s the impact for
physiologists and modelers? New Phytologist 233: 592–598.
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
Research
New
Phytologist
2084

Lamour J, Serbin SP. 2021. LeafGasExchange. doi: 10.5281/zenodo.4545818.
Lawson T, Blatt MR. 2014. Stomatal size, speed, and responsiveness impact on
photosynthesis and water use efficiency. Plant Physiology 164: 1556–1570.
Lawson T, Vialet-Chabrand S. 2019. Speedy stomata, photosynthesis and plant
water use efficiency. New Phytologist 221:93–98.
Leach JE, Woodhead T, Day W. 1982. Bias in pressure chamber measurements
of leaf water potential. Agricultural Meteorology 27: 257–263.
Leakey ADB, Bernacchi CJ, Ort DR, Long SP. 2006. Long-term growth of
soybean at elevated [CO
2
] does not cause acclimation of stomatal conductance
under fully open-air conditions. Plant, Cell & Environment 29: 1794–1800.
Leuning R. 1995. A critical appraisal of a combined stomatal-photosynthesis
model for C
3
plants. Plant, Cell & Environment 18: 339–355.
Li Q, Serbin SP, Lamour J, Davidson KJ, Ely KS, Rogers A. 2022.
Implementation and evaluation of the unified stomatal optimization approach
in the Functionally Assembled Terrestrial Ecosystem Simulator (FATES).
Geoscientific Model Development 15: 4313–4329.
Lin Y-S, Medlyn BE, Duursma RA, Prentice IC, Wang H, Baig S, Eamus D, de
Dios VR, Mitchell P, Ellsworth DS et al. 2015. Optimal stomatal behaviour
around the world. Nature Climate Change 5: 459–464.
Long SP, H¨
allgren J-E. 1993. Measurement of CO
2
assimilation by plants in the
field and the laboratory. In: Hall DO, Scurlock JMO, Bolh`
ar-Nordenkampf
HR, Leegood RC, Long SP, eds. Photosynthesis and production in a changing
environment: a field and laboratory manual. Dordrecht, the Netherlands:
Springer, 129–167.
Machado R, Loram-Lourenc
¸o L, Farnese FS, Alves RDFB, Sousa LF, Silva FG,
Filho SCV, Torres-Ruiz JM, Cochard H, Menezes-Silva PE. 2021. Where do
leaf water leaks come from? Trade-offs underlying the variability in minimum
conductance across tropical savanna species with contrasting growth strategies.
New Phytologist 229: 1415–1430.
Manzoni S, Vico G, Katul G, Fay PA, Polley W, Palmroth S, Porporato A.
2011. Optimizing stomatal conductance for maximum carbon gain under
water stress: a meta-analysis across plant functional types and climates: optimal
leaf gas exchange under water stress. Functional Ecology 25: 456–467.
Martı´nez Cano I, Muller-Landau HC, Wright SJ, Bohlman SA, Pacala SW.
2019. Tropical tree height and crown allometries for the Barro Colorado
Nature Monument, Panama: a comparison of alternative hierarchical models
incorporating interspecific variation in relation to life history traits.
Biogeosciences 16: 847–862.
McAdam SAM, Brodribb TJ. 2015. The evolution of mechanisms driving the
stomatal response to vapor pressure deficit. Plant Physiology 167: 833–843.
McDowell NG, Adams HD, Bailey JD, Hess M, Kolb TE. 2006. Homeostatic
maintenance of ponderosa pine gas exchange in response to stand density
changes. Ecological Applications 16: 1164–1182.
Medlyn BE, Duursma RA, Eamus D, Ellsworth DS, Prentice IC, Barton CVM,
Crous KY, De Angelis P, Freeman M, Wingate L. 2011. Reconciling the
optimal and empirical approaches to modelling stomatal conductance. Global
Change Biology 17: 2134–2144.
Meidner H, Mansfield TA. 1968. Physiology of stomata. London, UK: McGraw-
Hill.
Meinzer FC, Goldstein G, Andrade JL. 2001. Regulation of water flux through
tropical forest canopy trees: Do universal rules apply? Tree Physiology 21:19–26.
Meinzer FC, James SA, Goldstein G, Woodruff D. 2003. Whole-tree water
transport scales with sapwood capacitance in tropical forest canopy trees: water
transport scales with sapwood properties in tropical trees. Plant, Cell &
Environment 26: 1147–1155.
Meng F-R, Arp PA. 1993. Net photosynthesis and stomatal conductance of red
spruce twigs before and after twig detachment. Canadian Journal of Forest
Research 23: 716–721.
Missik JEC, Oishi AC, Benson MC, Meretsky VJ, Phillips RP, Novick KA.
2021. Performing gas-exchange measurements on excised branches –evaluation
and recommendations. Photosynthetica 59:61–73.
Miyazawa Y, Tateishi M, Komatsu H, Kumagai T, Otsuki K. 2011. Are
measurements from excised leaves suitable for modeling diurnal patterns of gas
exchange of intact leaves? Hydrological Processes 25: 2924–2930.
Norman JM. 1979. Modeling the complete crop canopy. In: Barfield G, ed.
Modification of the aerial environment of crops. St Joseph, MI, USA: American
Society of Agricultural Engineers, 249–277.
Osnas JLD, Katabuchi M, Kitajima K, Wright SJ, Reich PB, Van Bael SA, Kraft
NJB, Samaniego MJ, Pacala SW, Lichstein JW. 2018. Divergent drivers of
leaf trait variation within species, among species, and among functional groups.
Proceedings of the National Academy of Sciences, USA 115: 5480–5485.
Paton S. 2020. Yearly reports_San Lorenzo Crane. Balboa Ancon, Panama:
Smithsonian Tropical Research Institute. [WWW document] URL https://
smithsonian.figshare.com/articles/dataset/Yearly_Reports_San_Lorenzo_
Crane/11799309/2 [accessed 15 February 2021].
Pinheiro J, Bates D, DebRoy S, Sarkar D, Heisterkamp S, Bert VW. 2020.
Package ‘NLME’. [WWW document] URL https://cran.r-project.org/web/
packages/nlme/index.html [accessed 15 February 2021].
Pivovaroff A, McDowell N, Davies S, Detto M, Wolfe BT, Wright J, Zambrano
A. 2020. Sap flow data from San Lorenzo, Panama (PA-SLZ) from January–June
2020. 1.0. NGEE Tropics Data Collection (dataset). doi: 10.15486/ngt/1678733.
R Core Team. 2013. R: a language and environment for statistical computing.
Vienna, Austria: R Foundation for Statistical Computing.
Resco de Dios V, Anderegg WRL, Li X, Tissue DT, Bahn M, Landais D, Milcu
A, Yao Y, Nolan RH, Roy J et al. 2020. Circadian regulation does not
optimize stomatal behaviour. Plants 9: 1091.
Resco de Dios V, Loik ME, Smith R, Aspinwall MJ, Tissue DT. 2016. Genetic
variation in circadian regulation of nocturnal stomatal conductance enhances
carbon assimilation and growth: on the function of nocturnal stomatal
conductance. Plant, Cell & Environment 39:3–11.
Ricciuto D, Sargsyan K, Thornton P. 2018. The impact of parametric
uncertainties on biogeochemistry in the E3SM land model. Journal of Advances
in Modeling Earth Systems 10: 297–319.
Rodriguez-Dominguez CM, Forner A, Martorell S, Choat B, Lopez R, Peters
JMR, Pfautsch S, Mayr S, Carins-Murphy MR, McAdam SAM et al. 2022.
Leaf water potential measurements using the pressure chamber: synthetic
testing of assumptions towards best practices for precision and accuracy. Plant,
Cell & Environment 45: 2037–2061.
Rogers A, Allen DJ, Davey PA, Morgan PB, Ainsworth EA, Bernacchi CJ, Cornic
G, Dermody O, Dohleman FG, Heaton EA et al.2004.Leaf photosynthesis and
carbohydrate dynamics of soybeans grown throughout their life-cycle under free-
air carbon dioxide enrichment. Plant, Cell & Environment 27:449–458.
Rogers A, Medlyn BE, Dukes JS, Bonan G, von Caemmerer S, Dietze MC,
Kattge J, Leakey ADB, Mercado LM, Niinemets
¨
Uet al. 2017. A roadmap for
improving the representation of photosynthesis in earth system models. New
Phytologist 213:22–42.
Rogers A, Serbin SP, Ely K, Wu J, Wolfe BT, Dickman LT, Collins AD, Detto
M, Grossiord C, McDowell N et al. 2022. Diurnal leaf gas exchange survey,
Feb2016-May2016, PA-SLZ, PA-PNM: Panama. 2. NGEE Tropics Data
Collection (dataset). doi: 10.15486/ngt/1411972.
Santiago LS, Mulkey SS. 2003. A test of gas exchange measurements on excised
canopy branches of ten tropical tree species. Photosynthetica 41: 343–347.
van Schaik CP, Terborgh JW, Wright SJ. 1993. The phenology of tropical
forests: adaptive significance and consequences for primary consumers. Annual
Review of Ecology and Systematics 24: 353–377.
Schlesinger WH, Jasechko S. 2014. Transpiration in the global water cycle.
Agricultural and Forest Meteorology 189–190: 115–117.
Scholander PF, Hammel HT, Hemmingsen EA, Bradstreet ED. 1964.
Hydrostatic pressure and osmotic potential in leaves of mangroves and some
other plants. Proceedings of the National Academy of Sciences, USA 52: 119–125.
Sirri NF, Libalah MB, Momo Takoudjou S, Ploton P, Medjibe V, Kamdem
NG, Mofack G, Sonk´
e B, Barbier N. 2019. Allometric models to estimate leaf
area for tropical African broadleaved forests. Geophysical Research Letters 46:
8985–8994.
Speckman H. 2019. AQUAFLUX: swift and transparent analysis of plant-level
transpiration. [WWW document] URL https://rdrr.io/github/
HeatherSpeckman/AquaFlux/man/AquaFlux.html [accessed 10 March 2021].
Sperry JS. 2013. Cutting-edge research or cutting-edge artefact? An overdue
control experiment complicates the xylem refilling story: xylem-refilling
artefacts. Plant, Cell & Environment 36: 1916–1918.
Sperry JS, Venturas MD, Anderegg WRL, Mencuccini M, Mackay DS, Wang Y,
Love DM. 2017. Predicting stomatal responses to the environment from the
optimization of photosynthetic gain and hydraulic cost: a stomatal
optimization model. Plant, Cell & Environment 40: 816–830.
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
New
Phytologist Research 2085

Tobin AK, Rogers WJ. 1992. Metabolic interactions of organelles during leaf
development. In: Tobin AK, ed. Seminar series-society for experimental biology.
Plant organelles. Cambridge, UK: Cambridge University Press, 293–323.
Van Bodegom PM, Douma JC, Witte JPM, Ordo˜
nez JC, Bartholomeus RP,
Aerts R. 2012. Going beyond limitations of plant functional types when
predicting global ecosystem-atmosphere fluxes: exploring the merits of traits-
based approaches: merits of traits-based vegetation modelling. Global Ecology
and Biogeography 21: 625–636.
Venturas MD, Sperry JS, Hacke UG. 2017. Plant xylem hydraulics: what we
understand, current research, and future challenges: plant xylem hydraulics.
Journal of Integrative Plant Biology 59: 356–389.
Verryckt LT, Van Langenhove L, Ciais P, Courtois EA, Vicca S, Pe ˜
nuelas J,
Stahl C, Coste S, Ellsworth DS, Posada JM et al. 2020. Coping with branch
excision when measuring leaf net photosynthetic rates in a lowland tropical
forest. Biotropica 52: 608–615.
Von Caemmerer S. 2013. Steady-state models of photosynthesis: steady-state
models of photosynthesis. Plant, Cell & Environment 36: 1617–1630.
Way DA, Pearcy RW. 2012. Sunflecks in trees and forests: from photosynthetic
physiology to global change biology. Tree Physiology 32: 1066–1081.
Wingler A, Mar`
es M, Pourtau N. 2004. Spatial patterns and metabolic
regulation of photosynthetic parameters during leaf senescence. New Phytologist
161: 781–789.
Wolf A, Anderegg WRL, Pacala SW. 2016. Optimal stomatal behavior with
competition for water and risk of hydraulic impairment. Proceedings of the
National Academy of Sciences, USA 113: E7222–E7230.
Wolz KJ, Wertin TM, Abordo M, Wang D, Leakey ADB. 2017. Diversity in
stomatal function is integral to modelling plant carbon and water fluxes. Nature
Ecology & Evolution 1: 1292–1298.
Wong SC, Cowan IR, Farquhar GD. 1979. Stomatal conductance correlates
with photosynthetic capacity. Nature 282: 424–426.
Wong SC, Cowan IR, Farquhar GD. 1985. Leaf conductance in relation to rate
of CO
2
assimilation. Plant Physiology 78: 821–825.
Wu J, Albert LP, Lopes AP, Restrepo-Coupe N, Hayek M, Wiedemann KT,
Guan K, Stark SC, Christoffersen B, Prohaska N et al. 2016. Leaf
development and demography explain photosynthetic seasonality in Amazon
evergreen forests. Science 351: 972–976.
Wu J, Rogers A, Albert LP, Ely K, Prohaska N, Wolfe BT, Oliveira RC, Saleska
SR, Serbin SP. 2019a. Leaf reflectance spectroscopy captures variation in
carboxylation capacity across species, canopy environment and leaf age in
lowland moist tropical forests. New Phytologist 224: 663–674.
Wu J, Serbin SP, Ely KS, Wolfe BT, Dickman LT, Grossiord C, Michaletz ST,
Collins AD, Detto M, McDowell NG et al. 2019b. The response of stomatal
conductance to seasonal drought in tropical forests. Global Change Biology 26:
823–839.
Wu J, Serbin SP, Xu X, Albert LP, Chen M, Meng R, Saleska SR, Rogers A.
2017. The phenology of leaf quality and its within-canopy variation is essential
for accurate modeling of photosynthesis in tropical evergreen forests. Global
Change Biology 23: 4814–4827.
Zhang J, Jiang H, Song X, Jin J, Zhang X. 2018. The responses of plant leaf
CO
2
/H
2
O exchange and water use efficiency to drought: a meta-analysis.
Sustainability 10: 551.
Zhou S-X, Medlyn B, Sabat´
e S, Sperlich D, Prentice IC, Whitehead D. 2014.
Short-term water stress impacts on stomatal, mesophyll and biochemical
limitations to photosynthesis differ consistently among tree species from
contrasting climates. Tree Physiology 34: 1035–1046.
Zhou S-X, Medlyn BE, Prentice IC. 2016. Long-term water stress leads to
acclimation of drought sensitivity of photosynthetic capacity in xeric but not
riparian Eucalyptus species. Annals of Botany 117: 133–144.
Supporting Information
Additional Supporting Information may be found online in the
Supporting Information section at the end of the article.
Fig. S1 Histograms of the atmospheric vapor pressure deficit
(VPD
air
) observed during survey gas exchange measurements.
Fig. S2 Regression of stomatal response from the two different
measurement methods.
Fig. S3 Observed vs predicted fluxes of canopy-scale transpira-
tion (E) for Guatteria dumetorum,Miconia borealis,Terminalia
amazonia, and Vochysia ferruginea using three different methods
to derive stomatal parameters.
Fig. S4 Stomatal intercept (g
0
) estimates by species, phenological
stage, and measurement method.
Fig. S5 Relationships between estimated stomatal and leaf func-
tional traits.
Fig. S6 Relationship between leaf water potential (Ψ
leaf
) and
model residual (difference between observed and predicted stom-
atal conductance (g
s
)) for the survey model including a time-of-
day effect.
Fig. S7 Relationship between leaf phenological stage and photo-
synthetic traits.
Fig. S8 Simulations of daily canopy transpiration (E) by species
when using different leaf phenology distributions.
Fig. S9 Comparison of fitted values of stomatal slope (g
1
) and
stomatal intercept (g
0
) by species for curves conducted at differ-
ent times of day.
Fig. S10 Estimated stomatal slope (g
1
) vs length of response
curve measurement.
Fig. S11 Stomatal slope (g
1
) as estimated from response curves
collected with decreasing (saturating to 0 μmol PAR) and
increasing (0 μmol PAR to saturating) levels of irradiance.
Fig. S12 Observed vs predicted stomatal conductance (g
s
) from
the calibration phase of model development.
Fig. S13 Range and distribution of individual observations of
leaf-level conditions collected during survey gas exchange.
Fig. S14 Mean estimated intercellular carbon mole fraction (C
i
)
by species and leaf age.
Fig. S15 Conceptual illustration of the link between stomatal
slope and water use efficiency.
Fig. S16 Observed vs predicted stomatal conductance (g
s
) when a
leaf water potential (Ψ
leaf
) penalty is applied (upper panels) and
when no penalty is applied (lower panels).
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
Research
New
Phytologist
2086

Notes S1 Description of mixed effects models and their parame-
ter estimates.
Notes S2 Description of the leaf scale photosynthesis, conduc-
tance, and energy balance model from the R package LEAFGASEX-
CHANGE (Lamour & Serbin, 2021).
Notes S3 Description of the leaf water potential (Ψ
leaf
) penalty
function introduced by Wolf et al.(2016) and tested by Ander-
egg et al.(2017).
Table S1 Parameter estimates and their standard error for the
response curve model.
Table S2 Parameter estimates and their standard error for the
survey model.
Table S3 Parameter estimates and their standard error for the
survey +time-of-day model.
Table S4 Species-specific Weibull function values corresponding
to xylem conductivity vs stem water potential (Ψ
stem
).
Please note: Wiley is not responsible for the content or function-
ality of any Supporting Information supplied by the authors. Any
queries (other than missing material) should be directed to the
New Phytologist Central Office.
Ó2022 The Authors
New Phytologist Ó2022 New Phytologist Foundation
New Phytologist (2023) 237: 2069–2087
www.newphytologist.com
New
Phytologist Research 2087