Carbon pools in Mediterranean forests: comparing eddy
covariance and GOTILWA+ model results.
Carlos Gracia1,2, Santi Sabaté1,2, Trevor Keenan2
1.Department of Ecology, University of Barcelona, Barcelona, Spain.
2. CREAF, Autonomous University of Barcelona, Barcelona, Spain.
[email protected]
Abstract
Modeling and monitoring the processes involved in terrestrial carbon sequestration are often thought to be
independent events. In fact, rigorously validated modern modeling techniques can prove very useful tools in the
monitoring of the carbon sequestration potential of an ecosystem through simulation, by highlighting key areas
for study of what is a complex dynamical system. This is ever more important in the light of climate change,
where it becomes essential to have an understanding of the future role of terrestrial ecosystems as potential sinks
or sources in the global carbon cycle.
The lack of understanding of the effect of water stress on ecosystem function leads to big discrepancies when
modelling under these conditions. Recent advances allow us to successfully reproduce water and carbon balances
during periods of stress. This will be of particular importance for modelling future scenarios, where water stress
is expected to increase in Mediterranean ecosystems.
The analysis of FLUXNET data gathered at four Mediterranean forest sites made it possible to separate stomatal
and non-stomatal limitations to assimilation under drought conditions on the canopy scale. From the analysis,
new process descriptions were developed which allow for the accurate simulation of carbon and water fluxes at
each site.
Three sites in Mediterranean Europe and one in Mediterranean California were chosen, including both deciduous
and evergreen species - Quercus cerris, Quercus ilex, Pinus ponderosa, and Fagus sylvatica, and varying degrees
of water stress and ecosystem responses.
The effectiveness of ecosystem process based modelling using the forest growth model GOTILWA+ was
assessed, and both observed and modelled canopy responses to stressed conditions were analysed, by confronting
the models with FLUXNET data from each site.
The results show that the currently accepted hypothesis of stomatal control over photosynthesis does not hold in
water stressed conditions, and reveal the involvement of non-stomatal limitations to net photosynthesis which
come into play under more severe water stress. It is shown that, applying non-stomatal limitations to assimilation,
carbon and water fluxes can now be modelled to an equal degree of accuracy in both stressed and non stressed
conditions.
Keywords: Gotilwa+, Ecosystem Models, FLUXNET, Modelling, drought-stress, Mediterranean climate
Introduction
Changes in regional and global climate have already been observed, and projections of climate change
suggest that higher temperatures, and increased potential evapotranspiration, as well as changes in
seasonal precipitation patterns (IPCC AR4, 2007) will probably aggravate the seasonal drought stress
characteristic to Mediterranean ecosystems. As soil water availability is already the main limiting
factor to Mediterranean production, this could thereby push certain ecosystems beyond their climatic
limits of existence. However, despite recent advances, our knowledge of the effect of water stress on
carbon and water fluxes is limited.
In modelling the vulnerability of Mediterranean vegetation, it is vital to know how their rate of
assimilation and thus their capacity to store carbon is affected by drought. Only through a thorough
understanding of the key processes will it be possible to design process-based models that confidently
predict the effect of potential future changes in these ecosystems
Process-based models incorporate the relevant ecosystem processes to successfully simulate the
sensitivity of ecosystem functioning to drought stress at all relevant time scales and are thus
indispensable tools to assess the vulnerability of Mediterranean ecosystems. However, previous
modelling studies have highlighted that, at regional to global scales, models consistently perform worse
in Mediterranean ecosystems than in any others, mainly related to difficulties in reproducing the effects
of seasonal drought on carbon and water fluxes (Morales et al., 2005; Reichstein et al., 2007; Jung et
al., 2007).
The effect of water stress on plant photosynthesis and stomatal conductance has been widely studied in
leaf level laboratory experiments (Wilson, Baldocchi & Hanson 2000; Chaves et al., 2002). Empirical
relationships based on the stomatal control of photosynthesis have been developed to scale the
functioning from the leaf level to the canopy of ecosystems (Field, 1995). Such relationships are
employed by process-based ecosystem models to simulate the effect of water stress on ecosystem
fluxes. Recent studies suggest that these relationships do not hold under severe water stress (Reichstein
et al., 2003; Misson et al., 2006). The most widely adhered paradigm for the control of photosynthesis
during water stressed periods is stomatal closure, that reduces leaf level water loss at the expense of
reducing the supply of carbon to the intercellular spaces required for continued photosynthesis (Chaves
et al., 2002).
More recent studies also show the existence of non-stomatal limitations (Wilson, Baldocchi & Hanson,
2000), which have been hypothesised to relate to changes in the conductivity of the mesophyll, or
metabolic changes. There has been much discourse over the weight of the roles played by stomatal and
non-stomatal limitations during drought periods (Jones, 1985; Ni & Pallardy, 1992; Kubiske & Adams,
1993; Wilson et al., 2000), with many studies giving conflicting results and great uncertainty as to
which limitation is strongest under drought conditions (Lawlor, 1995; Tezara et al., 1999; Lawlor &
Cornic, 2002; Flexas & Medrano, 2002). Here we provide a careful data analysis that allows for the
weight and timing of stomatal limitations to assimilation to be assessed against that of non-stomatal
limitations. This is the first study to look at the dynamic between the two limiting factors on the canopy
scale, including several ecosystems.
The utility of the understanding derived from the observations is evaluated using Gotilwa+, a
biogeochemical forest growth model (Gracia et al., 1999), first at Puechabon forest and then to four
different Mediterranean sites. Three sites are situated in Europe (Puechabon, France; Roccarespampani,
Italy; Collelongo, Italy), under the CarboEurope-EUROFLUX project, and one site in the United States
(Blodgett, California), under the AMERIFLUX project. They are mostly monospecific forest sites, and
include Quercus ilex, Quercus cerris, Fagus sylvatica, and Pinus ponderosa, respectively.
Gotilwa+ is an individual based forest growth model, designed to simulate carbon and water fluxes for
individual forest stands, accounting for tree species and site characteristics. Simulations are preformed
at each site, as a validation of the model, and a comparison is made of their performance using the
standard process assumptions with their performance using the inclusion of the new process
descriptions. We evaluate the effect this has on site and regional scale simulations, and discuss the
generality of the findings.
2
Materials and Methods
Model Approach
Process based models are complex simulators that attempt to mimic the real world. The aim is to
include mathematical descriptions of both the processes that govern a system, and their interactions,
thus recreating the system in a virtual environment. Each process in the system is described separately,
and dynamically interacts with other processes. Given an accurate description of each processes
separately, it is argued that a better description of the ecosystem in general, through the interaction of
these processes, can be achieved. Due to their detail, a large number of parameters are necessary. The
parameters determine the response of each function describing an individual process, and are based on
detailed field work or lab experiments. This allows an accurate description of all factors affecting a
process, but such parameters are not always available. This can be a problem, and a lack of data often
leads to assumptions and approximations, but the approach leads to a model with a wide applicability.
The detail and dynamic characteristic of process based models allows them, theoretically, to function
as effective exploratory tools and they should be fully applicable under new conditions and scenarios.
GOTILWA+: A Process-Based Model
Gotilwa+ (Growth of Trees Is Limited by WAter), is a highly mechanistic process based forest growth
model. It comprises of a two layer canopy photosynthetic model, coupled with a carbon allocation and
growth model and a soil respiration and hydrology model. It describes monospecific stands, which can
be even or uneven aged. The key environmental forcing factors taken into account are precipitation, air
temperature, vapour pressure, global radiation, wind speed, and atmospheric carbon dioxide
concentration.GOTILWA+ has been implemented to simulate forest growth processes and to explore
how these processes are influenced by climate, tree stand structure, management techniques, soil
properties and climate change. The Gotilwa+ model simulates carbon and water fluxes through forests
in different environments, for different tree species, under changing environmental conditions, either
due to climate or to management regimes.
Climate
Tree & Stand structure
Pipe Model
Pipe Model
Soil Traits & Processes
Ecosystem Physiology
Figure 1. A schematic graph of processes and interactions accounted for in the GOTILWA+ model
3
Results of GOTILWA+ are computed separately for 50 DBH classes and they are integrated at the
stand level. The processes are described with different sub-models that interact and integrate the results
of simulated growth and the evolution of the whole tree stand through time (hourly calculations
integrated at a daily time step). Horizontal space is assumed homogeneous and the vertical profile
distinguishes two canopy layers (sun and shade conditions).
Input and Output Variables
The input data includes: climate (max. and min. temperatures, rainfall, VPD, wind speed, global
radiation and atmospheric CO2 concentration); stand characteristics (tree structure including the
structure of the canopy; DBH class distribution); tree physiology (photosynthetic and stomatal
conductance parameters, specific growth and maintenance respiration rates), site conditions including
soil characteristics and hydrological parameters and also forest management criteria.
Many output variables can be extracted from the model. These can be separated into three main
categories: canopy variables, tree and stand structural variables, and root and soil variables. Canopy
variables include: gross primary production, net primary production, net ecosystem exchange, leaf area
index, transpiration, water use efficiency, leaf production, leaf respiration, leaf biomass, growth
activity, the length of the growing period and volatile organic compound emissions.
Tree and stand structural variables include: tree density, sapling density, basal area, sapwood area,
mean quadratic tree diameter, vigour index, tree height, wood production, wood respiration, mobile
carbohydrates, tree ring width, aboveground biomass, the weight of the sapwood column, wood
volume, dead wood volume and yield (when considering management). Root and soil variables
include: soil temperature, water stored in soil, fine and gross litter fall, soil organic carbon, fine root
biomass, fine root production, fine root respiration, heterotrophic respiration, maintenance respiration
and growth respiration.
How GOTILWA+ copes with processes.
Process based models start at the very basic physiological leaf level, combining and describing the
different processes involved. Figure 2 shows a schematic of the most fundamental compartment, the
leaf. Here, photosynthesis is calculated dynamically, based on internal and external conditions.
Photosynthesis is calculated using a two layer model which splits the available leaf area index into sun
and shade leaves, depending on the time of the day, leaf area angle and the canopy’s ellipsoidal
distribution. Assimilation rates depend on the direct and diffuse radiation intercepted, the species
specific photosynthetic capacities, leaf temperature, available carbon, the extent of stomatal opening,
and the availability of soil water. Growth and the allocation of mobile carbon for tree maintenance are
considered through three compartments: Leaf respiration, Sapwood respiration, and Fine roots
respiration. Respiration rates respond to water stress.
Fine litter fall (e.g. leaves), gross litter fall (e.g. bark, branches) and the mortality of fine roots add to
the soil organic carbon content. The soil in Gotilwa+ is divided into two layers, an organic layer and a
mineral layer, with a rate of transfer between them. Soil organic carbon is decomposed depending on to
which layer it belongs, with both decomposition rates depending on a Q10 function taking into account
soil water content and soil temperature. Soil temperature is calculated from air temperature using a
moving average of 11 days. The amount of soil water available for organic layers is calculated taking
into account the cumulated rainfall of the previous 30 days and soil water availability for mineral layers
depends on the soils water filled porosity which in turn is a function of the organic matter present in
soil. Soil water content is described as one layer, taking inputs though precipitation less leaf
interception, which is evaporated, (stem interception, or stemflow, is not evaporated, but directed to the
soil), and outputs though drainage, runoff, and transpiration. Surface evaporation only occurs when the
canopy is not closed.
4
Flux calculations are performed hourly, whereas slower processes such as growth and other state
variables are calculated daily. Gotilwa+ comprises of a two layer canopy photosynthetic model (Wang
& Leuning, 1998), coupled with a carbon allocation and growth model and a soil respiration and
hydrology model. The photosynthesis submodel splits the available leaf area index into sun and shade
leaves, with intercepted radiation depending on the time of the day, leaf area angle and the canopy’s
ellipsoidal distribution. Assimilation rates are calculated using the Farquhar approach (von Caemmerer
& Farquhar, 1981) and depend on the direct and diffuse radiation intercepted, the species specific
photosynthetic capacities, leaf temperature, available internal carbon, the extent of stomatal opening.
Stomatal conductance is calculated using the Leuning, Ball and Berry model, (Leuning et al., 1995).
Growth and the allocation of mobile carbon for tree maintenance are considered through three
compartments: Leaf respiration, Sapwood respiration, and fine roots respiration. Stand characteristics
are taken into account and species specific parameters, to give highly accurate predictions of forest
growth and carbon or water fluxes in the system.
Q
transpiration rate
Leaf energy balance
Q-(R+C+λ E)< ε
YES
NO
stomatal conductance
leaf temperature
f(θ)
assimilation rate
estimation of Vc, J and other
temperature-dependent
parameters of photosynthesis
Ci=f(C a, gs , E, A)
soil water storage
Figure 2. A schematic diagram of the model representation of the photosynthetic assimilation rate. Leaf
Temperature is the result of the energetic balance which depends on the radiation which reaches the leaf and the
amount of transpired water. Leaf temperature influences the rate of assimilation, which is also dependent on the
stomatal conductance. These interactions constitute the basis of the coupling of the eco-physiological processes
of the leaf with it’s surrounding atmosphere. In Gotilwa+, the leaf is initially assigned an arbitrary temperature
(tair+5ºC). With this temperature, it is possible to determine the values of Vc, J and the resulting photosynthetic
parameters whose values determine the rate of assimilation. The rate of assimilation is in equilibrium with the
stomatal conductance and, as a result, the rate of transpiration, which is in turn dependent on the quantity of
water in the soil. Once the rate of transpiration of determined, the energetic balance of the leaf is recalculated
and a check is made to see if the radiation absorbed and the radiation emitted by the leaf differ by less than a
fixed value ε (normally ε<0.001ºC). The leaf temperature is modified progressively until a point is reached
where the energetic balance, temperature, assimilation, conductance and transpiration couple together.
Fine litter fall (i.e. leaves), gross litter fall (i.e. bark, branches) and the mortality of fine roots add to the
soil organic carbon content. The soil in Gotilwa+ is divided into two layers, an organic layer and a
5
mineral layer, with a rate of transfer between them. Soil organic carbon is decomposed depending on to
which layer it belongs, with both decomposition rates depending on a Q10 function taking into account
soil water content and soil temperature. Soil temperature is calculated from air temperature using a
moving average of 11 days. The amount of soil water available for organic layers is calculated taking
into account the cumulated rainfall of the previous 30 days and soil water availability for mineral layers
depends on the soils water filled porosity which in turn is a function of the organic matter present in
soil.
Soil water content is described as one layer, taking inputs though precipitation less leaf interception,
which is evaporated, (stem interception, or stemflow, is not evaporated, but directed to the soil), and
outputs though drainage, runoff, evaporation and transpiration. Surface evaporation only occurs when
the canopy is not closed.
Flux calculations are performed hourly, whereas slower processes such as growth and other state
variables are calculated daily.
The input data includes: half-hourly meteorological variables (temperature, precipitation, vapour
pressure deficit, wind speed, global radiation and atmospheric CO2 concentration), which have been
taken from site observations; and site conditions including soil characteristics and hydrological
parameters (Table 1). Observed or reported values from the literature of photosynthetic were also used
for each site and model. Stomatal conductance parameters were calculated from the data (Table 2). In
addition Gotilwa+ requires descriptions of stand characteristics (tree structure including the structure of
the canopy; DBH class distribution), and also allows for the possibility of forest management, as well
as some tree physiology parameters (specific growth and maintenance respiration rates).
Fluxnet Site Data and Data Manipulation
Simulations were run at 3 EUROFLUX sites in Mediterranean Europe, and one AMERIFLUX site in
Mediterranean California, covering a total of 12 years. The sites were chosen to include a range of
species, including Deciduous, Coniferous and Broadleaf Evergreen species, at various altitudes, and
varying levels of water stress in summer.
Table 1. Characteristics of the FluxNet sites chosen. Physiological functional types (PFT) considered are
temperate broadleaved evergreen (TeBE), needle-leaved evergreen (TeNE) and broadleaved-summergreen
(TeBS). Max LAI - Maximum Leaf Area Index; SD – Soil Depth; SWHC – Soil Water Holding Capacity.
Site
Period
Puéchabon,
France
Rocarespampani,
Italy
Collelongo,
Italy
Blodgett Forest,
California
20012004
20022004
19981999
20012004
Longitude
Latitude
Altitude
3º 35’
43º 44’
270
11º 55’
42º 23’
223
13º 35’
41º 50’
1560
-120º 37’
38º 53’
1315
Max LAI
(m2/ m2)
2.9 - 3.2
4.0 - 5.0
4 – 5.5
2.4 – 4.2
SD
(m)
SWHC
(Kg/ m2)
4.5
172
4.5
485?
4
287
4
583
Species/PFT
Quercus ilex
(TeBE)
Quercus cerris
(TeBS)
Fagus sylvatica
(TeBS)
Pinus ponderosa
(TeNE)
Fluxnet provides continuous measurement of carbon dioxide and water fluxes on a seasonal basis with
half-hourly discrimination (Wofsy et al., 1993), and new flux separation techniques now give the
improved level 4 data set (Reichstein et al., 2005) which has been used in this study. Data was gap
6
filled using the Artificial Neural Network method (Papale & Valentini, 2003), with carbon flux
measurements broken down into assimilation rates using estimated ecosystem respiration.
Although the methodology is highly refined, and eddy covariance techniques have developed to apply
very sophisticated algorithms and techniques for the back calculation of many variables, it should be
kept in mind that the resulting measurements of ecosystem variables are results of modelling
techniques, not to mention gap filling techniques employed. It is also often the case that the energy
budget does not close (Wilson et al., 2002). Thus, any comparison between models such as Gotilwa+
and FLUXNET data is, in reality, an inter-model comparison. That said, the data generated by the
FLUXNET eddy covariance techniques is probably as close as we will ever get to the truth, and
certainly is as close as we can get right now.
Estimation of fractional soil water storage, canopy conductance and leaf inter-cellular carbon
concentration
Fractional soil water storage
Understanding the response of observed carbon and water fluxes to changes in soil moisture requires
the seasonal evolution of soil water content to be known. In the absence of such measurements, daily
soil moisture content at each site was reconstructed by inverting the measured latent heat flux, given
the observed maximum soil water holding capacity, and inputs of water into the soil from rainfall (less
interception). A simple ‘bucket’ soil water model, accounting for drainage and runoff as a function of
soil moisture (Gracia et al., 1999), was employed prescribing the evapotranspiration loss of soil water
using the observed latent heat fluxes.
The method to reconstruct soil moisture can in addition be applied to prescribe soil moisture in the
ecosystem models. Decoupling the simulation of soil and canopy processes by removing the water
volume equivalent of the observed latent heat flux at each time step instead of the simulated
evapotranspiration, allows for the validation of the canopy physiological process description, and thus
to assess the modelling of physiological responses to soil moisture independent of potential
inaccuracies in the simulated hydrological budgets of the ecosystem models.
Calculation of Canopy conductance and internal leaf carbon dioxide concentrations
In order to gauge ecophysiological responses to drought, it is also necessary to calculate canopy
conductance. Canopy conductance can be estimated from observed latent heat flux provided that the
canopy is dry and well coupled to the atmosphere, and assuming furthermore that evaporation from
bare soil in an ecosystem with a closed canopy is minimal. Latent heat flux under these conditions can
be taken as equivalent to gross canopy transpiration.
By inverting the McNaughton and Black approach for calculating stand transpiration (McNaughton &
Black, 1973) a proxy for canopy conductance, Gc , can be obtained:
Gc =
LH .ε .λ .γ
ρ .Cp.vpd
eq. 1)
where LH is the observed latent heat flux, ε the coefficient for the conversion of latent heat to its
water flux equivalent, λ is the latent heat of vaporisation of water, γ is the psychometric constant, ρ
is the air density, Cp is the heat capacity of air, vpd is the observed vapour pressure deficit.
Using the estimated canopy conductance, observed rates of net carbon assimilation from the eddycovariance measurements, and atmospheric CO2 concentrations, the canopy average level of leaf
internal carbon dioxide can be calculated using a simple supply and demand algorithm.
7
2.1.2 The response of Conductance and Assimilation to soil water stress
Response of canopy conductance to soil water stress.
Traditional modelling approaches estimate canopy conductance as a function of irradiance, the balance
of the supply and demand of carbon, and vapour pressure deficit. The Ball, Berry & Leuning approach,
applies the following equation for the calculation of canopy conductance Gc :
Gs = Gs 0 + bi
( An − Rd )
⎛
vpd ⎞
(Ca − Γ* ). ⎜1 +
⎟
⎝ GsD 0 ⎠
eq. 2)
where Gs 0 is the residual conductance (or intercept); An is the rate of Assimilation; Rd is the rate of
dark respiration; Ca is the atmospheric concentration of CO2; Γ* is the photosynthetic compensation
point; GsD 0 is a parameter that describes the sensitivity of conductance to vpd ; b is the slope, an
empirical factor. Studies often model canopy conductance in periods of drought by imposing changes
in the slope and/or intercept of equation 2, thus controlling the internal carbon available for
photosynthesis. To evaluate changes in canopy conductance due to soil moisture availability, the
intercept and slope of the Ball & Berry equation (eq. 2) were estimated over a range of soil moisture
availability levels using canopy conductance as calculated in equation 1 and solving eq. 2 for Gs0 and b.
Non conductance related limitations of photosynthesis due to soil water stress:
To evaluate whether or not there are non-stomatal limitations to photosynthesis during drought periods,
the observed net assimilation flux was filtered to extract assimilation rates within a ‘non-limiting’
range of leaf internal CO2 concentrations (220 to 300 ppm ) at different levels of soil water availability.
The filtering criteria also took into account a site-specific range of similar atmospheric conditions (air
temperature, radiation level and vapour pressure deficit).
Thus, changes in assimilation rates could be directly related to changes in soil water content, excluding
any effects of climate or of changes in conductance.
2.3 Techniques used for the analysis of model performance
Two types of model comparisons were made. Firstly, Daily Primary Production, Soil water content,
Ecosystem respiration and Net Ecosystem Exchange modelled by Gotilwa+, compared against field
data gathered by the FLUXNET network at the Puechabon site. Secondly, simulations from Gotilwa+
by applying the stress functions to both stomatal conductance and photosynthetic potential
simultaneously, were compared against Fluxnet data for each site. Statistics outlined below were used
to assess model goodness of fit.
Statistics
Various statistical methods were employed to assess model performance. The models were evaluated
based on r2, the Mean Squared Error, and the Root Mean Squared Error. To this was added the statistic
Model Efficiency, which is a complement to the r2 statistic. The modelling efficiency statistic (MEF) is
interpreted as the proportion of variation explained by the fitted line. This statistic has been extensively
used in hydrology models (Byers et al., 1989; Loague & Green, 1991), but can certainly be used in
biogeochemical models. In a perfect fit, both r2 and the MEF would result in a value equal to one. The
upper bound of MEF is one and the (theoretical) lower bound of MEF is negative infinity. If MEF is
lower than zero, the model-predicted values are worse than the observed mean (Loague & Green,
8
1991). The MEF statistic is more sensitive than the r2 to systematic deviations and can be a good tool
in the assessment of goodness of fit (Mayer & Butler, 1993).
Results
Observed vs. simulated values at Puechabon forest.
Several validation exercises were carried out modellng daily primary production, soil water content,
ecosystem respiration and net ecosystem exchange using Gotilwa+ model. The simulations where
carried out using the structural data of Puechabon Quercus ilex forest and the climate data (2001-2004)
gathered by the FLUXNET network at the Puechabon site.
GPP - Measured Vs Modelled
10
2
GPP Modelled (gC/m /day)
12
8
6
4
y = 0.9775x + 0.2026
r2 = 0.7027
2
0
0
2
4
6
8
10
12
2
GPP Measured (gC/m /day)
Figure 3. Daily Primary Production (10 day smoothed) for years 2002 and 2003, modelled by Gotilwa+,
compared against field data gathered by the FLUXNET network at the Puechabon site.
The results where compared against FLUXNET network field data. The figures 3 and 4 give an
example of the performance of the model. Simulated Gross Primary production (figure 3) compares
well with the FLUXNET measured data as well as Soil Water (figure 4) content does. Ecosystem
respiration and Net ecosystem exchange are slightly overestimated by the model (figure 4). The
analyses of the data show that these overestimation is mainly due to the fact that under water stress
conditions, which are common under Mediterranean climate, non-stomatal limitations to
photosynthesis limit the assimilation rates.
In the second approach, oriented to evaluate whether or not there are non-stomatal limitations to
photosynthesis during drought periods simulations were run at each of the four sites, covering a total of
12 years of half hourly data and four separate species. There was a wide range of inter-annual and
inter-site variability in climatic variables for the four chosen sites. This lead to varying levels of
drought, with a particularly strong drought experienced in 2003 at the European sites.
9
Soil Water Comparison
2
Soil Water Calculated (kg/m )
200
160
120
80
y = 0.8443x + 20.042
2
R = 0.9773
40
0
0
50
100
150
200
2
Soil Water Modelled (kg/m )
Ecosystem Respiration - Modelled Vs Measured
2
Measured Ecosystem (gC/m /day)
10
8
6
4
2
y = 0.7357x + 0.5156
R2 = 0.6386
0
0
2
4
6
8
10
2
Modelled Ecosystem Respiration (gC/m /day)
Net Ecosystem Exchange Comparison
4
2
NEE Measured (gC/m /day)
6
2
0
-2
y = 0.5946x - 0.0775
R2 = 0.2085
-4
-6
-6
-4
-2
0
2
Figure 4. Water in Soil, Ecosystem respiration,
and Net Ecosystem Exchange (bottom) modelled
by Gotilwa+, compared against field data gathered
by the FLUXNET network at the Puechabon site.
Simulated Gross Primary production compares
well with the FLUXNET measured data as well as
Soil Water content does. Ecosystem respiration
and Net ecosystem exchange are slightly
overestimated by the model. The analyses of the
data show that these overestimation is mainly due
to the fact that under water stress conditions,
which are common under Mediterranean climate,
non-stomatal limitations to photosynthesis limit
the assimilation rates.
4
2
NEE Modeled (gC/m /day)
10
Soil Water Content
Soil water content was calculated at each site as described above. Figure 5 shows the evolution of
reconstructed relative soil water content for each simulated site and year. The Blodgett site soil water
content shows little inter-annual variability due to the lack of inter-annual variability in its climate
during the studied period. In comparison, at Puechabon, soil water varies over a large range, with levels
reaching a prolonged low during 2003 due to the notable drought experienced in Europe in that year.
This drought period is also reflected at the Roccarespampani site, with soil water levels in 2003
reaching more than half of those in 2004. These levels correspond well with levels reported in other
studies at these sites (Hoff et al., 2002; Rambal et al., 2003).
Figure 5. Reconstructed relative soil water content for the simulated periods at each of the
studied sites, separated by year.
Stomatal Limitations
Calculated values for the slope (b, Eq.2) and intercept (Residual conductance - G s 0 , Eq.2) of the Ball,
Berry & Leuning model at high soil water content for each site are given in Table 2. It was found that
residual conductance did not change significantly with changes in available soil water in any of the
sites. This is contrary to results previously reported at the Blodgett site (Misson et al., 2004), but in line
with the accepted understanding of the affects of soil water on stomatal conductance.
Table 2. Parameters for the calculation of Stomatal Conductance, and water stress parameters applied to
Stomatal Conductance and Photosynthetic potential for each site.
Site
Stomata - Wfacstoma
Photosynthesis - Wfacphoto
Wfac
Puéchabon
Rocca
Collelongo
Blodgett
Slope
Intercept
smax
9
11.2
10.5
10.5
0.0017
0.0015
0.000025
0.00002
80
95
95
85
Wfac
WFac
s min
q
smax
s min
q
10
10
0
5
0.15
0.22
0.23
0.18
65
70
75
45
15
10
5
5
0.5
0.85
0.3
0.2
WFac
11
The stomatal conductance levels declined gradually with soil water levels due to the closure of the
stomata. A similar response was found at all sites, suggesting that a general model could be applied to
all species.
Non - Stomatal Limitations
Non-stomatal limitations begin to take effect at lower soil water content than do stomatal limitations.
Thus at high soil water contents, and at the onset of drought, stomatal limitations play the largest part
in limiting photosynthesis at otherwise optimal conditions, through the control of the concentration of
internal carbon available for assimilation. Once the severity of the drought reaches a sufficient level,
which was found to be different for each of the studied species, the role of non-stomatal limitations
increases, and in severe drought can eventually come to limit photosynthesis to the same extent as
stomatal limitations, or more (as is the case for the two Quercus species in Puechabon and
Roccarespampani). This dynamic between stomatal and non-stomatal limitations agrees with recent
laboratory results reported (Grassi & Magnani, 2005), and is the first to be reported across various
ecosystems on the canopy scale.
Simulations
The effect of soil water stress on canopy conductance (Wfacstoma) and non-stomatal limitations of
photosynthesis (Wfacphoto) was modelled using a limitation function derived by fitting the following
function separately to the filtered assimilation rates and the slope of the Ball & Berry equation for
conductance, in dependence of soil moisture (Porporato et al., 2002):
⎧
⎪
⎪
Wfac = ⎨
⎪
⎪
⎩
if S (t ) ≥ S max
1,
(eq. 3)
q
⎡ s (t ) − s min ⎤
⎢
⎥ ,
⎣ s max − s min ⎦
if S (t ) < S max
where q is a measure of the non-linearity of the effects of soil moisture deficit on plant conditions,
s max the point at which reductions are first evident, and s min is the soil water level which gives the
maximum effect on the process in question (conductance, or non-stomatal limitations on assimilation).
The parameters of this function have been estimated for each site independently, both for conductance
and non stomatal limitations, to account for potential differences between vegetation types. These
functions are then used within Gotilwa+ to adjust conductance and photosynthetic potential in drought
conditions. Conductance is affected by soil water content in the model by applying Wfacstoma in a
modified version of eq. 2:
Gs = Gs 0 + Wfacstoma ibi
( An − Rd )
⎛
vpd ⎞
(Ca − Γ* ). ⎜ 1 +
⎟
⎝ GsD 0 ⎠
(eq. 4)
Non-stomatal limitations are simulated by applying Wfacphoto to photosynthetic potential as follows:
Vc max = Vc max* Wfac photo
J max = J max*Wfac photo
(eq. 5)
12
Where Vcmax, and Jmax are the maximum rate of RuBP carboxylation, and the maximum rate of electron
transport.
Comparing approaches
Four Simulations were run for each site. The first applied the original process description and
parameters, the second applied the restrictions which were calculated from the Fluxnet data, presented
in Table 2, to photosynthesis, the third to stomatal conductance, and the fourth to both stomatal
conductance and photosynthetic potential.
The simulation of the diurnal courses of carbon and water fluxes during periods of high water
availability was relatively unaffected by the modelling approach chosen, due to the fact that the
approaches only differ in their treatment of responses to water stress. During periods of drought,
responses to water stress were highly dependent on the chosen response description. All sites showed
marked improvements in diurnal simulation accuracy when applying a reduction in both stomatal
conductance capacity and photosynthetic capacity.
The average diurnal cycle for 20 days in both wet and dry periods at the Collelongo site is shown in
figure 6, using each modelling approach. The correlations between each of the simulations and the
Euroflux data correspond closely, due to the effective capturing of the shape of the diurnal cycle. The
different approaches gave marked differences however in the mean average error. Applying the
calculated water stress restrictions presented in Table 2 to both stomatal conductance and
photosynthetic potential lead to a reduction of 61.33% in the mean average error for assimilated carbon
when compared to the original model parameterisation.
Figure 6. 20 average diurnal courses for the modelled (Gotilwa+) and observed net assimilation (An, in
umol/m2/s) and latent heat flux (LHF, in mm/hour) for Collelongo, using 4 different modelling approaches: 1)
Applying factors from Table 2 to both Stomatal conductance and Photosynthetic potential, 2) Applying the
factors to stomatal conductance only. 3) Applying the factors to Photosynthetic potential only, and 4) Applying
the original parameterisations. On the left, well watered conditions. On the right, strongly water stressed periods.
13
Applying the water stress restrictions solely to stomatal control lead to a marked increase of 76.8% in
the mean average error, whilst applying the restrictions to photosynthetic potential decreased the mean
average error by 18.44%. Latent heat fluxes were similarly affected, with the most effective modelling
approach (considering the water use efficiency) being the application of the stress limiting factor to
both the stomatal conductance and photosynthetic potential. This pattern was repeated at each site, with
the optimal modelling strategy being the application of limiting functions to both stomatal control and
photosynthetic potential.
Carbon and water fluxes were accurately modelled in Collelongo. Gotilwa+ compared well, with both
the shape and the magnitude of the diurnal cycle in both dry and wet conditions being accurately
captured.
Simulation at Roccarespampani was made difficult by the existence of an under story and phenology,
but when focussing on Golden days it can be seen that accurately reproduced carbon assimilation.
Drought events at the Blodgett site are less severe than at the other chosen sites. However, although the
seasonal drought does not lead to a drop in assimilation rates and conductance it does prevent the rates
of assimilation from reaching their optimal maximum. This was well captured by Gotilwa+, with water
use efficiency accurately reproduced. A full set of statistics comparing the model against the Fluxnet
data for each site is given in Keenan et al. (in press).
Conclusions
It has recently become more evident that the long accepted classical approach of modelling carbon and
water fluxes in the water stressed Mediterranean environment is not capable of reproducing data
gathered in the field. The implementation of algorithms in which the photosynthetic response to
drought solely depends on stomatal control over photosynthesis invariably overestimates water use
efficiency, and fails to capture both the timing and extent of the photosynthetic response to water
stress. This is not due to an inaccurate description of stomatal conductance using the Ball-Berry
approach, but an incomplete description of the mechanisms controlling photosynthetic response to
water stress.
It has been shown here that other mechanisms play a role in controlling photosynthesis during drought,
and the weight of these controls has been gauged through the analysis of Fluxnet data. The
identification of these mechanisms is outside the scope of this study, but it has been hypothesised that
the possible controls are: 1) Changes in mesophyll conductance or 2) Damage to the photosynthetic
apparatus. Most likely both of these possibilities play a part in the control of photosynthesis, depending
on the degree of water stress encountered. The dynamic of the timing and extent of stomatal and non
stomatal limitations to photosynthesis observed here reflects those reported in other recent studies, with
stomata beginning to close before the non stomatal limitation on photosynthesis is observed. Non
stomatal limitations were seen to be highly species dependent, and shown to become as or more
limiting to photosynthesis as stomatal limitations, depending on the degree of water stress encountered
and the species in question.
With the inclusion of the knowledge learned through the analysis of the Fluxnet data at each site, fluxes
of carbon and water during water stressed periods can now be simulated to as high a degree as fluxes
simulated in well watered conditions. This is only possible through the application of water stress
responses to both stomatal conductance and photosynthetic capacity independently.Gotilwa+ model
compared well against the Fluxnet data. To the best of the author’s knowledge, this is the first time
eco-physiological process based models have been able to achieve such accuracy in ecosystems which
suffer strong seasonal drought cycles and this has greatly improved our ability to model and predict
carbon and water fluxes in Mediterranean forest ecosystems.
14
5. Acknowledgments
This study was funded through GREENCYCLES, the Marie-Curie Biogeochemistry and Climate
Change Research and Training Network (MRTN-CT-2004-512464) supported by the European
Commissions Sixth Framework program for Earth System Science.
Bibliography
Ball, J.T.; Woodrow, I.E.; Berry, J.A.; 1987. A model predicting stomatal conductance and its contribution to the
control of photosynthesis under different environmental conditions. In: Progress in Photosynthesis Research, ed.
J Biggins. M Nijhoff, Dordrecht, Vol. 4, pp 221-224.
Bréda, N.; Huc, R.; Granier, A. & Dreyer, E.; 2006. Temperate forest trees and stands under severe drought: a
review of ecophysiological responses, adaptation processes and long-term consequences. Ann. For. Sci. 63: 625644.
Blyth, E.M.; Dolman, A.J.; Noilhan, J.; 1994. The effect of forest on mesoscale rainfall – An example from
Hapex-Mobilhy. Journal of Applied Meteorology 33 (4): 445-454.
Byers, F.M.; Schelling, G.T.; Greene, L.W.; 1989. Development of growth functions to describe energy density
of growth in beef cattle. In: Proceedings of Energy Metabolism of Farm Animals, ( eds. van-der Honing, Y.,
Close, W.H.) 11. Pudoc. 43, pp. 195–198.
Chaves, M.M.; Pereira, J.S., Maroco, J.; Rodrigues, M.L.; Ricardo, C.P.P.; Osorio, M.L.; Carvalho, I.; Faria, T.;
Pinheiro, C.; 2002. How plants cope with water stress in the field. Photosynthesis and growth. Annals of Botany
89: 907-916.
Collatz, G.J.; Ribas-Carbo, M.; Berry, J.A.; 1992. Coupled Photosynthesis-Stomatal conductance model for
leaves of C4 plants. Australian Journal of Plant Physiology 19 (5): 519-538.
Farquhar, G.D.; Caemmerer, S.V.; Berry, J.A.; 1980. A Biochemical model of Photosynthetic CO2 assimilation
in leaves of C3 species. Planta 149 (1): 78-90.
Field, C.B.; Randerson, J.T.; Malmstrom, C.M.; 1995. Global Net Primary Production – Combining ecology and
remote sensing. Remote Sensing and the Environment 51 (1): 74-88.
Flexas, J. & Medrano, H.; 2002. Drought inhibition of photosynthesis in C3 plants: Stomatal and non-stomatal
limitations revisited. Annals of Botany 89: 183-189.
Friedlingstein, P.; Joel, G.; Field, C.B.; Fung, I.Y.; 1999. Toward an allocation scheme for global terrestrial
carbon models. Global Change Biology 5 (7): 755-770
Gracia C.A.; Tello, E.; Sabate, S.; Bellot, J.; 1999. GOTILWA+: An Integrated Model of Water Dynamics and
Forest Growth. Ecology of Mediterranean Evergreen Oak Forests. In: Ecology of Mediterranean evergreen oak
forests. (ed. Rodà F, Retana J, Bellot J, Gracia CA.), pp. 163-178. Springer, Berlin.
Granier, A.; Reichstein, M.; Bréda, N.; Janssens, I.A.; Falfe, E.; Ciais, P.; Grünwald, T.; Aubinet, M.; Berbigier,
P.; Bernhofer, C.; Buchmann, N.; Facini, O.; Grassi, G.; Heinesch, B.; Ilvesniemi, H.; Keronen, P.; Knohl, A.;
Köstner, B.; Lagergren, F.; Lindroth, A.; Longdoz, B.; Loustau, D.; Mateus, J.; Montagnani, L.; Nys, C.; Moors,
E.; Papale, D.; Peiffer, M.; Pilegaard, K.; Pita, G.; Pumpanen, J.; Rambal, S.; Rebmann, C.; Rodrigues, A.;
Seufert, G.; Tenhunen, J.; Vesala, T.; & Wang, Q.; 2007. Evidence for soil water control on carbon and water
dynamics in European forests during the extremely dry year. Agricultural and Forest Meteorology 143: 123-145.
Grassi, G.; Magnani, F.; 2005. Stomatal, mesophyll conductance and biochemical limitations to photosynthesis as
affected by drought and leaf ontogeny in ash and oak trees. Plant, Cell and Environment 28 (7): 834-849
15
Harley, P.C. & Tenhunen, J.D.; 1991. Modelling the photosynthetic response of C3 leaves to environmental
factors. In: Modeling Crop Photosynthesis – from Biochemistry to Canopy (eds Boot, K.J & Loomis, R.S.), Vol.
19, pp. 17-39. America Society and Crop Science, Madison, WI, USA.
Hoff, C.; Rambal, S. & Joffre, R.; 2002. Simulating carbon and water flows and growth in a Mediterranean
evergreen Quercus ilex coppice using the FOREST-BGC model. Forest Ecology and Management 164: 121-136.
IPCC; 2007. Climate Change 2007 Synthesis Report. A contribution of working groups I,II and III to the fourth
Assessment Report of the Intergovernamental Panel on Climate Change (Watson R.T. and the Core Writing
Team (eds). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
Jones, H.G.; 1985. Partitioning stomatal and non-stomatal limitations to photosynthesis. Plant, Cell and
Environment 8: 95-104.
Jung, M.; Le Maire, G.; Zaehle, S.; Luyssaert, S.; Vetter, M.; Churkina, G.; Ciais, P.; Viovy, N.; Reichstein, M.;
2007. Assessing the ability of three land ecosystem models to simulate gross carbon uptake of forests from boreal
to Mediterranean climate in Europe. Biogeochemical Discussions 4, 1353-1375.
Kramer, K.; Leinonen, I.; Baterlink, H.H.; Berbigier, P.; Borguetti, M.; Bernhofer, Ch.; Cienciala, E.; Dolman,
A.J.; Froer, O.; Gracia, C.A.; Granier, A.; Grünwald, T.; Hari, P.; Jans, W.; Kellomakï, S.; Loustau, D.; Magnani,
F.; Markkanen, T.; Matteucci, G.; Mohren, M.J.; Moors, E.; Nissinen, A.; Peltola, H.; Sabaté, S.; Sanchez, A.;
Sontag, M.; Valentini, R.; Vesala, I.; 2002. Evaluation of six processes-based forest growth models using eddycovariance measurements of CO2 and H2O fluxes at six forest sites in Europe. Global Change Biology 8: 213230.
Krinner, G.; Viovy de Noblet-Ducoudré, N.; Ogée, J.; Polcher, J.; Friendlingstein, P.; Ciais, P.; Sitch, S.; &
Prentice, I.C.; 2005. A dynamic global vegetation model for studies of the coupled atmosphere-biosphere system.
Global Biogeochemical Cycles 19: GB1015, doi: 10.1029/2003GB002199.
Kubiske, M.E. & Adams, M.D.; 1993. Stomatal and non-stomatal limitations of photosynthesis in 19 temperate
tree species on contrasting sites during wet and dry years. Plant, Cell and Environment 16: 1123-1129.
Lawlor, D.W.; 1995. The effects of water deficit on photosynthesis. In: Environment and Plant Metabolism (ed
N. Smirnoff), pp. 129-160. Bios Scientific Publishers, Oxford, UK.
Lawlor, D.W. & Cornic, G.; 2002. Photosynthetic carbon assimilation and associated metabolism in relation to
water deficits in higher plants. Plant, Cell and Environment 25: 275-294
Leuning, R.; Kelliher, F.M.; Depury, D.G.G.; Schulze, E.D.; 1995. Leaf Nitrogen, Photosynthesis, Conductance
and Transpiration – Scaling from leaves to canopies. Plant, Cell and Environment 18 (10): 1183-1200.
Leuning, R.; Montcrieff, J.; 1990. Eddy-Covariance CO2 flux measurements using open-path and closed-path
CO2 analysers – corrections for analyser water vapour sensitivity and damping of fluctuations in air sampling
tubes. Boundary later meteorology 53 (1-2): 63-76.
Loague, K. & Green, R.E.; 1991. Statistical and graphical methods for evaluating solute transport models:
Overview and application. Journal of Contaminant Hydrology 7: 51–73.
Mayer, D.G. & Butler, D.G. 1993. Statistical validation, Ecological Modelling 68: 21–32.
Misson, L.; Panek, J.A.; Goldstein, A.H.; 2004. A comparison of three approaches to modeling leaf gas exchange
in annually drought-stressed ponderosa pine forests. Tree Physiology 24 (5): 529-541
Misson, L.; Tu, K.P.; Boniello, R.A.; Goldstein, A.H.; 2006. Seasonality of photosynthetic parameters in a multispecific and vertically complex forest ecosystem in the Sierra Nevada of California. Tree Physiology 26 (6): 729741
Morales, P.; Sykes, M.T.; Prentice, I.C.; Smith, P.; Smith, B.; Bugmann, H.; Zierl, B.; Friedlingstein, P.; Viovy,
N.; Sabate, S.; Sanchez, A.; Pla, A.; Gracia, C.A.; Sitch, S.; Arneth, A.M Ogee, J.; 2005. Comparing and
16
evaluating process-based ecosystem model predictions of carbon and water fluxes in major European forest
biomes. Global Change Biology 11 (12): 2211-2233.
Mc. Naughton, K.G. & Black, T.A.; 1973. Evapotranspiration from a forest: A micrometeorological study. Water
Resour. Res. 9: 1579-1590.
Ni, B.R. & Pallardy, S.G.; 1992. Stomatal and non-stomatal limitations to net photosynthesis in seedlings of
woody angiosperms. Plant Physiology 99: 1502-1508
Papale, D., & Valentini, A.; 2003. A new assessment of European forests carbon exchanges by eddy fluxes and
artificial neural network spatialization. Global Change Biology 9 (4): 525-535
Porporato, A. & Rodriguez-Iturbe, I.; 2002. Ecohydrology - a challenging multidisciplinary research perspective.
Journal of Hydrological Sciences 47 (5): 811-821.
Rambal, S.; Joffre, R.; Ourcival, J.M.; Cavender-Bares, J. & Rocheteau, A.; 2004. The growth respiration
component in eddy CO2 flux from a Quercus Ilex Mediterranean forest. Global Change Biology 10: 1460-1469.
Rambal S; Ourcival J.M.; Joffre R; Mouillot F; Nouvellon Y.; Reichstein M. &. Rocheteau A; 2003. Drought
controls over conductance and assimilation of a Mediterranean evergreen ecosystem: scaling from leaf to canopy.
Global Change Biology 9: 1813-1824.
Reichstein, M.; Tenhunen, J.; Roupsard, O.; Ourcival, J.M; Rambal, S.; Miglietta, F.; Peressotti, A.; Pecchiari,
M.; Giampiero, T. & Valentini, R.; 2003. Inverse modeling of seasonal drought effects on canopy CO2/H2O
exchange in threee Mediterranean ecosystems. J. Geophys Res, 108(D23), 4726, doi:10.1029/2003JD003430.
Reichstein, M.; Falge, E.; Baldocchi, D.; Papale, D.; Aubinet, M.; Berbigier, P.; Bernhofer, C.; Buchmann, N.;
Gilmanov, T.; Granier, A.; Grunwald, T.; Havrankova, K.; Ilvesmiemi, H.; Janous, D.; Knohl, A.; Laurila, T.;
Lohila, A.; Loustau, D.; Matteucci, G.; Meyers, T.; Miglietta, F.; Ourcival, J.M.; Pumpanen, J.; Rambal, S.;
Rotenberg, E.; Sanz, M.; Tenhunen, J.; Seufert, G.; Vaccari, F.; Vesala, T.; Yakir, D.; Valentini, R.; 2005. On the
separation of net ecosystem exchange into assimilation and ecosystem respiration: review and improved
algorithm. Global Change Biology 11 (9): 1424-1439
Reichstein, M.; Papale, D.; Valentini, R.; Aubinet, M.; Bernhofer, C.; Knohl, A.; Laurila, T.; Lindroth, A.;
Moors, E.; Pilegaard, K.; Seufert, G.; 2007. Determinants of terrestrial ecosystem carbon balance inferred from
European eddy covariance flux sites. Geophysical Research Letters 34, L01402, doi:10.1029/2006GL027880.
Sitch, S.; Smith, B.; Prentice, I.C.; Arneth, A.; Bondeau, A.; Cramer, W.; Kaplan, J.O.; Levis, S.; Lucht, W.;
Sykes, M.T.; Thonicke, K.; Venevsky, S.; 2003. Evaluation of ecosystem dynamics, plant geography and
terrestrial carbon cycling in the LPJ dynamic global vegetation model. Global Change Biology 9 (2): 161-185.
Schimel, D.S.; 1995. Terrestrial ecosystems and the carbon cycle. Global Change Biology 1: 77-91
Tezara, W.V.J.; Mitchell, S.P.; Driscoll, S.P.; Lawlor, D.W.; 1999. Water stress inhibits plant photosynthesis by
decreasing coupling factor and ATP. Nature 401: 914-917.
Tuzet, A.; Perrier, A. & Leuning, R.; 2003. A coupled model of stomatal conductance, photosynthesis and
transpiration. Plant, Cell, and Environment 26: 1097-1116.
Von Caemmerer, S.; Farquhar, G.D.; 1981. Some relationships between the biochemistry of photosynthesis and
gas exchange of leaves. Planta 153: 376-387
Wang, Y.P. & Leuning, R.; 1998. A two-leaf model for canopy conductance, photosynthesis and partitioning of
available energy I: Model description and comparison with a multi-layered model. Agricultural and Forest
Meteorology 19: 89-111.
Wilson, K.B.; Baldocchi, D.D. & Hanson, P.J.; 2000. Quantifying stomatal and non-stomatal limitations to
carbon assimilation resulting from leaf aging and drought in mature deciduous tree species. Tree Physiology 20
(12): 787-797.
17
Wilson, K.B.; Goldstein, A.; Falge, E.; Aubinet, M.; Baldocchi, D.; Berbigier, P.; Bernhofer, C.; Ceulemans, R.;
Dolman, H.; Field, C.; Grelle, A.; Ibrom, A.; Law, B.E.; Kowalski, A.; Meyers, T.; Moncrieff, J.; Monson, R.;
Oechel, W.; Tenhunen, J.; Valentini, R.; Verma, S.; 2002. Energy balance closure at FLUXNET sites.
Agricultural and Forest Meteorology 113: 223 – 243.
Wofsy, S.C.; Goulden, M.L.; Munger, J.W.; Fan, S.M.; Bakwin, P.S.; Daube, B.C.; Bassow, S.L.; Bazzaz, F.A.;
1993. Net Exchange of CO2 in a mid-latitude forest. Science 260 (5112): 1314-1317.
18