You are on page 1of 17

Ecosystems (2010) 13: 949–965

DOI: 10.1007/s10021-010-9365-y
 2010 Springer Science+Business Media, LLC

Modeling Productivity in Mangrove


Forests as Impacted by Effective Soil
Water Availability and Its Sensitivity
to Climate Change Using Biome-BGC
Zhongkui Luo,1,2 Osbert Jianxin Sun,1* Enli Wang,3 Hai Ren,4
and Hualin Xu5

1
MOE Key Laboratory for Silviculture and Conservation, Beijing Forestry University, Beijing 100083, China; 2State Key Laboratory of
Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China; 3CSIRO Land and
Water, Black Mountain, Canberra, Australian Capital Territory 2601, Australia; 4South China Botanical Garden, Chinese Academy of
Sciences, Guangzhou 510650, China; 5Neilingding-Futian National Nature Reserve of Guangdong, Shenzhen 518040, China

ABSTRACT
Ecosystem dynamics and the responses to climate (SW) stress index to account for the impact of the
change in mangrove forests are poorly understood. effective SW availability in the coastal wetland. Our
We applied the biogeochemical process model modified Biome-BGC well predicted the dynamics
Biome-BGC to simulate the dynamics of net pri- of NPP and LAI in the mangrove forests at the study
mary productivity (NPP) and leaf area index (LAI) sites. We found that the six mangrove systems
under the present and future climate conditions in differed in sensitivity to variations in the effective
mangrove forests in Shenzhen, Zhanjiang, and SW availability. At the ecosystem level, however,
Qiongshan across the southern coast of China, and soil salinity alone could not entirely explain the
in three monocultural mangrove stands of two limitation of the effective SW availability on the
native species, Avicennia marina and Kandelia obov- productivity of mangrove forests. Increasing
ata, and one exotic species, Sonneratia apetala, in atmospheric CO2 concentration differentially
Shenzhen. The soil hydrological process of the affected growth of different mangrove species but
model was modified by incorporating a soil water only had a small impact on NPP (<7%); whereas a
doubling of atmospheric CO2 concentration asso-
ciated with a 2C temperature rise would increase
NPP by 14–19% across the three geographically
separate mangrove forests and by 12% to as much
as 68% across the three monocultural mangrove
stands. Our simulation analysis indicates that
Received 25 December 2009; accepted 5 July 2010; temperature change is more important than
published online 7 August 2010
increasing CO2 concentration in affecting produc-
Electronic supplementary material: The online version of this article
(doi:10.1007/s10021-010-9365-y) contains supplementary material, tivity of mangroves at the ecosystem level, and that
which is available to authorized users. different mangrove species differ in sensitivity to
Author Contributions: Zhongkui Luo carried out the field measure- increases in temperature and CO2 concentration.
ments and data analysis, ran simulation, and wrote the manuscript. Os-
bert J. Sun designed the experiment and wrote the manuscript. Enli
Wang contributed to modeling and manuscript writing. Hai Ren and Key words: climate change; effective soil water
Hualin Xu contributed data and helped with field measurements and data availability; leaf area index; mangroves; net pri-
collections.
mary production; salinity; simulation.
*Corresponding author; e-mail: sunjianx@bjfu.edu.cn

949
950 Z. Luo and others

INTRODUCTION cates that the primary production of Rhizophora


mangle Roxb., a neotropical mangrove species, de-
Mangrove forests play a vital role in conserving creases with rising air temperature (Miller 1972).
biodiversity and buffering disturbance in coastal Elevated atmospheric CO2 concentrations in con-
regions. A clear understanding of the factors that junction with increases in air temperature are
regulate the growth of mangroves and knowledge commonly found to enhance C gain of most plants,
of the dynamics of mangrove ecosystems under but the effects differ on different plant types and in
future climate changes are critically important to different habitats. Photosynthesis-related eco-
devising management strategies aimed for conser- physiological attributes, such as leaf conductance to
vation of this unique and threatened biotic system CO2 and water vapor, also change markedly with
(Field 1995; Valiela and others 2001). However, latitude, temperature, precipitation, and fresh wa-
under the prediction that the global mean air ter supply in mangroves (Duke and others 1998;
temperature will be likely to rise by 1.1–6.4C Alongi 2002). A study by Ball and others (1997)
above the current level and the atmospheric CO2 shows that although an elevated CO2 concentra-
concentration may reach between 730 and tion at 700 ppm enhances the growth rate in Rhi-
1020 ppm at the end of this century (IPCC 2007), it zophora apiculata Blume and Rhizophora stylosa
is yet uncertain how mangroves may respond to Griff., the relative responses of the two species vary
these future climate changes (Alongi 2002; Luo and with soil salinity and air humidity, indicating
others 2007). alteration of the inter-specific relationships among
Numerous studies have indicated that increasing mangrove species along salinity–aridity gradients
atmospheric CO2 concentration and rising surface by elevated CO2. A long-term study with R. mangle
air temperature, individually or in combination, indicates that elevated CO2 accelerates growth and
can have significant impacts on ecosystem pro- alters growth attributes (Farnsworth and others
cesses, including vegetation zonation and commu- 1996). Through the various effects on eco-physio-
nity succession, species interactions within and logical attributes of mangrove species, the rising air
among trophic levels, and carbon (C) assimilation temperature and an increase in atmospheric CO2
and turnover, across various upland ecosystems concentration may change the distributional pat-
(Field and others 2007; Hyvönen and others 2007; tern, species composition, and C cycling of man-
Tylianakis and others 2008). Although there is grove ecosystems.
evidence that mangrove ecosystems are function- Because of the technical difficulties in field
ally similar with upland forests (Ellison 2002), the experimentation and lack of long-term data, studies
submerged growth environment and high salinity that address the interactive effects of increasing
in coastal wetlands may significantly differentiate CO2 concentration and rising air temperature on
the two systems in response to future climate the structure and functioning of mangrove forests
changes. In comparison with canopy processes, soil are lacking (Field 1995; Alongi 2002), particularly
processes in mangrove ecosystems may undergo when considering the variability among species and
little alterations by direct impact of changes in over large scales as well as in terms of interactive
atmospheric environment due to the confounding effects with other environmental factors. Process-
effect of poor aeration under submerged or water- based simulation modeling can overcome this lim-
saturated soil conditions, leading consequently to itation as it is an effective tool for analyzing the
differentially regulated physiological processes and complex interactions through integration of the
patterns of water and nutrient utilizations under responses in the ecological and physiological pro-
future climate changes in mangrove ecosystems. cesses to various environmental constraints (Clark
Temperature is known as the dominant factor and Gelfand 2006). Its application can provide in-
limiting the distribution and growth of mangroves sight into the interactions of mangrove ecosystems
(Duke and others 1998; Alongi 2002; Stuart and with changing environmental conditions when
others 2007), and considerable variation can be direct experimental evidence is lacking.
found among mangrove species in their sensitivity A number of process models have been devel-
to changes in temperature (Field 1995). The oped and used for simulation of mangrove forests
majority of mangrove species reach maximal shoot focusing on specific ecological processes; examples
growth when mean air temperature rises to 25C of these types of models include FORMAN (Chen
(Hutchings and Saenger 1987), whereas the maxi- and Twilley 1998), KiWi (Berger and Hildenbrandt
mum rates of photosynthesis generally occur at leaf 2000), and NUMAN (Chen and Twilley 1999).
temperatures around 30C (Ball and others 1988). FORMAN is a gap dynamics model developed by
In southern Florida, USA, a modeling study indi-
Modeling Mangrove Forests Using Biome-BGC 951

Chen and Twilley (1998) for simulation of stand forests may be constrained by lack of appropriate
dynamics in mangrove forests. KiWi is also a stand representations of the eco-physiological responses
dynamic model incorporating population growth, to simulate the spatiotemporal dynamics in man-
competition, and demographic processes (Berger grove ecosystems. Some modifications of the pro-
and Hildenbrandt 2000). NUMAN, a process-based cesses in Biome-BGC are undoubtedly required to
biogeochemical model with a simplified manipu- make it applicable to mangrove forests.
lation of aboveground processes, has been devel- It is known that soil water (SW) availability in
oped for simulation of soil organic matter terms of salinity stress can significantly affect on
decomposition and nutrient cycling (Chen and growth and physiological processes in mangroves
Twilley 1999). For simulations of the effects of (Ball and Pidsley 1995; Kathiresan and Bingham
salinity in coastal wetlands, a model has been 2001). López-Hoffman and others (2007) found
developed for projecting changes in salinity con- that mangroves growing at low soil salinity had
centrations and species dominance in the coastal higher stomatal conductance, leaf-level transpira-
habitats of the Everglades, southern Florida, USA tion, and internal CO2 concentrations than those
(Teh and others 2008). However, for simulation at growing at high salinity, and thus had greater rates
the ecosystem level, incorporation of various of net photosynthesis, growth, and survival. When
above- and belowground ecological processes cap- the same mangrove species grow at different
turing the energy, water, and nutrient dynamics is severities of salinity, their growth forms may show
crucial for achieving a mechanistic understanding distinct features in response to salinity level. For
on the responses of mangroves to changing envi- example, in R. mangle, tall trees (5–10 m) are found
ronmental conditions. The existing simulation in habitats with lower salinity, whereas dwarf trees
models for mangroves are all individual-based and (1 m or less) are common in habitats with high
used for describing simple ecosystem processes, salinity due to an apparent impact of water stress
primarily concerning biotic regulations (Berger and (Hao and others 2009). Therefore, accounting for
others 2008). Although these types of models are the effects of salinity appears to be critical to
useful for simulating specific ecosystem processes accurately model the dynamics of mangrove forests
in mangrove forests for management purposes and/ and the responses to environmental variations
or study of individual processes, they are not useful using the models developed based on ecological
for wider applications in the context of global cli- processes in upland ecosystems.
mate change. In this study, we synthesized data from the lit-
Compared with those for mangrove forests, most erature and field surveys to parameterize the bio-
of the process-based simulation models for upland geochemical model Biome-BGC to simulate the
ecosystems have been developed with more dynamics of mangrove forests in response to cli-
sophisticated structures and descriptions of more mate change. We modified the model to consider
complex ecosystem processes and interactions, the water stress effect of salinity and applied the
benefitting from the availability of more detailed modified model to simulate net primary produc-
information on, and better understanding of, the tivity (NPP) and leaf area index (LAI) in mangrove
upland ecosystem processes. Biome-BGC is a well- forests at three separate locations (that is, Shenz-
established biogeochemical model for simulating hen, Zhanjiang, and Qiongshan) along the south-
the storage and fluxes of water, C, and nitrogen ern coast of China, under the current and predicted
(N), all in a coupled manner and interactively future climate conditions. We also tested the per-
linking the above- and belowground processes in formance of the modified model to simulate NPP
terrestrial ecosystems at a daily time step and var- and LAI of three monocultural mangrove stands in
ious ecological scales (Churkina and others 2003). Shenzhen, including two native species, Avicennia
One of the most common applications of the marina (Forssk.) Vierh. and Kandelia obovata Sheue,
Biome-BGC model is for assessment of the Liu and Yong, and one exotic species, Sonneratia
dynamics of forest productivity and ecosystem C apetala Buch.–Ham. Our objectives were to: (1)
storage in response to disturbances and changes in examine the ability of the modified Biome-BGC to
climate (Law and others 2003, 2004). This makes predict NPP and LAI of mangrove forests, (2) test
the model a relevant tool for studying the responses the hypothesis that effective SW availability, in
of mangrove forests to future climate changes. term of salinity stress, was a major constraint to
However, as mangroves are a special group of productivity in mangrove ecosystems, and (3)
plants that grow in environments normally not predict the responses of mangrove forests of
habitable for most of the terrestrial vascular plants, different localities and community structures to
direct application of Biome-BGC to mangrove climate change.
952 Z. Luo and others

MATERIALS AND METHODS species usually occupy distinct zones along the
intertidal regions from the sea to upland. In
Study Sites and Mangrove Systems Zhanjiang, the composition of mangrove commu-
Three natural mangrove systems were selected for nity and the stand structure is relatively simple;
ecosystem level simulation. They were from natural mangrove forests occur along the coastline
Shenzhen, Zhanjiang (latitude 2130¢N, longitude and extend hundreds of meters seaward from the
10941¢E), and Qiongshan, along the southern coast upland. Aegiceras corniculatum (L.) Blanco., B. gym-
of China (Figure 1). In addition, three monocultural norrhiza, K. obovata, and R. stylosa are the most
mangrove stands in Shenzhen were used for stand dominant species at this site. In Shenzhen, natural
level simulation. Most of the mangrove forests at mangrove forests occur along the coastline and
these locations are in natural reserves and represent extend 50–150 m seaward from the upland. Three
the main mangrove communities across subtropical native species, A. corniculatum, A. marina, and K.
and tropical climate zones in China. The site condi- obovata, dominate the mangrove community, and
tions differ among the three locations: Shenzhen is usually occur as mosaics of monocultural stands
the northern-most site (latitude 2232¢N, longitude with open understory (Luo and others 2010). In
11400¢E) and has the coarsest soils (clay 29%, silt 1993, two exotic mangrove species, S. apetala and
59%, and sand 12%) and intermediate soil salinity Sonneratia caseolaris (L.) Engl., were introduced to
(15&), with an average soil depth at 0.9 m; Qiong- Shenzhen from Hainan Province (originally intro-
shan is the southern-most site (latitude 1951¢N, duced to Hainan from Bangladesh in 1985), China,
longitude 11024¢E) and has the finest soils (clay and cultivated successfully and dispersed rapidly.
61%, silt 30%, and sand 9%) and lowest soil salinity We used the monocultural stands of A. marina,
(11&), with an average soil depth at 1.2 m; while K. obovata, and S. apetala in Shenzhen for stand
Zhanjiang is characterized by the highest soil salinity level simulation of different mangrove species. The
(18&) and intermediate soil texture (clay 37%, silt three types of stands have contrasting characteris-
57%, and sand 6%), with an average soil depth at tics: A. marina is a small tree and forms stands with
0.9 m. The average stand ages are approximately an average age of 68 years and canopy height of
65 years at Shenzhen, approximately 80 years at approximately 4.5 m at the study site; K. obovata is
Zhanjiang, and approximately 70 years at Qiong- a small or medium tree and forms stands with an
shan. average age of 63 years and canopy height of
Of the three locations, Qiongshan has the highest approximately 6 m; S. apetala is a fast-growing tree
mangrove species diversity, where mangrove and forms stands with an average age of 13 years
communities usually associate with other salt-tol- and canopy height of approximately 10 m.
erant species and compose ‘‘mangrove associates’’
(Mao and others 2003; Wang and others 2010). Field Measurements and Literature
The dominant species include Bruguiera gymnorrh-
Survey
iza (Linn.) Savigny, Bruguiera sexangula (Lour.)
Poir., K. obovata, and R. stylosa. Those mangrove For the three monocultural mangrove stands in
Shenzhen, field surveys and measurements were
conducted in 2005 and 2006. Measurements were
made on stand biomass, LAI, and selected leaf
traits.

Stand Biomass
Four 10 9 10 m2 plots were established for each
stand. Tree diameter at breast height (DBH, 1.3 m)
and the average canopy height (H) were measured.
Biomass was then estimated from DBH and H using
species-specific allometric equations (Tam and
others 1995; Weng 1999).

LAI
Hemispherical photographs, taken with a Nikon
Coolpix 4500 camera with a FC-E8 fisheye lens
Figure 1. Location of study sites. at 1 m above the ground surface, were used to
Modeling Mangrove Forests Using Biome-BGC 953

analyze the understory light environment and Leaf-level C and water fluxes are scaled to the
calculate the canopy LAI. True color photographs canopy level based on LAI. A detailed description of
were taken at multiple locations on each plot under the model can be found in Churkina and Running
diffuse light conditions (that is, at dawn and dust or (1998) and Churkina and others (2003).
under overcast sky conditions). These photos were We used the Biome-BGC 5.0 for Excel (http://
analyzed using the imaging software GLA 2.0 www.ntsg.umt.edu/models/bgc/index.php), and
(Frazer and others 1999). Separating the canopy focused on the ability of the model to simulate
from the sky is one of the most critical steps, which annual NPP and LAI of the six mangrove systems
requires a binary transformation usually done by under the present environment and future condi-
setting a manual threshold. We used the software tions based on climate change scenarios.
SideLook 1.1 to calculate automatic thresholds
thereby avoiding subjectivity and improving the Model Modification
accuracy of the analyses (Nobis and Hunziker
Mangroves mainly grow under water-logged con-
2005).
ditions, and thus the quantity of water supply is
generally not an issue without a substantial modi-
Leaf Traits
fication of the habitat. However, salinity-induced
Leaf samples were collected from three randomly water stress could greatly affect water uptake by
selected trees on each plot for measurements of C:N mangrove roots and constrain the water transpor-
ratio and specific leaf area (SLA, m2 [kg C]-1). Total tation from roots to leaves along the water poten-
C and N concentrations were determined on oven- tial gradient, causing leaf water deficits (Hao and
dried leaf samples (70C to constant mass) using others 2009). We therefore modified the water
wet-combustion and Kjeldahl methods (Gallaher stress index (SI) by introducing a parameter
and others 1976), respectively. accounting for the impact of water salinity.
For the mangrove forests at the three different In Biome-BGC, leaf water potential is one of the
locations, we conducted a literature survey to ob- dominant factors relating to SW condition and
tain data required for the Biome-BGC simulation controlling the capacity of leaf C assimilation. In
(see appendix in Supplemental materials). the original model, a SI based on leaf and SW po-
tential is used to regulate maximum stomatal
conductance and then leaf photosynthesis (module
Model Simulations BGC_FUNCTIONS CanopyET):
Description of Biome-BGC LWPc  SWP
fLWP ¼ ; 0  fLWP  1; ð1Þ
Biome-BGC is a biogeochemical model that simu- LWPc  LWPs
lates the storage and fluxes of water, C and N in
where LWPs is leaf water potential when leaf con-
terrestrial ecosystems using a daily time step and at
ductance starts to decline, LWPc is leaf water po-
different ecological scales (Kimball and others
tential when the stomata are closed, and SWP is
1997; Churkina and others 2003). It requires daily
SW potential, which changes with SW content and
weather data as inputs, including radiation, maxi-
texture, and is calculated as (module BGC_FUNC-
mum and minimum temperatures, precipitation,
TIONS SoilPSI):
and daytime vapor pressure deficit (VPD). It also
 
requires information describing the soil properties SWC a
and eco-physiological traits of vegetation. SWP ¼ SWPs  ; ð2Þ
SWCs
In Biome-BGC, NPP is defined as the difference
between gross primary productivity (GPP) and where SWPs is SW potential at saturation water
autotrophic respiration (Ra), where GPP is a func- content (that is, SWCs), both SWPs and SWCs are a
tion of air temperature (T), VPD, SW content, solar function of soil texture (that is, relative distribution
radiation at the top of the canopy (SRAD), CO2 of clay, silt, and sand), SWC is the actual SW con-
concentration, and LAI; Ra is a function of T, leaf N tent, and a is a factor relating to soil texture. The-
content (LN), and root N content (RN). There is oretically, total water potential, w, can be expressed
only one layer in the canopy, but the leaf area is as the sum of several components (Brady and Weil
divided into shaded and sunlit fractions. Photo- 2008):
synthesis is based on the Farquhar equation (Far- w ¼ wg þ wm þ wp þ wo ; ð3Þ
quhar and others 1980). Leaf conductance is
regulated by air temperature, radiation, leaf water where the subscripts g, m, p, and o represent
potential, LN, and atmospheric humidity and CO2. gravitational potential, matric potential, pressure
954 Z. Luo and others

potential, and osmotic potential, respectively. Os- Climate Data


motic potential occurs when solutes are dissolved
Data on the daily maximum, minimum, and aver-
in water. It is linearly related to the solute con-
age temperatures and precipitation were obtained
centration of the soil (that is, soil salinity in this
from the China Meteorological Data Sharing Ser-
study). Equation (2) assumes that SW potential is
vice System (http://cdc.cma.gov.cn; Figure 2).
only regulated by SW content and soil texture, and
Other meteorological variables (for example, radi-
no salinity impact is considered. In the case of
ation and daytime vapor pressure) required by the
mangrove forests, we assume that SW content
Biome-BGC were derived using the microclimate
remains saturated. Thus,
simulator MTCLIM-XL, which provides an estima-
SWC ¼ SWCs : ð4Þ tion of the missing meteorological variables based
on measurements of temperature, precipitation,
SW potential is additionally regulated by soil and geographical and topographical information
salinity (that is, wo). According to Eqs. (3) and (4), such as latitude, longitude, elevation, slope, and
Eq. (2) can be simplified as: aspect (Thornton and Running 1999; Thornton and
 
SWCs a others 2000).
SWP ¼ SWPs  þ wo ¼ SWPs þ wo ; ð5Þ
SWCs
Model Calibration and Validation
where wo is assumed to be linearly related to soil
We used the following formula to set the best-fit SI
salinity (Ss) and given as:
(SIopt, that is, the value when observed variable
wo ¼ b  Ss ; ð6Þ and the modeled variable have the best agree-
ment):
where b is an empirical parameter describing the sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
impact of salinity that regulates SW potential in Pn 2
i¼1 ðxm  xo Þ
mangrove forest soils. Hence, fLWP under the RMSE ¼ ; ð8Þ
n
mangrove systems is calculated as:
where xm is the modeled value (mainly for NPP in
LWPc  ðSWPs  b  Ss Þ
fLWP ¼ : ð7Þ this study), and xo is the observed value (that is,
LWPc  LWPs observed NPP), n is the number of the observed
Here, we refer to fLWP as the SI of SW availability values, RMSE is the square root of the mean of the
on leaf water potential. A SI of 1 indicates that squared prediction errors. We derived the value of
leaf water potential is not limited by SW avail- SIopt when RMSE was at the minimum. In fact,
ability; a SI of 0 indicates that the leaf cannot use SIopt shows the real stress of SW on mangroves in
SW. situ. We also set the threshold value of SI (SItlv, that
is, the tolerance of mangroves to SW stress) when
modeled NPP became 0, that is, GPP = Ra. For the
Model Parameterization mangrove forests at the three different locations
Many years of research on mangroves in China and the three monocultural stands in Shenzhen,
have accumulated a rich amount of information we assumed that each site and species had a specific
on the eco-physiological characteristics of the SIopt and SItlv.
mangrove species (see reviews by Lin 1997; Luo In the model run process, we first conducted a
and others 2007; Chen and others 2009). Eco- spin-up run for each site to get initial C and N state
physiological parameterization of the modified variables, based on the assumption that the eco-
Biome-BGC was made using data collected from system is in equilibrium with the long-term cli-
the literature for mangrove forests at the three mate. For the three monocultural stands in
different locations without differentiating species Shenzhen, we used the initialization results based
(Table 1). Species-specific parameterization was on the spin-up run of the same site. As the target
carried out for the three monocultural mangrove mangrove forests of the simulations are all at cer-
stands in Shenzhen using data both from the lit- tain ages, the tree C and N pools were consequently
erature and the field measurements in this study reinitialized with the minimum initial values pos-
(Table 2). Average values were used if multiple sible to represent a young stand (Churkina and
data sources were found for the same attributes. If others 2003). For all the six target mangrove sys-
the site- and/or species-specific information were tems, initial C and N pools of live stems and roots
not available, the generic values of White and were set to zero. Alternatively, the initial leaf C
others (2000) were used as default (Table 1). pool was set to the minimum possible values so that
Table 1. Values Used in the Eco-physiological Parameterization for Simulation of NPP and LAI in Mangrove Forests in Qiongshan, Zhanjiang, and
Shenzhen, South China, using the Biome-BGC Model
Parameter Unit Value Data source

Annual turnover rates


Annual leaf and fine root turnover fraction y-1 0.75 Lin and others (1985, 1990, 1992); Zheng and others (1995b, c)
Annual live wood turnover fraction y-1 0.7 White and others (2000)
Annual whole-plant mortality fraction y-1 0.03 Lin and others (1990, 1992); Zheng and others (1994, 1995b, c, 1998); Zheng and
Lin (1995); Peng and others (2005)
Allocation ratios
New fine root C to new leaf C DIM 1 White and others (2000)
New stem C to new leaf C DIM 0.6 Lin and others (1985, 1990, 1992, 1995, 1998); Yin and Lin (1993b); Zheng and
Lin (1995); Zheng and others (1995b, c); Liao and others (1999); Wen (1999);
Zan and others (2001b); Han and others (2004)
New live wood C to new total wood C DIM 0.31 Lin and others (1985, 1990, 1992, 1995, 1998); Yin and Lin (1993b); Zheng and
Lin (1995); Zheng and others (1995b, c); Zan and others (2001b); Han and
others (2004)
New coarse root C to new stem C DIM 1.5 Lin and others (1985, 1990, 1992, 1995, 1998); Yin and Lin (1993b); Zheng and
Lin (1995); Zheng and others (1995b, c); Liao and others (1999); Zan and
others (2001b); Han and others (2004)
Current growth to storage growth DIM 0.5 White and others (2000)
C/N ratios
Leaves g C g-1 N 26 This study; Yin and Lin (1993a, c); Zheng and others (1994); Zheng and others
(1995a, b, c)
Leaf litter after retranslocation g C g-1 N 52 Yin and Lin (1993c); Lu and others (1994); Zheng and others (1994, 1995b, c)
Fine roots g C g-1 N 49 Yin and Lin (1993a); Yin and Lin (1993c); Zheng and others (1994, 1995b, c)
Live wood g C g-1 N 49 Yin and Lin (1993a, c); Zheng and others (1994, 1995b, c)
Dead wood g C g-1 N 225 Yin and Lin (1993c); Zheng and others (1994, 1995b, c)
Labile, cellulose and lignin fractions in decomposition pools
Leaf litter labile proportion DIM 0.32 White and others (2000)
Leaf litter cellulose proportion DIM 0.44 White and others (2000)
Leaf litter lignin proportion DIM 0.24 White and others (2000)
Fine root labile proportion DIM 0.34 White and others (2000)
Fine root cellulose proportion DIM 0.44 White and others (2000)
Fine root lignin proportion DIM 0.22 White and others (2000)
Dead wood cellulose proportion DIM 0.76 White and others (2000)
Dead wood lignin proportion DIM 0.24 White and others (2000)
Canopy parameters
Water interception coefficient LAI-1 d-1 0.045 White and others (2000)
Light extinction coefficient DIM 0.65 This study
Modeling Mangrove Forests Using Biome-BGC

All-sided to projected leaf area ratio DIM 2 White and others (2000)
Average SLA (projected area basis) m2 kg-1 C 25 This study; Lin and Wang (2001)
955
956 Z. Luo and others

1996, 1998); Yang and others (2001); Zan and others (2001a); Jing and others
the model can run meaningfully (that is, the min-

Zheng and Lin (1990); Lei and Lin (1995, 1996, 1998); Jing and others (2005)
Zheng and Lin (1990, 1992, 1997); Lu and others (1994); Lei and Lin (1995,
imum leaf biomass that is needed to assimilate C
and maintain the stand growth); and the corre-

Zheng and Lin (1990); Lei and Lin (1996, 1998); Jing and others (2005)
sponding LN pool was set according to the leaf C:N
ratio (Tables 1, 2) . The model was then run to the
number of years to the age of the specific stand
with continuously increasing atmospheric CO2
concentration according to the attached
CO2_Worksheet in the model.
For the ecosystem level simulation, we assumed
that mangrove growth was mainly regulated by soil
salinity (Ss), whereas other eco-physiological
parameters remained the same. We performed a
calibration to derive the b value when the best-fit
SI was attained, that is, when RMSE was at the
minimum, using the dataset from the site in
(2005); Chen and others (2008)

Shenzhen with site-specific soil salinity. Sensitivity


analysis was done at different SI values, and RMSE
was calculated for each SI. Using the calibrated b
White and others (2000)
White and others (2000)

White and others (2000)


White and others (2000)
White and others (2000)
White and others (2000)

value, we then applied the modified model to


mangrove forests in Shenzhen and Qiongshan to
validate the model performance. Sensitivity anal-
ysis was also performed to determine the best-fit SI
Data source

and the threshold value of SI at both sites.


For the stand level simulation, species-specific
eco-physiological parameters (Table 2) were used
with the same soil salinity condition (Ss). The ef-
fects of soil salinity on growth were assumed to be
species-specific. For each monocultural stand, the b
value and the best-fit SI were derived when RMSE
0.00006
0.033

0.006

was at the minimum.


0.01
–0.6
–3.9
Value

Model outputs of NPP and LAI were compared


2500
350
2

with the observed values for each of the six man-


grove systems. Values of observed NPP for man-
Refer to appendix in Supplemental material for a list of references used for data collection.

grove forests at different locations were estimated


m s-1

m s-1
m s-1

as the sum of increases in biomass and litterfall over


Unit

DIM
DIM

MPa
MPa

a given period using data from the literature,


Pa
Pa

ignoring the biomass consumed by herbivores. In


studies where the litterfall data were not reported,
Conductance parameters (projected area basis)

Vapor pressure deficit: complete gs reduction

the average of the known values was applied. The


Leaf water potential: complete gs reduction
Vapor pressure deficit: start of gs reduction

stand-specific values of NPP from field measure-


Leaf water potential: start of gs reduction
Maximum stomatal conductance (gsmax)

ments were used as the observed NPP for the three


monocultural mangrove stands in Shenzhen. We
Ratio of shaded SLA to sunlit SLA

compared the observed accumulative NPP with the


model outputs without explicit description of the
Boundary layer conductance
Fraction of leaf N in Rubisco

site locations for the three geographically separate


mangrove forests. If multiple values of NPP were
Cuticular conductance
continued

reported for the same stand age of a given site, the


average value was calculated. For each of the study
sites, observed LAI values were calculated as the
average of the data collected from the literature,
irrespective of the stand age, and the model output
Parameter
Table 1.

of LAI was the maximum LAI during the develop-


ment of the mangrove forest. For each of the
monocultural stands in Shenzhen, observed LAI
Table 2. Values Used in the Species-Specific Eco-Physiological Parameterization for Simulation of NPP and LAI in Three Monocultural Stands in
Shenzhen, South China, Using the Biome-BGC Model
Parameter (unit) Value (data source, as listed in Appendix)

Avicennia marina Kandelia obovata Sonneratia apetala

Annual turnover rates


Annual leaf and fine root turnover 0.88 (Lin and others 1995 and 1998) 0.9 (Lin and others 1985) 0.65 (Han and others 2004)
fraction (y-1)
Annual whole-plant mortality 0.03 (Lin 1997) 0.05 (Lin and others 1985; Lu and 0.014 (Han and others 2004; Peng and
fraction (y-1) others 1988; Peng and others 2005) others 2005)
Allocation ratios
New fine root C to new leaf C (DIM) 0.2 (Lin and others 1995, 1997, 1998; 0.6 (Lin and others 1985; Zheng and Lin, 2.4 (Zan and others 2001b; Han and
Miao and others 1998) 1995; Miao and others 1998; Zan and others 2004)
others 2001b)
New stem C to new leaf C (DIM) 0.62 (Lin and others 1995, 1997, 1998; 1.1 (Lin and others 1985; Zheng and Lin 4 (Liao and others 1999, Zan and
Miao and others 1998; Wen, 1999) 1995; Liao and others 1999; Miao and others 2001b; Han and others 2004)
others 1998; Wen, 1999; Zan and others
2001b)
New live wood C to new total wood C 0.9 (Lin and others 1995, 1997, 1998; 0.2 (Lin and others 1985; Zheng and Lin 0.09 (Zan and others 2001b; Han and
(DIM) Miao and others 1998; Wen 1999) 1995; Miao and others 1998; Wen 1999; others 2004)
Zan and others 2001b)
New coarse root C to new stem C 1 (Lin and others 1995, 1997, 1998; 0.5 (Lin and others 1985; Zheng and Lin 0.25 (Zan and others 2001b; Liao and
(DIM) Miao and others 1998) 1995; Liao and others 1999; Miao and others 1999; Han and others 2004)
others 1998; Zan and others 2001b)
Canopy and conductance parameters (proportion)
C:N of leaves (g C g-1 N) 14.54 (this study) 25.54 (this study) 15.95 (this study)
Canopy average SLA (projected area 16.53 (this study) 15.75 (this study) 26.68 (this study)
basis; m2 kg-1 C)
Canopy light extinction coefficient 0.64 (this study) 0.58 (this study) 0.65 (this study)
(DIM)
Maximum stomatal conductance 0.006 (Zheng and Lin 1997; Chen and 0.009 (Lei and Lin 1995, 1996, 1998; 0.013 (Jing and others 2005; Chen and
gsmax (m s-1) others 2008) Zan and others 2001a; Chen and others 2008)
others 2008)
Leaf water potential: start of gs -1.1 (Zheng and Lin 1997; Chen and -1.7 (Zheng and Lin 1990; Lei and Lin -0.6 (Chen and others 2008)
reduction (MPa) others 2008) 1996)
Leaf water potential: complete gs -3.2 (Zheng and Lin 1997; Chen and -3.3 (Zheng and Lin 1990; Lei and Lin -3.9 (Chen and others 2008)
reduction (MPa) others 2008) 1996)
Vapor pressure deficit: start of gs 300 (Zheng and Lin 1997) 400 (Zheng and Lin 1990; Lei and Lin 350 (Jing and others 2005)
reduction (Pa) 1995, 1996, 1998)
Modeling Mangrove Forests Using Biome-BGC

Vapor pressure deficit: complete gs 2400 (Zheng and Lin 1997) 2500 (Zheng and Lin 1990; Lei and Lin 2500 (Jing and others 2005)
reduction (Pa) 1995, 1996, 1998)

Refer to appendix in Supplemental material for a list of references used for data collection.
957
958 Z. Luo and others

Model Simulations Under Predicted Future Climate


Change Scenarios
In the latest IPCC (2007) assessment on climate
change, it was reported that the global atmospheric
CO2 concentration was 379 ppm in 2005; the glo-
bal surface warming is likely to be in the range of
2–4.5C following a doubling of CO2 concentration.
Here, we started with a temperature increase of
2C, and assumed further increases in surface
temperature with increasing atmospheric CO2
concentration. A sensitivity test was performed
with temperature increases by 0.5, 1, 1.5, and 2C
and the corresponding CO2 concentration at 427.5,
475, 570, and 760 ppm, respectively. Sensitivity
analysis was also performed at these CO2 concen-
trations without interactions with temperature in-
creases. Relative changes in NPP as a result of
climate change from the present conditions to the
predicted future scenarios were calculated.

RESULTS
Model Performance
Application of the original Biome-BGC overesti-
Figure 2. Mean annual temperature (A) and annual mated NPP (data not shown) with large RSME
rainfall (B) in Qiongshan, Zhanjiang and Shenzhen, (Figures 3, 5). Using the modified Biome-BGC with
southern China. alteration to the soil hydrological process, the
modeled and observed NPP reached the best
agreement when the values of SI and b were 0.25
was measured in the field, and modeled LAI and 205, respectively, at the ecosystem level in
referred to the modeled LAI in the corresponding Shenzhen (Ss = 15&, Figure 3). Using the cali-
stand ages when we carried out this study. brated value 205 for b, SI was estimated to be 0.16

Figure 3. Comparison of modeled NPP and observed NPP at Shenzhen (A), and the square root of the mean of the
squared prediction errors (RMSE) for fitting the observed and the modeled NPP using different SI values (B). Filled circle
(d) shows RMSE when reaching the threshold limit value of SI (SItlv, mangrove cannot survive beyond this value), gray-
filled circle shows RMSE when reaching the best-fit stress index (SIopt), and filled club-shaped symbol (¤) shows the RMSE
under the simulation without incorporating soil water stress, that is, b = 0. See details in the text for the SIopt, SItlv, b, and
RMSE.
Modeling Mangrove Forests Using Biome-BGC 959

Figure 4. Observed and


modeled NPP of
mangrove forests in
Zhanjiang (A) and
Qiongshan (B) under
b = 205.

in Zhanjiang (Ss = 18&) and 0.50 in Qiongshan exhibited a similar trend as observed LAI with site
(Ss = 11&); and values of NPP were underesti- differences in temperature.
mated to some extent by the modified Biome-BGC
in both the locations (Figure 4). Sensitive analysis
indicates that modeled and observed NPP reach the
Sensitivity of NPP to Climate Change
best agreement when the value of SI is 0.31 in Increasing atmospheric CO2 concentration had
Zhanjiang and 1 in Qiongshan (data not shown). negligible effects on NPP of mangrove forests at the
For the monocultural stands of A. marina, K. ecosystem level across the three locations (Fig-
obovata, and S. apetala in Shenzhen, the values of SI ure 7A). The relative change in NPP was less than
were 0.09, 0.60, and 1, with the corresponding b 7% at all CO2 levels for the Shenzhen site, and less
values of 200, 209, and 54, respectively, when the than 2% for the Qiongshan and Zhanjiang sites
modeled and observed NPP reached the best (Figure 7A).
agreement (Figure 5). Although we did not vali- The three monocultural stands in Shenzhen
date the model prediction due to lack of sufficient showed distinct responses to increasing CO2 con-
data, our calibration results showed that the mod- centration (Figure 7C). In K. obovata, the effects of
ified model simulated NPP reasonably well for the increasing CO2 concentration on NPP were negli-
three monocultural stands in Shenzhen when the gible; whereas in A. marina, NPP increased by 58%
best-fit SI was attained (Figure 5A, C, E). at the CO2 level of 427.5 ppm, followed by a steady
The sensitivity of NPP to changes in SI differed and slight increase until reaching a CO2 level of
among the mangrove forests at different locations 760 ppm; the stand of the exotic species, S. apetala,
as well as among the three monocultural stands in showed a gradual increase of NPP with increasing
Shenzhen (Figures 3, 5B, D, F). The threshold of SI CO2 concentration, and increased by as much as
(that is, SItlv), an indicator of the maximum toler- 35% at a CO2 level of 760 ppm.
ance of mangroves to water stress, varied from 0.14 When coupling the increasing CO2 concentration
in Zhanjiang to 0.17 in Qiongshan, and to 0.22 in with a rise in air temperature, NPP markedly in-
Shenzhen (Figure 3, the data for sensitivity analy- creased across the three locations and in the three
sis were not shown for Zhanjiang and Qiongshan), monocultural stands (Figure 7B, D). For all the six
and were 0.005, 0.10, and 0.065 for the stands of A. mangrove systems, NPP showed a logarithmic re-
marina, K. obovata, and S. apetala, respectively, in sponse to increasing CO2 concentration in associa-
Shenzhen (Figure 5B, D, F). tion with an increase in temperature. The relative
In the mangrove forests at the ecosystem level in changes of NPP with different climate change sce-
the three different locations, modeled maximum narios were comparable among the three locations,
LAI in the time interval of the corresponding stand and increased by 14% in Qiongshan, 19% in
age were slightly overestimated in comparison with Zhanjiang, and 18% in Shenzhen, at the CO2 level
the average of observed LAI (Figure 6); whereas in of 760 ppm in association with a temperature in-
the three monocultural stands, modeled LAI in crease of 2C (Figure 7B). On the contrary, there
2005 were underestimated in comparison with the were large differences among the three monocul-
observed value (Figure 6). Overall, modeled LAI tural stands in Shenzhen; the model simulations
960 Z. Luo and others

Figure 5. Left panels


observed and modeled
NPP of three
monocultural mangrove
stands in Shenzhen,
southern China. Dotted
lines are the 1:1 line. Right
panels the square root of
the mean of the squared
prediction errors (RMSE)
for fitting the observed
and modeled NPP using
different b values in the
calibration process. Filled
circle (d) shows RMSE
when reaching the
threshold limit value of SI
(SItlv, mangrove cannot
survive beyond this
value), gray-filled circle
shows RMSE when
reaching the best-fit stress
index (SIopt), and filled
club-shaped symbol (¤)
shows the RMSE under
the simulation without
incorporating soil water
stress, that is, b = 0. See
details in the text for the
SIopt, SItlv, b, and RMSE.

indicate an increase in NPP by 68% in A. marina, present climate conditions and the predicted future
12% in K. obovata, and 31% in S. apetala, at the CO2 climate change scenarios at the stand and ecosys-
level of 760 ppm in association with a temperature tem levels. We considered the effective SW avail-
increase of 2C (Figure 7D). In the stand of the ability as the primary factor differentiating the
exotic species, S. apetala, a concurrent increase in mangrove forests and the upland ecosystems. After
both CO2 concentration and temperature increased modifying the SW process to account for salinity-
NPP by 10% less than did an increase in CO2 induced SW stress, the Biome-BGC generally well
concentration alone (Figure 7C, D). predicted the dynamics of NPP of the six mangrove
systems in our study. Our simulation results indi-
cate that different mangrove systems and species
DISCUSSION
could well differ in response to the variability in the
Using a modified biogeochemistry model, Biome- effective SW availability climate conditions.
BGC, NPP and LAI of mangrove forests along the Our simulation was based on the assumption
southern coast of China were analyzed under the that mangrove ecosystems follow similar rules as
Modeling Mangrove Forests Using Biome-BGC 961

Figure 6. Observed and modeled maximum LAI in


mangrove forests in Qiongshan, Zhanjiang, and Shenz-
hen, southern China, and three monocultural stands in
Shenzhen. All modeled LAI values were the maximum
LAI when the square root of the mean of the squared
prediction errors (RMSE) was at a minimum in fitting the
observed and modeled NPP. Vertical bars show the stan-
dard deviation.

upland ecosystems (Ellison 2002) except for the


SW process. The good agreement between the
Figure 7. Relative changes of NPP in mangrove forests in
modeled and the observed NPP using our modified
Qiongshan, Zhanjiang, and Shenzhen, southern China,
Biome-BGC shows that ecosystem-process models
and in three monocultural stands in Shenzhen, with
developed in upland systems could perform well in different climate change scenarios. All simulations were
simulating the productivity of mangroves by performed when the square root of the mean of the
introducing a SW stress factor into the model. In squared prediction errors (RMSE) was at a minimum in
Biome-BGC, LAI is a principle variable controlling fitting the observed and the modeled NPP under the
biomass growth and canopy development (Churk- normal run (on temperature and CO2 concentration in-
ina and Running 1998; Churkina and others 2003). crease).
Optimization of the b value, which is a factor
describing the impact of salinity, resulted in good tivity of mangrove forests did not appear to be af-
predictions of LAI in all the six mangrove systems. fected by effective SW availability (SIopt = 1), likely
This generally agrees with Ellison (2002) that because of the relatively low salinity (11&). This is
‘‘ecological theory developed in upland forests can consistent with results from many other studies
be readily applied to mangrove forests,’’ but mod- showing that a low level of soil salinity only has
ification of the key processes that differentiate slight effects on mangrove plant performance (Ball
mangroves and upland ecosystems appears to be 1988; Zheng and Lin 1992; Ball and Pidsley 1995;
necessary for this technological extension. Suárez and Medina 2005). In comparison, the va-
There is sufficient evidence that salinity is one of lue of SIopt was 0.31 in Zhanjiang and 0.25 in
the main constraints to the growth and physiolog- Shenzhen, where the soil salinity was relatively
ical processes in mangroves (Ball and others 1997; high (that is, 18& in Zhanjiang and 15& in
Chen and Twilley 1998; Suárez and Sobrado 2000; Shenzhen). These results further demonstrate that
López-Hoffman and others 2007; Hao and others the modified model could well capture the stress of
2009). In our simulation, we used two indexes, SW availability on the growth of mangroves.
SIopt and SItlv, to reflect the real stress of SW and Sensitivity analysis of SItlv suggests that different
the tolerance of mangroves to SW stress. The sim- mangrove systems and species can differ in toler-
ulation results indicate that mangroves can be ance to SW stress. A number of field studies that
highly susceptible to effective SW availability, compared the performance of different species
which varies with environmental conditions and under the same soil salinity gradients showed
species. For instance, in Qiongshan, the produc- similar patterns (for example, Ball and others 1988;
962 Z. Luo and others

Ye and others 2005). In the monocultural K. obov- responses of mangroves to soil salinity. The stand
ata and A. marina stands in Shenzhen, for example, level simulation of the three mangrove species
the values of SItlv were 0.10 and 0.005, respec- highlights the importance of species level response
tively, indicating possibly the greater tolerance of to soil salinity. Lastly, because of a lack of site- and
A. marina to salinity-induced SW stress than species-specific data, we used the same set of values
K. obovata. Controlled pot experiments show that for C allocation across the three locations, possibly
K. obovata (also known as K. candel, see Sheue and contributing to uncertainties in the modeled NPP
others 2003) is more sensitive to soil salinity than by not accounting for the impact of changes in C
A. marina in the activity of ATPase, the function of allocation with species and site conditions. It has
the plasma membranes, membrane potential, and been found that C allocation among plant parts
transmembrane proton gradient (Zhao and others varies with soil salinity (Rasse and others 2005;
2004). Further evidence of the greater sensitivity to Cherry and others 2009).
salinity in K. obovata when compared to A. marina is Our modeling results demonstrate that increas-
provided by a study using the d13C technique (Wei ing atmospheric CO2 concentration only has slight
and others 2008). Those experimental results positive effects on the net C gain of mangrove
indirectly verify our modeling predictions of K. ecosystems. This differs with results obtained for
obovata and A. marina in response to salinity. upland ecosystems in which elevated atmospheric
However, future studies directly comparing the CO2 concentration shows marked effects on C
response of growth performance (for example, assimilation (Churkina and others 2003; Shaw and
biomass and growth rate) of different mangrove others 2002; Körner and others 2005). As different
species to SW stress gradients may help to validate mangrove species can have distinct responses to
our postulations based on the modeling study. increasing atmospheric CO2 concentration, that is,
We found that soil salinity (Ss) alone could not from neutral to a large increase, an increase in CO2
entirely explain the response of mangrove growth concentration under future climate conditions
to the effective SW availability. Simulations using might have moderate stimulating effects on the
the same b value that was calibrated at the productivity of mangroves at the ecosystem level.
Shenzhen site overestimated NPP at the other two Furthermore, when the ambient CO2 concentra-
locations. This might be attributable to several tion is sufficient for photosynthesis, C assimilation
reasons. First, we assumed that SW potential was could be limited by other factors such as LN con-
linearly related to soil salinity. As mangroves are centration, water use efficiency, enzyme activity,
relatively tolerant to salinity, this assumption and stomatal conductance. The contrasting effects
might mask some information on the growth re- between elevated CO2 concentration and adverse
sponses of mangroves at either low (that is, in environmental conditions such as salinity stress
Qiongshan with Ss value of 11&) or high soil might cause the responses of mangrove plants to
salinity (that is, in Zhanjiang with Ss value of 18&), climate change to significantly differ from that of
which emphasizes the need for more experimental upland ecosystems. It has been previously found
data on the relationship between soil salinity and that elevated CO2 concentration has little effect on
the effective SW availability for mangroves at dif- mangrove growth rate in the presence of high
ferent scales. Secondly, soil salinity may not be the salinity stress (Ball and others 1997).
only factor that determines growth responses of In this study, we found that the impact of
mangroves to effective SW availability. Other soil increasing CO2 concentration on mangrove forests
properties, such as temperature, nutrients, aera- differs with locations; the greater stimulating effects
tion, and chemical composition of the soil solution, on NPP were demonstrated for the Zhangjiang and
can interplay with soil salinity, confounding the Shenzhen sites, where mangroves suffer greater
effects of soil salinity on mangrove growth. Both N soil salinity stress (SI), than at the Qiongshan site
and phosphorus have been suggested to limit with the lowest soil salinity. This is consistent with
mangrove growth (McKee and others 2002; Feller the findings that enhancement of NPP may in-
and others 2003a, b). In our simulation study we crease with water stress (Kimball and others 1993;
did not consider the impacts of soil nutrient avail- Wullschleger and others 2002; Cherry and others
ability and balances, which may account for some 2009), possibly resulting from increased water use
biases in our model outputs. Thirdly, we used the efficiency when plants are exposed to elevated CO2
common generic average values of parameters for concentration in a water-limiting growth environ-
the three ecosystem level simulations, neglecting ment (Erickson and others 2007; Robredo and
the possible influences of variations in species others 2007). In a brackish marsh community,
composition and site conditions on the growth Cherry and others (2009) found that elevated
Modeling Mangrove Forests Using Biome-BGC 963

CO2 stimulated biomass production of C3 plants as which might contribute to distinct responses to
a result of amelioration of negative effects of soil rising temperatures and increasing atmospheric
salinity stress. These results, together with our CO2 concentrations in the future climate.
findings, indicate that the interactions between Our simulation study demonstrates that Biome-
environmental stresses (that is, water availability in BGC can be modified to predict the productivity
this study) and species-specific responses to and growth of mangroves by incorporating a SW
increasing CO2 concentration can greatly compli- SI. Future climate change could change the per-
cate the prediction of the dynamics in wetland formance of mangrove forests depending on local
plants and ecosystem processes with future climate environmental conditions and mangrove species.
change (Rasse and others 2005). When considering This highlights the need to consider both species
a concurrent increase in both atmospheric CO2 and environmental conditions when devising con-
concentration and temperature, we found that the servation and management strategies. Application
net C assimilation capability greatly increased in all of ecological models can be a useful and valuable
the six mangrove systems. In general, the optimal tool for understanding and predicting the dynamics
temperature for photosynthesis is about 30C for of mangrove forests under different environmental
plants in tropical forests, a reduction in the rates of conditions across different scales if the underlying
photosynthesis may occur at temperatures above or processes that differentiate mangroves with upland
below this level (Lloyd and Farquhar 2008). All our ecosystems are better understood. It needs to be
study sites had average air temperatures far below noted, however, our simulation was based on the
the optimal level for photosynthesis. An increase in assumption that the growth of mangrove plants
temperature may facilitate a greater rate of photo- responds linearly to soil salinity, which downplays
synthesis in mangroves of subtropical China, and the possible impacts of a greater range of soil
hence a greater requirement for atmospheric CO2 salinity at both the stand and the ecosystem levels.
as an explanation for the positive interaction be- Moreover, the interaction between soil salinity and
tween temperature rise and increasing atmospheric soil nutrient availability (that is, N and phospho-
CO2 concentration on NPP. rus) can have significant effects on mangrove
Results in this study showed that the exotic growth (Ball and others 1997; Feller and others
mangrove species S. apetala had much higher LAI 2003a, b; Rasse and others 2005; Cherry and others
and annual NPP than the other two native species 2009), which needs to be explicitly addressed be-
in Shenzhen. Mangroves play a key role in main- fore a confident prediction can be made on the
taining and shaping the unique habitat as a nursery dynamics of mangroves in response to changes in
for biodiversity. The higher productivity of the climate and other environmental constraints. More
exotic mangrove species could profoundly change sophisticated application of modeling approaches
the stand structure and energy transformation depends on our understanding of these topics and
among the ecosystem components. This potential the ability of models to capture these phenomena.
alteration of ecosystem structure might break the
original balance of trophic cascades (Tilman and ACKNOWLEDGEMENTS
others 1997) and nutrient cycling (Ehrenfeld 2003;
This study was jointly funded by the Chinese
Allison and Vitousek 2004) in native forests.
Academy of Sciences (grant KSCX2-SW-132), the
Contrary to our findings, Chen and others (2008)
State Forestry Administration of China (2008
found that S. apetala had fewer competitive
04001), and the Ministry of Science and Technol-
advantages and lower C assimilation capacities
ogy of China (2008BADB0B0302). Financial sup-
than other native mangrove species. In our simu-
port from a joint PhD program to Zhongkui Luo
lation study, a concurrent increase in both atmo-
under the CSIRO-MOE (Ministry of Education,
spheric CO2 concentration and temperature had
China) Scientific Exchange Agreement is also
less effect in promoting the growth of S. apetala
greatly acknowledged. We thank Luzhen Chen and
than increasing atmospheric CO2 concentration
Xueqing Zeng for field assistance, Charlie Chen for
alone. We do not yet have a logical explanation for
help with the graphic drawing, and Futian Man-
this phenomenon. Rather, the contradictory results
grove Natural Reserve for site access permission
stress the need for longer term studies to better
and technical support.
understand this species’ invasive potential as well
as C assimilation process. S. apetala has different
physiological characteristics from the native man- REFERENCES
grove species, such as a lower gas exchange rate Allison SD, Vitousek PM. 2004. Rapid nutrient cycling in leaf
and water use efficiency (Chen and others 2008), litter from invasive plants in Hawai¢i. Oecologia 141:612–19.
964 Z. Luo and others

Alongi DM. 2002. Present state and future of the world’s man- Erickson JE, Megonigal JP, Peresta G, Drake BG. 2007. Salinity
grove forests. Environ Conserv 29:331–49. and sea level mediate elevated CO2 effects on C3–C4 plant
Ball MC. 1988. Salinity tolerance in the mangroves Aegiceras interactions and tissue nitrogen in a Chesapeake Bay tidal
corniculatum and Avicennia marina. I. Water use in relation to wetland. Glob Change Biol 13:202–15.
growth, carbon partitioning and salt balance. Aust J Plant Farnsworth EJ, Ellison AM, Gong WK. 1996. Elevated CO2 alters
Physiol 15:447–64. anatomy, physiology, growth, and reproduction of red man-
Ball MC, Cochrane MJ, Rawson HM. 1997. Growth and water grove (Rhizophora mangle L.). Oecologia 108:599–609.
use of the mangroves Rhizophora apiculata and R. stylosa in Farquhar GD, von Caemmerer S, Berry JA. 1980. A biochemical
response to salinity and humidity under ambient and elevated model of photosynthetic CO2 assimilation in leaves of C3
concentrations of atmospheric CO2. Plant Cell Environ species. Planta 149:78–90.
20:1158–66. Feller IC, McKee KL, Whigham DF, O’Neill JP. 2003a. Nitrogen
Ball MC, Pidsley SM. 1995. Growth responses to salinity in vs. phosphorus limitation across an ecotonal gradient in a
relation to distribution of two mangrove species, Sonneratia mangrove forest. Biogeochemistry 62:145–75.
alba and S. lanceolata, in northern Australia. Funct Ecol Feller IC, Whigham DF, McKee KL, Lovelock CE. 2003b.
9:77–85. Nitrogen limitation of growth and nutrient dynamics in a
Berger U, Hildenbrandt H. 2000. A new approach to spatially disturbed mangrove forest, Indian River Lagoon, Florida.
explicit modeling of forest dynamics: spacing, aging and Oecologia 134:405–14.
neighborhood competition of mangrove trees. Ecol Model Field CB, Lobell DB, Peters HA, Chiariello NR. 2007. Feedbacks
132:287–302. of terrestrial ecosystems to climate change. Annu Rev Environ
Berger U, Rivera-Monroy VH, Doyle TW, Dahdouh-Guebas F, Resour 32:1–29.
Duke NC, Fontalvo-Herazo ML, Hildenbrandt H, Koedam N, Field CD. 1995. Impact of expected climate change on man-
Mehlig U, Piou C, Twilley RR. 2008. Advances and limitations groves. Hydrobiologia 295:75–81.
of individual-based models to analyze and predict dynamics of Frazer GW, Canham CD, Lertzman KP. 1999. Gap Light Ana-
mangrove forest: a review. Aquat Bot 89:260–74.
lyzer (GLA), Version 2.0: Imaging software to extract canopy
Brady NC, Weil RR. 2008. The nature and properties of soils. structure and gap light transmission indices from true-colour
14th edn. Upper Saddle River, NJ: Pearson Education, Inc. fisheye photographs, users manual and program documenta-
p 975. tion. Simon Fraser University, Burnaby, British Columbia,
Chen L, Tam NFY, Huang J, Zeng X, Meng X, Zhong C, Wong Y, and the Institute of Ecosystem Studies, Millbrook, New York.
Lin G. 2008. Comparison of ecophysiological characteristics Gallaher RN, Weldon CO, Boswell FC. 1976. A semiautomated
between introduced and indigenous mangrove species in procedure for total nitrogen in plant and soil samples. Soil Sci
China. Estuar Coast Shelf Sci 79:644–52. Soc Am J 40:887–9.
Chen L, Wang W, Zhang Y, Lin G. 2009. Recent progresses in Hao G, Jones TJ, Luton C, Zhang Y, Manzane E, Scholz FG, Bucci
mangrove conservation, restoration and research in China. SJ, Cao K, Goldstein G. 2009. Hydraulic redistribution in
J Plant Ecol 2:45–54. dwarf Rhizophora mangle trees driven by interstitial soil water
Chen R, Twilley RR. 1998. A gap dynamic model of mangrove salinity gradients: impacts on hydraulic architecture and gas
forest development along gradients of soil salinity and nutri- exchange. Tree Physiol 29:697–705.
ent resources. J Ecol 86:37–51. Hutchings P, Saenger P. 1987. Ecology of mangroves. St Lucia:
Chen R, Twilley RR. 1999. A simulation model of organic matter University of Queensland Press. p 388.
and nutrient accumulation in mangrove wetland soils. Bio- Hyvönen R, Agren GI, Linder S, Persson T, Cotrufo MF, Ekblad
geochemistry 44:93–118. A, Freeman M, Grelle A, Janssens IA, Jarvis PG, Kellomaki S,
Cherry JA, McKee KL, Grace JB. 2009. Elevated CO2 enhances Lindroth A, Loustau D, Lundmark T, Norby RJ, Oren R,
biological contributions to elevation change in coastal wet- Pilegaard K, Ryan MG, Sigurdsson BD, Stromgren M, van
lands by offsetting stressors associated with sea-level rise. Oijen M, Wallin G. 2007. The likely impact of elevated [CO2],
J Ecol 97:67–77. nitrogen deposition, increased temperature and management
Churkina G, Running SW. 1998. Contrasting climate controls on on carbon sequestration in temperate and boreal forest eco-
the estimated productivity of global terrestrial biomes. Eco- systems: a literature review. New Phytol 173:463–80.
systems 1:206–15. IPCC. 2007. In: Solomon S, Qin D, Manning M, Chen Z, Marquis
Churkina G, Tenhunen J, Thornton P, Falge EM, Elbers JA, M, Averyt KB, Tignor M, Miller HL, Eds. Climate Change
Erhard M, Grünwald T, Kowalski AS, Rannik Ü, Sprinz D. 2007: The Physical Science Basis. Contribution of Working
2003. Analyzing the ecosystem carbon dynamics of four Group I to the Fourth Assessment Report of the Intergov-
European coniferous forests using a biogeochemistry model. ernmental Panel on Climate Change. Cambridge University
Ecosystems 6:168–84. Press, Cambridge, UK and New York, NY, USA.
Clark JS, Gelfand AE. 2006. A future for models and data in Kathiresan K, Bingham BJ. 2001. Biology of mangroves and
environmental science. Trends Ecol Evol 21:375–80. mangrove ecosystems. Adv Mar Biol 40:81–251.
Duke NC, Ball MC, Ellison JC. 1998. Factors influencing biodi- Kimball JS, Mauney JR, Nakayama FS, Idso SB. 1993. Effects of
versity and distributional gradients in mangroves. Global Ecol increasing CO2 on vegetation. Vegetatio 104(105):65–75.
Biogeogr Lett 7:27–47. Kimball JS, Thornton PE, White MA, Running SR. 1997. Sim-
Ehrenfeld JG. 2003. Effects of exotic plant invasions on soil ulating forest productivity and surface-atmosphere carbon
nutrient cycling processes. Ecosystems 6:503–23. exchange in the BOREAS study region. Tree Physiol
Ellison AM. 2002. Macroecology of mangroves: large-scale pat- 17:589–99.
terns and processes in tropical coastal forests. Trees-Struct Körner C, Asshoff R, Bignucolo O, Hättenschwiler S, Keel SG,
Funct 16:1194–981. Peláez-Riedl S, Pepin S, Siegwolf RTW, Zotz G. 2005. Carbon
Modeling Mangrove Forests Using Biome-BGC 965

flux and growth in mature deciduous forest trees exposed to Suárez N, Sobrado MA. 2000. Adjustments in leaf water rela-
elevated CO2. Science 309:1360–2. tions of mangrove (Avicennia germinans) seedlings grown in a
Law BE, Sun OJ, Campbell J, Van Tuyl S, Thornton P. 2003. salinity gradient. Tree Physiol 20:277–82.
Changes in carbon storage and fluxes in a chronosequence of Tam NFY, Wong YS, Lan CY, Chen GZ. 1995. Community
ponderosa pine. Glob Change Biol 9:510–24. structure and standing crop biomass of a mangrove forest in
Law BE, Turner D, Campbell J, Sun OJ, Lefsky M, Guzy M, Van Futian Natural Reserve, Shenzhen, China. Hydrobiologia
Tuyl S. 2004. Disturbance and climate effects on carbon stocks 295:193–201.
and fluxes across the forested region of Oregon USA. Glob Teh SY, DeAngelis DL, Sternberg LdSL, Miralles-Wilhelm FR,
Change Biol 10:1429–44. Smith TJ, Koh HL. 2008. A simulation model for projecting
Lin P. 1997. Mangrove ecosystem in China. Beijing: Science changes in salinity concentrations and species dominance in
Press. p 271. the coastal margin habitats of the Everglades. Ecol Model
Lloyd J, Farquhar GD. 2008. Effects of rising temperature and 213:245–56.
[CO2] on the physiology of tropical forest trees. Philos Trans R Thornton PE, Running SW. 1999. An improved algorithm for
Soc B 363:1811–17. estimating incident daily solar radiation from measurements
of temperature, humidity, and precipitation. Agric For Mete-
López-Hoffman L, Anten NPR, Martı́nez-Ramos M, Ackerly D.
2007. Salinity and light interactively affect neotropical man- orol 93:211–28.
grove seedlings at the leaf and whole plant levels. Oecologia Thornton PE, Hasenauer H, White MA. 2000. Simultaneous
150:545–56. estimation of daily solar radiation and humidity from ob-
Luo ZK, Huang JH, Sun JX. 2007. On ecological functions and served temperature and precipitation: an application over
complex terrain in Austria. Agric For Meteorol 104:255–71.
conservation of mangrove forest. J Subtrop Resour Environ
2:37–47 (in Chinese with English abstract). Tilman D, Knops J, Wedin D, Reich P, Ritchie M, Siemann E.
Luo ZK, Sun OJ, Xu HL. 2010. A comparison of species com- 1997. The influence of functional diversity and composition
position and stand structure between planted and natural on ecosystem processes. Science 277:1300–2.
mangrove forests in Shenzhen Bay, South China. J Plant Ecol. Tylianakis JM, Didham RK, Bascompte J, Wardle DA. 2008.
doi:10.1093/jpe/rtq004. Global change and species interactions in terrestrial ecosys-
Mao L, Zhang Y, Bi H. 2003. Mangrove vegetation changes since tems. Ecol Lett 11:1351–63.
mid-Holocene in a coastal swamp of northern Hainan Island, Valiela I, Brown JL, York JK. 2001. Mangrove forests: one of the
China. J Geosci Osaka City Univ 46:135–55. world’s threatened major tropical environments. Bioscience
51:807–15.
McKee KL, Feller IC, Popp M, Wanek W. 2002. Mangrove iso-
topic (d15N and d13C) fractionation across a nitrogen vs. Wang LM, Mu MR, Li XF, Lin P, Wang WQ. 2010. Differentia-
phosphorus limitation gradient. Ecology 83:1065–75. tion between true mangroves and mangrove associates based
Miller PC. 1972. Bioclimate, leaf temperature, and primary on leaf traits and salt contents. J Plant Ecol. doi:10.1093/jpe/
rtq008.
production in red mangrove canopies in South Florida. Ecol-
ogy 53:22–45. Wei LL, Yan CL, Ye BB, Guo XY. 2008. Effects of salinity on leaf
Nobis M, Hunziker U. 2005. Automatic thresholding for hemi- d13C in three dominant mangrove species along salinity gra-
spherical canopy-photographs based on edge detection. Agric dients in an estuarine wetland, Southeast China. J Coast Res
24:267–72.
For Meteorol 128:243–50.
Weng Y. 1999. Biomass and productivity of five mangrove
Rasse DP, Peresta G, Drake BG. 2005. Seventeen years of ele-
vated CO2 exposure in a Chesapeake Bay Wetland: sustained communities in Yingluo Bay of Guangxi. Guangxi Sci.
but contrasting responses of plant growth and CO2 uptake. 6:142–7 (in Chinese with English abstract).
Glob Change Biol 11:369–77. White MA, Thornton PE, Running SW, Nemani RR. 2000.
Parameterization and sensitivity analysis of the BIOME-BGC
Robredo A, Perez-Lopez U, dela Maza HS, Gonzalez-Moro B,
Lacuesta M, Mena-Petite A, Munoz-Rueda A. 2007. Elevated terrestrial ecosystems model: net primary production controls.
CO2 alleviates the impact of drought on barley improving Earth Interactions 4:1–85.
water status by lowering stomatal conductance and delaying Wullschleger SD, Tschaplinski TJ, Norby RJ. 2002. Plant water
its effects on photosynthesis. Environ Exp Bot 59:252–63. relations at elevated CO2—implications for water-limited
Shaw MR, Zavaleta ES, Chiariello NR, Cleland EE, Mooney HA, environments. Plant Cell Environ 25:318–31.
Field CB. 2002. Grassland responses to global environmental Ye Y, Tam NFY, Lu CY, Wong YS. 2005. Effects of salinity on
changes suppressed by elevated CO2. Science 298:1987–90. germination, seedling growth and physiology of three salt-
secreting mangrove species. Aquat Bot 83:193–205.
Sheue CR, Liu HY, Yong JWH. 2003. Kandelia obovata (Rhizo-
phoraceae), a new mangrove species from eastern Asia. Taxon Zhao ZQ, Zheng HL, Zhu YG. 2004. Changes of plasma membrane,
52:287–94. ATPase activity, membrane potential and transmembrane
Stuart SA, Choat B, Martin KC, Holbrook NM, Ball MC. 2007. proton gradient in Kandelia candel and Avicennia marina seed-
The role of freezing in setting the latitudinal limits of man- lings with various salinities. J Environ Sci-China 16:742–5.
grove forests. New Phytol 173:576–83. Zheng WJ, Lin P. 1992. Effects of salinity on the growth and
Suárez N, Medina E. 2005. Salinity effect on plant growth and some ecophysiological characteristics of mangrove Bruguiera
leaf demography of the mangrove, Avicennia germinans L. sexangula seedlings. Chin J Appl Ecol 3:9–14 (in Chinese with
Trees-Struct Funct 19:722–8. English abstract).

You might also like