Abstract
A lack of explicit information on differential controls on net primary productivity (NPP) across regions and ecosystem types is largely responsible for uncertainties in global trajectories of terrestrial carbon balance with changing environment. The objectives of this study were to determine how NPP of different forest types would respond to inter-annual variability of climate and to examine the responses of NPP to future climate change scenarios across contrasting forest types in northern China.
We investigated inter-annual variations of NPP in relation to climate variability across three forest types in northern China, including a boreal forest dominated by Larix gmelinii Rupr., and two temperate forests dominated by Pinus tabulaeformis Carr. and Quercus wutaishanica Mayr., respectively, and studied the responses of NPP in these forests to predicted changes in climate for the periods 2011–40, 2041–70 and 2070–100 under carbon emission scenarios A2 and B2 of Intergovernmental Panel on Climate Change. We simulated the responses of NPP to predicted changes in future climate as well as inter-annual variability of the present climate with the Biome-BGC version 4.2 based on site- and species-specific parameters. The modeled forest NPP data were validated against values in literature for similar types of forests and compared with inter-annual growth variations reflected by tree-ring width index (RWI) at the study sites.
Inter-annual variations in modeled NPP during the period 1960–06 were mostly consistent with the temporal patterns in RWI. There were contrasting responses of modeled NPP among the three forest types to inter-annual variability of the present climate as well as to predicted changes in future climate. The modeled NPP was positively related to annual mean air temperature in the L. gmelinii forest (P < 0.001), but negatively in the P. tabulaeformis forest (P = 0.05) and the Q. wutaishanica forest (P = 0.03), while the relationships of modeled NPP with annual precipitation for the three forest types were all positive. Multiple stepwise regression analyses showed that temperature was a more important constraint of NPP than precipitation in the L. gmelinii forest, whereas precipitation appeared to be a prominent factor limiting the growth in P. tabulaeformis and Q. wutaishanica. Model simulations suggest marked, but differential increases in NPP across the three forest types with predicted changes in future climate.
INTRODUCTION
With a continuously rising concentration of atmospheric greenhouse gases, an increase in mean annual air temperature of 4–5°C is predicted to occur by the end of the 21st century (Christensen et al. 2007; IPCC 2007). In China, the mean annual air temperature increased at an average rate of 0.6°C/decade over the past two decades (Piao et al. 2011), which was far greater than the global average of ~0.27°C/decade (IPCC 2007). Associated with this warming, significant changes in seasonal precipitation were also observed (Piao et al. 2005). Recent changes in climate have been particularly evident in northern China (Ren et al.2008, 2012).
Forests are considered to play an important role in global carbon balance because they account for up to 80% of the total aboveground carbon and some 40% of belowground carbon in the entire terrestrial ecosystems (Dixon et al. 1994). A persistent warming and changes in climate can lead to unprecedented changes in forest growth and productivity (Myneni et al. 2001; Nemani et al. 2003; Piao et al. 2007; Tan et al. 2007), enhancing or weakening the forest carbon sinks. However, assessment of regional and global forest carbon balance under changing climate can be complicated by differential responses of forest productivity to climate variability across stand types and geographical locations (Law et al. 2004; Lindner et al. 2010; Sun et al. 2004). Therefore, a better understanding on the local controls and constraints of forest response to climate variability is vital when efforts are made for reducing uncertainties in assessment of the regional carbon budget and forest feedbacks to global climate change.
Net primary productivity (NPP) is a central component of ecosystem carbon cycle and a primary indicator of ecosystem functioning. Direct field measurements often provide most accurate estimate of forest NPP at stand level, but the operation is more costly and time consuming with increasing spatial and temporal coverage (Clark et al. 2001; Gower et al. 2001; Houghton 2005). Ground NPP estimates based on forest growth data from permanent sampling plots or forest inventory analysis typically rely on measurements taken at intervals ranging from 5 to 10 years; the resultant increment data provide mean values for specified time periods while neglecting the inter-annual variations in NPP due to climate fluctuations (Churkina et al. 2003; Hasenauer et al. 2012; Houghton 2005). For identification of the climatic controls on temporal dynamics of forest NPP, explicit information on the responses of tree growth and stand dynamics to inter-annual variability of climate and other environmental conditions is critical (Bousquet et al. 2003; Cao et al. 2003; Keenan et al. 2012). Process-based modeling provides a mean of integrating data collected across a spectrum of spatial and temporal scales for studying the regional carbon balance and assessing potential changes in ecosystem carbon dynamics with changing environment (Law et al.2003, 2004).
Process-based models have inherent advantages in predicting how future climate change may influence forest NPP compared with field-based approaches (Melillo et al. 1993; Morales et al. 2005; Scurlock et al. 2002; Wang et al. 2012). Several models have been used to assess forest NPP and responses to climate change in China (Cao et al. 2003; Fang et al. 2003; Gao and Zhang 1997; Jiang et al. 1999; Peng et al. 2009; Wang et al. 2005; Xiao et al. 1998; Zhu et al. 2007). Most of them use meteorological data and remote-sensing products of vegetation as model inputs with minimum parameterization of the models and only a few have examined temporal dynamics of forest NPP on an annual basis (Cao et al. 2003; Fang et al. 2003; Piao et al. 2005). The inter-annual variations and long-term trend of NPP have not been comprehensively assessed or reported for forests in northern China.
Determination of the accuracy and validation of model predictions are important steps in modeling ecosystem carbon cycle and understanding critical drivers of regional NPP under changing environment. Availability of long-term data on tree growth and forest productivity is a prerequisite for model validation (Tao et al. 2005; Turner et al. 2005). Tree-ring chronologies provide long-term records of ad hoc tree growth under natural conditions and, therefore, can be used as indicators of NPP for evaluations of the impacts of climate change and atmospheric CO2 enrichment on forest productivity (Rathgeber et al. 2000, 2003). Tree-ring width index (RWI) has been proven to be an effective indicator of forest NPP because tree radial increment is proportional to annual NPP for a wide variety of forest types (Graumlich et al. 1989; Krakauer and Randerson 2003; Rathgeber et al. 2000, 2003). RWI provides a robust testing of model simulations of inter-annual variations in forest NPP.
In this study, by using the ecosystem process model Biome-BGC (BioGeochemical Cycles) with site- and/or species-specific parameters, we investigated inter-annual variability of NPP under present climate, and simulated responses of NPP to predicted changes of future climate and increasing atmospheric CO2, in three representative forest types (Larix gmelinii Rupr., P. tabulaeformis Carr. and Quercus wutaishanica Mayr. [aka. Q. liaotungensis Koidz.]) across two climate zones in northern China. The objectives of this study were (i) to determine how NPP of different forest types would respond to inter-annual variability of climate and (ii) to examine the differential responses of NPP to future climate change scenarios across the three contrasting forest types in northern China.
MATERIALS AND METHODS
Study sites
The study was located at two forest sites: one in the Genhe Nature Reserve on the northwest slope of Daxing’anling mountains (Genhe site), Inner Mongolia (latitude 50°49ʹ–50°51ʹN; longitude 121°30ʹ–121°31ʹE; elevation 784–1142 m a.s.l.) and the other in the Mt Linkong of the Taiyue mountains (Taiyue site), Shanxi province (latitude 36°31ʹ–43ʹN; longitude 121°01ʹ–121°15ʹE; elevation 1542–1734 m a.s.l.).
The Genhe site is situated in the boreal region of northeastern China. Annual mean air temperature at the site is about –5.4°C and annual precipitation fluctuates between 450 and 550mm, of which 85–90% falls during the growing season (Gower et al. 2001). Soils are of dark brown earth (i.e. Cryumbreps in the US soil classification system or Humiccambisols in the UN–FAO soil classification system) developed on bedrocks of granite and basalt. Continuous or discontinuous permafrost is commonly found in the region, which reaches down to 3 m below the ground surface and lasts for 8 months each year. The dominant forest tree species in the Genhe area is L. gmelinii, often forming communities with Ledum palustre or Rhododendron dahuria as dominant companion shrubs in the understory and thick Sphagnum layer on forest floor.
The Taiyue site has a warm-temperate and continental monsoon climate with an annual mean air temperature of 8°C. Annual precipitation varies from 600 to 650mm, occurring mostly during summer months (Yi et al. 2008). The soil is identified as Cinnamon, which matches the alfisol type in the US soil classification system. The dominant tree species are evergreen needleleaf P. tabuliformis and deciduous broadleaf Q. wutaishanica, which occur widely in northern China (Yu and Sun 2013). Shrubs are represented by Lespedeza bicolor Turcz. and Spiraea salicifolia, occurring mostly in the open or on forest edge.
Establishment of stand chronologies
Tree-ring chronologies were developed for the three forest types by using increment coring method and used for determination of inter-annual variations in tree growth.
Increment cores were collected from previously established plots for the three forest types at the two study sites, including three plots for the L. gmelinii forest at the Genhe site, and four plots for the P. tabulaeformis forest and three plots for the Q. wutaishanica forest, respectively, at the Taiyue site. Plots at the Genhe site, each of the dimension 20×50 m, represented the three most common forest community types of the region, namely L. gmelinii–R. dahuria community, L. gmelinii–L. palustre community and L. gmelinii–L. palustre–Sphagnum community. The plots at the Taiyue site are representatives of pure P. tabulaeformis and Q. wutaishanica stands, and were laid-out in the regular dimension of 20×20 m each.
On each plot, increment cores were collected from pre-designated sampling trees at breast height (130cm above ground) with a Presler increment borer. Altogether, the trees sampled for increment cores consisted of 30 individuals for L. gmelinii, 82 individuals for P. tabulaeformis and 65 individuals for Q. wutaishanica, which were randomly selected across age classes >50 years. The cores were mounted on grooved boards under a layer of wood glue, air dried and then sanded with successively finer grades of sandpaper until all rings became easily identifiable. The ring widths were measured with a linear digitizing tablet coupled to a computer at a resolution of 0.001mm. Dating and measurement errors were further checked and identified by using the computer program COFECHA of Holmes (1983). Cores with apparent missing rings or severe distortion were excluded from the analysis. After standardizing each chronology, all tree-RWI were averaged to generate a standard chronology for each tree species by using the program ARSTAN (Cook and Holmes 1986). Several descriptive statistics commonly adopted in dendrochronology were used to assess the quality of the standard chronologies, including the mean sensitivity (MS) and standard deviation (SD) for assessing the high-frequency variations (Fritts 1976), and the express population signal (EPS) for expressing the confidence of the site chronologies (Briffa and Jones 1990). EPS is a measure to express the common signal in a time series; a value of 0.85 or higher is commonly considered as acceptable (Wigley et al. 1984).
Model description and parameterization
We used the version 4.2 of Biome-BGC model (http://www.ntsg.umt.edu/project/Biome–BGC) developed by the Numerical Terradynamic Simulation Group at the University of Montana, USA, to simulate the forest NPP under present climate and predicted changes in future climate of the study areas. The model simulates the mass-balanced dynamics of carbon, nitrogen, water and energy in forest and non-forest terrestrial ecosystem ranging from individual plots to global scale (Churkina et al. 2003; Luo et al. 2010; Running and Coughlan 1988; Thornton et al. 2002; White et al. 2000). It runs on a daily time-step with the input data of daily meteorological records and information on the general environment and the ecophysiological traits of the vegetation to be simulated.
The Biome-BGC was provided with default parameter sets of ecophysiological characteristics for the major biome types, such as evergreen needleleaf forest, deciduous needleleaf forest and deciduous broadleaf forest (White et al. 2000). In this study, the model was parameterized with site- and/or species-specific information compiled for this study by conducting literature survey. Average values were used if multiple data sources were found for the same attributes. If the species-specific data on some of the parameters were not found from literature, the default values of deciduous needleleaf forest were used as representation of L. gmelinii, the default values of evergreen needleleaf forest as representation of P. tabulaeformis and the default values of deciduous broadleaf forest as representation of Q. wutaishanica, respectively. The ecophysiological parameters for simulations of NPP of the three forest types are listed in Supplementary Tables A1–A3.
Meteorological data
Daily records of maximum and minimum temperatures and precipitation for the period 1960–2006 were obtained from the China Meteorological Data Sharing Service System (CMDSSS; http://cdc.cma.gov.cn). The missing daily data for vapor pressure deficit, incident shortwave radiation and daytime solar irradiance were generated using the microclimate simulator MT–CLIM version 4.3 (http://www.ntsg.umt.edu/project/mtclim), corrected for differences in slope, aspect and elevation between the base meteorological station and the study site (Thornton and Running 1999; Thornton et al. 2000).
Values of the annual mean air temperature and annual precipitation were directly derived from the daily meteorological observations archived in the database of CMDSSS. The annual evapotranspiration was computed based on meteorological and vegetation data using the Penman–Monteith equation (Monteith 1973).
Modeling procedures
The model simulations were comprised of two phases: the self-initialization, or spin-up simulation, and the normal simulation. Biome-BGC is a mechanistic biogeochemical model; the endpoint of any given simulation depends on the starting values of state and flux variables, or simply on initial conditions (Ruelle 1978), as commonly required in ecosystem models (Thornton and Rosenbloom 2005). In the absence of information for the initialization of the state variables, model simulations are required for the spin-up run based on standard procedures (White et al. 2000). Using pre-industrial carbon dioxide concentration of 294.842 p.p.m., approximating the level at the end of the nineteenth century and for the nitrogen deposition levels at 0.0001kg N m2 year−1 (Holland et al. 1999), we performed the spin-up procedure until reaching a steady state among climate, vegetation, ecophysiological traits, soil organic matter and nutrient pools (Thornton et al. 2002). The endpoint of the spin-up run was used as the initial conditions for the normal model simulations performed for a 47-year time period, from 1960 to 2006, with historical daily metrological data and the ambient CO2 concentration.
Future climate and CO2 scenarios
Global temperature increase is predicted to be greatest in the northern mid- and high-latitude regions (IPCC 2007). This general pattern of climate change also applies to China (Ding et al. 2006). To develop future trajectories of global climate change, the Intergovernmental Panel on Climate Change (IPCC) outlined four carbon emissions scenarios, designated as A1, A2, B1 and B2, based on distinct directions for global development through the year 2100 (IPCC 2000). The scenarios A2 and B2 are generally adopted in the China’s National Assessment Report on Climate Change (ECNARCC 2007) and were used for evaluation in this study. A2 describes a world with increased population growth, slow economic development and slow technological change, whereas B2 reflects a scenario with an intermediate population and economic growth, addressing local solutions to economic, social and environmental sustainability (IPCC 2000).
We used predicted changes in air temperature and precipitation in the China’s National Assessment Report on Climate Change (ECNARCC 2007) for the periods 2011–40, 2041–70 and 2071–100, against the reference period 1961–90 for the three forest types (Table 1). The CO2 concentrations of the three time periods were set to values estimated using the Bern carbon model based on the IPCC carbon emissions scenarios A2 and B2 (Joos et al. 2001). We performed simulations to examine the sensitivity of each forest type to future climate change and rising CO2 concentrations.
predicted changes in annual mean air temperature and annual precipitation under the climate change scenarios SRES A2 and B2 of IPCC (2000) for the periods 2011–40, 2041–70 and 2071–100 against the reference period 1961–90 and the corresponding atmospheric CO2 concentrations used in the simulation
| Study sites | Period | A2 | B2 | ||||
|---|---|---|---|---|---|---|---|
| Temperature (°C) | Precipitation (%) | CO2 (p.p.m.) | Temperature (°C) | Precipitation (%) | CO2 (p.p.m.) | ||
| Genhe | 2011–40 | +1.7 | +1 | 386–481 | +2.1 | +3 | 386–490 |
| 2041–70 | +3.8 | +5 | ~620 | +3.4 | +8 | ~524 | |
| 2071–100 | +6.1 | +13 | ~836 | +4.5 | +12 | ~611 | |
| Taiyue | 2011–40 | +1.6 | –1 | 386–481 | +1.8 | +2 | 386–490 |
| 2041–70 | +3.2 | +2 | ~620 | +2.9 | +4 | ~524 | |
| 2071–100 | +5.3 | +11 | ~836 | +3.8 | +10 | ~611 | |
| Study sites | Period | A2 | B2 | ||||
|---|---|---|---|---|---|---|---|
| Temperature (°C) | Precipitation (%) | CO2 (p.p.m.) | Temperature (°C) | Precipitation (%) | CO2 (p.p.m.) | ||
| Genhe | 2011–40 | +1.7 | +1 | 386–481 | +2.1 | +3 | 386–490 |
| 2041–70 | +3.8 | +5 | ~620 | +3.4 | +8 | ~524 | |
| 2071–100 | +6.1 | +13 | ~836 | +4.5 | +12 | ~611 | |
| Taiyue | 2011–40 | +1.6 | –1 | 386–481 | +1.8 | +2 | 386–490 |
| 2041–70 | +3.2 | +2 | ~620 | +2.9 | +4 | ~524 | |
| 2071–100 | +5.3 | +11 | ~836 | +3.8 | +10 | ~611 | |
predicted changes in annual mean air temperature and annual precipitation under the climate change scenarios SRES A2 and B2 of IPCC (2000) for the periods 2011–40, 2041–70 and 2071–100 against the reference period 1961–90 and the corresponding atmospheric CO2 concentrations used in the simulation
| Study sites | Period | A2 | B2 | ||||
|---|---|---|---|---|---|---|---|
| Temperature (°C) | Precipitation (%) | CO2 (p.p.m.) | Temperature (°C) | Precipitation (%) | CO2 (p.p.m.) | ||
| Genhe | 2011–40 | +1.7 | +1 | 386–481 | +2.1 | +3 | 386–490 |
| 2041–70 | +3.8 | +5 | ~620 | +3.4 | +8 | ~524 | |
| 2071–100 | +6.1 | +13 | ~836 | +4.5 | +12 | ~611 | |
| Taiyue | 2011–40 | +1.6 | –1 | 386–481 | +1.8 | +2 | 386–490 |
| 2041–70 | +3.2 | +2 | ~620 | +2.9 | +4 | ~524 | |
| 2071–100 | +5.3 | +11 | ~836 | +3.8 | +10 | ~611 | |
| Study sites | Period | A2 | B2 | ||||
|---|---|---|---|---|---|---|---|
| Temperature (°C) | Precipitation (%) | CO2 (p.p.m.) | Temperature (°C) | Precipitation (%) | CO2 (p.p.m.) | ||
| Genhe | 2011–40 | +1.7 | +1 | 386–481 | +2.1 | +3 | 386–490 |
| 2041–70 | +3.8 | +5 | ~620 | +3.4 | +8 | ~524 | |
| 2071–100 | +6.1 | +13 | ~836 | +4.5 | +12 | ~611 | |
| Taiyue | 2011–40 | +1.6 | –1 | 386–481 | +1.8 | +2 | 386–490 |
| 2041–70 | +3.2 | +2 | ~620 | +2.9 | +4 | ~524 | |
| 2071–100 | +5.3 | +11 | ~836 | +3.8 | +10 | ~611 | |
Statistical analysis
Relationships between the modeled NPP and tree-RWI were examined with linear regression analyses. The relationships between the modeled NPP and the selective climatic variables (i.e. temperature and precipitation) were examined with both the linear regression analyses for single-factor effect and the multiple stepwise regressions for the interactive effects between precipitation and temperature. All data analyses were performed with R3.0.1 software (R Development Core Team 2009).
RESULTS
Historical climate analysis of the study sites
During the period 1960–2006, the annual mean air temperature significantly increased at an average rate of 0.07°C a−1 (P < 0.001) with relatively large inter-annual variations (coefficient of variation [CV] = 30.1%) at the Genhe site and at 0.02°C a−1 (P = 0.001) with a relatively narrow range of inter-annual variations (CV = 5.8%) at the Taiyue site (Fig. 1A). The annual precipitation was moderately variable inter-annually (CV = 17.7%) without displaying apparent trend of change at the Genhe site, while it declined marginally (P = 0.06) with large inter-annual variations (CV = 23.6%) at the Taiyue site (Fig. 1B).
temporal variation in (A) annual mean air temperature and (B) annual precipitation during the period 1960–2006 for the two study sites in northern China.
Tree-ring chronologies and associated descriptive statistics
Among the three forest types, the L. gmelinii forest had the longest chronology and was lowest in the MS, SD and EPS, whereas the P. tabulaeformis forest was highest in the MS, SD and EPS (Table 2). The values of EPS of the chronologies for all three forest types exceeded the recommended threshold of 0.85; hence, the chronologies were judged to be suitable for use in evaluating the impacts of climate change variability and rising atmospheric CO2 concentration on forest productivity (Table 2).
general descriptive statistics of the standard tree-ring chronologies for the three forest types
| Forest types | Trees sampled | Chronology time span | MS | SD | EPS |
|---|---|---|---|---|---|
| Larix gmelinii forest | 30 | 1743–2009 | 0.17 | 0.22 | 0.95 |
| Pinus tabulaeformis forest | 69 | 1895–2010 | 0.27 | 0.31 | 0.98 |
| Quercus wutaishanica forest | 54 | 1904–2010 | 0.25 | 0.24 | 0.97 |
| Forest types | Trees sampled | Chronology time span | MS | SD | EPS |
|---|---|---|---|---|---|
| Larix gmelinii forest | 30 | 1743–2009 | 0.17 | 0.22 | 0.95 |
| Pinus tabulaeformis forest | 69 | 1895–2010 | 0.27 | 0.31 | 0.98 |
| Quercus wutaishanica forest | 54 | 1904–2010 | 0.25 | 0.24 | 0.97 |
general descriptive statistics of the standard tree-ring chronologies for the three forest types
| Forest types | Trees sampled | Chronology time span | MS | SD | EPS |
|---|---|---|---|---|---|
| Larix gmelinii forest | 30 | 1743–2009 | 0.17 | 0.22 | 0.95 |
| Pinus tabulaeformis forest | 69 | 1895–2010 | 0.27 | 0.31 | 0.98 |
| Quercus wutaishanica forest | 54 | 1904–2010 | 0.25 | 0.24 | 0.97 |
| Forest types | Trees sampled | Chronology time span | MS | SD | EPS |
|---|---|---|---|---|---|
| Larix gmelinii forest | 30 | 1743–2009 | 0.17 | 0.22 | 0.95 |
| Pinus tabulaeformis forest | 69 | 1895–2010 | 0.27 | 0.31 | 0.98 |
| Quercus wutaishanica forest | 54 | 1904–2010 | 0.25 | 0.24 | 0.97 |
Biome-BGC simulations of NPP and comparison with RWI
Among the three forest types, the L. gmelinii forest had the lowest values of modeled NPP with more constrained inter-annual variability (CV = 10.8%), ranging from 280 to 496g C m−2 a−1, whereas the Q. wutaishanica forest had the highest values of modeled NPP with intermediate inter-annual variability (CV = 17.8%), ranging from 334 to 848g C m−2 a−1 (Fig. 2A). The P. tabulaeformis forest ranked intermediate in the modeled NPP with the greatest inter-annual variations (CV = 18.8%), ranging from 235 to 710g C m−2 a−1 (Fig. 2A). The mean values of modeled NPP for the period 1960–2006 based on the Biome-BGC simulations are in general agreement with those reported in literature for the same area or other similar forest types in China and worldwide (Tables 3–5).
comparison of modeled NPP with RWI in the three forest types of northern China for the period 1960–2006. (A) Time series of modeled NPP and RWI; (B) relationships between modeled NPP and RWI.
summary of NPP values from various studies for boreal or deciduous coniferous forests
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Larix gmelinii | Genhe | 424b | Biome-BGC simulations | This study |
| Larix gmelinii | Genhe | 425 | Field measurements | Feng and Yang (1985) |
| Larix gmelinii | Genhe | 448 (417–478)c | Field measurements | Liu et al. (1994) |
| Larix gmelinii | Daxing’anling | 437 (389–490) | Field measurements | Shi et al. (2002) |
| Larix spp | Boreal forest, China | 516 (189–868) | Field measurements | Ni et al. (2001) |
| Boreal forest | China | 181–780 | Field measurements | Jiang et al. (1999) |
| Deciduous coniferous forest | Eastern China | 419 | Field measurements | Li et al. (2011) |
| Deciduous coniferous forest | China | 432 | CASA simulations | Piao et al. (2001) |
| Deciduous coniferous forest | Northeast China | 451 | GLOPEM–CEVSA simulations | Zhao et al. (2011) |
| Larix gmelinii forest | Daxin’anling | 510 | Remote sensing | Jiang et al. (1999) |
| Larix spp | Northern China | 447 (152–626) | Remote sensing | Zhu et al. (2007) |
| Boreal forest | Siberian | 314–445 | Field measurements | Jarvis et al. (2001) |
| Boreal forest | Worldwide | 424 | Field measurements | Gower et al.(2001) |
| Deciduous coniferous forest | Worldwide | 400 (200–1000) | MIAMI simulations | Lieth (1975) |
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Larix gmelinii | Genhe | 424b | Biome-BGC simulations | This study |
| Larix gmelinii | Genhe | 425 | Field measurements | Feng and Yang (1985) |
| Larix gmelinii | Genhe | 448 (417–478)c | Field measurements | Liu et al. (1994) |
| Larix gmelinii | Daxing’anling | 437 (389–490) | Field measurements | Shi et al. (2002) |
| Larix spp | Boreal forest, China | 516 (189–868) | Field measurements | Ni et al. (2001) |
| Boreal forest | China | 181–780 | Field measurements | Jiang et al. (1999) |
| Deciduous coniferous forest | Eastern China | 419 | Field measurements | Li et al. (2011) |
| Deciduous coniferous forest | China | 432 | CASA simulations | Piao et al. (2001) |
| Deciduous coniferous forest | Northeast China | 451 | GLOPEM–CEVSA simulations | Zhao et al. (2011) |
| Larix gmelinii forest | Daxin’anling | 510 | Remote sensing | Jiang et al. (1999) |
| Larix spp | Northern China | 447 (152–626) | Remote sensing | Zhu et al. (2007) |
| Boreal forest | Siberian | 314–445 | Field measurements | Jarvis et al. (2001) |
| Boreal forest | Worldwide | 424 | Field measurements | Gower et al.(2001) |
| Deciduous coniferous forest | Worldwide | 400 (200–1000) | MIAMI simulations | Lieth (1975) |
aData on biomass in the original literature were all converted to carbon stock assuming the conversion factor of 0.5 for wood, foliage and roots. Values in parentheses indicate the range of variation in NPP.
bThe modeled NPP in this study is an average for the period 1960–2006.
cThe study provided only the aboveground NPP, which was converted to the total NPP using the ratio of shoot NPP to root NPP reported in Feng and Yang (1985).
summary of NPP values from various studies for boreal or deciduous coniferous forests
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Larix gmelinii | Genhe | 424b | Biome-BGC simulations | This study |
| Larix gmelinii | Genhe | 425 | Field measurements | Feng and Yang (1985) |
| Larix gmelinii | Genhe | 448 (417–478)c | Field measurements | Liu et al. (1994) |
| Larix gmelinii | Daxing’anling | 437 (389–490) | Field measurements | Shi et al. (2002) |
| Larix spp | Boreal forest, China | 516 (189–868) | Field measurements | Ni et al. (2001) |
| Boreal forest | China | 181–780 | Field measurements | Jiang et al. (1999) |
| Deciduous coniferous forest | Eastern China | 419 | Field measurements | Li et al. (2011) |
| Deciduous coniferous forest | China | 432 | CASA simulations | Piao et al. (2001) |
| Deciduous coniferous forest | Northeast China | 451 | GLOPEM–CEVSA simulations | Zhao et al. (2011) |
| Larix gmelinii forest | Daxin’anling | 510 | Remote sensing | Jiang et al. (1999) |
| Larix spp | Northern China | 447 (152–626) | Remote sensing | Zhu et al. (2007) |
| Boreal forest | Siberian | 314–445 | Field measurements | Jarvis et al. (2001) |
| Boreal forest | Worldwide | 424 | Field measurements | Gower et al.(2001) |
| Deciduous coniferous forest | Worldwide | 400 (200–1000) | MIAMI simulations | Lieth (1975) |
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Larix gmelinii | Genhe | 424b | Biome-BGC simulations | This study |
| Larix gmelinii | Genhe | 425 | Field measurements | Feng and Yang (1985) |
| Larix gmelinii | Genhe | 448 (417–478)c | Field measurements | Liu et al. (1994) |
| Larix gmelinii | Daxing’anling | 437 (389–490) | Field measurements | Shi et al. (2002) |
| Larix spp | Boreal forest, China | 516 (189–868) | Field measurements | Ni et al. (2001) |
| Boreal forest | China | 181–780 | Field measurements | Jiang et al. (1999) |
| Deciduous coniferous forest | Eastern China | 419 | Field measurements | Li et al. (2011) |
| Deciduous coniferous forest | China | 432 | CASA simulations | Piao et al. (2001) |
| Deciduous coniferous forest | Northeast China | 451 | GLOPEM–CEVSA simulations | Zhao et al. (2011) |
| Larix gmelinii forest | Daxin’anling | 510 | Remote sensing | Jiang et al. (1999) |
| Larix spp | Northern China | 447 (152–626) | Remote sensing | Zhu et al. (2007) |
| Boreal forest | Siberian | 314–445 | Field measurements | Jarvis et al. (2001) |
| Boreal forest | Worldwide | 424 | Field measurements | Gower et al.(2001) |
| Deciduous coniferous forest | Worldwide | 400 (200–1000) | MIAMI simulations | Lieth (1975) |
aData on biomass in the original literature were all converted to carbon stock assuming the conversion factor of 0.5 for wood, foliage and roots. Values in parentheses indicate the range of variation in NPP.
bThe modeled NPP in this study is an average for the period 1960–2006.
cThe study provided only the aboveground NPP, which was converted to the total NPP using the ratio of shoot NPP to root NPP reported in Feng and Yang (1985).
summary of NPP values from various studies for evergreen coniferous forests
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Pinus tabulaeformis forest | Taiyue | 515b | Biome-BGC simulations | This study |
| Pinus tabulaeformis forest | Taiyue | 490 | Field measurements | Zhai et al. (1992) |
| Pinus tabulaeformis forest | northern China | 589 | Field measurements | Fang et al. (2006) |
| Temperate evergreen coniferous forest | China | 635 | Remote sensing | Jiang et al. (1999) |
| Temperate evergreen coniferous forest | Northeast China | 470 | GLOPEM–CEVSA simulations | Zhao et al.(2011) |
| Evergreen coniferous forest | China | 179–806 | Field measurements | Zhu et al. (2007) |
| Temperate evergreen coniferous forest | China | 625 | Field measurements | Jiang et al. (1999) |
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Pinus tabulaeformis forest | Taiyue | 515b | Biome-BGC simulations | This study |
| Pinus tabulaeformis forest | Taiyue | 490 | Field measurements | Zhai et al. (1992) |
| Pinus tabulaeformis forest | northern China | 589 | Field measurements | Fang et al. (2006) |
| Temperate evergreen coniferous forest | China | 635 | Remote sensing | Jiang et al. (1999) |
| Temperate evergreen coniferous forest | Northeast China | 470 | GLOPEM–CEVSA simulations | Zhao et al.(2011) |
| Evergreen coniferous forest | China | 179–806 | Field measurements | Zhu et al. (2007) |
| Temperate evergreen coniferous forest | China | 625 | Field measurements | Jiang et al. (1999) |
aData on biomass in the original literature were all converted to carbon stock assuming the conversion factor of 0.5 for wood, foliage and roots. Values in parentheses indicate the range of variation in NPP.
bThe modeled NPP in this study is an average for the period 1960–2006.
summary of NPP values from various studies for evergreen coniferous forests
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Pinus tabulaeformis forest | Taiyue | 515b | Biome-BGC simulations | This study |
| Pinus tabulaeformis forest | Taiyue | 490 | Field measurements | Zhai et al. (1992) |
| Pinus tabulaeformis forest | northern China | 589 | Field measurements | Fang et al. (2006) |
| Temperate evergreen coniferous forest | China | 635 | Remote sensing | Jiang et al. (1999) |
| Temperate evergreen coniferous forest | Northeast China | 470 | GLOPEM–CEVSA simulations | Zhao et al.(2011) |
| Evergreen coniferous forest | China | 179–806 | Field measurements | Zhu et al. (2007) |
| Temperate evergreen coniferous forest | China | 625 | Field measurements | Jiang et al. (1999) |
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Pinus tabulaeformis forest | Taiyue | 515b | Biome-BGC simulations | This study |
| Pinus tabulaeformis forest | Taiyue | 490 | Field measurements | Zhai et al. (1992) |
| Pinus tabulaeformis forest | northern China | 589 | Field measurements | Fang et al. (2006) |
| Temperate evergreen coniferous forest | China | 635 | Remote sensing | Jiang et al. (1999) |
| Temperate evergreen coniferous forest | Northeast China | 470 | GLOPEM–CEVSA simulations | Zhao et al.(2011) |
| Evergreen coniferous forest | China | 179–806 | Field measurements | Zhu et al. (2007) |
| Temperate evergreen coniferous forest | China | 625 | Field measurements | Jiang et al. (1999) |
aData on biomass in the original literature were all converted to carbon stock assuming the conversion factor of 0.5 for wood, foliage and roots. Values in parentheses indicate the range of variation in NPP.
bThe modeled NPP in this study is an average for the period 1960–2006.
summary of NPP values from various studies for deciduous broadleaved forests
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Quercus wutaishanica forest | Taiyue | 645b | Biome-BGC simulations | This study |
| Quercus wutaishanica forest | Northern China | 349–1364 | Field measurements | Jiang (1997) |
| Quercus wutaishanica forest | Northern China | 700 | LPJ–GUESS simulations | Liu et al. (2009) |
| Deciduous broadleaved forest | Northern China | 638 | GLOPEM–CEVSA simulations | Zhao et al. (2011) |
| Deciduous broadleaved forest | China | 663 | Remote sensing | Zhu et al. (2007) |
| Temperate deciduous forest | China | 715 (502–1036) | TEM simulations | Xiao et al. (1998) |
| Temperate deciduous forest | World | 620 (81–978) | TEM simulations | Melillo et al. (1993) |
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Quercus wutaishanica forest | Taiyue | 645b | Biome-BGC simulations | This study |
| Quercus wutaishanica forest | Northern China | 349–1364 | Field measurements | Jiang (1997) |
| Quercus wutaishanica forest | Northern China | 700 | LPJ–GUESS simulations | Liu et al. (2009) |
| Deciduous broadleaved forest | Northern China | 638 | GLOPEM–CEVSA simulations | Zhao et al. (2011) |
| Deciduous broadleaved forest | China | 663 | Remote sensing | Zhu et al. (2007) |
| Temperate deciduous forest | China | 715 (502–1036) | TEM simulations | Xiao et al. (1998) |
| Temperate deciduous forest | World | 620 (81–978) | TEM simulations | Melillo et al. (1993) |
aData on biomass in the original literature were all converted to carbon stock assuming the conversion factor of 0.5 for wood, foliage and roots. Values in parentheses indicate the range of variation in NPP.
bThe modeled NPP for each forest type is an average for the period 1960–2006.
summary of NPP values from various studies for deciduous broadleaved forests
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Quercus wutaishanica forest | Taiyue | 645b | Biome-BGC simulations | This study |
| Quercus wutaishanica forest | Northern China | 349–1364 | Field measurements | Jiang (1997) |
| Quercus wutaishanica forest | Northern China | 700 | LPJ–GUESS simulations | Liu et al. (2009) |
| Deciduous broadleaved forest | Northern China | 638 | GLOPEM–CEVSA simulations | Zhao et al. (2011) |
| Deciduous broadleaved forest | China | 663 | Remote sensing | Zhu et al. (2007) |
| Temperate deciduous forest | China | 715 (502–1036) | TEM simulations | Xiao et al. (1998) |
| Temperate deciduous forest | World | 620 (81–978) | TEM simulations | Melillo et al. (1993) |
| Forest type | Location | NPPa (g C m−2 a−1) | Method of estimation | Reference |
|---|---|---|---|---|
| Quercus wutaishanica forest | Taiyue | 645b | Biome-BGC simulations | This study |
| Quercus wutaishanica forest | Northern China | 349–1364 | Field measurements | Jiang (1997) |
| Quercus wutaishanica forest | Northern China | 700 | LPJ–GUESS simulations | Liu et al. (2009) |
| Deciduous broadleaved forest | Northern China | 638 | GLOPEM–CEVSA simulations | Zhao et al. (2011) |
| Deciduous broadleaved forest | China | 663 | Remote sensing | Zhu et al. (2007) |
| Temperate deciduous forest | China | 715 (502–1036) | TEM simulations | Xiao et al. (1998) |
| Temperate deciduous forest | World | 620 (81–978) | TEM simulations | Melillo et al. (1993) |
aData on biomass in the original literature were all converted to carbon stock assuming the conversion factor of 0.5 for wood, foliage and roots. Values in parentheses indicate the range of variation in NPP.
bThe modeled NPP for each forest type is an average for the period 1960–2006.
The inter-annual variability in modeled NPP closely matched the inter-annual variability in RWI for the three forest types (Fig. 2A; CV = 10.8% vs. 10.4% in L. gmelinii forest, 18.8% vs. 19.6% in P. tabulaeformis forest and 17.8% vs. 17.4% in Q. wutaishanica forest, respectively). However, significant relationships between modeled NPP and RWI were only found in the two forest types at the Taiyue site, namely the P. tabulaeformis forest (P < 0.05) and the Q. wutaishanica (P < 0.001) forest (Fig. 2B).
Climatic constraints of NPP under present conditions
The modeled NPP based on the present climate was significantly (P < 0.05) related to both annual mean air temperature and annual precipitation. However, the relationships differed among the three forest types as well as between the two study sites. The modeled NPP was positively and closely related to annual mean air temperature in the L. gmelinii forest, but negatively in the P. tabulaeformis forest and the Q. wutaishanica forest (Fig. 3). The relationships of modeled NPP with annual precipitation for the three forest types were all positive (Fig. 3). Among the three forest types, the Q. wutaishanica forest responded most strongly in modeled NPP to both annual mean air temperature and annual precipitation, whereas the L. gmelinii forest responded the least to both climatic variables (Fig. 3). Multiple stepwise regression analyses showed that temperature was a more important constraint of NPP than precipitation in the L. gmelinii forest, whereas precipitation appeared to be a prominent factor limiting the growth in P. tabulaeformis and Q. wutaishanica (Table 6).
relationships of modeled NPP with annual precipitation and annual mean air temperature in the three forest types of northern China for the period 1960–2006.
summary of multiple stepwise regressions of modeled NPP with precipitation and temperature
| Forest type | Variable | Estimate | Standard error | t-value | P |
|---|---|---|---|---|---|
| Larix gmelinii | (Intercept) | 420.494 | 36.88 | 11.402 | 0.000 |
| Precipitation | 0.183 | 0.07 | 2.559 | 0.014 | |
| Temperature | 18.618 | 4.43 | 4.201 | 0.000 | |
| R2 | 0.36 | ||||
| F-value | 12.57 | ||||
| P | <0.001 | ||||
| Pinus tabulaeformis | (Intercept) | 219.612 | 43.86 | 5.007 | 0.000 |
| Precipitation | 0.438 | 0.06 | 6.905 | 0.000 | |
| R2 | 0.52 | ||||
| F-value | 12.57 | ||||
| P | <0.001 | ||||
| Quercus wutaishanica | (Intercept) | 262.022 | 43.27 | 6.055 | 0.000 |
| Precipitation | 0.569 | 0.06 | 9.09 | 0.000 | |
| R2 | 0.65 | ||||
| F-value | 82.62 | ||||
| P | <0.001 |
| Forest type | Variable | Estimate | Standard error | t-value | P |
|---|---|---|---|---|---|
| Larix gmelinii | (Intercept) | 420.494 | 36.88 | 11.402 | 0.000 |
| Precipitation | 0.183 | 0.07 | 2.559 | 0.014 | |
| Temperature | 18.618 | 4.43 | 4.201 | 0.000 | |
| R2 | 0.36 | ||||
| F-value | 12.57 | ||||
| P | <0.001 | ||||
| Pinus tabulaeformis | (Intercept) | 219.612 | 43.86 | 5.007 | 0.000 |
| Precipitation | 0.438 | 0.06 | 6.905 | 0.000 | |
| R2 | 0.52 | ||||
| F-value | 12.57 | ||||
| P | <0.001 | ||||
| Quercus wutaishanica | (Intercept) | 262.022 | 43.27 | 6.055 | 0.000 |
| Precipitation | 0.569 | 0.06 | 9.09 | 0.000 | |
| R2 | 0.65 | ||||
| F-value | 82.62 | ||||
| P | <0.001 |
summary of multiple stepwise regressions of modeled NPP with precipitation and temperature
| Forest type | Variable | Estimate | Standard error | t-value | P |
|---|---|---|---|---|---|
| Larix gmelinii | (Intercept) | 420.494 | 36.88 | 11.402 | 0.000 |
| Precipitation | 0.183 | 0.07 | 2.559 | 0.014 | |
| Temperature | 18.618 | 4.43 | 4.201 | 0.000 | |
| R2 | 0.36 | ||||
| F-value | 12.57 | ||||
| P | <0.001 | ||||
| Pinus tabulaeformis | (Intercept) | 219.612 | 43.86 | 5.007 | 0.000 |
| Precipitation | 0.438 | 0.06 | 6.905 | 0.000 | |
| R2 | 0.52 | ||||
| F-value | 12.57 | ||||
| P | <0.001 | ||||
| Quercus wutaishanica | (Intercept) | 262.022 | 43.27 | 6.055 | 0.000 |
| Precipitation | 0.569 | 0.06 | 9.09 | 0.000 | |
| R2 | 0.65 | ||||
| F-value | 82.62 | ||||
| P | <0.001 |
| Forest type | Variable | Estimate | Standard error | t-value | P |
|---|---|---|---|---|---|
| Larix gmelinii | (Intercept) | 420.494 | 36.88 | 11.402 | 0.000 |
| Precipitation | 0.183 | 0.07 | 2.559 | 0.014 | |
| Temperature | 18.618 | 4.43 | 4.201 | 0.000 | |
| R2 | 0.36 | ||||
| F-value | 12.57 | ||||
| P | <0.001 | ||||
| Pinus tabulaeformis | (Intercept) | 219.612 | 43.86 | 5.007 | 0.000 |
| Precipitation | 0.438 | 0.06 | 6.905 | 0.000 | |
| R2 | 0.52 | ||||
| F-value | 12.57 | ||||
| P | <0.001 | ||||
| Quercus wutaishanica | (Intercept) | 262.022 | 43.27 | 6.055 | 0.000 |
| Precipitation | 0.569 | 0.06 | 9.09 | 0.000 | |
| R2 | 0.65 | ||||
| F-value | 82.62 | ||||
| P | <0.001 |
Responses of NPP to predicted climate change
Future climate changes were predicted to substantially increase NPP for all the three forest types (Table 7). The magnitude of changes, however, appeared to vary over different time periods and with forest types and climate change scenarios. Under the projected climatic and atmospheric CO2 conditions of SRES A2, the two needleleaf forests would increase NPP by more than 40% over the period 2071–100 against the reference period 1961–90, whereas only half of the NPP increase would be expected in the Q. wutaishanica forest as compared with the other two forest types. Between the two needleleaf forest types, the L. gmelinii forest was predicted to have a much greater initial response in NPP to future changes in climate than the P. tabulaeformis forest (Table 7). However, with time the difference in the response of NPP between the two forest types would diminish. In contrast, the two temperate forest types started with similar responses in NPP to future changes in climate but continued with much diverged temporal patterns (Table 7).
predicted NPP for the periods 2011–40, 2041–70 and 2070–100 under the IPCC (2000) climate change scenarios SRES A2 and B2 for the three forest types in northern China, based on simulations with Biome-BGC version 4.2, and the relative changes in NPP against the reference period 1961–90
| Climate scenario | Period | Larix gmelinii forest | Pinus tabulaeformis forest | Quercus wutaishanica forest | |||
|---|---|---|---|---|---|---|---|
| NPP (g C m−2 a−1) | Change (%) | NPP (g C m−2 a−1) | Change (%) | NPP (g C m−2 a−1) | Change (%) | ||
| SRES A2 | 1961–90 | 416 | — | 515 | — | 634 | — |
| 2011–40 | 474 | +14.1 | 559 | +8.6 | 692 | +8.8 | |
| 2041–70 | 539 | +29.7 | 621 | +20.5 | 737 | +15.9 | |
| 2071–100 | 606 | +45.7 | 727 | +41.1 | 761 | +19.8 | |
| SRES B2 | 1961–90 | 416 | — | 515 | — | 634 | — |
| 2011–40 | 487 | +17.2 | 575 | +11.6 | 713 | +12.2 | |
| 2041–70 | 528 | +27.0 | 617 | +19.9 | 785 | +23.5 | |
| 2071–100 | 562 | +35.5 | 671 | +30.0 | 874 | +37.5 | |
| Climate scenario | Period | Larix gmelinii forest | Pinus tabulaeformis forest | Quercus wutaishanica forest | |||
|---|---|---|---|---|---|---|---|
| NPP (g C m−2 a−1) | Change (%) | NPP (g C m−2 a−1) | Change (%) | NPP (g C m−2 a−1) | Change (%) | ||
| SRES A2 | 1961–90 | 416 | — | 515 | — | 634 | — |
| 2011–40 | 474 | +14.1 | 559 | +8.6 | 692 | +8.8 | |
| 2041–70 | 539 | +29.7 | 621 | +20.5 | 737 | +15.9 | |
| 2071–100 | 606 | +45.7 | 727 | +41.1 | 761 | +19.8 | |
| SRES B2 | 1961–90 | 416 | — | 515 | — | 634 | — |
| 2011–40 | 487 | +17.2 | 575 | +11.6 | 713 | +12.2 | |
| 2041–70 | 528 | +27.0 | 617 | +19.9 | 785 | +23.5 | |
| 2071–100 | 562 | +35.5 | 671 | +30.0 | 874 | +37.5 | |
predicted NPP for the periods 2011–40, 2041–70 and 2070–100 under the IPCC (2000) climate change scenarios SRES A2 and B2 for the three forest types in northern China, based on simulations with Biome-BGC version 4.2, and the relative changes in NPP against the reference period 1961–90
| Climate scenario | Period | Larix gmelinii forest | Pinus tabulaeformis forest | Quercus wutaishanica forest | |||
|---|---|---|---|---|---|---|---|
| NPP (g C m−2 a−1) | Change (%) | NPP (g C m−2 a−1) | Change (%) | NPP (g C m−2 a−1) | Change (%) | ||
| SRES A2 | 1961–90 | 416 | — | 515 | — | 634 | — |
| 2011–40 | 474 | +14.1 | 559 | +8.6 | 692 | +8.8 | |
| 2041–70 | 539 | +29.7 | 621 | +20.5 | 737 | +15.9 | |
| 2071–100 | 606 | +45.7 | 727 | +41.1 | 761 | +19.8 | |
| SRES B2 | 1961–90 | 416 | — | 515 | — | 634 | — |
| 2011–40 | 487 | +17.2 | 575 | +11.6 | 713 | +12.2 | |
| 2041–70 | 528 | +27.0 | 617 | +19.9 | 785 | +23.5 | |
| 2071–100 | 562 | +35.5 | 671 | +30.0 | 874 | +37.5 | |
| Climate scenario | Period | Larix gmelinii forest | Pinus tabulaeformis forest | Quercus wutaishanica forest | |||
|---|---|---|---|---|---|---|---|
| NPP (g C m−2 a−1) | Change (%) | NPP (g C m−2 a−1) | Change (%) | NPP (g C m−2 a−1) | Change (%) | ||
| SRES A2 | 1961–90 | 416 | — | 515 | — | 634 | — |
| 2011–40 | 474 | +14.1 | 559 | +8.6 | 692 | +8.8 | |
| 2041–70 | 539 | +29.7 | 621 | +20.5 | 737 | +15.9 | |
| 2071–100 | 606 | +45.7 | 727 | +41.1 | 761 | +19.8 | |
| SRES B2 | 1961–90 | 416 | — | 515 | — | 634 | — |
| 2011–40 | 487 | +17.2 | 575 | +11.6 | 713 | +12.2 | |
| 2041–70 | 528 | +27.0 | 617 | +19.9 | 785 | +23.5 | |
| 2071–100 | 562 | +35.5 | 671 | +30.0 | 874 | +37.5 | |
The model simulations produced similar values of the relative change in NPP over the period 2071–100 against the reference period 1961–90 among the three forest types under the climatic and atmospheric CO2 conditions of SRES B2 (Table 7). However, the temporal patterns of the relative change in NPP are differentiated between the two study sites; the L. gmelinii forest at the Genhe site was predicted to have greater initial response to the future climate change under SRES B2 than other two forest types (Table 7).
DISCUSSION
Forest responses to climate change are determined to a large extent by the susceptibility of tree growth to temporal fluctuations in climatic conditions. Different forest types may vary in the patterns of inter-annual variability in productivity owing to differential responses to climatic signals (Law et al. 2004). Simulations of inter-annual variability of terrestrial ecosystem carbon cycle has, therefore, imposed a challenge for most biogeochemical models (Keenan et al. 2012; Mahecha et al. 2010). In this study, based on simulations with the Biome-BGC model parameterized with species- and site-specific information and validated with tree-ring chronologies, we found contrasting responses of NPP among three forest types typical of northern China to inter-annual climate variations as well as to predicted changes in future climate under the carbon emission scenarios A2 and B2 of IPCC (2000). Among the three forest types, the L. gmelinii forest represents a typical boreal forest in the northeastern China, whereas both P. tabulaeformis and Q. wutaishanica forests naturally occur under temperate climate over a large range in central and northern China (Yu and Sun 2013). Our results showed that the modeled NPP was positively related to annual mean air temperature in the L. gmelinii forest, but negatively in the P. tabulaeformis forest and the Q. wutaishanica forest. The relationships of modeled NPP with annual precipitation for the three forest types were all positive but the linear fitting was much stronger for the two forest types of temperate climate than the boreal L. gmelinii forest. Model simulations also suggest marked, but differential increases in NPP across the three forest types in response to predicted changes of climate in northern China.
The average values of NPP for the three forest types based on our simulations with present climate are in agreement with values reported in literature for similar forest types in China and worldwide (Tables 3–5). In absence of long-term field data on NPP in our study, we used tree-ring chronologies to reflect temporal variations in forest productivity. The established chronologies were examined for quality concerning their correlations with climate signals using descriptive statistics MS, SD and EPS, all with satisfactory performance as judged by commonly accepted criteria (Briffa and Jones 1990; Fritts 1976; Wigley et al. 1984; Wu 1990). The chronologies for all three forest types had relatively high values in MS, SD and EPS (Table 2), confirming their qualifications for use in studying the correlation between tree growth and climatic factors. RWI has been found to correlate well with NPP (Graumlich et al. 1989; Krakauer and Randerson 2003; Rathgeber et al. 2000). In this study, however, we found significant relationships of the modeled NPP with RWI only in the two forest types of temperate climate; there was no clear relationship between the modeled NPP and RWI in the boreal L. gmelinii forest. Similar problem has been found in the study of Su et al. (2007) with a Picea schrenkiana forest under cold climate. There were large mismatches between the modeled NPP and RWI series in the L. gmelinii forest for the periods 1965–68 and 1999–2002, which coincided with two major drought events in northeastern China (Xiao et al. 2009). The extreme weather conditions of the two periods could have resulted in modifications of the permafrost, imposing indirect effects on tree growth via edaphic processes. However, the Biome-BGC version 4.2 used in this study does not incorporate the precise coupling between the permafrost dynamics and biogeochemical cycle. In fact, most process-based models focus mainly on the general descriptions of fundamental physiological and biogeochemical processes; they may lack predictive capabilities during extreme weather conditions (Vetter et al. 2008). Therefore, our simulations were constrained for taking into consideration of the impacts of climate variability on forest productivity through modification of belowground processes.
Ni et al. (2001) showed that the NPP of Chinese forests is highly correlated with both annual mean air temperature and precipitation. There are also studies suggesting a stronger control of precipitation than temperature on forest NPP in China (e.g. Piao et al. 2001; Cao et al. 2003). In this study, however, we show that temperature was a more important constraint of NPP than precipitation in the L. gmelinii forest, whereas precipitation appeared to be a prominent factor limiting the growth in P. tabulaeformis and Q. wutaishanica. Contrasting responses of NPP to annual mean air temperature were found between the boreal L. gmelinii forest and other two temperature forest types. The modeled NPP was positively related to annual mean air temperature in the L. gmelinii forest, but negatively in the P. tabulaeformis and Q. wutaishanica forests. Our results further demonstrate the complex interactions between climate change and forest responses (Churkina et al. 1999; Shaw et al. 2002).
Marked increases in NPP across the three forest types were predicted to occur in response to predicted changes in climate and atmospheric CO2 through this century under both the carbon emissions scenarios SRES A2 and B2 of IPCC (2000). It is well understood that temperature and precipitation are dominant controls on plant photosynthesis (Dai and Fung 1993; Lieth 1975). Raising air temperature may either increase NPP through metabolically enhanced photosynthesis as well as by increasing nutrient availability through accelerated decomposition, or inhibit NPP by enhancing plant respiration and decreasing soil moisture. Increasing precipitation tends to alleviate the water stress on plant growth, hence imposing positive effects on NPP in water-limited regions. NPP in boreal forest is characteristically low compared with temperate forests because of reduced solar radiation, cold climate and shorter growing seasons at higher latitudes (Keuper et al. 2012; Vanhala et al. 2008). Increased temperature may markedly enhance NPP in the boreal L. gmelinii forest. However, the thawing of permafrost with climate warming and drought could impose some adverse impacts on tree growth in boreal forests. Long-term field experimental studies are certainly required to closely examine the interactions between climate change and forest responses in boreal regions with consideration of the dynamics and impacts of permafrost. The much conserved responses in the two temperate forest types suggest that temperature is not a critical factor in constraining forest NPP and that increasing temperature may impose some negative impact on NPP by inducing water deficit, thus partially offsetting the growth response to elevated atmospheric CO2 concentration and slightly increasing precipitation. Our results showed that more drastic changes in climate and atmospheric CO2 concentration under the carbon emissions scenario SRES A2 would benefit relatively little to the deciduous broadleaved Q. wutaishanica forest compared with other two types of coniferous forests. The contrasting responses to inter-annual variations and changes in climate across the three forest types suggest some genetic controls of tree growth in response to changing environmental conditions (Law et al. 2004; Sun and Sweet 1996; Sun et al. 1995).
The magnitudes of increase in forest NPP predicted in this study are in line with predictions in other similar studies (Fang 2000; Ji et al. 2008; Peng et al. 2009), suggesting the adequacy of using Biome-BGC in simulating forest productivity in the region with the aid of localized parameterization. Some of the discrepancies between the simulated and measured patterns of inter-annual variations in tree growth illustrate the complex control of climate on forest productivity and emphasize the need for long-term experimental studies addressing fundamental mechanisms underlying interactions among climate change, site abiotic perturbations and forest responses. Lack of representation of the dynamics and impacts of permafrost in our simulations may explain the poor relationships between the modeled NPP and RWI in the boreal L. gmelinii forest.
SUPPLEMENTARY MATERIAL
Supplementary material is available at Journal of Plant Ecology online.
FUNDING
Public Welfare Forestry of the State Forestry Administration of China (201104008); Beijing Municipal Commission of Education for development of Key Laboratory for Silviculture and Conservation.
ACKNOWLEDGEMENTS
We wish to express our thanks to the Numerical Terradynamic Simulation Group (NTSG) at the University of Montana for providing the source code of model Biome-BGC version 4.2. We are also grateful to Drs Zhongkui Luo and Hongxin Su for their assistance in performing model simulations. Insightful comments by the handling editor and two anonymous reviewers helped greatly with improving the manuscript.
Conflict of interest statement. None declared.
REFERENCES
