EN中文

Journal of Geographical Sciences >

2021 , Vol. 31 >Issue 7: 938 - 964

DOI: https://doi.org/10.1007/s11442-021-1879-z

Research Articles

Evaluation of the carbon sequestration of Zhalong Wetland under climate change

  • YU Chenglong ,
  • LIU Dan , * ,
  • ZHAO Huiying
Expand
  • Heilongjiang Province Institute of Meteorological Sciences, Harbin 150030, China
* Liu Dan (1974-), PhD, specialized in ecological meteorology. E-mail:

Yu Chenglong (1973-), PhD, specialized in ecological meteorology. E-mail:

Received date: 2020-06-09

  Accepted date: 2020-11-13

  Online published: 2021-09-25

Supported by

Science Foundation of Heilongjiang Province (General Program)(D2018006)

National Natural Science Foundation of China(41665007)

National Natural Science Foundation of China(41165005)

CMA/Northeast China Innovation and Open Laboratory of Eco-meteorology(stqx2017zd01)

CMA/Northeast China Innovation and Open Laboratory of Eco-meteorology(stqx2018zd03)

Copyright

Copyright reserved © 2021.
Fold

Abstract

Wetland ecosystems are crucial to the global carbon cycle. In this study, the Zhalong Wetland was investigated. Based on remote sensing and meteorological observation data from 1975-2018 and the downscaled fifth phase of the coupled model intercomparison project (CMIP5) climate projection dataset from 1961-2100, the parameters of a net primary productivity (NPP) climatic potential productivity model were adjusted, and the simulation ability of the CMIP5 coupled models was evaluated. On this basis, we analysed the spatial and temporal variations of land cover types and landscape transformation processes in the Zhalong Nature Reserve over the past 44 years. We also evaluated the influence of climate change on the NPP of the vegetation, microbial heterotrophic respiration (Rh), and net ecosystem productivity (NEP) of the Zhalong Wetland and predicted the carbon sequestration potential of the Zhalong Wetland from 2019-2029 under the representative concentration pathways (RCP) 4.5 and RCP 8.5 scenarios. Our results indicate the following: (1) Herbaceous bog was the primary land cover type of the Zhalong Nature Reserve, occupying an average area of 1168.02 ± 224.05 km 2, equivalent to 51.84% of the total reserve area. (2) Since 1975, the Zhalong Nature Reserve has undergone a dry-wet-dry transformation process. Excluding several wet periods during the mid-1980s to early 1990s, the reserve has remained a dry habitat, with particularly severe conditions from 2000 onwards. (3) The 1975-2018 mean NPP, Rh, and NEP values of the Zhalong Wetland were 500.21±52.76, 337.59±10.80, and 162.62±45.56 gC·m-2·a-1, respectively, and an evaluation of the carbon balance indicated that the reserve served as a carbon sink. (4) From 1975-2018, NPP showed a significant linear increase, Rh showed a highly significant linear increase, while the increase in the carbon absorption rate was smaller than the increase in the carbon release rate. (5) Variations in NPP and NEP were precipitation-driven, with the correlations of NPP and NEP with annual precipitation and summer precipitation being highly significantly positive (P < 0.001); variations in Rh were temperature-driven, with the correlations of Rh with the average annual, summer, and autumn temperatures being highly significantly positive (P < 0.001). The interaction of precipitation and temperature enhances the impact on NPP, Rh and NEP. (6) Under the RCP 4.5 and RCP 8.5 scenarios, the predicted carbon sequestration by the Zhalong Wetland from 2019-2029 was 2.421 (± 0.225) × 1011 gC·a-1 and 2.407 (± 0.382) × 1011 gC·a-1, respectively, which were both lower than the mean carbon sequestration during the last 44 years (2.467 (± 0.950) × 1011 gC·a-1). Future climate change may negatively contribute to the carbon sequestration potential of the Zhalong Wetland. The results of the present study are significant for enhancing the abilities of integrated eco-meteorological monitoring, evaluation, and early warning systems for wetlands.

Key words： net ecosystem productivity; CMIP5; Zhalong Wetland

Cite this article

YU Chenglong , LIU Dan , ZHAO Huiying . Evaluation of the carbon sequestration of Zhalong Wetland under climate change[J]. Journal of Geographical Sciences, 2021 , 31(7) : 938 -964 . DOI: 10.1007/s11442-021-1879-z

1 Introduction

Wetlands provide the highest ecosystem service value per unit area among all ecosystem types and play a key role in the global carbon cycle (Burkett et al., 2000; Anderson et al., 2016). Although peatlands only cover approximately 3% of the Earth’s land surface, they store approximately 30% of the world’s soil organic carbon. Research on carbon sequestration by wetlands has been conducted mostly in the Arctic region (Mueller et al., 2017; Yu et al., 2018), tropics (Saunders et al., 2012; Howard et al., 2017), and central North America (Hossler et al., 2010; Bernal et al., 2017), with few studies focused on temperate Asia. The Zhalong Wetland, located along the lower reaches of the Wuyur River in Heilongjiang Province in the northern temperate zone, is the largest reed wetland in the world. This important wetland was the first National Nature Reserve in China and was one of the first wetlands added to the List of Wetlands of International Importance in China. This wetland is in a region sensitive to climate change. Statistics (data source: dataset of daily climate data from Chinese surface stations V3.0, National Meteorological Information Centre, China, http://data.cma.cn/) indicate that the Zhalong Wetland region has experienced an increase in average annual temperature at a rate of 0.32℃/10a since 1951. This increase is slightly lower than that of Heilongjiang Province (0.37℃/10a), but much higher than the national average (0.22℃/10a) (Zhou et al., 2013).
In addition, the Special Report on Global Warming of 1.5℃ (SR15) published by the Intergovernmental Panel on Climate Change (IPCC) in October 2018 (IPCC, 2018) predicts an increase of 0.3-4.8℃ in global temperature by 2100 (compared with 1900-1982) based on integrated assessment models. Since the timescale in which global warming reaches 1.5℃ is entirely dependent on the amount of anthropogenic greenhouse gases released into the atmosphere (Zhao et al., 2019), the Zhalong Wetland may experience a 1.5℃ increase within 80 years due to heightened CO2 concentrations. Global warming and elevated CO2 levels will influence the NPP of the vegetation (Gang et al., 2015; Schippers et al., 2015) and soil microbial activity (Miki et al., 2016; Wang et al., 2018). Furthermore, because the Zhalong Wetland is situated near industrial and agricultural areas, it is frequently affected by atmospheric nitrogen deposition and nitrogen fertiliser application from the surrounding farmland (Luo et al., 2014). Although the influence of available nitrogen on ecosystem carbon accumulation is unclear, studies have demonstrated that increased nitrogen availability can lead to a rise in soil microbial activity, thereby increasing Rh and the amount of carbon released by the soil into the atmosphere (Allison et al., 2008; Ma et al., 2018). In addition, available nitrogen also promotes photosynthesis and enhances the carbon absorption of ecosystems (Ivanov et al., 2018). With the co-action of climate change, natural disturbances, and anthropogenic disturbances, uncertainties in the carbon sequestration potential of temperate wetlands may influence our predictions of future climatic conditions. Therefore, analysis of the variations in the NPP and Rh of wetlands due to climate change, estimation of the NEP of these wetlands, and determination of the response of wetland carbon sequestration potential to increased temperatures are of great significance in enhancing the abilities of integrated eco-meteorological monitoring, evaluation, and early warning systems for wetlands.
Numerous studies have been conducted on carbon sequestration in wetland soils. For instance, Mueller et al. (2019) assessed the long-term carbon sequestration potential of semi-natural salt marshes in the European Wadden Sea. In another study, Bellen et al. (2011) proposed the reconstruction of carbon sequestration patterns based on multiple cores from three ombrotrophic peatlands in the Boreal forest of Quebec, Canada, which was used for the quantification of total carbon accumulation. A study by Packalen et al. (2016) explored the relationships of climate and peat type with the spatial variation of the peatland carbon mass in the Hudson Bay Lowlands of Canada. Sim et al. (2019) used paleoecological techniques to investigate the response of Canadian High Arctic wetlands to a mid-20th century increase in growing degree days. These studies are of great significance to the elucidation of wetland carbon sequestration mechanisms, exploration of the influence of environments on these sequestration abilities and the relevant feedback mechanisms, and carbon accounting in wetlands on the regional to global scales. However, most research on the controlling factors of ecosystem respiration and wetland carbon sequestration potential has been conducted based on laboratory experimentation or observation-point data, which has led to limitations in the time series or the regional representativeness of such studies. The development of remote sensing, NPP estimation, and Rh modelling techniques have provided technical support to evaluate the carbon sequestration potential of regional ecosystems. In the present study, the Zhalong Wetland was selected as the study area for the analysis of the various characteristics of land cover and landscape patterns within the reserve from 1975-2018 and the estimation of NPP, Rh, and carbon sequestration potential for the reserve from 1975-2029. This study was conducted to: 1) understand the change in carbon sequestration in Zhalong Wetland from 1975 to 2018; 2) estimate the natural capacity of Zhalong Wetland from 2019 to 2029; and 3) explain the impact of climate change on the carbon sequestration capacity of the Zhalong Wetland. In addition to providing a quantitative evaluation of the influence of climate change on the ecosystem of the Zhalong Nature Reserve, the present study provides cumulative historical data for subsequent quantitative monitoring and the evaluation of ecosystem status in the reserve and can serve as a reference for wetland resource management and conservation as well as future research on regional wetlands under global climate change.

2 Study area and methods

2.1 Study area

Zhalong Nature Reserve (46°52'-47°32'N, 123°47'-124°37'E), located along the lower reaches of the Wuyur River in western Heilongjiang Province of China, is a wetland reserve covering a total area of approximately 2250 km2 (Figure 1). The Wuyur River serves as the predominant water source for the formation and maintenance of the wetland ecosystem of the Zhalong Nature Reserve (Guo et al., 2004). The soil within the wetland is mostly comprised of chernozem, meadow soil, and bog soil. The reserve, surrounded by grassland, farmland, and human-made fishponds, functions as a conservation base for rare birds, such as the red-crowned crane, and wetland ecosystems. It is also the most complete, primitive, and open wetland ecosystem in this latitude region of North China. According to the definition of wetlands in the narrow sense stated in the Ramsar Convention (Hawkins et al., 1983), the boundaries of wetlands may change over time. To enhance the comparative analysis of the carbon sequestration potential of the Zhalong Wetland, we selected the Zhalong Nature Reserve as the study area. The Administration Bureau of Zhalong National Nature Reserve of Heilongjiang Province (http://www.chinazhalong.gov.cn/) provided data regarding the boundaries and functional zones of the study area from the Zhalong National Nature Reserve Functional Zoning Map. The reserve is divided into the core zone (739 km2), buffer zone (699 km2), and experimental zone (812 km2), as shown in Figure 1. The region in which the reserve is located has a temperate continental monsoon climate. Based on the 1980-2010 daily average data from meteorological observation stations within the Zhalong Nature Reserve and a 100-km buffer zone (data source: dataset of daily climate data from Chinese surface stations V3.0), the relevant climate data are as follows: average annual temperature: 4.34 °C, annual precipitation: 429.00 mm, average sunshine duration: 7.41 h/d.
Figure 1 Location of the Zhalong Nature Reserve

2.2 Data sources and processing

2.2.1 Remote sensing data
Landsat and HJ1B/CCD1 data were used to extract land use types. Landsat data were obtained from the United States Geological Survey (USGS, http://glovis.usgs.gov/), and the HJ satellite data were obtained from the China Centre for Resources Satellite Data and Application (CRESDA, http://www.cresda.com/CN/). The data used for the respective periods were as follows: 1975-1981: Landsat2/MSS; 1982-1986: Landsat4/MSS; 1987- 1998: Landsat5/MSS; 1999: Landsat7/ETM+; 2000-2011: Landsat5/TM; 2012: HJ1B/CCD1; and 2013-2018: Landsat8/OLI_TIRS. The Landsat2/MSS, Landsat4/MSS, and Landsat5/ MSS data has a spatial resolution of 60 m, whereas the Landsat5/TM, HJ1B/CCD1, and Landsat8/OLI_TIRS data has a spatial resolution of 30 m. The original coordinate system of the data was WGS84/UTM zone 51, and the projection of the data was converted to the Albers projection with a spatial resolution of 30 m.
2.2.2 Meteorological data
The 1961-2018 meteorological data of the study area were obtained from the Heilongjiang Meteorological Bureau, and the 1961-2100 data for future scenarios were obtained from the downscaled climate projection dataset of CMIP5 (http://stdown.agrivy.com/##). In the present study, daily maximum temperature, minimum temperature, and precipitation data were acquired from six meteorological observation stations in the Wuyur River basin. The geographic locations of the meteorological observation stations are shown in Figure 1.
CMIP was organised by the World Climate Research Programme, and CMIP5 was launched in September 2008 after several phases of CMIP1, CMIP2 and CMIP3. The experimental data involved in the project were widely used in the research of climate change mechanisms and the prediction of climate change characteristics, and these results have become one of the primary results in the assessment report of the IPCC. (Xin et al., 2012) Climate models form an important tool to predict climate change. Compared with previous models, the resolution, physical process and parameterisation of the climate coupled model used in CMIP5 have been improved, resulting in improved temperature simulation (Chen et al., 2015). However, the improvement in precipitation simulation errors is minor (Zhao et al., 2014). Liu et al. (2012) established the downscaled CMIP5 climate projection dataset by downscaling monthly CMIP5 data using a stochastic weather generator (WGEN) (Richardson et al., 1984). Historical monthly CMIP5 data were used for the extrapolation of specific WGEN parameters, which were then used to downscale and obtain future daily CMIP5 climate data to increase simulation accuracy. The dataset includes the meteorological elements (including maximum temperature, minimum temperature, precipitation, etc.) of two greenhouse gas emission and radiative forcing scenarios (RCP4.5 and RCP8.5) simulated by 33 models for 736 meteorological stations in China.
2.2.3 Geographic information data
The Advanced Land Observing Satellite (ALOS) Global Digital Surface Model, a global dataset with a horizontal resolution of 30 m, consists of data acquired by a panchromatic remote sensing instrument on ALOS. Relevant data were obtained from the ALOS Research and Application Project for use in land cover classification. The county-level administrative division data required by the present study were obtained from the 1:250,000 basic geographic information data published by the China Meteorological Administration, and the topology was established to eliminate gaps in the county boundaries. Data on the surface observation station locations were obtained from vector data published by the China Meteorological Administration. The data mentioned above was converted to the Albers projection.
2.2.4 Investigation of sampled ground
The data were used to verify the results of the land use classification and NPP estimation. At the end of August and beginning of September 2017, 55 sampling points (30 m × 30 m in the control area) were set up through the establishment of sample plots (1 m × 2 m) for field investigation and sampling, most of which were selected within the control area. The observation items include vegetation species, above-ground biomass, longitude, latitude and altitude of quadrats, vegetation coverage, and average height of vegetation canopy. The above-ground portion of the vegetation in each quadrat was harvested, and its dry weight was measured. The average value was calculated after drying. The spatial distribution of sampling points is shown in Figure 1, of which 21 sampling points are herbaceous bog swamp vegetation and 34 sampling points are grassland.

2.3 Study methods

2.3.1 Land cover classification
The Landsat/OLI_TIRS data (with atmospheric correction) acquired on June 30, 2015 were used as a reference for the geographic correction of other data to achieve errors of less than 0.5 pixels. Using the land cover classification method proposed by Yu et al. (2018), the band characteristics of each type of remote sensing image data were analysed to select the optimum classification bands, construct the water land index from 1987-2018, and calculate the normalised difference vegetation index (NDVI) from 1975-2018. Subsequently, a random forest classification model was constructed to extract the land cover types of the Zhalong Nature Reserve. A total of five land cover types were identified, including herbaceous bog, grass, water bodies, cultivated land, and building land and unused land. An accuracy validation was performed on each classification result by extracting 662 validation points with interpoint distances of 0.02° from each study area image and determining the land cover type of each point by visual interpretation. The results of the accuracy validation indicated an overall classification accuracy of >90% and a Kappa coefficient of >0.87. The annual classification accuracy and Kappa coefficient are shown in Table 1. The land cover classification results were transformed into vector data.
Table 1 Remote sensing classification accuracy of Zhalong Nature Reserve from 1975 to 2018
Time Classification accuracy (%) Kappa coefficient Time Classification accuracy (%) Kappa coefficient
1975 90.44 0.8756 1997 90.80 0.8916
1976 90.46 0.8778 1998 90.10 0.8711
1977 90.48 0.8824 1999 91.02 0.8985
1978 90.83 0.8920 2000 92.15 0.9153
1979 91.15 0.9022 2001 91.46 0.9074
1980 91.61 0.9125 2002 90.98 0.8968
1981 91.65 0.9131 2003 92.19 0.9156
1982 90.36 0.8711 2004 91.01 0.8981
1983 90.43 0.8735 2005 91.25 0.9051
1984 90.47 0.8801 2006 90.55 0.8841
1985 90.85 0.8929 2007 90.46 0.8771
1986 90.86 0.8959 2008 91.57 0.9086
1987 90.88 0.8965 2009 92.41 0.9178
1988 91.10 0.9005 2010 90.01 0.8701
1989 91.21 0.9043 2011 91.48 0.9076
1990 91.45 0.9063 2012 92.26 0.9171
1991 91.00 0.8974 2013 92.35 0.9174
1992 91.04 0.8985 2014 91.57 0.9086
1993 91.17 0.9025 2015 91.99 0.9142
1994 91.19 0.9036 2016 92.22 0.9162
1995 91.23 0.9046 2017 91.56 0.9082
1996 91.44 0.9057 2018 90.80 0.8916
2.3.2 Evaluation of the simulation ability of the CMIP5 coupled models
The 1961-2018 meteorological observation station data, 1961-2018 simulated experimental data, and 2019-2029 simulated data were used as the observation, historical simulation, and future simulation fields, respectively. The completeness of the temperature and precipitation data for the selected meteorological observation stations allowed the data of 31 coupled models to be used (The CNRM-CM5 and IPSL-CM5A-LR models lack data for this study area). The deviations between the historical simulation and observation fields were measured using the root-mean-square error (RMSE). The standard deviation (SD) was used to measure the degree of dispersion of the historical simulation field or observation field. The correlation between the historical simulation field and the observation field was measured using the correlation coefficient (r), and Taylor plots (Taylor et al., 2001) were used to evaluate the simulation ability of the various models. For each model, the distances were calculated between each historical single model point and the corresponding observation point in 3D space (RMSE, SD, and r) in the Taylor plots. Shorter distances indicated better simulation ability. Subsequently, the systematic classification tool in SPSS Statistics software. SPSS was used to classify the calculated distances. The unweighted average of the shortest distance class was used as the model-set result. Similarly, the distance between each model-set point and the corresponding observation point was calculated, and the distances among all single model points, model-set points, and observation points were compared. The simulation results of the models (or model sets) with the shortest distances were selected for subsequent analysis.
Figure 2 shows the Taylor plots of the simulation and observation fields for two emission scenarios, 31 single models, and 1 model-set. The single model points were relatively concentrated but separated from the observation points by a consistent distance. Table 2 lists the models belonging to the shortest distance class, the corresponding distances between the model points and observation points, and the distances between the model-set points and observation points. Since the shortest distance was achieved between the model-set point and observation point for all meteorological elements, it can be deduced that the simulation results of the model sets provided a suitable approximation of the observed data.
Figure 2 Taylor plot of meteorological simulation field data relative to meteorological observation field data around the Zhalong Nature Reserve from 1961 to 2018 (a. Maximum air temperature under RCP4.5; b. Maximum air temperature under RCP8.5; c. Minimum air temperature under RCP4.5; d. Minimum air temperature under RCP8.5; e. Precipitation under RCP4.5; f. Precipitation under RCP8.5) (, , and symbolise pattern-point data, where is the type with the shortest distance from the observation-point data, and is the mode-set data, while n symbolises the observation-point data.)
Table 2 Distance between model points and observation points with different meteorological elements and emission scenarios in the Wuyur River basin
Meteorological element Emission scenarios Pattern Distance from model point to observation point Meteorological element Emission scenarios Pattern Distance from model point to observation point
Maximum temperature RCP4.5 CESM1-BGC 1.06 Minimum temperature RCP4.5 GFDL-ESM2G 1.13
CMCC-CM 1.07 GFDL-ESM2M 1.13
GISS-E2-H 1.04 MIROC5 1.10
GISS-E2-H-CC 1.03 NorESM1-M 1.10
GISS-E2-R 1.03 Multimodel set 0.96
GFDL-ESM2M 1.07 RCP8.5 CanESM2 1.12
MIROC5 1.03 GFDL-ESM2G 1.11
Multimodel set 0.79 NorESM1-M 1.13
RCP8.5 CESM1-BGC 1.12 Multimodel set 0.97
FIO-ESM 1.12 Precipitation RCP4.5 CMCC-CM 141.98
GISS-E2-H 1.11 EC-EARTH 141.53
GISS-E2-H-CC 1.09 GFDL-ESM2G 141.72
GISS-E2-R 1.08 Multimodel set 117.88
INM-CM4 1.13 RCP8.5 CESM1-BGC 140.94
Multimodel set 0.86 GFDL-ESM2G 137.90
MPI-ESM-LR 140.68
Multimodel set 116.40
2.3.3 Meteorological observation station data gridding
To improve the regional representativeness of meteorological observation data and CMIP5 climate prediction downscaling data, the meteorological factors were converted from point data to grid data with a spatial resolution of 500 m by using the Inverse Distance Weighted (IDW) tool of ArcGIS, corresponding to the NPP data simulated by the Carnegie- Ames-Stanford Approach (CASA) model in space. The meteorological factors include annual radiative aridity, potential evapotranspiration rate, annual mean biological temperature (℃), annual precipitation (mm), daily mean temperature less than 30℃ and greater than 0℃, and monthly mean temperature less than 30 ℃ and greater than 0℃ from 1975 to 2029.
2.3.4 Calculation of NPP
(1) Estimation of vegetation biomass in investigation quadrat
The root-shoot ratio (the ratio of biomass dry weight or fresh weight of underground part to the above-ground portion) of reed in the northeastern wetland in September was 1.487 (Jia et al., 2006). This was used to estimate the total biomass of reeds in the plots investigated in the field, and the NPP of reeds in each plot was then estimated according to the conversion coefficient of biomass dry weight to NPP, which is commonly used as 0.45 (Fang et al., 1996). The NPP of the reeds in each quadrat was estimated to be in the range 223.82-637.92 gC∙m-2∙a-1, with a mean of 423.15 Cg∙m-2∙a-1 and a SD of 87.74 gC∙m-2∙a-1.
(2) Screening the NPP calculation model
The annual net primary productivity (NPP) of Zhalong Nature Reserve was calculated using both a climatic productivity model and a remote sensing model. The climatic productivity model includes the Zhou Guangsheng-Zhang Xinshi model (Zhou et al., 1996), the Miami model (Leith et al., 1972), the Thornthwaite Memorial model (Leith et al., 1972), the Beijing model (Zhu et al., 1993) and the Chikugo model (Efimova, 1983). Remote sensing models include the CASA model (Potter et al., 1993) and the terrestrial ecosystem carbon flux (TEC) model (Yan et al., 2015). The NPP estimation results under the same grid were compared with the measured values on the ground, and the most suitable calculation method for Zhalong Nature Reserve NPP was selected.
The average measured NPP ($\overline{NPP}$) was 423.15 gC∙m-2∙a-1 (Figure 3), which is the closest to the CASA model estimate ($\overline{NPP}$ = 423.68 gC∙m-2∙a-1), followed by the Zhou Guangsheng-Zhang Xinshi model ($\overline{NPP}$ = 444.08 gC∙m-2∙a-1). The predicted values of the TEC model ($\overline{NPP}$ = 326.56 gC∙m-2∙a-1), Beijing model ($\overline{NPP}$ = 154.41 gC∙m-2∙a-1) and Thornthwait Memorial model ($\overline{NPP}$ = 229.56 gC∙m-2∙a-1) were much lower than the measured values, while the Chikugo model ($\overline{NPP}$ = 648.324 gC∙m-2∙a-1) and Miami model ($\overline{NPP}$ = 583.60 gC∙m-2∙a-1) were much higher than the measured values.
Figure 3 NPP Box distribution obtained using the actual measurement, and the CASA, TEC, Beijing, Chikugo, Thornthwaite Memorial, Miami and Zhou Guangsheng-Zhang Xinshi models, represented by distributions 1-8 respectively
The CASA model calculation result has the highest correlation coefficient with the measured value (R = 0.229), and has passed the significance test with a confidence level of 95% (sig. = 0.049) (Table 3). The results of an independent-sample t-test show that there was no significant difference between the CASA model and the Zhou Guangsheng-Zhang Xinshi model. The other model calculation results have no significant correlation with the measured value.
Table 3 Comparison between measured values and calculated results of various models
NPP acquisition method R Sig. Independent-sample t-test
CASA model 0.229 0.049 Sig.=0.969
TEC model -0.134 0.329 Sig.=0.002
Beijing model 0.049 0.722 Sig.<0.001
Chikugo model 0.055 0.692 Sig.<0.001
Thornthwaite Memorial model 0.135 0.327 Sig.<0.001
Miami model 0.134 0.329 Sig.<0.001
Zhou Guangsheng-Zhang Xinshi model 0.117 0.396 Sig.=0.077
(3) Zhou Guangsheng-Zhang Xinshi model
The 1975-2029 NPP of the vegetation of the Zhalong Wetland was estimated using the Zhou Guangsheng-Zhang Xinshi model, which is an NPP estimation model based on the water balance and heat balance equations, the observed NPP data and the corresponding meteorological elements, and the physiological and ecological characteristics of plants. The model is represented by the following equations:
$NPP=RD{{I}^{2}}\frac{r(1+RDI+RD{{I}^{2}})}{(1+RDI)(1+RD{{I}^{2}})}\exp [-{{(a+bRDI)}^{0.5}}]$
$RDI={{(c+dPER-ePE{{R}^{2}})}^{2}}$
$PER=\frac{58.93\cdot BT}{r}$
$BT=\frac{\sum t}{365}\ \text{or}\ BT=\frac{\sum T}{12}$
where NPP is the net primary productivity of the vegetation (tDM·hm-2·a-1)),
a, b, c, d, and e are coefficients (a = 9.87, b = 6.25, c = 0.629, d = 0.237, e = 0.00313) (Zhou et al., 1996),
RDI is the radiation drought index,
PER is the potential evapotranspiration rate,
BT is the average annual biological temperature (℃),
r is the annual precipitation (mm),
t is the daily average temperature (0℃ < t < 30℃),
T is the monthly average temperature (0℃ < T < 30℃).
(4) Parameter adjustment of Zhou Guangsheng-Zhang Xinshi model
The CASA model cannot be used to calculate NPP due to the difficulty in obtaining the early (1970s) monthly NDVI data. Earlier accurate meteorological observation data can be easily obtained, thus based on the results of previous research, we chose the Zhou Guangsheng-Zhang Xinshi model to calculate NPP. Since this model was originally constructed to estimate the distribution of NPP on regional to global scales, the differences in NPP among different vegetation types in small regions with minute differences in temperature and precipitation were not considered. Using the NPP data acquired from field surveys, the present study adopted the estimation results of the CASA model (Potter et al., 1993) as the reference standard. The land cover classification data of the Zhalong Nature Reserve was also used as the basis for adjusting the parameters of the Zhou Guangsheng-Zhang Xinshi model according to vegetation type. In particular, data from 2000-2009 were used to adjust the model parameters, and data from 2010-2018 were used for model validation.
The Zhou Guangsheng-Zhang Xinshi model is based on the relationship between the water and heat balance equations as well as the physiological and ecological characteristics of plants. Since the latter portion of the equation ($\text{exp }\!\![\!\!\text{ }-{{(a+bRDI)}^{0.5}}\text{ }\!\!]\!\!\text{ }$) is closely related to vegetation type, adjustments were made to the parameters a and b according to the following steps. (1) Using the Create Fishnet tool in ArcGIS, a sample point document with interpoint distances of 0.01° was created, and the NPP calculated by the CASA model and the Zhou Guangsheng-Zhang Xinshi model were extracted, which provided basic data for parameter adjustment and verification of the Zhou Guangsheng-Zhang Xinshi model. (2) The NPP values of the Zhalong Wetland were calculated using the CASA model with a spatial resolution of 500 m. For each sample point, the NPP value was extracted and recorded as NPPc. (3) The NPP values of the surrounding meteorological observation stations of the Zhalong Wetland were calculated using the Zhou Guangsheng-Zhang Xinshi model, and point data were converted to surface data using the IDW tool in ArcGIS. For each sample point, the NPP value was extracted and recorded as NPPz. (4) An independent samples t-test was performed with the NPPc and NPPz values of the herbaceous bog, grass, and crop. The results indicate that the mean NPPc and NPPz values of the grass was significantly different (P < 0.05), whereas the mean NPPc and NPPz values of the herbaceous bog and crop were not significantly different (P > 0.05). Therefore, the NPP estimation parameters were only adjusted for grass in the present study. (5) Using the nonlinear regression tool in SPSS, the regression was performed with NPPc as the dependent variable, NPPz as the independent variable, and the Zhou Guangsheng-Zhang Xinshi model as the model expression. The adjusted values of a and b obtained from the regression were 13.390 and 2.582, respectively. An independent samples t-test indicated that there was no significant difference in the mean NPPc and NPPz values of the grass when the adjusted a and b values were used (P > 0.05).
2.3.5 Landscape transformation analysis
In the present study, the extent of landscape transformation of land cover types in the Zhalong Nature Reserve was defined as the extent of conversion of a particular landscape type to other landscape types during the wetland landscape transformation process. The calculation is as follows:
$M=\sum{\{(j-i)\times {{s}_{ij}}\}}$
where M is the extent of landscape transformation; sij is the transformed area in the landscape transformation matrix during different years; i and j are the ith and jth landscape types, respectively; (j - i) is the transformation coefficient, which was defined according to the method reported by Zhang et al. (2015). M > 0 indicates a dry-to-wet landscape transformation direction, whereas M < 0 indicates a wet-to-dry landscape transformation direction. The absolute value of (j - i) is smaller for similar land cover types and larger for dissimilar land cover types, as shown in Table 4.
Table 4 Landscape conversion coefficient for the Zhalong Nature Reserve
Conversion coefficient Building land and
unused land		 Cultivated land Grass Herbaceous bog Water bodies
Building land and unused land 0 1 2 3 4
Cultivated land -1 0 1 2 3
Grass -2 -1 0 1 2
Herbaceous bog -3 -2 -1 0 1
Water bodies -4 -3 -2 -1 0
The landscape transformation coefficient was used as the primary landscape change indicator for the analysis of the dynamic changes in the landscapes within the Zhalong Nature Reserve through the following method. First, pairwise overlays were performed with landscape type maps of different years to determine the landscape transformation coefficient of each grid. Temporal variation sequences of the landscape transformation coefficients of all grid cells were determined using statistical methods, which were subsequently used to obtain the extents and directions of landscape transformation in the reserve.
2.3.6 Heterotrophic respiration model
Soil respiration (Rs) includes plant autotrophic respiration (Ra) and Rh (Werth et al., 2008). Since Rh is found to account for 46 (Hu et al., 2008) - 48% (Kelting et al., 1998) of Rs in meadows, an average value of 47% has been adopted in our study. Heterotrophic respiration rates were estimated using the simple empirical exponential model of soil respiration proposed by Van’t Hoff with the adoption of the model parameters fitted by Yang et al. (2011) for different land cover types in the Zhalong Nature Reserve, as shown in the following equations:
$R{h}'=1.32{{e}^{0.0500k}}\times 47.00% & grass \\ 0.99{{e}^{0.0620k}}\times 47.00% & Herbaceous\text{ }bog,crop$
where Rh′ is the heterotrophic respiration rate, which is a characterisation of CO2 flux (μmol·m-2·s-1), and k is the air temperature (℃).
2.3.7 Estimation of net ecosystem productivity
The NEP was estimated using the NPP and Rh values obtained from the previous sections, as shown in the following equation:
$NEP=NPP-Rh$
where NEP is the net ecosystem productivity, with NEP > 0 indicating a net carbon sink and NEP < 0 indicating a net carbon source, NPP is the net primary productivity (gC·m-2·a-1), and Rh is the heterotrophic respiration rate (gC·m-2·a-1).
2.3.8 Calculation of mean NPP, Rh and NEP values of Zhalong Wetland
The area-weighted sum method was used to calculate the mean NPP, Rh, and NEP values for the Zhalong Wetland, as shown in the following equation:
$y=\sum\limits_{i=1}^{n}{{{w}_{i}}{{x}_{i}}}$
where y is the mean NPP, Rh, or NEP value, wi is the percentage area of the ith land cover type, i.e., the percentage of the total meadow area accounted by grass or herbaceous bog, and xi is the NPP, Rh, or NEP value of the ith land cover type.
2.3.9 Rate of change
A linear regression equation was established using the time series of the meteorological elements by setting time as the independent variable and the meteorological elements as the dependent variables, as shown in the following equation:
$y(x) = b_1 x + b_0$
where y is a particular meteorological variable, x is time (year or series number), bl × 10 is the change tendency rate (℃·10a-1, mm·10a-1, or gC·m-2·10a-1).
The coefficient in the equation b0 can be determined using the least-squares method.
2.3.10 Principle and application of the geographic detector method
The geographic detector method measures the driving effect of a single factor or two-factor interaction by studying the differentiation characteristics of target factors and detecting the influence of factors and their interactions. The core of the model is to explain the degree of spatial differentiation of target factors by impact factors, measured by the q value, and the equation is given below (Wang et al., 2010):
$q=1-\frac{\sum\limits_{h=1}^{L}{{{N}_{h}}\sigma _{h}^{2}}}{N{{\sigma }^{2}}}$
where h = 1, 2, …; L is the stratification of impact factors; Nh and N are the number of units in the layer h and the whole area, respectively; and σ2h and σ2 are the variances of the layer h and the target value of the whole zone, respectively. The value range of q is [0, 1]. The closer the q value is to 1, the stronger the dominant driving effect of the impact factor is.
To correlate the carbon exchange components (NPP, Rh, NEP) and environmental factors (temperature, precipitation, land cover type) in space, we used ARCGIS10.2 to generate sampling points with equal spacing of 0.02° and extracted the carbon exchange components and environmental factors of the study area from 1975 to 2018. Each factor extracted 15759 valid points as the operation data of the geographic detector. The input variables of the geographical detector were required as category data, and the continuous variables need to be discretised. According to the value range of each meteorological factor, the factors are divided into ten grades at equal intervals (the maximum value of temperature is 2.54℃, the minimum value is 5.66℃, with an interval of 0.31℃; the maximum value of precipitation is 278.61 mm, the minimum value is 685.60 mm, with an interval of 40.70 mm). The land cover type is directly assigned “1” for cultivated land, “2” for grassland, and “3” for others.

3 Results and analysis

3.1 Climate variation characteristics of the Zhalong Nature Reserve

From 1975-2018, the average annual temperature of the Zhalong Nature Reserve (Figure 4a) varied within the range 2.54-5.66℃ with a mean value of 4.00 ± 0.77℃ and exhibited a significant increasing trend with a change tendency rate of 0.30°C/10a. Annual precipitation (Figure 4b) varied within the range 278.61-685.60 mm with a mean value of 451.46 ± 96.86 mm and exhibited a weaker increasing trend.
Figure 4 Variation of average annual temperature and annual precipitation in the Zhalong Nature Reserve from 1975 to 2018
Table 5 illustrates that the average temperature during the summer and autumn seasons showed a highly significant increasing trend over the last 44 years (P < 0.01), whereas the temperature during spring showed a significant increase (P < 0.05), and the temperature during the winter season did not exhibit a significant trend. The trending rate of climate change (i.e., rate of temperature increase) was the highest for the autumn season (0.35℃/10a) and the lowest for the winter season (0.25℃/10a). Precipitation exhibited a highly significant increasing trend in winter (P < 0.01), while the other seasons showed statistically insignificant increasing trends (P≥0.05).
Table 5 Variation of average temperature and precipitation in four seasons in the Zhalong Nature Reserve from 1975 to 2018
Meteorological factors Season Mean value Maximal value Minimum value Standard deviation Propensity of change
Air temperature (℃/10a),
Precipitation (mm/10a)		 P-value
Temperature (℃) Spring 5.88 3.40 8.54 1.30 0.33 0.031
Summer 21.67 19.65 23.49 0.85 0.30 0.002
Autumn 4.36 2.13 6.43 1.13 0.35 0.008
Winter -16.27 -20.10 -11.53 1.81 0.25 0.246
Precipitation (mm) Spring 57.31 16.47 135.56 29.28 3.80 0.279
Summer 317.10 159.56 526.33 85.01 7.89 0.441
Autumn 69.01 20.40 194.45 35.73 2.86 0.506
Winter 8.27 2.64 25.61 4.69 1.72 0.001

3.2 Variation in the characteristics of land cover types in the Zhalong Nature Reserve

3.2.1 Temporal variation characteristics
Figure 5 shows the variation in the area occupied by various land cover types in the Zhalong Nature Reserve from 1975-2018. Figure 5a shows that the area occupied by construction/ unused land varied within the range of 30.88-553.51 km2 with a mean value of 255.07 ± 156.92 km2, indicating a highly significant increasing trend with an average rate of 10.92 km2•a-1. Construction/unused land occupied a maximum area of 553.51 km2 in 2000. This was primarily attributed to the occurrence of severe drought in the Zhalong Nature Reserve and its primary water source, the Wuyur River basin, during 1999-2000, as well as a substantial fire event in 2000, which lasted for over 10 days and burned reeds down to the roots in the wetland (Cui, 2008). Consequently, the area of construction/unused land in 2000 was 2.17 times the mean value from 1975-2018.
Figure 5 Change in the area occupied by various land cover types in the Zhalong Nature Reserve from 1975 to 2018
The area occupied by farmland in the reserve varied within the range of 107.48-345.77 km2 with a mean value of 219.84 ± 90.92 km2 and showed a highly significant increasing trend, increasing at an average rate of 6.74 km2·a-1. Three distinct stages of increase could be observed over 44 years: (1) an average increase of 107.57 km2 during 1975-1982; (2) an average increase of 150.51 km2 during 1983-1995; and (3) an average increase of 298.08 km2 during 1996-2018 (Figure 5b).
Figure 5c shows that the area occupied by water bodies varied within the range of 51.36-330.07 km2 with a mean value of 123.58 ± 56.05 km2 and showed a weak decreasing trend. The maximum area, which was 1.56 times the mean value from 1975-2018, was observed in 1998.
The area occupied by grass varied within the range of 137.12-978.24 km2, with a mean value of 625.14 ± 162.42 km2. Figure 5d illustrates considerable inter-year fluctuations and a highly significant decreasing trend, at an average rate of 5.00 km2•a-1. Herbaceous bog occupied an area of 686.08-1544.35 km2 with a mean value of 1026.22 ± 224.05 km2 over the last 44 years, making this the land cover type occupying the largest area in the Zhalong Nature Reserve. The inter-year fluctuations of herbaceous bog were smaller than those of grass and a highly significant decreasing trend, with an average rate of 12.54 km2•a-1 was observed (Figure 5e). Figure 5f shows the variations in the total area of herbaceous bog and grass. The total meadow area, comprising herbaceous bog and grass, varied within the range of 1285.83-2002.29 km2, with a mean value of 1651.35±241.73 km2, exhibiting smaller inter-year fluctuations and a highly significant decreasing trend, with an average rate of 17.53 km2•a-1.
3.2.2 Spatial variation characteristics
To determine the general distribution of land cover types in the Zhalong Nature Reserve, the distribution maps of land cover types from 1975-2018 were overlaid. The general distribution was obtained by determining the land cover type with the highest distribution frequency in each cell, as shown in Figure 6. Herbaceous bog accounted for the most extensive land cover type, covering 1168.02 km2 or 51.84% of the total reserve area, followed by grass, covering 533.09 km2, or 23.66% of the reserve. Water bodies accounted for the smallest land area, covering only 88.96 km2, or 3.95% of the reserve.
Figure 6 Distribution of land cover types with the largest frequency in the Zhalong Nature Reserve from 1975 to 2018
When each eco-functional zone was analysed separately, the core zone was found to be comprised primarily of meadows (herbaceous bog and grass), covering 695.54 km2, or 94.09% of the area of the core zone, with herbaceous bog forming the predominant meadow type (78.74% of the area of the core zone). Meadows were also the primary land cover type in the buffer zone (533.84 km2, 76.40% of the area of the buffer zone), although the proportion covered by meadows was smaller than in the core zone. Similarly, herbaceous bog was also the predominant meadow type, covering 50.63% of the area of the buffer zone, although the proportion covered by herbaceous bog was lower than that of the core zone. In the experimental zone, meadows still accounted for a substantial area (471.73 km2, 57.87% of the experimental zone), but the proportion of meadow area was significantly reduced, while the proportion of the area occupied by farmland and construction land increased concurrently.
3.2.3 Characteristics of the landscape transformation
From the variations in their extent (Figure 7), the landscape transformations in Zhalong Nature Reserve over the last 44 years can be broadly divided into four stages. (1) From 1975-1983, the net extent of transformation of the wetland landscape was smaller (within -109.95 to 49.95 km2), and the cumulative extent of transformation remained below 0 (-28.25 to -192.29 km2), indicating a gradual transformation of the wetland landscape to a dry habitat during this stage. (2) From 1984-1991, the net extent of transformation increased significantly (within -384.83 to 626.33 km2). The cumulative extent of transformation was greater than zero throughout this period except from 1988-1989, which indicates a transformation to a wet habitat during this stage. (3) From 1992-2000, the net extent of transformation remained substantial (within -1343.50 to 598.26 km2), and the cumulative extent of transformation was below zero for most years, indicating a relatively sizeable extent of transformation to a dry habitat. (4) From 2001-2018, the net extent of transformation was within -227.21 to 244.01 km2, and the cumulative extent of transformation remained below 0 (-1318.52 to -940.54 km2), which indicates the continued transformation towards a dry habitat during this stage, although the extent of the transformation was reduced.
Figure 7 Landscape transformation range of the Zhalong Nature Reserve from 1975 to 2018
(The range of landscape transformation is the net change in the landscape drying-wetting transformation in the Zhalong Nature Reserve. The ascending curve indicates the development of wet habitat, and the descending curve indicates the development of a dry habitat. The cumulative landscape transformation range is the cumulative quantity of the landscape transformation range. Values > 0 indicate a wet habitat, while values < 0 indicate a dry habitat.)

3.3 Variation in the characteristics of the carbon exchange components in Zhalong Nature Reserve

3.3.1 Temporal variation in the characteristics of the carbon exchange components from 1975-2018
As shown in Table 6, the annual mean of NPP > mean Rh, and the annual mean of NEP > 0 during the last 44 years, indicating that the reserve served as a carbon sink. When NPP, Rh, and NEP were normalised for linear regression analysis, it was found that NPP and Rh exhibited significant (P < 0.05) and highly significant (P < 0.01) trends of increase, respectively. However, the normalised rate of change of NPP was lower than that of Rh, which signifies that the increasing trend in the carbon absorption rate was smaller than that of the carbon release rate. These results may explain the lack of significant temporal variation in NEP (P≥0.05).
Table 6 Statistical characteristics of average annual NPP, Rh, and NEP in the Zhalong Wetland from 1975 to 2018
Items NPP Rh NEP
Mean value (gC·m-2·a-1) 500.21 337.59 162.62
Standard deviation (gC·m-2·a-1) 52.76 10.80 45.56
Linear regression correlation coefficient 0.319 0.650 0.155
Normalised tendency rate (gC·m-2·10a-1) 0.06 0.14 0.003
Sample number 44 44 44
Significance 0.034 <0.001 0.314
3.3.2 Spatial variability in the characteristics of carbon exchange fractions from 1975 to 2018
It can be seen that the core zone of wetland had the highest NPP of 501.08 ± 53.24 gC∙m-2∙a-1, followed by the buffer zone, with an NPP of 500.40 ± 52.87 gC∙m-2∙a-1, while the experimental zone had the lowest NPP, at 499.14 ± 52.19 gC∙m-2∙a-1 (Figure 8a). The NPP in most areas showed an increasing trend (Figure 8b), and the areas showing an increasing trend accounted for 88.37% of the total area of the region. However, the areas exhibiting significant changes were fewer (Figure 8c), accounting for only 0.46% of the total area.
Figure 8 Spatial variation of NPP (a-c), Rh (d-f) and NEP (g-i) in the Zhalong Nature Reserve from 1975 to 2018
As can be seen in Figure 8d, the core zone had the smallest carbon emissions (average value was 285.65 ± 51.95 gC∙m-2∙a-1). The buffer zone had the next highest carbon emissions (302.21 ± 17.37 gC∙m-2∙a-1), with larger carbon emission patches embedded within the smaller carbon emission patches. The experimental zone was the largest carbon emissions area (424.90 ± 44.65 gC∙m-2∙a-1). Comprehensive analysis of Figures 8e and 8f shows that carbon emissions in most of the core zone and the northern part of the buffer zone showed a significant increasing trend (p < 0.05), and the area with a significantly increasing trend of Rh accounted for 54.04% of the total area of the reserve.
Analysis of Figure 8g showed that the NEP in the core region was the largest (215.43 ± 49.97 gC∙m-2∙a-1) and was more evenly distributed. The buffer zone had the second-highest NEP (198.19 ± 45.88 gC∙m-2∙a-1), with greater carbon emissions in the north and south than in the centre. The experimental zone has the smallest NEP (74.24 ± 41.31 gC∙m-2∙a-1), with larger carbon emission patches embedded in the smaller ones. Combining this information with Figures 8h and 8i, it can be seen that most of the NEP in the core zone showed an increasing trend, and the increasing and decreasing trends of NEP in the remaining areas showed a mosaic pattern. However, the change in most areas did not pass the significance test; in fact, the areas passing the significance test only accounted for 0. 53% of the total protected area.
3.3.3 Variation characteristics of carbon sequestration
Based on the data on average annual NPP and areas occupied by the various land cover types, the annual mean carbon sequestration by all vegetation types in the Zhalong Wetland was calculated for the various periods shown in Table 7. For 2019-2029, it was assumed that the mean areas of the various land cover types during the last 10 years (2009-2018) will be maintained, and the meteorological data obtained from simulations using the optimised CMIP model sets in Table 1 under the RCP 4.5 and RCP 8.5 scenarios were used for the estimation of carbon sequestration. This analysis showed that the annual mean carbon sequestration by the Zhalong Wetland from 1975-2018 was 2.467 (±0.950) × 1011 gC·a-1, with carbon sequestration showing a weak decreasing trend with time (P≥0.05). In particular, carbon sequestration was the highest during 1980-1989, at 1.65 times that from 2000-2009, which had the lowest carbon sequestration rate. The predicted carbon sequestration by the Zhalong Wetland during the next 10 years (2019-2029) under the RCP 4.5 and RCP 8.5 scenarios was lower than the annual mean carbon sequestration during the last 44 years, with lower carbon sequestration under RCP 8.5 than under RCP 4.5. These results indicate that climate change reduces the carbon sequestration ability and suggests that future climate change may provide a negative contribution to the carbon sequestration potential of the Zhalong Wetland.
Table 7 Current (1975-2018) and future (2019-2029) carbon sequestration in the Zhalong Wetland
Time Climate scenario Carbon sequestration (1011 gC·a-1) Standard deviation (1011 gC·a-1)
1975-1979 Reality 2.096 0.772
1980-1989 3.126 1.024
1990-1999 2.811 0.917
2000-2009 1.899 0.793
2010-2018 2.404 0.737
2019-2029 RCP4.5 2.421 0.225
RCP8.5 2.407 0.382

3.4 Responses of carbon exchange fractions to environmental factors

3.4.1 Response to the change of meteorological factors
Figure 9 shows that the annual NPP of the Zhalong Wetland had a highly significant positive linear correlation with annual precipitation but was not significantly correlated with the average annual temperature. The analysis of the individual seasons indicated that the annual NPP had a highly significant positive linear correlation with summer precipitation (y = 0.31x + 483.45, R2 = 0.7607, P < 0.001, n = 44) but had neither significant linear relationships with precipitation during the other seasons nor with the average temperatures of each season (P≥0.05). Rh had a highly significant linear correlation with the average annual temperature and a highly significant linear correlation (y = 12.00x + 87.86, R2 = 0.6342, P < 0.001, n = 44) and significant linear correlation (y = 5.01x + 325.76, R2 = 0.2046, P = 0.002, n = 44) with summer and autumn average temperatures, respectively. The annual NEP had a highly significant positive linear correlation with annual precipitation and no significant correlation with average annual temperature.
Figure 9 Relationship between NPP, Rh, NEP, and meteorological factors in the Zhalong Wetland from 1975 to 2018
The analysis of NEP and the individual seasons demonstrated that NEP was significantly correlated with summer precipitation (y = 0.60, R2 = 0.8194, P < 0.001, n = 44) but not with the precipitation during other seasons or temperature in any season. Therefore, we can deduce that variations in NPP and NEP were precipitation-driven, whereas the variations in Rh were temperature-driven.
3.4.2 Response to the change of land cover types
By analysing the relationship between NPP, Rh, NEP and cultivated land area in Zhalong Nature Reserve (Figures 10a-10c), it was found that there was a very significant linear positive correlation between Rh and cultivated land area (p < 0.001). When cultivated land area increased by 1 km2, Rh increased by 9.56 × 104 gC. There was no significant linear correlation among NPP, NEP, and cultivated land area. At the same time, there was a very significant negative linear correlation between Rh and herbaceous bog and grass area (p < 0.001), where Rh decreased by 3.40 × 104 gC when herbaceous bog and grass area increased by 1 km2 (Figures 10e). However, NPP and NEP had no significant linear correlation with the herbaceous bog and grass area (Figures 10d and 10f). Combined with the land cover types (Section 3.2.1) and carbon exchange components (Section 3.3.1) of the Zhalong Na-ture Reserve, it can be seen that increase of farmland area is a reason for the increase in Rh in Zhalong Nature Reserve.
Figure 10 Relationship between NPP, Rh, NEP and land cover types in Zhalong Wetland from 1975 to 2018
3.4.3 Interaction of meteorological factors and land cover change on carbon exchange components
It can be seen that the interaction of temperature and precipitation, and the interaction of precipitation and land cover type had a greater impact on carbon exchange components than any single environmental factor, and the impact of meteorological factors on carbon exchange components was greater than that of land cover type. It was also found that the interaction of two environmental factors had the greatest impact on NPP, followed by NEP and Rh, as shown in Table 8.
Table 8 Effects of interaction of environmental factors on carbon exchange components in Zhalong Nature Reserve
Dominant interaction NPP Rh NEP
Dominant interaction 1 q Temperature×Precipitation
0.881		 Temperature×Precipitation
0.218		 Temperature×Precipitation
0.854
Dominant interaction 2 q Precipitation×Land use type
0.862		 Precipitation×Land use type
0.116		 Precipitation×Land use type
0.842
Dominant interaction 3 q Precipitation
0.850		 Temperature×Land use type
0.106		 Precipitation
0.837
Dominant interaction 4 q Temperature×Land use type
0.008		 Temperature
0.104		 Temperature×Land use type
0.016
Dominant interaction 5 q Temperature
0.005		 Precipitation
0.088		 Temperature
0.015
Dominant interaction 6 q Land use type
0.002		 Land use type
0.001		 Land use type
0.001

The value range of q was [0, 1], and the closer the value of q was to 1, the stronger the dominant driving effect of the impact factors was. The number of samples involved in the calculation was n = 15759.

4 Conclusions and discussion

4.1 Discussion

(1) Comparison of heterotrophic respiration rates
Significant differences exist among the estimated heterotrophic respiration rates due to various influencing factors, such as vegetation type, soil type, climatic conditions, and estimation method. The comparison of the heterotrophic respiration values reported in several studies by Chinese researchers (Table 9) indicates that the results obtained for the same ecosystem type in various regions using different experimental methods were within the same order of magnitude. These results indicate that the estimated values were comparable. In addition, the estimated respiration rates of the present study were similar to the reported results for the same ecosystem type in neighbouring regions, which demonstrates the reliability of the model simulation results.
Table 9 Comparison of results of heterotrophic soil respiration
Land cover type Method Microbial heterotrophic respiration
(μmol·m-2·s-1)		 Research area
Farmland
ecosystem		 Closed-chamber soil carbon flux system (LI-8100) March to November: 0.79-1.20 Shaanxi Province (Zhang et al., 2019)
Root biomass
extrapolation		 Growing season: 1.11-1.96 Liaoning Province (Han et al., 2009)
Static chamber method Full year: 0.58 ± 0.08 for waterlogged fields, 0.75 ± 0.10 for dry fields Sanjiang Plain, Heilongjiang
Province (Hao et al., 2007)
Model calculations Growing season: 0.90-2.42 Present study
Model calculations Full year: 0.12-2.42 Present study
Forest ecosystem Closed-chamber alkali absorption method Full year: 0.59-1.37 Fujian Province (Yang et al., 2006)
Closed-chamber soil carbon flux system (LI-8100) Full year: 0.82-7.11 Fujian Province (Yang et al., 2018)
Alpine meadow Soil respiration chamber (Li6400-09) Growing season: 0.47-0.63 Tibetan Plateau (Zhang et al., 2006)
Static chamber method Growing season: 0.95-2.53 Tibetan Plateau (Hu et al., 2008)
Grassland Root biomass
extrapolation		 Growing season: 1.54-4.42 Inner Mongolia (Shi et al., 2014)
Wetland
ecosystem		 Indoor cultivation, gas chromatography Full year: 0.41 ± 0.22 for the
deposition promotion zone,
0.07 ± 0.02 for the natural state		 Chongming Island, Shanghai
(Tang et al., 2010)
Model calculations Growing season: 0.50-3.67 Present study
Model calculations Full year: 0.05-3.67 Present study
(2) Differences in optimum CMIP5 model across different study areas
Due to uncertainties in the simulations by global models, CMIP5 models must be evaluated for specific regions. In the present study, the CMIP5 models for the Zhalong Nature Reserve were evaluated using Taylor plots and a systematic classification was adopted to differentiate between the two greenhouse gas emission scenarios, RCP 4.5 and RCP 8.5, to screen the optimum models for maximum temperature, minimum temperature, and precipitation, and obtain model sets through the unweighted average method. The results of the validation demonstrated that the simulation ability of the model sets was superior to that of any single model, and similar conclusions were reported by relevant studies (Chen et al., 2013; Tao et al., 2016; Jiang et al., 2017). However, as the model simulation effects vary significantly across different regions (Yao et al., 2012), the optimum models for each region will differ as well.
(3) Assignment of weights to various models in CMIP5 model sets
In the present study, the abilities of the CMIP5 models to simulate the maximum temperature, minimum temperature, and precipitation of the Zhalong Nature Reserve were systematically evaluated, and models that showed an excellent ability to simulate the average meteorological observation fields were screened based on different RCPs and meteorological elements and combined to form model sets. Although this method partially enables the reduction of biases in the model simulations, it is unable to eliminate these biases. Furthermore, equal weights were assigned to the models during the formation of the model sets. In future research, data mining techniques can be employed for models with different weights to enhance the simulation ability of the model sets.
(4) Insufficient time series of remote sensing data affected the spatial estimation accuracy of NPP
In our study, actual data from field surveys were used for the evaluation of the simulation abilities of commonly used NPP estimation models. The results indicated that the CASA model and the Zhou Guangsheng-Zhang Xinshi model exhibited excellent ability to simulate the NPP of the Zhalong Wetland, with the CASA model being superior to the Zhou Guangsheng-Zhang Xinshi model in terms of data distribution. However, as we were unable to acquire monthly NDVI data for the 1970s, the CASA model could not be used in the present study. Although the estimation accuracy of the Zhou Guangsheng-Zhang Xinshi model was enhanced in this study through the adjustment of relevant parameters, the ability of the model to differentiate spatial differences in the NPP for the same land cover type remained limited, which affected the spatial simulation accuracy for the NPP.
(5) Analysis of biases in the results of future carbon sequestration potential estimation for Zhalong Wetland
During the estimation of the future carbon sequestration potential for the Zhalong Wetland, we assumed that the mean areas of the land cover types from the last 10 years (2009-2018) would be maintained. Since the NPP varies significantly among different land cover types and the CMIP5 models yield several errors, there will be uncertainties in the simulation of NPP based on the CMIP5 models. However, an evaluation of the results produced by the CMIP5 model sets has shown that the correct variation trends were obtained for minimum temperature, maximum temperature, and precipitation. Because the selected study area is a national nature reserve, it is not likely to undergo drastic changes in land cover and landscape patterns in the near future owing to the various policies implemented by the government to limit anthropogenic damage to the natural landscapes of the wetlands. Therefore, the estimations of this study can serve as a reference for future evaluations of carbon sequestration potential of the region. In addition, although the rationality of the results of NPP and Rh has been demonstrated in this paper, the NEP results still contain uncertainties because NEP is calculated using NPP and Rh and lacks supporting results. In the future, we can use eddy covariance measurements to decompose the net ecosystem exchange data to obtain the NEP, and the NEP obtained by the two methods can be compared, improving the accuracy of NEP calculation results.

4.2 Conclusions

(1) Herbaceous bog formed the land cover type with the largest area in the Zhalong Nature Reserve, covering a mean area of 1168.02 ± 224.05 km2 and accounting for 51.84% of the total reserve area. The total area of herbaceous bog exhibited a highly significant trend of decrease since 1975.
(2) The Zhalong Nature Reserve has undergone a dry-wet-dry landscape transformation process. Excluding several wet periods during the mid-1980s to the early 1990s, the reserve has remained a dry habitat, with especially severe conditions from 2000 onwards.
(3) From 1975-2018, the mean NPP of the Zhalong Wetland was 500.21±52.76 gC·m-2·a-1, with NPP values exhibiting a significant overall linear increase. The mean Rh was 337.59±10.80 gC·m-2·a-1, with overall Rh values showing a highly significant increasing trend. However, the trend of an increase in the carbon absorption rate was smaller than the increase in the carbon release rate.
(4) An evaluation of the carbon balance of the Zhalong Wetland in the last 44 years revealed that the wetland served as a net carbon sink. The annual mean carbon sequestration was 2.467 (± 0.950) × 1011 g C•a-1, and the carbon sequestration exhibited an overall weak decreasing trend. Variations in NPP and NEP were precipitation-driven, whereas the variations in Rh were temperature-driven.
(5) The influence of the interaction of temperature and precipitation on NPP, Rh and NEP was greater than that of any single meteorological factor. The influence of precipitation was dominant for NPP and NEP, while the effect of temperature was dominant for Rh.
(6) Under the RCP 4.5 and RCP 8.5 scenarios, the predicted carbon sequestration by the Zhalong Wetland during the next 10 years (2019-2029) was 2.421 (± 0.225) × 1011 gC•a-1 and 2.407 (± 0.382) × 1011 gC•a-1, respectively, which were both lower than the mean carbon sequestration during the last 44 years.
(7) Climate change over the past 40 years has negatively affected the carbon sequestration capacity of the Zhalong Wetland, and climate change in the next 10 years may result in a negative effect on the carbon sequestration capacity of the Zhalong Wetland. Simultaneously, we found that water was the key environmental factor affecting the carbon sequestration capacity of Zhalong Wetland. Therefore, improving the water environment of Zhalong Wetland, reducing the water consumption of human activities around the wetland, and increasing upstream water supplementation may an effective means to improve the carbon sequestration capacity of the wetland.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Allison S D, Czimczik C I, Treseder K K, 2008. Microbial activity and soil respiration under nitrogen addition in Alaskan boreal forest. Global Change Biology, 14(5):1156-1168.

DOI

[2]
Anderson F E, Bergamaschi B, Sturtevant C et al., 2016. Variation of energy and carbon fluxes from a restored temperate freshwater wetland and implications for carbon market verification protocols. Journal of Geophysical Research: Biogeosciences, 121(3):777-795.

DOI

[3]
Bellen S V, Dallaire P L, Garneau M et al., 2011. Quantifying spatial and temporal Holocene carbon accumulation in ombrotrophic peatlands of the Eastmain region, Quebec, Canada. Global Biogeochem Cycles, 25(2):1-15.

[4]
Bernal B, Megonigal J P, Mozdzer T J, 2017. An invasive wetland grass primes deep soil carbon pools. Global Change Biology, 23(5):2104-2116.

DOI

[5]
Burkett V, Kusler J, 2000. Climate change: Potential impacts and interactions in wetlands of the United States. Journal of the American Water Resources Association, 36(2):313-320.

DOI

[6]
Chen H P, 2013. Projected change in extreme rainfall events in China by the end of the 21st century using CMIP5 models. Chinese Science Bulletin, 58(12):1462-1472.

DOI

[7]
Chen W, Jiang Z, Li L, 2015. Probabilistic projections of climate change over China under the SRES A1B scenario using 28 AOGCMs. Journal of Climate, 24(17):4741-4756.

DOI

[8]
Cui L D, Zha L, 2008. Wetland in Heilongjiang Province owes 100 million yuan in water debt for five emergency water replenishment. Heilongjiang Life Daily, 2008-05-25. (in Chinese)

[9]
Efimova H A, 1983. Wang B Z trans. Radiation Factors of Plant Yield. Beijing: China Meteorological Press.

[10]
Fang J Y, Liu G H, Xu G L, 1996. Biomass and net production of forest vegetation in China. Acta Ecologica Sinica, 16(5):497-508. (in Chinese)

[11]
Gang C, Zhou W, Wang Z et al., 2015. Comparative assessment of grassland NPP dynamics in response to climate change in China, North America, Europe and Australia from 1981 to 2010. Journal of Agronomy and Crop Science, 201(1):57-68.

DOI

[12]
Guo Y D, He Y, Deng W et al., 2004. Research on eco-environmental water demand of Zhalong national natural wetlands. Journal of Soil and Water Conservation, 18(6):163-174. (in Chinese)

[13]
Han G X, Zhou G S, Xu Z, 2009. Seasonal dynamics of soil respiration and carbon budget of maize (Zea mays L.) farmland ecosystem. Chinese Journal of Eco-Agriculture, 17(5):874-879. (in Chinese)

DOI

[14]
Hao Q J, Wang Y S, Song C C et al., 2007. CO2 budget in agro-ecosystems in the Sanjiang Plain. Journal of Agro-Environment Science, 26(4):1556-1560. (in Chinese)

[15]
Hawkins C F, Smith O E, 1983. The convention on wetlands of international importance especially as waterfowl habitat. Environmental Policy and Law, 10(2):70-71.

[16]
Hossler K, Bouchard V, 2010. Soil development and establishment of carbon-based properties in created freshwater marshes. Ecological Applications, 20(2):539-553.

DOI

[17]
Howard J, Mcleod E, Thomas S et al., 2017. The potential to integrate blue carbon into MPA design and management. Aquatic Conservation: Marine and Freshwater Ecosystems, 27(Suppl.1):100-115.

DOI

[18]
Hu Q W, Wu Q, Cao G M et al., 2008. Growing season ecosystem respirations and associated component fluxes in two alpine meadows on the Tibetan Plateau. Journal of Integrative Plant Biology, 50(3):271-279.

DOI

[19]
IPCC, 2018. Special Report on Global Warming of 1.5℃ (SR15). Cambridge: Cambridge University Press, 785.

[20]
Ivanov A A, Kosobryukhov A A, 2018. Cooperation of photosynthetic and nitrogen metabolism. In: Fominan I R, Biel K Y, Soukhovolsky V G. Adaptation and Tolerance to Extreme Environments. Hoboken: John Wiley & Sons, Inc., 329-437.

[21]
Jia Q Y, Zhou L, Xie Y B et al., 2006. Study on biomass dynamics of Phragmites communis community in Panjin wetland. Journal of Meteorology and Environment, 22(8):25-29. (in Chinese)

[22]
Jiang S, Jiang Z H, Li W et al., 2017. Evaluation of the extreme temperature and its trend in China simulated by CMIP5 models. Climate Change Research, 13(1):11-24. (in Chinese)

[23]
Kelting D L, Burger J A, Edwards G S, 1998. Estimating root respiration, microbial respiration in the rhizosphere, and root-free soil respiration in forest soils. Soil Biology & Biochemistry, 30(7):961-968.

DOI

[24]
Leith H, Box E, 1972. Evapotranspiration and primary productivity. Publications in Climatology, 25(1):37-46.

[25]
Liu D L, Zuo H P, 2012. Statistical downscaling of daily climate variables for climate change impact assessment over New South Wales, Australia. Climatic Change, 115:629-666.

DOI

[26]
Luo J M, Yin X G, Ye Y J et al., 2014. Impacts of edaphic environments on water quality under artificial disturbance in the Zhalong wetland. Chinese Journal of Soil Science, 45(2):450-456. (in Chinese)

[27]
Ma S Y, Verheyen K, Props R et al., 2018. Plant and soil microbe responses to light, warming and nitrogen addition in a temperate forest. Functional Ecology, 32(5):1293-1303.

DOI

[28]
Miki T, Doi H, 2016. Leaf phenological shifts and plant-microbe-soil interactions can determine forest productivity and nutrient cycling under climate change in an ecosystem model. Ecological Research, 31(2):263-274.

DOI

[29]
Mueller P, Granse D, Nolte S et al., 2017. Top-down control of carbon sequestration: Grazing affects microbial structure and function in salt marsh soils. Ecological Applications, 27(5):1435-1450.

DOI

[30]
Mueller P, Laniges N, Jack A et al., 2019. Assessing the long-term carbon-sequestration potential of the semi-natural salt marshes in the European Wadden Sea. Ecosphere, 10(1):e02556.10.

[31]
Packalen M S, Finkelstein S A, Mclaughlin J W, 2016. Climate and peat type in relation to the spatial variation of the peatland carbon mass in the Hudson Bay Lowlands, Canada. Journal of Geophysical Research: Biogeosciences, 121(4):1104-1117.

DOI

[32]
Potter C S, Randerson J T, Field C B et al., 1993. Terrestrial ecosystem production: A process model based on global satellite and surface data. Global Biogeochemical Cycles, 7(4):811-841.

DOI

[33]
Richardson C W, Wright D A, 1984. WGEN: A model for generating daily weather variables. U.S. Department of Agriculture, Agricultural Research Service ARS-8.

[34]
Saunders M J, Kansiime F, Jones M B, 2012. Agricultural encroachment: Implications for carbon sequestration in tropical African wetlands. Global Change Biology, 18(4):1312-1321.

DOI

[35]
Schippers P, Sterck F, Vlam M et al., 2015. Tree growth variation in the tropical forest: understanding effects of temperature, rainfall and CO2. Global Change Biology, 21(7):2749-2761.

DOI PMID

[36]
Shi J J, Geng Y B, 2014. Study on the distinguishing of root respiration from soil microbial respiration in a Leymus chinensis steppe in Inner Mongolia, China. Environmental Science, 35(1):341-347. (in Chinese)

DOI

[37]
Sim T G, Swindles G T, Morris P J et al., 2019. Pathways for ecological change in Canadian High Arctic wetlands under rapid twentieth century warming. Geophysical Research Letters, 46(9):4726-4737.

DOI

[38]
Tang Y S, Wang L, Jia J W et al., 2010. The effect of artificial disturbance such as siltation promotion on soil microbial respiration of tidal wetlands in the Yangtze River estuary. Acta Ecologica Sinica, 30(18):5022-5032. (in Chinese)

[39]
Tao C W, Jiang C, Sun J X, 2016. Projection of future changes in climate in Northeast China using a CMIP5 multi-model ensemble. Chinese Journal of Geophysics, 59(10):3580-3591. (in Chinese)

[40]
Taylor K E, 2001. Summarizing multiple aspects of model performance in a single diagram. Journal of Geophysical Research Atmospheres, 106(D7):7183-7192.

DOI

[41]
Wang J F, Li X H, Christakos G et al., 2010. Geographical detectors-based health risk assessment and its application in the neural tube defects study of the Heshun region, China. International Journal of Geographical Information Science, 24(1):107-127.

DOI

[42]
Wang Q K, Liu S G, Tian P, 2018. Carbon quality and soil microbial property control the latitudinal pattern in temperature sensitivity of soil microbial respiration across Chinese forest ecosystems. Global Change Biology, 24(7):2841-2849.

DOI

[43]
Werth M, Kuzyakov Y, 2008. Root-derived carbon in soil respiration and microbial biomass determined by 14C and 13C. Soil Biology & Biochemistry, 40(3):625-637.

DOI

[44]
Xin X G, Wu T W, Zhang J, 2012. Introduction of CMIP5 experiments carried out by BCC climate system model. Progressus Inquisitiones de Mutatione Climatis, 8(5):378-382. (in Chinese)

[45]
Yan H, Wang S Q, Billesbach D et al., 2015. Improved global simulations of gross primary product based on a new definition of water stress factor and a separate treatment of C3 and C4 plants. Ecological Modelling, 297:42-59.

DOI

[46]
Yang K, Liu G D, Liu F et al., 2011. Soil restoration in riparian zone of Zhalong wetland, northeastern China. Earth Science Frontiers, 18(6):94-100. (in Chinese)

[47]
Yang Y S, Chen G S, Xie J S et al., 2006. Soil heterotrophic respiration in native Castanopsis kawakamii forest and monoculture Castanopsis kawakamii plantations in subtropical China. Acta Pedologica Sinica, 43(1):53-61. (in Chinese)

[48]
Yang Z J, Zheng Y X, Chen S D et al., 2018. High resolution temporal analysis of heterotrophic respiration in the soil of a subtropical forest by using continuous wavelet transformation. Acta Ecologica Sinica, 38(14):5078-5086. (in Chinese)

[49]
Yao Y, Luo Y, Huang J B, 2012. Evaluation and projection of temperature extremes over China based on 8 modeling data from CMIP5. Progressus Inquisitiones de Mutatione Climatis, 8(4):250-256. (in Chinese)

[50]
Yu C L, Liu D, 2018. Spatiotemporal evolution of land cover types and response to climate change in Zhalong Wetland. Ecology and Environmental Sciences, 27(11):2117-2126. (in Chinese)

[51]
Yu Z L, Genet H, He Y J et al., 2018. The role of environmental driving factors in historical and projected carbon dynamics of wetland ecosystems in Alaska. Ecological Applications, 28(6):1377-1395.

DOI

[52]
Zhang D Q, Shi P L, He Y T et al., 2006. Quantification of soil heterotrophic respiration in the growth period of alpine steppe-meadow on the Tibetan Plateau. Journal of Natural Resources, 21(3):458-464. (in Chinese)

[53]
Zhang Y H, Su L Y, Yu W H et al., 2015. Characteristics of landscape dynamic changes in Zhalong wetland. Acta Geographica Sinica, 70(1):131-142. (in Chinese)

[54]
Zhang Y J, Guo S L, 2019. Effect of environmental factors on variation characteristics of soil microbial respiration and its temperature sensitivity. Environmental Science, 40(3):1446-1456. (in Chinese)

[55]
Zhao T B, Chen L, Ma Z G, 2014. Simulation of historical and projected climate change in arid and semiarid areas by CMIP5 models. Chinese Science Bulletin, 59(4):412-429.

DOI

[56]
Zhao Z C, Luo Y, Huang J B, 2019. Thinking again on global warming of 1.5℃ after the IPCC published SR15. Climate Change Research, 15(2):212-216. (in Chinese)

[57]
Zhou G S, Zhang X S, 1996. Study on NPP of natural vegetation in China under global climate change. Acta Phytoecologica Sinica, 20(1):11-19. (in Chinese)

[58]
Zhou X J, Wang F L, Wu Y Y et al., 2013. Analysis of temperature change characteristics of Heilongjiang Province, Northeast China and whole country in recent 60 years. Journal of Natural Disasters, 22(2):124-129. (in Chinese)

[59]
Zhu Z H, 1993. Estimation model of net primary productivity of natural vegetation. Chinese Science Bulletin, 38(15):1422-1426. (in Chinese)

Options
Outlines

/

〈 〉