Ecological Modelling 306 (2015) 16–23

Contents lists available at ScienceDirect

Ecological Modelling
journal homepage: www.elsevier.com/locate/ecolmodel

Surface modeling of ecosystem responses to climatic change in Poyang

Lake Basin of China
Tian-Xiang Yue a, *, Zheng-Ping Du a , Ming Lu a,c , Ze-Meng Fan a , Chen-Liang Wang a ,
Yong-Zhong Tian b , Bing Xu c,d
a
State Key Laboratory of Resources and Environment Information System, Institute of Geographical Sciences and Natural Resources Research, 11A, Datun
Road, Anwai, Beijing 100101, China
b
School of Geographical Sciences, Southwest University, China
c
School of Environment, Tsinghua University, Beijing 100084, China
d
College of Global Change and Earth System Science, Beijing Normal University, Beijing 100875, China

A R T I C L E I N F O A B S T R A C T

Article history: Surfaces of mean annual temperature and mean annual precipitation during the period from 1951 to
Available online 27 September 2014 2010 in Jiang-Xi province (Poyang Lake Basin) of China are simulated by means of a method for high
accuracy surface modeling (HASM), using data from 106 weather observation stations scattered over and
around Poyang Lake Basin. Methodologically, we analyzed errors of HASM by comparing with classical
Keywords: methods. The analysis indicated that HASM has a much higher accuracy than the classical methods. The
Climate change
simulation results from HASM show that mean annual temperature was respectively 17.05
C and 17.46
C
HASM
in the periods from 1951 to 1980 (P1) and from 1981 to 2010 (P2). Mean annual precipitation was
Accuracy
Holdridge life zone
1602 mm and 1718 mm, respectively in the P1 and P2. In other words, climate has become warmer and
Mean center wetter in recent 60 years in general. The ecosystems are very sensitive to relatively small changes in
Response surface meteorology. Warm temperate moist forest and subtropical moist forest were the dominant HLZ
types, accounted for 94.99% of total area of Jiang-Xi province. The proportion of warm temperate moist
forest decreased from 30.81% to 16.84%, while subtropical moist forest increased from 64.79% to 76.73%.
The mean centers of the rare HLZ types, cool temperate wet forest, cool temperate rain forest and
subtropical wet forest, moved respectively 291 km, 104 km and 122 km. Warm temperate wet forest and
subtropical moist forest had a little shift, moved respectively 21 km and 17 km during the period from
P1 to P2.
ã 2014 Elsevier B.V. All rights reserved.

1. Introduction changes in the composition and productivity of these forests (Reed

and Desanker, 1992). An individual-based forest simulator
Warming of the climate system is unequivocal, and since the suggested that forest zones could be shifted on the order of
1950s, many of the observed changes are unprecedented over 500–1000 m in elevation, which could lead to the local extirpation
decades to millennia (IPCC, 2013). Simulating ecosystem responses of some high-altitude species (Urban et al., 1993).
to climatic change is a central interest of terrestrial ecosystem Global warming created a zigzag pattern of biomass distribu-
studies. The related studies can be exhibited in chronological tion along a latitudinal gradient, i.e., an increase in the cooler-side
sequence as follows. boundary of forest types and a decrease in the warmer-side
A simulated result suggested that forests may respond boundary (Kohyama and Shigesada, 1995). The timing and
dramatically and rapidly to global warming, with significant and magnitude of future changes in forest systems will depend on
readily observable changes in forests of mid-latitudes occurring by environmental factors such as a changing global climate, an
the turn of the century or shortly thereafter (Botkin and Nisbet, accumulation of CO2 in the atmosphere, and increase global
1992). A study in northern Michigan of USA indicated that climate mineralization of nutrients such as nitrogen and sulfur (Dixon and
changes could lead to ecologically and commercially significant Wisniewski, 1995). Depending on the global change scenario used,
4–9 species would potentially move out of the United States to the
north. Nearly half of the species assessed (36 out of 80) showed the
* Corresponding author. Tel.: +86 10 64889041. potential for the ecological optima to shift at least 100 km to the
E-mail address: [email protected] (T.-X. Yue). north (Iverson and Prasad, 1998). A simulation by an ecosystem

http://dx.doi.org/10.1016/j.ecolmodel.2014.09.015
0304-3800/ ã 2014 Elsevier B.V. All rights reserved.
 T.-X. Yue et al. / Ecological Modelling 306 (2015) 16–23 17

process model with a spatial landscape model indicated that boreal from 1938 to 2009 in northeastern Siberia and the rising
species would disappear from the landscape in 200–300 years and temperatures over the second half of the 20th century enhanced
approximately same amount of time for a southern species to plant growth along the northeast Eurasian boreal margin (Berner
become common (He et al., 1999). Changes in both climate and CO2 et al., 2013). Results from evaluating the relative impact of climate
will lead to changing phenological relationships between intro- and economic changes on mountain regions in Switzerland imply
duced and native forest insects and their host trees, which could that forest will be strongly influenced by the direct impact of
either exacerbate or attenuate damage caused by these pests climate change, but that changes in land use will have a
(Simberloff, 2000). comparatively small impact (Bodin et al., 2013). Responses of
Response surface simulations indicated that changes in the individual trees and whole-forest ecosystems to CO2 and climate
potential distribution of tree and shrub taxa in North America in manipulations often vary by age, implying that forests of different
response to future climate change will be large and that ranges will ages will respond differently to climate change (Anderson et al.,
shift not only northward and upward in elevation but in all 2013). Time-series remote sensing imagery of Poyang Lake from
directions (Shafer et al., 2001). The simulations for GCM-projected 2000 to 2012 showed that areas of submerged vegetation
future climate scenarios with doubled atmospheric CO2 concen- expanded with a significant average annual change rate of
tration predicted that broadleaf forests would increase, but conifer 23.51 km2 (Chen et al., 2014).
forests, shrubs and grasses would decrease; and that deciduous In this paper, a method for high accuracy surface modeling
forests would have the largest relative increase, but evergreen (HASM) is used to create surfaces of mean annual temperature
shrubs would have the largest decrease (Yu et al., 2002). A meta- (MAT), mean annual precipitation (MAP) and potential evapo-
analysis showed that the range limits of species have moved on transpiration ratio (PER); then Holdridge life zone (HLZ) model is
average 6.1 km per decade northward (Parmesan and Yohe, 2003). operated on these HASM-created surfaces to simulate spatial
A review indicated that tropical forest responses to ongoing distribution of ecosystems. This approach is termed as HASM-HLZ,
environmental change are complex because effects of many which is a combination of deterministic and statistically based
simultaneously changing environmental factors are integrated models.
by the plants, and their responses can involve significant lags and
non-linearities (Clark, 2007). A simulation inference is that the 2. Material and methods
forest as a whole may be buffered from extreme drought events
that only affect early life-history stages (McMahon et al., 2009). 2.1. Data
Simulation from perspectives of ecophysiology, resource availabil-
ity, and climate change showed that boreal forest responses to 106 meteorological stations were available over and around the
climate change are complex due to the inherent edaphic variations Jiang-Xi province during the period from 1951 to 2010 (Fig. 1), of
and ecophysiological responses (Ise and Moorcroft, 2010). [(Fig._1)TD$IG]
Increases in the frequency, duration, and/or severity of drought
and heat stress associated with climate change could fundamen-
tally alter the composition, structure, and biogeography of forests
(Allen et al., 2010).
According to results from a combination of an integrated
environmental model and climate envelope models, more than
35% of the species composition in 2100 will be new in Northern
Europe and up to 25% of the species now present will have
disappeared in Southern Europe under the climatic circumstances
forecasted for 2100 (Alkemade et al., 2011). Warming climate
might eventually transform the subalpine Abies fabric forest into
Betula utilis forest that is widely distributed in the eastern Tibetan
Plateau mountainous areas with the relatively lower elevation,
while subalpine forests might move to higher and colder areas,
which are currently tundra (Wang et al., 2011). According to
TRIPLEX-Management model, climate change at the southern edge
of the boreal forest in North America may enhance the C
sequestration rate if there are no severe drought events (Wang
et al., 2012). An investigation in Peninsular Spain indicated that
recent warming has reduced growth rate and C accumulation,
especially in wet areas (Veyreda et al., 2012). Simulation of current
and future distributions of 15 tree species in Iberian Peninsula
showed that large changes in species ranges and forest communi-
ties might occur, not only at high elevations within Mediterranean
mountains but also along the entire elevation gradient throughout
this region, particularly at low and mid-elevations (Ruiz-Labour-
dette et al., 2012).
It is widely predicted that the geographic distributions of tree
species and forest types will undergo substantial shifts in future
(Vanderwel and Purves, 2013). Forest mortality related to climate
change is an increasingly common global phenomenon (Kerhoulas
et al., 2013). Tropical forests and subtropical forests are threatened
by their lack of resilience against long-term climate change owing
to predicted reductions in rainfall and increased tree mortality Fig. 1. DEM and meteorological stations (marked as dots) scattered over and around
(Zhou et al., 2013). Mean summer temperatures increased 1.0
C Jiang-Xi province.
18 T.-X. Yue et al. / Ecological Modelling 306 (2015) 16–23

which 85 stations are located within Jiang-Xi province and trend surface to the residue surfaces optimized respectively by
21 stations around Jiang-Xi province. These stations can provide HASM, OK, IDW and Spline (Yue et al., 2013). A cross-validation
daily temperature and precipitation data. The mean annual method is employed to test the target surface of HASM by
temperature (MAT) and the mean annual precipitation (MAP) at comparing with the ones of OK, IDW and Spline.
every meteorological station are calculated in terms of the daily Cross-validation is comprised of four steps: (i) 10% of the
temperature and precipitation data during the period from 1951 to meteorological stations in Jiang-Xi province were removed for
2010. We downscaled a digital elevation model (DEM) of Jiang-Xi validation prior to model creation, (ii) MAT and MAP from 1951 to
province at a spatial resolution of 90 m  90 m from the Shuttle 2010 are respectively simulated at a spatial resolution of 250 m
Radar Topography Mission (SRTM) (http://srtm.csi.cgiar.org). The  250 m using the remaining 90% of meteorological stations in
DEM at a spatial resolution of 250 m  250 m is developed by Jiang-Xi province and 21 stations around Jiang-Xi province,
means of a cubic convolution resampling approach. The DEM was (iii) MAE and MRE are calculated using the 10% validation set,
used as auxiliary data to interpolate MAT and MAP from the and (iv) the 10% validation set is returned to the pool of available
stations to grid cells at the spatial resolution of 250 m  250 m. station for the next iteration. This process is repeated until MAT
and MAP at all meteorological stations have been simulated and
2.2. Mathematical models the simulation error statistics for each station can be calculated.
The average errors are formulated as follows,
Three mathematical models are employed in analyzing !
1 X X
J 8
responses of ecosystems to climate change, which are the MAE ¼ jsf i;j  f i;j j (5)
improved-HLZ model (Yue et al., 2005), a scaling diversity model 8  J j¼1 i¼1
(Yue and Li, 2010) and a mean center model (Yue et al., 2007). The
HLZ model is defined by mean annual biotemperature (MABT) in
!
1 X X
degrees centigrade, mean annual precipitation (MAP) in milli- J 8
sf i;j  f i;j
meters, and potential evapotranspiration ratio (PER) logarithmi- MRE ¼ j j (6)
8  J j¼1 i¼1 f i;j
cally, i.e.,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi where MAE represents mean absolute error; MRE mean relative
di ðx; yÞ ¼ ðTðx; yÞ  T i Þ2 þ ðPðx; yÞ  Pi Þ2 þ ðPEðx; yÞ  PEi Þ2 (1) error; sfi,j is the simulated value of the ith validation grid cell in the
jth sampling group; fi,j is the observed value of the ith validation
where T(x,y) = ln MABT(x,y), MABT(x,y) is MABT at a grid cell of (x, grid cell in the jth sampling group; J represents the repeat times.
y); P(x,y) = ln MAP(x,y), MAP(x,y) represents MAP at a grid cell of (x, The cross-validation results indicate that HASM has a much
y); PE(x,y) = ln PER(x,y), PERðx; yÞ ¼ 58:93MABTðx; yÞ=MAPðx; yÞ higher accuracy comparing with the classic methods such as IDW,
represents PER at a grid cell of (x,y); Ti,Pi and PEi are respectively OK and Spline (Table 1). During the period from 1951 to 2010, MAE
standards of MABT logarithm, TAP logarithm and PER logarithm at of the MAT simulated by HASM is 0.27
C and lower respectively 4%,
the central point of the i’th life zone in the hexagonal system of 7% and 30% than the ones by IDW, OK and Spline; MAE of the MAP
HLZs; when dk ðx; y; tÞ ¼ mini fdi ðx; y; tÞg, the site (x,y) is classified simulated by HASM is 37.21 mm and less 11%, 6% and 41% than the
into the k’th life zone. ones by IDW, OK and Spline, respectively.
The scaling diversity model is formulated as,

e;r;tÞ 
!2 3. Results and discussion
X
mð 1=2
ln pi ðe; r; tÞ
3.1. Change trend of mean annual temperature
Dðe; r; tÞ ¼  i¼1
(2)
lne
In terms of the simulation results from HASM, MATs were
where pi(e,r,t) is proportion of area of the ith ecotope to area of the
17.05
C in the period from 1951 to 1980 and 17.46
C from 1981 to
whole investigation region; m(e,r,t) is total number of HLZs under
2010. The MAT increased by 0.41
C. Although the difference of
investigation; t represents time; e = (e + A)1, A is area of the
MAT between the grid cell with the highest MAT and the grid cell
investigation region measured by hectare; r is spatial resolution of
with lowest MAT is 12.27
C in P1 and 12.92
C in P2, 99.975% of
the used data set; and e equals 2.71828.
grid cells became warmer in Jiang-Xi province, in which P1 and
The mean center model is formulated as,
P2 represent the two sub-periods from 1951 to 1980 and 1981 to
X Ij
sij ðtÞ  X ij ðtÞ 2010, respectively.
xj ðtÞ (3) In the sub-period of P1, grid cells at which MAT was less than
Sj ðtÞ
i¼1 16
C mainly distributed in Jiu-Ling mountains, Mu-Fu mountains,
Huai-Yu mountains, Wu-Yi mountains, Wu-Gong mountains, Luo-
Xiao mountains and Yu mountains, accounting for 14.73% of total
X
Ij
sij ðtÞ  Y ij ðtÞ area of Jiang-Xi province. Grid cells at which MAT was warmer than
yj ðtÞ ¼ (4)
i¼1
Sj ðtÞ 18
C mainly located in valleys and hilly areas of Gan-Zhou and Ji-
An, accounting for 16.58% of the total area. MAT in most part of
where t is the variable of time; Ij(t) is patch number of the HLZ type
j; sij(t) is area of the ith patch of the HLZ type j; Sj(t) is total area of
HLZ type j; (Xij(t), Yij(t)) is longitude and latitude coordinate of the
geometric center of the ith patch of the HLZ type j; (xj(t), yj(t)) is the
Table 1
mean center of HLZ type j. Comparative analysis of HASM errors.

2.3. Cross-validation Weather element Errors HASM IDW Kriging Spline

MAT MAE(
C) 0.27 0.28 0.29 0.35
A trend surface is created by operating the geographically MRE (%) 1.72 1.83 1.86 2.16
MAP MAE (mm) 37.21 41.42 39.37 52.54
weighted regression on the data transformed by Box–Cox MRE (%) 2.29 2.53 2.39 3.20
approach. Target surfaces are defined respectively by adding the
 T.-X. Yue et al. / Ecological Modelling 306 (2015) 16–23 19
[(Fig._2)TD$IG]
Jiang-Xi province was between 16
C and 18
C, accounting for 59%
(Figs. 2 and 3).
In the sub-period of P2, grid cells less than 16
C distributed in
Jiu-Ling mountains, Mu-Fu mountains, Huai-Yu mountains, Wu-Yi
mountains, Wu-Gong mountains and Luo-Xiao mountains as
before, but they accounted for only 10.57% of total area, shrunk
back 28.24% comparing with the one in P1. In addition to Gan-Zhou
and Ji-An, MAT became higher than 18
C at many grid cells in
Ying-Tan Jing-De-Zhen, Fu-Zhou, Nan-Chang, Yi-Chun and Xin-Yu,
which accounted for 34.78% of the total area and increased by
109.77% comparing with the ones in P1. MAT in 55% of the total
area remained between 16 and 18
C.

3.2. Change trend of mean annual precipitation

MAPs in the two sub-periods of P1 and P2 were respectively

1602 mm and 1718 mm, increased by 116 mm from the period of P1
to P2. In P1, the minimum MAP at the driest grid cells was 1288 mm
and the maximum MAP at the wettest grid cells was 2557 mm. In
P2, the minimum one was 1283 mm and the maximum was
2583 mm. The difference between the minimum and the maxi-
mum was 1269 mm in P1, but 1300 mm in P2.
In P1, grid cells with MAP less than 500 mm were mainly located
in basins and valleys of Jiu-Jiang, Nan-Chang, Ji-An, Yi-Chun and
Gan-Zhou, accounting for 24.96% of the total area of Jiang-Xi
province (Fig. 4a). Grid cells with MAP higher than 1800 mm
located in mountainous areas of Huai-Yu, Wu-Yi, Jiu-Ling, Luo-Xiao
and Yu mountains, accounting for 9.39% of the total area. MAP was
between 1500 mm and 1800 mm in the most part of Jiang-Xi
province, about 65.65% of the total area.
In P2, grid cells where MAP was lower than 1500 mm mainly
distributed in southern Ji-An, northern Gan-Zhou, and around Lu-
Shan mountain in northern Jiu-Jiang, of which area accounted for
Fig. 2. The districts of Jiang-Xi province.
5.59% of total area and decreased by 77.6% comparing with the one
in P1. Grid cells with MAP higher than 1800 mm mostly located in

[(Fig._3)TD$IG]

Fig. 3. (a) MAT at a spatial resolution of 250 m  250 m in the sub-period of P1, (b) surface of MAT change from P1 to P2.
 20 T.-X. Yue et al. / Ecological Modelling 306 (2015) 16–23
[(Fig._4)TD$IG]

Fig. 4. (a) MAP in the sub-period of P1 at a spatial resolution of 250 m  250 m, (b) surface of MAP change from P1 to P2.

Jin-De-Zhen, Shang-Rao Ying-Tan Fu-Zhou and Yi-Chun as well as 3.3. Climate change trend
mountainous areas of Jiu-Ling, Mu-Fu and Luo-Xiao mountains, of
which area accounted for 28.95% of the total area and increased by The biggest MAT increase happened in the northern and the
208.36% comparing with the one in P1. The area with MAP between eastern Jiang-Xi as well as in Gan-Zhou (Fig. 3b). But some grid
1500 mm and 1800 mm accounted for 65.46%. cells in Luo-Xiao mountains had a MAT decrease. There was the

[(Fig._5)TD$IG]

Fig. 5. Spatial pattern of HLZ ecosystems: (a) in the period of P1; (b) in the period of P2.
 T.-X. Yue et al. / Ecological Modelling 306 (2015) 16–23 21

Table 2
Area change of ecosystems in Jiang-Xi province.

HLZ type 1951–1980 1981–2010 Area change (km2) Area change rate (%)
2 2
Area (km ) Proportion (%) Area (km ) Proportion (%)
CTWF 1092.65 0.66 12.87 0.01 1079.87 98.82
CTRF 77.87 0.05 79.24 0.05 1.37 1.76
WTMF 51443.15 30.81 28113.08 16.84 23332.18 45.35
WTWF 6165.13 3.69 10640.16 6.37 4475.44 72.59
STMF 108167.6 64.79 128089.8 76.73 19923.94 18.42
STWF 0.56 0.00 11.87 0.01 11.32 2021.43

largest MAP rise in Jing-De-Zhen, valley flat of eastern Shang-Rao province; 13.63 km2 of CTRF was converted into WTWF, distributing
Mu-Fu mountains and Jiu-Ling mountains as well as in some areas in high altitude areas of Huai-Yu mountains and Wu-Yi mountains.
of Nan-Chang (Fig. 4b), while MAP became smaller in Huai-Yu Caused by becoming warmer and wetter at grid cells accounting
mountains, Yu mountains and Wu-Yi mountains as well as in some for 0.002% of Jiang-Xi province, 3.06 km2 of WTMF transferred into
areas of southern Jiang-Xi province. STWF distributing in middle altitude areas of Wu-Yi mountains.
In Jiang-Xi province, 96.815% of grid cells became warmer and Owing to the decrease of MAT and the increase of MAP at grid cells
wetter, 3.16% warmer and drier, 0.024% cooler and wetter, and accounting for 0.005% of total area of Jiang-Xi province, 8.75 km2 of
0.001% cooler and drier. The grid cells becoming warmer and drier WTWF changed to CTRF, distributing sporadically in high altitude
distributed in Huai-Yu mountains, Wu-Yi mountains, Yu moun- areas of Luo-Xiao mountains. Because of the decease of MAP,
tains and some areas of southern Jiang-Xi province; the ones 0.56 km2 of STWF converted into STWF distributing sporadically in
becoming cooler and wetter appeared in high altitude areas of Luo- mountainous areas of southern Gan-Zhou; 12.06 km2 of CTRF
Xiao mountains; the ones cooler and drier were in high altitude changed into CTWF and mainly distributed in high altitude areas
areas of Wu-Yi mountains. of Wu-Yi mountains.
82.81% of the ecosystems were changed because of warming,
15.29% because of wetting,1.58% because of drying, 0.28% because of
3.4. Conversion of ecosystems
warming and drying, and 0.01% because of warming and wetting
(Fig. 6, Table 3).
6 HLZ types appeared in Jian-Xi province. They were cool [(Fig._6)TD$IG]
temperate wet forest (CTWF), cool temperate rain forest (CTRF),
warm temperate moist forest (WTMF), warm temperate wet forest
(WTWF), subtropical moist forest (STMF) and subtropical wet
forest (STWF) (Fig. 5). In terms of the scaling diversity index,
ecotope diversity decreased from 0.061 in P1 to 0.055 in P2.
In the period of P1, CTWF was 1092.75 km2, mainly distributed
in alpine areas of Jiu-Ling, Mu-Fu and Wu-Gong mountains as well
as Lu-Shan mountain; in P2, CTWF shrank to 12.88 km2, of which
area decreased by 99%; larger part of the changed CTWF was
converted into WTMF and smaller part of the changed CTWF was
converted into CTRF. CTRF appeared in alpine areas of Jiu-Ling, Luo-
Xiao, Wu-Yi and Huai-Yu mountains; its area was increased from
77.88 km2 in P1 to 79.25 km2 in P2 (Table 2).
WTMF and STMF were dominant ecosystems, accounting for
96% of the total area in P1 and 94% in P2; they distributed in plains
and low mountain areas of Jiang-Xi province. WTMF decreased
from 51,447.81 km2 in P1, which accounted for 33% of the total area,
to 28,115.63 km2 in P2, which accounted for 17% of the total area.
Most of the changed WTMF was converted into STMF. STMF was
expanded from 108,177.44 km2 in P1 to 128,101.38 km2 in P2; STMF
proportion increased from 65% to 77% of the total area.
WTWF mostly distributed in subalpine areas of Jiu-Ling, Mu-Fu,
Wu-Yi, Huai-Yu and Luo-Xiao mountains. Its area was expanded
from 6165.91 km2 in P1 to 10,641.13 km2. Much MTMF and CTWF
was converted into WTWF because of the increases of MAT and
MAP. WTWF increased by 73% from P1 to P2. STWF was the most
stable ecosystem in Jiang-Xi province and had almost no change in
the past 60 years.
Owing to impact of warming mainly, 1073.6 km2 of CTWF was
converted to WTWF, distributing in high altitude areas of Lu
mountain, Mu-Fu mountains, Jiu-Ling mountains and Wu-Gong
19866.63 km2 of WTMF was converted to STWF, distributing in low
hills of northern, western and middle Jiang-Xi province; 8.81 km2 of
WTWF was changed to STWF, distributing in southern Gan-Zhou.
Because it became warmer and drier at grid cells accounting for
0.042% of Jiang-Xi province, 56.75 km2 of WTWF was changed to Fig. 6. Conversion of ecosystem types between the periods from 1951 to 1980 and
STMF, which mainly distributed in southern boundary of Jiang-Xi from 1981 to 2010.
22 T.-X. Yue et al. / Ecological Modelling 306 (2015) 16–23

Table 3 mountains. STWF shifted 122 km toward southwest and its

Conversion of ecosystem types between the periods from 1951 to 1980 and from
location moved from east of Gan-Zhou to southwestern Gan-Zhou.
1981 to 2010 (unit: number of grid cells).
In short, the ecosystems are very sensitive to climate change
CTWF CTRF WTMF WTWF STMF STWF such as warming and wetting in Jiang-Xi province. Changes and
CTWF 13 193 0 0 0 0 potential changes of the ecosystems are similar to many parts of
CTRF 293 835 0 140 0 0 the world such as Europe, North America, Siberia and Asia. CTWF
WTMF 0 0 443657 6193 0 0
and CTRF moved to higher and colder areas. The forests have
WTWF 17178 218 61593 91269 0 0
STMF 0 0 317866 908 1730839 9
buffered from extreme drought and extreme cold events.
STWF 0 0 49 141 0 0
4. Conclusions

3.5. Shift of ecosystem mean center

In Jiang-Xi province, MAT and MAP were respectively 17.23
C
and 1661 mm from the year of 1951 to 2010. MATs were 7.45
C and
Mean center of CTWF was located in northern Jiu-Ling
19.74
C, respectively at the grid cells with the lowest temperature
mountains in P1. It moved 291 km toward southeast in P2 and
and the highest temperature. MAP was 1367 mm at the grid cell
then located in the northern Wu-Yi mountains, which means that
with the minimum precipitation and 2461 mm at the grid cell with
most of CTWF in Jiu-Ling, Mu-Fu and Wu-Gong mountains as well
maximum precipitation. Climate became in general warmer and
as Lu mountain disappeared in P2, but mainly distributed in Wu-Yi
wetter during the periods from P1 to P2. MAT became 0.41
C
mountains (Fig. 7). CTRF was distributed in the northwestern Yu
higher and MAP increased by 116 mm in the 60 years.
mountains in P1 and shifted 104 km toward southwest in P2; then
Owing to the climate change, areas of 6 HLZ types had a great
it was located in the southeast of Wu-Gong mountains, which
change. Ecotope diversity decreased from 0.061 to 0.055. CTWF
indicates that area of CTRF had considerably decreased in Wu-Yi,
had the contraction in area from 1092.65 km2 in P1 to 12.87 km2,
Huai-Yu and Yu mountains. The mean centers of WTMF and STMF,
which almost disappeared. STWF expanded from almost none, of
which were the dominant ecosystems in Jiang-Xi province, had a
which 0.56 km2 might be created by error, to 11.87 km2. STRF had
little shift; they moved respectively 21 km toward southwest and
almost no change. Proportion of WTWF increased from 3.69% to
17 km toward north during the period from P1 to P2. WTWF shifted
6.37%.
79 km toward northwest in P2 comparing with in P1, and moved
In terms of the average results from the year of 1951 to 2010,
from northeastern Yu mountains to northeast of Wu-Gong
WTMF and STMF were the dominant HLZ types in Jiang-Xi province.
They accounted for 94.99% of the total area of Jiang-Xi province. The
[(Fig._7)TD$IG] proportion of WTMF decreased from 30.81% in P1 to 16.84% in P2,
while STMF increased from 64.79% in P1 to 76.73% in P2.
The mean centers of the dominant HLZ types were relatively
stable, while the ones of the rare HLZ types had a considerable
shift. CTWF moved 291 km toward southeast, CTRF 104 km toward
west and STWF 122 km toward southwest.
Our further work includes, (1) the expected change in biomes
by the end of this century under consideration of the climate
scenarios in Jiang-Xi province, and (2) the biomes-change trend is
to be verified with historical investigation data and time-series
remote sensing imagery.

Acknowledgments

This work is supported by National High-tech R&D Program of

the Ministry of Science and Technology of the People’s Republic of
China (2013AA122003), by National Basic Research Priorities
Program (2010CB950904) of Ministry of Science and Technology
of the People’s Republic of China, and by CSTC (2010BB0146). We
would like to acknowledge the two anonymous reviewers.

References

Alkemade, R., Bakkenes, M., Eickhout, B., 2011. Towards a general relationship
between climate change and biodiversity: an example for plant species in
Europe. Reg. Environ. Change 11, S143–S150.
Allen, C.D., Macalady, A.K., Chenchouni, H., et al., 2010. A global overview of drought
and heat-induced tree mortality reveals emerging climate change risks for
forests. For. Ecol. Manage. 259, 660–684.
Anderson, T.K.J., Miller, A.D., Mohan, J.E., Hudiburg, T.W., Duval, B.D., Delucia, E.H.,
2013. Altered dynamics of forest recovery under a changing climate. Global
Change Biol. 19, 2001–2021.
Berner, O.T., Beck, P.S.A., Bunn, A.G., Goetz, S.J., 2013. Plant response to climate
change along the forest-tundra ecotone in northeastern Siberia. Global Change
Biol. 19, 3449–3462.
Bodin, J., Badeau, V., Bruno, E., Cluzeau, C., Moisselin, J.M., Walther, G.R., Dupouey, J.
L., 2013. Shifts of forest species along an elevation gradient in Southeast France:
climate change or standmaturation? J. Veg. Sci. 24, 269–283.
Botkin, D.B., Nisbet, R.A., 1992. Forest response to climatic change: effects of
parameter estimation and choice of weather patterns on the reliability of
Fig. 7. Shift of mean center of ecosystem. projections. Clim. Change 20, 87–111.
 T.-X. Yue et al. / Ecological Modelling 306 (2015) 16–23 23

Chen, L.F., Jin, Z.Y., Michishita, R., Cai, J., Yue, T.X., Chen, B., Xu, B., 2014. Dynamic Shafer, S.L., Bartlein, P.J., Thompson, R.S., 2001. Potential changes in the distributions
monitoring of wetland cover changes using time-series remote sensing of Western North America tree and shrub taxa under future climate scenarios.
imagery. Ecol. Inf. 24, 17–26. Ecosystems 4, 200–215.
Clark, D.A., 2007. Detecting tropical forests' responses to global climatic and Simberloff, D., 2000. Global climate change and introduced species in United States
atmospheric change: current challenges and a way forward. Biotropica 39, forests. Sci. Total Environ. 262, 253–261.
4–19. Urban, D.L., Harmon, M., Halpern, C.B., 1993. Potential response of Pacific
Dixon, R.K., Wisniewski, J., 1995. Global forest systems: an uncertain response to Northwestern forests to climatic change, effects of stand age and initial
atmospheric pollutants and global climate change? Water Air Soil Pollut. 85, composition. Clim. Change 23, 247–266.
101–110. Veyreda, J., Martinez-Vilalta, J., Gracia, M., Retana, J., 2012. Recent climate changes
He, H.S., Mladenoff, D.J., Crow, T.R., 1999. Linking an ecosystem model and a interact with stand structure and management to determine changes in tree
landscape model to study forest species response to climate warming. Ecol. carbon stocks in Spanish forests. Global Change Biol. 18, 1028–1041.
Modell. 114, 213–233. Vanderwel, M.C., Purves, D.W., 2013. How do disturbances and environmental
IPCC, 2013. In: Stocker, T.F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S.K., Boschung, J., heterogeneity affect the pace of forest distribution shifts under climate change?
Nauels, A., Xia Bex, Y.V., Midgley, P.M. (Eds.), Climate Change 2013: The Physical Ecography 36, 1–11.
Science Basis. Contribution of Working Group I to the Fifth Assessment Report of Wang, X.D., Cheng, G.W., Zhong, X.H., 2011. Assessing potential impacts of climatic
the Intergovernmental Panel on Climate Change. Cambridge University Press, change on subalpine forests on the eastern Tibetan Plateau. Clim. Change 108,
Cambridge, United Kingdom and New York, NY, USA. 225–241.
Ise, T., Moorcroft, P.R., 2010. Simulating boreal forest dynamics from perspectives of Wang, W.F., Peng, C.H., Kneesh, D.D., Larocquec, G.R., Song, X.Z., Zhou, X.L., 2012.
ecophysiology, resource availability, and climate change. Ecol. Res. 25, 501–511. Quantifying the effects of climate change and harvesting on carbon dynamics of
Iverson, L.R., Prasad, A.M., 1998. Predicting abundance of 80 tree species following boreal aspen and jack pine forests using the TRIPLEX-Management model. For.
climate change in the eastern United States. Ecol. Monogr. 68 (4), 465–485. Ecol. Manage. 281, 152–162.
Kerhoulas, L.P., Kolb, T.E., Hurteau, M.D., Koch, G.W., 2013. Managing climate change Yu, M., Gao, Q., Liu, Y.H., Xu, H.M., Shi, P.J., 2002. Responses of vegetation structure
adaptation in forests: a case study from the U.S. Southwest. J. Appl. Ecol. 50 (6), and primary productionof a forest transect in eastern China to global change.
1311–1320. Global Ecol. Biogeogr. 11, 223–236.
Kohyama, T., Shigesada, N., 1995. A size-distribution-based model of forest Yue, T.X., Fan, Z.M., Liu, J.Y., 2005. Changes of major terrestrial ecosystems in China
dynamics along a latitudinal environmental gradient. Vegetation 121, 117–126. since 1960. Global Planet. Change 48, 287–302.
McMahon, S.M., Dietze, M.C., Hersh, M.H., et al., 2009. A predictive framework to Yue, T.X., Fan, Z.M., Liu, J.Y., 2007. Scenarios of land cover in China. Global Planet.
understand forest responses to global change. Ann. N.Y. Acad. Sci. 1162, Change 55 (4), 317–342.
221–236. Yue, T.X., Li, Q.Q., 2010. Relationship between species diversity and ecotope
Parmesan, C., Yohe, G., 2003. A globally coherent fingerprint of climate change diversity. Ann. N.Y. Acad. Sci. 1195, E40–E51.
impacts across natural systems. Nature 421, 37–42. Yue, T.X., Zhao, N., Ramsey, R.D., Wang, C.L., Fan, Z.M., Chen, C.F., Lu, Y.M., Li, B.L.,
Reed, D.D., Desanker, P.V., 1992. Ecological implications of projected climate change 2013. Climate change trend in China, with improved accuracy. Clim. Change 120,
scenarios in forest ecosystems in northern Michigan, USA. Int. J. Biometeorol. 137–151.
36, 99–107. Zhou, G.Y., Peng, C.H., Li, Y.L., Liu, S.Z., Zhang, Q.M., Tang, X.L., Liu, J.X., Yan, J.H.,
Ruiz-Labourdette, D., Nogues-Bravo, D., Ollero, H.S., Schmitz, M.F., Pineda, F.D., 2012. Zhang, D.Q., Chu, G.W., 2013. A climate change-induced threat to the ecological
Forest composition in Mediterranean mountains is projected to shift along the resilience of a subtropical monsoon evergreen broad-leaved forest in Southern
entire elevation gradient under climate change. J. Biogeogr. 39, 162–176. China. Global Change Biol. 19, 1197–1210.