Differences in functional traits of D. fruticosa between two habitats
Differences in the fine root functional traits
Compared to the shady slope, the SRL and SRA on the sunny slope were larger by 117.5% and 7.5%, respectively (Fig.1a 1c). Compared to the sunny slope, the RAD and RTD on the shady slope were significantly larger (p<0.05) by 60% and 56.7%, respectively (Fig.1b 1d). Results clearly indicated significant differences between the SRL and RAD on the shady and sunny slopes (p<0.05) (Fig.1b 1c).
Differences in the leaf functional traits
Our results demonstrated that LMA, Nmass, Pmass changed significantly(p<0.05) between the shady and sunny slopes (Fig. 2a, 2b, 2c). Additionally, Nmass, Pmass, PPUE, PNUE on the shady slope were greater by 54.44%, 89.38%, 14.34% and 17.14% over the sunny slope (p<0.05) (Fig. 2b, 2c, 2d, 2e). LMA on the shady slope was smaller by 42.31% compared with sunny slope (p<0.05) (Fig. 2a).
Differences in population characteristics and soil physicochemical properties of D. fruticosa in different habitats
Our result indicated that plant density, height and cover changed significantly (p<0.05) between the two habitats. Density, height and cover of D. fruticosa on the shady slope were 2.46, 0.19 and 7.85 times higher than the sunny slope (Table S1). One-way AVONA analysis also showed that shady slopes have higher SOC, STP, STN, pH, demonstrating significant differences (p<0.05) between the shady and sunny slopes (Table S2).
Differences in branch and leaf traits of D. fruticosa between two habitats
Differences in the branch vulnerability curve between shady and sunny slopes
From the stem vulnerability curves for the shady and sunny slopes, it can be seen that the P50 value is -0.97MPa on shady slope (Fig. 3a), and -1.43MPa on the sunny slope (Fig. 3b). This results indicate a higher embolism resistant capacity for plants growing on the sunny slope compared to those on the shady slope.
Differences in hydraulic and photosynthetic traits
Leaf and stem traits vary significantly between the sunny and shady slopes. KL, KS, HV, π0, ΨL, Aa, Am, Tr, Pn, gs on the sunny slope were significantly lower than for the shady slope (p<0.05) (Table 1). However, ε and WD on the shady slope was significantly lower than for the sunny slope (p<0.05). WUE on the shady slope was also lower than for the sunny slope(p>0.05)(Table 1).
Table 1 Differences in hydraulic and photosynthetic traits between shady and sunny slope (mean ± SE). See Abbreviations for the following traits.
Traits
|
units
|
Shady slope
|
Sunny slope
|
t
|
p
|
KS
|
g m-1 s-1 MPa-1
|
26.60±9.70
|
11.64±3.52
|
-5.240
|
0.000
|
KL
|
g m-1 s-1 MPa-1
|
1.10±0.31
|
0.03±0.02
|
-12.441
|
0.000
|
HV
|
cm2 cm-2
|
4.59±1.85
|
2.62±1.92
|
-2.967
|
0.012
|
WD
|
g cm-3
|
0.63±0.09
|
0.71±0.07
|
2.767
|
0.011
|
π0
|
MPa
|
-0.59±0.14
|
-1.03±0.17
|
-7.406
|
0.000
|
ε
|
MPa
|
15.71±2.47
|
22.46±2.40
|
7.192
|
0.000
|
ΨL
|
MPa
|
-1.79±0.48
|
-2.10±0.31
|
-1.996
|
0.057
|
Aa
|
umol.m-2. s-1
|
21.64±1.10
|
15.74±1.28
|
-12.867
|
0.000
|
Am
|
nmol. g-1. s-1
|
222.49±75.01
|
118.76±28.58
|
-4.675
|
0.000
|
Tr
|
mmol.m-2. s-1
|
4.69±0.29
|
2.93±0.46
|
7.199
|
0.000
|
Pn
|
umol.m-2. s-1
|
13.25±0.67
|
8.76±1.30
|
6.850
|
0.000
|
gs
|
mol.m-2. s-1
|
0.49±0.03
|
0.28±0.07
|
5.925
|
0.000
|
WUE
|
umol. mmol-1
|
2.82±0.26
|
3.13±1.01
|
1.111
|
0.277
|
Ci:Ca
|
/
|
0.88±0.01
|
0.86±0.03
|
1.270
|
0.240
|
Soil factors influencing vegetation cover and variations in fine-roots functional traits
Our results indicated that the presence of D. fruticosa shrubs was associated with significant differences in soil nutrients (Fig. S1). SWC, SOC, STN, STP and pH were all positively related to the degree of shrub cover (Fig. S1).
The PCA showed that D. fruticosa in different habitats are mainly affected by SRL and RAD (Fig. 4, Table S3). In addition, the stepwise multiple regression analyses showed that STN and STP were determined as the most affected factors on RAD and SRL (Table 2).
Table 2 Summarized results of the stepwise multiple regression analyses with root functional traits (SRL, SRA, RAD, RTD) as dependent variables and soil nutrients (SOC, STP, STN, SWC, pH) as independent variables. Coefficients (β) and P-values for each independent variable as well as coefficient of determination (R2) and P-values and F-values in each model are shown. STP and STN were predictors and SWC, SOC and pH were excluded variables in model. See Abbreviations for soil properties.
Relationships between functional traits
Our result showed that KL and Ks were positively related to Aa, Am, Pmass, Nmass, gs, and RAD (p<0.05) (Fig. 5e-l) and negatively correlated with SRL, WD, RWCtlp, ε, π0 (Table S4). Whereas WD was positively correlated with ε, π0 (p<0.05) (Table S4). SRL was negatively correlated with RAD, Pmass and RTD (p<0.01) (Fig. 5b-5d). There was no significance correlation with RAD and RTD (p>0.05) (Table S4). Additionally, SRL was negatively related to Aa and gs (p<0.05), and RAD was positively related to Pmass, Aa, Am and gs (p<0.05) (Table S4). LMA was negatively related to KL, KS, Am, PPUE, PNUE (Fig 5n-5p) and positively correlated with RWCtlp, ε, π0(p<0.05) (Fig. 5m) (Table S4).
Principal component analysis between functional traits and shrub encroachment of different habitats
The Principal Component Analysis (PCA) indicated that PC1 and PC2 explained 58.50% of the total variance. The first PC accounted for 43.5% of the total variation, the second PC accounted for 15.0% of the total variation (Table 3). The PC axis 1 was loaded by Cover, KL, KS, HV, Aa, Am, Pn, gs, Pmass, Nmass, PPUE, PNUE, RAD, RTD. the PC axis 2 was loaded by π0, ε, RWCtlp, ΨL, SRA, SRL, LMA, WD and WUE (Fig. 6). D. fruticosa on the shady slope was clustered to the right of the PCA plot, whereas D. fruticosa on the sunny slope was clustered to the left of the PCA plot. These results indicated that D. fruticosa shrubs on shady slopes were positively correlated with KL, KS, Aa, Am, gs, Pmass, whereas D. fruticosa shrubs on sunny slopes were postively correlated with π0, ε, RWCtlp (Fig. 6). Additionally, tested traits of D. fruticosa on sunny and shady slopes were distributed along the diagonal line between the PC axis 1 and 2, meaning that the functional traits of different organs together determine the different adaptive strategies by which D. fruticosa are able to flourish in different habitats (Fig. 6).
Table 3 The Eigenvalue of each axis of PCA ordination of different habitats and correlation coefficients of functional traits with each axis
Principal component
|
PC1
|
PC2
|
PC3
|
Eigenvalue
|
10.00
|
3.45
|
2.35
|
Percentage of total variance (%)
|
43.50
|
15.01
|
10.22
|
Cumulative percentage (%)
|
43.50
|
58.50
|
68.72
|
Factor loading
|
|
|
|
KL
|
0.30
|
0.07
|
-0.02
|
KS
|
0.22
|
0.13
|
-0.11
|
WD
|
-0.16
|
-0.05
|
0.15
|
HV
|
0.20
|
-0.01
|
0.08
|
RWCtlp
|
-0.23
|
0.20
|
-0.08
|
π0
|
-0.29
|
0.09
|
0.15
|
ε
|
-0.26
|
-0.02
|
0.28
|
ΨL
|
-0.15
|
0.20
|
-0.26
|
SRL
|
-0.18
|
-0.30
|
-0.10
|
SRA
|
-0.06
|
-0.36
|
-0.15
|
RAD
|
0.19
|
0.13
|
0.17
|
RTD
|
0.02
|
0.27
|
-0.07
|
Nmass
|
0.22
|
-0.14
|
-0.39
|
Pmass
|
0.25
|
0.22
|
-0.22
|
LMA
|
-0.17
|
0.35
|
0.11
|
Aa
|
0.30
|
0.02
|
0.13
|
Am
|
0.26
|
-0.26
|
-0.04
|
PNUE
|
0.11
|
-0.20
|
0.39
|
PPUE
|
0.04
|
-0.49
|
0.17
|
gs
|
0.26
|
0.11
|
0.21
|
WUE
|
-0.04
|
-0.08
|
-0.21
|
Ci:Ca
|
0.15
|
0.12
|
0.47
|