Rainfall
Temperature
Maximum temperature
Minimum temperature
700
35
600
30
500
25
400
20
300
15
200
10
100
5
0
0
J
(b)
F
M
A
M
Irradiance
J
J
A
S
O
N
D
Day-length
600
16
14
Irradiance (MJ/m 2)
500
12
400
10
300
8
6
200
4
100
Day length (hr)
Precipitation (mm)
(a)
Temperature (ºC)
Figures
2
0
0
J
F
M
A
M
J
J
A
S
O
N
D
Fig. S1. Main climatic variables at the Fushan subtropical rain forest, Taiwan (22-year
average, 1991–2012). (a) Mean monthly rainfall and temperature (including mean,
maximum, and minimum temperature); (b) total monthly irradiance and mean
monthly day length. All climatic variables were measured at the Fushan
meteorological station, located 3 km to the east of Fushan FDP.
1
Fig. S2. Monthly time series of the local climatic variables (including mean,
maximum, and minimum temperature, rainfall, and irradiance) at the Fushan
subtropical rain forest, Taiwan, and the ENSO indices (including ENSO34, ENSO4,
and SOI indices) from September 2002 to August 2012.
2
Fig. S3. Correlation matrix for the five local climatic variables (mean, maximum, and
minimum temperature, rainfall, and irradiance) at the Fushan subtropical rain forest.
The correlation coefficients and 95% confidence limits are listed in the upper
triangular matrix. The 95% confidence limits is computed by using stationary
bootstrap (5000 times). Correlation coefficients significantly different from zero are
indicated in red. The lower triangular matrix is the scatter plots between the paired
climatic variables.
3
Fig. S4. Cross correlations between the number of species flowering (left
panel)/fruiting (right panel) and the local climatic variables recorded in the Fushan
weather station (panels from top to bottom: irradiance, rainfall, mean temperature,
maximum temperature, and minimum temperature). Significant and insignificant
correlations are indicated by solid and open circles, respectively. Shaded areas are the
95% confidence limits.
4
Fig. S5. Cross correlations between the number of species flowering (left
panel)/fruiting (right panel) and the indices of the El Niño Southern Oscillation
variables (panels from top to bottom: ENSO34, ENSO4, SOI). Significant and
insignificant correlations are indicated by solid and open circles, respectively. Shaded
areas are the 95% confidence limits.
5
Fig. S6. Cross correlations between the local climatic variables recorded in the
Fushan weather station (panels from top to bottom: irradiance, rainfall, mean
temperature, maximum temperature, and minimum temperature) and the indices of the
El Niño Southern Oscillation variables (panels from left to right: ENSO34, ENSO4,
SOI). Significant and insignificant correlations are indicated by solid and open circles,
respectively. Shaded areas are the 95% confidence limits.
6
Fig. S7. Frequency distributions of the best predictive time window of six climatic
variables for flower (a) and seed (b) production. Different time windows are indicated
by different colors.
7
Fig. S8. Phylogenetic distribution of correlations between flower production and
climatic variables on the PHYLOMATIC phylogenetic tree for 44 studied species.
Branches are shaded in proportion to the weighted mean of trait values of descendent
tips simply for the illustration, and ancestral states should not be over-interpreted.
Species abbreviation can be found in Table S1.
8
Fig. S9. Phylogenetic distribution of correlations between seed production and
climatic variables on the PHYLOMATIC phylogenetic tree for 28 studied species.
Branches are shaded in proportion to the weighted mean of trait values of descendent
tips simply for the illustration, and ancestral states should not be over-interpreted.
Species abbreviation can be found in Table S1.
9
Fig. S10. Cross correlations between the number of species flowering (left
panel)/fruiting (right panel) and the local climatic variables with different
moving-average windows (panels from top to bottom: irradiance, rainfall, mean
temperature, maximum temperature, and minimum temperature). Significant and
insignificant correlations are indicated by solid and open circles, respectively. Shaded
areas are the 95% confidence limits.
10
Fig. S11. Cross correlations between the local climatic variables recorded with
different moving-average windows (panels from top to bottom: irradiance, rainfall,
mean temperature, maximum temperature, and minimum temperature) and the indices
of the El Niño Southern Oscillation variables (panels from left to right: ENSO34,
ENSO4, SOI). Significant and insignificant correlations are indicated by solid and
open circles, respectively. Shaded areas are the 95% confidence limits.
11