1
Table S1
2
Pearson's correlation between soil properties and rice yield
Yie2005
Yie2006
Yie2007
Yie2008
Yie2009
Yielda
Nitrate
0.578
0.590
0.674
0.619
0.634
0.639
Ammonium
-0.662
-0.490
-0.471
-0.650
-0.685
-0.611
Available phosphorous
0.857
0.752
0.711
0.858
0.930
0.849
pH
-0.392
-0.298
-0.196
-0.447
-0.503
-0.383
Total carbon
0.691
0.771
0.800
0.700
0.600
0.731
Total nitrogen
0.594
0.720
0.744
0.638
0.561
0.670
Available potassium
-0.154
-0.064
0.088
-0.174
-0.189
-0.104
Polyphenol oxidase
0.466
0.548
0.526
0.522
0.572
0.545
Denitrification enzyme
0.229
0.143
0.182
0.105
0.184
0.172
Arylsulfatase
0.028
0.199
0.172
0.047
0.047
0.101
Catalase
0.804
0.804
0.762
0.867
0.893
0.854
Acid phosphomonoesterase
-0.072
0.166
0.083
0.007
-0.029
0.032
Urease
0.597
0.794
0.707
0.713
0.686
0.723
Dehydrogenase
0.262
0.287
0.280
0.234
0.311
0.284
Biomass N
0.455
0.555
0.578
0.495
0.477
0.527
Biomass P
0.477
0.570
0.542
0.517
0.546
0.548
Biomass C
0.485
0.594
0.526
0.508
0.577
0.556
Cellulase
0.611
0.748
0.699
0.721
0.706
0.721
Nitrate reductase
0.332
0.192
0.189
0.200
0.305
0.250
-0.418
-0.166
-0.253
-0.337
-0.402
-0.325
Nitrification potential
3
a
4
Bold text indicate significant P values (<0.05)
5
6
Average annual rice yield of 2005 to 2009
7
1.0
PC2 (35.28%)
0.5
0.0
-0.5
Control
N
NP
NK
NPK
-1.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
PC1 (41.25%)
8
9
Figure S1
10
Ordination plot produced from principal-component analysis (PCA) of geochemical
11
data for all the samples. The geochemical parameters used for PCA include: total
12
carbon, total nitrogen, nitrate, ammonium, available phosphorous, available
13
potassium, and pH.
14
15
16
17
14000
a
b
10000
-1
Yield (kg ha )
12000
8000
c
c
c
6000
4000
2000
0
Control
N
NP
NK
NPK
18
Figure S2
19
Average rice yields over 5 years from 2005 to 2009 in the long-term fertilization
20
experiments.
21
22
23
24
25
Figure S3
26
Average percentages of total 16S rRNA gene sequences classified to each phylum for
27
each treatment. Phyla are displayed if they represent at least 1% of the total
28
sequences in at least one treatment. “Others” includes Armatimonadetes,
29
Cyanobacteria,
30
Fusobacteria, and OP11, with no more than 1% each.
31
32
Actinobacteria,
Chlorobi,
Bacteroidetes,
BRC1,
Chlamydiae,
OTU_00267 Bacteria
OTU_00472 Syntrophobacteraceae
OTU_01238 Syntrophobacterales
OTU_00443 Thermodesulfovibrio
OTU_00003 Rhizobiales
OTU_00264 Bacteria
OTU_00601 Desulfobacteraceae
OTU_00080 Anaerolineaceae
OTU_00004 Bacteria
OTU_02406 Bacteria
OTU_00032 Bradyrhizobium
OTU_00043 Syntrophorhabdus
OTU_00023 Steroidobacter
OTU_00276 Bacteria
OTU_00246 Bacteria
OTU_00187 Rhodospirillaceae
OTU_00006 Anaerolineaceae
OTU_00137 Bacteria
OTU_01077 Desulfonema
OTU_00010 Rhodoplanes
OTU_00036 Bacteria
OTU_00099 Syntrophorhabdus
OTU_00294 Thermodesulfovibrio
OTU_00062 Anaerolineaceae
OTU_00044 Bradyrhizobium
37
800
600
400
Number of Sequences
1200
1000
200
0
33
34
Figure S4
35
25 most abundant OTUs in the samples. Each is classified to the lowest level possible
36
with 0.5 confidence.
3000
Control
N
NK
NP
NPK
2500
chao1
2000
1500
1000
500
A
0
0
500
1000
1500
2000
2500
3000
4
Control
N
NK
NP
NPK
2
H-alpha
6
8
Sequences per sample
0
B
0
0.25
0.5
1
2
4
8
Inf
alpha
38
39
40
Figure S5
41
A: chao1 estimators showing the phylogenetic diversity of bacterial community of all
42
the samples. B: Renyi diversity profiles by treatment. There were no differences in
43
species richness (alpha=0) among treatments, but richness (alpha=Inf) did differ
44
among treatments with those receiving P being the lowest.
45
46
PC2 (8.90%)
0.2
0.1
Control
N
NP
NK
NPK
A
0.0
-0.1
-0.2
-0.2
-0.1
0.0
0.1
0.2
PC1 (9.04%)
B
47
48
Figure S6
49
A: Ordination plot produced from principal coordinates analysis (PCoA) of 16S rRNA
50
gene based pyrosequencing data. B: Principal components analysis of Hellinger
51
transformed OTU counts based on 16S rRNA gene based pyrosequencing data. The
52
second axis correlates with a gradient in available phosphorous and separates
53
samples according to whether or not P was added. The lines indicate the
54
phosphorous gradient, increasing from the bottom to the top of the figure. The
55
symbol sizes are proportional to the phosphorous concentration for that sample.
56
57
58
59
60
Figure S7
61
Correlations between sequence abundances and available phosphorus. The top three
62
panels are for individual OTUs. The bottom three panels are for data agglomerated
63
by genus. The three OTUs or genera with the greater numbers of total sequences
64
were selected from those having significant (P <= 0.05) correlations with available
65
phosphorus.
66
67
40000
A
b
b
b
b
bc
b
bc
b
b
a
a
a
a
a
a
a
a
30000
20000
10000
0
2500
2000
B
c
1500
1000
500
Diversity
0
4000
C
c
3000
2000
1000
0
800
D
600
b
400
200
0
600
400
a
E
bc
c
N
NP
ab
d
200
0
Control
NK
NPK
68
Figure S8
69
Diversity (inverse simpson) of functional genes. A: total diversity, B: Carbon cycling, C:
70
Nitrogen cycling, D: Phosphorus cycling, E: Sulfur cycling. Bars with the same letter
71
do not differ at P < 0.05.
72
73
74
75
76
Figure S9
77
The relative changes of the detected genes involved in the N cycle (Control vs. NPK
78
treatment). The percentage of a functional gene in a bracket was the sum of signal
79
intensity of all detected sequences in one gene divided by the grand sum of signal
80
intensity of all detected N cycle genes, and weighted by the fold change of the signal
81
intensity of this gene in NPK fertilized treatment to that at control. For each
82
functional gene, red means that this gene had a higher signal intensity at treatment
83
than at Control and their significance was indicated with two stars (**) at p<0.05 and
84
one star (*) at 0.05<p<0.1, while blue means that this gene had a lower signal
85
intensity. Gray-colored genes were not targeted by this GeoChip, or not detected in
86
those samples.
87
88
89
90
91
92
Figure S10
93
The relative changes of the detected genes involved in the N cycle (Control vs. N
94
treatment). The percentage of a functional gene in a bracket was the sum of signal
95
intensity of all detected sequences in one gene divided by the grand sum of signal
96
intensity of all detected N cycle genes, and weighted by the fold change of the signal
97
intensity of this gene in N treatment to that at control. For each functional gene, red
98
means that this gene had a higher signal intensity at treatment than at Control and
99
their significance was indicated with two stars (**) at p<0.05 and one star (*) at
100
0.05<p<0.1, while blue means that this gene had a lower signal intensity.
101
Gray-colored genes were not targeted by this version of GeoChip, or not detected in
102
those samples.
103
Normalized signal intensity
25
20
15
a
Control
N
NP
NK
NKP
d
ab ab ab
b
b
ab ab ab
10
5
a a a a
a
0
phytase
ppk
ppx
104
Figure S11
105
The normalized average signal intensity of the detected key functional genes involved
106
in P cycling. phytase for phytate degradation; ppx encoding exopolyphosphatase; ppk
107
encoding polyphosphate kinase; All data were presented as the mean ± SE. Bars with
108
the same letter do not differ at P < 0.05.
109
110
111
112
6
Control
N
NP
NK
NKP
Normalized signal intensity
a a
5
b
4
c
d
3
b
a a
a a
2
a b
c b b
1
0
mcrA
mmoX
pmoA
113
Figure S12
114
The normalized average signal intensity of the detected functional genes involved in
115
methane metabolism. mcrA encoding alpha subunit of methyl coenzyme M
116
reductase; mmoX encoding particulate methane monooxygenase; pmoA encoding
117
methane monooxygenase. All data were presented as the mean ± SE. Bars with the
118
same letter do not differ at P < 0.05.
119