Influence of B. ceiba on soil nutrient contents
The values of whole soil nutrient properties were highest at D0 among all reproductive stages of rice, suggesting B. ceiba may not compete with rice for soil nutrients (Fig. 2). Results of One-way ANOVA showed that B. ceiba had a significant influence on soil nutrient properties in rice paddy, and this influence varied with different reproductive stages of rice (Fig. 2 & Table 1). During booting stage of rice, both SOC (P < 0.05) and TN (P < 0.05) were significantly higher at D0 than D5 but not D1. Similarly, this pattern could be found on SOC (P < 0.05) and TP (P < 0.05) during heading stage of rice. Further, we found that AN (P < 0.05) and AK (P < 0.05) were significantly higher at D0 than D1 and D5 during mature stage of rice (Fig. 2). These results indicated that rice paddy closer to the B. ceiba would have higher soil nutrients.
Table 1
Two-way ANOVA of soil nutrient properties between distance gradients to Bombax ceiba and reproductive stages of rice.
Variance sources
|
SOM
|
AN
|
AP
|
AK
|
TN
|
TP
|
TK
|
Distance to B. ceiba (DB)
|
6.577**
|
1.604
|
2.825
|
6.549**
|
4.314*
|
4.489*
|
0.187
|
Reproductive stage of rice (RSR)
|
0.856
|
10.147**
|
4.825*
|
1.047
|
1.977
|
53.160**
|
17.028**
|
DB × RSR
|
0.189
|
2.324
|
0.813
|
0.856
|
0.352
|
1.417
|
0.204
|
Values in the table are F value, and “*” represent P < 0.05 while “**” represent P < 0.01 |
Influence of B. ceiba on soil bacterial and fungal alpha diversity and its links with soil nutrient properties
One-way ANOVA showed that B. ceiba significantly affected bacterial richness at booting stage of rice and bacterial Shannon index at the whole reproductive stages of rice (Fig. 3). Specifically, these alpha diversity indices were higher at D1 than D0 (Fig. 3), suggesting that B. ceiba may promote the increase of bacteria in rice paddy soil. In contrast to bacteria, fungal richness and Shannon index were not obviously different among distance gradients to B. ceiba (Fig. 3). These results indicated that B. ceiba had different effect on rice paddy soil microorganisms.
Correlation analysis showed that bacterial richness was negatively correlated with AP and AK while Shannon index was negatively correlated with AP and TN. In addition, both the bacterial alpha diversity indices positively correlated with TK (Table S1). However, no significant relationship was found between fungal alpha diversity and soil chemical properties (Table S1).
Furthermore, multiple regression analysis showed that bacterial richness was mainly predicted by distance to B. ceiba, AP and TK (R2 = 0.177, 0.132 and 0.193, respectively; P < 0.05, Table 2), whereas the bacterial Shannon index was mainly predicted by distance to B. ceiba, reproductive stage of rice, and AP (R2 = 0.223, 0.381 and 0.175, respectively; P < 0.05, Table 2). In contrast, both richness and Shannon index of fungi were principally explained by reproductive stage of rice and AN (R2 = 0.256, 0.129 and 0.177, 0.178, respectively; P < 0.05, Table 2).
Table 2
Explanatory variables retained in the best models for explaining bacterial and fungal alpha diversity
|
Alpha diversity index
|
Variables
|
Individual contribution of variable (%)
|
P
|
|
Model P
|
Bacteria
|
Richness
|
Distance to B. ceiba
|
17.72
|
<0.05
|
0.496
|
<0.01
|
|
|
AN
|
5.67
|
<0.05
|
|
|
|
|
AP
|
13.19
|
<0.1
|
|
|
|
|
SOM
|
5.28
|
<0.05
|
|
|
|
|
TK
|
19.33
|
<0.01
|
|
|
|
Shannon index
|
Distance to B. ceiba
|
22.28
|
<0.001
|
0.825
|
<0.0001
|
|
|
Reproductive stage of rice
|
38.09
|
<0.01
|
|
|
|
|
AP
|
17.54
|
<0.05
|
|
|
|
|
AK
|
4.36
|
=0.102
|
|
|
|
|
SOM
|
3.00
|
<0.05
|
|
|
|
|
TP
|
2.61
|
<0.05
|
|
|
Fungi
|
Richness
|
Reproductive stage of rice
|
25.58
|
<0.01
|
0.379
|
<0.01
|
|
|
AN
|
17.72
|
<0.01
|
|
|
|
|
AP
|
4.12
|
=0.462
|
|
|
|
Shannon index
|
Reproductive stage of rice
|
12.93
|
<0.05
|
0.217
|
<0.05
|
|
|
AN
|
17.78
|
<0.01
|
|
|
The relative roles of B. ceiba, rice stage and soil nutrient properties in determining soil bacterial and fungal community composition
NMDS plot showed that bacterial community composition was different among distance gradients to B. ceiba (Fig. 4A), which was further demonstrated by ANOSIM (R = 0.521; P = 0.0001) and PERMANOVA (F = 7.992; P = 0.0001) methods (Table 3). Based on a relative abundance > 1%, Proteobacteria, Acidobacteria, Firmicutes, Chloroflexi, Actinobacteria, Myxococcota, Verrucomicrobia, Desulfobacteria, Gemmatimonadetes, Bacteroidetes, Planctomycetes, Nitrospira, Methylomirabilota were identified as dominant bacterial phyla in the studied plots (Fig. 4B). The relative abundance of most of these phyla were significantly different among distance gradients to B. ceiba (Table S2). For example, the relative abundance of Chloroflexi phylum was significantly higher at D1 and D5 than D0, while the relative abundance of Actinobacteria phylum was significantly higher at D0 than D1 and D5 during the whole reproductive stages of rice (Table S2). In addition, it cannot be ignored that rice stages also significantly affected some bacterial dominant phyla. For instance, the relative abundance of Firmicutes phylum was significantly higher at heading and mature stages than at booting stage, while the relative abundance of Acidobacteria phylum was significantly higher at booting stage than at heading and mature stages (Table S2).
Table 3
Variations of bacteria and fungi community composition among distance gradients to Bombax ceiba or in different reproductive stages of rice, as tested with ANOSIM and PERMANOVA methods
|
Variables
|
ANOSIM
|
PERMANOVA
|
R
|
P
|
F
|
P
|
Bacteria
|
Distance to B. ceiba
|
0.5211
|
0.0001
|
7.9924
|
0.0001
|
|
Reproductive stage of rice
|
0.1483
|
0.0133
|
2.0714
|
0.0325
|
Fungi
|
Distance to B. ceiba
|
0.5861
|
0.0001
|
5.7872
|
0.0001
|
|
Reproductive stage of rice
|
0.0440
|
0.1714
|
1.1771
|
0.2220
|
Similarly, fungal community composition also varied with distance to B. ceiba (Fig. 5A), which was confirmed by ANOSIM (R = 0.586; P = 0.0001) and PERMANOVA (F = 5.787; P = 0.0001, Table 3). Of the identified phyla, Ascomycota, unidentified fungi, Basidiomycota, Mortierellomycota, Rozellomycota, Chytridiomycota and Glomeromycota phyla dominated the fungal communities in this study plots (Fig. 5B). The relative abundance of Ascomycota phylum was significantly higher at D5 than D0 and D1 at heading stage of rice while higher at D5 and D1 than D0 at mature stage of rice (Table S2), which suggesting that soil fungal community in rice paddy closer to B. ceiba would be simultaneously affected by B. ceiba and rice. Furthermore, Veen diagram showed that more bacterial and fungal OTUs were shared at D0 and D1 rather than D0 and D5 (Fig. S1).
Mantel correlation analysis showed that both bacterial and fungal community composition were significantly correlated with AN, AP, AK, SOC and TN (Table S3). In RDA, the whole variables respectively explained 52.4% and 39.8% of the variations of bacterial and fungal community (Fig. 6). Furthermore, the relative contribution of each variable was quantified by “rdacca.hp” function. Distance to B. ceiba (R2 = 0.217, P < 0.01) and AK (R2 = 0.076, P < 0.05) were revealed to be factors significantly affecting bacterial community composition, while distance to B. ceiba (R2 = 0.194, P < 0.01), SOC (R2 = 0.050, P < 0.05) and TN (R2 = 0.037, P < 0.05) significantly affected fungal community composition (Fig. 6).