3.1 Richness and diversity of bacterial and fungi communities in vineyard plots before and after AMF inoculation and irrigation systems application.
Soil samples were collected from 16 plots in the Oakville vineyard before and after the treatment application (Additional file 1: Figure S1). Each treatment was characterized by the combination of mycorrhizal inoculation (I or NI) and irrigation systems (FI or HI) with four replicates (see “Material and Methods” section). Three kingdoms were identified in the samples (fungi, archaea and bacteria) (Additional file 3). Venn diagrams were generated before removing the less common species found in the trial (see “Material and Methods” section) to assess the distinct and common bacterial and fungal species among different treatments (Additional file 1: Figure S2). There were 97 bacterial and 273 fungal OTUs specific to the NT plots. On the other hand, there were 9 bacterial species were specific to FINI and HII whereas 19 bacterial OTUs were found in HINI as well as in FII plots. Surprisingly, no specific fungal OTUs for plots after treatments were found. Thus, the number of bacterial species shared by all the plots were the 46.4 % whereas for fungi accounted for the 21.4 % of the total observations.
The analysis of α-diversity indices in different plots showed differences according to the LMEM (Figure 1). Regarding bacterial communities, species richness was decreased in the FINI plots comparing to NT ones (Figure 1A). However, no differences were found between the different plots regarding the Shannon or the inverse of Simpson indices (Figure 1C and E). Fungal communities were significantly affected by the treatments. Thus, the number of fungal species identified was significantly lower after treatment application, especially in the FII plots (Figure 1B). Shannon index was decreased after treatment in the NI plots (Figure 1D). On the other hand, inverse of Simpson index increased after treatment application in all the plots (Figure 1F).
In order to find possible effects of competition between bacteria and fungi, correlation between α-diversity indices was conducted. Pearson correlation across all samples was negligible for Shannon and inverse of Simpson indices (R = 0.16, p value = 0.39 and R = -0.095, p value = 0.6, respectively) (Additional file 1: Figure S3A and B). Similarly, relationships of the diversity indices within each treatment/ plots was insignificant suggesting no competition between fungal and bacterial communities (Additional file 4: Table S1).
Correlations between microbiome structure and composition and the applied treatments were studied by computing the between-sample diversity using Bray-Curtis distance (Figure 2). For bacteria, the Principal Coordinates Analysis (PCoA) showed dissimilarities between samples from different treatments that clustered separately in three groups, NT, the full irrigated plots (FINI and FII) and half irrigated (HINI and HII), respectively (Figure 2A). The first two components explained the 28.1 % and the 18.8 % of the variation. Additionally, all the applied treatments (AMF inoculation and irrigation amounts), time of sampling and their combination affected the bulk soil bacterial communities as showed Table 1. The ordination of the data in a non-metric multidimensional scaling (NMDS) indicated that all samples clustered closely suggesting similar bacterial communities (Figure 2B). The application of treatments significantly affected beta-diversity where PERMANOVA test showed a clear distinction (F=4.879, p=0.001).
Table 1. F and p values of PERMANOVAs comparing different irrigation systems (I), AMF inoculation (M) and time course (T) of Merlot vineyard microbial beta diversity.
|
|
Bray-Curtis
|
|
|
Bacteria
|
Fungi
|
Treatment
|
Comparison
|
F
|
p value
|
F
|
p value
|
Irrigation (I)
|
FI vs. HI
|
4.727
|
0.001
|
6.480
|
0.001
|
AMF inoculation (M)
|
NI vs. I
|
3.528
|
0.002
|
5.446
|
0.005
|
Time course (T)
|
T0 vs. T1
|
9.688
|
0.001
|
26.762
|
0.001
|
I × M
|
FINI vs. FII vs. HINI vs. HII
|
3.424
|
0.001
|
4.598
|
0.001
|
I × T
|
FI_T0 vs. FI_T1 vs. HI_T0 vs. HI_T1
|
8.788
|
0.001
|
14.134
|
0.001
|
M × T
|
I_T0 vs. I_T1 vs. NI_T0 vs. NI_T1
|
5.312
|
0.001
|
14.092
|
0.001
|
I × M × T
|
FINI_T0 vs. FINI_T1 vs. HINI_T0 vs. HINI_T1 vs. FII_T0 vs. FII_T1 vs. HII_T0 vs. HII_T1
|
4.879
|
0.001
|
7.615
|
0.001
|
On the other hand, the PCoA for fungi communities highlighted changes in composition due to treatments where NT samples were clearly separated from the treatment samples (Figure 2C). The first two components accounted for explaining the 55.8 % of the total variance. Moreover, PERMANOVA tests comparing the different treatment combination showed clear distinctions between them (Table 1). Likewise, the NMDS showed two clusters with NT clearly separated from the samples after treatments (Figure 2D). PERMANOVA test highlighted how treatments affected fungal beta-diversity (F = 7.615, p = 0.001). These results suggested that the main factor affecting the bacterial and fungal composition was the time of sampling and that bacterial composition is more responsive to different irrigation amounts than fungal composition.
3.2 Bacterial and fungal taxa distribution in the Merlot vineyard bulk soil is significantly affected by AMF inoculation, irrigation treatments and time.
The taxonomic affiliations of the bacterial OTUs core showed vineyard bulk soil hosted 18 phyla, 38 classes, 71 orders, 130 families and 298 genera (Additional file 3). Different samples were dominated by Proteobacteria phylum that accounted, on average, for more than the 35% in the bacterial communities. Other common phyla were Actinobacteria (ranged between 15.3% and 23.9 %), Verrumicrobia (between 8 and 12.2%), Gemmatimonadetes (between 4.8 and 9.8%), Firmicutes (between 2.3 and 7.7%), Planctomycetes (between 5.6 and 8.8%), and Chloroflexi (between 1.5 and 2.9%), as well as the phylum Crenarchaeota belonging to Archaea (ranged between 0 and 16%) (Figure 3A).
Irrigation, AMF inoculation and time affected the abundances at phylum levels as subjected the significance of the LMEM conducted (Additional file 4: Table S2A). Thus, FI had increased the proportion of Proteobacteria (p ≤ 0.0001) while reduced applied water with HI increased the abundances of Actinobacteria (p = 0.032), Chloroflexi (p = 0.019) and others (p ≤ 0.0001).
The most abundant 12 bacterial classes, which accounted for more than 90 % of the relative abundance in all treatments, are presented in Figure 3C. Irrigation and AMF inoculation treatments affected the proportion of these bacterial classes as highlighted with the significance of the LMEM (Additional file 4: Table S2B). Therefore, we observed increased abundance of Bacilli and decreased abundance of Gemmatimonadetes comparing to the abundances of NT (p ≤ 0.0001 for both classes). On the other hand, FI and/ or AMF inoculation increased the abundances of Alphaproteobacteria, Betaproteobacteria and Gammaproteobacteria. On the other hand, HINI decreased Alphaproteobacteria and Betaproteobacteria and increased Gammaproteobacteria and Actinobacteria classes. Finally, full irrigation decreased the abundance of Thaumarchaeota and other bacterial classes (p ≤ 0.0001 and p = 0.003, respectively, Additional file 4: Table S2B).
Regarding the bulk soil mycobioma different plots showed 4 phyla, 13 classes, 37 orders, 67 families and 121 genera (Additional file 4). The most abundant phyla we found were Basidiomycota, which, on average, accounted for the 50% of the fungal communities, Ascomycota (ranged between 21.6 and 37.5 %) and Zygomycota (ranged between 3.3 and 17.6 %) (Figure 3B).
Relative abundances of the fungal phylum were highly responsive to the treatments as indicated by the LMEM (Additional file 4: Table S3A). Thus, Ascomycota decreased in HINI plots (p = 0.007) while Basidiomycota increased after treatment application in all plots (p ≤ 0.0001). The abundance of Zygomycota phylum increased with the combination of treatments, especially in FI and inoculation with AMF (p ≤ 0.0001) whereas, the abundances of other less abundant phyla decreased (Figure 3C, Additional file 4: Table S3A).
Within fungal classes the most abundant were Agaricomycetes, Dothideomycetes, Eurotiomycetes, Leotiomycetes, Pezizomycetes, Saccharomycetes, Sordariomycetes, and Tremellomycetes, which accounted for the 50 % of the fungal abundance in the NT plots, and between the 62% and 92% in the plots after treatment (Figure 3D). However, distribution of fungal classes was strongly affected by treatments (Additional file 4: Table S3B). Non-inoculated plots decreased the abundances of Agaromycetes (p = 0.008), Leotimycetes (p ≤ 0.0001) and Saccharomycetes (p = 0.001) especially in the FI plots. On the other hand, all the treatments increased the abundance of Tremellomycetes in detriment of Sordariomycetes and Dothideomycetes classes. Finally, comparing to NT, plots after treatment application decreased abundances of other fungal classes (p ≤ 0.0001, Figure 3D; Additional file 4: Table S3B).
3.3 Bacterial and fungal clade proportions in the Merlot vineyard after treatments differed due to AMF inoculation and/or irrigation treatments.
The study of the differences on the median proportions of the microbiome families cladograms due to treatments are presented in the Additional file 1: Figures S4-S10. Within the Proteobacteria phylum (Additional file 1: Figure S4), HI increased the proportion of Oceanospirillaceae and Cellvibrionaceae families. FII increased the Marinicellaceae, Anaeromyxobacteriaceae, Rickettsiaceae and decreased Neisseriaceae. The main changes in Actinobacteria (Additional file 1: Figure S5) were increments of the proportions of Streptoporangiales and Cryptosporangiaceae due to HI, and enhancement of Micromonosporaceae, Acidimicrobiaceae and Actinospicaceae in FII, HINI and HII plots. Firmicutes phylum proportions were highly affected by treatments (Additional file 1: Figure S6). Thus, HI increased the proportion of members of the family XVII and Erysipelotrichaceae. FII increased Syntrophomonadaceae and Gracilibacteraceae proportions. Finally, comparing to FINI plots, FII, HII and HINI enhanced Halanaerobiaceae. Within the other less abundant phyla (Additional file 1: Figure S7), the main differences were due to HI, which, increased families from Chlamydiales and Spirochaetia clades and enhanced the proportions of families belonging to Archaea Kingdom. On the other hand, FII increased the proportion of Deinococcaceae family.
Regarding the effect of different treatments on the proportion of fungal families, members of the Ascomycota phylum were highly affected by treatments (Additional file 1: Figure S8). FII increased the proportion of Glomerellaceae and Togniniaceae, HII increased Sporormiaceae, Tubeufiaceae and decreased Eremomycetaceae, whereas FI led to increased proportions of Leptosphaeriaceae. Basidiomycota phylum cladogram (Additional file 1: Figure S9) shows that different irrigation treatments affected the Malasseziaceae (increased in HI) and Sparassidaceae (increased in FI) families. On the other hand, AMF inoculation increased the proportions of Hydnodontaceae under FI conditions and Psathyrellaceae and Cortinariaceae under HI conditions. Zygomycota families were not highly affected by treatments (Additional file 1: Figure S10).
3.4 Edaphic factors barely affected in the bacterial and fungal distribution.
The SWC decreased with the HI treatment that accounted for a decreased soil evaporation, especially in HII plots, although no interactive effect between factors was evident. The edaphic factors of different plots slightly influenced the bacterial and fungal communities as showed the canonical correspondence analysis (CCA) (Figure 4, Additional file 4: Table S4 and S5).
Table 2. Edaphic factors of Merlot vineyard soil subjected to different irrigation amounts (FI, Full Irrigated or HI, Half Irrigated), and AMF inoculation (I, inoculated; or NI, non-inoculated).
|
Soil pH
|
Relative humidity (RH)
|
Soil evaporation (mmol m-2 s-1)
|
Soil temperature (℃)
|
SWC (g/g)
|
Bulk density (g/cm3)
|
FI
|
5.90 ± 0.10
|
17.4 ± 1.2
|
0.41 ± 0.03 a
|
39.1 ± 1.0
|
0.06 ± 0.01 a
|
1.11 ± 0.02
|
HI
|
5.74 ± 0.09
|
16.1 ± 0.5
|
0.24 ± 0.02 b
|
39.2 ± 0.5
|
0.05 ± 0.01 b
|
1.12 ± 0.01
|
NI
|
5.74 ± 0.05
|
16.2 ± 0.9
|
0.29 ± 0.03 b
|
39.7 ± 0.5
|
0.05 ± 0.01
|
1.11 ± 0.02
|
I
|
5.89 ± 0.12
|
17.3 ± 1.0
|
0.36 ± 0.04 a
|
38.6 ± 0.9
|
0.06 ± 0.01
|
1.11 ± 0.02
|
LMEM
|
|
|
|
|
|
|
Irrigation (I)
|
0.194
|
0.232
|
0.0001
|
0.862
|
0.029
|
0.591
|
AMF inoculation (M)
|
0.155
|
0.267
|
0.017
|
0.199
|
0.094
|
0.935
|
I x M
|
0.267
|
0.780
|
0.413
|
0.645
|
0.331
|
0.264
|
Values represent means ± SE (n = 8) separated by Kenward–Roger method and Tukey’s p-value adjustment (P ≤ 0.05). Different letters indicate significant differences as affected by treatment application (FI, HI, NI, or I) according to the main factors in the linear mixed-effect model.
However, Figure 4A shows that soil temperature and bulk density, soil evaporation and soil water content correlated with the bacterial beta diversity, and explained the separation between HI and FI treatments. This separation was related with the abundances of Nitrososphaera sp., Jatrophihabitans sp., Actinophytocola sp., Pseudonocardia sp., Geodermatophilus sp., Actinomycetospora sp., Rugosimonospora acidiphila or Micromonospora hermanusense species as shown the correlation with CCA1 (Additional file 4: Table S4C).
On the other hand, the treatments applied did not differ in their fungal composition driven by edaphic factors, and all the treatments clustered together (Figure 4B). However, the soil evaporation and the SWC were negatively correlated with the fungal composition that could be related to the abundances of Penicillium, Aspergillus, Crytococcus and Cladorrhinum members (Additional file 4: Table S5C).
3.5 Co-occurrence patterns in bacterial and fungal communities were affected by time and AMF inoculation and/or irrigation treatments.
The analyses of the co-occurrence bacterial and fungal networks in the bulk soil of the Merlot vineyard showed different connectivity patterns influenced by time (before and after treatment application; Figure 5, Table 3) and by the different treatments (FINI, FII, HINI and/or HII; Figure 6, Table 4). Before treatment, just 973 (0.7 %) of the 141796 pair-wise comparisons yielded statistically significant co-occurrence, comprising 615 positive and 358 negative associations (Table 3). Similarly, after treatment application, from the 135356 pair-wise comparisons, only 1049 were statistically significant with 795 positive and 254 negative co-occurrences, respectively. Thus, although the total number of co-occurrences did not differ between samples before and after treatment application, the latter showed more positive and less negative associations comparing to the pre-treatment samples (Table 3, Figure 5). This shift was likely influenced by the enhancement of the positive associations between bacteria species and the diminution of the negative fungal associations in soil samples after treatment application (Table 3).
Table 3: Degree of connection for bacterial and fungal communities found in Merlot vineyard plots before (not treated, NT) and after treatments, different irrigation amounts (FI, Full Irrigated or HI, Half Irrigated), and AMF inoculation (I, inoculated; or NI, non-inoculated).
|
|
Not treated (NT)
|
After treatment
|
Positive connections
|
|
|
|
Total
|
615
|
795
|
|
Bac-Bac
|
281
|
350
|
|
Fun-Fun
|
126
|
205
|
|
Bac-Fun
|
208
|
240
|
Negative connections
|
|
|
|
Total
|
358
|
254
|
|
Bac-Bac
|
93
|
85
|
|
Fun-Fun
|
85
|
18
|
|
Bac-Fun
|
180
|
151
|
Total connections
|
|
|
|
Total
|
973
|
1049
|
|
Bac-Bac
|
374
|
435
|
|
Fun-Fun
|
211
|
223
|
|
Bac-Fun
|
388
|
391
|
|
Total analyzed pairs
|
141796
|
135356
|
|
Percentage of non-random
|
0.7
|
0.8
|
We also conducted co-occurrence analyses to assess the effect of AMF inoculation and irrigation treatments (Figure 6, Table 4). In the FI plots, 229 co-occurrences, were found to be statistically significant, majority of them were bacterial associations, bacteria-fungi associations and less frequent the associations between fungal species. The 57% (130) of the co-occurrences in FI plots were positive while the 43% were negative (99, Table 4). Figure 6A showed that the significant negative co-occurrences happened between bacterial and fungal species separately. For instance, Chloroflexi and Acidobacteria with Proteobacteria or connections between Ascomycota species. The positive connectivity found in FI plots was mainly explained by associations between bacteria species, such as the links of Bacteroidetes, Firmicutes, Proteobacteria and Actinobacteria species. On the other hand, HI plots had 300 significant associations, the 77% (231) of the connections between microbial species were positive while the 23% (69) were negative. In these plots, half of the connections were identified as fungi-bacteria associations. Thus, a great connectivity between species of Acidobacteria or Proteobacteria with Ascomycota and Basidiomycota members was found (Figure 6B).
Table 4: Degree of connection for bacterial and fungal communities in Merlot vineyard soils subjected to different irrigation amounts (FI, Full Irrigated or HI, Half Irrigated) or AMF inoculation (I, inoculated; or NI, non-inoculated).
|
|
FI
|
HI
|
NI
|
I
|
Positive connections
|
|
|
|
|
|
Total
|
130
|
231
|
176
|
196
|
|
Bac-Bac
|
72
|
52
|
97
|
66
|
|
Fun-Fun
|
24
|
81
|
26
|
62
|
|
Bac-Fun
|
34
|
98
|
53
|
68
|
Negative connections
|
|
|
|
|
|
Total
|
99
|
69
|
95
|
67
|
|
Bac-Bac
|
30
|
24
|
25
|
28
|
|
Fun-Fun
|
21
|
0
|
17
|
0
|
|
Bac-Fun
|
48
|
45
|
53
|
39
|
Total connections
|
|
|
|
|
|
Total
|
229
|
300
|
271
|
263
|
|
Bac-Bac
|
102
|
76
|
122
|
94
|
|
Fun-Fun
|
45
|
81
|
43
|
62
|
|
Bac-Fun
|
82
|
143
|
106
|
107
|
|
Total analyzed pairs
|
120299
|
128975
|
128097
|
126817
|
|
Percentage of non-random
|
0.2
|
0.8
|
0.2
|
0.2
|
Regarding NI plots, 271 significant association were found with 176 being positive and 95 being negative (Table 4). Majority of the associations were between bacteria species (about 45%, Table 4) or between bacteria and fungi (39%). Negative associations were more frequent between species belonging to Acidobacteria, Proteobacteria, Bacteroidetes or Chloroflexi phyla whereas positive associations were found between Actinobacteria, Verrumicrobia, Proteobacteria and other bacteria phyla (Figure 6C). Moreover, a great degree of connectivity (either positive or negative) between Acidobacteria and Proteobacteria with Ascomycota, Basidiomycota and Zygomycota species was highlighted. Finally, inoculated plots showed 263 significant associations with the 75% being positive co-occurrences (196) and the 25% negative (67, Table 4). Again, the majority of connections were within bacteria species or between bacteria and fungi species. After inoculation, associations between fungi and bacteria were found, such as Proteobacteria with Ascomycota or Verrumicrobia with Basidiomycota. Contrarily, the connectivity between fungal species was very low. It is noteworthy the increase of the negative connectivity network of Acidobacteria species, especially with Verrumicrobia and bacteria belonging to other clades, and Actinobacteria phyla with fungal and bacterial species (Figure 6D).