Table 2 Multiple stepwise regression model results for dominant bacterial and fungal phyla (and classes for Proteobacteria) that significantly correlated with a secondary metabolites. Soil origin was also included as an explanatory variable to discern metabolite versus soil effects. Raw abundances were centered log-ratio transformed prior to analysis. Type-1 error rates given are FDR-corrected

67.3

< 0.01

13.07

0.19

< 0.01

−2.84

< 0.01

−2.67

< 0.01

17.99

< 0.01

21.99

0.38

< 0.01

−7.33

< 0.01

1.74

0.09

−2.17

0.03

70.39

< 0.01

51.94

0.6

< 0.01

−10.26

< 0.01

−1.49

0.14

2.66

< 0.01

45.92

< 0.01

25.79

0.42

< 0.01

8.48

< 0.01

2.25

0.03

−1.85

0.07

62.91

< 0.01

22.79

0.39

< 0.01

−5.68

< 0.01

−2.56

0.01

2.29

0.02

36.34

< 0.01

10.82

0.22

< 0.01

−3.49

< 0.01

−1.82

0.07

2.51

0.01

48.68

< 0.01

41.9

0.45

< 0.01

−8.98

< 0.01

2.15

0.03

45.06

< 0.01

7.65

0.12

< 0.01

−2.79

< 0.01

2.75

< 0.01

7.93

< 0.01

3

0.08

0.02

2.93

< 0.01

2.24

0.03

−1.53

0.13

−2.11

0.04

24.9

< 0.01

5.96

0.09

< 0.01

−2.42

0.02

2.41

0.02