Criteria for assigning nirK and nirS sequences to operational

When analyzing sequences of functional genes instead of 16S rRNA genes, it is crucial to define an empiric cutoff value that defines if two sequences are probably derived from two different species or belong to the same species or OTU (Purkhold et al. 2000). Values used for analyzing narG and nosZ were as published (Palmer et al. 2009). Values for nirK and nirS were not available and were therefore calculated in silico (2.5.12.1) prior to gene analyses in the earthworm gut and soil.

3.1.1.3.1.1. Phylogenetic correlation plots and comparative tree topologies of nirK and corresponding 16S rRNA genes

For nirK, phylogenetic correlation plots were constructed with 74 nirK sequences together with 73 corresponding 16S rRNA sequences (Pseudomonas palustris TIE-1 contained two copies of nirK) for both nirK gene and in silico translated nirK amino acid sequences (2.5.12.1, Figure 11). Linearity between 16S rRNA gene similarity and both nirK gene and amino acid similarity was particularly apparent for a 16S rRNA gene similarity of about ≥ 90 % (Figure 11). Some distantly related organisms (i.e., with a 16S rRNA gene similarity between 78 % and 83 %) carried highly similar nirK genes (i.e., their nirK gene and amino acid sequences were 90 % to 100 % identical). This feature was more pronounced for amino acid sequences (Figure 11B) than for gene sequences (Figure 11A).

Of all organisms with a ≥ 97 % 16S rRNA gene similarity, 90 % had a nirK similarity of

≥ 83 % (Figure 11A) and a nirK in silico translated amino acid sequence similarity of ≥ 91 % (Figure 11B). Thus, 83 % was defined as a cutoff value to create nirK gene sequence species-level OTUs, i.e., a dissimilarity of two nirK gene sequences of 17 %. For nirK amino acid sequences, this cutoff value was 91 %, i.e., a dissimilarity of 9 % between two nirK amino acid sequences. Both cutoff values are conservative estimates that indicate a minimum amount of species-level OTUs that can be expected.

Comparison of 16S rRNA gene phylogeny and nirK gene phylogeny showed that some taxa were separated in both phylogenetic trees (e.g., clusters 2 to 5 ) whereas other taxa were separated in the 16S rRNA tree only but clustered together in the nirK gene tree (e.g., cluster 1 and clusters 6 to 9) (Figure 12). The two nirK gene copies of Pseudomonas palustris TIE-1 clustered closely together in the phylogenetic tree (Figure 12B).

Figure 11: Phylogenetic correlation plots of gene (A) and in silico translated amino acid sequence (B) similarities of nirK versus 16S rRNA gene similarity.

Dotted vertical lines show the similarity values, below which two sequences always had less than 97 % 16S rRNA gene sequence similarity. Dashed vertical lines show the 90 % quantile of pairwise sequence comparisons with a 16S rRNA gene sequence similarity of at least 97 % (i.e., threshold similarity). The solid vertical lines show the 97 % 16S rRNA gene similarities. Modified from Depkat-Jakob et al. (2013).

Figure 12: Comparison of 16S rRNA gene (A) and nirK (B) phylogenies of different species.

Neighbor-joining trees of 16S rRNA gene (A) and nirK gene (B) sequences fragments were constructed. The percentage of replicate trees the associated taxa clustered together in the bootstrap test (10,000 replicates), are shown at the node of two branches (values below 50 % are not displayed). Numbers indicate the clustering of representative taxa in both trees. The 16S rRNA gene based taxa 1 and 6 to 9 cluster together in the nirK based tree. The asterisks indicate the two nirK copies of Rhodopseudomonas palustris TIE-1. The bars represent an estimated sequence dissimilarity of 0.01 (A) and 0.05 (B).

3.1.1.3.1.2. Phylogenetic correlation plots and comparative tree topologies of nirS and corresponding 16S rRNA genes

For nirS, phylogenetic correlation plots were constructed with 96 nirS sequences together with 95 corresponding 16S rRNA sequences (Thauera sp. 27 contained two copies of nirS) for both nirS gene and in silico translated nirS amino acid sequences (2.5.12.1, Figure 13). Linearity between 16S rRNA gene similarity and both nirS gene and amino acid

1

A

B

AM084004 Ochrobactrum sp. R-24653 16S NC 009667 Ochrobactrum anthropi ATCC 49188 16S

AF229879 Ochrobactrum sp. 4FB13 16S AM084018 Ochrobactrum sp. R-25055 16S AM231060 Ochrobactrum sp. R-26465 16S NZ ACQA01000001 Ochrobactrum intermedium LMG 3301 16S NC 013118 Brucella microti CCM 4915 16S

NC 009504 Brucella ovis ATCC 25840 16S NZ ACJD01000006 Brucella ceti str. Cudo 16S

NZ ACOR01000003 Brucella abortus 2308 A 16S NC 010167 Brucella suis ATCC 23445 16S NC 010104 Brucella canis ATCC 23365 16S NC 012442 Brucella melitensis ATCC 23457 16S AM084043 Rhizobium sp. R-24658 16S NC 007761 Rhizobium etli CFN 42 16S NC 012587 Rhizobium sp. NGR234 16S NZ ACKY01000036 Cardiobacterium hominis ATCC 15826 16S AB453731 Alcanivorax dieselolei N1203 16S

AF094722 Pseudomonas chlororaphis subsp. aureofaciens ATCC 13985 16S NC 004129 Pseudomonas fluorescens Pf-5 16S

NC 009668 Ochrobactrum anthropi ATCC 49188 nirK AM230815 Ochrobactrum sp. R-24653 nirK

NZ ACQA01000001 Ochrobactrum intermedium LMG 3301 nirK AY078252 Ochrobactrum sp. 4FB13 nirK

AM230839 Ochrobactrum sp. R-26465 nirK NC 010104 Brucella canis ATCC 23365 nirK NC 010167 Brucella suis ATCC 23445 nirK NC 013118 Brucella microti CCM 4915 nirK NC 012442 Brucella melitensis ATCC 23457 nirK

NC 009504 Brucella ovis ATCC 25840 nirK NZ ACOR01000007 Brucella abortus 2308 A nirK NZ ACJD01000006 Brucella ceti str. Cudo nirK AM230832 Rhizobium sp. R-24663 nirK NC 007766 Rhizobium etli CFN 42 nirK EF363545 Bosea sp. MF18 nirK

AB453733 Alcanivorax dieselolei N1203 nirK FN600574 Devosia sp. GSM-205 nirK

NC 011004 Rhodopseudomonas palustris TIE-1 nirK 2 GU332847 Rhodopseudomonas sp. 2-8 nirK

NC 011004 Rhodopseudomonas palustris TIE-1 nirK 1 GQ404514 Afipia sp. 4AS1 nirK

NC 004463 Bradyrhizobium japonicum USDA 110 nirK NC 009445 Bradyrhizobium sp. ORS278 nirK

Z21945 Pseudomonas chlororaphis subsp. aureofaciens ATCC 13985 nirK NC 004129 Pseudomonas fluorescens Pf-5 nirK

EF363542 Castellaniella sp. ROi28 nirK EF202175 Alcaligenes sp. CJANPY1 (A-II) nirK

EF202174 Alcaligenes sp. ESPY2 (A-III) nirK

NZ ACKY01000036 Cardiobacterium hominis ATCC 15826 nirK AF339044 Nitrosomonas sp. C-56 nirK

similarity was particularly apparent for a 16S rRNA gene similarity of about ≥ 90 % (Figure 13). The amount of distantly related organisms (i.e., with a 16S rRNA gene similarity ≤ 85 %) carrying highly similar nirS genes (i.e., their nirS sequences were 90 % to 100 % identical) was negligible for both nirS gene and amino acid sequences (Figure 13).

Of all organisms with a ≥ 97 % 16S rRNA gene similarity, 90 % had a nirS similarity of ≥ 82 % (Figure 13A) and a nirS in silico translated amino acid sequence similarity of ≥ 87 % (Figure 13B). Thus, 82 % was defined as a cutoff value to create nirS gene sequence species-level OTUs, i.e., a dissimilarity of two nirS gene sequences of 18 %. For nirS amino acid sequences, this cutoff value was 87 %, i.e., a dissimilarity of 13 % between two nirS amino acid sequences. Both cutoff values are conservative estimates that indicate a minimum amount of species-level OTUs that can be expected.

Comparison of 16S rRNA gene phylogeny and nirS gene phylogeny showed that some taxa were completely separated in both phylogenetic trees (e.g., clusters 2, 3, and 5) whereas other taxa were separated in the 16S rRNA tree only but clustered together in the nirS gene tree (e.g., clusters 1, 7, and 9) (Figure 14). Single sequences of some taxa clustered together in the 16S rRNA gene based tree but were split in the nirS gene tree (e.g., clusters 4 to 6) (Figure 14B). The two nirS gene copies of Thauera sp. 27 were placed in two distinct clusters (clusters 4a and 4c) (Figure 14B).

Figure 13: Phylogenetic correlation plots of gene (A) and in silico translated amino acid sequence (B) similarities of nirS versus 16S rRNA gene similarity.

Dotted vertical lines show the similarity values, below which two sequences always had less than 97 % 16S rRNA gene sequence similarity. Dashed vertical lines show the 90 % quantile of pairwise sequence comparisons with a 16S rRNA gene sequence similarity of at least 97 % (i.e., threshold similarity). The solid vertical lines show the 97 % 16S rRNA gene similarities.

Figure 14: Comparison of 16S rRNA gene (A) and nirS (B) phylogenies of different species.

Neighbor-joining trees of 16S rRNA gene (A) and nirS gene (B) sequences were constructed. The percentage of replicate trees the associated taxa clustered together in the bootstrap test (10,000 replicates), are shown at the node of two branches (values below 50 % are not displayed). Numbers indicate the clustering of representative taxa in both trees; some 16S rRNA gene based taxa are split and therefore indicated with a, b, and c after the number in the nirS based tree.The asterisks indicate the two nirS copies of Thauera sp. 27. The bars represent an estimated sequence dissimilarity of 0.02 (A) and 0.05 (B). AY838759 Thauera sp. 27 nirS clone 8 AM230913 Dechloromonas sp. R-28400 nirS NC 010170 Bordetella petrii DSM 12804 nirS AY078264 Thauera selenatis AX nirS AY838762 Thauera sp. 27 nirS clone 2

GQ384053 Halomonas sp. F15 nirS

NC 009952 Dinoroseobacter shibae DFL 12 nirS NC 008209 Roseobacter denitrificans OCh 114 nirS NC 006569 Ruegeria pomeroyi DSS-3 nirS NC 010170 Bordetella petrii DSM 12804 16S AM084024 Comamonas sp. R-25066 16S NR 028702 Halomonas campisalis 4A ATCC 700597 16S NR 026274 Halomonas desiderata DSM 9502 16S DQ088664 Pseudomonas qianpuensis C10-2 16S

NC 008209 Roseobacter denitrificans OCh 114 16S NC 003911 Ruegeria pomeroyi DSS-3 16S

NC 009952 Dinoroseobacter shibae DFL 12 16S AM084107 Paracoccus sp. R-24665 16S

3.1.1.3.2. Denitrifiers and dissimilatory nitrate reducers detected