amoA-based consensus phylogeny of ammonia-oxidizing archaea and deep sequencing of amoA genes from soils of four different geographic regions

Abstract

Ammonia-oxidizing archaea (AOA) play an important role in nitrification and many studies exploit their amoA genes as marker for their diversity and abundance. We present an archaeal amoA consensus phylogeny based on all publicly available sequences (status June 2010) and provide evidence for the diversification of AOA into four previously recognized clusters and one newly identified major cluster. These clusters, for which we suggest a new nomenclature, harboured 83 AOA species-level OTU (using an inferred species threshold of 85% amoA identity). 454 pyrosequencing of amoA amplicons from 16 soils sampled in Austria, Costa Rica, Greenland and Namibia revealed that only 2% of retrieved sequences had no database representative on the species-level and represented 30-37 additional species-level OTUs. With the exception of an acidic soil from which mostly amoA amplicons of the Nitrosotalea cluster were retrieved, all soils were dominated by amoA amplicons from the Nitrososphaera cluster (also called group I.1b), indicating that the previously reported AOA from the Nitrosopumilus cluster (also called group I.1a) are absent or represent minor populations in soils. AOA richness estimates on the species level ranged from 8-83 co-existing AOAs per soil. Presence/absence of amoA OTUs (97% identity level) correlated with geographic location, indicating that besides contemporary environmental conditions also dispersal limitation across different continents and/or historical environmental conditions might influence AOA biogeography in soils.

Fig. 1

Consensus tree illustrating the five major clusters of archaeal amoA (designations in brackets refer to frequently used corresponding cluster names in AOA 16S rRNA trees) (A) and the diversification of the Nitrososphaera, Nitrososphaera sister, and Nitrosotalea cluster at the second and third phylogenetic level (B). The tree was determined using 592 unambiguously aligned positions of a data set of 735 representing nucleic acid amoA sequences. Each reference sequence is representative for a group of amoA sequences with an identity of ≥ 97%. For the radial overview tree, numbers within major lineages represent numbers of representing sequences (A). For the detailed circular tree, numbers in circles represent the second phylogenetic level (e.g. Nitrososphaera subcluster 1), whereas the third phylogenetic level is directly indicated at the tree branch (e.g. Nitrososphaera subcluster 1.1); sequences that did not form stable subclusters of more than three representatives kept the phylogenetic affiliation of the higher phylogenetic level and are indicated by their NCBI accession number (B). A corresponding detailed circular tree of the Nitrosopumilus cluster is given in Fig. S1. The consensus tree and the source alignment of representing sequences can be found in File S1. The scale bar indicates 10% estimated sequence divergence based on a Jukes-Cantor corrected distance matrix analysis.

Fig. 2

Pairwise comparison of 16S rRNA gene and archaeal amoA identities of all metagenome fragments and Candidatus species from which both genes are known. Sequences with less than 99% identity at the 16S rRNA gene level are considered to belong to different species (Stackebrandt and Ebers, 2006).

Fig. 3

Relative abundance of sequences affiliated with the five major archaeal amoA clusters in the analysed soils (no Nitrosocaldus cluster representatives were detected). The combined analysis of forward and reverse sequenced archaeal amoA gene fragments retrieved by 454 pyrosequencing is shown.

Fig. 4

Principal component analysis based on presence/absence of OTUs (jackknifed unweighted UniFrac) and separating soils according to their geographic origin. For this analysis, observed amoA OTUs at 97% sequence identity were used (representing the highest possible phylogenetic resolution) and normalized to 1300 reads per soil and sequencing direction. The Austrian spruce forest soil was omitted from the analysis due to a sequence number of less than 1300 reads. Analysis of the forward sequences is shown; analysis of reverse sequences gave similar results (data not shown).