How many novel eukaryotic 'kingdoms'? Pitfalls and limitations of environmental DNA surveys

Abstract

Background: Over the past few years, the use of molecular techniques to detect cultivation-independent, eukaryotic diversity has proven to be a powerful approach. Based on small-subunit ribosomal RNA (SSU rRNA) gene analyses, these studies have revealed the existence of an unexpected variety of new phylotypes. Some of them represent novel diversity in known eukaryotic groups, mainly stramenopiles and alveolates. Others do not seem to be related to any molecularly described lineage, and have been proposed to represent novel eukaryotic kingdoms. In order to review the evolutionary importance of this novel high-level eukaryotic diversity critically, and to test the potential technical and analytical pitfalls and limitations of eukaryotic environmental DNA surveys (EES), we analysed 484 environmental SSU rRNA gene sequences, including 81 new sequences from sediments of the small river, the Seymaz (Geneva, Switzerland).

Results: Based on a detailed screening of an exhaustive alignment of eukaryotic SSU rRNA gene sequences and the phylogenetic re-analysis of previously published environmental sequences using Bayesian methods, our results suggest that the number of novel higher-level taxa revealed by previously published EES was overestimated. Three main sources of errors are responsible for this situation: (1) the presence of undetected chimeric sequences; (2) the misplacement of several fast-evolving sequences; and (3) the incomplete sampling of described, but yet unsequenced eukaryotes. Additionally, EES give a biased view of the diversity present in a given biotope because of the difficult amplification of SSU rRNA genes in some taxonomic groups.

Conclusions: Environmental DNA surveys undoubtedly contribute to reveal many novel eukaryotic lineages, but there is no clear evidence for a spectacular increase of the diversity at the kingdom level. After re-analysis of previously published data, we found only five candidate lineages of possible novel high-level eukaryotic taxa, two of which comprise several phylotypes that were found independently in different studies. To ascertain their taxonomic status, however, the organisms themselves have now to be identified.

Figure 1

Identification of the 48 distinct, non-chimeric eukaryotic phylotypes we obtained from our samples of the small river, the Seymaz (Geneva, Switzerland). (A) Phylogenetic positions of the 48 eukaryotic phylotypes we obtained. The tree shown is the result of a minimum evolution analysis of 68 partial SSU rRNA gene sequences, using the GTR + G model of evolution (see text). The number of phylotypes belonging to each higher-level eukaryotic group is indicated in brackets under the clade name. A fast-evolving lineage of undetermined taxonomic position is highlighted in blue. The tree was arbitrarily rooted on opisthokonts. Numbers at nodes are bootstrap support values following 10,000 replicates. All branches are drawn to scale. (B) Relative proportion of phylotypes belonging to each higher-level eukaryotic group.

Figure 2

Bayesian phylogeny of eukaryotes based on the analysis of 125 complete or nearly complete SSU rRNA gene sequences (1,175 positions), including 56 selected environmental phylotypes (indicated in bold). The number of phylotypes belonging to each higher-level eukaryotic group is indicated in brackets next to the clade name. Phylotypes previously considered as novel eukaryotic lineages, which are in fact fast-evolving members of known groups, are highlighted in orange. Phylotypes that could be identified thanks to an increasing taxon sampling are highlighted in green. The remaining phylotypes of undetermined taxonomic position are highlighted in blue. The tree is presented with a basal bifurcation between unikonts (Amoebozoa + opisthokonts) and bikonts. The GTR + G model of evolution was used, and the topology shown is a Bayesian consensus of 20,000 sampled trees (see text). The posterior probability of each resolved node is indicated above branches, while numbers under branches represent bootstrap support following 10,000 replicates of a minimum evolution analysis of the same dataset, using maximum likelihood-corrected estimates of the distances (dashes indicate bootstrap values under 50%). Branches are drawn to scale, except those marked with an asterisk (*), which were reduced by half for clarity.

Figure 3

Bayesian phylogeny of alveolates based on the analysis of 80 complete or nearly complete SSU rRNA gene sequences (1,325 positions), including 44 selected environmental phylotypes (indicated in bold). The number of phylotypes belonging to each of the five main alveolate lineages is indicated in brackets next to the clade name. Phylotypes previously considered as novel eukaryotic lineages, which are in fact fast-evolving members of known groups are highlighted in orange. Phylotypes that could be identified thanks to an increasing taxon sampling are highlighted in green. The tree is rooted with three stramenopile sequences. The GTR + G model of evolution was used, and the topology shown is a Bayesian consensus of 20,000 sampled trees (see text). The posterior probability of each resolved node is indicated. Branches are drawn to scale, except those marked with an asterisk (*), which were reduced by half for clarity.

Figure 4

Bayesian phylogeny of opisthokonts based on the analysis of 80 complete or nearly complete SSU rRNA gene sequences (1,395 positions), including 28 selected environmental phylotypes (indicated in bold). The number of phylotypes belonging to each opisthokont lineage is indicated in brackets next to the clade name. An as yet undetermined lineage is highlighted in blue. The tree is rooted with five amoebozoan sequences. The GTR + G model of evolution was used, and the topology shown is a Bayesian consensus of 20,000 sampled trees (see text). The posterior probability of each resolved node is indicated. All branches are drawn to scale.

Figure 5

Identification of the 289 published phylotypes we re-analysed. (A) As determined by their authors and (B) after our re-analysis, highlighting the relative proportion of previously undetected chimeras and the reduced number of phylotypes of undetermined taxonomic position, compared to the proportion of phylotypes belonging to each defined higher-level eukaryotic group. The phylotypes related to, respectively, Mastigamoeba invertens, Jakoba incarcerata, and the Carpediemonas + Retortamonas + diplomonads lineage were grouped together as 'Excavates'.