1 Goltikree

Thermodesulfobacteria Classification Essay

The increasing size of timetrees in recent years has led to a focus on diversification analyses to better understand patterns of macroevolution. Thus far, nearly all studies have been conducted with eukaryotes primarily because phylogenies have been more difficult to reconstruct and calibrate to geologic time in prokaryotes. Here, we have estimated a timetree of 11,784 ‘species’ of prokaryotes and explored their pattern of diversification. We used data from the small subunit ribosomal RNA along with an evolutionary framework from previous multi-gene studies to produce three alternative timetrees. For each timetree we surprisingly found a constant net diversification rate derived from an exponential increase of lineages and showing no evidence of saturation (rate decline), the same pattern found previously in eukaryotes. The implication is that prokaryote diversification as a whole is the result of the random splitting of lineages and is neither limited by existing diversity (filled niches) nor responsive in any major way to environmental changes.

diversification, timetree, prokaryote, evolution

Introduction

Large and nearly complete molecular phylogenies of eukaryotes have allowed evolutionary biologists to better understand patterns of macroevolution in recent years (Hedges et al. 2015). The expansion model states that diversity increases without limit and depends only on time and diversification rate, which is the balance between speciation and extinction rates (Cornell 2013). On the other hand, density-dependent species production, as from competition or resource limitation, will lead to saturated diversity characterized by a null diversification rate (Cornell 2013). Evidence supporting saturated (Rabosky et al. 2012; Rabosky 2013) or expanding diversity (Morlon et al. 2010; Venditti et al. 2010; Jetz et al. 2012) has been found in recent years for selected groups of eukaryotes. A more inclusive study with a wider sampling coverage (50,455 species) found support for a constant rate of diversification over time in eukaryotes (Hedges et al. 2015). That study also found evidence for saturation and accelerating or decelerating diversification in several eukaryote clades, suggesting that the globally constant rate overall may be the product of averaging many small and random pulses of diversification (Ricklefs 2014). Unfortunately, a species-level timetree of prokaryotes has not been available to conduct similar analyses in those organisms. Instead, existing timetrees are relatively small and primarily involve higher taxonomic groups (Battistuzzi et al. 2004; Battistuzzi and Hedges 2009a, b, c; Jun et al. 2010; Loren et al. 2014; Gubry-Rangin et al. 2015; Hedges et al. 2015).

Prokaryote diversification patterns are expected to differ from those of eukaryotes. For example, horizontal gene transfer (HGT), a widespread mechanism in prokaryotes, can favor the movement of a gene variant, whether adaptive or neutral, between species thereby lowering rates of extinction and confounding boundaries between species (Young 1989; Cohan 2001). Debate continues over the definition of a prokaryote “species”, with some suggesting that the term will eventually be abandoned (Doolittle and Zhaxybayeva 2009) while others see value in the concept, albeit redefined (Staley 2013), and new evidence (Bendall et al. 2016; Cohan 2016) adds fuel to the debate. Because it continues to be used in the literature and databases, we also use the term here, but do not imply any special meaning, and therefore our use of “species” is equivalent to “operational taxonomic unit”.

The few studies that have focused on the macroevolutionary diversification of prokaryotes have done so with small groups (15–153 species) and have obtained mixed results, with constant (Martin et al. 2004; Loren et al. 2014; Gubry-Rangin et al. 2015) or decreasing (Morlon et al. 2012) diversification rates over time. Some experimental observations support the second result, an explosive radiation followed by decay in diversification rates (MacLean 2005; Kassen 2009). However, they were conducted on small taxonomic groups and cannot be extrapolated to all prokaryotes. Indeed, small groups of species are more likely to show patterns of saturated diversity more often than larger groups (Hedges et al. 2015). To explore the global diversification rate of prokaryotes over time, a large and comprehensive timetree is needed.

In order to construct the most complete prokaryote timetree (PTT), we used a comprehensive small subunit (SSU) data set of 11,269 species (Munoz et al. 2011) supplemented by SSU sequences of 684 species of cyanobacteria from the National Center for Biotechnology Information (NCBI) database and in the Ribosomal Database project (RDP; Cole et al. 2013). The SSU gene sequences may be the most accurate way to establish genealogical relationships (Yarza et al. 2008). However, the SSU is subject to base compositional biases that can affect phylogeny, unless corrected (Battistuzzi and Hedges 2009a). Because of this, we constrained the deepest nodes, between families and above, according to phylogenies obtained with multi-protein datasets (Battistuzzi and Hedges 2009a,b,c). For comparison, we also used two other higher level topologies based on many protein orthologs to constrain the relationships (Lang et al. 2013; Rinke et al. 2013).

We studied the diversification patterns of prokaryotes with three approaches. First, we explored variation in net diversification rate over time, using two methods for the prokaryotes as a whole, as well as subclades. We also timed a multi-gene Bacilli phylogenetic tree to compare the diversification patterns obtained with one gene (SSU) versus many genes. Second, we evaluated branch length distribution, which is another way of testing whether the data are clock-like (exponential distribution) or non-clock like (other distributions). We also compared the branch-lengths of prokaryotes and eukaryotes in order to further explore their difference. Finally, because the SSU gene is the only marker available for all described species, we used simulations to investigate how a limited number of variable sites could influence our results.

Results

Phylogenies and Timetrees

The recently released SSU dataset, along with topological constraints (Battistuzzi and Hedges 2009b,c), were combined to produce a species-level PTT of 11,784 species (Topology A; fig. 1). Topological constraints from other studies produced timetrees of 11,771 species (Topology B; Lang et al. 2013) and 11,774 species (Topology C; Rinke et al. 2013). To evaluate our time estimates, we compared our results, node estimates of the PTT, with a timetree of 98 representative prokaryote species built with a different phylogenetic and timing method (Sheridan et al. 2003). They calibrated a Neighbor Joining distance tree, built from SSU rRNA sequences, using a minimum time of 2,650 Ma for the emergence of cyanobacteria. The tree used to calibrate the PTT (Battistuzzi and Hedges 2009b,c) was also built with different genes and calibration points (see the “Materials and Methods” section). We tested the relationship of 32 common node estimates between the PTT and the timetree from Sheridan et al. (2003) (supplementary fig. S2 and table S6, Supplementary Material online). Over the 32 common nodes, 11 were not used as calibration points in the timing process of the PTT because they were not reported in Battistuzzi and Hedges (2009b,c) and none of the 32 nodes was used to calibrate the timetree from Sheridan et al. (2003). When using the 32 common nodes we obtained a significant correlation (regression by origin: P-value < 2.2 × 1016, r2 = 0.95, slope = 0.95; unconstrained regression: P-value = 1.71 × 1006, r2 = 0.54, slope = 0.53). A strong correlation was also obtained with only the 11 nodes when we constrained the regression through the origin (P-value = 1.297 × 106, r2 = 0.91, slope = 1.11). However, without constraint the correlation was not significant (P-value = 0.23, r2 = 0.15, slope = 0.23). Constraining regressions through the origin reflects the age of tips (0 Ma) in both data sets. Without this constraint, the regressions showed a weaker or no correlation which might be explained by the different phylogenetic and timing methods used in both studies.

Diversification Analyses

A significant positive gamma statistic (Pybus and Harvey 2000) obtained for the main PTT did not indicate a decline in diversification rate through time (gamma statistic = 94.5, P-value < 2.2 × 1016). This result was confirmed by our analyses on diversification rates through time using the programs BAMM (Rabosky et al. 2014) and TreePar (Stadler 2013). The first analysis (BAMM: Bayesian Analysis of Macroevolutionary Mixtures) showed a constant net diversification rate over the major part of the PTT (fig. 2a) as well as for the multi-gene Bacilli timetree (fig. 2c). Concerning the PTT (fig. 2a) we also detected a sharp increase in net diversification rate around 30 Ma but this is explained by sampling bias (see section below).

Fig.2

Net diversification rate plots of the PTT (topology A) obtained with the programs BAMM (a) and TreePar (b). Net diversification rate plots of the multi-geneBacilli timetree obtained with the programs BAMM (c), and TreePar (d). Shading denotes confidence on evolutionary rate at ± 95%. Dotted lines represent tree section with <10 nodes involved.

Fig.2

Net diversification rate plots of the PTT (topology A) obtained with the programs BAMM (a) and TreePar (b). Net diversification rate plots of the multi-geneBacilli timetree obtained with the programs BAMM (c), and TreePar (d). Shading denotes confidence on evolutionary rate at ± 95%. Dotted lines represent tree section with <10 nodes involved.

The rate shift analyses (TreePar) gave us the same constant net diversification pattern for the major part of the PTT (between 100 and 3,720 Ma; fig. 2b) and the multi-gene Bacilli timetree (between 50 and 1,884 Ma; fig. 2d) but other sections of the tree showed rates shifts. For the PTT, a model with five rate shifts was not rejected in favor of a model with six rate shifts (P-value = 0.133) (supplementary table S1, Supplementary Material online). The five shifts were detected at 20, 40, 100, 3,720, and 4,180 Ma (supplementary table S1, Supplementary Material online). The parameters obtained for the five-shifts model for the section between 100 and 3,720 Ma were λ = 0.0483 and µ = 0.0469 with λ being the speciation rate and µ the extinction rate. For the multi-gene Bacilli timetree, a model with one rate shift was not rejected in favor of a model with two rate shifts (P-value = 0.439). The shift was detected at 50 Ma (supplementary table S1, Supplementary Material online). The parameters obtained for the one-shift model for the section between 50 and 1,884 Ma were λ = 0.0719 and µ = 0.0704 (fig. 2d). Similar results were obtained with the alternative topologies (B and C), when using 500,000 as the number of prokaryote species and when removing the archaea that might suffer for a higher bias regarding the estimation of the number of species (Castelle et al. 2015) (supplementary fig. S3, Supplementary Material online). The maximum a posteriori probability shift configuration of the PTT was determined with the program BAMM, showing 215 shifts between lineages. We also evaluated rate shifts within subclades of the PTT. We did not take into account plot intervals with <30 nodes involved (

The Thermodesulfobacteria are a phylum[1] of thermophilic[2]sulfate-reducing bacteria.

Phylogeny[edit]

See also: Bacterial taxonomy

The phylogeny based on the work of the All-Species Living Tree Project.[3]

Taxonomy[edit]

The currently accepted taxonomy is based on the List of Prokaryotic names with Standing in Nomenclature (LSPN)[4] and the National Center for Biotechnology Information (NCBI).[5]

Notes:
♠ Strain found at the National Center for Biotechnology Information (NCBI) but not listed in the List of Prokaryotic names with Standing in Nomenclature (LPSN)

References[edit]

  1. ^Vick TJ, Dodsworth JA, Costa KC, Shock EL, Hedlund BP (March 2010). "Microbiology and geochemistry of Little Hot Creek, a hot spring environment in the Long Valley Caldera". Geobiology. 8 (2): 140–54. doi:10.1111/j.1472-4669.2009.00228.x. PMID 20002204. 
  2. ^Jeanthon C, L'Haridon S, Cueff V, Banta A, Reysenbach AL, Prieur D (May 2002). "Thermodesulfobacterium hydrogeniphilum sp. nov., a thermophilic, chemolithoautotrophic, sulfate-reducing bacterium isolated from a deep-sea hydrothermal vent at Guaymas Basin, and emendation of the genus Thermodesulfobacterium". Int. J. Syst. Evol. Microbiol. 52 (Pt 3): 765–72. doi:10.1099/ijs.0.02025-0. PMID 12054236. 
  3. ^"16S rRNA-based LTP release 123 (full tree)"(PDF). Silva Comprehensive Ribosomal RNA Database. Retrieved 2016-03-20. 
  4. ^J.P. Euzéby. "Thermodesulfobacteria". List of Prokaryotic names with Standing in Nomenclature (LPSN). Retrieved 2016-03-20. 
  5. ^Sayers; et al. "Thermodesulfobacteria". National Center for Biotechnology Information (NCBI) taxonomy database. Retrieved 2016-03-20. 

Leave a Comment

(0 Comments)

Your email address will not be published. Required fields are marked *