Genome expansion in bacteria: the curious case of Chlamydia trachomatis
© Bohlin. 2015
Received: 28 January 2015
Accepted: 21 September 2015
Published: 30 September 2015
Recent findings indicated that a correlation between genomic % AT and genome size within strains of microbial species was predominantly associated with the uptake of foreign DNA. One species however, Chlamydia trachomatis, defied any explanation. In the present study 79 fully sequenced C. trachomatis genomes, representing ocular- (nine strains), urogenital- (36 strains) and lymphogranuloma venereum strains (LGV, 22 strains), in three pathogroups, in addition to 12 laboratory isolates, were scrutinized with the intent of elucidating the positive correlation between genomic AT content and genome size.
The average size difference between the strains of each pathogroup was largely explained by the incorporation of genetic fragments. These fragments were slightly more AT rich than their corresponding host genomes, but not enough to justify the difference in AT content between the strains of the smaller genomes lacking the fragments. In addition, a genetic region predominantly found in the ocular strains, which had the largest genomes, was on average more GC rich than the host genomes of the urogenital strains (58.64 % AT vs. 58.69 % AT), which had the second largest genomes, implying that the foreign genetic regions cannot alone explain the association between genome size and AT content in C. trachomatis. 23,492 SNPs were identified for all 79 genomes, and although the SNPs were on average slightly GC rich (~47 % AT), a significant association was found between genome-wide SNP AT content, for each pathogroup, and genome size (p < 0.001, R 2 = 0.86) in the C. trachomatis strains.
The correlation between genome size and AT content, with respect to the C. trachomatis pathogroups, was explained by the incorporation of genetic fragments unique to the ocular and/or urogenital strains into the LGV- and urogential strains in addition to the genome-wide SNP AT content differences between the three pathogroups.
Several studies have described the evolutionary process of genome reduction in which microbes undergo genome degradation as they become attached to a host, either in a parasitic or a mutualistic relationship . The genomes of these microbes tend to be marked by dysfunctional DNA maintenance proteins and genetic drift which in turn lead to accumulated mutations in genes that are not important for the microbe’s existence . Superfluous genes are eventually lost, since there appears to be an evolutionary drive in the direction of genome minimization . Although exceptions exists , the genomes of these symbionts/mutualists are to a large extent AT-rich due to the mutational bias towards AT-richness . A negative correlation has been found between genomic AT content and genome size in some prokaryotic phyla [5, 6] (i.e. genome size decrease with increasing AT content). However, it was recently shown that for strains within several microbial species there was a positive correlation between genomic AT content and genome size . This correlation was linked to horizontal gene transfer (HGT), DNA uptake and/or recombination of AT rich sequences , but the species Chlamydia trachomatis seemed to be an exception since it could not be shown that foreign DNA had been integrated into the genomes of any of the sequenced strains .
Chlamydia trachomatis is a Gram-negative, obligate intracellular pathogen leading a parasitic lifestyle presumably only in humans . Of the sequenced and assembled strains, the species’ AT content ranges from 58.5 to 58.7 % and the genome size from 1,038,310 to 1,083,890 bp. (See Additional file 1 for details). C. trachomatis has been divided into three distinct lineages, or pathogroups, based on serovars; one associated with trachoma (serovars A–C) another with urogenital infections (serovars D–K) and the third with lymphogranuloma venereum (LGV) and proctitis (serovars L1–L3) .
Although C. trachomatis is presumed to affect humans only, other animals may be infected by different species of Chlamydia. Since the focus of interest here is genome expansion in C. trachomatis the reader is referred elsewhere for more information concerning other species of Chlamydia [8, 9].
Hence, the aims of the study were to scrutinize the positive association between AT content and genome size in C. trachomatis to shed more light on the evolutionary processes involved.
Exploration of the different pathogroups
Genome size, base composition and mutational bias
Average genome size and AT content in the different pathogroups
Number of strains
Average genome size (mb) (±95 % CI)
Average AT content (±95 % CI)
All 79 genomes
Genes and proteins unique to the urogenital biovar
Gene fragment in serovars
Part of cytotoxin gene (AY647992.1)
Putative cytoadherence factor (CAX09718.1)
Cytotoxic genes (homologous to AY647998.1)
TcdA/TcdB catalytic glycosyltransferase domain (pfam12919)
TcdB toxin N-terminal helical domain (pfam12918)
DUF3491 superfamily (pfam11996)
Average size and AT content of fragments unique to pathogroups
Average size (bp) (95 % CI)
Average AT content (95 % CI)
Genetic region not found in LGV strains
Genetic region exclusive to the ocular strains
Genetic region found in both ocular and urogenital strains
AT content estimation of serovars A–C/D–K
Genomic fragment + serovar
Estimated AT content from fragments
Estimated AT content + SNP AT
Observed average AT content
Sr L1-L3 + fragment Sr A–C
Sr L1-L3 + fragment Sr D–K
Sr D-K + fragment Sr A–C
Evolutionary analyses of the C. trachomatis strains
Compelling evidence has been put forth that there has been an early split in C. trachomatis from the LGV strains and the urogenital strains into two distinct clades, or biovars . Figure 1 indicates that the passing of this coalescent event must be one of the earliest known occurrences of a split within C. trachomatis. Within the LGV clade there has been recorded only sporadic re-combination events and even fewer such events between the LGV and urogenital strains [16, 22]. Hence, the LGV clade has, at least until now, been largely isolated from the urogenital strains [16, 23]. Within the urogenital strains however, several splitting events have taken place, in addition to numerous recombination events [12, 16, 22]. The earliest splitting within the urogenital clade seems to have resulted in two new minor clades (T1 and T2) both comprising predominantly urogenital strains [12, 22, 23]. Figure 3 shows a distinct size difference between the C. trachomatis A–C serovars and the urogenital pathogroup, suggesting an even more recent split within the urogenital T2 clade into the ocular pathogroup. That the strains from the A–C serovars have had a recent common ancestor with the T2 clade can be seen from the SNP based phylogenetic tree in Fig. 1 as well as in numerous studies [12, 16, 23]. That the ocular strains have emerged only once from the T2 clade, as suggested by Harris et al.  seems therefore to be supported by the approximately 1670 bp genetic fragment predominantly found within the A–C serovars. Interestingly, this genetic fragment was found to be GC-rich and increased GC content in genes has recently been associated with increased fitness [24, 25].
The LGV strains do not appear to have undergone any further major splits, and the strains’ genomes seem to be more homogeneous genome-wise than the strains within the urogenital clades .
The last common ancestor of the LGV- and urogenital strains
Due to the genome-wise homogeneity within the LGV clade it is tempting to speculate that these strains are more similar to the last common ancestor of both the LGV and urogenital strains. Within the urogenital clade however, there is wide-spread recombination [12, 16, 22, 23] indicating far more genomic plasticity than what has been shown for the LGV clade [16, 22]. Horizontal transfer of DNA from distantly related bacteria to C. trachomatis does not seem to have taken place, at least not to a large extent. Indeed, the pan- and core genomes of the C. trachomatis strains have been found to be of similar size suggesting that genetic exchange takes place mostly within the same species . In fact, as was noted above, genetic exchange appears to predominantly occur within the same biovar , which suggests that the strains within the urogenital clade have changed considerably more genome-wise with respect to the last common ancestor of both urogenital- and LGV clades.
Analyses of C. trachomatis SNP AT content
From Fig. 5 it can be observed that the SNPs from the ocular strains are more AT rich than those from the urogenital strains, which in turn are more AT rich than those from the LGV strains. The largest difference detected in SNP AT content between the three lineages is within the ocular strains. This could mean that the split from the T2 clade mentioned above may be a consequence of a shift in environmental conditions leading to a substantial increase in genetic drift and/or a considerable reduction in population size . In this regard it is interesting to recall the fact that the 23,942 SNPs identified genome-wide for all 79 C. trachomatis genomes were slightly GC rich (see Fig. 5). The GC-richness of the SNPs may indicate that C. trachomatis is, in general, subjected to selective constraints [24–26]. Since increased GC content has been associated with intensified selective pressures [24, 25] the larger genome size of the progressively more AT rich genomes of the urogenital- and ocular strains, respectively, could indicate that the increase in genome size in these strains are in fact a consequence of reduced selective constraints that allowed the strains to incorporate the genetic fragments . Or, alternatively, the split into urogenital- and ocular pathogroups is so recent that the consequences of relaxed selective pressure has not started to result in reduced genome sizes. Nevertheless, the substantial increase in SNP AT richness for the ocular strains may support a recent dramatic split from the T2 clade .
C. trachomatis genome size and genomic fragments unique to the urogenital clade
Genome size in C. trachomatis was found to correlate with genomic AT content. A more detailed investigation, as can be seen in Fig. 3, revealed that the correlation between genome size and AT content was linked to the different pathogroups (excluding the artificially manipulated laboratory isolates). The correlation between AT content and genome size in the C. trachomatis pathogroups was largely explained by the genome size and the AT content of the genetic fragments unique to serovars A–C and D–K (see Table 4). On average, however, the addition of AT content from these genetic fragments to the genomic AT content of the LGV- and urogenital pathogroups resulted in slightly less AT rich genomes than what was observed from the strains incorporating these fragments. The average AT content differences between the pathogroups from the 23492 SNPs were therefore also incorporated into the AT content estimates of the pathogroups lacking the genetic fragments unique to servars A–C/D–K. This resulted in slightly more AT rich pathogroups on the whole suggesting that both mutational biases, as described by the SNPs, as well as the incorporation of the genetic fragments unique to the urogenital- and ocular pathogroups explained the correlation between genome size and AT content in C. trachomatis. The lack of correlation between AT content and genome size within the LGV- and urogenital lineages appears therefore to be explained by, first and foremost, the absence of the genetic fragments. The difference in AT content within all three lineages, as can be observed from both Figs. 3, 5, seems to be described by smaller recombination events and mutational biases [21, 22]. The genetic fragments, or lack thereof, as in the case of the LGV strains, have previously been linked to symptoms particular to each pathogroup, i.e. urogenital infections (serovar D–K) and trachoma (serovars A–C) [20, 23, 28].
The origins of the genetic fragments unique within the C. trachomatis urogenital biovar
The genetic fragments particular to the trachoma and urogenital strains discussed here are widely present in other Chlamydia-species known to infect other animals [8, 9, 14, 15]. Since the genetic fragments unique to the urogenital and/or oculars strains, described in the present work, are present in several other Chlamydia species a natural question is why they are not present in the LGV strains (and some only partly in the urogenital strains). One possibility is that the human infecting urogenital and LGV strains have undergone genome reduction from a common ancestor with the ocular strains. Since all the genetic fragments present in the ocular and urogenital strains are extant in several other Chlamydia species it is conceivable that they have been lost and re-gained in the C. trachomatis strains. However, this is difficult to document and if so this cannot have occurred in recent times since a BLAST search  with the genetic fragments unique to C. trachomatis ocular- and urogenital strains were found to have a base composition substantially different to that found in the other Chlamydia species [8, 9]. It is also at odds with recent genomic research on C. trachomatis, which provide compelling evidence that the urogenital and LGV biovars shared a last common ancestor [12, 16]. Hence, a more plausible scenario is that the genetic fragments exclusive to serovars A-C and the urogenital strains have been attained due to more recent recombination. This could indicate that the genetic fragments found only in the urogenital- and ocular pathogroups are present in other yet to be discovered C. trachomatis strains. Indeed, in Figs. 3, 5 one outlier from the urogenital lineage can be observed. This outlier was identified as C. trachomatis E SotoenE8. The strain appears to have a duplicated region in the plasticity zone  (approximate genomic position: 190,300–195,300, found to contain the proteins: PhnP, GloB, AlsT with the additional protein domains identified: SET, ABC_ATPase, PEBP_bact_arch, Ftsk_gamma) also present in the other urogenital strains. Interestingly, the same genetic fraction consisting of approximately 1670 nucleotides predominantly found in the ocular strains was also present in this urogenital strain. Yet, based on whole genome SNPs, the C. trachomatis E SotoenE8 strain cluster with the urogenital T1 lineage (see Fig. 1). This may indicate that the genetic fragment predominantly found in the A–C serovars is also available for the D–K serovars suggesting that the acquisition of this genetic fragment may not alone sufficiently explain the splitting of the T2 clade into the ocular pathogroup.
DNA uptake in Chlamydia sp.
An example of DNA uptake in Chlamydia sp. was observed in the previously described pig pathogen C. suis which has been found to carry a GI with antibiotic resistance genes . It is believed that C. suis may have picked up the GI as a consequence of co-infection with the γ-proteobacterium Aeromonas salmonicida and the ε-proteobacterium Helicobacter pylori . Although many co-infecting bacteria may not be directly pathogenic to their hosts the case of C. suis shows that they can provide genes conferring resistance to tetracyclines, even though they belong to different phyla and have very different base compositions (The average AT content of A. salmonicida is ~41.5 %, compared to ~58 % for C. suis) . The C. trachomatis laboratory isolates, mentioned previously, shows that gene transfer can take place between C. trachomatis and distantly related bacteria as well and therefore that the human pathogen can potentially be an even more serious threat to public health; not only in developing countries but also in developed countries as well since C. trachomatis infections are, in general, treated with tetracycline-based antibiotics [15, 28].
The correlation between genome size and AT content in C. trachomatis cannot be explained by the incorporation of AT-rich genetic fragments alone. Additionally accounting for SNP AT content differences, however, largely resolves the differences in AT content observed between the pathogroups. Whether the association between SNP AT content and genome size is due to relaxed selective forces remains to be identified.
Due to few samples, outliers and indications of heteroscedasticity, correlation/association between genomic AT content and genome size was performed using “MM” type regression , from the “robust” package in R, for increased reliability. For the same reasons, statistical comparisons between pathogroups and genome size was performed using a modified version of Tukey’s test with the method described by Hothorn et al. , and implemented in the R package “multcomp”, incorporating the vcovHC3 sandwich estimator  to compensate for the heteroscedastic variance. The average values found in Tables 1, 2 and 4 were computed using Welch’s t test with 95 % confidence intervals (95 % CI). The non-linear association observed between SNP AT content in the 79 C. trachomatis genomes and genome size was modeled using a spline-based generalized additive model (GAM)  with genome size as the response variable and AT content as the explanatory variable. Details regarding the statistical estimations can be found in Additional file 2. The estimation of AT content change in the LGV and urogenital pathogroups was calculated by adding the average AT content of the genetic fragment to the average AT content of the genomes in the pathogroups of which the genetic fragments were missing and divide by the total number of nucleotides (i.e. average genome size of the genetic fragments in question and C. trachomatis genomes, respectively). The smallest and largest genome sizes as well as the highest and lowest AT content were calculated to produce the minimum/maximum intervals for both genome fragment and genome-wide AT content and genome size. An analogous procedure was carried out to account for AT content differences due to SNPs between the pathogroups; the 23492 SNPs were divided by the corresponding average pathogroup genome size and multiplied by the difference in SNP AT content between the pathogroups. The resulting estimates were added to the above-mentioned calculations regarding the genetic fractions unique to the A–C/D–K serovars. All statistical assessments were carried out in R .
I wish to thank the reviewers for their many good and constructive remarks.
Compliance with ethical guidelines
Competing interests The author declare that he has no competing interests.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- McCutcheon JP, Moran NA. Extreme genome reduction in symbiotic bacteria. Nat Rev Microbiol. 2012;10(1):13–26. doi:https://doi.org/10.1038/nrmicro2670.Google Scholar
- Wernegreen JJ. Endosymbiont evolution: predictions from theory and surprises from genomes. Ann N Y Acad Sci. 2015;. doi:https://doi.org/10.1111/nyas.12740.PubMedGoogle Scholar
- Merhej V, Royer-Carenzi M, Pontarotti P, Raoult D. Massive comparative genomic analysis reveals convergent evolution of specialized bacteria. Biol Direct. 2009;4:13. doi:https://doi.org/10.1186/1745-6150-4-13.PubMed CentralView ArticlePubMedGoogle Scholar
- Hershberg R, Petrov DA. Evidence that mutation is universally biased towards AT in bacteria. PLoS Genet. 2010;6(9):e1001115. doi:https://doi.org/10.1371/journal.pgen.1001115.PubMed CentralView ArticlePubMedGoogle Scholar
- Mitchell D. GC content and genome length in Chargaff compliant genomes. Biochem Biophys Res C. 2007;353(0006-291; 1):207–10.View ArticlePubMedGoogle Scholar
- Li XQ, Du D. Variation, evolution, and correlation analysis of C + G content and genome or chromosome size in different kingdoms and phyla. PLoS One. 2014;9(2):e88339. doi:https://doi.org/10.1371/journal.pone.0088339.PubMed CentralView ArticlePubMedGoogle Scholar
- Bohlin J, Sekse C, Skjerve E, Brynildsrud O. Positive correlations between genomic % AT and genome size within strains of bacterial species. Environ Microbiol Rep. 2014;6(3):278–86. doi:https://doi.org/10.1111/1758-2229.12145.View ArticlePubMedGoogle Scholar
- Nunes A, Gomes JP. Evolution, phylogeny, and molecular epidemiology of Chlamydia. Infect Genet Evol. 2014;23:49–64. doi:https://doi.org/10.1016/j.meegid.2014.01.029.View ArticlePubMedGoogle Scholar
- Bachmann NL, Polkinghorne A, Timms P. Chlamydia genomics: providing novel insights into chlamydial biology. Trends Microbiol. 2014;22(8):464–72. doi:https://doi.org/10.1016/j.tim.2014.04.013.View ArticlePubMedGoogle Scholar
- Gomes JP, Bruno WJ, Borrego MJ, Dean D. Recombination in the genome of Chlamydia trachomatis involving the polymorphic membrane protein C gene relative to ompA and evidence for horizontal gene transfer. J Bacteriol. 2004;186(13):4295–306. doi:https://doi.org/10.1128/JB.186.13.4295-4306.2004.PubMed CentralView ArticlePubMedGoogle Scholar
- Bohlin J, Snipen L, Cloeckaert A, Lagesen K, Ussery D, Kristoffersen AB, et al. Genomic comparisons of Brucella spp. and closely related bacteria using base compositional and proteome based methods. BMC Evol Biol. 2010;10:249. doi:https://doi.org/10.1186/1471-2148-10-249.PubMed CentralView ArticlePubMedGoogle Scholar
- Harris SR, Clarke IN, Seth-Smith HM, Solomon AW, Cutcliffe LT, Marsh P et al. Whole-genome analysis of diverse Chlamydia trachomatis strains identifies phylogenetic relationships masked by current clinical typing. Nat Genet. 2012;44(4):413–9, S1. doi:https://doi.org/10.1038/ng.2214.
- Bohlin J, van Passel MW, Snipen L, Kristoffersen AB, Ussery D, Hardy SP. Relative entropy differences in bacterial chromosomes, plasmids, phages and genomic islands. BMC Genom. 2012;13:66. doi:https://doi.org/10.1186/1471-2164-13-66.View ArticleGoogle Scholar
- Jeffrey BM, Suchland RJ, Eriksen SG, Sandoz KM, Rockey DD. Genomic and phenotypic characterization of in vitro-generated Chlamydia trachomatis recombinants. BMC Microbiol. 2013;13:142. doi:https://doi.org/10.1186/1471-2180-13-142.PubMed CentralView ArticlePubMedGoogle Scholar
- Suchland RJ, Sandoz KM, Jeffrey BM, Stamm WE, Rockey DD. Horizontal transfer of tetracycline resistance among Chlamydia spp. in vitro. Antimicrob Agents Chemother. 2009;53(11):4604–11. doi:https://doi.org/10.1128/AAC.00477-09.PubMed CentralView ArticlePubMedGoogle Scholar
- Ferreira R, Antelo M, Nunes A, Borges V, Damiao V, Borrego MJ, et al. In silico scrutiny of genes revealing phylogenetic congruence with clinical prevalence or tropism properties of Chlamydia trachomatis strains. G3 (Bethesda). 2014;5(1):9–19. doi:https://doi.org/10.1534/g3.114.015354.Google Scholar
- Carlson JH, Hughes S, Hogan D, Cieplak G, Sturdevant DE, McClarty G, et al. Polymorphisms in the Chlamydia trachomatis cytotoxin locus associated with ocular and genital isolates. Infect Immun. 2004;72(12):7063–72. doi:https://doi.org/10.1128/IAI.72.12.7063-7072.2004.PubMed CentralView ArticlePubMedGoogle Scholar
- Carver TJ, Rutherford KM, Berriman M, Rajandream MA, Barrell BG, Parkhill J. ACT: the artemis comparison tool. Bioinformatics. 2005;21(16):3422–3. doi:https://doi.org/10.1093/bioinformatics/bti553.View ArticlePubMedGoogle Scholar
- Lawrence JG, Ochman H. Amelioration of bacterial genomes: rates of change and exchange. J Mol Evol. 1997;44(4):383–97.View ArticlePubMedGoogle Scholar
- Jeffrey BM, Suchland RJ, Quinn KL, Davidson JR, Stamm WE, Rockey DD. Genome sequencing of recent clinical Chlamydia trachomatis strains identifies loci associated with tissue tropism and regions of apparent recombination. Infect Immun. 2010;78(6):2544–53. doi:https://doi.org/10.1128/IAI.01324-09.PubMed CentralView ArticlePubMedGoogle Scholar
- Oren Y, Smith MB, Johns NI, Kaplan Zeevi M, Biran D, Ron EZ, et al. Transfer of noncoding DNA drives regulatory rewiring in bacteria. Proc Natl Acad Sci USA. 2014;111(45):16112–7. doi:https://doi.org/10.1073/pnas.1413272111.PubMed CentralView ArticlePubMedGoogle Scholar
- Joseph SJ, Didelot X, Rothschild J, de Vries HJ, Morre SA, Read TD, et al. Population genomics of Chlamydia trachomatis: insights on drift, selection, recombination, and population structure. Mol Biol Evol. 2012;29(12):3933–46. doi:https://doi.org/10.1093/molbev/mss198.PubMed CentralView ArticlePubMedGoogle Scholar
- Nunes A, Borrego MJ, Gomes JP. Genomic features beyond Chlamydia trachomatis phenotypes: what do we think we know? Infect Genet Evol. 2013;16:392–400. doi:https://doi.org/10.1016/j.meegid.2013.03.018.View ArticlePubMedGoogle Scholar
- Raghavan R, Kelkar YD, Ochman H. A selective force favoring increased G + C content in bacterial genes. Proc Natl Acad Sci USA. 2012;109(36):14504–7. doi:https://doi.org/10.1073/pnas.1205683109%3B10.1073/pnas.1205683109.PubMed CentralView ArticlePubMedGoogle Scholar
- Hildebrand F, Meyer A, Eyre-Walker A. Evidence of selection upon genomic GC-content in bacteria. PLoS Genet. 2010;6(9):e1001107. doi:https://doi.org/10.1371/journal.pgen.1001107.PubMed CentralView ArticlePubMedGoogle Scholar
- Rocha EP, Feil EJ. Mutational patterns cannot explain genome composition: are there any neutral sites in the genomes of bacteria? PLoS Genet. 2010;6(9):e1001104. doi:https://doi.org/10.1371/journal.pgen.1001104.PubMed CentralView ArticlePubMedGoogle Scholar
- Moran NA, McLaughlin HJ, Sorek R. The dynamics and time scale of ongoing genomic erosion in symbiotic bacteria. Science (New York, NY). 2009;323(5912):379–82. doi:https://doi.org/10.1126/science.1167140.View ArticleGoogle Scholar
- Seth-Smith HM, Thomson NR. Whole-genome sequencing of bacterial sexually transmitted infections: implications for clinicians. Curr Opin Infect Dis. 2013;26(1):90–8. doi:https://doi.org/10.1097/QCO.0b013e32835c2159.View ArticlePubMedGoogle Scholar
- Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990;215(3):403–10.View ArticlePubMedGoogle Scholar
- Benson DA, Clark K, Karsch-Mizrachi I, Lipman DJ, Ostell J, Sayers EW. GenBank. Nucleic Acids Res. 2014;42(1):D32–7. doi:https://doi.org/10.1093/nar/gkt1030.PubMed CentralView ArticlePubMedGoogle Scholar
- Darling AE, Mau B, Perna NT. progressiveMauve: multiple genome alignment with gene gain, loss and rearrangement. PLoS One. 2010;5(6):e11147. doi:https://doi.org/10.1371/journal.pone.0011147.PubMed CentralView ArticlePubMedGoogle Scholar
- Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. MEGA6: molecular evolutionary genetics analysis version 6.0. Mol Biol Evol. 2013;30(12):2725–9.PubMed CentralView ArticlePubMedGoogle Scholar
- Paradis E, Claude J, Strimmer K. APE: analyses of phylogenetics and evolution in R language. Bioinformatics. 2004;20(2):289–90.View ArticlePubMedGoogle Scholar
- Akaike H. A new look at the statistical model identification. IEEE Trans Auto Control. 1974;AC-19(6):716–23.View ArticleGoogle Scholar
- Katoh K, Standley DM. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol Biol Evol. 2013;30(4):772–80. doi:https://doi.org/10.1093/molbev/mst010.PubMed CentralView ArticlePubMedGoogle Scholar
- Yohai V, Stahel W, Zamar R. A procedure for robust estimation and inference in linear regression. The IMA volumes in mathematics and its applications. New York: Springer; 1991. p. 365–74.Google Scholar
- Herberich E, Sikorski J, Hothorn T. A robust procedure for comparing multiple means under heteroscedasticity in unbalanced designs. PLoS One. 2010;5(3):e9788. doi:https://doi.org/10.1371/journal.pone.0009788.PubMed CentralView ArticlePubMedGoogle Scholar
- Zeileis A. Econometric computing with HC and HAC covariance matrix estimators. J Stat Soft. 2004;11(10):1–17.Google Scholar
- Wood SN. Generalized additive models: an introduction with R. Chapman & Hall/CRC Texts in Statistical Science. CRC Press; 2006.Google Scholar
- R Development CT. R: a language and environment for statistical computing. 2007. http://www.r-project.org/.