Universal 'tree of life'

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.


Prokaryotic evolution was initially thought to be 'tree-like' and scientists believed that the reconstruction of a universal 'tree of life' was within reach. However this has now shown to be an inaccurate portrayal of prokaryotic evolution. The existence of horizontal gene transfer (HGT) has been known for many years, however it was widely disregarded and thought to be insignificant. The emergence of molecular techniques made genome sequencing possible. This enabled scientists to form methods of HGT detection and realise the frequency and impact of horizontal gene transfer (HGT) in prokaryotic evolution. After doing so, scientists were not only far from forming a universal 'tree of life', but they were in fact on the verge of destroying the notion that it portrays prokaryotic evolution. Different methods of HGT detection have allowed scientists to study the frequency, influence and distribution of HGT with different results and conclusions obtained. This lead some to believe that prokaryotic evolution is very much 'tree-like' while others suggesting that prokaryotic evolution has greater resemblance to a web-like structure. HGT is not an event which occurs exclusively in prokaryotes as it has been also detected in eukaryotes. As genome sequencing of prokaryotic organisms is an ongoing project, day by day scientists are able to draw a clearer picture of prokaryotic evolution and the methods of inheritance involved, which may save or completely destroy the 'tree of life'.


From the moment Darwin coined the term 'tree of life' to describe the pattern of phylogenetic evolution the race began to form a universal 'tree of life' which would represent all living species and their evolutionary lineages. The 'tree-like' structure involves the base or root of the tree being the last universal common ancestor (LUCA) which is the single common ancestor that all currently living organisms on earth have evolved from. This then leads to the trunk and splits numerous times to form a bifurcating tree with branches that reach the top being the species that are currently present (Barton et al, 2007). However at the time where some thought it was near and only a matter of time for it to be reconstructed, molecular techniques and DNA sequencing emerged. Through the use of these techniques, scientists were able to reveal the frequency and influence of HGT especially between prokaryotes. Now the presence of HGT has been consolidated and acknowledged, the question is how important is it and how much of an impact has it had on the evolution of life? Can a 'tree of life' still be formed or will we have to think of another way to depict species phylogenies?

The history of "tree of life":

The first known use of a 'tree-like' structure to depict the relationship between organisms was in 1801 by Augustin Augier. However it was not an evolutionary tree as it involved the presence of a creator. In 1809 Jean-Baptiste Lamarck described an evolutionary 'tree of life' for animals and can be thought of as the first person to form an evolution based 'tree of life' (Archibald, 2009). Other scientists then went on to develop phylogenetic trees. This included Charles Darwin, who in his landmark Origins of Species book presented what is widely recognised as the first diagrammatic representation of the 'tree of life'. About the subject he famously wrote:

"The affinities of all the beings of the same class have sometimes been represented by a great tree. I believe this simile largely speaks the truth. The green and budding twigs may represent existing species; and those produced during former years may represent the long succession of extinct species...The limbs divided into great branches, and these into lesser and lesser branches, were themselves once, when the tree was young, budding twigs; and this connection of the former and present buds by ramifying branches may well represent the classification of all extinct and living species in groups subordinate to groups...so by generation I believe it has been with the great Tree of Life." (Darwin.C, 1859; as quoted in Bapteste.E, 2009).

Using Darwins concepts, Ernst Haeckel (1866) was able to construct the first 'tree of life' using named species. His tree was branched into 3 main lineages; Plantae, Protista and Animalia with the root being Monera which is now under the 3 domain system, divided into Archaea and Bacteria (Barton, 2007).

Scientists continued to develop phylogenic trees particularly for multicellular organisms based on their morphology and phenotypic characteristics and soon, evolutionary relationships of plants, fish, mammals and many other organisms were portrayed using phylogenic trees which all followed a vertical inheritance of organisms. However the lack of a complex intracellular structure and morphology meant that a phylogenic tree for unicellular organisms or prokaryotes was much harder to form. As the phylogenetic 'tree of life' was refined and explored, it was divided into 2 main lineages (prokaryotes and eukaryotes) in 1937 by Edourd Chatton based on the absence or presence of a nucleus. In 1959 Robert Whittaker developed the five-kingdom tree which included prokaryotes (Monera), unicellular eukaryotes (Protista) and multicellular eukaryotes (Animalia, Fungi and Plantae) (Barton, 2007). As modern day molecular techniques developed, Woese and his colleagues were able to analyse and compare the nucleotide sequence of small subunit rRNA (SSU rRNA) in a large number of species. The SSU rRNA is present in nuclear, organellar and prokaryotic DNA, and has been conserved through evolution which made it a very reliable 'molecular chronometer' (Woese, 1987). Using their findings, they were able to modify the 'tree of life' into 3 main lineages; eukaryotes, bacteria and previously unrecognised archaebacteria (now called archaea) and in 1990 Woese formed a modified universal 'tree of life' with the three subdivisions assigned to a new taxanomic status as the 3 domains (Woese et al, 1977).

The emergence of horizontal gene transfer

After Woese used molecular analysis to form the most comprehensive phylogenetic tree seen so far, it was hoped that the advances of molecular techniques would take us closer to finally forming a universal 'tree of life'. However the opposite occurred and as gene sequencing increased in speed in the past 2 decades, questions have been raised over the use of the tree metaphor to portray the evolutionary process of organisms. In 1960 a group of Japanese scientists showed that antiobiotic resistance in bacteria is transferred from one species to another (Akiba et al, 1960). This was one of the first examples of horizontal gene transfer (HGT) seen, however for many years whilst the presence of HGT was acknowledged, the frequency and impact of it was ignored and thought to be insignificant. Scientists claimed that HGT was a rare occurrence involving the transfer of less-important operational genes whilst the 'core' genes involved in DNA replication and protein synthesis did not undergo HGT and are passed on as previously thought, through vertical inheritance. However, scientists began to show that HGT occurred more frequently between closely related and also in distantly related organisms, and in a wider range of genes between bacteria and archaea then once thought, which put a question mark over the validity of the universal 'tree of life' and the idea that only vertical inheritance is involved in evolution of species. After the whole genome of an E.coli strain was sequenced, Lawrence and Ochman (1998) analysed its GC content which although it varies between different species, the GC content of genes within a genome are quiet similar and so any significant differences in GC content of a gene compared to the rest of the genome suggests that it has been introduced through HGT. They found that almost 18% of E.coli gene content had been introduced through HGT since the E.coli species lineage diverged from Salmonella enterica around 100 million of years ago (Lawrence and Ochman, 1998). As scientists began to use a variety of gene markers, the use of SSU rRNA as the ideal gene marker was put under the spotlight. Studies showed that HGT also occurred in SSU rRNA and through the use of other, more precise gene markers, proved that many species were placed incorrectly on the SSU rRNA formed 'tree of life'. For example in a study done by Hirt et al (1999), used the largest sub-unit of RNA polymerase II to form a phylogenetic tree which showed microsporidia is closely related to fungi rather than an early divergent eukaryote which was previously thought from phylogenetic trees based on SSU rRNA.

Detecting HGT

Methods commonly used to detect horizontally transferred genes fall into 2 categories; parametric methods which are based on the detection of genes with atypical composition in comparison to the whole genome and phylogenetic methods which detect atypical distribution of a gene across organisms along a phylogenetic tree. Both methods of HGT identification have their advantages and disadvantages. Parametric methods work by using molecular techniques to analyse dinucleotide frequency (e.g GC content), codon usage biases or oligonucleotide usage to detect abnormal sequence composition within a genome. HGT detection using codon usage biases is based on the idea that each genome has a specific codon preference for an amino acid. So if a gene shows preference to a different triplet codon encoding the same amino acid compared to other genes in the same genome, then this suggests that it has come from a different species and the gene has been horizontally transferred into the genome. Similarly, oligonucleotide usage analysis for HGT detection is based on the fact that an oligonucleotide sequence from a gene is similar to other oligonucleotides in the same genome. Molecular techniques allow us to detect any significant differences in an oligonucleotide sequence compared to others in the genome which suggests that the gene has been transferred from another species. Mathematical models such as the Markov model and the Bayesian model can also be used to detect atypical genes (Dalevi et al, 2006). This form of detection is very rapid and only requires the genome sequence of only one species which is not the case in phylogenetic methods. This is because parametric methods compare genes within a single genome to detect any dissimilarities or presence of an atypical nucleotide sequence which would suggest HGT has occurred, while phylogenetic methods compare genes from different genomes. This also avoids systematic errors such as incorrect sequence data, which may occur in phylogenetic reconstruction. However this method is restricted and has been questioned as atypical patterns in a genome may be caused by events other than HGT such as base mutation. Also through time, this method cannot detect HGT which occurred many years ago and can only be relied for detection of recent HGT because of gene amelioration. This is when a gene which has been horizontally transferred many years ago, slowly through time, acquires molecular characteristics of the host gene therefore making it very difficult for parametric methods to detect any difference between the transferred gene and host genes. Another weakness in this method is the difficulty in determining the threshold which is used to distinguish between normal genes and atypical genes which can lead to false positive and false negative results (Eisen, 2000).

Phylogenetic methods compare phylogenetic trees formed using various genes and look for any significant incongruence. If all systematic errors and HGT were to be absent these phylogenetic trees would be congruent. This form of HGT detection identifies genes which are present (or highly resemble) in distantly related species while absent in closely related species. There are various phylogenetic methods which are commonly used; one is the likelihood-based tests of topologies such as the approximately unbiased test (AU) and Kishino-Hasegawa (KH) test which compares two phylogenetic trees to see if there is a difference between the two and if so, that it is not due to sampling errors (Goldman et al, 2000). Tree distance methods such as the Robinso and Foulds metric (RF) calculates the differences in the branching of the two trees and identifies bifurcations present in one tree and absent in the other (Gogarten and Popstova, 2007). The advantage of using phylogenetic methods for HGT detection lies in their ability to detect early ancestral gene transfers. It is also regarded as a more reliable method of HGT identification. However it is very time consuming to form phylogenetic trees and also analysing and comparing every gene to detect phylogenetic incongruence. The method can only be efficiently used if a reference phylogenetic tree (such as the SSU rRNA tree) is available which individual gene trees can be compared with. Without it the congruency of a gene tree cannot be determined while if it is present but it is inaccurate i.e. HGT or systematic biases are present then the results from the comparison and analysis will be unreliable and may lead to incorrect estimations of HGT levels. Also, phylogenetic incongruence may have been as a result of errors caused whilst reconstructing phylogenetic tree. These systematic errors include incorrect sequence data, incorrect methodology, misidentification of paralogs and orthologs, and long-branch attraction which is caused when 2 rapidly evolving lineages are thought to be closely related when they actually have separate evolutionary backgrounds. This is one of the reasons why phylogenetic methods do not work as well in detecting HGT between closely related species (Eisen, 2000; McInerney et al, 2008). Another reason might be that when there is a high rate of HGT between 2 different species during the reconstruction of the phylogenetic tree, they may be considered to be the same and are couple together, therefore no HGT is detected.

Frequency of HGT

Critics refused to acknowledge the impact of HGT in phylogenetics and retained that a universal 'tree of life' can be formed. Kurland (2000, 2003) wrote critical reviews in which he refuted claims of a "global rampant HGT" and disregarded the level of influence it has in organismal evolution as it does not occur frequently enough and its influence on overall specific characteristics and phylogeny of an organism is not substantial (Kurland, 2000). He also criticises the methodology used to detect HGT and questions their reliability, claiming that other reasons for the phylogenetic incongruence in HGT-supporting studies were not explored. In fact a study to test the reliability and consistency of methods detecting HGT was done and showed that HGT was detected using some methods while it wasn't using other types of methods, suggesting that some current methods used to identify HGT are not reliable and are not the best way to prove the presence and frequency of HGT (Ragan, 2001). Critics maintain the idea that vertical inheritance is the main mode of evolution and the universal genome phylogeny is very similar to the SSU rRNA phylogenetic tree formed by Woese. They believe that though HGT does occur, the transferred gene does not remain in the genome for a very long time and through evolution it is eventually deleted from the genome. This is because the transferred gene may not survive the conditions in host cell (e.g. toxicity) or if it is inactive and does not perform any useful function it may be deleted through random mutation. Kurland (2003) suggests that during early cellular evolution, a phase known as progenate, sequence adaptations occurs at a fast rate and this enables the transferred gene to become fixated to the host cell and become favourable for the host gene. However when the cell leaves this progenate phase, sequence adaptation becomes much slower and HGT is much less likely to occur as molecular and cellular barriers are established. This suggests that in primitive species HGT had a large role to play but vertical inheritance is dominant in modern species and HGT rarely occurs. On the other hand, some studies have shown that in fact frequency of HGT is remarkably high. Dagan and her colleagues (2008) compared more than half a million genes from 181 prokaryotic genes and found that on average approximately 81% of genes in each genome had at some point been involved in HGT. Also after the genome sequencing of various strains of E.coli, Welch et al (2002) compared the genome sequence of 3 strains and found that only 39.2% of their genes were shared between the 3 strains of E.coli. For this finding to concur with vertical inheritance the last common ancestor of the 3 strains would need to have all the genes in each strain. This would mean that it would have to have an unrealistically large genome to accommodate the genes from all 3 strains. A more logical explanation would be that approximately 60% of each genome which are different from each other was obtained separately through HGT.

Recent studies however, have shown that the frequency of HGT varies between different types of genes. Two sets of genes seem to be present which differ in their gene transfer ability. Studies suggest genes involved in information-processing actions such as transcription and translation, are less likely to be laterally transferred to another organism. These informational genes are commonly called 'core genes'. The other set of genes which are mainly involved in 'housekeeping' activities such as metabolic processes, are known as operational genes and are much more prone to lateral transfer. Among the many theories as to why there is a distinct difference in rate of HGT between the two groups of genes, Lake et al (1999) attempted to explain this using what they called 'the complexity theory'. They claim that informational genes are involved in much larger and more complex systems with many genes interacting together while operational genes are part of much simpler systems. For this reason informational genes are much more restricted and therefore less likely to undergo later gene transfer than operational genes while 'core genes' cannot or very rarely undergo HGT and for this reason do form a phylogenetic 'tree of life'. In a study by Soreck et al (2007), the suggestion that some genes are somehow restricted from HGT was found to be untrue. They searched for genes which can and cannot be transferred into E.coli by attempting to move 246,045 genes from 79 prokaryotic genomes into E.coli. Their results showed that due to several possible reasons such as toxicity, there were a select few genes which could not be transferred into E.coli from some genomes, however no single gene was non-transferrable from all genomes examined and they claimed that there is no clear barrier which prevents certain genes from being laterally transferred. Although it was a lab based experiment and this meant that the results cannot be relied on unequivocally (although all genes were transferred in lab it does not necessarily mean that it can occur in nature) however it is clear that further research is required in this field as dividing genes into two types based on function is not necessarily a completely reliable tool for HGT prediction. In fact a recent study by Hao and Golding (2008) showed that though there is a difference in lateral transfer rate between informational and operational genes, the difference between the two is not substantial and in fact there is greater variation in HGT within information and operational genes than between the two. Boto (2010) supports Hao and Goldings findings by suggesting that the rate and success of gene transfer depends on the complexity with which the protein encoded by the gene is involved in. If the protein interacts with several other components then gene transfer is much more unlikely and if the encoded product is involved with a much simpler system then the rate of HGT is higher with greater level of success. On the other hand, in a study by Wang et al (2005) the level of HGT in 'core' genes and found that it was as low as 2% and much less likely of an occurrence then in operational genes. They claimed that while HGT does occur, 'core' genes can be used to form the backbone of the 'tree of life'.

Taking these studies into account, categorising genes into informational and operational genes is not a completely reliable and absolute tool for transferred gene prediction.

Studies have also shown that the frequency of HGT occurrence changes depending on the distance of transfer (Barton, 2007). It seems that as the phylogenetic distance between species increases the less likely HGT will occur between them as there are more barriers for the gene to pass compared to a gene transferred between closely related species. These barriers include restriction enzymes degrading the transferred gene as they recognise it as being foreign DNA and replication machinery not being compatible. Wagner and De la Chaux (2008) compared 2,091 sequences from 438 prokaryotic genomes using phylogenetic methods and found only 30 cases of HGT between distantly related species with only seven of those being recent transfers. Further studies have shown that HGT does indeed occur between distantly related species and even between bacteria and archaea. Kanhere and Vingron (2009) studied 171 horizontally transferred gene and found that 118 of them were between archaea and bacteria. Interestingly the gene transferred between bacteria and archaea were mainly operational genes while genes transferred between bacteria were mainly informational genes. The higher rate of operational gene transfer between bacteria and archaea can be explained using the 'complexity theory' however the preference for informational genes transferring between bacteria has not been seen in other studies. Based on their results they suggest that the functional property of a gene that is transferred depends on the phylogenetic distance between the two species. Boto (2010) believes that though HGT may occur more in closely related species, it also occurs in distantly related species and much more than some studies suggest. This is because detection of HGT is much easier between closely related species and gene amelioration along other reasons makes it harder to detect HGT occurred between distantly related species.

HGT in eukaryotes:

It is now becoming clearer that HGT does not occur exclusively in prokaryotes but also in eukaryotic species which raises further questions about the possibility forming a universal 'tree of life'. Until now this has been overshadowed by the prominence of HGT in prokaryotes, endosymbiosis in eukaryotes and also the lack of genome sequences for large number of species. The origin of eukaryotes is not clear with one hypothesis suggesting that it was formed by fusion of a bacterium and an archaeon. This is because eukaryote informational genes are more similar to archaeal genes while their operational genes have a greater similarity with bacterial genes (Lake and Rivera, 2004). The most significant and recognisable case of HGT in eukaryotes is the transfer of many genes from mitochondria and plastids, into the hosts nuclear genome. This is alternatively known as endosymbiotic gene transfer as mitochondria and plastids were formed from endosymbionts, a-proteobacteria and cyanobacteria respectively which explains why mitochondrial and plastid genomes are much smaller than their ancestral genomes (Keeling and Palmer, 2008). There have been HGT cases identified between prokaryotes and- eukaryotes and also from eukaryotes to-eukaryotes. However it is thought to be much less frequent then in prokaryotes and also less influential in eukaryotic evolution. This is because there are many more barriers in eukaryotes for a gene to pass before it can fully integrate into another genome. Multi-cellular organisms have much more complex systems and many transferred genes do not remain for a long period of time. This is why most of the HGT identified has mainly been in unicellular eukaryotes (Lake and Rivera, 2004).

From tree to web?

Since the frequency and influence of HGT came to light, scientists have begun to question the presence of a universal 'tree of life' that relies on vertical as previously acknowledged 'tree of life' and vertical inheritance as the sole route of gene transfer no longer fits with the ever increasing evidence that HGT occurs and has a major influence on prokaryotic evolution. Scientist have now began reconstructing web like structures where the majority of genes are still transferred through vertical inheritance however there are horizontal links between these vertical branches connecting species from different lineages which represents genes that have a different evolutionary lineage to the rest of the genome and has undergone HGT. Contrary to this thought some scientists still believed that it was possible to form a universal 'tree of life' which is simply more refined, using 'core' genes which have not undergone HGT. Ciccarelli et al (2006) attempted to do this by comparing genome sequences of 191 organisms from all 3 domains of life (eukaryotes, bacteria and archaea). They found 36 concatenated genes which all of the organisms possessed and then reduced it to 31 genes which had not undergone HGT. These genes were primarily informational 'core' genes and were used to form a phylogenetic 'tree of life' [24]. However this study was criticised as producing a 'tree of life' which is not representative of the whole genome and Dagan and Martin (2006) labelled it "the tree of one percent of life" [25]. They mention that on average a prokaryote contains 3,000 genes and using only 31 genes to form a phylogenetic tree representing all living systems means that only around 1% of the whole genome in prokaryotes and even much less in multicellular eukaryotes is taken into consideration. They claim that if, after eliminating all the genes which are not present in all organisms and those that have undergone HGT or endosymbiotic gene transfer leaves only 1% of genes to form a bifurcating phylogenetic tree, than this 'tree-like' thinking must be reconsidered as it simply does not fit with the overall phylogenetic picture. The use of 'core' genes also can lead to mis-representation of species on the phylogenetic tree. The biggest case of this is the mislead beliefs about the relationship between archaea and eukaryotes and the suggestion that the two domains have more similarities than eukaryotes have with bacteria. This idea has wrongly become accepted because of the use of informational genes such as rRNA and other 'core' genes to represent the whole genome when forming phylogenetic trees. In fact Esser et al. (2004) compared 6,214 genes with 177,117 genes from 45 bacteria and 15 archaea. There results showed that 75% of eukaryotic genes have greater similarities with bacterial genes then they have with archaeal genes [26]. Some scientists have put this into consideration and attempted to show the simultaneous relationship between eukaryotes- archaea and eukaryotes-bacteria [22]. Instead of fitting their data into a tree-like structure they depict the evolution of 'life' using a ring which better represents the relationship between the 3 domains of life and the presence of HGT. As genome sequencing continues, the presence of HGT becomes clearer and more prominent in evolution.

It is clear now that prokaryotic evolution can no longer be considered strictly 'tree-like' and more scientists have now begun reconstructing network like structures instead of bifurcating trees to portray the evolution of life which both vertical inheritance and horizontal gene transfer are involved in.


  1. Barton, H.N., Briggs, D.E.G., eisen, J.A., Goldstein, D.B., Patel, N.H. (2007), EVOLUTION, New York, Cold Spring Harbor Laboratory Press, pp.120-123
  2. Archibald.J.D. (2009), 'Edward Hitchcock's Pre-Darwinian (1840) 'Tree of Life'.', Journal of the History of Biology, 42:pp.561-592
  3. Bapteste, E., O'Malley, M., Beiko, R., Ereshefsky, M., Gogarten, J.P., Hall, L.F., Lapointe.F.J., Dupre.J, Dagan.T, Boucher.Y.,Martin.W. (2009), 'Prokaryotic evolution and the tree of life are two different things', Biology Direct, 4(1):34
  4. Woese, C.R and Fox, G.E. (1977), 'Phylogenetic structure of the prokaryotic domain: the primary kingdoms.', Proc Natl Acad Sci USA 74 (11), pp.5088-90
  5. Akiba, T., Koyama, K., Ishiki, Y., Kimura, S. Fukushima, T., (1960), 'On the mechanism of the development of multiple-drug-resistant clones of Shigella.' Japanese journal of microbiology , 4, pp.219-227
  6. Lawrence, J.G. and Ochman, H. (1998), 'Molecular archaeology of the Escherichia coli genome.', Proc.Natl. Acad. Sci ESA , 95 pp.9413-9417
  7. Hirt, R.P. et al. (1999), 'Microsporidia are related to Fungi: Evidence from the largest subunit of RNA polymerase II and other proteins', Proc. Natl. Acad. Sci. USA, 96, pp.580-585.
  8. Dalevi, D., Dubhashi, D. And Hermansson, M., (2006), 'Bayesian classifiers for detecting HGT using fixed and variable order markov models of genomic signatures.', Bioinformatics, 22(5):pp.517-522
  9. Eisen, J.A., (2000), 'Horizontal gene transfer among microbial genomes: new insights from complete genome analysis', Curr Opin Genet Dev, 10(6):pp.606-611
  10. Goldman, N., et al., (2000), 'Likelihood-Based Tests of Topologies in Phylogenetics', 49(4): pp.652-670
  11. Gogarten, J.P. and Popstova, M.S., (2007), 'The power of phylogenetic approaches to detect horizontally transferred genes.', BMC Evol Biol, 7(1): pp.45+
  12. McInerney, J.O., Cotton, J.A. and Pisani, D, (2008), 'The prokaryotic tree of life: past, present... and future?', Trends in Ecology and Evolution, 23:pp.276-281
  13. Kurland, C.G. , (2000), 'Something for everyone- Horizontal gene transfer in evolution', EMBO Rep, 1: pp. 92-95
  14. Ragan, M.A., (2001), 'On surrogate methods for detecting lateral gene transfer', FEMS Microbiol Lett, 201(2): pp.187-91
  15. Kurland, C.G., Canback, B. and Berg, O.G., (2003), 'Horizontal gene transfer: A critical view', Proc.Natl.Acad.Sci. USA, 100: pp.9658-9662
  16. Dagan, T., Artzy-Randrup, Y. & Martin, W., (2008), 'Modular networks and cumulative impact of lateral transfer in prokaryote genome evolution.', Proc. Natl Acad. Sci., 105: pp.10 039-10 044.
  17. Welch, R.A., etal., (2002). `Extensive mosaic structure revealed by the complete genome sequence of uropathogenic Escherichia coli.'. Proc Natl Acad Sci USA, 99(26):pp.17020-17024
  18. Lake, J.A., Rivera, M.C., Jain, R., (1999), 'Horizontal gene transfer among genomes: The complexity hypothesis', Proc Natl Acad Sci USA, 96(7): pp.3801-3806
  19. Sorek R., et al., (2007), 'Genome-wide experimental determination of barriers to horizontal gene transfer.', Science, 318: pp.1449-1452.
  20. Hao, W., Golding, G. B., (2008), 'Uncovering rate variation of lateral gene transfer during bacterial genome evolution.' BMC Genomics, 9:235.
  21. Ge, F., Wang, L-S., Kim, J., (2005), 'The Cobweb of Life Revealed by Genome-Scale Estimates of Horizontal Gene Transfer.', PLoS Biol 3(10): e316. doi:10.1371/journal.pbio.0030316
  22. Lake, J.A., Rivera, M.C., (2004), 'The ring of life provides evidence for a genome fusion origin of eukaryotes.', Nature, 431: pp.152-155
  23. Keeling, P.J., Palmer, J.D., (2008), 'Horizontal gene transfer in eukaryotic evolution', Nature Reviews Genetics, 9: pp.605-618
  24. FD Ciccarelli, F.D., et al., (2006), 'Toward automatic reconstruction of a highly resolved tree of life.', Science, 311: pp.1283-87
  25. Dagan, T., Martin, W., (2006), 'The tree of one percent', Genome Biology, 7(10): pp.118
  26. Esser, C., et al., (2004), 'A genome phylogeny for mitochondria among alpha-proteobacteria and a predominantly eubacterial ancestry of yeast nuclear genes.', Mol Biol Evol, 21: pp.1643-1660.
  27. Woese, C.R., (1987), 'Bacterial evolution', Microbiol. Rev., 51(2): pp.221-271
  28. Kanhere A., Vingron M. 2009 Horizontal gene transfer in prokaryotes shows differential preferences for metabolic and translational genes. BMC Evol. Biol. 9, 9
  29. Wagner A., De la Chaux N.2008 Distant horizontal gene transfer is rare for multiple families of prokaryotic insertion sequences. Mol. Genet. Genomics 280, 397-408
  30. Boto, L., Horizontal gene transfer in evolution: facts and challenges Proc. R. Soc. B 22 March 2010 vol. 277 no. 1683 819-827