Review Article Archaea: The First Domain of Diversified Life - Hindawi Publishing Corporation

27 Pages·2015·4.19 MB·English

Hindawi Publishing Corporation
Archaea
Volume 2014, Article ID 590214, 26 pages
http://dx.doi.org/10.1155/2014/590214
Review Article
Archaea: The First Domain of Diversified Life
GustavoCaetano-Anollés,1ArshanNasir,1KaiyueZhou,1DerekCaetano-Anollés,1
JayE.Mittenthal,1Feng-JieSun,2andKyungMoKim3
1EvolutionaryBioinformaticsLaboratory,DepartmentofCropSciences,InstituteforGenomicBiologyandIllinoisInformaticsInstitute,
UniversityofIllinoisatUrbana-Champaign,Urbana,IL61801,USA
2SchoolofScienceandTechnology,GeorgiaGwinnettCollege,Lawrenceville,GA30043,USA
3MicrobialResourceCenter,KoreaResearchInstituteofBioscienceandBiotechnology,Daejeon305-806,RepublicofKorea
CorrespondenceshouldbeaddressedtoGustavoCaetano-Anolle´s;[email protected]
Received30September2013;Revised15February2014;Accepted25March2014;Published2June2014
AcademicEditor:CelineBrochier-Armanet
Copyright©2014GustavoCaetano-Anolle´setal. This is an open access article distributed under the Creative Commons
AttributionLicense,whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkis
properlycited.
Thestudyoftheoriginofdiversifiedlifehasbeenplaguedbytechnicalandconceptualdifficulties,controversy,andapriorism.It
isnowpopularlyacceptedthattheuniversaltreeoflifeisrootedintheakaryotesandthatArchaeaandEukaryaaresistergroups
toeachother.However,evolutionarystudieshaveoverwhelminglyfocusedonnucleicacidandproteinsequences,whichpartially
fulfillonlytwoofthethreemainstepsofphylogeneticanalysis,formulationofrealisticevolutionarymodels,andoptimizationof
treereconstruction.Intheabsenceofcharacterpolarization,thatis,theabilitytoidentifyancestralandderivedcharacterstates,
anystatementabouttherootingofthetreeoflifeshouldbeconsideredsuspect.Hereweshowthatmacromolecularstructureand
anewphylogeneticframeworkofanalysisthatfocusesonthepartsofbiologicalsystemsinsteadofthewholeprovidebothdeep
andreliablephylogeneticsignalandenableustoputforthhypothesesoforigin.Wereviewoveradecadeofphylogenomicstudies,
whichmineinformationinagenomiccensusofmillionsofencodedproteinsandRNAs.Weshowhowtheuseofprocessmodels
ofmolecularaccumulationthatcomplywithWeston’sgeneralitycriterionsupportsaconsistentphylogenomicscenarioinwhich
theoriginofdiversifiedlifecanbetracedbacktotheearlyhistoryofArchaea.
1.Introduction or plural events in history that must be formalized by
“transformations”thatcomplywithanumberofevolutionary
“Imagineachildplayinginawoodlandstream,pokingastick axioms[3]andinterfacewithaframeworkofmaximization
intoaneddyintheflowingcurrent,therebydisruptingit.But of explanatory power [4]. The fundamental statement that
theeddyquicklyreforms.Thechilddispersesitagain.Againit organismaldiversityistheproductofevolutionissupported
reforms,andthefascinatinggamegoeson.Thereyouhaveit! byanensembleofthreenestedprimaryaxiomsofthehighest
Organismsareresilientpatternsinaturbulentflow—patterns level of universality [3]: (i) evolution occurs, including
inanenergyflow”—CarlWoese[1]. its principle that history of change entails spatiotemporal
Understandingtheoriginofdiversifiedlifeisachalleng- continuity (sensu Leibnitz), (ii) only one historical account
ing proposition. It involves the use of ideographic thinking of all living or extinct entities of life and their component
that is historical and retrodictive, as opposed to nomo- parts exists as a consequence of descent with modification,
thetic explorations that are universal and predictive [2]. and(iii)featuresofthoseentities(characters)arepreserved
Experimental science for the most part is nomothetic; the through generations via genealogical descent. History must
searchfortruthcomesfromuniversalstatementsthatcanbe comply with the “principle of continuity,” which crucially
conceptualizedasbeingofgeneralpredictiveutility.Nomo- supports evolutionary thinking. The axiomatic rationale of
thetic explorations are in general both philosophically and “naturanonfacitsaltum”highlightedbyLeibnitz,Linnaeus,
operationally less complex to pursue than any ideographic andNewtonmustbeconsideredageneralityofhownatural
exploration. In contrast, retrodictions speak about singular thingschangeanda“fruitfulprincipleofdiscovery.”Wenote
2 Archaea
that this axiomatic generality, which we have discussed in thisframeworktodatadrawnfromstructuralandfunctional
the context of origin of life research [5], encompasses rare genomics and argue that the origin of cellular life involved
punctuations (e.g., quantum leap changes such as genome gradual accretion of molecular interactions and the rise of
duplications and rearrangements or the rare evolutionary hierarchicalandmodularstructure.Wediscussourfindings,
appearanceofnewfoldstructures)embeddedinafabricof which provide strong support to the very early rise of
gradual change (e.g., changes induced by point mutations). primordialarchaeal lineages and the emergence of Archaea
Bothgradualchangesandpunctuationsareinterlinkedand as the first domain of diversified cellular life (superking-
arealwaysexpressedwithinspatiotemporalcontinuity(e.g., dom). The term “domain” of life stresses the cohesiveness
structural punctuations in the mappings of sequences into of the organism supergroup, very much like domains in
structures of RNA [6]). This interpretative framework can proteinsandnucleicacidsstressthemolecularcohesiveness
explain novelty and complexity with principles of scientific of their atomic makeup. Instead, the term “superkingdom”
inquiry that maximize the explanatory power of assertions (superregnum)makesexplicitthefactthatthereisanested
aboutretrodictions. hierarchy of groups of organisms, many of which share
Phylogenetic theories are embodied in evolutionary commonancestors(i.e.,theyaremonophyletic).Wepropose
“trees” and “models.” Trees (phylogenies) are multidimen- that the rise of emerging lineages was embedded in a
sional statements of relationship of the entities that are primordial“evolutionarygrade”(sensuHuxley[11]),agroup
studied (phylogenetic taxa). Models are evolutionary trans- of diversifying organisms (primordial archaeons) in active
formationsofthebiologicalattributesexaminedindata(phy- transitionthatwereinitiallyunifiedbythesameandarchaic
logeneticcharacters),whichdefinetherelationshipsoftaxain levelofphysiologicalcomplexity.Ourdiscussionwillattempt
trees.Thetripartiteinteractionbetweencharacters,models, toreconcilesomedivergentviewsoftheoriginofdiversified
and trees must occur in ways that enhance retrodictive lifeandwillprovideagenericscenariofor“turningpoints”of
power through test and corroboration [4]. In other words, originthatmayberecurrentinbiology.
it must follow the Popperian pillars of scientific inquiry
or suitable philosophical analogs. We note that retrodictive
statementsallowdrawinginferencesaboutthepastbyusing 2.ATripartiteWorldofOrganismalDiversity
informationthatisextant(i.e.,thatwecanaccesstoday)and
isnecessarilymodern.Thechallengeoftravellingbackintime Carl Woese and his colleagues of the Urbana School were
rests on not only making inferences about archaic biology responsibleforthegroundbreakingdiscoverythattheworld
with information drawn from modern biological systems of organisms was tripartite; that is, it encompassed not
but also interpreting the retrodictive statements without two but three major “domains” of cellular life (Archaea,
conceptualrestrictionsimposedbymodernity.Thishasbeen Bacteria, and Eukarya). Two of the three “aboriginal” lines
an important obstacle to historical understanding, starting of decent were initially conceptualized as “urkingdoms” of
withgradinghypothesesinspiredbyAristotle’sgreatchainof deep origin that were microbial and qualitatively differ-
being,thescalanaturae. ent from eukaryotic organisms [12]. They corresponded to
It was Willi Hennig in the fifties who first formal- ArchaeaandBacteria.ThediscoveryofArchaeachallenged
ized retrodiction in quantitative terms [7]. Since then, his the established akaryote/eukaryote divide (we use the term
“phylogenetic systematics” has benefitted from numerous “akaryote” to describe a cell without a bona fide nucleus.
conceptualandbioinformaticsdevelopments,whicharenow Thistermcomplementstheword“eukaryote”(“eu,”good,and
responsible for modern phylogenetic analysis of systems “karyon,” kernel), which is ahistorical. The new term takes
of any kind: from molecules and organisms to language away the time component of the widely used “prokaryote”
and culture, from engineering applications to astrophysics. (“pro” before) definition, which may be incorrect for many
Astrocladistics, for example, focuses on the evolution and organismsofthemicrobialdomains)thatsupported“ladder”
diversificationofgalaxiescausedbytransformingeventssuch scenarios of gradual evolution from simplistic microbes to
asaccretion,interaction,andmergers(e.g.,[8]).Whilemajor “higher” organisms, which were tenaciously defended by
views have emerged in the “discovery operations” (sensu molecularbiologistsandmicrobiologistsofthetime.Woese
[2]) of the phylogenetic systematics paradigm, including andFox[12]madeitclear:“Evolutionseemstoprogressina
maximum parsimony and the frequentist and uncertainty “quantized”fashion.Onelevelordomainoforganizationgives
views of maximum likelihood and Bayesian thinking, the riseultimatelytoahigher(morecomplex)one.What“prokary-
major technical and philosophical challenges persist [9]. ote”and“eukaryote”actuallyrepresentaretwosuchdomains.
More importantly, as we will explain below, technical and Thus, although it is useful to define phylogenetic patterns
philosophicalaspectsoftheideographicframeworkinsome within each domain, it is not meaningful to construct phylo-
cases have been turned into landscapes of authoritarianism geneticclassificationsbetweendomains:Prokaryotickingdoms
and apriorism [3]. This insidious trend is pervasive in the are not comparable to eukaryotic ones.” The discovery was
“rootingofthetreeoflife”fieldofinquiry[10]thatunderlies revolutionary,especiallybecausescalanaturaedeeplyseated
theoriginsofbiochemistryandbiodiversityweherediscuss. the roots of the akaryote/eukaryote divide and microbes
Inthisopinionpaperweaddressthechallengesoffinding wereconsideredprimitiveformsthatdidnotwarrantequal
an origin to biodiversity and propose a new framework for standing when compared to the complex organization of
deepphylogeneticanalysisthatfocusesonthepartsofbiolog- Eukarya(see[13]forahistoricalaccount).Thesignificanceof
icalsystemsinsteadofthewhole.Wereviewtheapplicationof thetripartiteworldwasquicklyrealizedandvividlyresisted
Archaea 3
by the establishment. Its resistance is still embodied today ensureditcarriedanancientandoverridingmemoryofthe
innewproposalsoforigins,suchasthearchaeon-bacterium cellularsystemsthatwerestudied.Thiswasmadeevidentin
fusion hypothesis used to explain the rise of Eukarya (see the first ToL reconstructions. In contrast, many of the pro-
below). It is noteworthy that the root of the universal tree teins encoded by paralogous gene couples (e.g., translation
of cellular organisms, the “tree of life” (ToL), was initially factors)likelycarriedconvolutedhistoriesofproteindomain
not the driving issue. This changed when the sequences cooption or important phylogenetic biases induced, for
of proteins that had diverged by gene duplication prior to example,bymutationalsaturationintheirproteinsequences.
a putative universal common ancestor were analyzed with
The ribosome is indeed the central feature of cellular life:
phylogeneticmethodsandthecomparisonsusedtorootthe
the signature of “ribocells.” However, its embedded phylo-
ToL [14, 15]. Paralogous gene couples included elongation
genetic signatures are also convoluted. The constitution of
factors (e.g., EF-Tu and EFG), ATPases (𝛼 and 𝛽 subunits),
theribosomeisheterogeneous.Theribosomerepresentsan
signal recognition particle proteins, and carbamoyl phos- ensemble of 3-4 ribosomal RNA (rRNA) and ∼70 protein
phate synthetases, all believed to be very ancient (reviewed
(r-protein)molecules,dependingonthespeciesconsidered,
in [16]). In many cases, bacterial sequences were the first
andembodiesmultipleinteractionswiththecellularmilieu
to branch(appearedatthebase)in thereconstructedtrees,
needed for function (e.g., assembly and disassembly; inter-
forcingarchaealandeukaryalsequences to be sister groups
actions with the membrane of the endoplasmic reticulum).
to each other. This “canonical” rooting scheme of the ToL
A group of 34 r-proteins is present across cellular life [28].
(Figure1(a))wasacceptedasfactandwasquicklyendorsed
Ribosomal history has been shown to involve complex pat-
by the supporters of the Urbana School [17]. In fact, the
terns of protein-RNA coevolution within the evolutionarily
acceptanceofthe“canonical”rootinginBacteriabecameso
deepthatithasnowpromptedthesearchfortheoriginsof conservedcore[29].Thesepatternsareexpresseddistinctly
Eukaryainthemolecularandphysiologicalconstitutionsof in its major constituents. While both of its major subunits
theputativearchaealsistergroup[18].Forexample,Embley evolved in parallel, a primordial core that embodied both
andcoworkersgeneratedsequence-basedphylogeniesusing processiveandcatalyticfunctionswasestablishedquiteearly
conserved proteins and advanced algorithms to show that in evolution. This primordial core was later accessorized
Eukarya emerged from within Archaea [19–21] (refer to withstructuralelements(e.g.,accretionofnumerousrRNA
[22] for critical analysis). Importantly, these analyses suffer helical segments and stabilizing A-minor interactions) and
from technical and logical problems that are inherent in r-proteins (e.g., the L7/L12 protein complex that stimulates
sequence-based tree reconstructions. For example, proteins the GTPase activity of EFG) that enhanced its functional
such as elongation factors, tRNA synthetases, and other properties.Thisincludedexpansionelementsinthestructure
universal proteins used in their analyses are prone to high of rRNA that were specific not only to subunits but also to
substitution rates [23]. Mathematically, it leads to loss of individualdomainsoflife.Figure2,forexample,showsaphy-
information regarding the root of the ToL as shown by logenomicmodelofribosomalmolecularaccretionderived
SoberandSteel[24](referto[25]formorediscussion).On
from the survey of protein domain structures in genomes
theotherhand,paralogousrootingssometimescontradicted
andsubstructuresinrRNAmolecules.Theaccretionprocess
each other and were soon and rightly considered weak and
of component parts of the universal core appears to have
unreliable[23,26,27].Thevalidityofparalogy-basedrooting
been a painstakingly slow process that unfolded during a
methodology has proven to be severely compromised by period of ∼2 billion years and overlapped with the first
a number of problems and artifacts of sequence analysis
episodes of organismal diversification [30]. Despite of this
(e.g., long branch attraction, mutational saturation, taxon
complexity,thefocusofbiodiversitystudieswasfordecades
sampling,horizontalgenetransfer(HGT),hiddenparalogy,
the small subunit rRNA molecule [31]. This focus has not
andhistoricalsegmentalgeneheterogeneity).Consequently,
changed much in recent years. Consequently, the history
there is no proper outgroup that can be used to root a ToL
oforganismsandpopulationsiscurrentlyrecountedby the
that is built from molecular sequences, and currently, there
informationseated in thesmallsubunitrRNA molecule. In
arenopropermodelsofsequenceevolutionthatcanprovide
other words, the historical narrative generally comes from
areliable“evolutionaryarrow.”Becauseofthisfact,archaeal
andeukaryalrootingsshouldbeconsideredequallyprobable only ∼1% of ribosomal molecular constitution. This impor-
to the canonical bacterial rooting (Figure 1(c)). This is an tantandunacknowledgedbiaswasalreadymadeexplicitin
importantrealizationthatneedstobeexplicitlyhighlighted, early phylogenetic studies. For example, de Rijk et al. [32]
especiallybecauseitaffectsevolutionaryinterpretationsand demonstratedthatphylogeniesreconstructedfromthesmall
thelikelihoodofscenariosoforiginsofdiversifiedlife. and large subunit rRNA molecules were different and that
thereconstructionsfromthelargesubunitweremorerobust
and better suited to establish wide-range relationships. The
3.MiningAncientPhylogeneticSignal structureofthesmallandlargesubunitswasalsoshownto
inUniversalMolecules carrydistinctphylogeneticsignatures[33].However,onlyfew
evolutionarystudieshavecombinedsmallandlargesubunit
Woese’s crucial insight was the explicit selection of the rRNAforhistoryreconstruction.Remarkably,inallofthese
ribosome for evolutionary studies. The universality of the cases phylogenetic signal was significantly improved (e.g.,
ribosomalensembleanditscentralroleinproteinsynthesis [34]).
4 Archaea
Crown
E A B
Archaea
E A B
A
Urancestor
U a a
Ancestor Stem
?
a
U U
E B
Origin of
Eukarya Bacteria Time lineages
(a) (b)
E A B B A E E B A
a a a
Time U U U
Canonical rooting Eukaryal rooting Archaeal rooting
(c)
Figure1:Rootingthetreeoflife(ToL):anexercisefor“tree-thinkers.”(a)Anode-basedtreerepresentationoftheToLfocusesontaxa
(sampledorinferredverticesillustratedwithopencircles)andmodelsedgesasancestryrelationships.Theunrootedtreedescribestaxaas
conglomeratesofextantspeciesforeachdomainoflife,Archaea(A),Bacteria(B),andEukarya(E)(abbreviationsareusedthroughoutthe
paper).Addingauniversalancestorvertex(rootingtheToL)impliesaddingareconstructedentity(urancestor)thatrootsthetreebutdoes
notmodelancestryrelationships.Thevertexiseithera“mostrecentuniversalancestor”ifonedefinesitfromaningroupperspectiveora“last
universalcommonancestor”(LUCA)ifthedefinitionrelatestoanoutgroupperspective.Rootingthetreeenablesdefining“totallineages,”
whicharelistsofancestorsspanningfromtheancestraltaxontothedomaintaxa.(b)Astem-basedviewfocusesonedges(branches),which
aresampled,andinferredancestraltaxaareviewedaslineagesundertheparadigmofdescentwithmodification.Verticescorrespondto
speciationevents.Terminaledgesrepresentconglomeratesoflineagesleadingtodomainsoflife(theterminalnodesofnode-basedtrees)
andtheancestralstemrepresentsthelineageoftheurancestor(U)(doublearrowheadline).Totallineagesaresimplyachainofedgesthat
goesbackintimeandendsintheancestralstem.Bothnode-basedandstem-basedtreerepresentationsaremathematicallyisomorphicbut
theyarenotequal[154].Theychangetheconceptofmonophyleticandparaphyleticrelationships.Anode-basedcladestartswithalineage
attheinstantofthesplittingevent,incorporatingtheancestorintothemakeupoftheclade.Incontrast,astem-basedcladeoriginateswith
aplantedbranchonthetree,wherethebranchrepresentsalineagebetweentwolineagesplittingevents.Plantinganancestralstemdefines
anoriginoflineagesandthefirstspeciationeventintherecordoflife.Thisdelimitsacrowncladeoftwodomainlineagesinthestem-based
tree(labeledwithblacklines)thatincludestheancestorofthesistergroups,a,andastemdomaingroupatitsbase.(c)Thethreepossible
rootingsoftheToLdepictedwithstem-basedtreerepresentations.Terminaledgesarelabeledwiththinlinesandtheseconglomeratelineages
canincludestemandcrowngroups.
4.BuildingaTowerofBabelfrom Theresults that were obtained consistently supported the
aComparativeGenomicPatchwork tripartiteworld,backing-uptheclaimsoftheUrbanaSchool.
ofSequenceHomologies However,cluesaboutancestorsandlineagesleadingtoextant
taxawerestillmissing.
Molecular evolutionists were cognizant of the limitation With few exceptions that focused on the structure of
of looking at the history of only few component parts, RNA and protein molecules (see below), analyses based on
which by definition could be divergent. When genomic genomic sequences, gene content, gene order, and other
sequences became widely available, pioneers jumped onto genomiccharacteristicswereunabletoproducerootedtrees
thebandwagonofevolutionarygenomicsandthepossibility withoutthehelpofoutgroups;additionaladhochypotheses
of gaining systemic knowledge from entire repertoires of that are external to the group of organisms being studied
genesandmolecules(e.g.,[35–39]).Thegenomicrevolution, andgenerallycarrystrongassumptions.An“arrowoftime”
forexample,quicklymaterializedingenecontenttreesthat was not included in the models of genomic evolution that
reconstructed the evolution of genomes directly from their wereused.Aswithsequence,ToLsthatweregeneratedwere
evolutionaryunits,thegenes(e.g.,[37,39,40]),orthedomain unrooted(Figure1(a))andgenerallyrootedaposteriorieither
constituentsofthetranslatedproteins[35,41].Thesequences claiming that the canonical root was correct or making
of multiple genes were also combined or concatenated assumptions about character change that may be strictly
in attempts to extract deep phylogenetic signal [42–44]. incorrect. In a recent example, distance-based approaches
Archaea 5
Primordial Mature Processive Mature Present day
ribosome PTC ribosome ribosome ribosome
Ribosomal history
Origin of
proteins Ribosomal core
1 2 3
Urancestor Diversified life
Molecular age (nd) The present
3.8 3.0 2.0 1.0 0
History of life (time, billions of years)
Figure2:Theevolutionaryhistoryofribosomewastracedontothethree-dimensionalstructureofitsrRNAandr-proteincomponents.Ages
ofcomponentswerecoloredwithhuesfromred(ancient)toblue(recent).Phylogenomichistoryshowsthatprimordialmetabolicenzymes
precededRNA-proteininteractionsandtheribosome.Thetimelineoflifederivedfromauniversalphylogenetictreeofproteindomain
structureatfoldsuperfamilylevelofcomplexityisshownwithtimeflowingfromlefttorightandexpressedinbillionsofyearsaccording
toamolecularclockoffoldstructures.Theribosomaltimelinehighlightsmajorhistoricaleventsofthemolecularensemble.Historyof
theconservedribosomalcore[29]overlapswiththatofdiversifiedlife[48],suggestingthatepisodesofcooptionandlateraltransferhave
pervadedearlyribosomalevolution.
wereusedtobuilduniversalnetworktreesfromgenefamilies problem. For example, analysis of genomic insertions and
definedbyreciprocalbestBLASThits[45].Thesenetworks deletions(indels)thatarerareinparalogousgenesetsrooted
showed a midpoint rooting of the ToL between Bacteria the ToL between the group of Actinobacteria and Gram-
andArchaea.However,thisrootinginvolvesacomplexopti- negative bacteria and the group of Firmicutes and Archaea
mizationofpathlengthsinthesplitnetworksandcritically [51]. Unfortunately, little is known about the dynamics of
assumes that lineages evolved at roughly similar rates. This indelgeneration,itsbiologicalroleingeneduplication,and
diminishestheconfidenceofthemidpointrooting,especially itsinfluenceonthestructureofparalogousproteinpairs(e.g.,
whenconsideringtheuncertaintiesofdistancesinferredfrom [52]). If the dynamics are close to that of sequences, indels
BLAST analyses and the fact that domains in genes hold should be considered subject to all the same limitations of
different histories and rates of change. Another promising sequenceanalyses,includinglongbranchattractionartifacts,
but ill-conceptualized case is the rooting of distance-based character independence, and inapplicable characters. Given
trees inferred by studying the frequency of l-mer sets of a tree topology, parsimony was also used to optimize the
amino acids in proteins at the proteome level [46]. The evolutionarytransformationofgenomicfeatures.Forexam-
compositional data generated a ToL that was rooted in ple, when considering the occurrence of protein domains
Eukarya when randomized proteome sequences were used thatareabundant,variantsofdomainstructuresaccumulate
asoutgroups.Theassumptionofrandomnessassociatedwith graduallyingenomes;operationallythedomainstructureper
therootofaToLishoweverunsupportedorprobablywrong, se cannot be lost once it had been gained. This enabled the
especiallybecauseproteinsequencespaceanditsmappingsto use of a variant of the unrooted Dollo algorithmic method
structurearefarfromrandom[47].Moreimportantly,alarge (e.g., [53]); the Dollo parsimony model [54] is based on
fractionofmodernproteinshadalreadyevolvedproteinfold the assumption that when a biological feature that is very
structureswhenthefirstdiversifiedlineagesarosefromthe complex is lost in evolution it cannot be regained through
universalcommonancestorofcellularlife[48].Theseevolved vertical descent. The method however could not be applied
repertoirescannotbeclaimedtoberandom. toakaryoticgenomes,asthesearesubjecttoextensivelateral
The inability of genomics to provide clear answers to genetransfer,anduntilrecently[55]wasnotextendedtoToL
rooting questions promoted (to some extent) parsimony reconstructions.
thinking and the exploration of evolutionary differences of Parsimony thinking was also invoked in “transition
significance that could act as “anchors” and could impart analysis,” a method that attempts to establish polarity of
“polarity”totreestatements[49].Aswewilldescribebelow, character changeby, forexample, examininghomologiesof
we applied parsimony thinking to the evolution of protein proteins in proteolytic complexes such as the proteasome,
andRNAstructures[41,50],andforoveradecadewehave membraneandcellenvelopebiochemicalmakeup,andbody
been generating rooted ToLs that portray the evolution of structures of flagella, sometimes aided by BLAST queries
proteomes with increasing explanatory power. This work [56, 57]. The elaboration however is restricted to the few
howeverhasbeenforthemostpartunacknowledged.Instead, molecular and cellular structures that are analyzed, out of
molecularevolutionistsfocusedalmostexclusivelyonmolec- thousands that populate the akaryotic and eukaryotic cells.
ular sequences in their search for solutions to the ToL The approach is local, has not yet made use of an objective
6 Archaea
analyticphylogeneticframework,anddoesnotweighgains,
losses,andtransferswithalgorithmicimplementations.Thus,
manystatementscanfallprayofincorrectoptimizationsor
processesofconvergenceofstructureandfunction,including
cooptions that are common in metabolic enzymes [58].
Transitionsalsofailtoconsiderthemolecularmakeupofthe
complexes examined (e.g., evolution of the photosynthetic
reaction centers), which often hold domains with hetero-
geneous histories in the different organismal groups. For
example, the 𝐹1/𝐹𝑜 ATP synthetase complex that powers
cellularprocesseshasalonghistoryofaccretionofdomains
that span almost 3.8 billion years of history, which is even
olderthanthatoftheribosome[59].Finally,establishingthe
validity of evolutionary transitions in polarization schemes
can be highly problematic; each transition that is studied
requires well-grounded assumptions [60], some of which
havenotbeenyetproperlyelaborated.
Theinabilitytosolvetherootingproblemandtheinsis-
tenceonextractingdeepphylogeneticsignalfrommolecular
sequences that are prone to mutational saturation raised
skepticism about the possibility of ever finding the root
of the ToL. Bapteste and Brochier [60] made explicit the
conceptual difficulties claiming scientists in the field had Figure 3: The “Confusion of Tongues,” an engraving by Gustave
adoptedAgrippa’slogicofdoubt.Misunderstandingsonhow Dore´(1832–1883),wasmodifiedbyD.Caetano-Anolle´stoportray
to conduct ideographic inquiry in evolutionary genomics theeventofdiversificationthathaltedtheconstructionoftheTower
of Babel. In linguistics, the biblical story inspired tree thinking.
had effectively blocked the reerection of a Tower of Babel
We take the metaphor as the fall of the urancestor of cellular life
thatwouldexplainthediversificationofgenomes(Figure3).
andthereplacementof“scalanaturae”bybranchingprocessesof
Instead,therewas“ConfusionofDoubts.”Unfortunately,the
complexity.
impasse was aggravated by aprioristic tendencies inherited
from systematic biology [3, 10], disagreements about the
evolutionary role of vertical and lateral inheritance and
theproblemofhomology[61],andcurrentlydisagreements
about the actual role of Darwinian evolution in speciation acrucialsourceofreticulation.Furthermore,therelationship
andtheToLproblem[13,62]. between differential gene loss and HGT has not been ade-
Duringthedecadeofevolutionarygenomicdiscovery,the quatelyformalized,especiallyforgenesthatareveryancient.
effectsofHGTonphylogeny[63]tookfrontcover.HGTwas Establishingpatternsoflossrequiresestablishingpolarityof
invoked as an overriding process. However, little attention changeandrootedtrees,whichaswehavediscussedearlier
waspaidtoalternativeexplanationssuchasdifferentialloss remains unattained for ToLs reconstructed from molecular
of gene variants, ancient or derived, and there was little sequences.
concern for other sources of reticulation. HGT is certainly
animportantevolutionaryprocessthatcomplicatesthe“tree” (ii)Interpretation.HGTgenerallymaterializesasamismatch
concept of phylogenetic analysis and must be carefully betweenhistoriesofgenesingenomesandhistoriesoforgan-
studied(e.g.,[64–66]).Insomebacterialtaxonomicgroups, isms.Sincegenomesarebyfunctionaldefinitioncollections
suchastheproteobacteria,HGTwasfoundtobepervasive ofgenes,reticulationsaffecthistoryofthegenesandnotof
andchallengedthedefinitionofspecies[67].Caseslikethese theorganisms,whichgivenevolutionaryassumptionsresult
prompted the radical suggestion that the ToL should be from hierarchical relationships (e.g., [73]). Thus and at first
abandonedandthata“weboflife”shouldbeusedtodescribe glance, reticulation processes (HGT, gene recombination,
the diversification of microbes and multicellular organisms gene duplications, and gene loss) do not obliterate vertical
[68]. However, the problem has two aspects that must be phylogeneticsignal.Theproblemhoweverremainscomplex.
separatelyconsidered. A ToL can be considered an ensemble of lineages nested
within each other [74]. Protein domain lineages will be
(i)Mechanics.ThewidespreadandimpactfulnatureofHGT nested in gene lineages; gene lineages will be nested in
mustbeaddressed.Doesitsexistencetrulycompromisethe lineages of gene families; gene families will be nested in
validity of phylogenetictree statements? While HGT seems organismal lineages; organisms will be nested in lineages
important for some bacterial lineages [69], its evolutionary ofhigherorganismalgroupings(populations,communities,
impacts in Archaea and Eukarya are not as extensive (e.g., ecosystems, and biomes), and so on. All lineage levels
[61, 70]) and its global role can be contested [71]. In turn, are defined by biological complexity and the hierarchical
the role of viruses and RNA agents in genetic exchange organizationoflifeandfollowafractalpatterndistribution,
continues to be understudied (e.g., [72]) and could be eachlevelcontributingverticalandlateralphylogeneticsignal
Archaea 7
tothewhole.However,sublineagesinonehierarchicallevel [80].ReticulationsandultimatelyHGTwereforcedtoexplain
may hold different histories compared to the histories of chimeric patterns. Fusion hypotheses diverted the central
higherandlowerlevels.Historicalmismatchesintroduce,for issue of the rooting of the ToL and prompted the separate
example,lineagesortingandreticulationproblemsthatrep- analysisofeukaryoticoriginsandakaryoticevolution.
resentacomplicationtophylogeneticanalysis.Forexample, Dagan and Martin [64] denounced that the ToL was
mismatchesbetweengeneandorganismalphylogeniesexist a “tree of one percent” because only a small fraction of
in the presence of differential sorting of genomic compo- sequences could be considered universal and could be
nents due to, for example, species extinctions or genomic minedfordeepphylogeneticsignal.Therestwouldaccount
rearrangements. These mismatches violate the fundamental for lateral processes that confounded vertical descent. The
cladisticassumptionthathistoryfollowsabranchingprocess, claimcamefundamentallyfromnetworksconstructedusing
but can be explained by homoplasy (convergent evolution), BLAST heuristic searches for short and strong matches
thehorizontal“trace”ofthehomology-homoplasyyin-yang in genomic sequences. Phylogenetic “forests” of akaryotic
of phylogenetic signal. The problem of homoplasy cannot genes later on boosted the claim [81]. These networks were
be solved by conceptually preferring one particular level built from 6,901 trees of genes using maximum likelihood
of the hierarchy [74]. A focus on the organismal level, for methods. Only 102 of these trees were derived from nearly
example, will not solve the problems of hybridization that universalgenesandcontributedverylittleverticalphyloge-
arebroughtbystrictorrelaxedsexualreproductionorviral- neticsignal.Theinitialconclusionwasstriking:“theoriginal
mediated genetic exchanges, or the problems of ancient tree of life concept is obsolete; it would not even be a tree
eventsoffusionsorendosymbiosis.However,homoplasyand of one percent” [81]. However, the fact that vertical signal
theoptimizationofcharactersandtreesofcladisticanalysis waspresentintheforestsmeritedreevaluation:“replacement
providearigorousframeworktodiscoverthemagnitudeand of the ToL with a network graph would change our entire
sourceofnonverticalprocessesinevolution.Thus,cladistics perception of the process of evolution and invalidate all
or the ToL are not invalidated by network-like signals. In evolutionary reconstruction” [82]. We note that in these
fact,recentdiscretemathematicalformulationsthattestthe studiesanultrametrictreeofakaryoteswasrecoveredfrom
fundamental axioms of tree construction have also proven vertical signal in a supertree of nearly universal genes.
that the bifurcating history of trees is preserved despite The rooted tree, which was used to simulate clock-like
evolutionaryreticulations[75].Whilethereisno“confusion behavior, revealed the early divergence of Nanoarchaeum
ofdoubts”atthislevel,thebabelofconfusionhasnotstopped. equitans and then Archaea [81]. While inconsistency of the
Genomics revealed evolutionary patchiness, which in- forest supertree increased at high phylogenetic depths, its
cited hypotheses of chimeras and fusions. Phylogenies of associatedsupernetworkshowedtherewerenoreticulations
genes were found highly discordant. Organisms that were betweenArchaeaandBacteria.Thus,thedeeprootingsignal
beingsequencedsharedveryfewgenesequences,andmore of archaeal diversification may be bona fide and worthy of
troublingly,genetreesthatweregeneratedpossesseddiffer- furtherstudy.
ent topologies, especially within akaryotic organisms. With While those that value tree thinking have contested on
time, the number of nearly universal genes decreased and many grounds the idea that the ToL is “obsolete” (e.g.,
the patchiness and discordancy increased. The comparative [13]),ourtakeinthedebateissimple.Homologiesbetween
genomic patchwork of sequence homologies showed that genesequencesestablishedthroughthe“emperor’sBLAST”
thereweregroupsofgenesthatwereonlysharedbycertain (sensu[71])arepoorsubstitutestophylogenetictreerecon-
domains of life. Thus, the makeup of genomes appeared structions from gene sequences. By the same token, phy-
chimeric. A number of hypotheses of chimeric origins of logenies reconstructed from sequences are poor substitutes
eukaryoteswereproposedbasedonthefactthateukaryotes to phylogenies that consider the molecular structure and
shared genes expressing sister group phylogenetic relation- function of the encoded products. Gene sequences are not
ships with both Bacteria and Archaea [64, 76]. One that onlypronetomutationalsaturationbuttheygenerallycome
is notable is the rise of eukaryotes from a “ring of life” inpieces.Thesepiecesrepresentevolutionaryandstructural
fusionofanarchaeonandabacterium(e.g.,[77]),whichin modules that host the functions of the encoded molecules.
a single blow defeated both the tree-like and the tripartite Protein domains, for example, are three-dimensional (3D)
nature of life. Under this school, homology searches with arrangements of elements of secondary structure that fold
BLASTagainstadatabaseof∼3.8millionakaryoticsequences autonomously[83],arecompact[84],andareevolutionarily
allowedtoassignarchaeal,bacterial,orambiguousancestries conserved[85,86].Thelandscapeofevolutionaryexploration
to genes in the human genome and explain homology changes when the role of protein domains in function and
patternsasrelicsoftheakaryoticancestorsofhumans[78]. evolution is considered [87]. Changes at sequence level,
Remarkably,archaealgenestendedtobeinvolvedininforma- including substitutions, insertions, and deletions, can have
tionalprocesses,encodedshorterandmorecentralproteins, littleimpactonstructure,andviceversa;sequencechangesin
and were less likely to be involved in heritable human dis- crucialsitescanhavedevastatingconsequencesonfunction
eases.Thechimericoriginofeukaryotesbyfusionshasbeen andfitness.However,thereisnodetailedanalysisofhistorical
rightfullycontested;thereisnoproperevidencesupporting mismatchesbetweengenesequencesanddomainstructures
itsexistence [79]. Unfortunately,chimerismopenedaflood at the ToL level. Remarkably, while HGT seems rampant
of speculations about reticulation. The history of life was at sequence level, its impact at the domain structural level
seenbymanythroughthelensofa“forest”ofgenehistories is limited [88]. This makes trees derived from domains
8 Archaea
effectively“treesof99percent”andtheiruseverypowerful. regions of the sequence will be fixed (will be structurally
The fact that domains diversify mostly by vertical descent canalized [6]) and will offer little if any phylogenetically
(e.g.,[89–91])suggeststhatgenereticulationssimplyreflect meaningful signal to uncover, for example, the universal
thepervasiveandimpactfulcombinatorialeffectofdomain rRNAcore.Inturn,mutationalsaturationofunconstrained
rearrangementsinproteins[92]andperhapslittleelse.This regionswillquicklyerasehistory.
important claim must be carefully evaluated. It offers the The mutational saturation problem was made mathe-
possibility of dissecting how levels of organization impact matically explicit by Sober and Steel [24] using “mutual
processesofinheritanceinbiology. information” and the concept of time as an information
destroyingprocess.Mutualinformation𝐼(𝑋,𝑌)betweentwo
randomvariables𝑋and𝑌isdefinedby
5.OutoftheImpasse:PartsandWholesin
theEvolvingStructureofSystems
𝑃(𝑋=𝑥,𝑌=𝑦)
TherootingoftheToLisclearlymuddledbythehighdynamic 𝐼(𝑋,𝑌)=∑𝑃(𝑋=𝑥,𝑌=𝑦)log( ).
𝑃(𝑋=𝑥)𝑃(𝑌=𝑦)
natureofchangeinproteinandnucleicacidsequencesand 𝑥,𝑦
by the patchiness and reticulation complexities that exist at (1)
gene level. However, it is also possible that the problems
of the ToL are ultimately technical. We have made the
case that the use of molecular sequences is problematic on When 𝐼(𝑋,𝑌) approaches 0, 𝑋 and 𝑌 become independent
many grounds, including mutational saturation, definition andnomethodcanpredict𝑋fromknowledgeof𝑌.Impor-
of homology of sites in sequence alignments, inapplicable tantly,mutualinformationapproaches0asthetimebetween
characters,taxonsamplingandtreeimbalance,anddifferent 𝑋 and 𝑌 increases in a Markov chain. Regardless of the
historical signatures in domains of multidomain proteins useofaparsimony,maximumlikelihood,orBayesian-based
[93].Inparticular,violationofcharacterindependencebythe frameworkofanalysis,𝐼(𝑋,𝑌)willbeparticularlysmallwhen
mereexistenceofatomicstructurerepresentsaveryserious sequence sites are saturated by too many substitutions due
problemthatplaguesphylogeneticanalysisofsequences.We to high substitution rates or large time scales. Ancestral
here present a solution to the impasse. We show that the states at interior nodes of the trees cannot be established
ToL can be rooted with different approaches that focus on with confidence from extant information even in the most
structureandfunctionandthatitsrootiscongruentlyplaced optimalsituationofknowingtheunderlyingphylogenyand
inArchaea. themodelofcharacterevolution.Undersimplemodels,the
Epistemologically, phylogenetic characters must comply problem is not mitigated by the fact that the number of
withsymmetrybreakingandtheirreversibilityoftime[94]. terminalleavesoftreesandthesourcesofinitialphylogenetic
In other words, characters must establish transformational information increases with time. Since a phase transition
homologyrelationshipsandserveasindependentevidential occurswhensubstitutionprobabilitiesexceedacriticalvalue
statements. [96],onewayoutoftheimpasseistofindfeaturesinsufficient
numberthatchangeatmuchslowerpacethansequencesites
(i) Characters Must Be Homologous and Heritable across and test if mutualinformationis significant and overcomes
TreeTerminalUnits(Taxa).Characterhomologyisacentral Fano’s inequality. These features exist in molecular biology
and controversial concept that embodies the existence of and have been used for phylogenetic reconstruction. They
historical continuity of information [74]. Characters are are,forexample,the3Dfoldstructuresofproteinmolecules
“basic” evidentialstatementsthat makeupcolumnsin data [87] or the stem modules of RNA structures [97], features
matricesusedfortreereconstruction.Theyareconjecturesof that change at very slow rate when compared to associated
perceivedsimilaritiesthatareacceptedasfactforthedura- sequences. For example, protein 3D structural cores evolve
tion of the study, are strengthened by Hennigian reciprocal linearlywithaminoacidsubstitutionspersiteandchangeat
illumination,andcanbeputtothetestthroughcongruence 3–10timesslowerratesthansequences[98].Thishighcon-
withothercharacters,asthesearefittothetrees.Tobeuseful, servationhighlightstheevolutionarydynamicsofmolecular
charactersmustbeheritableandinformativeacrossthetaxa structure.Remarkably,ratesofchangeofproteinsperforming
rowsofthedatamatrix.Thiscanbeevaluated,forexample, a same function are maintained by functional constraints
withthecladisticsinformationcontent(CIC)measure[95]. butacceleratewhenproteinsperformdifferentfunctionsor
Findinginformativecharacterscanbeparticularlychalleng- contain indels. In turn, fold structural diversity explodes
ingwhenthefeaturesthatarestudiedchangeatfastpaceand intomodularstructuresatlowsequenceidentitiesprobably
whentaxasampleawideandconsequentlydeepphylogenetic triggered by functional diversification. Within the context
spectrum.WhenbuildingaToL,thehighlydynamicnature of structural conservation, the fact that fold structures are
ofchangeinthesequencemakeupofproteinornucleicacid structural and evolutionary modules that accumulate in
moleculeschallengestheabilitytoretrievereliablephyloge- proteomes by gene duplication and rearrangements and
neticsignaturesacrosstaxa,evenifmoleculesareuniversally spreadinbiologicalnetworksbyrecruitment(e.g.,[99])also
distributedandharborevolutionarilyconservedregionswith providesasolutiontotheproblemsofvanishingphylogenetic
deep phylogenetic signal (e.g., rRNA). The reason is that signal. Since fold accumulation increases with time in the
given enough time, functionally or structurally constrained Markovchain,mutualinformationmustincrease,reversing
Archaea 9
the “data processing inequality” that destroys information charactertransformationfromacharacterobservedtobemore
andenablingdeepevolutionaryinformation. general to a character observed to be less general, the more
general character is primitive and the less general advanced”
(ii) Characters Must Show at Least Two Distinct Character [103]. This “biogenetic law” appears powerful in that it
States.Oneofthesetwostates(transformationalhomologs) dependsonlyontheassumptionthatontogeniesofingroup
must be ancestral (the “plesiomorphic” state) and the other taxa are properly surveyed. It is also a “direct” method that
mustbederived(the“apomorphic”state)[74].Onlyshared relies exclusively on the ingroup. Consequently, it has the
andderivedfeatures(synapomorphies)provideverticalphy- potentialtoroottheToL.Unfortunately,Nelson’s“generality”
logenetic evidence. Consequently, determining the relative hasbeeninterpretedinnumerousways,especiallyasitrelates
ancestryofalternativecharacterstatesdefinesthepolarityof totheontogeneticsequence,leadingtomuchconfusion[102].
charactertransformationsandrootstheunderlyingtree.This It also involves comparison of developmentally nested and
is a fundamental property of phylogenetic inference. Polar- distinct life history stages, making it difficult to extend the
ization in tree reconstruction enables the “arrow of time” method(originallyconceptualizedforvertebratephylogeny)
(sensu Eddington’s entropy-induced asymmetry), solves the to the microbial world. However, Weston [100, 104] made
rooting problem, and fulfills other epistemological require- it clear that the ontogenetic criterion embodies a wider
ments. “generality criterion” in which the taxic distribution of a
Cladistically speaking, character polarity refers to the character state is a subset of the distribution of another.
distinctionbetweentheancestralandderivedstatesandthe In other words, character states that characterize an entire
identificationofsynapomorphies.However,anevolutionary groupmustbeconsideredancestralrelativetoanalternative
view of polarity also refers to the direction of character state that characterizes a subset of the group. Besides the
statetransformationsinthephylogeneticmodel.Historically, centralityofnestedpatterns,thegeneralitycriterionembeds
three accepted alternatives have been available for rooting thecoreassumptionthateveryhomologyisasynapomorphy
trees [100–102], the outgroup comparison, the ontogenetic innature’snestedtaxonomichierarchyandthathomologies
method, and the paleontological or stratigraphic method. in the hierarchy result from additive phylogenetic change
While the three methods do not include assumptions of [100].Weston’smoregeneralrulethereforestatesthat“given
evolutionary process, they have been the subject of much a distribution of two homologous characters in which one,
discussion and their interpretation of much controversy. 𝑥, is possessed by all of the species that possess its homolog,
The first two are however justified by the assumption that character 𝑦, and by at least one other species that does not,
diversityresultsinanestedtaxonomichierarchy,whichmay then 𝑦 may be postulated to be apomorphous relative to 𝑥”
or may not be induced by evolution. We will not discuss [104]. The only assumption of the method is that relevant
the stratigraphic method as it relies on auxiliary assump- character states in the ingroup are properly surveyed. This
tions regarding the completeness of the fossil record. The new rule crucially substitutes the concept of ontogenetic
midpoint rooting criterion mentioned earlier will not be transformation by the more general concept of homology
discussedeither.Theprocedureiscontestedbytheexistence and additive phylogenetic change, which can be applied to
of heterogeneities in rates of change across the trees and casesinwhichhomologousentitiesaccumulate“iteratively”
problemswiththeaccuratecharacterizationofphylogenetic in evolution (e.g., generation of paralogous genes by dupli-
distances. cation). Since horizontally acquired characters (xenologs)
In outgroup comparison, polarity is inferred by the arenotconsideredsynapomorphies,theycontributetowards
distributionofcharacterstatesintheingroup(groupoftaxa phylogenetic noise and are excluded after calculation of
of interest) and the sister group (taxa outside the group of homoplasyandretentionindices(i.e.,measuresofgoodness
interest).Inasimplecase,ifthecharacterstateisonlyfound offitofcharacterstothephylogeny).
in the ingroup, it must be considered derived. Outgroup Wehaveappliedthe“generalitycriterion”totherootingof
comparisonisbyfarthemethodofchoicebecausephyloge- theToLthroughpolarizationstrategiesthatembodyaxioms
neticiststendtohaveconfidenceinthesupportingassump- of evolutionary process. Figure 4 shows three examples.
tions: higher-level relationships are outside the ingroup, A rooted phylogeny describing the evolution of 5S rRNA
equivalent ontogenetic stages are compared, and character molecules sampled from a wide range of organisms was
statedistributionsareappropriatelysurveyed.Unfortunately, reconstructedfrommolecularsequenceandstructure[105].
themethodis“indirect”inthatitdependsontheassumption The ToL that was recovered was rooted paraphyletically in
of the existence of a higher-level relationship between the Archaea(Figure4(a)).Themodelofcharacterstatetransfor-
outgroupandtheingroup.Consequently,themethodcannot mationwasbasedontheaxiomthatevolvedRNAmolecules
root the ToL because at that level there is no higher-level areoptimizedtoincreasemolecularpersistenceandproduce
relationshipthatispresentlyavailable.Moreover,themethod highly stable folded conformations. Molecular persistence
initselfdoesnotpolarizecharacters.Itsimplyconnectsthe materializes in RNA structure in vitro, with base pairs
ingrouptotherestoftheToL[100]. associatinganddisassociatingatratesashighas0.5s−1[106].
The ontogenetic criterion confers polarity through the Thefrustratedkineticsandenergeticsofthisfoldingprocess
distributionofthestatesofhomologouscharactersinonto- enablesomestructuralconformationstoquicklyreachstable
geniesoftheingroup,generallybyfocusingonthegenerality states.Thisprocessisevolutionarilyoptimizedthroughstruc-
ofcharacterstates,withmorewidelydistributedstatesbeing tural canalization [6], in which evolution attains molecular
consideredancestral.Nelson’srulestates“givenanontogenetic functionsbybothincreasingtheaveragelifeandstabilityof
10 Archaea
Archaea Bacteria Eukarya
Taxa=6665S rRNA molecules of diverse organisms
Characters=sequence and structure of tRNA
Root Treelength=11,381steps
CI=0.072,RI=0.772, and g1=−0.13
(a)
AB B
E
AE A
B Eukarya (E)
EB E
A
Bacteria (B)
A E B A Archaea (A)
B
E
Ancestral Derived
Timeline
(b)
Eukarya
Bacteria
Taxa=150free-living proteomes
Characters=genomicabundance of1,510FSFs
Treelength=70,132steps
CI=0.155,RI=0.739, and g1=−0.28
Root Archaea
1000
(c)
Figure4:TreesoflifegeneratedfromthestructureofRNAandproteinmoleculescongruentlyshowarootinginArchaea.(a)Arooted
phylogenetic tree of 5S rRNA reconstructed from both the sequence and the structure of the molecules (from [105]). (b) Global most-
parsimonious scenario of organismal diversification based on tRNA (from [107]). A total of 571 tRNA molecules with sequence, base
modification,andstructuralinformationwereusedtobuildaToL,whichfailedtoshowmonophyleticgroupings.Ancestriesoflineages
weretheninferredbyconstrainingsetsoftRNAsintomonophyleticgroupsrepresentingcompeting(showninboxes)ornoncompeting
phylogenetic hypotheses and measuring tree suboptimality and lineage coalescence (illustrated with color hues in circles). (c) A ToL
reconstructedfromthegenomicabundancecountsof1,510FSFsasphylogeneticcharactersintheproteomesof150free-livingorganisms
sampledequallyandrandomlyfromthethreedomainsoflife(datatakenfrom[122,123]).Taxawerelabeledwithcirclescoloredaccording
tosuperkingdom.CI=consistencyindex,RI=retentionindex,and𝑔1=gammadistributionparameter.
selectedconformationsanddecreasingtheirrelativenumber. cyberneticmethodthatdecomposesareconstructablesystem
Thus,conformationaldiversitymeasured,forexample,bythe intoitscomponents[109],andusedittorootaToLderived
Shannon entropy of the base-pairing probability matrix or from tRNA sequence and structure (Figure 4(b)). The ToL
featuresofthermodynamicstabilityactas“evo-devo”proxy was again rooted in Archaea. The number of additional
of a generality criterion for RNA molecules, in which the steps required to force (constrain) particular taxa into a
criterion of similarity (e.g., ontogenetic transformation) is monophyleticgroupwasusedtodefinealineagecoalescence
of positional and compositional correspondence. Using a distance (𝑆) with which to test alternative hypotheses of
differentapproach,werecentlybroadenedtheuseofphyloge- monophyly. These hypotheses were then ordered according
neticconstraintanalysis[107,108],borrowingfromaformal to 𝑆 value in an evolutionary timeline. Since 𝑆 records