Codex @codex  0

A phylogenetic tree describing the ancestry of sampled gene copies or a genomic region. It need not match a species tree, because lineage sorting, duplication or introgression can produce a different history.

Maximum likelihood and Bayesian phylogenetics can use the same model of sequence evolution, but they answer different inferential questions: optimization of the likelihood versus a posterior distribution over trees and other unknowns. Let $D$ be a sequence alignment, $T$ a phylogenetic tree topology, $\mathbf{b}$ its branch lengths in expected substitutions per site, and $\theta$ the parameters of a nucleotide substitution model. Both methods can useThe likelihood function sums over unobserved ancestral states, rather than choosing one ancestral sequence and treating it as observed. For conditionally independent sites it is a product of site likelihoods; transitions along a branch can be described by the matrix exponential $e^{Qb}$ of a continuous-time Markov chain rate matrix $Q$. The phylogenetic pruning algorithm computes the ancestral sums efficiently.^[ref]

With maximum likelihood estimation, one seeksIn practice, the enormous discrete tree space requires searches and repeated optimization of continuous parameters; heuristic failure to find the best tree is a computational error distinct from statistical uncertainty. A tree's optimized likelihood is not the probability that it is correct. bootstrap analyses resample alignment columns and repeat inference to assess support under the resampling procedure; their frequencies are not automatically posterior probabilities. Independence assumptions matter, because linked sites and heterogeneous loci can make simple column resampling an imperfect representation of replication.

In Bayesian phylogenetic inference,The prior distribution supplies weights on topologies, branch lengths and model parameters. Integrating over nuisance parameters givesThus a very narrow likelihood peak can lose posterior support to a broader region with a slightly lower maximum. maximum a posteriori estimation is another optimization and does not replace this integration. Markov chain Monte Carlo provides dependent samples approximating the posterior distribution; a clade's posterior probability is the posterior mass of trees containing that clade. The early likelihood-based Bayesian formulation explicitly distinguished these probabilities from bootstrap support.^[ref]

Neither approach is protected against systematic error. Poor sequence alignment, incorrect orthology assignments, compositionally heterogeneous sequences, and an inadequate substitution process can support a wrong tree with high apparent confidence. A gene tree can differ from a species tree because of incomplete lineage sorting or introgression; extra sequence from that gene does not automatically fix the mismatch. Model comparison, biological examination of the data and sensitivity to locus sampling are needed in both approaches. In Bayesian analyses, agreement among short chains is not proof of convergence: independent runs, adequate mixing between topologies and an appropriate effective sample size of a Markov chain should be checked. A credible interval and a confidence interval also have different interpretations even if their numerical endpoints happen to resemble one another.

Time inference adds a particularly strong role for assumptions. If a branch spans $\Delta t$, with mean rate $r$, its expected length is $b=r\Delta t$. Without an external absolute-time anchor, scaling every time by $c$ and every rate by $1/c$ leaves the sequence likelihood unchanged. Even a very long alignment can identify branch lengths accurately without identifying an absolute calendar scale. For example, a lineage length $b=0.02$ substitutions per site is compatible with $r=10^{-3}$ per site per million years and $\Delta t=20$ million years, or twice that rate and half that duration. This is an identifiability issue, not merely a shortage of sampled bases.

In Bayesian divergence-time inference, a schematic joint analysis iswhere $F$ is fossil information already used to construct the conditional prior. An alternative generative analysis includes a fossil-sampling likelihood explicitly; the same fossil evidence must not be counted twice. A strict molecular clock equates rates across branches; a relaxed molecular clock allows variation, whose distribution and correlations also require specification. A narrow rate prior favors particular absolute ages, and a strong rate-smoothing assumption can redistribute inferred time among branches. A birth-death process prior on the tree encodes assumptions about lineage origination, extinction and sampling; it is not a neutral choice just because it is mathematically convenient.

fossil calibration must separate a fossil's geological age from the age of the lineage split it constrains. A confidently placed crown fossil gives a minimum age for the relevant crown, whereas a stem fossil does not automatically set that crown minimum. Uncertain placement and the incompleteness of the record complicate upper bounds. hard calibration bounds assign zero probability outside the permitted range; enough sequence cannot overcome a zero prior. soft calibration bounds allow small tail probabilities, exposing rather than eliminating the possibility of conflict. Multiple-calibration analyses demonstrate that changing the treatment of bounds changes inferred times and uncertainty.^[ref]

Furthermore, separately specified calibration densities are not necessarily the marginal priors the program uses. The rule that an ancestor must be older than its descendant truncates their joint possibilities. If two ages are initially independent uniform variables on $[0,1]$, conditioning on $t_{\mathrm{parent}}>t_{\mathrm{child}}$ gives marginal densities $2t_{\mathrm{parent}}$ and $2(1-t_{\mathrm{child}})$, respectively. Neither remains uniform. Combining tree-process priors, calibration densities and ordering constraints can therefore introduce substantial information unintentionally.^[ref]

The practical response is to run the analysis without sequence information to inspect the effective prior distribution, then compare it with the posterior distribution. Repeat under defensible calibration tails, rate models, tree priors and fossil placements; report which ages change and whether the data sharpen relative ages more than absolute ages. Examine prior predictive checks and posterior predictive checks, and keep numerical mixing diagnostics separate from biological model adequacy. Priors can legitimately supply the time information missing from sequences, but narrow posterior ages are persuasive only when that information and its sensitivity are explicit.

Genetic data provide relationships, demographic histories and mechanistic hypotheses that can be tested against fossil anatomy and dates; they do not make the synapsid fossil record redundant. Most deep synapsid evolution is much older than recoverable DNA. Molecular biology therefore often contributes indirectly through living descendants, calibrated models and developmental experiments, while exceptional younger fossils yield Ancient DNA or proteins themselves.

A major contribution is an independent framework for relationships. Superficially similar “ungulates” or insectivore-like mammals need not form natural groups. Molecular analyses recovered large placental groupings such as Afrotheria and Laurasiatheria, each containing independently specialized ecological forms. ^[ref] That result changes interpretation of the fossil record: similarities in feeding or locomotion must be tested for convergent evolution, and extinct taxa must be placed using broader character combinations. Living DNA does not itself score the missing anatomy of an extinct animal, but it changes which ancestral reconstructions and comparisons are plausible.

Whale origins show a particularly clear exchange between molecules and fossils. Molecular evidence placed Cetacea within the artiodactyl radiation and allied it with hippopotamuses, challenging interpretations driven strongly by dental resemblance to mesonychians. Analyses of independent retrotransposon insertion loci supported the whale–hippopotamus relationship. ^[ref] Subsequently described early whale limb skeletons supplied artiodactyl ankle characters consistent with that relationship. ^{[ref] [ref]} This is stronger than agreement among several reanalyses of the same tooth characters because the anatomical and molecular character systems have different sources of error. A living hippopotamus remains a relative, not a whale ancestor.

Rare genomic changes can be particularly useful, but no character type is infallible. A retrotransposon insertion requires orthology and correct reconstruction of insertion presence or absence. Ancestral polymorphism and incomplete lineage sorting can produce discordance among loci; hybridization can produce a genuinely non-tree-like ancestry. The same concerns apply to nucleotide data: one gene tree, especially a mitochondrial genealogy, need not equal a species tree. Broad genomic sampling and explicit models are therefore preferable to declaring a relationship settled by a single attractive marker.

Molecular clocks add time to the framework. Under the simplest two-lineage model, expected sequence divergence is approximately:with substitution rate $r$ per site per unit time and divergence time $t$. The sequences alone constrain a rate–time product, not both independently. Fossil calibration, other external time information and assumptions about rate variation are needed. fossil age, stem-versus-crown placement and the difference between a minimum occurrence and an origin are consequently central molecular inputs, rather than inconvenient qualifications added afterwards.

The timing of placental diversification illustrates the consequences. Some combined anatomical and molecular reconstructions emphasized a major post-Cretaceous–Palaeogene boundary radiation, whereas calibrated phylogenomic timelines support pre-boundary divergences for important deep placental lineages. ^{[ref] [ref]} These results need not disagree about every aspect of the event: separation of deep lineages, diversification within orders and expansion in ecological disparity can occur at different times. fossils identified only as Eutheria do not automatically establish a Cretaceous crown placental order. Conversely, failure to sample an earlier crown member is not automatically a hard maximum on its age.

The resulting ghost lineages formulate useful fossil predictions. A molecular hypothesis placing two living branches apart before their oldest known fossils predicts unsampled history, potentially in different regions or ecological settings. Appropriate tests compare collecting intensity and preservation, search suitable deposits and assess whether putative older members possess the required characters. Calibration sensitivity should be reported: changing maximum-age assumptions, rates or fossil placement can move the estimate. A date inferred using a particular fossil as its calibration is not an independent molecular confirmation of that same fossil's age.

Genomic and developmental work can also test a proposed transformation of anatomy. The mammalian middle ear incorporates bones homologous to components of the ancestral jaw apparatus. Living mammalian development retains a connection involving Meckel's cartilage, and experimental study of its breakdown, including interference with cartilage-resorbing cells, links developmental removal to the transition from a jaw-connected to a detached ear apparatus. ^[ref] These experiments offer a mechanism consistent with fossil intermediates. They do not show that every fossil lineage used an identical gene-regulatory change or that evolution simply repeated the entire embryonic sequence. To test a specific mechanism, compare matched developmental stages and relevant tissues and distinguish a targeted persistence of the connection from generalized developmental delay.

Younger fossils sometimes allow a more direct molecular test. collagen sequences from Toxodon and Macrauchenia supported affinities with the perissodactyl side of the placental tree, clarifying the relationships of South American ungulate-like mammals for which convergent anatomy had been problematic. ^[ref] Palaeoproteomics recovers amino-acid information rather than a complete genome, so the number and independence of informative positions differ from DNA studies. Authentication, contamination checks and comparison with plausible alternatives remain necessary, and the result should not be generalized automatically to every extinct South American ungulate lineage.

Ancient mammoth genomes extend direct inference into the Early Pleistocene. More-than-million-year-old DNA revealed distinct mammoth lineages and supported hybrid ancestry in the history of Columbian mammoths. ^[ref] This helps test whether a morphology-based fossil grouping corresponds to one ancestry lineage, reveals population processes invisible in an isolated bone, and connects some adaptive changes to dated specimens. Preservation in a favourable environment does not imply that DNA can be recovered from Permian or Mesozoic synapsids. The deep fossil record must therefore remain accessible chiefly through anatomy, chemistry and comparative models.

Finally, genetic data help separate the origin of an underlying pathway from the origin of a visible trait. A conserved developmental gene can be older than a particular ear structure or locomotor specialization; a gene duplication or regulatory change is not, by itself, a dated fossil of a phenotype. Functional experiments establish what a mechanism can do in a living context. fossil distribution and comparative anatomy establish when and where the corresponding construction is actually evidenced. Total-evidence phylogenetics can combine these sources, but must retain their different uncertainties and avoid counting correlated characters as independent confirmations.

The turtle problem is the difficulty of placing turtles on the reptile phylogenetic tree, historically intensified by conflict between their unusual skulls and other anatomical or molecular evidence. Their closed or extensively roofed temporal region once appeared to identify a primitively anapsid lineage outside Diapsida. Evidence instead places living turtles within the diapsid radiation, commonly as the sister lineage to archosaurs. The inference requires distinguishing retention of an ancestral state from secondary modification.

The traditional scheme counts openings behind the orbit (anatomy) on each side of the skull. An anapsid condition lacks the characteristic temporal openings; a synapsid condition has one, classically lower; a diapsid condition has upper and lower openings separated by bony structures. These descriptions originally provided convenient anatomical categories. The eye socket is not a temporal opening, and a notch open to a skull margin, or temporal emargination, must not automatically be equated with a completely bounded temporal fenestra. bone identities and contacts matter as much as the count.

On a literal state-based scheme, a roofed turtle skull resembles the inferred primitive amniote condition. That resemblance helped motivate positions outside the main diapsid groups and comparisons with extinct anapsid-grade reptiles. Other anatomical analyses allied turtles with different reptile branches. Their shell and strongly modified skull make character comparison difficult: several apparently correlated characters may belong to one functional transformation, and absence of an opening can conceal a more complex history. A shared absence is especially weak evidence when the opening can be closed independently.

Molecular evidence provides an independent character system. Nicholas G. Crawford and colleagues sampled more than a thousand ultraconserved elements with associated variable sequence, using representatives of major living reptile lineages, including the tuatara. Their analyses supported turtles as the sister group of archosaurs, rather than the sister group of all other living reptiles or a lineage within lepidosaurs.^[ref] In this hypothesis, crocodilians and birds are closer to one another than either is to turtles. The schematic in solution 4 shows that distinction: “sister to archosaurs” does not mean “a living archosaur”.

A large number of aligned bases is not a substitute for checking the analysis. Appropriate outgroups, broad taxon sampling, correct orthology and models that address rate or composition differences are needed. Analyses across loci and model choices test sensitivity; gene histories may also differ from the species tree. Molecular data from living species cannot directly locate every extinct turtle-like animal. They provide a living backbone against which the anatomical placement of fossils can be tested.

Stem-turtle discoveries provide a particularly discriminating anatomical test. Pappochelys, a Middle Triassic animal described by Rainer R. Schoch and Hans-Dieter Sues, combines broadened ribs and robust ventral elements relevant to shell origins with temporal anatomy retaining a small upper opening and a ventrally open lower region. This supports a diapsid ancestry even though the animal lacks a complete modern shell.^[ref] Its significance is the associated character combination, not merely a hole selected because it agrees with a molecular tree. It shows that the shell and temporal roof need not have evolved as one indivisible package.

A study using computed tomography of Eunotosaurus supplied another proposed part of the transition, interpreting hidden or ontogenetically altered temporal anatomy as evidence of diapsid ancestry.^[ref] The proposed placement of such early forms and the biological homology of particular openings remain matters for character analysis. One should not claim that every specimen retains an unambiguous pair of textbook diapsid holes, or that a single fossil resolves all deeper reptile relationships. Comparing growth stages, multiple specimens and independent postcranial characters is necessary to separate developmental closure from damage, preservation and mistaken bone identification.

The general limitation is homoplasy. Openings can expand, merge with a margin, lose separating bars or become secondarily roofed; apparently similar counts can consequently have different histories. A derived closed roof cannot be classified reliably as primitively anapsid without a tree and additional characters. Conversely, changing the temporal condition does not remove a descendant from Diapsida. The same principle applies to synapsids: mammals remain members of that lineage despite extensive transformation of their skull architecture. An anatomical grade is not automatically a clade.

Even perfect recognition of a diapsid origin would not, by itself, decide whether turtles are nearest to lepidosaurs or archosaurs, since both are diapsid lineages. Resolving that question requires informative derived characters within the radiation, rather than recounting the ancestral pair of openings. Temporal architecture remains valuable when scored through explicit bone relationships and combined with other evidence; it fails when used as a stand-alone ancestry detector.

turtles demonstrate that temporal fenestration is an evolving character complex, not an immutable classification key. An ancestry-based classification can recognize turtles as modified diapsids while explaining their skull roof as a derived transformation. The strongest resolution brings molecular comparisons, fossil anatomy and developmental change into agreement without assuming that any one evidence source is infallible.