Colors correspond to the three topologies. We compared the single-locus posterior probabilities we estimated through the Bayesian phylogenetic analyses with bootstrap support from maximum likelihood analyses to determine if the high support we obtained was due to the choice of methodology.
Bootstrap supports from the maximum likelihood analyses were highly correlated with the Bayesian posterior probabilities for each of the three topologies M. To measure phylogenetic discordance across all loci, we used the posterior probability distributions from all single-locus phylogenetic analyses as input for the Bayesian concordance analysis [71]. By incorporating the statistical uncertainty in phylogenetic reconstruction among the individual loci, we were able to estimate the number of loci across the genome that supported each individual topology.
Bayesian concordance analysis identified a primary phylogenetic history, placing M. This tree was supported by a concordance factor of 0. There was also substantial support for the two other possible histories. The M. Although the Bayesian concordance analysis estimates the proportion of loci supporting a particular topology, the analysis does not integrate the sizes of individual loci.
If the concordance factors are accurately reflecting the contributions of incomplete lineage sorting and gene flow, the median locus sizes supporting each of the three topologies should parallel the concordance factors. Median locus size followed the same rank order as the concordance factors, with the M. Bayesian concordance factors were calculated from the posterior probability distributions of 14, single-locus phylogenetic analyses. We also repeated our concordance analysis using maximum likelihood methods.
Of these trees, M. These concordance factors were similar to those resulting from Bayesian concordance analysis, indicating that our conclusions were robust to alternative analytical approaches. The slight differences likely arose because Bayesian concordance analysis used all loci and incorporated uncertainty across loci, whereas the maximum likelihood method only used half of the available loci and did not propagate uncertainty across loci.
Bayesian concordance analysis uses a prior probability of gene tree concordance, which could affect our estimates of a primary phylogenetic history. To address this issue, we recalculated concordance factors assuming two extreme priors: 1 a high probability of concordance among individual loci, and 2 complete independence among individual loci.
In both cases, patterns of discordance among the three histories remained largely unchanged Figure 3 , suggesting robustness to prior assumptions. This robustness was likely due to the large number of loci used in the concordance analysis combined with the high degree of support for single topologies at most individual loci.
Estimates of concordance might also be affected by the parameters used in the MDL partitioning the cost and the starting interval size. To investigate this possibility, we applied the maximum 3 and minimum 0. In all cases, partitioning the genome with the minimum cost roughly doubled the number of loci on each chromosome, but chromosome-wide concordance factors were not significantly altered Figure S6.
We also calculated concordance factors on chromosomes 18 and 19 using a range of starting interval sizes 25, 50, , , , and SNPs. For both chromosomes, concordance factors did not significantly differ for the three topologies with starting interval sizes of 25, 50, or SNPs when the full range of credibility intervals were taken into account Figure S7 , indicating that partitioning with starting intervals smaller than SNPs did not significantly alter the estimates of concordance.
The phylogenetic histories at individual loci may not be the true histories if the divergence time between rat and house mice is too great [72] , [73]. Under this scenario, the rat branch can pair with whichever mouse lineage has the greatest amount of divergence and the largest number of sequence similarities due to homoplasy rather than orthology long-branch attraction.
To evaluate this possibility, we randomized the nucleotides of the rat sequence at each locus, erasing any phylogenetic signal and further compounding the effect of long-branch attraction. The posterior probability distributions from each shuffled locus on chromosomes 18 and 19 were used as input for the Bayesian concordance analysis [71].
If the patterns we observed in the data were due to long-branch attraction, we would expect to recover similar patterns of discordance with an artificially lengthened branch. This indicated that the rat sequence provided strong phylogenetic signal.
Patterns of phylogenetic discordance within the X chromosome are expected to differ from those on the autosomes. Additionally, loci on the X chromosome exhibit reduced gene flow within and between species of house mice [30] , [32] , [33] , [36] , [38] , [39] , [74] , [75]. Both factors should reduce discordance across the X chromosome. In agreement with patterns for the autosomes, the primary phylogenetic history of the X chromosome was a M. As predicted, this history was supported by a higher concordance factor 0.
In addition, loci supporting a M. Although the concordance factor supporting a M. Median locus sizes matching each topology on the X chromosome also paralleled the concordance factors as observed on the autosomes , with the M.
Significantly less discordance is observed across loci of the X chromosome X: loci than the autosomes A: 13, loci. This is shown by significantly higher support for the primary phylogenetic history, the M.
Increased support for the M. The reduced discordance on the X chromosome persisted in these comparisons M. Reduced discordance on the X chromosome thus appears to reflect processes differentially affecting the X chromosome and the autosomes. A potential source of discordance in our results comes from ascertainment bias in SNP identification.
Consistent with such a bias, a deficiency of SNPs from the three species was documented [59]. The strongest bias was against M. This deficiency could reduce the number of loci supporting a M. To determine whether ascertainment bias would affect our ability to resolve a primary phylogenetic history, we simulated increased ascertainment bias against M.
All phylogenetic analyses were then repeated. These simulations modeled the effects of artificially increasing the false negative rate of SNP identification against M. For each chromosome, as ascertainment bias against M. Importantly, introducing ascertainment bias did not differentially affect the inferred concordance factors for the other two topologies; both factors increased at equal rates.
Although ascertainment bias against M. Varying proportions of M. The difference in concordance factors between M. This indicates that recovery of a M.
A second form of bias could also increase the branch length of M. Perlegen discarded SNPs that were polymorphic in only one of the 15 strains sequenced.
As a majority of the strains were M. If this bias increased the branch length of M. When the rat sequence was randomized, support for the M. However, on both chromosomes, the M. This result indicated that the M. Our genomic analysis revealed a primary phylogenetic history across the house mouse genome, placing M.
We also documented striking phylogenetic discordance on a genome-wide scale. Discordance was observed in previous phylogenetic studies of house mice based on a small number of loci [30] , [53] , [54]. In addition, gene trees reconstructed from large population samples have shown that reciprocal monophyly between subspecies is higher on the X chromosome than the autosomes [29] — [31] , a result that agrees with our genomic comparisons. In summary, our results extend previous observations from phylogenetic analyses of a few loci to the entire genome, thereby providing the power needed to resolve the history of these closely related subspecies for the first time.
Several additional factors might shape the discordance we observed. We now discuss the importance of each potential source of discordance in turn. Errors at several stages of phylogenetic reconstruction could generate phylogenetic discordance [76] — [78]. First, mis-estimated models of molecular evolution could introduce disagreement among loci. However, by statistically selecting the best-fitting model of molecular evolution separately for each locus, we minimized errors associated with assuming the same model across loci.
Second, the alignment with rat might have inflated discordance if the error rate in the whole-genome alignment was high. Contrary to this idea, randomizing the rat sequence in respect to the three mouse sequences across chromosomes 18 and 19 strongly reduced the posterior probabilities at individual loci and instead exacerbated discordance, suggesting that the rat sequence contributed a strong phylogenetic signal.
Third, estimates of concordance might be affected by the parameters used in the MDL partitioning. Applying the minimum cost against splitting concatenated fragments roughly doubled the number of loci, but concordance factors were not significantly altered. In addition, partitioning the genome using starting intervals less than SNPs had no significant effect on the concordance factors.
Finally, concordance factors might have been inaccurately estimated because many SNPs were missed by resequencing [57]. Although comparable analyses of complete genome sequences would likely reveal variation in the exact breakpoints of partitions, reduced numbers of informative sites did not seem to be responsible for the observed discordance. Our analyses demonstrated that MDL partitions the genome in a phylogenetically informative manner and that individual loci generally favor one history with high posterior probability.
In addition, we found a significant correlation between locus size and recombination rate across the genome as predicted by theory , suggesting that this dataset contains information about the evolutionary processes responsible for phylogenetic discordance.
Although we detected a significant negative correlation between locus size and recombination rate, the correlation coefficient was relatively low, indicating that most of the variation in locus size was explained by other variables. The weakness of this correlation was expected for several reasons. First, our data set was limited by the number of informative sites generated by the resequencing project.
Additional sequence data might change the locations of breakpoints inferred by the MDL partitioning, which would alter the locus sizes and the correlation with recombination rate.
Second, the recombination rate estimates came from crosses between other inbred strains of mice [64] , [65] , not the wild-derived strains used in our analyses. Pairwise divergence times between house mouse subspecies pairs are roughly similar when the full range of confidence intervals is considered [30] , suggesting a rapid, sequential splitting of the three subspecies. This scenario is expected to result in concordance factors that differ only slightly from 0.
Our results are consistent with these patterns, with a primary phylogenetic history supported by a concordance factor of 0. In contrast, three-taxon cases in Drosophila and primates feature phylogenies with longer internal branches, resulting in a greater proportion of the genome supporting the primary phylogenetic histories [17] , [22] , [23] , [25].
If incomplete lineage sorting is solely responsible for phylogenetic discordance, the two minor topologies should occur at equal frequencies in the genome [9] , [12] , [56] , [76] , and these frequencies should decrease at equal rates as effective population size decreases and the length of the internal branch increases [11] , [15]. In contrast, our analysis revealed asymmetric genomic proportions supporting the two minor topologies, indicating a strong deviation from the model of pure lineage sorting.
Similar patterns were observed in Drosophila species [17] , and on the X chromosome in primates [10] , [22]. Gene flow following divergence can drive asymmetries between the minor histories. Patterns of shared polymorphism among populations [29] — [31] and introgression across hybrid zones [30] , [32] — [39] , [74] , [75] indicate that gene flow differs among the subspecies pairs and across the genome. If the primary phylogenetic history M.
Significant levels of gene flow have only been detected between M. This introgression is expected to increase support for the M. In addition to gene flow in nature, sequencing error likely contributed to differences in concordance factors between the two minor histories. It has been suggested that sequencing errors could have caused differences in the genomic proportions supporting alternative minor histories in Drosophila and on the primate X chromosome [10].
Resequencing studies have detected a high false negative rate against M. This bias probably led us to underestimate the concordance factor for the M. Although we cannot separate the contributions of recent gene flow and ascertainment bias to the asymmetry between minor histories in our analyses, ascertainment bias seems to have played a larger role in producing this pattern. If the asymmetry between minor histories was mostly due to gene flow, we would expect it to be less apparent on the X chromosome because recent introgression has been relatively reduced on the X chromosome [30] , [34] — [36] , [39] , [74] , [75].
In contrast, differences between minor histories were similar for the X chromosome 0. Furthermore, the asymmetry was still present after the rat sequence was randomized at each locus across chromosomes 18 and Because the shuffling erased phylogenetic signal due to orthology, lowered support for the M. This result also supports the idea that ascertainment bias contributed to the difference in concordance factors between the two minor histories.
Although ascertainment bias appears to have affected the relative frequencies of the minor histories, it does not seem to have interfered with our identification of a primary phylogenetic history. Randomly removing M. Assuming the asymmetry was entirely driven by ascertainment bias, we adjusted the data according to the simulations by lowering the concordance factors of the M. Although these concordance factors do not include any effect of gene flow an unrealistic assumption , these rough estimates allowed us to calculate the length of the internal branch of the subspecies tree that would maximize the likelihood of our dataset under a model of pure lineage sorting [80].
The high level of phylogenetic discordance we observed suggests a rapid splitting of the three house mouse subspecies, consistent with close divergence times among the three subspecies estimated from large population samples [30]. Several additional lines of evidence support this conclusion. First, there is significantly higher support for this topology on the X chromosome relative to the autosomes where incomplete lineage sorting is expected to be reduced.
If either of the minor histories were the true subspecies tree, rates of gene flow would need to be higher on the X chromosome than on the autosomes to explain the difference in concordance factors. However, gene flow on the X chromosome is considerably lower [30] , [32] , [33] , [36] , [38] , [39] , [74] , [75]. Second, we observed increased support for the M. In species that experience gene flow after the initial development of reproductive isolation, loci underlying reproductive barriers might better reflect species history because discordance generated by gene flow is reduced in these regions [76] , [81].
Increased phylogenetic resolution of species history has been observed at loci associated with hybrid male sterility [82] — [84]. Within house mice, loci that affect hybrid male sterility have been mapped repeatedly to the X chromosome in crosses between M. As a preliminary examination of the association between reproductive isolation loci and phylogenetic history, we performed concordance analyses in sliding windows comprised of four contiguous loci across the X chromosome.
We identified several adjacent regions supporting a M. The highest peak spanned a 1. As might be predicted from the mapping results, concordance factors supporting the M.
We did not detect similar associations for other known hybrid sterility loci, including Hst1 on chromosome 17 [85]. However, the results from the X chromosome should motivate similar analyses across the entire genome once more information is available about the regions contributing to reproductive isolation in house mice. Discordance within a four-locus sliding window was calculated across the X chromosome and is plotted as the midpoint position of each window.
The entire Hstx1 interval is indicated by the purple line, whereas the peak of this quantitative trait locus is indicated by the orange line. The black line indicates the chromosome-wide concordance factor for the M. In addition to informing speciation studies, the phylogenetic history of house mouse subspecies has important implications for mouse genetics. The classical inbred mouse strains widely used in genetic studies of disease and other phenotypes are descended — in unequal proportions — from the three subspecies examined here [86] , [87].
Analyses of the Perlegen sequences documented substantial genomic variation in relationships among the classical strains and attempted to attribute the ancestry of different genomic regions to M. Our results suggest that much of this phylogenetic variation likely reflects incomplete lineage sorting and differential introgression in wild mice.
Genomic comparisons involving larger numbers of wild mice will be needed to interpret the patterns of genetic variation observed in the classical strains. The UCSC table browser data retrieval tool. Nucleic Acids Res. Bioinformatics 26 : — A maximum pseudo-likelihood approach for estimating species trees under the coalescent model. Liu L. BEST: Bayesian estimation of species trees under the coalescent model. Bioinformatics 24 : — Maddison WP.
Gene trees in species trees. Linking great apes genome evolution across time scales using polymorphism-aware phylogenetic models. PoMo: an allele frequency-based approach for species tree estimation. How reticulated are species? Bioessays 38 : — Convergence in pigmentation at multiple levels: mutations, genes and function.
Phil Trans R Soc B. Genome-wide evidence for speciation with gene flow in Heliconius butterflies. Genome Res. Epistasis and the dynamics of reversion in molecular evolution. Genetics — Gene tree discordance causes apparent substitution rate variation. Mirarab S Warnow T. Bioinformatics 31 : i44 — i Mita S Siol M. EggLib: processing, analysis and simulation tools for population genetics and genomics.
BMC Genet. Fitness conferred by replaced amino acids declines with time. Sequencing of the genus Arabidopsis identifies a complex history of nonbifurcating speciation and abundant trans-specific polymorphism. Oliver JC. Microevolutionary processes generate phylogenomic discordance at ancient divergences. Orr HA. The probability of parallel evolution.
Evolution 59 : — Pamilo P Nei M. Relationships between gene trees and species trees. Genome-wide signatures of convergent evolution in echolocating mammals. Phylogenomics reveals three sources of adaptive variation during a rapid radiation. PLoS Biol. Widespread discordance of gene trees with species tree in Drosophila: evidence for incomplete lineage sorting. PLoS Genet. Amino acid coevolution induces an evolutionary Stokes shift.
Sequence space and the ongoing expansion of the protein universe. Repeated elevational transitions in hemoglobin function during the evolution of Andean hummingbirds. Rasmussen M Kellis M. Unified modeling of gene duplication, loss, and coalescence using a locus tree. Direct 3 : 7. Rokas A Carroll SB. Frequent and widespread parallel evolution of protein sequences. The molecular basis of phenotypic convergence. Insights into hominid evolution from the gorilla genome sequence. Contingency and entrenchment in protein evolution under purifying selection.
Does gene tree discordance explain the mismatch between macroevolutionary models and empirical patterns of tree shape and branching times?
Stern DL. The genetic causes of convergent evolution. Storz JF. Causes of molecular convergence and parallelism in protein evolution. The dynamics of incomplete lineage sorting across the ancient adaptive radiation of neoavian birds. Determining the null model for detecting adaptive convergence from genomic data: a case study using echolocating mammals. Reconciliation with non-binary species trees. Whelan S Goldman N. A general empirical model of protein evolution derived from multiple protein families using a maximum-likelihood approach.
Yang Z. PAML 4: phylogenetic analysis by maximum likelihood. The BPP program for species tree estimation and species delimitation. Comparative genomics reveals insights into avian genome evolution and adaptation. Zhang J Kumar S. Detection of convergent and parallel evolution at the amino acid sequence level. Parallel molecular evolution in an herbivore community. Zou Z Zhang J. Are convergent and parallel amino acid substitutions in protein evolution more prevalent than neutral expectations?
No genome-wide protein sequence convergence for echolocation. Oxford University Press is a department of the University of Oxford.
It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide. Sign In or Create an Account. Sign In. Advanced Search.
Search Menu. Article Navigation. Close mobile search navigation Article Navigation. Volume Article Contents Abstract. Materials and Methods. Oxford Academic. Matthew W. Associate editor: Beth Shapiro.
Select Format Select format. Permissions Icon Permissions. Abstract Phenotypic convergence is an exciting outcome of adaptive evolution, occurring when different species find similar solutions to the same problem. Open in new tab Download slide.
Google Scholar Crossref. Search ADS. Google Scholar PubMed. All rights reserved. For permissions, please e-mail: journals. Issue Section:. Download all slides. Supplementary data. Supplementary Data - zip file. View Metrics. Email alerts Article activity alert. Advance article alerts. New issue alert. Receive exclusive offers and updates from Oxford Academic. Related articles in Web of Science Google Scholar. Citing articles via Web of Science Additional mechanisms and processes may also result in genealogical discordance, though less work has been done on these in the specific context presented here and, in some cases, whether they ought to be treated as modes of inheritance is more controversial.
That includes the impact of epigenetic or environmental niche inheritance across generations Jablonka and Lamb ; Laland et al. It is important to be attendant to the breadth of modes of genealogical discordance. One reason for that is the danger of over-attributing the presence of discordance to one of these modes.
For example, if high levels of genealogical discordance were discovered in a system, yet only HGT considered as an explanation for the presence of this discordance, then the level of HGT might be over-estimated for that system and other explanations under-estimated. This threatens to obscure the role other modes of production of genealogical discordance played in the history of that system Galtier and Daubin ; Haber Another reason to be attendant to the various modes of genealogical discordance is the opportunity it presents.
Different modes of genealogical discordance produce different patterns of discordance. As researchers better identify those distinct patterns and build models of inference that take advantage of this, we are in a better epistemic position to report on the complex histories of evolutionary systems e.
Methodological considerations are certainly an important component of how genealogical discordance impacts our approach to studying species. Yet, my primary interest here is broader than that, and in the next section I consider this through three framing questions that set research problems for how we think about species. Merely asking how genealogical discordance impacts our study of species is too imprecise.
There are so many ways to think about species, let alone how we might study them, that we need to introduce some structure to usefully move forward. Here I use three framing questions to facilitate that goal:. These framing questions are intended to produce more precise questions around the study of species—questions that, in turn, identify opportunities and challenges that researchers are or ought to be exploring.
For the first two, I use specific examples to highlight how an appreciation for genealogical discordance complicates and, in many cases, enriches the way we interrogate biological divergence and diversification. For the third framing question, I take the opportunity to be more speculative.
At least at a broad level, discordance can be viewed as challenging the very notion of coherently discrete or distinct taxonomic groups in biology at all. Darwin included only a single figure in his Origin figure 2 —a figure which may be interpreted in lots of ways and was, even by Darwin!
This figure helps him explain how, in a system of descent with modification, processes of divergence and diversification, coupled with selection, generate distinct groups. One natural reading of that is to treat those groups that are products of those processes as taxa typically species, but the reading may hold at other levels of hierarchy too.
Notoriously, fleshing out precisely what sorts of groups of organisms should be treated as species has been an ongoing source of disagreement. We also might disagree over what criteria we ought to use to determine which groups are groups of species as opposed to other kinds of groups , e.
We might be committed to a prospective approach that groups together as a species those things that we think will continue to persist as groups; in contrast, adopting a retrospective approach identifies as a species those groups that have historically held together by some specified criteria Harrison That is just a sampling of the many dimensions along which debates over the so-called species problem can take shape, though I have no interest in rehashing that here see Wilkins for a more exhaustive analysis.
Instead, in this section I want to make the case that genealogical discordance makes the species problem harder; it complicates and muddies that debate by troubling what has been a background assumption of much of that debate.
Genealogical discordance challenges the underlying assumption that evolution is generating singular distinct and discrete groups as the products of evolution. Namely, do we still see discrete groups in the context of a system that also exhibits enormous amounts of discordance? What kinds of groups do we see in a system like this?
In what way are they meaningfully discrete? What if multiple groups are being generated that are overlapping with one another in complex and interesting ways, perhaps even overlapping in dynamic ways if we introduce a temporal element? This may mean that the groupings we identify are multifaceted and complex in peculiar and surprising ways, but in ways that reflect the processes of divergence and diversification that we are hoping to study and understand see figure 3.
As with all things species, there are, of course, debates over what it means for a species to be cohesive, and what that implies for what kinds of things species are. Those debates concern the epistemic criteria for grouping e. This echoes similar sentiments voiced by many microbiologists though see Galtier and Daubin for a dissenting view. One response that may be offered to this is to observe that there are multiple mechanisms working for and against the emergence of such groups, and that the expression of those various mechanisms and their balance against one another is far from universal across clades.
This suggests that the distinctiveness and discreteness of groups or levels of hierarchy is an evolved feature of biological hierarchies that might not be expressed uniformly across the entirety of biology. Among other things, this supports a rank-free view of that hierarchy Griffiths ; Queiroz and Gauthier ; Ereshefsky ; Okasha , and, more radically, lends credence to views that the groups that did evolve above the level of organism may not be as similar or uniform as is often supposed and, thus, may fail to license inferences presuming that similarity Mishler Franklin's observation notwithstanding, variation across biology is not continuous, and there are meaningfully distinct groups that reflect underlying biological processes and mechanisms.
Yet, these groups are rarely, if ever, categorically and cleanly delineated. Precisely how genealogical discordance will shape the species debate is still an open question.
But it is a question that offers a rich research opportunity, and a sophisticated, nuanced treatment. However, it would be a mistake to think this entails the dissolution of meaningful groupings at all in biology. Groups must be understood as complex, dynamic, and multi-dimensional, but that, in turn, leads to a far richer understanding of evolutionary systems. These groups may fail to be natural kinds, [8] but that hardly means they are not theoretically relevant, causally efficacious, or explanatorily rich.
It is, thus, a mistake to think that only groups that are natural kinds, or that present clean categorical boundaries, are groups worth tracking, naming, or classifying e. How has growing appreciation of genealogical discordance impacted scientific practice? In this section I offer one answer to these questions, making the case that genealogical discordance pushes us to revise a centrally important methodological commitment in phylogenetics.
Lin et al. These phylogenetic hypotheses are in conflict, i. The traditional approach in cases like this is to resolve that conflict, understood as identifying which one of the available hypotheses is the best supported.
That is hardly surprising, and follows from the central commitments of phylogeny reconstruction. Felsenstein states this with characteristic clarity, putting it plainly:. If each site in a set of sequences has changed only once in the evolution of a group, then the newly-arisen base will be shared by all species descended from the lineage in which the change occurred.
If this were the case at all sites, then the sets of species having the new bases would be either perfectly nested or disjoint, never overlapping unless one set of species was included in the other. It would be possible to erect a tree on which we could explain the evolution of the group with only a single change at each site. This can be done by inspection of the sets of species defined at each varying site. If some of these sets of species overlap without being nested, then there is conflict between the information provided by different sites.
Most of the interesting issues in phylogeny reconstruction are in how to resolve these conflicts. The conflict described here by Felsenstein covers a lot of kinds of conflict we see in phylogenetics though, as is hopefully clear by now, not all the kinds of conflicts.
Regardless, the highlighted part of this passage captures an important commitment in phylogenetic inference, that the central task is to resolve conflicting phylogenetic signals.
Traditionally, this has meant identifying the single best-supported tree. That is certainly an important methodological commitment about the goal of phylogenetic analysis, and not one I am disputing here. However, I am suggesting that it needs to be revised in light of discoveries that the breadth and depth of genealogical discordance is far greater than anticipated—so much so that genealogical discordance ought to be regarded as simply a background feature of most, if not all, evolving systems.
The change this revision implies is subtle but impactful. When faced with conflicting information about phylogenies, phylogeneticists now must consider two different possible kinds of explanations. On the one hand, the conflict might be resolved in favor of one or the other phylogenies. This involves testing competing phylogenetic hypotheses for that which is best supported on competing models of phylogenetic inference. This approach reflects the methodological commitment as previously understood and articulated by Felsenstein.
On the revised commitment, and in recognition of genealogical discordance, phylogeneticists should also consider whether the conflict might be reconciled as reflecting the outcome of genealogical discordance; i.
That is, that the conflicting patterns correctly reflect the multiple histories a complex lineage might contain. This is not to criticize Lin et al. I have no evidence nor have I analyzed their data to determine whether a discordant explanation of reconciliation better explains the data than resolving the conflict; on the contrary, the study looks like a model of careful, good work that nicely reflects the central methodological commitments of phylogenetics.
Rather, it is a testament to how quickly science can shift that a study done in can now be viewed as working in a rather narrower framing question than we might now expect. An appreciation of genealogical discordance complicates what was a reasonable methodological commitment less than a decade ago.
It is an example of a shift in commitments in a research community, and, I believe, how our expectations of scientific methodology may advance or progress in response to empirical discoveries as well as conceptual and theoretical advances.
Because it enriches our understanding of evolutionary systems and explains a greater amount of data in response to disruptive challenges, I call advances like these productive disruptions. Advances in our understanding of genealogical discordance mean that a study like now include the possibility that what appear to be conflicts may be reconciled by appeal to genealogical discordance. Hailer et al.
Similarly, Willyard, Cronn, and Liston argue that patterns of incongruence in ponderosa pine nDNA and chloroplast haplotypes are best explained by introgression followed by a genetic bottleneck rather than incomplete lineage sorting, or other modes of genealogical discordance , concluding that single-locus analyses risk missing these biologically relevant patterns.
0コメント