The field is somewhat divided on this issue, although I am not sure if there is a majority view, nor which of the two viewpoints would be favored. There is a clear split between those who regard passerines as a "crown group" whose thousands of species adorn the terminal branches of the phylogenetic tree of birds, and those with a complementary view that passerines have an ancient origin that has given this group plenty of opportunities to make repeated branches without an extremely high mutation rate. The same raw data of DNA substitutions can lead to either picture of passerine phylogeny, depending on assumptions about mutation rates, so the question arises whether this circularity can be broken from within the DNA data themselves. Or do we need another, independent source of information to anchor the ambiguous sequence information?

I favor the idea that the passerines represent a recent successful clade. My evidence is primarily based on human judgments of their similarity.

James, I like the fact that you have informed your decision using data from outside DNA. There is another source of such information, the avian nervous system. This neurological "anchor" strongly supports an ancient origin for passerines. More on that later.

Why not have trust in fossil data? See e.g. Gerald Mayr's (2013) approach. Accordingly passerines are a relatively young group.

http://www.tandfonline.com/doi/abs/10.1080/14772000.2013.765521#.UYQhULXxqk4

Exactly. There is evidence that passerines now occupy niches formerly held by non-passerines. This suggests a relatively recent diversification. The human similarity judgment data suggests that passerines are prototypical birds, but harder to discriminate than non-passerines.

Fossils:The passerine fossil record is fragmentary, misleading and controversial, but if this must be dragged into the argument, one cannot avoid the neglected Protoavis. Chatterjee failed to document the relationships of these bone fragments he dissected from a Triassic nodule, thinking at first that he was dealing with a garden-variety reptilian nodule. He has been ostracized for that, but the heterocoelous vertebrae and the delicate, brainy skull are diagnostically avian, despite afew teeth. Albeit one of the most contentious avian fossils,  it is by far the oldest and is close to the passerine form in many ways, such as its small size, delicacy and large brain.

Fossils and DNA:  Gerald Mayr's work with Per Ericson and other colleagues on the reconciliation of the avian fossil record with DNA covers an impressive range of taxa and provides strong support for Alan Feduccia's proposal of a rapid radiation of ~70 odd families of modern birds at the K-T boundary. Those same data are fairly mute about the passerine question, although I note that the deepest divergence between any of the families was between lanid and tyrranid families, which could hardly be regarded as "ancient" passerine lineages and might have been expected to have diverged much more recently in the crown passerine formulation. 

Gondwanan  Passerines: Biogeographic data on the really ancient " Gondwanan " passerines shows that they were present deep in the Cretaceous, unlike the vast majority of modern bird families in the more than 70 shown in the studies just mentioned, all of which  first appeared soon after the end of the Cretaceous.

Brain Marker of the Tertiary Explosion of Modern Birds:     All of these families in the modern radiation of birds share a unique neurological marker that is not found in any any other vertebrate, nor in passerines, nor in a few outlying avian groups with links to the Cretaceous (such as rails, cranes and the ratite descendants of lithornithids).  Jeff Woodbury found no evidence that this complex brain trait ever reverses. If one member of a family in the modern radiation has this doubled, schizocerate, condition in the spinal cord, so will every genus and species in the family. This stability, without homoplasy, far excels any other character known in avian evolution, particularly DNA bases, which can switch back and forth. Although  the neurological schizocerate/leiocerate split does not correspond to any well-accepted avian categories, it does mirror the post  K-T explosion proposed by Feduccia and is further supported by the fossil+DNA study by Mayr and colleagues. Inter alia it can also illuminate a few minor conundrums in avian classification, such as the hoatzin and the bustard, which belong to the modern schizocerate radiation despite some confusing characters placing them elsewhere.

Future Resolution of the Passerine Problem:

It may be possible to use other tools beside the sparse fossils to help anchor the molecular phylogeny of birds, with its unresolved issues of mutation rate versus difference in the passerines and  "galloanserines"(to name only two such issues). The extremely conservative evolution of the brain and increasing sophistication in  biogeography could each contribute to a tightening of conventional DNA reconstructions like the methodology using fossils

To reply to James' suggestion that the recency of the passerines is supported by evidence that they have replaced non-passerines:- 

It is true that the passerine radiation of more than 5000 species is extraordinary, but I am not aware of evidence that this has been achieved by replacing non-passerine species, which are characterized by a radiation that is well-defined in time (Tertiary) and is remarkable for its expansion rather than its reductions. 

The argument about similarity between passerines is also problematical if it is based upon your experiments with naive observers and bird skins. One of the prominent features of passerines is their behavioral specialisations, including song, which can convincingly differentiate passerine species that are convergently similar in external appearance."Similarity" is a generalization with the implication that the 5000+ are not good species. It ignores differentiation that goes far beyond external appearance alone.

The question whether passerines form the crown group or branch off more deeply sounds like a usual tree rooting problem that can be solved by inclusion of more and/or better outgroup taxa. If done, the presence or loss of plesiomorph (molecular, morphological, behavioural or other) characteristics should enligthen the order of branching deep down in the avian tree. So arguably DNA sequence information of extant birds alone is not sufficient to solve the problem.

There is also the more general problem whether 'time' as a cause of differences in diversity has any explanatory value at all. More time also means more time for processes that reduce diversity.

Outgroups Can be Part of the Problem: The use of an outgroup is another method for "anchoring" or "rooting" a phylogeny that is badly needed in birds because of the small number of useful substitutions in the stem of the bush-like Tertiary radiation and the extremely heterogeneous rate variations.  The outgroup  uses the same DNA language as rest of the data, and so is presently convenient. This is in contrast to the use of fossil dates or neurological blades or tectonics etc, which are only in the process,  like Mr. Bayes, of becoming formal  phylogenetic reconstruction.

       The problems with the outgroup method for reconstructing a phylogeny are serious because the appropriate choice of the outgroup is commonly circular. It often presupposes that one already knows much about the phylogeny! To put this another way, it is common for the choice of the outgroup to be completely inappropriate and so confound the resulting phylogeny. Let me give an example of this which one can find repeatedly in the published literature:- if one chooses an outgroup that is in fact part of the input, the order of branching in the tree may be inverted compared to the actual tree revealed by later comprehensive analysis. This is because the  derived characters in the inappropriate outgroup will drag homologous derived characters, in the otherwise-crown taxa, toward the base of the tree where the false outgroup's position has forced, by definition. This is a particular problem in cases where the number of informative substitutions are small and the phylogeny is controversial, as in birds.

Putting an Outgroup into the Ingroup:

It can also work the other way, with a basal taxon that has good outgroup qualifications being included with the ingroup. To show how this can distort the analysis, I have chosen an example that uses one of the real anchoring characters I have been talking about, to make it easier to follow the argument. Real characters are a bit easier to follow than it is possible with the noisy flurry of changes in substitutions that characterizes avian phylogenetic reconstructions with DNA.

Gary W. Kaiser has dismissed the phylogenetic value of the neurological, schizocerate, trait that unites 4000+ modern non-passerines into the short-stemmed bush of Feduccia's Tertiary radiation,  ........ but excludes the passerines and a few scattered groups like paleognaths, cranes and rails. An uncomplicated view of all the evidence  is that the lack of this derived trait in passerines is shared with all other vertebrates. This would make passerines a suitable plesiomorphic candidate for an outgroup of birds, as some studies practice. Kaiser's thesis  is that passerines belong to the crown group but have lost the neurological character found in most non-passerines. The flaw in Kaiser's argument is that one cannot have one's cake and eat it. If passerines belong to the crown group, as Kaiser contends,  reversion of the neurological character would have had to have occurred on thousands of separate occasions. This would have astronomical odds, particularly when one considers the extraordinary stability of the derived neurological schizocerate trait, which has never been observed to reverse in any species from the 4000+ Tertiary radiation of non-passerines. It is therefore hard to escape the conclusion that passerines originated very early in avian phylogeny and that their many branches are the result of a long period of speciation, just as quantitative studies of the steady branching pattern imply.

Enormous Rate Variation Causes Artefacts in Avian Phylogeny:

There are other problems with outgroups, such as the problem of dealing with the huge variation in mutation rates across avian families, like that seen in the "galloanserines",  an artificial pair that shares little with the real world of birds and exists mainly in the minds of DNA algorithms  . The high rate of both galliforms and anseriforms tends to separate them from the Tertiary radiation of non-passerines, where they belong, into a more  basal position that tempts their inappropriate  use as an outgroup. Without some anchoring strategy for normalising and ordering these big variations in rate, distances are distorted and the phylogeny corrupted, as in the in-group/outgroup examples above. The standard "relative rate test" does not work well in birds because of the extreme rate heterogeneity. Where does one start, when any avian lineage could be slower or faster than any other?

Anchor Needed for Avian DNA:

In view of the fact that the outgroup is more likely than not to cause problems, as illustrated, I think that it is entirely appropriate to consider that the use of DNA sequence alone for the passerine problem is circular.

I think that the evidence shows that avian DNA requires an independent source of anchoring information, an opinion  suggested by the way I framed the original question about the origin of passerines.

Perhaps I should make my meaning plainer: If you have doubts about the principal relationships of major (extant) avian lineages (are ratites or passerines the group that branches off before all others?) and want to rely only on DNA sequence data, the outgroup to root the tree has to be non-avian (crocs, turtles, lepidosaurs) to avoid the problems you described above . But perhaps the relationships of these groups to birds is too distant to help solve the problem.

If you are willing to use osteological data in a phylogenetic analysis the problem of the tree root becomes somewhat smaller because there are numerous fossil representatives closer to birds than to crocodiles that can be included.

That's why I was suggesting to use fossil data. You can include them in different ways: (1) use first occurrences in the fossil record as time calibration points for molecular-data-based phylogenies, (2) use the order of first occurrences to distinguish among several equally parsimonious trees those with a good stratigraphic fit (more likely) from those with a bad fit (less likely), (3) use morphological data of fossil and extant taxa in addition or as alternative to molecular data. Indirectly (1) and (2) also include morphological data (to determine whether some fossil forms part of an extant group or not).

One idea of a strategy to solve your problem (passerines as an early- or late-diverging group?): (1) Compute an unrooted tree of extant birds on the basis of DNA sequence data. (2) Use morphological data and/or the order of first occurences to find the tree root. (I guess you will probably get a conventional tree scheme with ratites branching off early and passerines nested high up within a neoavian clade.)

Non-avian Outgroup:

There is no doubt about the appropriateness of the outgroup in this case, but this route is not often taken because a very large DNA sequence is required to ensure enough linking characters between the birds and these distant non-avian sequences.

There are a number of published studies that used this approach. Two of the three studies that I could find showed a very clear result with passerines at the root of all birds, even the paleognaths, which often occupy that place in avian phylogenies. The brain gives exactly the same topology. I note that passerines will occupy the same root position on the basis of the neurological character that it shares with non-avians but not with

The third study also gave a similar result, with passerines branching off before the paleognaths and then all other non-passerine birds, but was “marred” by an early branch by the “Galloanserines” , an anomalous and probably artificial assemblage to which I have drawn already drawn attention. There is always a problem distinguishing the number of DNA substitutions caused by a very high mutation rate from the number of substitutions that have resulted from a long period of evolution (both are “phylogenetic distance”), but in the case of the “Galloanserines” we can be fairly sure that their similar great phylogenetic distance from other birds is a product of an anomalous, high mutation rate because of relative rate tests using a variety of DNA markers. Moreover, there are none of the other shared derived characters expected between galliforms and anseriforms, beside the common high mutation rate itself.

Michael’s suggestion is excellent for the rooting of unrooted DNA phylogenetics with fossil, osteological and other data, such as neurological characters.

On the Question being discussed, it is worth emphasising that this informal approach already provides strong support for the basal position of passerines.

Unfortunately workers in the field do not all share the same high degrees of unaided deductive abilities, making a formal, quantitative method, like PAUP or Mr. Bayes, better able to convince sceptics. A move toward methods that can formally incorporate the different non-DNA data such as osteology and neurology into DNA sequence matrix would have two beneficial results:- 1. Powerful new data can be brought to bear on important problems, like the passerine problem discussed here, that is not even recognised as a problem by many ornithologists; 2. Successful outcomes of the new formal hybrid algorithms might help rescue other data sets from obscurity and help to counterbalance unthinking hegemony of DNA.

" All of these families in the modern radiation of birds share a unique neurological marker that is not found in any any other vertebrate, nor in passerines, nor in a few outlying avian groups with links to the Cretaceous (such as rails, cranes and the ratite descendants of lithornithids)."

Jack, I think that this is the most impressive evidence of the relative antiquity of the passerines and it is news to me. Thank you. However it suggests that the passerines have been extraordinarily successful for their tree of resemblances to have escaped pruning through extinction as has seemed to affect other groups. I will have to think about this.

I never meant to imply that passerines were not good species (although this can happen, witness the identification of more species of orchids than of composites because they are prettier and have attracted passionate OCD splitters) and I also acknowledge that a great deal of information is lost when you kill the birds, wrench their guts out, and turn them into the feathered tubes that are bird study skins. However, these as stimuli are still superior to photographs, drawings, etc. in allowing humans to discern the pattern of similarity observed by expert ornithologists.

I will dredge up the reference to the niche replacement evidence when I am not distracted by the necessity of writing a final exam for 100 students by tomorrow.

Good work Jack. You have generated an interesting discussion.

I have Attached a reading list for those desiring some more substantial sources.

The link above does not work, so I have re-attached the reading list.

A Reading List on the “Ancient Passerine” Proposal:

1. Schizocerate (“split horn”) finding of a doubled spinal cord dorsal horn in many modern bird families, but not in passerines, paleognaths, rails cranes and pigeons, which all share the leiocerate (normal spinal cord with one dorsal horn) condition found in all other amniotes.

Woodbury, Jeffery. C. 1998 Two spinal cords in birds: novel insights into early avian evolution. 
Proc. R. Soc. Lond. B 265: 1721-1729, doi: 10.1098/rspb.0494

2. Molecular studies showing that passerines are found at the root of the avian tree, with paleognaths. All the “cherry-picked” studies below agree on this, except iii, where passerines are still basal, but are usurped from the earliest branch by the Galloanserines, an artificial assemblage of two otherwise unrelated groups of birds (galliforms and anseriforms) that appear to be ancient only by virtue of shared, extraordinarily high, mutation rate.

i. MINDELL, D. P., M. D. SORENSON, D. E. DIMCHEFF, M. HASEGAWA, J. C. AST, AND T. YURI. 1999. Interor- dinal relationships of birds and other reptiles based on whole mitochondrial genomes. Syst. Biol. 48:138– 152. The basal position of passerines in this study was not generally accepted at first; “long branch attraction” was thought to be responsible for the heterodox phylogeny of passerines.

ii. JOHNSON KP (2001) Taxon Sampling and the Phylogenetic Position of Passeriformes: Evidence from 916 Avian Cytochrome b Sequences. Syst. Biol. 50(1):128–136 This study has a very large data set compared with i. and convincingly despatches the idea that “long branch attraction” accounts for the basal position of passerines.

iii. Kerryn E. Slack a,b, Frédéric Delsuc a,1, Patricia A. Mclenachan a, Ulfur Arnason b, David Penny (2007) Molecular Phylogenetics and Evolution 42: 1–13 Resolving the root of the avian mitogenomic tree by breaking up long branches. These authors are expert at correcting for biases in DNA that can influence phylogenetic reconstruction such as base composition and taxon sampling, but they have not been completely successful at correcting for large, circular, variations in mutation rate. For this reason, the basal position of passerines is confirmed, even lying outside the paleognaths, but the Galloanserines, with a very large rate anomaly, intrude (see 2 above).

iv. Slack KE, Jankea A, Penny D., Arnason U. (2003) Two new avian mitochondrial genomes (penguin and goose) and a summary of bird and reptile mitogenomic features. Gene 302 (2003) 43–52 In this study, Galloanserines are more appropriately placed than iii, but are still outside the paleognaths, outside of which are the passerines, in keeping with the present thesis. A nice feature of this study is the use of non-avian amniotes as outgroups. Passerines prove to be the closest sister group to the non-avian amniotes in their cladogram.