Deep-Time Demographic Inference Suggests Ecological Release as Driver of Neoavian Adaptive Radiation

Round 1

Reviewer 1 Report

This intriguing paper seeks to investigate, using genetic considerations, if basal neoavians experienced an "ecological release" following the K-Pg event,  in response to ecological opportunity measured by density compensation ---specifically -- whether lineages experienced ecological release in response to opportunity, evidenced by density compensation. They attempt to estimate population size (Ne) in ancient lineages using intels and gene trees. They find that some divergences near the extinction event involved unexpectedly high gene tree discordance relative to the estimated number of generations between speciation events---a most interesting discovery. Ne estimates revealed a pronounced spike near the K-Pg boundary. This paper thus provides a putative foundation for theoretical considerations for studying population dynamics and rapid evoluton following extinction. While this approach is in its infancy, the paper nevertheless provides a provocative potential method for gaining critical population information from the haze of deep time. 

 

A substantial contribution which should receive considerable attention. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Review for Deep-time Demographic Inference Suggests Ecological Release as a Driver of Neoavian Adaptive Radiation

 

In the present work, Houde, Braun and Zhou attempt to evaluate a problem which has plagued evolutionary ornithology for decades— what were the dynamics of the early avian lineages surviving the K-Pg mass extinction? In order to provide some insight into this question, the authors investigate the history of effective population size (Ne) to develop a line of evidence about adaptive radiation and potential ecological release. The authors rely on recently generated whole-genome scale datasets for birds, and in particular focus on patterns of variation in indels.

As I was reading through the introduction and methods discussed in this paper, I thought of all sorts of questions and concerns that I hoped the authors would eventually address— but these concerns were ultimately unwarranted — I’m not sure if I have ever read a paper that so thoroughly discusses and addresses the assumptions implicit in its analysis, thereby embracing potential criticisms, rather than trying to redirect or hide from them.

It is through this comprehensive adjudication of assumptions that I think the paper shines as an example of how one can artfully combine theory and empirical approaches in phylogenetics, even if when results are not totally definitive. Indeed, there is, and may always be much uncertainty here. Ultimately, I found the authors case to be compelling, appropriately qualified, and consistent with the idea that ecological release may have followed the K-Pg mass extinction for at least some lineages. In general, I appreciated the technical approach, and I was able to follow the authors’ narrative and justifications, even though I do not feel totally qualified to evaluate their approach from the perspective of theoretical population genetics.

I think one point to mention which may be helpful for the authors has to do with the initial presentation of the conceptual links between adaptive radiation and changes in population size. As I note below, I was initially a bit unsure of the logic connecting Ne to adaptive radiation as presented in the introduction, around line 135-137. But as I noted above, these ideas are expounded upon at great length in the discussion — and the authors’ meaning was clarified for me by the time I had gotten to those sections. Nevertheless, I think it would be helpful for the authors to consider moving some of the discussion that outlines that specific logic to the introduction.

More generally, I am curious about the implications that these results may have for understanding broader patterns of molecular evolution. For example, is the detection of a spike in Ne compatible with the idea that population sizes of lineages which actually survived the K-Pg may have been transiently and dramatically restricted? I think that scenario might imply an acceleration of the molecular clock (restriction in Ne), followed by a dramatic slowdown (increase in Ne), in some loci. Perhaps some limited speculation on this topic would be warranted.

Otherwise, my suggestions are fairly minor and mostly editorial:

Line 36: change “is” to “were”

Line 135-137: The logic connecting Ne to adaptive radiation is not clear to me at this point in the text. What is the connection between Ne and the ‘adaptive’ quality of adaptive radiation? Opportunity for increases in abundance implies resources allowing for abundance became available— but it doesn’t seem to require anything adaptive. If this is explained in the examples the authors cite, I recommend including a brief summary of that specific logic in their text.

Line 147: something wrong with “some many”

Line 156: As above -- “ecological release” does not seem synonymous with adaptive radiation

Line 158: swap “concatenated” and “2515”

Line 173: typo: “bape”

Line 218: Typo: “in indicates”

Line 227: Typo: “even the wt25”

Line 233-237: Not clear to me what these Ne estimates reflect— are these across particular internodes? I’m not sure this is the clearest way to present these results — a Tree figure would probably be clearer.

Like 586: typo: “thru”

Line 616-619: important caveats about inferring adaptive radiation from evidence of ecological release — I suggest moving some of these details to the introduction (lines 135-137).

Line 622-624: Cite supporting literature for Neoaves < Kpg

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

This study represents a big step forward in resolving the mystery of the early diversification of modern birds. Modern birds, and Neoaves in particular, show an striking pattern of very rapid diversification around KPg times. The leading hypothesis postulates that birds experience ecological release from competitors and predators and the resultant ecological opportunity triggered a rapid diversification event. Studies so far have been centered on the temporal coincidence of the rapid diversification event with the KPg extinction event, and secondarily on the magnitude of diversification rate estimates. The present study addresses a crucially important dimension: the demographics of the event. This is a very interesting study that represent a key contribution to the study of the early diversification of modern birds.

Having said that, I do have several comments and suggestions. Firs of all, the theoretical framework needs to be revised a bit. It is true that ecological opportunity is a classic element of the ecological theory of adaptive radiation (sensu Simpson and Schluter). However, ecological opportunity is not a sufficient elements for producing an adaptive radiation. They key tenet of the ecological theory of adaptive radiation is that ecological opportunity increases ecological speciation, and thus diversification (proliferation of lineages in multiple adaptive zones). In contrast, the current formulation of the present study refers to the effect of ecological release on population size, and this is a much more general phenomenon, predicted by basic ecological theory (niche theory and even neutral theory: with the extinction of a great portion of individual and species in a metacommunity, the surviving species will increase in population size as they take advantage of the available resources). A key aspect of this distinction, is that an increase in population size will decrease the probability of extinction of a lineage (again, a basic prediction from neutral ecological theory), and this suggests an alternative explanation for the explosion of early Neoaves: the increase in net diversification rates may be due, not to an increase in speciation rates but to a decrease in extinction rates due to exceptionally large population sizes. This contrasts with the ecological theory of adaptive radiation that postulates that the increase in diversification is produced by an increase in speciation rates (see Schluter). Therefore, I think that the theoretical framework should be expanded to incorporate these elements. I would posit that the results are fully consistent with the ecological neutral theory, not the ecological theory of adaptive radiation (although not inconsistent with it, a key element is missing: to prove that speciation rates increased).

 

Other comments:

The study would be much easier to read and understand if you concentrate on a chosen scheme (e.g. long indels and MP-EST topology) instead of cluttering the results and the discussion with all variations and combinations of analyses. Once the superiority of the long indels and MP-EST topology was established, perform the estimates of Ne using those only. Ne estimates for other indel sizes and tree topologies (and other variations) can be published as supplemental material.

Figures can be improved. Now they look like screen captures from an Excel preadsheet.

Line 59. Explain what are “neoavian birds.”

Line 67. Citing Oliveros et al. 2019 seems a stretch as they do not show anything about the neoavian radiation or a K-Pg effect.

Line 69 “Indeed, modern timetrees indicate that early divergences among lineages of modern orders and supraordinal clades appear to have occurred in rapid succession in as little as <1 Ma.” This figure is inaccurate. Taking the most recent common ancestor of Neoaves as a reference, the rapid diversification that occurred afterwards spanned about 10 million years for supraordinal groups, and including the origin of only a handful of the oldest orders. This is also evident in Table 2 of the results.

81 “For example, one could argue that Neoaves is actually characterized by a number of “ghost lineages” that extend into the Cretaceous.” Actually, most studies cited in the previous paragraph show that the rapid diversification of Neoaves started in the late Cretaceous, not after the K-Pg event.

123 What is the difference between “rapid” and “explosive”? Is “explosive” just “more rapid” or it is “simultaneous”?

143 Is “that” needed?

147 “some” or “many”?

147 Leave results to the Results section.

158 Correct : “..up to concatenated 2515…”

218 Delete “in”.

253 Most of the paragraph is repeating data that is in the table. Make a better effort to summarize the significance of the data instead of describing the data in a different way.

I don’t see the point of Tables 3 and 4. The data in them is not central to the main result. Can be published as supplemental material.

Figure 2. The scale of the Y axis is strange. It goes from 0 to 5e+6 in the first mark but then stays in the order of e+7 for the rest of the plot. Figs. 3 and 4 show similar anomalies. Issues with representing vastly different Ne values can be solved by using log scale. Figs. 3 and 4 should be in supplemental material.

277 What is “CBL”?

Table 5 can be transferred to supplemental.

Table 6. Delete

322 Spell out ILS.

345 Italicize “Ne”.

368 What do you mean by “uncertain to boot”?

369 What is CBL?

370 Estimates are done on branches or on nodes?

393 I did not understand the sentence: “In both cases, all one needs to make their data consistent with a radiation near the K-Pg boundary is invoke is a bias toward more ancient divergence time estimates.” Also, the rest of the paragraph is awkward in that it tries to argue that the chosen time frame seems correct because makes the Ne spike coincide with the K-Pg boundary. Sounds like circular reasoning.

407-420 The entire paragraph, which contains general statements about time-tree estimation seems superfluous and distracting. First, no one needs “a perfect fossil record” and “absolute precision” to make progress in avian macroevolution: we can use statistical methods to estimate required parameters. Second, divergence times ARE identifiable once informative calibrations are used together with appropriate Bayesian or maximum likelihood models; no need to create a straw-man here. Third, you mention improvements in amount of sequence data and models but forget a big piece: the quantity and quality of calibration fossils and how they are used to calibrate the tree. Finally, an easy way to take into account uncertainty in the ages of clades used in this study would be to incorporate the 95% highest posterior densities of the ages (from Jarvis et al) as whiskers in the figures.

437 The issue of species tree vs. gene tree divergence in estimating divergence time has bee published and discussed multiple times by now (e.g. Brown et al. BMC Biology 2008, 6:6 for an avian example). At least some of those references should be cited here and Fig. 6 seems unnecessary. A more interesting discussion would be about an apparent conflicting expectation: larger Ne predicts greater coalescent times, potentially biasing estimates towards longer branch lengths in naïve analyses, but, instead, branch lengths have been estimated to be very short for the basal neoavian radiation.  Explaining this apparent contradiction seem relevant.

457 Can you say “nearly homoplasy-free” instead of  “measurably more parsimony-consistent”?

474 Spell out QS or remind the reader what is QS. Use real words, “nucleotides”, instead of “nt” or “nts”.

489 What are “TE” insertions? (presumably “transposable elements”).

507 Spell out CBL or explain.

543 What is “PSMC”?

563 Those for the Red-billed quelea are estimates of Ne or of census population size?

565 If Ne is affected by all those factors, it seems crucial to discuss how your own estimates represent changes in population size or changes in some of those other factors, in particular, selection and range expansion may be drastically different after the K-Pg extinction.

573 The message in this paragraph is not clear. How hybridization may play a role here? Hybridization can certainly increase gene-tree incongruence than may be confounded with incomplete lineage sorting.

591 “Hypothesized” or “estimated”?

596 “Only bee-eater+woodpecker and Passerimorphae fall within UQ of their respective indel size partitions of Ne estimates more recently than 55 MYBP on either the MP-EST* tree or the TENT. A single exception is the finch+zebrafinch branch, which is also in the UQ for the >100 bp indels size partition on the TENT. In contrast, preceding 66 MYBP on the MP-EST* timetree only Passerea falls within the UQ in all indel size partitions, but Columbea, Otidimorphae, and Passerea minus Otidimorphae are within the UQ in 1-3 indel size partitions.”

Delete. These sentences may be more appropriate in the results. In the discussion, concentrate on the main finding and avoid numerical results and analytical jargon.

620 “…it is notable that Neoaves did not yield unusually high Ne estimates,…”  Confusing use of Neoaves. In the previous sentences “Neoaves” refers to a clade but in this sentence, Neoaves refers to the ancestor whose Ne was estimated (the basal node of Neoaves).

689 “It is widely held that most clades arise as adaptive radiations…” This is hardly true, and if true, it would undermine the interpretation of the results of this study. Many maintain instead that diversification occurs mostly through vicariance that results in rapid lineage-proliferation of ecological equivalents (no ecologically-divergence divergence). Others maintain that adaptive radiations are exceptional cases, representing the extreme of a continuum of diversification and adaptation. Critically, if most clades arise by adaptive radiation, its pattern would be widely distributed across the avian tree and the unusual peak in Ne should be produced by a different process, not adaptive radiation.

721 “If it were possible to assess ancient population size in the basal lineages of Neoaves, then it might be possible to find evidence of density compensation associated with both ecological release and adaptive radiation.” I don’t understand. Isn’t this what the present study has investigated?

725 – 734 Delete this mini-review of methods for estimating ancient Ne. I don’t see the point of it and clutters the discussion of the interesting aspect.

738 The characteristics of ecological release and ecological opportunity have already been listed in paragraph starting in line 714. It sounds repetitive here.

743 Why only the extinction of competitors is mentioned? The extinction of predators may have also played a role.

750 “A subsequent reduction in Ne in subordinate branches is necessary in order to demonstrate that this is indeed a transient spike in Ne in any given lineage.” This is a trivial inferential aspect, what is more important is the biological process: presumably, as birds diversified, the initial unusual abundance was apportioned to the multiple descendant linages, producing a reduction in the population size of each descendant lineage. Also, with a general increase in diversity of al biotas, predators and competitors also increased.

753 “Foremost, ecological release and density compensation need not result in radiation [54,57].” But an absence of “radiation” does not explain the absence of fossils. A superabundant species has higher chances of being fossilized (fossilization chances should depend more on abundance than on species richness). Ideas in this paragraph are not completely clear or connected.

765 The section “Conclusions and Future Directions” in highly unbalanced. This may be the only section some readers will read. First, you need to abandon all jargon (nt, QS, and even indel and Ne) and use standard English words (e.g.  “insertions and deletions”, “effective population size”). Second, the methodological issues may be relevant, as this is the first time this approach is attempted, but the biological implications are equally (if not more) important. This is the first time that a demographic phenomenon is associated with the Neoaves explosion around the KPg. State those results clearly. The Columbimorphae peculiarity does not belong into this section. Concentrate on the main result, not the anomaly.

Author Response

Please see the attachment.

Author Response File: Author Response.docx