A bony-crested Jurassic dinosaur with evidence of iridescent plumage highlights complexity in early paravian evolution

The Jurassic Yanliao theropods have offered rare glimpses of the early paravian evolution and particularly of bird origins, but, with the exception of the bizarre scansoriopterygids, they have shown similar skeletal and integumentary morphologies. Here we report a distinctive new Yanliao theropod species bearing prominent lacrimal crests, bony ornaments previously known from more basal theropods. It shows longer arm and leg feathers than Anchiornis and tail feathers with asymmetrical vanes forming a tail surface area even larger than that in Archaeopteryx. Nanostructures, interpreted as melanosomes, are morphologically similar to organized, platelet-shaped organelles that produce bright iridescent colours in extant birds. The new species indicates the presence of bony ornaments, feather colour and flight-related features consistent with proposed rapid character evolution and significant diversity in signalling and locomotor strategies near bird origins.

other feathered dinosaurs. The authors document for the first time in the fossil record the presence of platelet-shaped melanosomes.
This study is particularly significant among theropod and bird palaeontologists as it adds a new taxon close to the root of bird origins. In particular, this new taxon differs from all other Jurassic paravians in its combination of features (but see comment below on Pedopenna). The dentition (and serration pattern) is unique among Jurassic paravians, which has interesting implications on the discussion on the palaeocological context of bird origins. This was briefly mentioned by the authors, who have focused on plumage and ornamentation (see my comment below): the palaeocological implications of this new dental pattern among Jurassic paravians should be further discussed as they may result particular interesting among the broader palaeontological and biological community.
I have three main questions on this manuscript that would appreciate that the authors evaluate and discuss: 1-Anatomical description: My main concern relates to the claimed presence of a bony crest in this taxon. This is particularly relevant because the authors name the new species according to this feature, include it in the diagnosis of the taxon (and even mention it in the title of the manuscript). The authors report that the specimen bears the "lacrimal with [a] prominent dorsolaterally oriented crest", a feature previously unreported among Jurassic paravians, and usually absent among maniraptoran theropods. Lacrimal crests are widely distributed among basal (non-maniraptoran) theropods, but in paravians (as the authors note) lacrimal ornamentation is limited to lateral shelves along the anterior ramus, overhanging the antorbital fossa. A dorsolaterally oriented crest is thus an unusual (and unexpected) feature among these close relatives of birds. Although I have no a priori objections against a lacrimal crest in a basal paravian, I suspect this is a preservational artifact, and would like the authors to more carefully discuss this and add further evidence that it is a genuine biological feature. This is necessary given that, as I noted above, this feature is so relevant in the diagnosis (and in the species name) of the taxon. Before listing my reasons for asking a careful discussion of the lacrimal crest, I remark that even if this crest turns out to be a preservational artifact, the validity of this new species is not affected: the specimen shows several unique features in the premaxilla, maxillary pneumatisation, dentition, limb proportions that strongly support its status as a new valid species. Thus, the presence (or absence) of a bony crest is, in my opinion, not significant for the value and significance of this manuscript. But this absence/presence must be clearly and unambiguously supported in the manuscript in its actual form. My suspect that the bony projection in the lacrimal is an artifact is based on these lines of evidence (see also Fig. S5): -the orbital and postorbital parts of the skull are clearly dorsoventrally compressed: the palatine is dorsally displaced inside the antorbital fenestra, the postorbital broadly overlaps the jugal, the tip of the ascending ramus of jugal reaches the dorsal margin of the infratemporal fenestra. The skull appears quite long and low, differing from the more triangular outline in other basal paravians: given the displacement of the above mentioned bones, this shape is clearly preservational. This means that we cannot exclude that some unusual elements like the lacrimal crest are preservational artifacts due to deformation and displacement of bones.
-the element that the authors consider as the posterior ramus of the lacrimal might be the anterior end of the frontal. This element forms an acute corner with the ventral lacrimal ramus: this is unusual when compared to other basal paravians, where the posterodorsal ramus of lacrimal forms an obtuse corner with the rest of the lacrimal, and usually forms a broad arch of the large orbital fenestra. Such narrow anterodorsal corner of the orbit is clearly a taphonomic artifact, which indicates that the relative positions of the elements in that part of the skull do not show the original shape. Unfortunately, the Fig. S5 is not clear, but I note that the element claimed to be the posterodorsal ramus of lacrimal is aligned to the displaced frontal bone, with a large crack crossing between them: I suspect that the two elements are just parts of the same bone (the frontal): this is consistent with the larger contribution to the orbit made by the frontal among all other paravians.
In conclusion, I suspect that the "lacrimal crest" is simply the dorsal end of the preorbital ramus of the lacrimal as in all other paravians, and that no dorsolateral crest was present in this taxon. Accordingly, I suggest to not name the species according to this problematic element, and to remove this feature from the diagnosis of the species (and from the title of the manuscript). I may be wrong, I have based my arguments exclusively on the images and information provided by the authors: if they consider the lacrimal crest as a genuine feature, I would appreciate much additional evidence for supporting this. Therefore, my main request to the authors is to provide additional anatomical description and close-up images of the lacrimal (on both slabs and with different light angles) because this element is a pivotal feature in their diagnosis of the new species (and is even mentioned in both species name and manuscript title!). As I noted above, even in absence of a bony crest this is a significant and interesting specimen: its removal from the diagnosis and description is in my opinion a relatively minor change to the manuscript.
2-Reproducibility of the results: I replicated their main phylogenetic analysis (based on Xu et al. 2015), using the data matrix provided, and confirm the topology and tree statistics they discuss. Unfortunately, I was not able to replicate the second analysis (based on Brusatte et al. 2014). The authors stated that the analysis produced 100000 shortest trees of 3514 steps, where Caihong is found as sister taxon of Xiaotingia among Anchiornithinae. My re-anaysis, using the scores provided by the authors for Caihong and using the data set of Brusatte et al. (2014) resulted in 100000 shortest trees of 3399 steps long (much shorter than the value obtained by the authors): in the strict consensus tree, Caihong results among Anchiornithinae but in a not-resolved polytomy. I have tested different settings (in particular, setting all characters as non-additive, as in Xu et al. 2015), but have not being able to recover the results reported by the authors. Please, provide a complete matrix for the second analysis (and the character setting) to allow replication of their topology.
A note on the two phylogenetic analyses used. Xu et al. (2015) and Brusatte et al. (2014) used comparable sets of character statements, but differ each other in the use of additive (ordered) characters (not used by Xu et al., explicitly used by Brusatte et al.). This means that the two analyses follow different interpretations of the state transitions for the same characters. I encourage the authors to follow only one of the two interpretations, as they contradict each other in the homology assumptions. [Note that in TNT, character numeration starts from 0, thus -for example -"character 15" in the above string is "character 16" in the list of Xu et al. 2015] The result of this alternative analysis obtained a slightly different topology than the one discussed by the authors: Anchiornithinae results the sister-taxon of Dromaeosauridae+Troodontidae. This example shows how a priori assumptions on character transition settings may produce different topologies of the same data matrix. I encourage the authors to clearly state the hypotheses followed in their phylogenetic analyses, and to consistently follow the same setting on both analyses.

3-Comparison with Pedopenna:
Pedopenna is based on a fragmentary specimen preserving exclusively the tibiotarsus and pes. The authors stated that Caihong differs from Pedopenna in the relatively shorter first metatarsal (<15% of metatarsus length, compared to 25% in Pedopenna) and in "extensively feathered pedal phalanges". According to Xu and Zhang (2005: A new maniraptoran dinosaur from China with long feathers on the metatarsus. Naturwissenschaften 10.1007/s00114-004-0604-y), the first metatarsal of Pedopenna is not complete, but the preserved element is <10% of metatarsal III: this ratio is comparable to the condition reported in Caihong, and differ from other basal paravians. If the authors have new information on Pedopenna, this should be explicitly stated, otherwise, this feature actually cannot differentiate Caihong from Pedopenna. The second difference may be preservational, as documented among the different specimens of Anchiornis. So, apparently there are no actual features differentiating Caihong from Pedopenna. Nevertheless, Pedopenna is diagnosed by a very slender pedal phalanx I-1 (length/midshaft diameter ratio about 7.2): please, provide information on this element in Caihong since it may indicate whether it is distinct from Pedopenna.
Give the possible synonymy with Pedopenna, I encourage the authors to include Pedopenna in the data set of Xu et al. (2015) to test this hypothesis. Although Pedopenna is already included in the analysis of Brusatte et al. (2014, where it results a scansoriopterygid), the latter data set is not designed to test basal paravian relationships (it mostly focuses on basal coelurosaurs): in overall features Pedopenna is clearly an Anchiornis-grade paravian, as is implicitly assumed also by the authors of this study who included Pedopenna in the differential diagnosis of Caihong.
In conclusion, this is a well-written manuscript describing a new and interesting taxon of paravian theropod. The amount of modification to this manuscript from its actual form before final acceptance is relatively minor, and I see no reasons for not accepting it for publication once the above listed points have been discussed and implemented.

Andrea Cau Geological Museum "G. Capellini", Bologna (Italy)
We have revised the ms based on the referees' helpful comments. Most notably, we: 1. Moved the description section into the main article 2. Added more description of the lacrimal and particularly of the lacrimal crest 3. Revised the diagnosis 4. Added new phylogenetic analyses and revised relevant discussions 5. Added some discussions to, and made our assumptions clearer in, the melanosome section.
We thank the referees for their constructive comments that have greatly improved the ms. We have toned down all conclusions from the microstructures. We have done a truly massive study with novel statistical approaches and assessment of morphospaces of fossils in the same units. We have added extensive new data on extant birds as well as now run 3 different phylogenetic analyses at the request of other referees.
While we absolutely appreciate the immense value of a separate chemical study in the referee's (2) desired framework, and would be happy to collaborate with them to do this in the future, we stand with the value of the extensive work that we have done in its own right-although with major changes to wording to address concerns with identification of the structures.
Responses to the referees:

Reviewer #1 (Remarks to the Author):
The authors describe a new taxon of very bird-like dinosaur from northeastern China, which they call Caihong juji. Many feathered dinosaurs have come out of China recently, and show no signs of abating. But this one stands out and is clearly an important find. It provides the earliest evidence of flattened melanosomes in dinosaurs (associated with iridescence and variable/vivid hues in modern birds), the earliest possible evidence for an alula, and has a noteworthy combination of a fairly short arm, a forearm proportionally longer than the humerus, and a huge tail with long asymmetrical feathers. This is something a novel bauplan among early birds and their closest dinosaur relatives. It demonstrates that the new fossil is indeed a new taxon, and also provides more evidence of incredible diversity in flight and feather structures around the origin of flight. Strangely, the new animal also has a fairly large 'horn' on its lacrimal, a cranial display structure previously unknown in very bird-like dinosaurs. It is further evidence that the fossil is a distinct taxon, and also that these dinosaurs on the cusp of becoming birds were using both their bones and their feathers for display. Allin-all, this is an important find that further clarifies (or, in fact, muddies a little-but that is ok!) our understanding of the origin of birds and flight. It deserves a high profile publication.
I am pleased to see this paper resubmitted. I reviewed it for another journal and gave it a very good review and recommended publication, and was surprised that it was not published there. The authors took into account all comments in my original review, so I little else to recommend. This paper can essentially be published as-is. However, I will make one final suggestion: Nature Communications allows slightly longer papers than Nature and Science, but the manuscript is still written to the very short length of these journals. I suggest that the authors move some of the morphological description from the supplement into the main paper. That will ensure that readers see it. Also, the skull line drawing would work well in the main text.
We thank the referee for his positive comments. Following his suggestion, we have moved the morphological description and several illustrations including the skull line drawing into the main article.

Reviewer #2 (Remarks to the Author):
By tradition, the paleontological community is more forgiving towards 'liberal' interpretations than are scientists of other disciplines, particularly within natural sciences. However, we still need to base our conclusions on some sort of data.
We understand the referee's concerns with some practices of the paleotntological community. However, we have collected large amounts of data for this paper from both the fossil and from extant birds, including hundreds of new SEM images (fossil and new extant taxa) and thousands of measurements. We have then analyzed these data using rigorous statistical analyses that are not typically seen in the paleontological literature. So to say that our interpretations need "some sort of data" to back them up does not seem to be an entirely accurate appreciation of our work.
Hu et al. make a lot of claims pertaining to ancient colors and their role in sexual signaling; however, these are largely unsupported because they are based on the following (untested) assumptions: One cannot do palaeontology (or science in general) without making assumptions. However, as long as these assumptions are reasonable, backed by data or logic and are clearly stated, this is acceptable and should not prevent rigorous work from moving forward. In particular, when an assumption has been supported by tests from numerous independent groups (including by those who have been most skeptical of it), it is reasonable to partially rely on these results and not repeat them for every new case. Science works by building on previous findings, not endlessly repeating them. Otherwise, we would have a lot expense for very little benefit. In this case, we argue that our assumptions are backed by data and logic (see details below), but agree that in some cases we did not clearly state them.
We tone down our claims throughout by making our assumptions clear, addressing the referees concerns and cutting some text.  Schweitzer et al. 2015;Vinther 2016). Likewise, the interpretation of microbodies and their imprints in feathered dinosaurs is not a straightforward process (Lindgren et al. 2015a). Nonetheless, means exist to distinguish remnant melanosomes from other types of microbodies, such as e.g. microorganismal cells (e.g. Glass et al. 2012;Lindgren et al. 2015b We agree that our wording was perhaps too strong and we have now modified the main text to make clear these are microstructures, identified by us as melanosomes, that that is our assumption. Although this is certainly an assumption, it is based on data, the morphological similarity of these microstructures to modern melanosomes (see In summary, we appreciate the chemical work suggested by the referee but argue that, given 1) the morphological similarity of these microbodies to melanosomes, 2) previous morphological and chemical evidence in support of this interpretation and 3) the lack of support for any alternative interpretation, they are not needed in this case.
However, throughout the paper, we now make it clear that this is an assumption, refer to them as "structures" and generally tone down our wording.

If indeed fossilized pigment organelles, then there are no contributions from internal melanosomes.
It is assumed that all microbodies are epidermal rather than internal; however, internal melanosomes are abundant in modern animals. Moreover, they can fossilize (McNamara et al. 2014(McNamara et al. , 2016 and thus may be present within the body outline/cavity of Caihong. Means exist to chemically characterize the matrix in which the microbodies are embedded (e.g. Pan et al. 2016). Why has not this been done?
We argue that this is a safe assumption because we sample exclusively from the feathers of this fossil. It is extremely unlikely that melanosomes migrated from the internal organs through the body and into the feathers while demarcating the fine scale structure of the feather. In the unlikely event that internal melanosomes migrated outside of the body, we would expect them to randomly aggregate into a "halo" or pools (which are not seen in this fossil), and not infiltrate the feathers. Furthermore, although data on shape and size of melanosomes in internal organs are limited, we know of no evidence that any of them are flattened as are the ones we report here. We now note this in the main text.

The reconstructed coloration of Caihong is based on the assumption that, other than melanosomes, no other structures and/or pigments were originally present in the plumage of PMoL-B00175; however, in modern birds the coloration is the result of a variety of factors, including diet, keratin structure and coexpressed biochromes (e.g. Jawor & Breitwisch 2003). Why are these uncertainties not mentioned?
We appreciate this point, and indeed have addressed it in previous work (Li et al. 2012). However, we do not need to make this assumption here, as we do not attempt to assign specific colors to this fossil. Rather, our focus in this portion of the paper is on the presence of the unusual melanosome morphologies that are similar to those seen in some extant iridescent feathers. These indicate that the feathers likely had some form of iridescence, but the precise color is unknowable at this point, as we note.

Two of the authors (MDS and LD) have recently written a comprehensive review paper on mixed structural and pigmentary colors (Shawkey and D'Alba 2017)
, and found no example of an iridescent feather containing any additional pigments. Moreover, the presence of an additional pigment would not change the fact that these unusual morphologies are present in the fossil.

The flattened microbodies all represent platelet-shaped melanosomes.
Regarding the 'platelet-shaped melanosomes': The entire fossil is flattened as a result of compaction, and thus it is very likely that the same diagenetic process has affected also those microstructures preserved within it. Normally, preservation is highly variable, even at the microscale. For instance, this can be seen in a specimen of Anchiornis where differently preserved microstructures occur immediately adjacent to one another (see Lindgren et al. 2015a, fig. 2a This is an interesting point that we previously addressed this point in the supplement. Other microstructures from the same specimen and from other specimens from the same locality with the same geologic history show no compaction or flattening. Thus we do not consider this likely. Even if true, such flattening would not alter the 2D structure of the melanosomes that is also consistent with the 2D structure of flattened melanosomes from extant birds. Figure S12). My suggestion is that the authors instead try transmission electron microscopy (TEM), which retains the structural integrity of the microbodies (when embedded in epoxy resin).

We thank the referee for this suggestion. Since (1) we do not argue for hollowness, (2) hollowness, if present, would only make our results (affinities with hummingbird style iridescence) stronger and (3) hummingbird hollowness is clearly visible in SEM (see figure)
, these additional analyses would not affect our core results and are outside the purview of the work.

Reviewer #3 (Remarks to the Author):
The authors describe a new, well-preserved specimen of bird-like theropod dinosaur that preserves a significant part of the feather body covering. They also infer part of the plumage color using a methodology previously used (by some of the authors and others) for other feathered dinosaurs. The authors document for the first time in the fossil record the presence of platelet-shaped melanosomes.

This study is particularly significant among theropod and bird palaeontologists as it adds a new taxon close to the root of bird origins. In particular, this new taxon differs from all other Jurassic paravians in its combination of features (but see comment below on Pedopenna). The dentition (and serration pattern) is unique among Jurassic paravians, which has interesting implications on the discussion on the palaeocological context of bird origins. This was briefly mentioned by the authors, who have focused on plumage and ornamentation (see my comment below): the palaeocological implications of this new dental pattern among Jurassic paravians should be further discussed as they may result particular interesting among the broader palaeontological and biological community.
I have three main questions on this manuscript that would appreciate that the authors evaluate and discuss: 1-Anatomical description: My main concern relates to the claimed presence of a bony crest in this taxon. This is particularly relevant because the authors name the new species according to this feature, include it in the diagnosis of the taxon (and even mention it in the title of the manuscript). The authors report that the specimen bears the "lacrimal with [a] prominent dorsolaterally oriented crest", a feature previously unreported among Jurassic paravians, and usually absent among maniraptoran theropods. Lacrimal crests are widely distributed among basal (non-maniraptoran) theropods, but in paravians (as the authors note) lacrimal ornamentation is limited to lateral shelves along the anterior ramus, overhanging the antorbital fossa. A dorsolaterally oriented crest is thus an unusual (and unexpected) feature among these close relatives of birds.
Although I have no a priori objections against a lacrimal crest in a basal paravian, I suspect this is a preservational artifact, and would like the authors to more carefully discuss this and add further evidence that it is a genuine biological feature. This is necessary given that, as I noted above, this feature is so relevant in the diagnosis (and in the species name) of the taxon. Before listing my reasons for asking a careful discussion of the lacrimal crest, I remark that even if this crest turns out to be a preservational artifact, the validity of this new species is not affected: the specimen shows several unique features in the premaxilla, maxillary pneumatisation, dentition, limb proportions that strongly support its status as a new valid species. Thus, the presence (or absence) of a bony crest is, in my opinion, not significant for the value and significance of this manuscript. But this absence/presence must be clearly and unambiguously supported in the manuscript in its actual form. My suspect that the bony projection in the lacrimal is an artifact is based on these lines of evidence (see also Fig. S5): -the orbital and postorbital parts of the skull are clearly dorsoventrally compressed: the palatine is dorsally displaced inside the antorbital fenestra, the postorbital broadly overlaps the jugal, the tip of the ascending ramus of jugal reaches the dorsal margin of the infratemporal fenestra. The skull appears quite long and low, differing from the more triangular outline in other basal paravians: given the displacement of the above mentioned bones, this shape is clearly preservational. This means that we cannot exclude that some unusual elements like the lacrimal crest are preservational artifacts due to deformation and displacement of bones.
-the element that the authors consider as the posterior ramus of the lacrimal might be the anterior end of the frontal. This element forms an acute corner with the ventral lacrimal ramus: this is unusual when compared to other basal paravians, where the posterodorsal ramus of lacrimal forms an obtuse corner with the rest of the lacrimal, and usually forms a broad arch of the large orbital fenestra. Such narrow anterodorsal corner of the orbit is clearly a taphonomic artifact, which indicates that the relative positions of the elements in that part of the skull do not show the original shape. Unfortunately, the Fig.  S5

As I noted above, even in absence of a bony crest this is a significant and interesting specimen: its removal from the diagnosis and description is in my opinion a relatively minor change to the manuscript.
While we respect the referee's opinion, we do have strong evidence for the presence of prominent lacrimal crests. Indeed, it may difficult to understand the cranial morphology just based on illustrations, in particular given that the skull and mandible expose in an oblique way (ventrolaterally rather than laterally) and some cranial elements are displaced from their original anatomical positions and display many breakages. However, a close examination suggests that a prominent lacrimal crest is present anterodorsal to the left orbit (and presumably another one to the right orbit). The left lacrimal is tetra-radiated element (rather than tri-radiated as in most paravians except Austroraptor: besides the anterior, posterior, and descending processes, there is an additional process at the junction area of the three processes which extends laterally first and then curves dorsally. This process also has an expanded, horn-like shape, which is unlike any other lacrimal process. The posterior process has a gradual, continuous transition to the lacrimal crest, though the angle between the two processes is relatively sharp. Beneath the posterior process, there is a separate bone which represents the anterior portion of the frontal (sorry that in the previous line-drawing, we did not illustrate this, which we believe is the reason that the referee considers the possibility of the posterior process being a part of the frontal). We agree with the referee that the narrow anterodorsal corner of the orbit is a taphonoimc artifact, resulting from multiple factors, including the deformations and slight displacement of the bone and visual effect by the ventrolateral exposure of the skull.
Nevertheless, in the revised version, we provide more detailed description of the lacrimal morphology, and particularly of the lacrimal crest morphology, and we also revised the linedrawing to more accurately reflect the shape of the lacrimal crest and its relationships to the neighboring bones. We appreciate the referee's comments which make the description much clearer and we hope that the revised version will satisfy the referee. [Note that in TNT, character numeration starts from 0, thus -for example -"character 15" in the above string is "character 16" in the list of Xu et al. 2015] The result of this alternative analysis obtained a slightly different topology than the one discussed by the authors: Anchiornithinae results the sister-taxon of Dromaeosauridae+Troodontidae. This example shows how a priori assumptions on character transition settings may produce different topologies of the same data matrix. I encourage the authors to clearly state the hypotheses followed in their phylogenetic analyses, and to consistently follow the same setting on both analyses.
We thank the referee for the comments, and following the referee's suggestion, we have provided a complete matrix for the second analysis. also following the referee's suggestion, we ran the third analysis on Brusttate et al 2014 matrix with some multistate characters ordered, and we have amended the text to discuss the result of unordering/ordering these characters.
As shown in the revised version of the submission, our analyses produced basically the same results as our previous version. However, with some characters ordered, the analysis did produce unresolved phylogenetic relationships among a monophyletic anchiornithinae. Although the referee prefers a strategy of using ordered characters, there is no real therotical basis favoring ordered characters rather than unordered characters. So called nest homologies can be arbitrarily determined as in practice even the same working group may choose different multistate characters to be ordered (e.g., see different studies by the AMNH group); furthermore, developmentally, some assumed nest homologies can be produced by separately without a transitional state.
Nevertheless, we have provided results from analyses with both ordered or unordered characters, and the readers can make a judgement what results are more realistic. Finally, the results from different analyses have no effect on the conclusions drew in our paper.