# Review: Biomechanics of extreme neck elongation – a case study using the extinct reptile Tanystropheus

Michael P. Taylor <dino@miketaylor.org.uk>

<!-- md2toc -l 2 review.md -->
* [Overview](#overview)
* [Specifics by manuscript section](#specifics-by-manuscript-section)
    * [Introduction](#introduction)
    * [Materials and methods](#materials-and-methods)
    * [Results](#results)
    * [Discussion](#discussion)
    * [Conclusions](#conclusions)
    * [Acknowledgements](#acknowledgements)
    * [References](#references)
    * [Supplementary information](#supplementary-information)
* [Note on manuscript style](#note-on-manuscript-style)


## Overview

This is an important study on an important animal, which contains novel results. I welcome its submission. I am keen to see it published, but I'm afraid I can't recommend that it proceed in its current for for for three important reasons. I will summarise these in this overview, then go on to comment in more detail on the text and figures.

First, and most important, I am really uncomfortable that the model _Tanystropheus_ neck used for this study seems to have been based on bones alone and does not include articular cartilage. (Connective tissues were included in the FEA part of the study, as described in lines 193-207, but not the range-of-motion analysis). The authors recognise that this is not really good enough (see manuscript line 318-321, but proceed anyway, and I suspect it is largely for this reason that some of the results do not feel credible -- see below for details. In particular, I am sceptical about an arched neck in neutral posture, and about the lateral inflexibility of the neck compared with those of extant live animals.

There are two possible paths to take here. I think the manuscript could be brought up to the bar by including a candid discussion of the ways in which the absence of cartilage in the model has likely affected the results, and how this means that Figure 2 in particular cannot be taken at face value. But since that figure is the main thing that most readers are going to take away from this paper, I think it would be better if the authors redid the ROM analysis using a model with intervertebral cartilage inserted. Better still would be to run the ROM analysis both with and without intervertebral cartilage, and show the difference that it makes. I leave it to the authors to decide which of these paths to take, because I know how unwelcome it is when reviewers ask for new experiments.

The second major problem to my mind is the strong dorsal kink in the base of neck as shown in Figure 2 parts B0 and B1, which visually appears to be something like 80 degrees, when no more than five or six degrees is shown as possible for any other intervertebral joint. Can this be correct? Intuition suggests not, and the measurements shown in the first row (DV1/CV13) of Supplementary Table 2 (Values of the maximal rotation of the joints achieved in the ROM analysis) do not indicate anything resembling this kink. If this is an error, it should of course be corrected; if not, then the manuscript needs a detailed explanation of why this one join is so very flexible.

The third project, which seems to be a simple mistake, is the effect of ribs on ventroflexion. While Supplementary Table 2 shows that maximal rotation is much less when the cervical ribs are included, the grey and blue segments in Figure 2 say the opposite, suggesting the astonishing result that adding cervical ribs dramatically increases ventral flexibility.

One these three major issues are addressed, I will be pleased to see this paper published. The many more minor issues mentioned below should also be addressed, but are not significant enough to be a barrier to the paper going forward.


## Specifics by manuscript section


### Introduction

"13 vertebrae, some of which were more elongated than in any other animal". Can this be quantified? It's not immediately obvious that these vertebrae are more elongated that those of some azhdarchid pterosaurs or flamingoes.


### Materials and methods

It is more common for Institutional Abbreviations to come at the end of the Introduction than at the the starts of the Materials and Methods (unless this is a quirk of the journal's style).

"To differentiate the anterior, middle, and posterior subregions of the neck we followed Rytel et al. (2024a) and Rytel (2026)." It would be helpful to include here a very brief overview of their criteria for distinguishing the neck regions.

"The models of ribs and most postaxial vertebrae derive from various specimens of Tanystropheus conspicuus". I really think it's necessary to be more precise about this: which elements from which specimens?

"They are freely available at MorphoSource: https://www.morphosource.org/projects/000774422." I commend the authors for making their models available, but it's disappointing to see "Download restricted".

The explicitness of the Skeletal Reconstruction section is commndable. It would be worth adding a paragraph here, though, admitting to the uncertainly arising the the necessarily chimaeric model.

"Each cervical vertebra was positioned so that the articular surfaces of its postzygapophyses are aligned as parallel as possible to the corresponding surface of the prezygapophyses of the more posterior vertebra, and so that the overlap between them is maximised. The position of the vertebra was then further vertically adjusted, to make the corresponding articular surfaces of the centra face each other (to the maximal extent possible without the models intersecting)." It would be worth including an illustration here of the sequence of steps performed to articulate a pair of vertebrae according to these criteria. They are difficult to visualise.

"After assembling, the entire model was scaled to match the size of large T. conspicuus individuals (Rytel, 2026), from which the isolated elements originated, and measured 325 cm in total anteroposterior length." Does the 235 cm represent the total length of the individual, the length of its neck, or the length of the skull+neck+D1 model?

"Landmarks were added to mark the most anterodorsal,  posterodorsal, and posteroventral points on the centra along the neck. The position of the centre of rotation was set to the average location of the four mentioned landmarks placed on the corresponding articular surfaces." I don't understand this at all. It sounds like you out the joints at the midpoints of the centra, rather than at the intercentral articulations. I'm sure that can't be right. Another illustration (or another part to the other one I suggested adding) would help here.

"The atlas-axis complex and the skull were treated as a single rigid element." This is a reasonable simplification. It would be worth adding a note to the effect that, while there was a lot of flexibility in this area in life, it did not much affect the flexiblity of the neck specifically.

"Maximum ROM was tracked by summing the rotation values reached for the X (torsion), Y (lateral flexion), or Z (dorsoventral flexion) axes of each joint." This is confusing nomenclature, and X, Y and Z here do not correspond to the uses of these axes in the preceding paragraph.

"If the results were asymmetric, the higher maximum ROM value was used.". I don't understand what "results are asymmetric" means, especially as the model itself is symmetric by design.

"As the variation of the material properties did not result in any notable differences in bending behaviour ..." That seems impossible at face value. How can the mechanical properties not affect bending behaviour? Please explain.

Although there is some acknowledgement of "the simplifications inherent in the FEA models", I think more ought to be said about this. Bone, as a complex biological material, responds differently to loading in different directions, torsions, shears, etc., and it is important to recogtnise the limitations of applying a technique designed for homogenous materials such as metals and ceramics. It doesn't mean there is no value in doing the analysis, but it's critical that the reliability of the results is not oversold. This is especially true in the case of vertebrae, with their complex internal variations of density.

"Connective tissues were added in Blender between the centra of successive vertebrae". I assume this means cartilaginous intervertebral discs? If so, it may as well be made explicit.

"The cervical vertebrae were free to rotate and translate about all three axes (six degrees of freedom per vertebra)". I think that is three degrees, not six.

"In contrast to the FEAs for the individual rib models, we did not test for failure stress magnitudes but compared the stress distributions in the models with and without cervical ribs." This is a much more useful kind of result.

"Subsequently, the ribs were remeshed in Blender into a single object that preserved the outline of the ribs but functioned as a single entity for the biomechanical simulations." This means that the ability of the ribs to slide past each other was lost. That is almost certainly how they behaved in life, as the homologous longus colli ventralis tendons in the necks of birds do this. Having clumped the ribs into a single object does not make the results worthless, but this additional limitation should be clearly stated.


### Results

While I accect the ONP (_osteological_ neutral posture) recovered by this analysis, I am highly sceptical that the actual neutral posture, when taking into account intervertebral cartilage resembled it at all closely. Inserting a short cartilaginous disc in each join, of perhaps 3% the length of the preceding vertebra, would likely suffice to bring "neutral" posture to a straight line or something similar.

At a bare minimum, the manuscript should acknowledge this, and make it clear that Figure 2 is not even attempting to showing life posture. While I hate to ask authors to do additional work, I think the paper would be more useful and less misleading if cartilage was inserted, so that the illustration dorsiflexed and ventroflexed postures indicate something more like what could have been achieved in life. If the aithors choose to do this additional work, they should experiment with different cartilage thicknesses to find what most closely approximates a straight line as "neutral", rather than just guessing "3%" like I did.

"This arch-like reconstruction is further supported by the orientations of the articular surfaces of the centra (Supp. Tab. 3)." To my mind, supplementary table 3 tells the opposite story: two of the six vertebrae have an angulation difference of the opposite polarity to the other four, and that suggests variation in preservation. No strong conclusion can be warranted from such a small sample with such contradactions contained within it.

"The anterior surface of the centrum of the 13th cervical vertebra and the posterior surface of the centra of the 12th and 11th cervical vertebrae were oriented distinctly more dorsally than in the other specimens." I don't understand this at all. What other specimens? Only one model was made. Is this just a typo for "the other vertebrae"?

"The neck bends laterally in a gentle curve, up to a maximum angle of 73° with ribs and 74° without them (Fig. 2A)." Once more I am sceptical that this can be the correct result, given that the necks of every extant animal are more flexible than this. I suspect what we're seeing here is another consequence of omitting intervertebral cartilage and I urge the authors to see how this finding changes when a reasonable amount of cartilage is introduced.

"The maximum ventroflexion achieved by the neck is 130° when the ribs are included and 61° when the ribs are omitted." That makes no sense at all to me. The ribs should constrain bending, or at best have no effect. I can't imagine how their presence could make the neck _more_ flexible, let alone more than twice as flexible. If this is really correct, a great deal of discussion is warranted. The "Ventral" columns of Supplementary Table 2 strongly suggest this result is the wrong way around.

Figure 2A would be much more informative if it also showed the maximum lateral flexion _with_ cervical ribs -- perhaps in the different colour. (I know from the text that the two results are very similar, but _seeing_ them very similar is helpful.)

Similarly, figure 2B would benefit from showing maximum dorsiflexion both with and without cervical ribs (the caption doesn't say which of these B1 represents). Once more, it would be helpful to show the with-ribs and without-ribs postured in different colours.

In Figure 2B, posture B0 is very difficult to accept as ONP due to the extreme dorsiflexion in the most posterior intervertebral joint. Was the 1st dorsal vertebra maybe not held immobile as intended. If that is not the problem, then some explanation is needed for why one join is so very very much more flexible than all the others. Similar problems beset posture B1. The text says "These characteristics indicate the dorsal inclination of the base of the neck (Renesto, 2005; Rytel, 2026), which is also recovered within the model." If this is genuinely what the model finds, then we need an illustration showing in close-up the ONP articulations D1-C13 and C13-C12.

It took me a while to make sense of what the grey and blue arcs indicate -- the angle of the skull/atlas/axis unit. It would be worth spelling this out. (Also, as noted above, the result shown must surely be wrong: it suggests that the neck is _more_ flexible with cervical ribs than without. If this is really what the model shows, then it needs extensive discussion.)

I don't understand at all what Figure 2C is showing me.

In Supplementary Figure 3, the X axes are not labelled, nor are any numbers (or unit) specified. As a result, I really have no idea what I'm looking at. Also, the caption doesn't seem to belong to the figure: "areas marked with brighter colour (red) being a subject to more stress", but there aere no read areas. "White indicates that the region is experiencing stress beyond the current scale" but the background is white so such areas would not be visible

"Without ribs and with dorsally directed force the tensile stress is present across the ventral surfaces of the centrum and the zygapophyses, and the entire dorsal portions of  vertebrae are affected by compressive stress (Fig. 3). The opposite pattern can be observed for the ventrally directed force. With laterally directed force the lateral side of the vertebral column oriented towards it is affected by tensile stress, and the other lateral side by compressive stress (Fig. 3)." This is utterly unsurprising, even tautological. I understand that it's included primarily to set up what follows, but maybe add an "as expected" or similar.

It's difficult to see much in Figure 3 beyond the unsurprising result that in all cases, when tensile stresses are along one side of the neck, compressive stresses are along the other. It would be helpful to provide insets that zoom into places where more interesting results are observable. The exception is that the dramatic difference between the "with individual ribs" and "with rib bundles" parts show how unhelpful it is to make the simplification of bundling the ribs into a single element. A useful finding of the paper would be that this is an approach to avoid in future studies.

"The inclusion of the cervical ribs into the model affects  distribution of tensile and compressive stresses, but only when the ribs are bound with each other. With ribs treated as individual, non-intersecting elements, stress distribution is nearly identical to the model without ribs." To my mind this is the principal finding of the paper.

"With ribs modelled into bundles, this pattern changes, with a major portion of the stress being transferred from the middle vertebrae (and especially their articular regions) to ribs (Fig. 3)". The paper should note here that this is not what the situation would have been in life.


### Discussion

"Whereas the aim of this study is to provide general patterns, not definitive values of ROM, and differences in methodology make direct comparisons of the results less feasible ..." This disclaimer is admirable.

"With ribs included in the model, only the maximum ventroflexion value decreased substantially". Ribs as individua;l elements or as a bundle? This is crucial.

"Whereas the relatively small number of intervertebral articulations present in the neck of Tanystropheus makes this issue less consequential compared to animals with higher cervical vertebral counts ..." I don't agree with this at all. The relatively small number of cervical vertabrae merely means that the relative importance of each individual intervertebral joint's geometry is greater.a

"In life their ability to slide along each other could possibly be limited". What is the reason for thinking this? I don't know of any extant analogue in which the cervical ribs or homologous tendons do not slide past eaxch other.

"Nevertheless, a huge part of the maximum ROM, especially during ventroflexion, originates from the base of the neck". As noted above, this statements need a good deal more justification and explanation.

"Tanystropheus has proportionally the longest cervical ribs among all known vertebrates". Proprtional to what? Centrum length? Total body length?

"In the tensile member hypothesis the tensile forces were transferred instead, and the weight of the neck was reduced by shifting the cervical musculature towards the trunk, allowing for more neck flexibility". Not really more flexibiliy -- that is determined by other factors -- but a reduction in mass (and inertia) in the anterior neck due to the posterior displacement of the muscles.

"Even though in one of the variations of the model (Fig. 3, ventrally directed force) the ribs are shown as transferring compressive forces, this is biologically not plausible, as due to their hyperelongation they would be extremely prone to buckling when loaded in compression." And (I know I keep going on about this) the way the ribs would have slid past each other.

"Finally, the majority of the hyperelongate ribs terminate distally around the same section of the neck, ventrally to the posterior cervical vertebrae." This is a really useful observation!

"In all long-necked taxa with hyperelongated ribs the skull is relatively small, presumably due to the effects of gravity and drag." Ha! That had never occurred to me, but seems to be right. Nice.

"The necks of the former were likely much thicker, which would decrease drag during locomotion and feeding (Troelsen et al., 2019)". That is highly unintuitive. It might be worth very briefly summarising why a thicker neck produces less, rather than more, drag.

"This is also supported by taphonomic data – despite the rich fossil record of plesiosaurs, there is only one documented case of predatory targeting of their necks (Drumheller et al., 2026), which, in contrast, was documented for two specimens of Tanystropheus (Spiekman & Mujal, 2023). This indicates that the neck of the latter was much more slender." That is _a lot_ of conclusion to draw from a very very small data-set. "Indicates" should at least be downgraded to "suggests",

### Conclusions

"This is the first quantitative evidence supporting the tensile member hypothesis". Yes, it is. Thank you for this!


### Acknowledgements

"We wholeheartedly thank K. Przestrzelska, I. Janicki, and P. Bajdek for masterful specimen preparation." It's really nice to see preparators getting some of the recognition they deserve. Also field-workers mentioned further down the acknowledgements.


### References

Please note that I have not proof-read the references, nor checked that all referencesd or cited or that all citations have references. The authors should make these checks.


### Supplementary information

I urge the authors to merge the supplementary figures and tables into the main manuscript. They are just as more a core part of the paper as the other figures, and no purpose is served by separating them out into a second document. The effect on the overall length of the paper will not be major.


## Note on manuscript style

I am being picky here, but there are some places where the English is not incorrect, but reads a little clumsily. (One example: "These animals differed considerably in their cervical anatomy, with highly varying number and length of the individual vertebrae, skull size, and the presence of cervical ribs". I think that here "variable" would be more idiomatic than "varying" and "presence of cervical ribs" ought to "presence or absence". I will not highlight these individually, but invite the authors to re-read the manuscript.

It may also be worth inviting an LLM ("AI") to comment on the style. I would never allow an LLM to do rewrites: they produce stilted, characterless text and often introduce errors. But they can be good at pointing out places where there is room for improvement. For example, I gave the abstract to an LLM and asked it: "Point out any infelicities in the following text. Do not suggest rewrites: just explain what is clunky." Among the things it pointed out was:

> **“but also biomechanical challenges, such as flexibility, stability, lift, and inertia”**  
>  
> This is probably the most conspicuous conceptual infelicity in the opening. The items in the list are not entirely parallel in meaning. Flexibility, stability, and lift are properties or mechanical considerations, whereas inertia is a physical property of a body. Furthermore, it is not immediately clear whether these are challenges posed by the development of an elongated neck or biomechanical properties that must be accounted for in understanding its function.  
>  
> _Lift_ is also potentially ambiguous: it could refer to aerodynamic or hydrodynamic lift, or to the mechanical effort involved in supporting the neck.

I consider this helpful feedback. (There was also some _un_helpful feedback: the authors should definitely not make changes in response to everything an LLM points out! LLMs are much better at language than they are at palaeontology.)


