Stop-and-go mammals and “warm-blooded” dinosaurs

Ever since Thomas Huxley opined that dinosaurs and birds might be related, a debate has waxed and waned over whether dinosaurs were simply inflated, “cold-blooded” reptiles or active, “warm-blooded” creatures.  Tied into these metabolic “states” have been judgement calls as well: we mammals tend to have warmer feelings for endotherms than for ectotherms (pun very much intended).

Our best data for inferring dinosaur physiology comes to from studies of the microscopic structure of long bones in thin cross-sections, a science called histology.  Bone is a dynamic, living tissue that provides a record of the relative speed at which its tissues were laid down during growth and development.  Bone growth and development rely on a good blood supply, and in a typical long bone cross-section, one can see the conduits that once carried life-sustaining blood vessels.  These circular rings of bone are called Haversian canals.

The rate or speed at which bone tissues are laid down effects how they look in cross-section, an observation first made by the famous Italian researcher Amprino and known as Amprino’s Law.  When bone grows slowly, fewer blood vessels are necessary to sustain the growth, and consequently there are few Haversian canals.  In contrast, when bones grows quickly, many blood vessels are needed to supply fuel for the growth, and therefore Haversian canals abound.  Additionally, slow-growing bone has time to lay itself down in nice, concentric layers like the rings of a tree.  This type of bone is called lamellar because of its distinctive layering.  In contrast, fibrolamellar bone occurs when bone is being laid down quickly — somewhat like a mason throwing together a wall of bricks as fast as possible and in no particular order.  In other words, fibrolamellar bone can look a mess compared with lamellar bone.

Ectothermic vertebrates like an alligator tend to have lamellar bone with few Haversian canals, whereas endotherms such as a bird will show fibrolamellar bone pocked with many Haversian canals.  But there’s a wrinkle: in addition to these features, we will also see dark, concentric rings called Lines of Arrested Growth (LAGs) representing times when long bone growth temporarily ceased.  Traditionally, it was thought that slow growth and seasonal cessations of that growth produce LAGs, such as the seasonal warming and cooling of “reptiles” from Spring and Summer to Fall and Winter.  However, there have been sporadic reports of LAGs in otherwise fast-growing mammals and birds, and it has been suggested that changes in diet might play a role here.

And this is where a lot of controversy has come in: dinosaur long bones often show LAGs in section, yet are often fibrolamellar with many Haversian canals.  What gives?  If there is any consensus currently, it is apparent that all dinosaurs had some fibrolamellar bone, and most researchers recognize that this indicates a qualitatively rapid growth rate elevated at least somewhat above “reptiles.”  But what about those pesky LAGs?

Enter a new study by Kohler and colleagues published in Nature which shows that lines of arrested growth occur in numerous wild ruminants over a large range of climates.  That’s right — these are endothermic mammals, your cows and antelopes and other hoofed herbivores that have large, multichambered stomachs for processing grasses.  By examining the histological cross-sections of 115 femora from ruminants across Africa and Europe, Kohler and colleagues have shown many examples of LAGs in otherwise “good” endotherms.

What is going on here?  Seasons.  Kohler and colleagues matched the LAGs in their ruminant sample to average rainfall across the various biomes from which the ruminants lived.  What they discovered was that LAGs and slowed growth occurred during times when precipitation was low.  As you might expect, low rates of precipitation will effect plant biomass: fewer plants = fewer calories = less energy available for growth during the lean season.

LAGs in dinosaurs have often been pointed to as either definitive evidence of ectothermy or at least have raised flags concerning how well bone histology conserves information about growth and development.  This study by Kohler and colleagues pretty much ends that argument: LAGs are not associated with the underlying metabolic rate of the animal.  Instead, they reflect changes in bone growth due to the seasonal availability of food and water.

The authors actually say it best at the end of their paper:

The consistently seasonal formation of rest lines in homeothermic endotherms debunks the key argument from bone histology in support of dinosaur ectothermy. Our study instead suggests that the extensive vascularization of the fibrolamellar bone in most dinosaurs and other extinct vertebrates is tightly correlated with seasonal maxima of endogenous heat production, an association that should be explored in future studies. — Kohler et al. (2012), doi:10.1038/nature11264, p. 4

Is this “proof” that dinosaurs were endothermic?  Not exactly, but what is exciting about this new study is that it does indicate that what we are seeing in dinosaur long bone thin-sections is a realistic approximation of their growth rates and an indirect measure of their environments.

XROMM Days 3-5: Data, data, data

Again, if you’re just tuning in to this thread on XROMM, please refer to my previous posts on what XROMM is, why I am excited about using it, and what I have done so far, including 3-D animating a chewing pig.

The past few days have seen us busy and social.  We’ve now learned to do scientific rotoscoping, or the art of properly aligning three-dimensional, anatomically-accurate models of vertebrates with two X-ray movies shot from different angles.  I have previously described the basic steps involved in making these XROMM movies and coordinating and calibrating the X-ray movies, so I refer you to my previous post on this.

Unlike the method of XROMM I described previously that uses bone markers to synch the moving images and the bones, in scientific rotoscoping you manually align bones frame by frame with x-ray movies.  It can be tedious at times, but essentially it is, to quote Steve Gatesy, quoting Ken Dial, fitting a digital key (the bones) into a visual lock (the X-ray movies).  In other words, you are posing the skeleton in three-dimensional coordinates based on the constraints imposed by X-ray movies.  Put simply, if your digital skeleton deviates from the reality of the X-ray movies, you are doing something wrong.

What is perhaps most important to emphasize in all of this is that because you are fitting bones into a virtual reality space of what happened when the animal was filmed, you can now recover data about skeletal and joint movements.  That’s right: you can actually retrieve reliable, repeatably scientific data on gait and movement from these 3-D animations.  For example, you assign joint markers and then measure how much the mouth of a pig opens, or how much rotation is happening in a bird knee, or even how parts of a fish skull move in relation to the skeleton and muscles.

So, you are not just making a nice skeleton movie — you are recovering what would normally be unrecoverable data.  The sort of data that allows you to more objectively describe what happens in living vertebrates, and hopefully, data that can be used as a baseline constraint for limiting what fossil vertebrates may or may not have been capable of.

All in all, this has been a great experience.  I wish to thank Beth Brainerd, Steve Gatesy, David Baier, Ariel Camp, Sabine Moritz, and Erika Giblin for all their help, information, and assistance this week.  It has been a pleasure.

XROMM Day 2: This little piggy bites … in 3-D

Again, if you missed previous posts on XROMM, please read those first for better context of the discussion that follows: see what XROMM is all about, why I’m excited to be learning it, and what I’ve already done.

On our second day in the course, we took the next step to synch two different, simultaneous X-ray movies of a mini-pig eating with a three-dimensional model of its skull in the MAYA program.  These X-ray films were taken on the C-arm x-ray machines I mentioned in my previous post by Dr. Beth Brainerd and colleagues, and we were essentially learning by replicating their process.

Dr. B. and the Mini-Pig

Me and mini-pig’s skull. The original movie was filmed back in 2006, so the mini-pig that stars in the X-rays has since passed on to piggy heaven. Here I am holding the actual skull of the animal that you will see in the animation.

In XROMM, you essentially have a work-flow like this:

  1. Film an animal behavior from two directions using calibrated X-ray cameras.  The animal usually has tiny, spherical beads implanted surgically into a few of the bones of interest prior to the filming.  As an important note here, all such films and surgeries are done under strict animal welfare protocols and the animals are not harmed: the X-ray dosages are as low or lower than that of humans exposed to x-rays for diagnosis, and the beads are tiny and biologically inert.
  2. After filming the animal’s behavior, the skeleton’s three-dimensional geometry is often CT-scanned from the animal.
  3. The CT-scan bone data are converted into geometrically-accurate 3-D representations of the bones of interest.
  4. In MAYA and MATLAB, the films from the two x-ray cameras are virtually “projected,” and the beads implanted into the animal subject show up as little dots.  These dots show up as little spheres in the CT-scanned bones you import into the MAYA program.
  5. Things then get more technical, but suffice it to say that the beads you see in the X-ray films and the spheres in the CT bones are synched.  The movements of the beads (spheres) are digitized in three-dimensions calibrated from the two camera views, and then the virtual bones are “cemented” to these spheres.  Then, as the spheres move, the bones follow, and what you get is a three-dimensional reconstruction of the 2-D x-ray films!

Here is a screen shot of the animation made in XROMM today.

Mini-Pig X-ray and Reconstruction

What you get from XROMM: to the left is one of the two X-ray films, and to the right is the 3-D skull reconstruction of the mini-pig.

Here is an animation of what you get — I still can’t believe we can do this!

Dr. B. Contemplates the Mini-Pig

Dr. B. Contemplates the mini-pig skull.

And we’re not even done yet.  Stay tuned …

XROMM Day 1: Pig heads and C-arms

If you’re just tuning in, you may want to read my previous posts on what XROMM is and why I am thrilled to be learning this technique.

Today was a good, busy, productive day at the XROMM course.  Never let it be said that “simply” synching X-ray “movies” (cineradiography) is easy!  The students in the course are learning how to animate the three-dimensional jaw movements of a mini-pig, based on the research of Dr. Beth Brainerd and colleaguesThe XROMM site provides movies and animations of what we are attempting to duplicate.

You don’t simply take an X-ray movie and then transpose that into a 3-D animation, of course.  If you’re going to match up the three-dimensional models of the mini-pig skull to the chewing motions recorded on as cineradiographs, you have to get the animation program (MAYA) to synch with the frames of these movies.  And that involves a number of techniques including correcting for distortion (the tube that transmits the X-ray images to the camera has convex ends, which give the original footage a fish-eye lens distortion) and “registering” the simultaneous side and top or bottom views of the moving animal to virtual cameras and screens in MAYA.

In science, the tedious parts come from making absolutely sure you are doing everything to account for error and noise in your data.  This usually pays off with dividends in the end, but getting there is the hard work. Its not enough in this case to do all the technical things necessary simply to capture a moving animal’s skeleton in two planes.  Then you have to spend hours and days matching that raw data to your virtual skeleton — and all for a sequence of maybe 2-20 seconds in length.

Today we also were able to visit the two XROMM facilities at Brown.

Dr. Bonnan at XROMM facility

Dr. Bonnan in one of the two labs of the XROMM facility a Brown University. Note the two large, mobile X-ray machines to Dr. Bonnan’s left.

Physics came back to haunt me today, as part of doing the science of XROMM correctly and safely is putting knowledge of photons and radiomagnetic waves to good use. Physics was always a tough subject for me, but it is amazing what you can learn and apply when you really want to do something and when doing it incorrectly will result in long-term injuries from X-ray irradiation!

One of the rooms had mobile C-arm X-ray machines that demonstrate very well the basic concept of what you do when you capture the motions of an animal.

Small C-arms at XROMM

Two small C-arm X-ray machines in one of the XROMM labs, positioned perpendicular to one another. One is facing left, and the other is pointed away from you in this photo.

All of the XROMM faculty and staff have been wonderful and saintly in their patience with us as we learn a technique that is often as frustrating to learn as it is to explain many times over and over again.  A big thank you for their patience and help today … and I think my eyes are starting to uncross now.

To give you an idea of the time investment necessary to convert 2-D cineradiographs into 3-D moving models, consider this: we spent most of our time today simply registering points and synching virtual camera views … there has been no animation yet!

I see great potential for research and student involvement with XROMM, and I’m looking forward to having a chewing mini-pig skull in the next few days.

Stay tuned …

Seeing through vertebrates to see through time

While waiting in the airport for my last flight (long story) to Providence, RI, and on to Brown University for the XROMM course, I obtained a good WiFi signal and so I’m writing a brief post.


Tomorrow will mark the beginning of learning new cineradiography techniques and skeletal modeling that I have jealously been wanting to do for a long time.  It is hard to convey in words how anxious and interested I am to begin learning and then using the XROMM techniques.  Perhaps this is a bit of an exaggeration, so forgive the hyperbole, but I feel somewhat like a physicist who first get access to an atom smasher or an astronomer learning for the first time how to peer into the cosmos through some technologically marvelous telescope.

For someone like myself who is interested in how the skeleton actually behaves as a machine, and how to apply this new XROMM technology to deciphering past vertebrates like dinosaurs, this is coming close to time travel.  Okay, perhaps a bit of an exaggeration again, but I believe that seeing through live vertebrates to understand quantitatively how their skeletons “tick” is seeing through time.  Conserved movement and novel functions in living relatives of dinosaurs help us realistically constrain and predict what those long-dead animals were doing when they moved, hunted, or vacuumed-up vegetation.

I’ll be updating this blog throughout the week, and of course you can follow me on twitter for up-to-the-minute thoughts and comments.

How to see the skeleton in action … in 3-D

I am interested in dinosaur locomotion.  In particular, I am interested in how the various parts of a dinosaur skeleton moved in relation to one another, especially when capped in cartilage, actuated by muscles and tendons, and ensheathed in flesh.  Of course, until time travel becomes reality or until Jack Horner brings to life his “chickenosaurus,” knowing just how dinosaurs moved is fraught with difficulties and unknown variables.

Take, for example, the recent, independent confirmations by Casey Holliday and colleagues, and that of my own group’s research, on archosaur long bone articular cartilage thickness and shape.  It turns out, frustratingly, that a good amount of joint shape and thickness are lost to time in dinosaurs.  Part of my current research focus is to develop empirical methods for reconstructing the missing cartilage shape at the ends of dinosaur long bones, but that is a long work in progress.

However, you would think that at the very least, we should be able to quite readily determine the range of movements and interactions of the living skeleton in crocodylians and birds.  Strange as it may seem, understanding how the skeleton of a vertebrate moves in real life is not easy.  An actual, physical skeleton of any given animal can only be articulated in one pose at a time, and in any case the limitations or freedom of movement resulting from soft tissues are very difficult to establish — just ask Larry Witmer and his group of dedicated students.  One could, of course, pose a dead animal in various ways and then CT-scan or take a radiograph (an X-ray photograph) of each pose, but dead animals don’t always assume natural poses.

Another technique used for a number of analyses of skeletal movement has been fluoroscopy combined with the capture of X-ray movies of live animals, called cineradiography.  Several papers have been published on various aspects of vertebrate animal movement, breathing, and other activities using cineradiography.  Whereas these previous studies have been invaluable in revealing heretofore unknown or unanticipated movements of the vertebrate skeleton (e.g., the furcula of birds acts as a flexible spring), they are limited in that they are 2-D.  All bones of the animal are compressed into a flat plane, and you have to be able to recognize and follow the movements of the bones you are interested in while they merge and pass over all the other bones in the movie.

Therefore, I am thrilled to report that I have been accepted into a 1 week course at Brown University where myself and other faculty, postdocs, and graduate students will learn a new, 3-D cineradiographic technology that enables accurate three-dimesnional animations of vertebrate skeletons!  Called XROMM (X-ray Reconstruction of Moving Morphology), this new technology uses combined, multiple fluoroscopic and standard video sequences of a moving animal in combination with three-dimensional representations of its bones to produce a scientifically accurate moving skeletal model of particular behaviors and motions.  The three-dimensional bones are obtained using standard CT-scan or laser-scanning technology.  Check out the following examples under the Movies link on the XROMM site:

  • Mini-pig feeding
  • Ray-finned fish mouth mechanics
  • Iguana breathing and rib movements
  • Bird hindlimb movements
  • Duck feeding

The advantages of XROMM over traditional fluoroscopic movies or the attachment of external markers to skin is easy to appreciate.  For the first time, you can watch the movement of bones in three dimensions from any angle and in any perspective.  Moreover, you can easily export the ranges of movements generated by the animation for further quantification and analysis of how the bones actual move in space, in time, and in relation to one another.

I will be blogging and tweeting (@MattBonnan) about my experiences in this course, which will take place Monday, June 11 through Friday, June 15, 2012.  This sort of experience will not only open new doors for the research of all those involved in the course, but it will inform projects I will conduct in the future with my students.

I wish thank the members of the XROMM course for such an amazing opportunity:

  • Beth Brainerd
  • Steve Gatesy
  • Dave Baier
  • Ariel Camp
  • Sabine Moritz

It is wonderful to be a paleontologist in the 21st century!