How did birds get their wings?

This was originally posted at:

How many fingers do you have? Hopefully, 5. Do you think that’s the normal condition for all animals? Do you think that’s air you’re breathing right now? … OK, so I watched the Matrix last night, but still, do you think all tetrapods (dudes with 4 feet, including you, and anything else with four flippers, wings, or feet) have 5 digits on each limb? Actually, there’s a pattern within tetrapods of limb reduction in various lineages – our earliest ancestors seemed to experiment with digit numbers and went a bit berserk by growing extra fingers from their fishy flippers.

Some early tetrapod limb bone patterns

Some early tetrapod limb bone patterns

One of the most interesting, and therefore controversial-enough-to-be-used-against-palaeontologistss-by-whacko-creationists-and-anti-evolutionists-who-probably-still-live-in-the-attic-with-their-mothers, aspects of this whole menagerie of tetrapods trying to decide how is best to give the finger, is the transition from dinosaurs to birds. In case you hadn’t realised, this is quite an important evolutionary jump, as to go from something that doesn’t fly to something that does, you have to do something quite special to your body rather than gluing bits of plasterboard on to your arms and flapping real hard. Of course, this can work if you’ve had enough beer, but birds don’t drink. As such, they had to develop some pretty cool anatomical modifications to learn how to fly.

The dinosaurian ancestors of birds, the tetanuran theropods (the big ones who like eating lawyers, according to the latest scientific information), had three clawed digits designed for grasping prey. For a long time, these were considered to be anatomically identical to the three digits you get on modern birds and their avian-line ancestors. This is called homology, and is what palaeontologists use along with sophisticated programs to determine the evolutionary relationships of organisms. In birds and dinosaurs, it wasn’t this simple though. It never is.

Different evolutionary hypotheses for the origins of  bird digits (source, click for larger)

Different evolutionary hypotheses for the origins of bird digits (source, click for larger)

Scientists are superbly anal. This goes far enough that they actually put numbers on almost everything they can. This includes fingers, and for once, actually came in useful. It turns out, that if you look at the anatomical similarities between the digits of birds and dinosaurs, something odd happens. Dinosaurs have what are known as digits I, II and III of a ‘perfect’ 5-digit hand (known as the Lateral Reduction Theory (LDR)). Birds, on the other hand (ha. ha.) have a formula of II, III, and IV (written as II-III-IV, known as the Bilateral Digit Reduction theory (BDR)), based on their individual anatomies and through looking at how they develop in the embryonic stage. The battle to reconcile this pattern, or refute the entire theory of evolution based on it, has lasted for slightly longer than it takes to push forward a sensible bill through supreme courts (about 200 years). The transitional theory is known as the ‘homeotic frame shift’, which is only worth including here as it sounds awesome, identified by the way in which certain bird genes express themselves during embryology.

Alternative hypothesis suggesting that digits 'fused' themselves to masquerade as others (source)

Alternative hypothesis suggesting that digits ‘fused’ themselves to masquerade as others (source)

In dinosaurs, you can track a reduction of digits IV and V through time. This is different to many other tetrapod lineages, where you can track the gradual loss of I and V, such as in turtles. So if you consider just the pretty fossils, then really there’s no problem. Birds match the dinosaurs toe for toe. Yeah, these crap puns aren’t going anywhere. There are some dinosaurs that break this rule though. Limusaurus is a minisaurus from the Late Jurassic of China (about 140 million years ago), and has a reduced digit I, which means it follows the bird version of BDR. Some early tetanurans also have the bird-like II-III-IV formula, so what is going on?

Limusaurus reconstruction (source)

Limusaurus reconstruction (source)

What this tells us to begin with, is that to get a proper killer story about evolution, you have to combine the fossil record with developmental biology, embryology, and just every other cool branch of science out there to understand properly.

Genes are cool. One in particular, known as sonic hedgehog, is a gene that expresses itself in cells and acts to control the development of digits, particularly that of the pinky, or digit V. You can use the expression of these genes to determine the difference between different digits that look the same morphologically, but are actually different structures – this happens quite a bit in biology, for example with cryptic species, or those which look identical but are actually totally different organisms! In mice, you can actually fiddle with the sonic hedgehog gene, and change their developments to lose various digits during growth, including the middle ones.

Another way of looking at these developmental patterns is at the bone itself. Bone begins growth as a condensation of pre-cartilage cells along the limb axis. Digit IV is typically the first to begin life in tetrapods, and therefore can act as a reference point for when the others start to pop up and give an idea as to which digit is which. Oddly, what analysis of this suggests is that birds have a I-II-III identity, completely different to that known from morphological and genetic data!

In mice again, digit III is the last condensation to form. According to Moore’s Law of developmental biology (specifically of evolutionary loss of structures), the last element to form during embryonic growth is the first to be lost in an evolutionary trajectory. If this were the case, it could imply that tetanurans actually have a I-II-IV pattern. If this were the case, it would go some way to aligning the apparently different bird and dinosaur signals.

So as it stands, there isn’t really a consensus at the present. Superficially, birds look like they have the II-III-IV patterns, but developmental and genetic data actually suggests this is more likely to be I-II-III. Tetanurans are considered to have II-III-IV or I-II-III, depending on how you interpret what it is birds have and under the criteria of parsimony (the solution with the fewest ‘steps’ is the most likely). What is needed in the future is a re-examination of the tetanuran fossil record in the light of new genetic and developmental data, and of course, more fossils!

 Further reading:


Fossils – So Much More Than Just Pretty Dead Things

What do fossils tell us? It’s an obvious question, commonly phrased as ‘What is the point in studying fossils?’, but often can be one of the more difficult ones to answer objectively. The most prominent reason, that I’m sure a lot of people will agree with, is that we want to know what ancient and often extinct organisms looked like. This promise of discovering the unknown is what captivates people from a young age, and often motivates them in to studying fossils as a profession. From fossils, we can infer ecological aspects such as behavioural interactions, feeding strategies, and predator-prey relationships, and how these factors all changed through time. Tracking and reconstructing the co-evolution of the Earth and its biota is one of the most magical and beautiful stories ever to be told.

However, fossils can provide so much more than just aesthetic pleasure. If this wasn’t so, it would make grant proposals incredibly difficult – people don’t usually like giving away money just so a fanatic can play with fossils all day. So, palaeontologists have developed numerous excuses to satisfy funding bodies, to show that studying fossils actually has some scientific value.

The real motive is going “WT*expletive deleted*????” Sheer curiosity. Then, this develops into seeing that there was a *expletive deleted*load of stuff that was WAY different – how did this work? What does this tell us about how stuff works today? Mallison (2011)

Following are additional reasons why the study of fossils is not only awesome, but also indispensible in our understanding of biological and geological evolution.

Lineage Reconstruction

This is perhaps the most important use that fossils have for evolutionary biologists and palaeontologists. While genetic analysis might tell you about the particular history of a gene or genome, or the genetic evolution of species or populations (there are key fundamental differences between gene-trees and species-trees, something which molecular systematists miss out A LOT), they tell you virtually nothing about the phenotypic, or morphological evolution within a lineage. We don’t have many fossilised genetic markers (except in exceptional circumstances from permafrost-preserved mammoths), and thus must default to morphological analysis when tracking lineage evolution. While methods do exist for estimating and modelling the temporal evolution of species with respect to their genetic make-up, these can never provide such solid evidence as fossils can in terms of reconstructing ancient organisms, and the evolutionary trajectories leading to what we see surrounding us today.

The next two points are largely based on cladistic methodology. For a nice summary of cladistics, it’s worth quickly checking the following Wikipedia entry here. Essentially, cladistic analysis is the primary method for reconstructing cladograms, or trees, that represent the systematic and hierarchical classification of organisms. Note, that cladograms are not to be confused with phylogenetic trees, in which explicit evolutionary trends are inferred (i.e., patterns of ancestor-descendant relationships).

Novel Extinct Morphologies

Cladistics is based on the analysis of characters, which are formally broken down into character states. A character is essentially an aspect of morphology which can be expressed as a number of mutually exclusive variables, or character states. This forms the basis for analysis of species’ relationships and homology assessment. An example of how this can be expressed is:

Maxilla, anterior process, length: shorter (0) or longer (1) than the posterior process (taken from Sereno, 2007)

Now, if you want to reconstruct the phylogeny of any extant group with extinct members using just living members of the group and using just morphology, then you would directly neglect the unique character combinations that fossil species exhibit. This is important because, as a general rule (there are exceptions) the more characters included in a cladistic analysis, the greater the resolution achieved. Fossils can also provide transitional morphologies between species and additional information in areas of low resolution, and therefore resolved relationships are more evolutionarily stringent. Missing out the morphological information contained within fossils constitutes a severe case of neglect, and also disregards one of the most important aspects of any evolutionary analysis: time.

Character Polarity

As shown above, characters are broken down into various character states representing variations of a particular aspect of morphology. One of the main goals of cladistic analysis is to resolve the sequence of evolutionary transformation of these particular character states. If we increase the complexity slightly to include three variables, the character becomes known as ‘multi-state’. Keeping in line with the example shown above, one possible character is:

Maxilla, anterior process, length: shorter (0), identical (1) or longer (2) than the posterior process

Note that this is a purely hypothetical example to illustrate the point. To ‘transform’ from one of these character states to an adjacent one (i.e., 0<->1 or 1<->2) it costs one ‘step’ with the implication that it costs more to transform from 0<->2, and must pass through a transitional stage, character state 1. This is known as character ordering, and represents the directionless sequence of evolutionary transformation. However, what we want to know is the direction of character state transformation, to tell if a particular character state is the derived (apomorphic) or primitive (plesiomorphic) condition. This is achieved by polarising characters, and is where fossils play their part. As fossils are explicitly related in terms of chronostratigraphic age, this can automatically impose an evolutionary trajectory on character state polarity (i.e., the older fossils have the plesiomorphic state). This can also be achieved by ‘rooting’ a cladogram through outgroup assignment, which is an a priori determination of the plesiomorphic conditions through fossils; this is actually explained quite nicely here. The main point is that fossils perform a critical role in inferring sequences of phenotypic evolution.

Sampling Diversity

Now, one thing I’m sure palaeontologists are tired of hearing over and over is that the fossil record is biased in numerous ways (i.e., regarding sampling biases). Numerous studies have recently been undertaken to overcome these apparent biases, the most recent and critical of which is Hannisdal and Peters (2011). This paper explains how many of the patterns of fossil diversity we observe during the Phanerozoic can be explained by covariation between ancient biotas, sedimentation rates, and Earth system dynamics (e.g., ocean redox). Thus fossils, and the way in which we interpret them, are proving to be influential in how we interpret the co-evolution of, for example, biochemical and tectonic patterns, and contiguous biota assemblages.

The fact remains that, yes, the fossil record is biased. But now we can compensate for and use it to nurture our understanding of geological processes in deep time. On the other hand, we have molecular systematists who consistently use the excuse of the ‘incomplete and biased’ nature of the fossil record to completely disregard the use of fossils, and assume that DNA-based analyses are adequate. This is actually pretty ironic, considering extant organisms (i.e., those we can extract DNA from) represent a single time slice containing a fraction of the total species that have existed on the Earth since life began, and is therefore the most biased sample of all. Hypocrisy, thy name is deoxyribonucleic acid. A recent example of this is Ericson (2011), in which fossils are neglected from the study entirely (with only a brief mention), thus compromising the accuracy of all results obtained (making inferences about Mesozoic palaeobiogeographic patterns without consulting the fossil record is pretty offensive). So, analysing and incorporating fossils into diversity analyses actually decreases relative sampling bias, and increases the empirical and theoretical validity of studies. Ignoring the fossil record for a biogeographical, phylogenetic, or any other evolutionary study is counter-productive, and pretty much blasphemy.

Breaking of Long Branches

Long branch attraction is a fairly common side-effect of genetic-based phylogenetic analysis, typically occurring in when invoking parsimony. It arises as the result of highly rapid divergence between multiple lineages, and due to the limitations of nucleotide substitution (i.e., four possible character states) can lead to misinterpretation of homoplastic sites (e.g., through reversals, parallelisms, or convergences of states) as homologous (orthologous) sites. This can lead to erroneous inferences about the evolutionary (i.e., topological) distances between lineages. Although using advanced modelling methods such as Maximum Likelihood or Bayesian analysis can partially resolve this issue with genetic data, fossils can also be used to ‘break up’ long branches by calibration against a particular lineage in deeper time (specifically in morphological analysis), or by providing information in areas of limited information, ultimately improving phylogenetic accuracy. This is another example of the limitations of molecular-based analyses, with analogous issues in morphological analysis being quite well understood and resolved (see Cobbett et al., 2007 for a nice discussion about including fossils in cladistic analysis).

Calibrating Molecular Phylogenies

Molecular phylogenies are becoming increasingly used to estimate divergence times of major clades and as the basis for assessing temporal dynamics in within- and between-group diversification. However, using site-substitution rates alone to estimate the temporal origin of a clade (i.e., a node) is a poor estimation, regardless of the complexity of the models employed. Therefore, fossils with a strongly supported or well-defined taxonomic status can be used to calibrate the minimum origins of a particular clade, in a strict spatio-temporal context. This bypasses several assumptions made by models, such as stochastic or constant rates of site substitution, and is therefore an invaluable tool for accurately reconstructing phylogenies. Accordingly, the integration of ‘metadata’ (such as stratigraphy, or relative or absolute ages) is essential in reconstructing accurate phylogenetic relationships. It can also reveal additional crucial factors, such as the rates of phenotypic evolution, and how particular functional characters or morphological domains covary through geological time.

The above examples are just several of the more significant reasons why studying fossils is crucial, and how upon further critical analysis can yield unparalleled detail about the evolutionary history of life on Earth. It is worth noting that, although there are drawbacks and advantages to studying either the fossil record or the genetic evolution of extant taxa, it is when both are integrated that a more complete picture of global evolution emerges. Fossils are prominent in this reconstruction based on the unequivocal increased accuracy gained, but possibly at the cost of decreased resolution, due to the incomplete, patchy and biased nature of the fossil record.

To finish, I’ll quickly mention the concept of uniformitarianism: “the present is the key to the past”. This may be, in many cases of natural processes, but the past is also key to unlocking how it is the present transpired, and furthermore, in predicting future patterns of biotic diversification. To neglect the fossil record is to discard the one solid piece of evidence that we have in understanding global biotic responses to very real scenarios such as global warming.


Further reading

Butler, R. J., Benson, R. B. J., Carrano, M. T., Mannion, P. D. and Upchurch, P. (2011) Sea level, dinosaur diversity and sampling biases: investigating the ‘common cause’ hypothesis in the terrestrial realm, Proceedings of the Royal Society, Biological Sciences, 278, 1165-1170

Cobbett, A., Wilkinson, M. and Wills, M. A. (2007) Fossils impact as hard as living taxa in parsimony analyses of morphology, Systematic Biology, 56(5), 753-766

Ericson, P. G. P. (2011) Evolution of terrestrial birds in three continents: biogeography and parallel radiations, Journal of Biogeography, doi:10.1111/j.1365-2699.2011.02650.x

Hannisdal, B. and Peters, S. E. (2011) Phanerozoic Earth system evolution and marine biodiversity, Nature, 334, 1121-1124

Sereno, P. C. (2007) Logical basis for morphological characters in phylogenetic analysis, Cladistics, 23(6), 565-587