‘Archaeopteryx is not a bird (possibly)!’ and ‘Archaeopteryx is a bird again!’ In a pair of recent evaluations, Faculty Member Andrew Clarke of the British Antarctic Survey in Cambridge, UK, highlights how science continually revises its theories in light of new data.
His first review, classified as ‘new finding,’ discusses work by Xing Xu and colleagues (published in Nature in July) that controversially suggests that the Archaeopteryx, long believed to be the earliest known bird, is actually a dinosaur. This hypothesis stemmed from numerous discoveries of feathered, bird-like dinosaurs over the past fifteen years, including that of an Archaeopteryx-like theropod from China. Clarke encourages further investigation in order to follow up on the Xu et al.’s cladistical analysis that focused on flight features among related ancestry; yet, the possibility of having to reclassify “this most iconic of fossils” and re-evaluate possibilities in multiple origins of flight was, as Clark said, nothing short of “revolutionary”.
About three months later, another analysis on the same dataset of the Xu et al. paper caught Clarke’s attention, this time as a ‘refutation.’ That finding, by Michael Lee and Trevor Worthy, presented in Biology Letters, used maximum-likelihood and Bayesian methods and reverted back to the orthodox hypothesis that Archaeopteryx was indeed a primitive bird. Furthermore, Clarke highlights,
The likelihood tree implies a single origin for typical (that is forewing-powered) flight. The likelihood and Bayesian trees are supported by several derived characters which are unique to Archaeopteryx and other birds, but which are homoplasious in the parsimony tree of Xu and colleagues. This study illustrates the value of likelihood-based phylogenetic methods for palaeontology, and the need to continually reassess our view of evolutionary history as knowledge increases.
In several areas of science, one can choose multiple possible methods with which to work toward generating a theory of a phenomenon. This is certainly the case in developing a phylogenetic tree, where there are several routes that help to classify relatedness among groups of organisms. Interestingly, subtle changes or additions in methodology can present radically different, and even opposite, hypotheses. This seems a nice case in point of how scientific theories remain provisional, where hypotheses are never really fully proven. Karl Popper’s famous legacy in philosophy of science was to say that scientific theories should strive to be questioned, opened constantly to the possibility that they can be proven false owing to a contrary observation or experiment. It seems that has never been more true.