evo-devo

A study by Di Vincenzo, Steven Churchill and Giorgio Manzi has fallen into the early drawer of the Journal of Human Evolution: "The Vindija Neanderthal scapular glenoid fossa: Comparative shape analysis suggests evo-devo changes among Neanderthals" [1]. The authors do a very nice job taking a long-studied anatomical feature and reframing its variation within a new context. Reading through its discussion, I find much to like in the way Di Vincenzo and colleagues deal with the variation of late Neandertals and integrate the concept of introgressive gene flow among Late Pleistocene populations.

The glenoid fossa is the part of the scapula that articulates with the head of the humerus. It's the base of the "socket" in the ball-and-socket joint of the shoulder -- indeed, "glenoid" comes from the Greek word for "socket". Roughly shaped like a rounded teardrop, the glenoid is narrower in early hominins and relatively broad in recent people. Neandertals have an intermediate form compared to earlier and later humans.

Figure 1 from Di Vincenzo et al [1]. Original caption: "The scapular fragment VI-209 and its stratigraphic position (arrow) within the Mousterian layers of complex G of Vindija cave (left) according to Malez et al. (1980). On the right, the configuration of the 60 semi-landmarks used in the analysis is superimposed on the SGF profile. Sliding points are filled. The stratigraphic column is from Janković et al. (2006). Photograph by Milford H. Wolpoff."

The main point of the study is that the Vindija glenoid specimen, Vi-209, has a more humanlike form than other Neandertals. Another conclusion based on the comparative sample is that the sample of glenoids from late Neandertals is intermediate between early Neandertals and recent people. Likewise, Upper Paleolithic and Mesolithic-era European specimens are intermediate between late Neandertals and recent people. Here's a graph with the first and second principal components of the variation; I've highlighted these groups.

Figure 2a from Di Vincenzo et al. [1]. Altered to include sample names: Krapina, "Classic" and West Asian Neandertals, Vi-209, and Upper Paleolithic/Mesolithic. X-axis is the first principal component of variation based on analysis of the whole sample, Y-axis the second principal component.

The first principal component basically depends on the relative breadth of the glenoid fossa, with living people being much broader and Australopithecus (represented by Sterkfontein Sts 7 and Malapa MH2) being much narrower relative to the overall size of the fossa. The authors tested and rejected the hypothesis that the apparent trend could be a simple effect of size. This test was carried out relative to glenoid size, and since Australopithecus had relatively large shoulders compared to Homo, size does not vary much across the hominin sample. It would be useful to consider whether body size might matter, but body size would not by itself explain the relations of the later members of the genus Homo.

The authors emphasize that the data are consistent with a single evolutionary trend within the genus Homo, so that the Neandertal-human difference should be interpreted within the context of this broader pattern. They propose a specific developmental hypothesis.

Therefore, it seems reasonable that heterochronic factors related to the prolonged developmental pattern of our species (Smith et al., 2007a), which contrasts with the faster growth rates of Neanderthals and other ‘archaic’ hominins (Smith et al., 2007b; but see; Guatelli-Steinberg et al., 2005), led to longer periods of bone deposition along the inferior-lateral edge of the SGF [scapular glenoid fossa]. This could explain the observed variation along PC1 (and/or CV1) for different morphs of the genus Homo, reaching in H. sapiens the greatest extent in width of the SGF and, particularly, of its scapular portion. This is also consistent with the observation by Churchill and Trinkaus (1990) that much of the variability of the glenoid surface is a function of size variation of the joint itself, which can be viewed as forming a single functional matrix sensu Moss and Young (1960). Thus, the overall reduction in developmental rates in the genus Homo (relative to those of other hominoids) across the Pleistocene may account for the general evolutionary trend in SGF shape seen in the fossils, with more marked changes in developmental rates between archaic (including Neanderthals) and early modern humans, producing somewhat more dramatic differences between these groups in joint shape. Green et al. (2010) suggest that some of the differences between Neanderthals and modern humans in shoulder and thoracic morphology (particularly those related to clavicular length) are attributable to differences in the RUNX2/CBFA1 gene. The temporal pattern observed here would suggest that, with respect to SGF shape at least, that some differences are due to overall differences in developmental schedules (rather than specific differences in genes controlling development of the shoulder, such as RUNX2/CBFA1 or HoxC6).

By suggesting at least one actual genetic substitution in recent humans, they lend some plausibility to the idea. I am more hesitant to accept the assumption that Neandertals had faster developmental schedules than recent people, although it could be true. This specific assumption is not necessary to support the idea of heterochronic change in the glenoid, which could be caused by much more focused developmental processes. If glenoid shape reflects heterochronic developmental changes, the data suggest that those changes were ongoing in global populations during the Holocene. Indeed, the difference between recent people in the study and Upper Paleolithic Europeans is as great as the difference between late Neandertals and Upper Paleolithic Europeans. The study's recent human sample covers a broad geographic distribution but is relatively small in numbers; a fuller comparison of recent people might uncover a more interesting pattern of change.

The scapula has long figured in discussions of Neandertal genetic persistence. Neandertal scapulae often have a sulcus (groove) on the dorsal (back) aspect of the axillary border, and this feature is also found in a high fraction of early Upper Paleolithic skeletons [2] The axillary border morphology probably has no functional or developmental correlation with the glenoid morphology, so these features are best viewed as separate issues. I mention the axillary border only because of one significant commonality with the glenoid as considered here: We don't know how much variation in the trait may be explained by environment. Maybe the way an individual uses her arms when growing will affect the form of the scapula? With the axillary border, this question has occupied many researchers who tried to determine why some humans resemble some Neandertals and vice versa [3]. The current consensus is that a dorsal axillary sulcus probably reflects early developmental processes that are substantially influenced by genetics instead of shoulder activity pattern, but the consensus is not without detractors.

In this study, the authors consider the role of introgressive gene flow among Pleistocene populations as a way to maintain the apparently continuous trend:

The morphology of the SGF [scapular glenoid fossa] is unlikely to be under the genetic control of a single locus. Thus, it is more likely that regulatory genes controlling developmental rates overall produce pleiotropic effects throughout the skeleton. The introduction of these and other (non-regulatory) alleles into the Neanderthal populations of the Near East, and their movement by gene flow across Neanderthal demes into southern Europe (well in advance of the actual in-migration of modern humans) could account for mosaic morphology seen in the Vindija G3 Neanderthals, including the Vi-209 scapula. Introgression and subsequent gene flow would not be expected to have affected early Neanderthal populations (those predating the admixture), nor late Neanderthal populations from western (trans-Alpine) Europe, because they were separated by geographic barriers ( [Fabre et al., 2009] and [Degioanni et al., 2011] ), and/or protected from gene flow by distance (as hypothesized by Voisin, 2006).

There is as yet no evidence that the Vindija Neandertal genomes have genetic introgression from the African populations from which present non-Africans derive most of their genetic heritage. Green and colleagues [4] tested explicitly for this kind of gene flow, from "modern" into Neandertal populations and found none.

And yet, the latest Neandertals are consistently similar to recent people in ways that earlier Neandertals were not. The glenoid fossa of Vi-209 is not an isolated case, it joins many other characteristics in this sample (as noted in the quote above) and other Neandertal samples after 45,000 years ago.

Frankly, I expect that the admixture estimates presented thus far will prove to be wrong. I could be wrong in this expectation, but there are many assumptions underlying genetic analyses of admixture, and it's easy for an incorrect assumption to give rise to an incorrect conclusion. I take the morphological evidence very seriously as a possible "reality-check" about the validity of genetic comparisons. After all, the morphological comparisons predicted introgression from Neandertals in the first place...

Another reaction to the study by Zachary Cofran: "Evo-devo of the human shoulder?"

Fabio Di Vincenzo and colleagues analyzed the shape of the outline of the glenoid fossa on the scapula (not to be confused with the glenoid on your skull), from Australopithecus africanus to present day humans. The glenoid fossa is essentially the socket in the ball-and-socket joint of your shoulder. The authors found that there is pretty much a single trend of glenoid shape change from Australopithecus through the evolution of the genus Homo: from the fairly narrow joint in Australopithecus africanus and A. sediba, to the relatively wide joint in recent humans. The overall size and shape of the joint influences/reflects shoulder mobility, so presumably this shape change hints that more front-to-back arm motions became more important through the course of human evolution (authors suggest throwing in humans from the Late Pleistocene onward).

I think Cofran takes this in an interesting direction with respect to his own dissertation work on development in earlier hominins.

References

I'm just going to quote from a press release that fell into my inbox. It's about a talk being given at the American Society for Human Genetics meeting by Raymond Clarke, who identified a gene disrupted in a family sharing a disorder of the vocal tract. They call the gene tospeak:

The most exciting breakthrough in their research came when Clarke’s group discovered that the tospeak gene was unique to primates. Most of the human genome contains genes that are older (i.e., conserved over generations) and can also be found in other mammals, including the mouse. However, the tospeak gene is a relatively young gene that is only found in primates. Further excitement came when the group discovered that the tospeak gene has a special control region, known as a promoter, which is only found in humans.

“The discovery that a unique and more powerful human gene/promoter was disrupted in this vocally impaired family is of particular interest to the field of evolutionary genetics, since humans are the only creatures that have developed the capacity to speak,” said Dr. Clarke.

Clarke provided the following example as a comparison to help explain this new discovery: “Unlike GDF6, a bone protein gene which has existed since the dawn of vertebrate evolution, the tospeak gene is only found in primates. The best indication of the role of tospeak in human vocal development is that it was the only gene disrupted in a large family with a severe vocal disorder, altered composition of the vocal cords, and malformation of the voice box.”

That is a good story, as described, and I'd say it points strongly to the hypothesis that this gene was a target of selection in Homo for its role in vocal development. But this gene can't be alone, and the appearance of the promoter in humans doesn't necessarily suggest a change in its vocal-specific function. I'd like to know more about the gene's variation in humans, and whether there are other functional polymorphisms of the gene in primates. Vocal anatomy is quite variable, with a few very distinctive outliers.

At the same time, let's see some expression data -- the gene probably does other stuff, too, and that might be the target of selection.

Science has a very important paper in the current issue about the evolution of a gene enhancer in hominids, expressed in forelimb development and concentrated toward the first digit. The enhancer is a conserved sequence named HACNS1, it exhibits a stronger signature of recurrent selection on the human lineage than any other conserved enhancer sequence. In transgenic mice, the human version of this enhancer triggers gene expression in the forelimb, concentrated toward the thumb side, and some other parts of the body, notably the pharyngeal arches (which give rise to elements of mouth, throat and larynx), eye and ear. The research is by Shyam Prabhakar and others at Lawrence Berkeley National Lab, and involves Edward Rubin and James Noonan, otherwise prominent in the Neandertal genome sequencing.

I think this is an extraordinarily important result. You don't see me write those words very often. This is a paper that every biological anthropologist should read. It gives an extremely good example of the importance of developmental regulation to human evolution. We will see many more papers like this one in the coming years. This is one of the genes that makes us human.

Ed Yong of Not Exactly Rocket Science has written a nice online review of the research, and Science has accompanied it with a perspective piece by Gregory Wray and Courtney Babbitt. Here's a quote from that article:

To test the function of this region, they genetically engineered mouse embryos to express a construct composed of human HACNS1, the promoter element of a heat shock gene, and a reporter gene. Their results show that human HACNS1 drives expression in the mesenchyme of the early developing forelimb, and later developing hindlimb, in these mouse embryos. A comparison of expression patterns driven by macaque, chimpanzee, and human orthologs of HACNS1 revealed that consistently strong forelimb expression is a unique property of the human version. By testing various combinations of human and chimpanzee HACNS1 sequences, the authors narrowed down the relevant functional mutations to an 81-base pair region containing 13 substitutions that arose during human evolution. This concentration of substitutions is highly unusual relative to the genome as a whole, implying positive selection on this region during human origins.

The press are going with the story that the evolution of this gene may underlie the unique evolution of human manual dexterity. It's a good hypothesis, but I think there is a more accurate way of putting the situation. We see that the enhancer has effects in different areas of the developing embryo. Its action is therefore pleiotropic: changing its function in one area might well screw up its action somewhere else. So at the very least, this is an enhancer that must satisfy multiple constraints. Strong evolutionary change in its sequence may reflect changes in one of those functions, or more than one. But at the very least, it implies that the hominid developmental program not only satisfies different fitness constraints than in the human-chimpanzee common ancestor, but that these changes required repeated changes.

We don't know how long it would have taken all these nucleotide substitutions to happen. But we might find signs in the fossil record of such a sequence of events, if we had enough bones, and if we had more information about the effects of different forms of the gene on the adult phenotype. For example, the relatively long thumbs of the Hadar hominids (compared to chimpanzees and gorillas) suggest that the sequence of changes started early in hominid evolution. There's a hypothesis.

But like I said, I wouldn't rule out other possible functions of the enhancer as targets for selection. It is plausible (as a hypothesis) that the enhancer with the most selected substitutions on the human lineage might be more likely than others to have been selected for multiple functions. And we have plenty of reasons to suspect selection on its other targets, particularly the developing mouth, throat and ear.

It may even be that the evolution of human thumbs was a side effect of evolution in the throat, or vice versa. That's the kind of weird world evo-devo makes for us!

References:

Prabhakar S and 9 others. 2008. Human-specific gain of function in a developmental enhancer. Science 321:1346 - 1350. doi:10.1126/science.1159974

Wray GA, Babbitt CC. 2008. Enhancing gene regulation. Science 321:1300-1301. doi:10.1126/science.1163568

Seed is running a little article on the evolution of language, by lingust Juan Uriagereka:

A quasi-paradox has persisted within the field of linguistics, because the sudden emergence of such a complex, limitless system in a single species is hard to rationalize in terms of standard evolution. Its rapid spread makes language seem more like a viral epidemic that swept through the human population rather than a trait inherited through the typical dynamics of evolution.

Luckily, two recent advances have made it possible to rigorously address the problem of language's evolution for the first time. Molecular biology (including the publication of the human genome) and the so-called evo-devo paradigm now permit us to establish new and often quite unexpected connections among very different species. In addition, linguists' understanding of syntax -- how words are strung together into grammatical sentences—has developed to the point where language can be broken down into its basic procedural components. These components can now be seen to resemble traits observed in other species -- with functions that appear to be completely unrelated to familiar thought processes. Language may indeed be unique to humans, but the processes that underlie it are not.

It hits on many of the big topics, including the comparative biology of communication in finches, the regulatory role of FoxP2 in birdsong, and the brain processes underlying syntax.

I will differ from Uriagereka on this point:

The publication of the Neanderthal genome should tell us just how different their FoxP2 gene really is from our own.

Human FoxP2 differs from chimpanzees by two derived amino acid substitutions. If Neandertals were different from us (which seems likely, given the recent evidence of selection on the gene), then they would have had only one of these substitutions. It's an answer we don't actually need the Neandertal genome for. Now, if only we could start thinking about some other language-related genes.