Header image

john hawks weblog

paleoanthropology, genetics and evolution

Photo Credit: Endocast reconstruction of the "Neo" cranium of Homo naledi

Human brain evolution looks gradual. If you ignore enough data...

Bernard Wood’s research group has a new paper on brain size evolution in hominins, led by Andrew Du in Proceedings of the Royal Society, Series B: “Pattern and process in hominin brain size evolution are scale-dependent”.

In this paper, I notice that the researchers have done a really weird thing: Their analyses include only hominin fossils before 500,000 years ago.

Here’s their main figure:

Figure 2 from Du and colleagues showing endocranial volumes of hominin fossils over time
Figure 2 from Du and colleagues (2018).

Each of the symbols in this figure represents a single fossil hominin specimen that has an estimate of endocranial volume. The specimens reflect every hominin species from Australopithecus afarensis up to “Homo heidelbergensis”. Modern humans and Neanderthals have been left out of the dataset—they don’t fall within the pre-500,000-year time range.

On the basis of this dataset, the authors conclude that the entire hominin lineage is compatible with a single pattern of gradual evolutionary increase over time:

Figure 3b from Du and colleagues showing the data fitting a gradual model of brain size increase over time
Figure 3b from Du and colleagues (2018). Each point here is the data average in each 200,000-year-long interval, connected by a line. The shaded area indicates a range consistent with gradual linear increase over time.

The mean for each time interval of 200,000 years is plotted here, and none of the means fall outside the predicted error range for a gradual linear increase.

So what’s weird about this?

Let’s look at what the data show if we don’t impose an artificial limit of 500,000 years ago:

Same figure as above with hominin endocranial volume data, but this time with Homo naledi and Homo floresiensis added
Same figure as above, but this time with Homo naledi and Homo floresiensis added.

Hey, look at that! There are two species entirely missing from the data examined by Du and colleagues. The fossil records of endocranial volume in Homo naledi and Homo floresiensis both date to the last 300,000 years. When you include them, they both reject the notion of gradual monotonic increase in brain size.

It is likely that Homo naledi branched from the lineage of Neanderthals and modern humans more than a million years ago, maybe much longer. In other words, H. naledi ancestors must have existed throughout much of the last phase of our evolutionary history, and we haven’t found them yet. The same is true of H. floresiensis—and I don’t assume that Flores is the only place where such a population may have existed. The real message of these species is that the Lower and Middle Pleistocene records must undersample hominin diversity.

In neither case is it clear whether the small brains of these hominins resulted from a reversal from a larger-brained ancestor, or whether their small brain size was retained from the common ancestor of Homo.

The paper doesn’t justify its exclusion of this key evidence, so I don’t really know why the authors chose to ignore the data. There are other strange decisions underlying the analyses here—including the strange assumption that a gradual monotonic increase is an appropriate model across many species that are not a single ancestor-descendant lineage. I think that what looks like a “fit” is actually just an illustration of how weak the data are.

What’s obvious that the conclusions would be different if all the evidence were included.