Header image

john hawks weblog

paleoanthropology, genetics and evolution

Photo Credit: 3D printing a Homo naledi cranium. John Hawks CC-BY-NC-ND

A laboratory at Kyoto University has been maintaining a long-term evolution experiment on fruit flies that started in 1954. Now the flies are adapted to living in pitch black darkness:

To keep the flies away from light, they are reared in vials kept in a large pot painted black on the inside and covered with a blackout cloth. When the vials and food need to be changed, the researchers tend to the flies in the pitch dark, then use a feeble red light to check on their work. Fruit flies can’t see this light because the species lacks those light receptor proteins that absorb red wavelengths.
When [Syuiti] Mori retired, he passed on the precious fly stocks to his colleagues at Kyoto University, who have maintained them continuously to this day. The stock of flies has now spent more than 1,500 generations without light. In human terms, that would be like sequestering generations of our ancestors in the dark for 30,000 years.

It’s an interesting type of experiment, similar to the Long Term Experimental Evolution project that has kept E. coli cultures under a constant environmental regime for more than 64,000 generations. The populations adapt to their environmental conditions, different from the natural situations in which their ancestors evolved.

Culture, mathematical models, and Neandertal extinction

I am not philosophically opposed to building a mathematical model of Neandertal populations. Some of my best work has involved mathematical model-building. Models have an important place in helping us to understand evolutionary history. But when it comes to understanding Neandertal and modern human interactions, we have had lots and lots and lots of models and few testable predictions.

When you assume that modern human populations grew faster than Neandertal populations, you will conclude that modern human populations could have out-reproduced the Neandertals. This is not a very deep piece of circular logic. and so I get a little frustrated at the number of papers that really say nothing more than this.

Find x meme

Modeling is a start, but cultural systems are complicated. Clever innovations can help a population grow, but a population can co-evolve with its culture, yielding not only more growth but also greater difference in growth between populations. A cultural innovation can be tied to a particular landscape or raw material substrate, making it difficult to apply outside the context where it was invented.

The populations that we call “modern humans” really did out-reproduce the Neandertals. That’s why living people have only a small fraction of Neandertal ancestry today. But is culture a sufficient explanation? Were modern humans just smarter than Neandertals? Or were other factors important to the interactions between these populations?

Differential equations and Neandertals

I’m taking up this subject today in response to a paper by William Gilpin, Marcus Feldman and Kenichi Aoki, who have investigated a differential equation model for population growth with feedbacks from competitive interactions. The abstract summarizes the paper well:

Archaeologists argue that the replacement of Neanderthals by modern humans was driven by interspecific competition due to a difference in culture level. To assess the cogency of this argument, we construct and analyze an interspecific cultural competition model based on the Lotka−Volterra model, which is widely used in ecology, but which incorporates the culture level of a species as a variable interacting with population size. We investigate the conditions under which a difference in culture level between cognitively equivalent species, or alternatively a difference in underlying learning ability, may produce competitive exclusion of a comparatively (although not absolutely) large local Neanderthal population by an initially smaller modern human population. We find, in particular, that this competitive exclusion is more likely to occur when population growth occurs on a shorter timescale than cultural change, or when the competition coefficients of the Lotka−Volterra model depend on the difference in the culture levels of the interacting species.

The Lotka-Volterra system of differential equations is one in which two components of a system change over time, in such a way that the amount of change in one component depends upon the magnitude of the other component. It has been most famously applied as a model for predator and prey populations, where predator population growth is coupled to prey population size. The behavior in this system can be cyclical—for example, if a predator population crashes, the growth of a prey population can resume, driving subsequent growth in the predator population.

This kind of feedback between components of the system is determined by the differential equation coefficients. The utility of this kind of system is that varying the coefficients allows us to investigate the conditions under which the system can exhibit stable, cyclical, or degenerate behavior.

Gilpin and colleagues assume the Neandertal and modern human populations to have been in a competitive interaction, where the rate of growth of the Neandertal population is smaller when the modern human population is larger, and vice-versa. Each population grows logistically up to a carrying capacity, which is determined by a parameter that they refer to as “culture level”. This “culture level” increases with population size, and it increases faster for the modern human population than for Neandertals—in other words, in this model Neandertals are stupid.

In this system, modern human populations grow faster when they wipe out the local Neandertals. They innovate faster when the population gets larger. If the difference in the rate of change of “culture level” is sufficiently large, Neandertals are doomed.

In principle, the model allows the authors to investigate the way that a second parameter, “culture level”, might constitute an advantage if Neandertals had a lower rate of increase. But the paper does not actually deploy this model in a way that would inform us about the importance of this second parameter. As a result, the conclusion is boring. If we assume that Neandertals were stupid, we don’t need a differential equation to tell us what happened to them.

Neandertal in a suit

Is “culture level” relevant?

The problem is not that we lack models to show how culture may have helped modern humans beat the Neandertals. The problem is that the archaeological data suggest that culture alone may be a poor explanation.

For one thing, the Neandertals persisted in Europe and central Asia long beyond the entry of modern humans into Asia. Initial modern humans in Asia exhibited no obvious cultural superiority over other Middle Paleolithic people, who were presumably archaic humans. “No cultural superiority” is maybe an understatement: Archaeologists have trouble finding any consistent material culture differences between people in West Asia before 50,000 years ago.

Tens of thousands of years later, when modern humans did start to enter Europe, they seem to have mixed with Neandertals more extensively. The later Neandertals were making symbolic artifacts, using pigments, feathers and other ornaments. The people who made the earliest Aurignacian, often assumed to be the earliest modern humans in Western Europe, did not have the intensity of symbolic artifacts of later Aurignacian and Gravettian people. Instead they seem to have been sparse and little different in most cultural practices from Neandertals.

In other words, at the critical time when modern humans entered Europe and their population apparently grew, there was little cultural difference between them. There is even less evidence that there was any cultural advantage to modern humans who spread across southern Asia prior to 50,000 years ago.

What gives? If we assume that “culture level” was a continuous variable, and that “modern humans” had a higher rate of increase than Neandertals, we get a very simple pattern. The data are not a simple pattern. So the “culture level” model seems like a bad model to account for the complexity of what actually happened.

Vital forces

“Culture level” is an archaeological version of “vital force”. Plenty of archaeologists think there is something special about “modern human behavior”, and believe that a “spark” entered the human population. After this vital spark entered the modern population, they were able to grow their population, spread around the world, and conquer the earth with their cultural adaptability. Some have written that this “spark” was a key mutation, some believe it was fully human grammar, some believe that it was a special demographic or ecological setting.

They are not thinking like biologists. The evolution of human cognition was not magic, and it was not caused by a “spark”.

I don’t object to the idea that Neandertals may have been cognitively different than modern humans—in fact, I think this is likely. The idea that Neandertals were fixed for stupid and modern humans fixed for smart is biologically incredible. Instead, we need to consider that if many Neandertals had challenges learning to work with some cultural innovations, many modern humans should have had such challenges as well. Key innovations, if rare, must have been stochastic.

To understand human cognitive evolution, we must consider how specific behaviors may have contributed to reproductive success. Useful cultural innovations tend to be transferred readily across groups, and so make unlikely vehicles for a continuous growth of one population at the expense of another. Was the ability to learn some cultural behaviors heritable? If so, it is unlikely that the ability to learn two behaviors was equally heritable; some must have been more influenced by genes than others. To the extent that behaviors are learned by exposure to skilled individuals, this exposure causes the selection in favor of the ability to learn to weaken as the trait becomes more commonly expressed.

Cultural selection to enforce cultural uniformity can be very effective in fixing cultural traits in a population, but is not especially likely to enhance the traits that really matter for adaptation to new environments. Indeed, cultural selection in many recent human groups has been depressingly conservative, preventing innovation and reducing population growth by imposing various handicaps.

In the real world, some archaic people—including the Neandertals—really were more successful than most early modern human groups. Neandertals as a population contributed more DNA to people around the world than their “conquerors”, the Upper Paleolithic people of Europe. Some modern human populations have massively grown during the last 50,000 years at the expense of others, often for cultural reasons. When we look at the diversity of those situations, we can see that culture is not easily broken down into a linear variable.

Non-linear culture

What makes culture more than a simple system of accumulating knowledge?

There are conditions under which a cultural system may be hard to transfer across groups. A cultural system may have co-evolved with some genetic variants, like dairying and lactase persistence. The cultural traits in such a system, even if learned, might not have had the advantage for Neandertals as for the modern humans that developed it. A cultural system may rely upon some elaborate codification of social behavior, like religious rules, that are not readily adopted by new cultural groups. Again, the behaviors that seem tied to reproductive advantage in such a system may not be as advantageous to people who lack the essential cultural background.

What we lack is some empirical demonstration of what cultural factors among Late Pleistocene people actually led to higher reproductive success. Archaeologists have proposed several, but have tested few. Most, like evidence for symbolic behavior, have subsequently been found in the Neandertals themselves, making them poor explanations for Neandertal extinction.

Later modern humans exhibited greater material culture diversity and more symbolic expression than earlier modern humans, thousands of years after the Neandertals were gone. This is true not only in Europe and central Asia, it is also true in other places long after the first modern humans appeared there: in southern Africa, west Asia, and southeast Asia. In every part of the world, the evidence for elaborate symbolic culture occurs long after the earliest evidence of modern humans. And in most of these areas, the first evidence for symbolic culture occurs substantially before the earliest evidence of modern humans.

To me, this means it was not just “culture level” that made a difference. I can imagine that there may have been some specific aspects of culture, which may not have been archaeologically visible, that made a key difference. Archaeologically visible material culture may reflect demographic growth, but not necessarily the key aspects that mattered to the initial dispersal of populations.

But I can also imagine that non-cultural factors were more important. For example, disease has been a key factor underlying the survival or replacement of populations during the past 10,000 years. It is not a stretch to imagine that disease influenced the Neandertals and other archaic peoples differently from each other and from modern humans. The evidence for selection on genes related to immunity from these archaic humans is now strong, and some of these may reflect pathogens or parasites that were important at the time of population contacts among these people.

Exploration needed

This is why we need more data, more exploration, more archaeology. I don’t mind if people continue to think of mathematical models, as they may help us to understand which factors are important. Listing all the possible factors doesn’t necessarily get us closer to a test of which of those factors were crucial to our evolution, and including every factor in a model will make it untestable.

But we are past the point where a simple model is going to tell us something we don’t already know. Neandertals are gone. Their cultures did not persist. Yet they are among our ancestors. What is necessary is to test models against the timeline of modern human dispersal as we currently understand it, and to take note of those predictions that we have not yet observed. It is the novel predictions of a model that make it valuable to the future.


Gilpin W, Feldman MW, Aoki K. 2016. An ecocultural model predicts Neanderthal extinction through competition with modern humans. Proc Nat Acad Sci USA (online) doi:10.1073/pnas.1524861113

Aleš Hrdlička, in the concluding paragraphs of The Most Ancient Skeletal Remains of Man, his 1914 review of the fossil evidence of human evolution:

The gradually accumulating finds which throw light on the physical past of man, have naturally stimulated further exploration in the same lines; and the various failures and uncertainties connected with some of the finds in the past have impressed all investigators in the field with the necessity of the most careful and properly controlled procedure. Besides men of science, the educated public, engineers controlling public works, and even many among the workmen in Europe have been impressed by these remarkable discoveries, and in hundreds of instances are doubtless watching for new treasures. Under these conditions we are justified in hoping that from time to time we shall receive additions to the precious material already in our hands; that these additions will fill the existing vacua, and gradually extend farther back to the more strictly intermediary forms between man and his ancestral stock, and perhaps eventually even to the source of these link-forms themselves, to the peculiar morphologically unstable family of the anthropogenous primates.
While the anthropologist is thus painfully and slowly reconstructing the past physical history of man, he is also with every new fact adding another imperishable block to the foundation upon which will stand not only the knowledge of the future in regard to man himself, but also the laws of his further physical development, and radically even those of his beliefs and his moral behavior. This is a part of the service of anthropology to humanity.

I think I’m going to take that phrase, “another imperishable block”. Seems ripe for mischief-making…

Ed Yong has an article in Atlantic on the mysterious evolution of the human chin: “We’re the Only Animals With Chins, and No One Knows Why”. He builds on a recent review paper by James Pampush.

It’s a nice read about a trait that has occupied more than the usual amount of paleoanthropological mindshare over the last hundred years.

For example, during human evolution, our faces shortened and our posture straightened. These changes made our mouths more cramped. To give our tongues and soft tissues more room, and to avoid constricting our airways, the lower jaw developed a forward slope, of which the chin was a side effect. The problem with this idea is that the chin's outer face doesn't follow the contours of its inner face, and has an exceptionally thick knob of bone. None of that screams “space-saving measure.”

There are some true “chin nerds” out there who will likely think the article doesn’t treat their favorite hypothesis fairly. The fact is, it’s sort of embarrassing that we have not yet come up with a persuasive test of any of these hypotheses.

Christopher Henshilwood has written a short article for The Conversation describing the archaeological importance of the finds from Blombos and elsewhere in southern Africa: “What excavated beads tell us about the when and where of human evolution”.

Modern human behaviour can be defined as behaviour that is brought about by socially constructed patterns of symbolic thinking, actions and communication. This allows for material and information exchange and cultural continuity between and across generations and contemporaneous communities. The capacity for symbolic thought is not the key defining factor for modern human behaviour. It is rather the use of symbolism to organise behaviour that defines us.
In other words, early humans were first behaviourally modern when symbols became an intrinsic part of their daily lives.

The tranformation of archaeology toward understanding the complexity of Middle Stone Age assemblages in southern Africa is impressive. I think it may be premature to write off the behavior of earlier people as non-symbolic. After all, the production of symbolic objects can only be functional within a society in which symbolic communication is near-universal. Language provides such a basis. Even simple forms of vocal communication in other primate species involve arbitrary learned relationships between sounds and concepts, which is the definition of symbolic communication. Auditory symbol use must have been present in the earliest forms of human language as well, and evidence from the vocal and auditory channels place the origin of vocal language much earlier than the Neandertals.

This is not modern human, I would say it is simply human.

I think we should take seriously the hypothesis that the differences between Middle Pleistocene populations were quantitative and not qualitative. Symbolic behavior did not emerge instantaneously; it evolved within a context of complex social interactions of earlier archaic human populations. The use of symbolic artifacts is an important clue to ancient social systems, evidencing one aspect of complexity. The artifacts give us one kind of evidence about the connections among ancient groups and the persistence of symbolic traditions.

I endorse this message from Holly Dunsworth:

Everyone’s likely heard it or seen it written on a protest sign: “I didn’t evolve from a monkey.” It’s a well-worn refrain of those who resist the evolutionary perspective. The pat response we often hear is, “You’re right! We didn’t evolve from monkeys. We share ancestors with them.” However, this talking point isn’t entirely honest.
Yes, we share ancestors with monkeys; we share ancestors with every living thing. But, also, to be clear: We did evolve from monkeys.

That’s from her new blog, “Origins”, from the new website, SAPIENS. It’s a new public communication portal sponsored by the Wenner-Gren Foundation for Anthropological Research, and I’ll look forward to seeing more of what they have in store.

How long does it take to publish new hominin species after discovery?

In an earlier post, I looked at descriptions of new hominin species during the last 25 years, to see how long they took from submission to acceptance in the journal where each was published (“Hominin species and time in peer review”). The data show that these papers have taken a median time of 70 days to review. But as I noted, the time in peer review probably doesn’t tell us very much about the quality of review. What stands out is that the duration of review has not changed appreciably across the 25-year span.

Au. sediba
Australopithecus sediba was 20 months from first discovery to publication. Photo credit: Lee Berger

A more interesting time interval is from discovery to publication. This time includes not only the peer review and other editorial processes, but also the primary scientific work. Technicians prepare and conserve the fossils, specialists in anatomy take systematic measurements and observations, and they carry out comparisons with other samples of fossils and skeletal collections.

Analysis can take a lot of time. Some fossils require considerable reconstruction. Today such reconstruction can be carried out virtually after 3D scanning of the material, which sounds like it should save time but somehow always seems to take longer. If a specimen preserves a rare piece of anatomy, the comparable anatomical areas may not have been well-reported in other fossil samples, and therefore one or more researchers may need to make special research trips to study fossils from other parts of the world.

Considering the diversity of fossil preservation across different field sites, you might expect the process of publishing diagnoses of new fossil species to be equally diverse in how long they take to prepare.

The data show the opposite: the duration of scientific work preceding the diagnosis of new hominin taxa has generally been between one and two years, and has remained pretty much the same over the last 25 years.

Discovery dates

When was a species found? This is not a simple question, because the evidence for a new species is rarely limited to a single specimen. A team may find fragments that suggest the existence of a new species in one field season, and later find additional evidence that enhances the case.

For example, the earliest specimen now attributed to Australopithecus anamensis to have been discovered is a humerus fragment from Kanapoi, KNM-KP 271, which was found by Bryan Patterson’s research team in 1966. This fragment was known to represent a very early hominin, but does not present anatomical features that would have enabled a clear diagnosis relative to other known hominin species. Only much later did Maeve Leakey and colleagues uncover more remains from Kanapoi and Allia Bay that enabled a diagnostic comparison with other species.

Every formal description of a new species designates a single specimen, known as the “holotype”, that will serve as an anatomical reference for future scholars. Under the rules of the International Code of Zoological Nomenclature, the holotype is forever tied to the formal species name; it can never be recycled to serve as the holotype for any other species name. For Au. anamensis, Leakey and colleagues designated the specimen KNM-KP 29281, discovered in 1994, as the holotype of the species. So although the first specimen now attributed to Au. anamensis was found in 1966, the holotype was found 28 years later.

Even the holotype discovery date is not really the time of “species discovery”. Generally, a team of scientists tests the hypothesis that a new fossil assemblage belongs to an existing species. If the evidence rejects this hypothesis, a team may move toward formally defining a new species. “Discovery” of this new species may happen more-or-less gradually during the analysis of fossil remains, as researchers develop evidence in comparison with other fossil samples. In some cases the team may recognize this distinctiveness very rapidly, in others more slowly, depending upon the quality of the evidence and the difficulty of making the comparisons. Sometimes additional field seasons may be necessary to add more fossil specimens and thereby broaden the scope of comparisons. In some cases, it is the holotype that is unearthed during a later field season, well after a field team has other specimens that make them think a new species exists.

Looking at the records available in formal taxonomic diagnoses, the only practical alternative is to consider the time of holotype discovery. Papers usually report this date and do not reliably list the discovery dates of paratype specimens. Even with holotypes, the reporting is uneven. Some papers have reported a single day of discovery, in which case it is simple to calculate the time from discovery to publication. In many cases, however, only a month of discovery (e.g., “March 2014”) is provided. There several reasons why a single date may not be available. A holotype may have taken several days to uncover during excavation, or it may have been reconstructed from fragments found over a range of dates. In extreme cases, parts of a holotype might have been found in successive field seasons.

In a few cases, the paper gives only the year of discovery (for example, the only date given for KNM-KP 29281 is “1994”). For my comparisons here, if the paper gives no indication of the timing of the field season, I have assigned a date of January 1 to these discoveries. The resulting timeline is necessarily longer than the real time taken by a research team to describe its discovery, in theory up to 12 months longer if the specimen was actually discovered in December.

The results

Number of months from holotype discovery to publication for hominin species since 1990
Asterisks (**) indicate species for which the discovery year was supplied but not the date; the point in this graph assumes discovery on January 1 of that year, making this the maximum possible time from discovery to publication. In reality they may be as much as 12 months less than indicated here.

In this chart, H. cepranensis is an outlier, as discussed below. The remainder of the data show no reduction or increase in the time from discovery to publication over the last 25 years. Out of the fifteen formally named taxa here, ten were published within two years after the discovery of the holotype specimen. Only two of those appeared within one year after discovery. The median time from discovery to description is 20 months.

The x-axis of the chart is the discovery date, not the publication date. There may be taxa that have been discovered in the past few years that have not yet been published, and obviously any such species would not be represented in the chart. H. gautengensis, published in 2010 but based on a holotype discovered in 1976, is not included in the chart.

This period of two years or less includes the time spent by the species in peer review and revision, which I discussed in the previous post. In the case of Homo floresiensis, for example, the time from submission to acceptance of the manuscript was more than 6 months, and the publication of the paper was still only 14 months after the discovery. The paper describing Australopithecus ramidus was slightly more than two months from submission to acceptance; the total timeline from discovery of the holotype to publication was only 9 months.

In light of discussion about the publication pace of Homo naledi, these data may surprise people. H. naledi may be remarkable for the quantity of anatomical evidence, but not the time from discovery to publication. It took substantially longer to move from discovery to publication for H. naledi than for Au. ramidus (9 months), S. tchadensis (12 months), or O. tugenensis (3 months), and even longer than H. floresiensis (14 months) and Au. garhi (17 months).

What explains the consistency in time to publication?

A varied array of research teams around the world have composed effective and highly-cited diagnoses on a varied array of fossil assemblages, with all necessary research and editorial handling and peer review within two years or less. Setting aside diagnoses of new taxa based on previously-published fossil assemblages, the timeline of the procedure has been remarkably consistent.

I think this consistency of timeline can in large part be attributed to the consistency of content.

  • Formal diagnoses follow a common recipe, with a stereotypical block giving essential diagnostic information, and a discussion that places the new taxon into a phylogenetic and adaptive context.

  • The addition of a new taxon generally must reiterate the key information of the next most similar taxon.

  • Most diagnoses of hominin taxa are relatively short, with a median of 7 text pages. Only Homo naledi and Homo gautengensis were diagnosed in papers longer than 11 pages.

  • Although scientific papers have undergone a major trend during the past 15 years to add supplementary information in addition to the main text, this has affected very few of the hominin taxa included here. Only Au. deyiremeda, Au. sediba, and H. floresiensis were accompanied by supplementary information of substantial length.

Some of these species are now represented by fossil samples that were difficult to reconstruct and analyze. But in nearly all cases, the reconstruction and analysis was a second phase of work that followed the formal diagnosis of the new taxon. Therefore most of the difficult work of reconstruction could follow formal diagnosis. This was perhaps most famously the situation with Au. ramidus, where the most famous specimen is not the holotype. But to one degree or another, the process of later, deeper description has been a routine part of the study of most of these species.

The exceptions

What about the exceptions, the species that took much longer than two years to diagnose?

Five taxa required more than two years from discovery to publication. The diagnosis of one of those taxa, H. cepranensis, was based on a holotype specimen that had been described in a peer-reviewed publication several years earlier. To this we can add several similar cases of new formal diagnoses of taxa published in the last 25 years based on previously published fossil material.

  • Curnoe (2010) based the diagnosis of H. gautengensis on StW 53, which was found in 1976, with a description published by Hughes and Tobias (1977).

  • White et al. (1995) based their diagnosis of the genus Ardipithecus on the species diagnosis of Au. ramidus published in 1994.

  • Haile-Selassie et al. (2004) based their diagnosis of the species Ar. kadabba upon the diagnosis of the subspecies Ar. ramidus kadabba that Haile-Selassie had previously published in 2001.

I don’t think these cases are directly comparable to the original first assessment of newly excavated material. Such secondary study is also a process of discovery, sometimes undertaken due to the recovery of additional material (as in the case of Ardipithecus), but the timeline of such research is extremely variable.

Two of the taxa that took longer than two years to diagnose were subspecies: Ar. ramidus kadabba and H. sapiens idaltu. Providing such formal diagnoses of subspecies is a relatively new innovation in hominin taxonomic practice. One motivation to name a subspecies is to preclude the use of a holotype specimen for other taxonomic diagnosis in the future. As the case of Ar. kadabba shows, future discoveries may force a reassessment of the variation presented by a sample, sometimes prompting the promotion of such a subspecies to a species-level taxon.

I can think of two hypotheses to explain why the diagnosis of these subspecies has occupied a longer period of time than species or genera. One possibility is that the relatively subtle anatomical variation that distinguished a subspecies may require more time to fully understand and characterize. A second possibility is that researchers may work on new species and genus-level diagnoses with greater intensity than a subspecies diagnosis. These are not mutually exclusive, and probably there are other possibilities as well. But again, it is not clear that researchers treat the formal diagnosis of a subspecies as the same kind of task as the diagnosis of a higher-level taxon.

This leaves only Au. deyiremeda and H. antecessor, which required approximately 4 years and 3 years from holotype discovery to description, respectively. Both of these are instances in which additional field seasons may have added more information from additional specimens, and as I move forward, I’ll consider whether the sample size and anatomical regions preserved in the holotype and paratype specimens help to explain the timeline of either of these taxa.

Is everyone “rushing” their research?

The standards by which I judge the quality of science are replicability and originality, not speed. The formal diagnosis of a taxon is one of the least creative exercises in paleoanthropology, with several highly standardized parts. These essential ingredients have emerged through history as a way of ensuring the replicability of the key observations that contribute to attributing other fossils in the future. Considering that many hominin fossils are practically inaccessible to independent scientists, scientists must insist that the formal description of such fossils will meet a high standard of replicability.

With that in mind, is there a correlation between replicability and speed of publication?

I don’t think so. Among the papers that are published within four years of holotype discovery, I just don’t see any obvious correlation between time to publication and replicability.

Keep in mind that this is a very small sample to try to find such a correlation. Certainly there are papers here that omit crucial data, there are papers that have given rise to years of controversy, and there are papers that have led to relatively little subsequent reassessment of the broader phylogenetic pattern of hominins. Any set of independent scientists would probably have a wide variety of “favorites” or choices as “best hominin description EVAH”.

But even if we take the strongest critics of new species, I don’t think we see any relationship between a person’s preferences about which diagnoses are replicable and the timeline of publication.

As an example, Tim White has been publicly critical of many of these species as examples of “taxonomic inflation”. Such “inflated” species include H. naledi, Au. deyiremeda, Au. sediba, K. platyops, O. tugenensis, S. tchadensis, Au. anamensis, and H. georgicus. I may be missing several. The median time from discovery to publication of these eight taxa is at most 20 months. The median time for the three taxa published by White himself is 17 months. Whatever the standard of quality and replicability applied by White, time is not an explanation.

Production of a taxonomic assessment for newly discovered hominin fossils is a basic responsibility of field research. Assessing whether a fossil or an assemblage belong to a previously-known taxon is relatively straightforward. That assessment may rely upon small anatomical details, but a review of the anatomy of a fossil will not likely miss those details if they are present. So there’s nothing about the procedure that in principle should take many years to accomplish.

In light of the evidence from the last 25 years of formal taxonomic diagnosis in hominins, it is clear that in most cases, this process is very efficient.


  1. As in the previous post, I am missing data for Australopithecus bahrelghazali.


References for this post are the same as listed in “Hominin species and time in peer review”.

From chapter 2 of Sharing Publication-Related Data and Materials: Responsibilities of Authorship in the Life Sciences, a publication of the National Research Council of the National Academy of Sciences, U. S. A., in 2003.

Community standards for sharing publication-related data and materials should flow from the general principle that the publication of scientific information is intended to move science forward. More specifically, the act of publishing is a quid pro quo in which authors receive credit and acknowledgment in exchange for disclosure of their scientific findings. An author’s obligation is not only to release data and materials to enable others to verify or replicate published findings (as journals already implicitly or explicitly require) but also to provide them in a form on which other scientists can build with further research. All members of the scientific community—whether working in academia, government, or a commercial enterprise—have equal responsibility for upholding community standards as participants in the publication system, and all should be equally able to derive benefits from it.

The entire book, reporting on the NRC “Committee on Responsibilities of Authorship in the Biological Sciences” in 2002, is available online for free.

I spent part of the week preparing for the beginning of classes next week here at UW-Madison. When I’m teaching introductory classes, I’m always more attentive to the basic information that’s available for students.

Neil Shubin reminds us that the Howard Hughes Medical Institute has made all three episodes of the Tangled Bank documentary, “Your Inner Fish”, available for streaming:

As a consultant for the production, I’m very enthusiastic about this. I’ll be using the book, Your Inner Fish, in my course on Evolutionary Biology this semester. It’s such a great way to look at how evolution has transformed vertebrates into our own unique body form.

I’m less enthusiastic about the pictures of hominin relationships that teachers have to show students:

My post on the timeline of peer review for hominin species got some attention:

Some tweets about the new archaeological finds on Sulawesi:

On the subject of data access and replicability in science, first another hopeful sign that African countries are moving toward adoption of open access principles:

And my reminder that I do not consider access to data as an optional add-on to good science.

This week, another series of revelations about sexual harassment by professors of astronomy and astrophysics has dominated my tweetstream. So many voices there that need hearing, many sharing their own harrowing personal experiences. One summarizes the problem:

Why do universities cover up high-profile harassment? Look for the money

How can professors get away with years of sexual harassment or abuse of graduate students without their universities taking any action?

I don’t think money is the entire answer, but I do not think it is a coincidence that these professors generated millions in grants for their institutions.

This week, a story of the disciplinary action against Christian Ott at Caltech was told by Azeen Ghorayshi from Buzzfeed (“He Fell In Love With His Grad Student — Then Fired Her For It”). Last fall Ghorayshi also broke the story of long-term sexual harassment by astronomer Geoffrey Marcy at UC-Berkeley. People reading the Ott story are rightly amazed that a department would fail to take action as a professor has a succession of nine graduate advisees leave in a span of a few years. Different fields and institutions have different standards, but this remarkably high attrition is a clear signal that something serious has gone wrong. So how could an entire department ignore the signs?

The story of Timothy Slater at the University of Arizona was made public this week by Congresswoman Jackie Speier, as detailed in a Mashable story: “Congresswoman reveals prominent astronomy professor’s history of sexual harassment”. Speier also expresses frustration and amazement at the failure of universities to take effective action to end instances of harassment, as quoted in a Wired interview (“Rep Jackie Speier on Why She’s Taking on Sexual Harassment in Science”):

In my work on sexual assault in the military and sexual assault on college campuses, the pattern is there. Typically predators are in environments where there is a closed institution, where they have their own code of conduct—whether the military code of justice or the code of conduct at a university. Often times these cases are not handled appropriately. They sweep them under the rug. It allow sexual predators to reoffend. Often times it’s six or seven times before they are actually caught because everyone believes it’s a one off situation.

In the three cases that have been made public, the universities (the University of California-Berkeley, University of Arizona and California Institute of Technology) investigated and took disciplinary action against the professors in question. In response to the public relevations this week (and with Geoffrey Marcy last fall), people have been speaking out to criticize these institutions’ responses, focusing on the weakness of the disciplinary actions, the lack of protection for future students or adequate response to the injustice suffered by past students. In two of the cases, the faculty member successfully moved to another institution in a senior high-salary position despite sexual harassment investigations or disciplinary action at their previous institution.

These cases are outrageous. These are not cases where a department made a bad hire that no one could have anticipated. High-salary scientists do not make mid-career moves into endowed positions without the special involvement of university administrators. Each of these universities has profited handsomely from its association with these scientists. The three cases now public involve more than $17 million in NSF and NASA funding.

I’ve been surprised at how little attention has been given to the financial aspect of these cases. Federal grants are public records, and we are not talking about small amounts of money. When people ask why university administrators have not been forthcoming about sexual harassment and abuse of students, I believe we must look at how those administrators continue to benefit from looking the other way.

A hint about the importance of funding in this issue comes from the Mashable article that discusses Slater’s case.

Slater began his career as a Kansas high school science teacher in 1989 and has since become one of the most renowned names in astronomy education. Slater said that, over the course of his career, he has received more than $30 million in federally funded grants, and in developing the curriculum for a new generation of astronomy teachers, has received several awards and prestigious appointments.
In 1996, he took a job as a physics professor at Montana State University, where he founded the Conceptual Astronomy and Physics Education Research (CAPER) program that his wife Stephanie today runs as an independent nonprofit. In 2001, he was hired as an astronomy professor at the University of Arizona.

I did a public records search for NSF and NASA awards for each of these three scientists. I could not confirm the $30 million claim based on searching for awards at NSF and NASA. My search is only as good as the online systems for federal grant searches, and it is probable that Slater has significant funding from other federal grant sources. I found a total of $5,450,737 in NSF awards credited to Timothy Slater as either Principal Investigator or Co-Principal Investigator. Additionally, $652,637 was awarded to Slater by NASA at the University of Wyoming and $85,977 at the University of Arizona.

NSF records credit Christian Ott at Caltech with $5,106,789 for grants upon which he is listed as Principal Investigator (PI) or Co-Principal Investigator (co-PI). NASA credits him with $150,000 as PI.

The total in NSF awards credited to Marcy as PI or co-PI is $3,636,399. NASA additionally credits $2,612,545.98 to Marcy as PI since 2005. As large as these amounts are, they are small compared to the $100 million in funding from the Breakthrough Prize Foundation (“Internet investor Yuri Milner joins with Berkeley in $100 million search for extraterrestrial intelligence”). Marcy’s work on extrasolar planets was publicly recognized as a factor behind this award, and he was prominent in the public announcement.

These scientists are not merely successful at grant-getting, they are rainmakers for their universities and departments. In times of increasing challenges for state support for these research universities, the funding brought in by such scientists keeps the budget afloat.

Many people who are not familiar with the U.S. federal grant system may not know that much of this money goes directly into university budgets instead of the project for which the funds were granted. Nature News did a nice article on indirect costs in 2014 that explains them well: “Indirect costs: Keeping the lights on”. When a federal agency funds a grant, the agency provides an additional amount of funding directly to the institution, above and beyond the direct costs of the project budget. These “indirect costs” are intended to help fund the institution’s research enterprise—buildings, laboratory space, electricity, personnel to administer the grant accounting and compliance with federal regulations. Each university negotiates an indirect cost rate with the federal government, which is applied to every federal grant awarded to the university.

Negotiated indirect cost rates differ greatly among institutions. At Caltech, the current indirect cost rate for on-campus research is 64.3% of direct costs, meaning that every budgeted dollar of direct costs comes with a supplementary fund of 64.3 cents to the university. At the University of Arizona, the current rate is 53.5% for on-campus research. Off-campus research indirect cost rates are much lower, in the mid-twenties. Indirect cost rates have consistently increased year-over-year, so grants funded 10 years ago may have had indirect costs from 45% to 55%, not the 50%–65% common today.

Indirect costs mean that when we look at the amount of grant funding awarded to the three scientists in these cases, each university has taken in millions of additional dollars of federal funding to its bottom line. That money was not allocated to these scientists’ particular projects, it solved budget problems for administrators.

Each case that becomes public teaches us more about a culture of neglect when it comes to sexual harassment and abuse of students. Harassers are not the majority, they are a tiny handful of scientists. But they are powerful, and too many departments are full of faculty who do nothing to stop them. That culture comes from treating grants and publications as the only important standards of performance assessment.

I am discouraged that every time I hear about one of these cases, the accused faculty member invariably has been a big NSF grantwinner.

Funding and training are strongly connected at universities. Many federal grants of this kind include direct funding for the salaries of graduate student research assistants and postdoctoral scientists. Most applications address graduate student training as part of their intellectual merit and broader impact. Additionally, when a university’s policies are not protecting its graduate students, we should consider other ways that the federal government is funding graduate education at the institution. Among the most pertinent is the IGERT program by which NSF directly funds interdisciplinary graduate work: more than $6 million at UC-Berkeley and $400,000 at the University of Arizona.

What kind of oversight could NSF and NASA implement to change this? Astronomy is not alone. Anthropologists are now discussing our field’s history of sexual harassment in light of the SAFE study, which showed the extraordinarily high fraction of students who are sexually harassed or abused during fieldwork research experiences. A research project should not be a fiefdom in which the PI has droit de seigneur. But this will persist as long as PIs are rewarded for this culture of abusive behavior.

The culture needs to change. The harm to students is irreparable. The harm to science will be a legacy across the next thirty years haunted by the shadows of promising careers that were stopped before they began. Grant reviewers and panels need to take seriously their responsibility to training the next generation of scientists. At a minimum, that means demanding evidence of a project’s demonstrated success in training students, including evidence about a university’s or department’s track record.

Rep. Speier has rightly criticized the universities for keeping the cases quiet. Each university profits from silence. Universities will hide this behavior as long as administrators believe that future funding will keep rolling in. If we want to end the secrecy, we need to stop the money.

We know about the cases only because victims have spoken out, and particularly we owe much to dogged investigative reporting by Buzzfeed and Ghorayshi. Many important voices have yet to be heard, including some who may draw attention to additional cases both inside and outside of astronomy. We need to keep listening.

More reading:

Ed Yong’s new piece in National Geographic on the evolution of eyes across different branches of animal life is a great one: “Inside the Eye: Nature’s Most Exquisite Creation”. Awesome photographic comparisons of eyes from different kinds of creatures and discussion of their evolutionary diversity.

The same can’t be said for other eye components. Take lenses. Almost all of them are made from proteins called crystallins, which improve their owners’ vision by focusing light onto underlying photoreceptors. But unlike opsins, with their single dynasty, crystallins are unified by name only. Yours are unrelated to those of a squid or a fly. Different animal groups have independently evolved their own brand of crystallins by co-opting proteins that had very different jobs, unrelated to vision: Some broke down alcohol; others dealt with stress. But all were stable, easy to pack, and capable of bending light—perfect for making lenses.

The eye is the evolutionary gift that keeps on giving to those of us who teach about homology and convergence.

A few years ago, historian of science and creationism Ronald Numbers did a great interview with Steve Paulson, which is on Salon: “Seeing the light — of science”. Numbers is the foremost academic expert on the history and growth of the creationism movement, both in the U.S. and globally, and he is one of my University of Wisconsin-Madison colleagues.

This interview looks at Numbers’ own history as a Seventh-Day Adventist, his journey from biblical literalism, and his historical perspective on the strategies of recent creationists.

Well, most people who reject evolution do not see themselves as being anti-scientific in any way. They love science. They love what science has produced. It’s allowed the conservative Christians to go on the airwaves, to fly to mission fields. They’re not against science at all. But they don’t believe evolution is real science. So they’re able to criticize one of the primary theories of modern science and yet not adopt an anti-scientific attitude. A lot of critics find that just absolutely amazing. And it’s a rhetorical game that has been played fairly successfully for a long time. In the latter part of the 19th century, when Mary Baker Eddy came up with her system that denied the existence of a material world — denying the existence of sickness and death, which flew in the face of everything that late 19th century science was teaching — what did she call it? “Christian science.” The founder of chiropractic thought that he had found the only true scientific view of healing. The creationists around 1970 took the view that’s most at odds with modern science and called it “creation science.” They love science! And they want to partake in the cultural authority that still comes to science.

Hat tip: Matt Sponheimer via Twitter.

Somebody was on Sulawesi before 118,000 years ago

Gerrit van den Burgh and colleagues reported in Nature this week that they have recovered artifacts from an early habitation of Sulawesi. Like Flores, Sulawesi was isolated from the mainland of Southeast Asia throughout the Pleistocene, so any humans who reached the island must have crossed Wallace’s Line by boat.

After the discovery of human remains at Liang Bua on Flores, and the much earlier archaeological evidence from the Soa Basin (“Earlier arrival of stone tools on Flores”), many speculated that hominins may also have reached Sulawesi at a similarly early date. The water crossing from Borneo to Sulawesi during the Pleistocene lowstand was substantially longer than the crossing from Bali to Lombok, but the Makassar Strait is long and parallel-sided, making a successful crossing relatively likely. A wider range of terrestrial mammals made the crossing to Sulawesi during the Pleistocene when compared to Flores, including macaques and pigs. All things considered, Sulawesi seems more likely human territory than other parts of Wallacea.

The late Mike Morwood set out to test the hypothesis by investigating stone tool industries on the island, and this work has been continued by van den Burgh and colleagues. Their current paper revolves around four artifacts recovered from a deep trench near the village of Talepu.

From T4 [Trench 4] we recovered 41 stone artefacts from the topsoil and colluvium down to a depth of 120 cm. However, four in situ silicified limestone artefacts were in exposed older strata within the silt of sub-unit E2 (Fig. 2c), and provide the stratigraphically earliest evidence for human activity at Talepu. Two are unmodified flakes (2.2–2.4 m depth) (Fig. 3l–m) and two are angular scatter fragments (3.0–3.1 m depth) (Fig. 3j–k), probably created in percussion flaking. The latter are made of a distinctive mottled silicified limestone and appear to have been removed from the same core. The artefacts bear no evidence of water transport; indeed, unit E did not yield any clasts indicative of high-energy water flow.

These artifacts are not diagnostic of any particular industry, so there is no possibility of connecting them to any culture. Their value is their depth.

The four samples analysed from T2 have ages in stratigraphic order 103 ± 9 kyr at 3 m depth to 156 ± 19 kyr at 10 m depth (Fig. 2, Methods: Optical Dating, and Supplementary Table 4). These results suggest that the Talepu cultural sequence ends at ~100 kyr, or possibly earlier (see Methods). The sediments dated to 156 ± 19 kyr were deposited near the top of unit D, which overlies the sedimentary layer (unit E) from which the deepest artefacts were excavated (more than 3 m below). The oldest securely dated evidence for stone artefacts at Talepu is, therefore, 118 kyr to 194 kyr in age at the 95% confidence interval (2σ), although human occupation of the site clearly occurred earlier given the recovery of artefacts from greater stratigraphic depths (Fig. 2c).

The deepest artifacts are tiny, a bit more than a centimeter in their longest dimension, and I wonder whether vertical movement of these flakes within the sediments may have occurred. Setting this possibility aside, the authors’ argument for a minimum age of 118,000 is otherwise very conservative—it marks the youngest end of the 95% confidence interval on a sediment layer that substantially overlies the unit containing the four deep artifacts. These four deep artifacts within unit E2 may be much older.

The paper ends with a discussion of who made the tools. The authors consider that early modern humans may have been in southern China before 100,000 years ago, so it seems barely possible that a very early modern human migration may have reached Sulawesi by 118,000 years ago. They discuss the hypothesis that an earlier population may have been present, by analogy with Flores.

In my view, it is premature to think that we understand much if anything about the first appearance of modern humans in Southeast Asia. There are too few pieces of evidence. None of the fragments yet found in China from earlier than 80,000 years ago are very persuasive of affinity with modern humans. They aren’t Neandertals, but in truth we have no idea what other populations of later archaic humans may have looked like, how much genetic diversity they may have had. Conceivably, a very early emergence of African or West Asian people—maybe even those evidenced at Qesem Cave—may have spread into India or Southeast Asia, bringing signs of affinities with a more modern-like population. We have no reason to assume that other populations, such as the Denisovans, would not be mistaken for modern humans, certainly based on the fragments that have so far been unearthed.

We know now that the Neandertals were arrayed across the western half of Eurasia, in a set of populations that collectively were highly endogamous and subject to strong genetic drift. Across a broad geographic range, such high inbreeding could only be accomplished by high mobility on a millennial scale, a population structure in which no long-isolated population persisted for long. This was a dynamic and adaptable population, probably relying upon ecotones at the edges of the wide Eurasian steppes, seemingly more successful in hills, mountains and forest edges than in open country.

Southeast Asia, including Indonesia, is a very different biogeographic story from Europe and Central Asia. At low sealevel, Sundaland together with Indochina formed a continental area larger in habitable area than Europe. Across this region the most uniform habitat is the coastline, which sometimes, cut through the interior creating dispersal corridors, as it does today. Inland forested areas in historic times have remained islands of high faunal and cultural endemism. These resource-rich environments persisted throughout the spread of rice agriculture, and are yielding only now to development, including industrial-scale palm oil plantations. It seems unlikely that the entire region was home to a single Neandertal-like population that maintained high inbreeding across its range by high mobility and dispersal. Instead, the vast land mass of Southeast Asia was probably home to many human populations of high regional endemism. The biogeographic islands within Southeast Asia may have retained some Homo erectus-like populations, but almost certainly the dispersal corridors across the region would have been dominated by events coming from the large population of the Indian subcontinent to the west.

I’m very enthusiastic about Sulawesi. It may be a beautiful test of the biogeography of early Homo across its southern range. If archaic humans were effectively using coastal habitat as a dispersal corridor, we may expect that they repeatedly reached Sulawesi—by 120,000 years ago, they may even have been in continuous contact.

Or if Southeast Asia was full of human populations with high endemism, some founded by Homo erectus-like populations, then Sulawesi may have been home to such a population. Unlike Flores, the resource base on Sulawesi was richer and island’s size would have enabled a relatively large human population, possibly large enough to avoid the mutational meltdown possibilities of the smaller island population.

Artifacts from a single deep trench do not establish either possibility, and so far the artifactual record across Southeast Asia as a whole gives hardly any hint of the cultural diversity that must have existed in the Middle Pleistocene. It is a large question mark. But the fact that artifacts turned up in a trench dug through sediments of the right age is promising. If the habitation of the island was intensive enough to salt the traces of human activity this densely into the earth, there is a good chance that further survey will turn up more interesting things.


van den Bergh GD, Li B, Brumm A, Grün R, Yurnaldi D, Moore MW, Kurniawan I, Setiawan R, Aziz F, Roberts RG, Suyono, Storey M, Setiabudi E, Morwood MJ. 2016. Earliest hominin occupation of Sulawesi, Indonesia. Nature 529:208–211. doi:10.1038/nature16448

Notable paper: Pitulko, V. V., Tikhonov, A. N., Pavlova, E. Y., Nikolskiy, P. A., Kuper, K. E., Polozov, R. N. 2016. Early human presence in the Arctic: Evidence from 45,000-year-old mammoth remains. Science 351:260-263. doi:10.1126/science.aad0554

Synopsis: Vladimir Pitulko and colleagues excavated an exceptionally well-preserved mammoth skeleton with characteristic damage from human hunting, from the northernmost end of the Yanisey River in Siberia. The bone is radiocarbon-dated to more than 45,000 years ago, making this the earliest evidence of human activity above the Arctic Circle. They also report on fauna from a site further east, Bunge-Toll, at around the same age, including a wolf with forelimb trauma that they interpret as a result of an injury from a human hunter.

Grisly detail: The hunters seem to have broken up the mammoth’s tusk in an effort to create sharp-edged ivory flakes that they could use to butcher the animal.

Not said: Were these Neandertals or modern humans? Neither find includes any artifacts. At least one high-latitude site further to the west and later in time, Byzovaya, has been interpreted as a terminal Mousterian that may have been made by Neandertals some 30,000–35,000 years ago, while a somewhat earlier site, Mamontovaya Kurya, has a marked piece of mammoth tusk, and has been likened to Szeletian (my earlier posts: “Who colonized the European Arctic?”, “Neandertals of the North”). The Ust’-Ishim femur shows that a modern human population without substantially greater Neandertal ancestry was present in the Ob drainage by 45,000 years ago (The Ust’-Ishim genome”), and a very early Upper Paleolithic was present in the Altai by 43,000 radiocarbon years ago, for example at Kara Bom. In light of this, it seems likely that one or more modern human populations had developed cultures with the logistical wherewithal to use the mammoth steppes, perhaps 20,000 years earlier than the mammoth hunters of Eastern Europe. The question about whether Neandertals were using the northern parts of the European Arctic remains unresolved.

Hominin species and time in peer review

In 2015, two new hominin species were published: Australopithecus deyiremeda and Homo naledi. One of the criticisms I’ve seen of both discoveries is the idea that they had not been given sufficient peer review. Since I was involved in the H. naledi discovery and analysis, this criticism took me by surprise.

So I have been taking some time to look at earlier hominin discoveries to look at the time they took to publish. This record provides a lot of evidence about how we have defined and published new species during the last 25 years.

What is the criticism?

Here is a section of Carl Zimmer’s New York Times article describing Au. deyiremeda last May: “The Human Family Tree Bristles With New Branches”.

But some hominid experts remain unconvinced that the road to Homo took so many turns. Tim D. White, a paleoanthropologist at the University of California, Berkeley, argues that most of the new studies have been rushed into publication without careful peer review.
The 3.3-million-year date for the ancient stone tools, for example, “seemed quite sketchy to me,” Dr. White said. The tools could have been made hundreds of thousands of years later, he said.
Dr. White is also skeptical that the new fossils represent a wealth of new species. He suspects that most of them, including Australopithecus deyiremeda, are just Australopithecus afarensis.
“Lucy’s species just got a few more new fossils,” he said of Wednesday’s announcement.

Many similar comments have been made about the Homo naledi publication, and White’s remarks here are not extreme. I wanted to point to this one because obviously this criticism is not specifically about H. naledi, nor is it about open access journals. Au. deyiremeda and the 3.3-million-year-old tools from Lomekwi 3 were both published in Nature.

Time from submission to acceptance

Is it true? Fortunately, we can consider some data from papers describing hominin species about the timeline of review and editorial decision. The paper describing Au. deyiremeda spent 6 months in the editorial process, while the Lomekwi 3 paper was in review from November 1, 2012 to April 13, 2015—more than 2 years and 5 months.

I’ve examined most of the research papers with new formal descriptions of hominin taxa since 1990, and have noted the date of submission and acceptance of each paper. The total list includes some with near-universal acceptance and some with relatively little acceptance from other professionals, but they all were peer-reviewed and satisfy the formal requirements for naming species under the International Code of Zoological Nomenclature.

Here are the data:

Time spent in peer review by hominin species diagnoses

The median time to acceptance for these papers is 70 days. The shortest time, 6 days, was taken by the paper describing Orrorin tugenensis in Comptes Rendus, while Homo floresiensis was 189 days, or more than 6 months, from submission to acceptance in Nature. Au. deyiremeda is the second longest time from submission to acceptance; H. naledi is in the second quartile with Au. garhi, Au. ramidus and H. antecessor.

I’ve looked at a number of ways in which these species descriptions might vary, and have found no obvious correlations yet with the time from submission to acceptance.

  1. The time does not seem to depend on whether the holotype of the taxon is being newly described, or whether it was described in a previous publication (possibly by other authors), as was the case for H. gautengensis and H. cepranensis.

  2. The time does not depend on the number of authors of the paper, which ranges from one (H. gautengensis) to 47 (H. naledi). Only 2 of the papers have more than 9 authors, and 9 of the papers have between 6 and 9 authors.

  3. Two of the taxa were later formally revised by their authors: Australopithecus ramidus to Ardipithecus ramidus and Ardipithecus ramidus kadabba to Ardipithecus kadabba. The editorial time for the initial descriptions was not unusually long or short compared to other species here.

  4. New taxa with Tim White on the author list do not have significantly different review times than those without him on the author list.

  5. One possible correlation: new genus names (Orrorin, Sahelanthropus, Kenyanthropus) are drawn from the shorter times in the distribution. Ardipithecus was named in a correction to White et al. 1994 that does not indicate the time of submission or acceptance, so I cannot test this one.

What can we learn from editorial time?

I’ve included several caveats below with the notes. Keeping these in mind, what does the editorial time tell us? The data can provide evidence concerning the question of whether a paper was “rushed into publication” as some critics may say. But this information is limited. The time from submission to acceptance is a measure only of time, and not necessarily quality or depth of review.

Nearly all of these papers are in three journals, Nature, Science, and Comptes Rendus, all known for a rapid review and editorial timeline. A referee for one of these journals typically agrees to return her review in a short time, sometimes as short as a week to 10 days. For experienced referees, returning a review in 10 days is a matter of ensuring they have time in their schedule, it is not a rush job. A careful review for a journal should not take longer than this.

If peer review shouldn’t take so long, why does it routinely take longer for most journals? I can comment on this from my own perspective as a working scientist and associate editor for PLoS ONE for many years. I schedule editorial and review time very deliberately. It is not every week that I can spare several hours to make a careful review of a manuscript checking for possible problems. As an editor, I often want the opinions of scientists who themselves have many competing demands for their time. When they graciously agree to look carefully at a manuscript, I work with them to do so on as reasonable a timeline as is possible. In my experience, the quality of reviews does not depend upon the time that a referee takes to return them. I have had work accepted in both Nature and Science, and work rejected by each of them. In cases where the work was sent for review, my experience is that these reviews are approximately the same length as reviews I’ve received from more specialized journals. I have only rarely had a referee for one of my papers for Nature or Science reveal her identity to me, but my perception is that they draw from the same pool of reviewers as more field-specific journals.

So I do not think we can draw conclusions from the data above about the quality of reviews, beyond the simple observation that the hominin species published in the past few years are not different in their editorial timeline from other species published as early as 1994. This may be the most remarkable fact. There is no trend evident in the data, and both the shortest and longest editorial processes happened in the middle of the time series. There were no “good old days” in which referees took months and months to carefully review the descriptions of new hominin species.

It’s surprising that the duration of the editorial process remains virtually unchanged despite the vast increase in the speed of communication during the last 25 years. In 1994 when the Australopithecus ramidus description was published, journal submission happened through the post, and peer review required sending papers by fax or post to referees. Ann Gibbons in her 2006 book, The First Human: The Race to Discover Our Earliest Ancestors, describes the submission and editorial process of the Au. ramidus description. A highlight of her story is her account of Nature editor Henry Gee meeting Berhane Asfaw in the lobby of a hotel in South Kensington, London, to be hand-delivered the manuscripts “like a character out of a John le Carré novel, receiving state secrets”. As Gibbons explains, the real motive for the face-to-face meeting was to save time. Today papers are submitted electronically, and instantly, but the process of peer review takes basically the same time.

It may surprise people that the more transparent referee process in eLife still yielded a time between submission and acceptance for Homo naledi that is only slightly less than the median for similar papers during the last 25 years. In my experience, the eLife process led to a faster consideration of the crucial issues with the study, but the referees spent the same time and effort that they would have done in any other journal.

Peer review in context

Peer review has become a cornerstone of the public’s trust in science. An expert can read an argument in a paper and evaluate it from the evidence it contains, but non-experts rely upon assessment by expert referees to confirm that a paper presents sufficient evidence to justify its conclusions.

Still, most professionals have at some time in their careers seen work published by colleagues with what seem to be obvious flaws. When this happens we sometimes ask, “How could that have gotten through peer review?” I’ve done it several times on the blog, focusing on specific criticisms of published papers. For example, in 2009, I called Science “a rinky-dink journal” for allowing the publication of a description of Ardipithecus ramidus material without including standard dental measurements (“Whoa, who stole the data?”). I have sometimes noticed misleading figures in a published paper and wondered how they were not caught and corrected in peer review. Improperly manipulated figures are one of the most common reasons for retraction of scientific papers, so peer reviewers should be especially on the watch for them.

Part of being a scientist is being willing to answer critics who challenge the reliability of research. Anyone who has been a referee for many papers has probably given advice to authors or an editor that was not ultimately taken, sometimes letting through what seem like obvious flaws. Peer review cannot remove all errors; and if we expect peer review to make papers error-free, we are expecting far too much.

But it’s obvious that part of being an effective critic is to obtain the best information and evidence possible. There is no value to blanket assertions that today’s papers are not receiving adequate peer review. The evidence is the opposite: the time and effort in peer review of new hominin taxa has remained basically the same across at least the past 25 years.


  1. I’m waiting on a physical copy of the description of Australopithecus bahrelghazali by Brunet and colleagues (1996), so I haven’t yet included this species.

  2. I’ve listed all taxonomic names based on their first formal description. Some of these have been revised by their original authors.

  3. The list includes both new species based on never-before-described material and new species based on previously-described material.

  4. In my experience publishing scientific papers, the dates for submission and acceptance provided by a journal are often unreliable. It is in the journal’s interest to advertise a rapid publication timeline, so that scientists will be more likely to submit important, time-sensitive work. So journals sometimes print the date only of the most recent or final submission, and do not provide the date of initial submission. They almost never indicate the date of initial editorial inquiry, which may precede formal submission by many months. As a result, the data available for these papers are probably biased to a shorter time of review and editorial guidance than the authors actually experienced.

  5. Also, some of these papers were first submitted to a different journal. The Homo naledi description, for example, was submitted in a different form to another journal before its submission to eLife. In such cases, the editorial timeline in the journal of publication does not reflect the time spent in peer review and editorial correspondence.

  6. The reference list below includes a much longer set than cited in this post; I will be continuing this inquiry with papers from earlier in the history of the field.


Arambourg, C. (1954). [The fossil man of Ternifine (Algeria).]. Comptes rendus hebdomadaires des seances de l'Academie des sciences, 239(15), 893-895.

Ascenzi, A., Biddittu, I., Cassoli, P. F., Segre, A. G., & Segre-Naldini, E. (1996). A calvarium of late Homo erectus from Ceprano, Italy. Journal of Human Evolution, 31(5), 409-423.

Asfaw, B., White, T., Lovejoy, O., Latimer, B., Simpson, S., & Suwa, G. (1999). Australopithecus garhi: a new species of early hominid from Ethiopia. Science, 284(5414), 629-635.

Berger, L. R., de Ruiter, D. J., Churchill, S. E., Schmid, P., Carlson, K. J., Dirks, P. H., & Kibii, J. M. (2010). Australopithecus sediba: A new species of Homo-like australopith from South Africa. Science, 328(5975), 195-204.

Bermúdez de Castro, J. M., Arsuaga, J. L., Carbonell, E., Rosas, A., Martınez, I., & Mosquera, M. (1997). A hominid from the Lower Pleistocene of Atapuerca, Spain: possible ancestor to Neandertals and modern humans. Science, 276(5317), 1392-1395.

Broom, R. (1936). New fossil anthropoid skull from South Africa. Nature, 138, 486-488.

Broom, R. (1938). The Pleistocene anthropoid apes of South Africa. Nature, 142, 377-379.

Broom, R. (1949). Another new type of fossil ape-man. Nature, 163, 57.

Broom, R., & Robinson, J. T. (1949). A new type of fossil man. Nature, 164, 322-323.

Brown, P., Sutikna, T., Morwood, M. J., Soejono, R. P., Saptomo, E. W., & Due, R. A. (2004). A new small-bodied hominin from the Late Pleistocene of Flores, Indonesia. Nature, 431(7012), 1055-1061.

Brunet, M., Beauvilain, A., Coppens, Y., Heintz, E., Moutaye, A. H., & Pilbeam, D. (1995). The first australopithecine 2,500 kilometres west of the Rift Valley (Chad). Nature 378, 273-275.

Brunet, M., Beauvilain, A., Coppens, Y., Heintz, E., Moutaye, A. H. E., & Pilbeam, D. (1996). Australopithecus bahrelghazali, une nouvelle espèce d'Hominidé ancien de la région de Koro Toro (Tchad). Comptes rendus de l'Académie des sciences. Série 2. Sciences de la terre et des planètes, 322(10), 907-913.

Brunet, M., Guy, F., Pilbeam, D., Mackaye, H. T., Likius, A., Ahounta, D., ... & De Bonis, L. (2002). A new hominid from the Upper Miocene of Chad, Central Africa. Nature, 418(6894), 145-151.

Curnoe, D. (2010) A review of early Homo in southern Africa focusing on cranial, mandibular and dental remains, with the description of a new species (Homo gautengensis sp. nov.). HOMO - Journal of Comparative Human Biology, 61, 151–177.

Dart, R. A. (1948). The Makapansgat proto‐human Australopithecus prometheus. American Journal of Physical Anthropology, 6(3), 259-284.

Dirks, P. H., Kibii, J. M., Kuhn, B. F., Steininger, C., Churchill, S. E., Kramers, J. D., ... & King, G. C. (2010). Geological setting and age of Australopithecus sediba from southern Africa. Science, 328(5975), 205-208.

Gabounia, L., de Lumley, M. A., Vekua, A., Lordkipanidze, D., & de Lumley, H. (2002). Découverte d’un nouvel hominidé à Dmanissi (Transcaucasie, Géorgie). Comptes Rendus Palevol, 1(4), 243-253.

Groves, C. P. (1999). Nomenclature of African Plio-Pleistocene hominins. Journal of Human Evolution, 37, 869-872.

Haile-Selassie, Y. (2001). Late Miocene hominids from the middle Awash, Ethiopia. Nature, 412(6843), 178-181.

Haile-Selassie, Y., Suwa, G., & White, T. D. (2004). Late Miocene teeth from Middle Awash, Ethiopia, and early hominid dental evolution. Science, 303(5663), 1503-1505.

Haile-Selassie, Y., Gibert, L., Melillo, S. M., Ryan, T. M., Alene, M., Deino, A., ... & Saylor, B. Z. (2015). New species from Ethiopia further expands Middle Pliocene hominin diversity. Nature, 521(7553), 483-488.

Hughes, A. R., & Tobias, P. V. (1977). A fossil skull probably of the genus Homo from Sterkfontein, Transvaal. Nature, 265(5592), 310-312.

Leakey, L. S. B. (1959). A new fossil skull from Olduvai. Nature, 184(4685), 491-493.

Leakey, L. S. B., Tobias, P. V., & Napier, J. (1964). A new species of Homo from Olduvai Gorge. Nature, 202, 7-9.

Leakey, M. G., Feibel, C. S., McDougall, I., & Walker, A. (1995). New four-million-year-old hominid species from Kanapoi and Allia Bay, Kenya. Nature, 376(6541), 565-571.

Leakey, M. G., Spoor, F., Brown, F. H., Gathogo, P. N., Kiarie, C., Leakey, L. N., & McDougall, I. (2001). New hominin genus from eastern Africa shows diverse middle Pliocene lineages. Nature, 410(6827), 433-440.

Leakey, R. E. F., & Wood, B. A. (1973). New evidence of the genus Homo from East Rudolf, Kenya. II. American Journal of Physical Anthropology, 39(3), 355-368.

Mallegni, F., Carnieri, E., Bisconti, M., Tartarelli, G., Ricci, S., Biddittu, I., & Segre, A. (2003). Homo cepranensis sp. nov. and the evolution of African-European Middle Pleistocene hominids. Comptes Rendus Palevol, 2(2), 153-159.

Oppenoorth, W. F. (1932) Homo (Javanthropus) soloensis, en Pleistocene mensch van Java. Wetenschapelijke Medische Dienst Mijnbouw Nederlandsch Indië 20, 49–74.

Senut, B., Pickford, M., Gommery, D., Mein, P., Cheboi, K., & Coppens, Y. (2001). First hominid from the Miocene (Lukeino formation, Kenya). Comptes Rendus de l'Académie des Sciences-Series IIA-Earth and Planetary Science, 332(2), 137-144.

White, T. D., Asfaw, B., DeGusta, D., Gilbert, H., Richards, G. D., Suwa, G., & Howell, F. C. (2003). Pleistocene Homo sapiens from Middle Awash, Ethiopia. Nature, 423(6941), 742-747.

White, T.D., WoldeGabriel, G., Asfaw, B., Ambrose, S., Beyene, Y., Bernor, R.L., Boisserie, J.R., Currie, B., Gilbert, H., Haile-Selassie, Y. and Hart, W.K., 2006. Asa Issie, Aramis and the origin of Australopithecus. Nature, 440(7086), pp.883-889.

White, T. D., Suwa, G., & Asfaw, B. (1994). Australopithecus ramidus, a new species of early hominid from Aramis, Ethiopia. Nature, 371(6495), 306.

WoldeGabriel, G., White, T. D., Suwa, G., Renne, P., de Heinzelin, J., Hart, W. K., & Heiken, G. (1994). Ecological and temporal placement of early Pliocene hominids at Aramis, Ethiopia. Nature, 371(6495), 330-333.

Woodward, A. S. (1921) A new cave man from Rhodesia, South Africa. Nature 108, 371–372.