Submerged Continental Shelves: the corridors of Hominin migrations

When we look back at the first humans as they spread across the globe, we are usually constrained by the coastal areas marked on maps. But people didn't only move through the heart of continents, they also trekked along their margins, by the coast, using the resources provided by the sea.

During the Ice Ages, vast quantities of water was trapped in the thick ice shields that covered the polar parts of the world, from 1.5 to 3 km (1-2 mi.) thick, the ice held so much water that the sea levels dropped, exposing areas now submerged along the continental shelves around the globe.

These areas were dry land for thousands of years. The exposed seabed thrived with plants and animal life. Trees, grasslands, rivers that flowed across them offered an environment suitable for our ancestors to live.

Some of the best known exposed continental shelves are: Beringia (spanning Bering's Strait between northeastern Siberia and alaska), Sahul (encompassing Australia, Tasmania, and New Guinea, Sundaland which included insular South East Asia: Borneo, Java, Sumatra, Flores, Bali, and the Malay Peninsula), and Doggerland (now submerged by the North Sea between Britain, Denmark, Belgium, Germany, and the Netherlands).

But there were many more spots that were used by humans, and also by our ancestors, the Denisovans and Neanderthals, to migrate, hunt, and live.

The forgotten Submerged World of our Ancestors

During the peak of each Ice Age, sea levels dropped up to 120 meters (400 ft.) and this exposed up to "15–20 million km² of land" (5.8 to 7.7 million square miles) this is equivalent to 50-66% of the surface of Africa, two to three times the surface of Australia, or two to three times the area of the lower 48 states of the U.S. (21 to 28 times the area of Texas).

Ice age sea levels. Source.

The map above shows a large area along the north coast of Siberia, Europe and Canda, but this was under the ice sheets, and the map with its Mercator projection tends to inflate the size of polar areas and diminsh those of mid and equatiorial latitudes. As you can see, Australia, Sunda, Japan-Taiwan-Korea, Southern South America, the mouth of the Amazon River, the persian Gulf, the Aegean Sea, the Gulf of Mexico, and the Atlantic Seabord of USA have the largest emerged areas.

The area of the continents is around 148 million km², so the ice ages exposed an additional 10 to 13.5% of dry land, now submerged beneath the waves. A territory that is unexplored.

Archaeology under the sea is almost non-existent, and except for the rare artifact brought up by chance during fishing (like the Denisovan jawbone Penghu-1 from Taiwan) or the Homo erectus bones dredged by Berghuis, 2025 in Java) we know little of what exists on the continental shelves.

A paper by Norman et al., 2024 described the now submerged shelf along the northwestern coast of Australia, modelling its environment and population. The authors mention the importance of subaquatic archaeology, and also noted that the flooding after the end of the last ice age was very fast, sea levels grew by 1 meter (3 ft. 4 in.) per century and accelerated to 4-5 m (13-16 ft) per hundred years for over 400 years around 14.1 kya. This was followed by a second flood, that lasted 3000 years, beginning 12 kya, which was also fast. The flooding pushed the people living on the sea shores into the interior of the continent, a period of stress and adaptation to new enviornments and resources which is reflected in the rock art styles that changed and seem to depict "conflict/battle scenes."

Americas

The submerged coastal areas of North America along the shores of Canada, Alaska, Washington, Oregon, and California surely contain sites with artifacts and campgrounds of the first people to reach America. Although narrow, this shelf holds evidence about the earliest humans in America. On the Atlantic coast, the continental shelf is far wider. In a recent post I mentioned the 22,000-year-old Cinmar site off the coast of Virginia, in the Atlantic Ocean, 47 mi (75 km) from Chesapeake Bay.

The Pacific coastline, is steeper in Central and South America, as the Andes run very close to it, and the seabed plunges into the ocean depths fairly quickly. But on the Atlantic coastline of Argentina, Uruguay and southern Brazil, there was a vast flat area that stretched for hundreds of kilometers eastwards from the current sea shore. Below is a map that depicts the current, and the now submerged terrain of South America (the white dots are sites where pollen was collected for an analysis of the plant coverage).

South America during the LGM. Source

The paper in which the map shown above, was published, by Díaz Pinaya et al., 2024, describes the "Southern Atlantic Continental Shelf Connectivity (SACS)" had trees of many species, like "Araucaria and Drimys, forming plant assemblages with Arecaceae, Ericaceae, Ilex, Myrsine, Myrtaceae, Podocarpus, Symplocos and Weinmannia. It includes Atlantic Rainforest sites extending from the Mantiqueira Mountains, the modern cerrado region of Minas Gerais into the Serra do Mar highlands along the coast of southeastern and southern Brazil. Considering the LGM landscape, this floristic connectivity extended from the Central South American Connection (CSAM) onto the then-exposed Atlantic Shelf spanning from 23 to 56°S in latitude. We hypothesize that this large exposed area was vegetated by cold and humid successional forests as a consequence of the downward migration of montane taxa into the coastal lowlands, a scenario supported by pollen records." During this period, the araucaria of the Paraná region (not the Patagonian variety of monkey-puzzle) is believed to have expanded "into Uruguay and northern Argentina, especially onto the exposed continental shelf. Drimys, on the other hand, appears to display high niche suitability centered in southern/southeastern Brazil, on the Argentinian shelf and in a large area between 36 and 48°S on the Pacific coast of Chile. Drimys winteri... an Andean endemic, could have populated the latter two regions"

This paper estimates the emerged continental shelf along the coasts of South America at 1.94 million km², which is "equivalent in size to the combined areas of France, Spain, Germany, Italy and the United Kingdom and that in southeastern/southern Brazil/Uruguay and Argentina the past coastline was 200–250 km and 500 km distant from its present location, respectively."

The Araucaria and Drimys (in the Patagonian forests, the latter survives as the Winter Bark, or Canelo, Drimy winteri) would have expanded along the southern part of the shelf, colder and wetter. In the other parts, further north it would have been similar to the present day Atlantic Rainforest with different levels of trees, possibly five tree layers with different tree-heights, from 50 meters (150 ft) to tree ferns and shrubs at the lowest level. The authors point out that "the South American Continental Shelf, which could have created an important migration corridor for different southern Andean plant species to migrate northwards and colonize areas of the Brazilian Atlantic coast. Our data suggests that this vast coastal corridor was possibly covered by temperate-like forest with prevalent Andean floristic affinities in the south and a more Atlantic floristic composition in the north."

This scenario of forests and jungles is very different from the grasslands and prairies of Uruguay, and the Pampas region of Buenos Aires in Argentina, and the arid steppes of Patagonia. The people living in South America during this period (if we accept them having reached the region 25-30 kya) would have enjoyed a rich environment that is now totally submerged. I

The difficulties and limitations of underwater archaeology

Unfortunately, most of this submerged area is part of Argentina's continental shelf. Research funds are limited and scarce in Argentina, so very few underwater sites have been studied (see Ciarlo, 2009). Most of the underwater sites involve post-European-Discovery sites like ships or inland places in lakes. In Uruguay, where the situation is similar, there have been "increasing discoveries of archaeological material from submerged sites redeposited on beaches by storm events" (see López Mazz, 2004) these findings include fossils of megafaunal animals from the former deltaic area of the River Plate, now submerged by the sea.

Underwater archaeology is complex, expensive, and requires scuba diving gear and special equipment such as pumps, support shipes, etc. Easton, Moore, and Mason, 2020 (free access) describe the experience along the coastal area of British Columbia in Canada, showing that although it is feasible, it will not become mainstream in the forseeable future.

Underwater sites often lack stratigraphic context, and objects dredged up from the seabed are isolated from their original emplacement making it hard to reach useful conclusions about them. For instance the "Chinese anchors" dredged off the ocean floor close to Catalina Island, California in November 1973 have been controversial. Brought up from the bottom by a United States Geological Survey expedition from the Patton Escarpment, the donught-shaped stone was believed to be part of Chinese Junk dated to 3,000 years old. It was followed by the discovery of stones shaped like Chinese anchors at Palos Verdes Peninsula, close to Los Angeles by Baldwiyn and Meistrell in 1975 (see these two references: Mc Ginty, 1983 p. 54, and University of San Diego, 1980 p. 4). However, Frank Frost, 1982, suggested they were recent, not more than 100-years-old.

Comments

With the main migration corridors, and sites linked to the shores of lower sea levels, now submerged and out of reach, how can be so sure about the dates for the movement of people around the world?

It is reasonable to assume that the first migrants exploited the coastal environments before spreading inland into the continents, yet we can only access sites located on dry land, which are certainly younger than those under the sea.

---

Patagonian Monsters - Cryptozoology, Myths & legends in Patagonia Copyright 2009-2026 by Austin Whittall © 

The earliest modern humans outside Africa: more older than expected

For the last year, more papers have been published suggesting an earlier age for Homo sapiens and an earlier Out of Africa migration. Today I read a paper in Science that pushes our "Out of Africa" event even further back in time (not that I believe in the OOA event) in other words Humans were in Asia long before the date currently accepted for that event (did we originate in Asia and move into Africa?).

This is the paper: The earliest modern humans outside Africa, Israel Hershkovitz et al. Science 26 Jan 2018:Vol. 359, Issue 6374, pp. 456-459 DOI: 10.1126/science.aap8369

And this is the free info (the rest is behind Science magazine's paywall:

Earliest modern humans out of Africa
Recent paleoanthropological studies have suggested that modern humans migrated from Africa as early as the beginning of the Late Pleistocene, 120,000 years ago. Hershkovitz et al. now suggest that early modern humans were already present outside of Africa more than 55,000 years earlier (see the Perspective by Stringer and Galway-Witham). During excavations of sediments at Mount Carmel, Israel, they found a fossil of a mouth part, a left hemimaxilla, with almost complete dentition.
The sediments contain a series of well-defined hearths and a rich stone-based industry, as well as abundant animal remains. Analysis of the human remains, and dating of the site and the fossil itself, indicate a likely age of at least 177,000 years for the fossil—making it the oldest member of the Homo sapiens clade found outside Africa.
Science, this issue p. 456; see also p. 389

Abstract
To date, the earliest modern human fossils found outside of Africa are dated to around 90,000 to 120,000 years ago at the Levantine sites of Skhul and Qafzeh. A maxilla and associated dentition recently discovered at Misliya Cave, Israel, was dated to 177,000 to 194,000 years ago, suggesting that members of the Homo sapiens clade left Africa earlier than previously thought. This finding changes our view on modern human dispersal and is consistent with recent genetic studies, which have posited the possibility of an earlier dispersal of Homo sapiens around 220,000 years ago. The Misliya maxilla is associated with full-fledged Levallois technology in the Levant, suggesting that the emergence of this technology is linked to the appearance of Homo sapiens in the region, as has been documented in Africa.

End of Science info.

So now we have a date of 177 to 194,000 years BP for humans in the Middle East. Meaning they got there even earlier -or did by chance did the teeth belong to the first arrival?. Now I ask why suppose that the H. sapiens went "OUT" of Africa instead of reaching Israel from some place in Asia and then going "INTO" Africa.

As usual, more discoveries will force change on closed minds

By the way, Happy New Year and a great 2018 for all of you.

---

Patagonian Monsters - Cryptozoology, Myths & legends in Patagonia Copyright 2009-2014 by Austin Whittall © 

Biases in Genetic Models that are generally overlooked

Before continuing with the Y chromosome C haplogroup and its peculiar global range, I want to dedicate today's post to "Models" in general (and those used in genetics in particular), and how they shape the way we believe things work.

As an engineer I have a clear notion that a model is a simplification of the real world, we make assumptions and find that models allow us to make predictions that are reasonably similar to the real world. This allows us to build a bridge that will not collapse and yet use a minimum amount of steel, or shoot a cannon ball from A and hit a target in B. The models can be more complex (relativity is taken into account to make your GPS work correctly and quantum mechanics in all things electronic), but all of them are just that: a "model", an approximation to reality, not reality itself, merely a simplification of the real world.

Models and Genetics a critical overview

Let's review some causes of errors and the hidden biases in genetic models and methods:

Populations

Some studies use strange populations such as "MXL, Mexican Ancestry from Los Angeles USA" or "ACB, African Carribbeans in Barbados", [10] which are far less significant than a native aboriginal individual in his or her homeland: such as an "Alakaluf from Magallanes, Chile" for instance.

What conclusions can be reached on the peopling of America by looking at the genome of a person living in LA, of Mexican ancestry? Mexico has several native groups, later overalaid by Spanish conquistadors (Spanish are themselves a mixture of many ethnic groups... aboriginal Iberians, Basques, Celts, Romans Carthaginians, Arab invaders, Goths, etc.), the African slaves they brought over to work in their plantations and a touch of other European and Southern and Mesoamerican groups. In other words the Mexicans are very admixed population. MXL are actually irrelevant or former slaves of African origin living in Barbados!

Then we have studies that ignore the New World completely. Samplings cover populations in Africa and Eurasia (sometimes including Australia and PNG), seldom the Americas. This of course limits the usefulness of these studies and maybe conceals interesting findings that will remain ignored until American samples are contrasted with Old World ones.

The Sampling within the populations

Once the populations have been identified (with all the caveats mentioned above), its ancestry is studied by means of a sample of individuals that in theory (but not in practice) is drawn from it in a random manner.

In other words a sample of size "n" is taken from a population with a size "N". In most populations "N" is several orders of magnitude lager than "n" (imagine a sample of 200 Italians from Tuscany out of the 61 million people living in Italy). This implies that we may have a sampling bias, and leave out some unique or even critical genetical sequences that appear at low frequencies in a given population.

In other cases the sample is not a random one (Ascertainment bias); for instance, in the case of small tribes: the sample "n" is small because "N" is also small and perhaps encompasses a clan or family group. Thus diversity is low and, as we will see below, calculations based on these samples will be affected by this sampling bias.

Ascertainment bias also arises when researchers take samples from databases in which some populations are missing, or some regions are under represented. Or from samples obtained from volunteers (i.e. a University class) which are definitively non-random samples.

The Typing of Haplogroups in those samples

Once we have sample of our population, then we sequence the Y chromosomes (or in the case of mtDNA, the mitochondrial DNA) for known markers (remember this: known, we will get back to it later) and type the individuals based on these markers. We don't read the whole sequence of 60 million base pairs in a Y chromosome and compare them all. Instead we choose certain ones which we believe are markers and base our analysis on these markers. This introduces another ascertainment bias: the choice of markers is not random.

Look at it this way, we have a million books printed in different languages, we take a sample of fifty from the pile of books and compare them based on a few of the words printed in them: If we find the word "Chapter" we will place them in the English group, "Capítulo" in the Spanish one, "Haupstück" in the German one. All the other words are ignored. We may even find a book printed in Armenian that by chance carries the word "Chapter" and we will place it in the "English" group and ignore that an "Armenian" group even exists. We may have not sampled a Chinese book and also ignore its existence.

Yes, the analogy is faulty (well it is a model after all), the markers we look for in genetics are placed in specific locations within a chromosome. We compare the markers in specific positions with each other to define similarity. In the books example the analogy falls through because the word Chapter can appear on any page and not on a specific page. Yet, what I wanted to point out is that even though there are millions of base pairs in a chromosome, we only use a handful to define haplogroups.

Let's look at this in depth:

Y chromosome SNPs, the Haplogroups

First of all, the Y chromosomes are sequenced and the Single Nucleotide Polymorphism (SNP) that identifies a given haplogroup (hg) is identified. An example of an SNP is shown below:

• AAGCCTA - ancestral
• AAGCTTA - derived

The fifth nucleotide, the "C" in the ancestral variant, mutated to a "T" in the derived variant. This SNP ocurrs in a specific location. These mutations are assumed to be apparently rare, and once they happen, they remain in the DNA and are inherited by all descendants of the mutated individual. This allows us to build trees based on which SNPs are present in the populations.

The SNP that identifies the whole Hg C is marker RPS4Y711. The different haplotypes also have their own markers: M8 (for C1), M38 (for C2), M217 (for C3), M93 (for C3a), and so on.

But, as I mentioned in a previous post, it is possible that these SNP mutations revert (in that post I mention the Motala remains sequenced for a Q haplotype which lacked a key marker but had all the others), so this may also introduce additional errors.

Haplotypes and Paragroups

So, in our sample which already may have a sampling bias, we will find that most individuals will belong to a given haplogroup (i.e. Y chromosome hg C) and to a given haplotype (i.e. if they tested positive for marker M38 we will be confident that they belong to the C2 haplotype).

But... Some cases will not test positive for the known markers (i.e. M8 -excluding C1, M217 -excluding C3, M347 -excluding C4 or M356 -excluding C5), so we will assume that they form a paragroup that includes all C-other lineages, and we will identify it as C* (with an asterisk).

So these pargroups clump together a large variety of haplotypes for which we have not yet identified specific "unique" markers which would set them apart as new haplotypes.

Hidden diversity

This is not a trivial matter. For instance paragroup C* is found all across the Eastern edge of the Old World, from Australia, across Indonesia, China, Japan and India, to Bering at frequencies that range from 0.3% to 10% of the population. (some are even higher: C3* among Mongols reaches a frequency of 18% [6]).

To assume that they all belong to the same group is erroneous, the paragroup masks subclades (haplotypes) which have not yet been discovered: Maybe C* in Australia is a not yet identified C8 hg, while C* in Central Asia is a yet undiscovered C9 hg for example. In other words, there is hidden diversity out there waiting to be discovered.

Bias in the choice of the markers

The SNPs are not chosen in a random manner, they are defined by geneticists to type haplogroups. This means that the real distribution of polymorphisms in a population may differ from those shown in a study. The reason is simple: The genotyping arrays (or chips)used to identify the markers contain biased sets of pre-ascertained SNPs. These SNPs tend to be older than the majority of the SNPs in a given population, and are found in many populations besides the one being studied. These hand-picked SNPs act as sieves, classifying the samples and causing alterations such as "[shifts in] allele frequency distributions ... towards intermediate frequency alleles", furthermore, "estimates of linkage disequilibrium are modified" [11].

In other words, this increases the frequency of the most commonly polymorphic loci and eliminates other markers (loci that are less polymorphic in the screening panel). This ascertainment bias in the SNP arrays strongly skews the estimates of genetic diversity by ignoring those that are not included in them.

Variety within the Haplotypes

Once the haplogroup and haplotype have been identified, we can take a look at the Microsatellites to check out even further diversity. Microsatellites are repeats (2 to 6 nucleotides long) that repeat "n" times ( n= 5 to 100). An example: the sequence "AT" repeated 25 times in a row (this is expressed as follows: (AT)₂₅).

These microsatellites are found across species and mutate quicker than single point mutations (SNPs) and for this reason they are used as markers to define the subclades within haplotypes. An example would be a mutation in (AT)₂₅ to (AT)₂₄ or to (AT)₂₆.

In the case of our Y chromosome, we will use special microsatellites known as Short Tandem Repeats (STR). These are named as "DYS-Number" (i.e. DYS393) which indicates a position for the STR. The STR, is a series of repeats of a dinucleotide (two nucleotides).

Below is a real example, for haplogroup C, for four individuals, two from Colombia, one Korean and a Kalkh:

A real set of STRs for four individuals. Copyright © 2014 by Austin Whittall

We can see in the image above that some DYS markers differ in the quantity of repeats (they are shaded pale blue and yellow).

The Evolutionary sequence

Now come the questions: Does the Korean derive from the Colombians or is it the other way round? What about the Mongolian? Note the differences at DYS392 and DYS393 between Koreans and Colombians (in yellow) and the difference between Colombians and Mongolians at the other DYSs (pale blue).

Without an A priori theory you can't answer that question just by looking at the repeats. Some additional assumptions are necessary, and computer programs are used to build the phylogenetic trees that link these individuals.

We could assume that humans came from Asia and peopled America, so the Americans are more recent than Asians. And as Koreans are Asians, they predate the Colombians so they accumulated mutations, passing from 11 to 12 in DYS392 and 13 to 15 in DYS393. So far so good, but then we have a Khalkh, from Mongolia, who are also Asians, but have less accumulated mutations than the Colombians or the Koreans. This way of comparing STRs is faulty.

Building a Phylogenetic Tree

The individuals are placed on phylogenetic trees using other assumptions that consider the differences between individuals, and are based on the different STRs, but it uses a different reasoning process to the one used above.

Distorting elements

Nevertheless we must remember that each DYS may have its own mutation rate, so if there are several DYS that differ, then they all have to be considered to calculate the "distance" between individuals.

An additional complication is that the "real" evolutionary history of any given set of individuals may differ from the "inferred" evolutionary history. As can be seen in the following image where only three (3) mutations out of twelve (12) are detected during analysis. The other nine (9) are ignored and remain undetected. This affects the estimations on divergence since the mutations are underestimated and the split time between the ancestor and its descendants is underestimated:

How mutations are Underestimated. Adapted from Fig. 2 in [4]

So "corrections" are introduced such as the Jukes-Cantor Model [4] which fiddle with the equations used in the models to make them fit better to reality.

So, how are the differences calculated?

Computer algorithms add even more assumptions

Algorithms are used. They are run on computers and compare the individuals in a pair-wise manner. There are several algorithms, each with their pros and cons. The two basic classes are:

• Distance-Based Methods - Neighbor Joining (NJ), which initially form an unresolved star-like tree and compare the branch length sum in a pairwise manner. It then groups as "close" those that have the minimum sum. This pair is linked in a branch and then the process begins again, iterating until all individuals have been grouped.
• Maximum Parsimony method, uses certain features (substitutions in the sequences) to work out a most likely evolutionary relationship among individuals. It builds the tree using the least substitutions chain from the common ancestor to the individuals being located on the tree. So it scores each possible option and minimiizes the mutation number to buid the tree.

The trees are then rooted by comparing them with some outgroup species (ie. chimpanzees are used for human and hominin comparisons). Of course this requires the assumption that molecular clocks are valid and that the divergence date with the outgroup species is well known (more on this below - see clock ticking out of time).

An example: Batwing

Batwing [8] (acronym which stands for Bayesian Analysis of Trees with Internal Node Generation), is a widely used computer program for analysis of genetic data. It has some implicit assumptions that I list below which are not mentioned in the papers that use it, but which impact on the outcome of the program's analysis. By the way, the authors of the program clearly point out that "Natural populations are unlikely to satisfy BATWING's modelling assumptions" [8].

• The data is a random sampling from the population (we have seen above it is not usually the case)
• The population is panmitic (not frequent in human groups)
• Splitting between populations is instantaneous (actually it takes plenty of time)
• There is no subsequent migration events between populations (there is always posterior admixture due to migrations between populations that have split)

Batwing uses different mutation models, but the default setting is the Stepwise Mutation Model or SSM, which we analyse in detail below.

Comment, the TMRCA (time to most recent common ancestor), Ť, is calculated under the Simple SSM model using the expression: Ť = Δ ⁄ μ.

Where Δ is the average squared difference in the number of repeats between all sampled Y chromosome and the founder haplotype, averaged over STR loci, and μ is the Simple SSM mean mutation rate per generation averaged over loci. But, if, as we will see below SSM is not very reliable, then how can clade age estimates be reliable?.

The Stepwise Mutation Model

This Stepwise Mutation Model (SSM) [2] was proposed in 1973 by Ohta and Kimura and has been widely adopted as the model for microsatellite evolution:

Microsatellites are believed to evolve neutrally: natural selection does not influence the number of repeats so, the SMM premise is: "In one generation the repeat number can only increase or decrease by one, and the probability is equal".

But this assumption is not exactly so for several reasons:

• Actually, the probability of mutation is larger for longer microsatellites [1][3].
• A long set of repeats (n larger than 20) may cause physical instability in the microsatellites and hamper its further growth, actually leading to their contraction. [3][2]
• Some microsatellites are interrupted and have lower mutation rates.
• The repeat unit also influences mutation rates: dinucleotides mutate slower than tetranucleotides.
• The motif of the dinucleotide (i.e. TG vs. TA) also plays a role: certain motifs are much longer than others.
• Variable mutation rates (those that change the repeat by more than 1) are not uncommon and happen about 15 to 22% of the time [3], in other words, the model ignores a big chunk of mutations.

Add to this that insertions or deletions next to the microsatellites also influence their lenghts. [3]

Even the "neutrality" of satellites is questionable since some repeats take place in promoter regions and may influence protein building [3] and thus be subject to natural selection. Some microsatellite repeats have been linked to certain diseases (myotonic dystrophy, Huntingtons' disease, etc.) making their neutrality doubtful too.

There are also "point mutations" that interrupt a repeat; an example: (AT)₁₈ may suffer a chance point mutation where "A" mutates to "G" in position 10, causing the new sequence to be: (AT)₉ GT (AT)₈. Transormation which may go undetected in sequence analysis, altering the mutation rate estimates.

Panmitic populations

Last but not least, the SMS model assumes that the individuals come from a random sample form a single panmitic population of constant size "N", and this is not the case. [5] Application to expanding populations or those with mixing due to migration may provide different results.

A panmictic population allows random mating without any restrictions of any kind (due to age, genes, behavior, social, environment, etc.), which is seldom the case in human populations, past or present.

Migrations

When an ancestral population splits into two groups, they are subjected to two opposing processes (see image below):

How genetic drift and migration affect allele frequencies. Copyright © 2014 by Austin Whittall

• Genetic Drift. It arises because N_e (number of effective breeders) which contribute their genes to the next generation is smaller than the total population, so they pass on their genes only and since this is a random sampling process, the frequency of these genes will differ from that of the previous generation. The smaller N_e, the larger the drift. This effect accumulates with each successive generation and separates the diverging subpopulations as time passes.

• Migration. Exchanges between the populations as they diverge will limit the drift, keeping them similar. The proportion of migrants "m" if larger will have a higer impact on stability.

Comment on drift and lack of migration: The "Beringian Standstill" was invented to justify the strangely unique American haplogroups, completely absent in the purported Asian homeland of the Native Americans.
The Standstill theory first suggested by Bonatto & Salzano (1997) and perfected by Tamm et al., 2007, is based on a one-in-a-million "Founding effect" that isolates a group of "founding fathers" in Beringia, cut off from their Asian relatives and from the vast empty Americas by ice sheets for about 15,000 years. During this long period of time they mutated their Asian mtDNA and NRY haplogroups into new ones and then in a quick wave covered America swiftly so as not to allow new diversity to arise.
Furthermore, their Asian relatives all died off, leaving no trace on the Asian side of Beringia.
Yes, I know it sounds improbable, yet even though the odds are against this kind of event, several papers apart from Tamm et al, support the theory.

But let's get back to the Stepwise Mutation Model: it is quite weak, to put it mildly.

Summary: SMM is unreliable

All these phenomena make SMM a very rough approximation to reality, yet it is used as if it was 100% reliable!

Just as an example of this lack of reliability is the quote below (Nebel a., et al., 2001) [7]:

"the behaviour of DYS388 appears to be inconsistent with the SMM, as was shown in two populations of Middle Eastern origin. Additionally, another widely used microsatellite, DYS392, has recently been demonstrated to deviate from the SMM" [7]

The Clock that ticks out of time

I have posted on the useless genetic clocks in the past, so I will not bore you, just highlight my previous objections to clocks:

Divergence from Chimps. Scientists devise clocks to calculate mutation rates. To do so we estimate the divergence dates of the human line from the chimpanzee line. But the date of this event is uncertain, and has been increasing since the 1970s from an estimated 5 Mya to 6 - 7 Mya, and in June 2014, to 13 Mya [9], this recent change should surely impact on the dating of human origins!

Assumptions are also made regarding the duration of a generation (what can we know about how long a generation was 50 kya? did females mature earlier or later? what about males? was it 27 years or 35?). Population sizes and their trends (expansion, migration, admixture with other groups as well as bottlenecks and founder effects) also should also be factored in.

When those clocks are calibrated against real mutation rates measured in (again a discrete sample) familes over the last few hundred years strong discrepancies arise: these family (pedigree) calculated mutation rates usuall differ from the former ones (evolutionary). But these differences remain unexplained in the papers. They merely show both figures but avoid explaining the causes (i.e. the mutational clock does not tick at a regular pace).

The mutation rates are also calibrated against the estimated dates for the peopling of certain regions based on the information provided by archaeology. i.e. 40 kya for Australia or 17-20 kya for America. However just by looking at the published error margins we can see the uncertainty involved in these calculations.

Diversity is taken as an indicator of antiquity so if Region X has a large variety of haplotypes while Region Z has fewer, population in Z is assumed to be younger. But, actually what happens is that if the people in "Z" are a subset of population from Region "X", the fact that they are a subset means that they will have less diversity than the original group. This does not mean that they are more recent, it means that they left certain genes behind. Add to this the pressure of natural Selection (and chance i.e. genetic drift) and the genes of certain individuals within the subset at "Z" will get lost too. So if we measure "X" against "Z" by their diversity we would incorrectly judge "Z" to be more recent, when they are really just as ancient as "X".

Frequency, Migrations and antiquity

Often the current distributions of haplogroups occur at differing frequencies in certain territories. What does this mean? That the less frequent "A" hg is a recent arrival of a small group carrying it, entering the territory of the prevailing "B" hg.? Did "A" exist in the same population as "B", but in very low frequencies, and those have been maintained or even decreased?

Or is "A" an ancient colonizer that suffered attrition over thousands of years and has been gradually losing ground to better equipped newcomers with hg. "B"? Maybe "A" and "B" were found in equal proportions in the original colonizers but "B" grew due to genetic drift or natural selection...

Questions like those are seldom asked or answered in mainstream papers. It is clear that the choice of the correct answer requires an in depth analysis which is not found in the academic literature (I have read tens of papers and these matters are not even addressed).

The low diversity among Amerindians is always invariably attributed to a founder effect or a bottleneck during the peopling of America event. The massive death of millions of Natives (virtually a genocide) during the process of discovery and conquest of the New World between 1492 and 1560 is ignored. Disease and war acted selectively wiping out tribes without leaving a trace of them, but this issue is simply ignored and the "lack of diversity" is assumed to be due to the original peopling event some 15 kya.

Dogma

The unidirectional migratory route from Africa to the World has some inconsistencies which can only be explained by back-migrations. These into Africa migrations are reluctantly accepted by orthodoxy but, fortunately, are gradually altering the OoA picture with a more parsimonious explanation. Sometimes I get the feeling that OoA is supported because it is politically correct and assuages the guilt complex of the Western world for the tragic crimes of Slavery and colonialism perpetrated against Africa.

Two issues requiring a serious review are: The East Siberian void of putative ancestors to the Amerindians, which remains unexplained and The "Beringian standstill" justification for Amerindian uniqueness, which also requires a critical analysis due to its improbability.

Closing comments

What I have tried to express in today's post is that there are many assumptions underlying the "facts" expressed by mainstream geneticists regarding human diversity and evolution.

Models are simple representations of reality and not reality itself. They should be taken as such and not as truth written in stone.

Algorithms and simulations run on models are only as reliable as the models they are based on. And we have seen the flaws in some of these models and programs. Flaws that introduce errors in their output yet are not explicitly mentioned in the papers that basethemselves on them.

The complexities in the statistical assumptions mentioned in papers (those pages or paragraphs, full of equations that you skip when reading a paper) mask some very evident biases that skew the results and produce patterns that do not correctly reflect reality, which is richer and much more varied than what these papers show us.

Sources

[1] Esra Ruzgar and Kayhan Erciyes, Phylogenetic Tree Construction for Y-DNA, Haplogroups.
[2]Amke Caliebe et al., (2010). A Markov chain description of the stepwise mutation model : Local and global behaviour of the allele process. Journal of Theoretical Biology 266(2010)336–342
[3] Peter Calabrese and Raazesh Sainudiin, (2004) Models of Microsatellite Evolution
[4] Yan Li, Phycs498BIO Assignment 2, How to Build a Phylogenetic Tree
[5] Valdes, Ana M. Slatkin M. and Freimer N., (1993). Allele Frequencies at Microsatellite Loci: The Stepwise Mutarion Model Revisited. Genetics 133: 737-749 March 1993
[6] Boris Malyarchuk, et al., (2010). Phylogeography of the Y-chromosome haplogroup C in northern Eurasia. Annals of Human Genetics (2010) 00,1–8 doi: 10.1111/j.1469-1809.2010.00601.x
[7] Nebel A., et al.,(2001). Haplogroup-specific deviation from the stepwise mutation model at the microsatellite loci DYS388 and DYS392. Eur J Hum Genet. 2001 Jan;9(1):22-6
[8] Ian Wilson, David Balding and Mike Weale, (2003), Batwing User Guide. See pt. 1.2.
[9] Oliver Venn et al., (2014). Strong male bias drives germline mutation in chimpanzees. Science 13 June 2014: Vol. 344 no. 6189 pp. 1272-1275 DOI: 10.1126/science.344.6189.1272
[10] www.1000genomes.org.
[11]Lachance J, Tishkoff SA. et al., (2013). SNP ascertainment bias in population genetic analyses: why it is important, and how to correct it. Bioessays. 2013 Sep;35(9):780-6. doi: 10.1002/bies.201300014. Epub 2013 Jul 9.
<(p>

---

Patagonian Monsters - Cryptozoology, Myths & legends in Patagonia Copyright 2009-2014 by Austin Whittall © 

A critical post on the current methods of genetics when applied to human ancestry

If you have been reading my posts over the last couple of years, you will notice that time and time again I write about the "early peopling of America". The reason for this is, one one hand, because I believe that our ancestral relatives, either Homo habilis, Homo erectus or even the Neanderthals reached the New World long before the 30 - 15 kya window currently accepted by orthodox mainstream science as the date of arrival of modern humans to the Americas. I also believe that H. sapiens reached America much earlier than that date. However, proof in the shape of stones and bones is lacking, or if produced, is dismissed as geofacts (for the tools) or poorly dated. I have the hunch that Acheulian or Mousterian tools are not even found because nobody is on the look out for them as would be the case in the Old World.

Why do I investigate the early peopling of America?

My original objective for looking into archaic humans in America arose from the intriguing posibility that extant populations of these hominins could have survived until recently in the Americas and been the origin of Native American myths regarding wild men, ogres and bigfoots across the continent. The recent survival of the Flores Island hominin is proof that such an idea is not absurd. Unfortunately we have no proof yet, that any of those hominins ever reached the New World.

My dismay regarding research and a Critique

Being an amateur in this subject, I decided to look into the different findings, sites, dates, the pattern of human migrations, the current ideas regarding the evolution of our species and the hard, solid and factual evidence that is behind the current consensus among scholars that dedicate their lives to study these matters.

My research into these matters, with the open mindeness of a person well read in science, with a University degree in engineering and a sceptic but questioning mentality has opened my eyes to the methods of geneticists and archaeologists. Methods which are often surprising becaus they are unquestioning regarding the evidence they base research on. Some papers are based on educated guesses or cite papers which cite other papers that do the guessing and therefore appear well backed by research and proven by peer review.

Most papers on ancestry and haplogroup datings or admixture between modern and archaic humans are dense models based on computer run simulations with pages of statistical formulae and probabilistic assumptions (mutation rates, migratory events) and equations dredged from previous papers, cited again and again. Seldom are these basic axioms questioned or scrutinized with an impartial approach.

Taking all of the above into account it is very logical to assume that they will all produce similar scenarios supporting the currently prevailing Out of Africa theory.

The model of an Out of Africa migration of modern humans, after they evolved in that continent, and then dispersed across the globe, replacing previous migrants without leaving trace of them has evolved since its inception in the early 1990s, was based on weak premises. But it was acclaimed by all and has adapted: it now admits introgression and admixture with archaic hominins in both Africa and Eurasia (though their mtDNA and Y chromosome haplogroups have vanished without a trace but not so their autosomal DNA-funnily nobody really sees this as unusual...). The African origin theory goes mainly unquestioned.

A dissident Voice against "Out Of Africa"
By the way, there is an excellent paper criticising the Out of Africa theory focusing on many of the things that I have pointed out in some of my previous papers: self-fulfilling circular-logic papers where data is bent or tinkered with to prove something (i.e. calibrate a clock based on a date arising from archaeology and then, via complex genetic inferences and hand-picked data, reach that same date and wield it as proof.
It is a very interesting paper and I recommend reading it: Anatole A. Klyosov, (2014) Reconsideration of the “Out of Africa” Concept as Not Having Enough Proof. http://dx.doi.org/10.4236/aa.2014.41004 Advances in Anthropology 2014. Vol.4, No.1, 18-37 February 2014

The Chinese view of the world OOC

I have also come across some Sino-centric papers written by Chinese scientists at Chinese Universities which seem to promote an Out-Of-China origin of modern humans (Well, after all, Chinese invented the compass, paper, the printing press and gunpowder so it is natural to assume that modern humans come from China too). The story goes something like this: [1]

Our darkest and deepest ancestors appeared in Africa, but soon dispersed: H. habilis moved into Asia (Georgia, Sundaland and Nihewan, China), followed later by H. erectus. The latter evolved in Lantian in China to produce H. yunxianensis which back-migrated to West Eurasia 600 kya and originated H. heidelbergensis and later, the Neanderthals. They also trekked back into Africa, where they were the ancestors of H. rhodesiensis and ultimately the Homo sapiens in Africa, replacing in the process the hominins that had evolved in Africa and Europe (H. leakeyi and H. antecessor respectively).

The hominins in China in the meantime kept on evolving, originating archaic humans in China (H. mapaensis in the South and H. daliensis in the North) some 400 kya. The modern H. sapiens left Africa 120 kya and intermixed with the archaic Eurasians in China and India. They were followed by modern H. sapiens 60 kya which kept on admixing and this "led to the emergence of fully modern Chinese hominans by approximately 35,000 ya." [1]. I rest my case.

Please don't misinterpret me; I have nothing against China or science in China, what is more, I do agree on a dynamic model of human evolution with migrations and introgressions, and even with an Eurasian origin of mankind.

The current one with isolated branches either evolving into other hominins or dying stunted is too linear for me. There is no doubt that the branches criss-cross and merge, but the OOC (out of China) theory is a bit too much for me, lacking as it does, solid proof.

An example of Sino-centrism is a recent paper on the sequencing of the DNA from the Tianyuan remains (allegedly 40 kya, but younger according to others). The bones, from a site close to Beijing, China, show a moisaic of modern and achaic features. However its DNA admixture with Neanderthals is similar to current-day populations in the region (and it has no Denisovan genes). The Sino-centric part comes when the paper claims that its mtDNA is "ancestral to present-day haplogroup B" [2], insinuating that it is the "root" of all Amerindian and South East Asian humans belonging to hg. B. Placing a Chinese origin to the dispersal of these people. But we know that this haplogroup arose in West Asia, in the Caucasus, 50 kya, and not in China 10 ky later.

The findings are interesting: admixture if it happened took place long before 40 kya (otherwise Tianyuan would have more Neander or Denisovan admixture since it had less time to dilute). The Phylo Tree.org site indicates that the nucleotide mutations place it in the B6 haplotype within R11'B6 [3], which has a more regional distribution than the other B lineages spread across America and Asia:

• R11'B6 (occasionally also in Japan and South Asia)
  
  • R11 (recorded in Hainan but not in Laos.)
  • B6 (Centred on South China. Recorded in Laos but not in Hainan. )

By the way, the paper explicitly mentions in the mtDNA sequence, a long deleted block at positions 8281-8289 (this is the famous 9bp deletion), which if you search for it in the Phylo Tree page above will turn up 4 times! Meaning it is not so uncommon or perhaps that it is ancestral and the haplogroups should be reviewed to accomodate it in a better way. This is an interesting marker and will be scrutinized in future posts.

Surviving Homo erectus

Despite my misgivings, I do find some papers very interesting, like the one that claims that Homo erectus survived until very recently in Northern China [4]:

The Lantian site, close to Xi'an (renown for its imperial tomb with hundreds human-sized terracota warriors) has unveiled stone artifacts of a clear Acheulean technology. The site is located on the terraces of the Bahie River and its toolage has been dated to the period between 70 and 30 kya. (Wang et al., 2014) [4].

This relatively modern dating (and the type of tools, which includes hand-axes), implies that Homo erectus was alive in the Lantian region "from the early and middle Pleistocene to the later period of the late Pleistocene" [4].

There is also a paper [5] (more below) that hints at a recent introgression, in Africa between archaic humans and modern ones some 37 kya. These archaics had split from our line between 700 and 2,100 kya, meaning they could be any hominin from H. erectus onwards.

But, except these two papers, Homo erectus is seldom mentioned in any of these papers concerning admixture; they are regarded as a dead end, perhaps extinct by the time modern humans reached East Asia. Not many follow up on the clues on the possible admixture from H. erectus into modern humans, as suggested by Prüfer et al., 2013 [6]: that Denisovans carried an introgression from an archaic hominin that split from our line over 1 Mya. (H. erectus perhaps?).

(Well, some do follow up, sort of, as you will see below in point 3 or 4)

And the possibility of H. erectus entering America, migrating across Northeastern Asia is, of course, completely ignored (excuses given: too dim witted to deal with the cold Arctic conditions, not able to navigate, etc.).

Competing Theories are good for Science

But, fortunately for Science, new ideas appear, and they are analysed from different points of view and with different tools; this is good for science and for the advancement of our knowledge.

It seems that almost everyone now agrees that there was admixture between Neanderthals and Modern Humans (While not too long ago -see Herrera et al., 2009 [7], there were serious doubts and scepticism regarding admixture). Then we have the mysterious Denisovans whose genome is found in high proportion of Melanesians and Papuans but not among South East Asians.

We are know asking: Who admixed with who? did Africans admix with Neanderthals? What is the degree of all this admixture?

And the answers differ and even have opposite points of view, as we can see below (from a very small sampling on the matter), and from these differences, Science will grow stronger:

1. Neanderthal DNA Only in Europeans: 3.6% introgression in Europeans with a decreasing cline as you move away from Europe. Neanderthal genes in Africans' (as well as Denisovan presence in Melanesians and Papuans) is not due to admixture but to "the retention of ancient mutations in these populations" [8], that is, the archaic genes survived in them -but, may I ask, why did they disappear in Europeans?
2. 40% More Neanderthal DNA in East Asians (9.6%) than Europeans (6.4%). The Maasai of East Africa have a small but significant fraction of Neanderthal DNA but not through direct contact with them, but with Eurasians humans carrying Neanderthal DNA in them; a "Back into Africa Migration".
   Admixture with Neanderthals happened at least twice or, over a long period of time 50,000 to 80,000 ya., with Europeans splitting off first and therefore being less exposed to Neanderthal genes. [11]
3. Archaic admixture in Africa: "African populations contain a small proportion of genetic material (~2%) that introgressed ~35 kya from an archaic population that split from the ancestors of anatomically modern humans ~700 kya.... [which] introgressed into modern Africans from a now-extinct taxon that may have lived in central Africa." [5]
   This study sets the split between our line and the "archaics" at 1.25 Mya (95% CI, 0.7–2.1 Mya) and an admixture time of 37 kya (95% CI, 1–137 kya). Note the very very large Confidence Intervals. [5]
   When CI are large, I begin to wonder about the soundness of the statistical tools used...
4. Not one, but many introgressions. This one is interesting; it finds "introgressions from two unknown archaic hominins whom diverged with modern humans approximately 859 and 3,464 thousand years ago. The latter unknown archaic hominin contributed to the genomes of the common ancestors of modern humans and Neanderthals. In total, archaic hominin introgressions comprised 2.4% of Eurasian genomes" [9]
   These episiodes of admixture are shown in the image below. The "E" hominin may be H. erectus or even "australopithecines, which inhabited Africa; it was proposed that those hominins might have migrated out of Africa at ~3 Mya" [9]. Note however that "E" does not admix with modern humans either in Africa or Asia, it is an indirect influx via Neanderthals or the unknown X hominin (E_N + N or X arrows) or an admixture with the pre-OOA humans (E_A arrow), which takes us to the case proposed in point 3 of an introgression in Africa.

Figure 3, from [9]

This last paper is interesting but we don't see an admixture between H. erectus and humans in Eurasia, only indirect introgression via Neanderthals or X hominin. In this context, the mtDNA extracted from the remains of Sima de los Huesos, Spain, belonging to a Mid Pleistocene H. heidelbergensis about 300 ky old. may give some clues:

It shows that these archaich Spaniards shared descent with the Denisovans, implying that the Denisovans of Altai and the Sima de los Huesos people formed a clade with a common ancestor some ~800 to 900 kya, which is not the same ancestor of Neanderthals and modern humans. [10]

Could this ancestor be H. erectus who had been in Eurasia for nearly 1 My at that time? or the descendants of Georgian H. habilis? Is there any other candidate from out of Africa at that time?

Interestingly we have the H. heidelbergensis in East Europe which must have extended its range or even migrated into Western and Central Asia to originate our mysterious Denisovans. Did this happen before the appearance of the Neanderthals in the Middle East? or did they overlap?

Now if Neanderthals did not originate from the European H. heidelbergensis as thought until now, where did they evolve? In the Levant, and migrated from there into Europe and West Asia?

Regarding the genesis of these clades, notice that we see very deep splits between mtDNAs not in Africa, but in Eurasia: so much for the OOA. But then, these are mtDNA lineages, what do the autosomal and the Y chromosomes have to say, we have no input yet. But it will be interesting when we do.

And these questions, the migrations and admixtures of our ancestors, take me back to a post (First Asians were not H. erectus) (Sept. 2011), where I mentioned the initial departure of H. habilis from Africa for Georgia in Asia. The image below is an update of the original one from that post.

Hypothetical routes of our ancestors out of Africa. Copyright © 2014 by Austin Whittall

The paths drawn above start with an Out of Africa event, long ago: H. habilis leaves Africa, reaches Asia where it evolves into H. erectus. Some H. habilis reach Indonesia the ancestors of Flores hominin, others may have reached America. The H. erectus cross Europe becoming the ancestors of the Sima de los Huesos people, and also re-enter Africa and they go East too, into South Asia and China. Some march on, into America. The Afro-European hominins evolve into Neanderthal, the Eurasians (now it seems that it is the H. heidelbergensis of Europe) into Denisovans.

They all admix as they criss cross their territories. Somewhere (Africa? Asia? or why not, America?) modern humans appear and also mix with the surviving archaics.

This possible story outlined above, is of course a wild conjecture, with no proof to back it. A fancy tale.

But who knows, maybe if I phrased it using formulae with Greek letters (θ, ρ and so on,) framed with obscure statistical phrases ("the likelihood function f(γ), describing the probability -p(x,y,n)- of the data under different parameter values; and f(z) is the total probability of the data summed and integrated over the parameter space considering a bias scatter ratio in the Bayesian inference based on the...") and ran a Computer program to validate it ("We then ran (n=1,000) simulations on GATACCA® using 150,000 ARGs for each parameter value, storing approximations of the summary statistic distribution, discarding outliers below conformed threshold ratios...")... would someone buy it?

Mind you I am an engineer with a solid foundation in statistics, advanced maths and am particularly fond of quantum mechanics but in all honesty, most genetics papers nowadays are based on statistical simulations.

And we should not forget that simulations are only as robust as the models they are based on, and models tend to oversimplify reality using assumptions that may or may not be correct. The ample confidence intervals informed in many papers is a clear indication of uncertainty in the models and the variability within the sample employed.

Other factors such as migrations influx, population sizes and growth, generation duration and mutation rates also impact on the models' predictions. Explaining anomalies away aducing bottlenecks, founder effects and genetic drifts is like invoking ether for light transmission in the pre-relativity days of modern physics. Predictions should be made and validated by the models to prove reliability.

Thanks for bearing me out!

Sources

[1] Dennis A. Etler, LI Tianyuan, A Multiple Dispersal Hypothesis for Interpreting the Pattern of Hominan Evolution in China
[2] Qiaomei Fu, et al., (2012). DNA analysis of an early modern human from Tianyuan Cave, China. PNAS vol. 110 no. 6 2223–2227, doi: 10.1073/pnas.1221359110
[3] van Oven M, Kayser M., (2009). Updated comprehensive phylogenetic tree of global human mitochondrial DNA variation. Hum Mutat 30(2):E386-E394. doi:10.1002/humu.20921. http://www.phylotree.org/
[4] Shejiang Wang et al., (2014). Newly discovered Palaeolithic artefacts from loess deposits and their ages in Lantian, central China. Chinese Science Bulletin Jan- 2014. 10.1007/s11434-013-0105-5
[5] Michael F. Hammer et al., (2011). Genetic evidence for archaic admixture in Africa. PNAS. vol. 108 no. 37 15123–15128, doi: 10.1073/pnas.1109300108
[6] Prüfer K.,et al., (2013). The complete genome sequence of a Neanderthal from the Altai Mountains. Nature 505: 43-49.
[7] Kristian J. Herrera et al., (2009). To what extent did Neanderthals and modern humans interact?. Biol. Rev. (2009), 84, pp. 245–257. doi:10.1111/j.1469-185X.2008.00071.x
[8] Lowery RK et al., (2013). Neanderthal and Denisova genetic affinities with contemporary humans: introgression versus common ancestral polymorphisms. Gene. 2013 Nov 1;530(1):83-94. doi: 10.1016/j.gene.2013.06.005. Epub 2013 Jul 19
[9] Ya Hu, et al. (2014). Genome-wide Scan of Archaic Hominin Introgressions in Eurasians Reveals Complex Admixture History. arXiv preprint arXiv:1404.7766
[10] Matthias Meyer et al., (2013). A mitochondrial genome sequence of a hominin from Sima de los Huesos Nature (2013) doi:10.1038/nature12788
[11] Jeffrey D. Wall et al., (2013). Higher Levels of Neanderthal Ancestry in East Asians than in Europeans. Early Online February 14, 2013, doi: 10.1534/genetics.112.148213 Genetics May 1, 2013 vol. 194 no. 1 199-209

---

Patagonian Monsters - Cryptozoology, Myths & legends in Patagonia Copyright 2009-2014 by Austin Whittall ©