Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Save to My Library
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
Ancestors Rohde-MRCA-Two

Ancestors Rohde-MRCA-Two

Ratings: (0)|Views: 112|Likes:
Published by ozdare

More info:

Published by: ozdare on Jul 05, 2012
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

01/29/2014

pdf

text

original

 
On the Common Ancestors of All Living Humans
Douglas L. T. Rohde
Massachusetts Institute of TechnologyNovember 11, 2003
Abstract
Questions concerning the common ancestors of allpresent-day humans have received considerable attentionof late in both the scientific and lay communities. Princi-pally, this attention has focused on ‘Mitochondrial Eve,’defined to be the woman who lies at the confluence of ourmaternal ancestry lines, and who is believed to have lived100,000–200,000 years ago. More recent attention hasbeen given to our common paternal ancestor, ‘Y Chromo-some Adam,’ who may have lived 35,000–89,000 yearsago. However, if we consider not just our all-female andall-male lines, but our ancestors along all parental lines,it turns out that everyone on earth may share a commonancestor who is remarkably recent.This study introduces a large-scale, detailed computermodel of recent human history which suggests that thecommonancestorofeveryonealivetodayverylikelylivedbetween 2,000 and 5,000 years ago. Furthermore, themodel indicates that nearly everyone living a few thou-sand years prior to that time is either the ancestor of noone or of all living humans.
1 Introduction
Advances in genetics have sparked interest in our com-mon ancestors, the individualsfrom which all present-dayhumans descend. Initial interest focused on the topic of ‘Mitochondrial Eve,’ who is defined to be the most re-cent female ancestor from whom all individuals descendalong strictly maternal lines (Cann, Stoneking, & Wilson,1987; Vigilant, Stoneking, Harpending, Hawkes, & Wil-son, 1991). An approximate date for Mitochondrial Eveof 100,000to 200,000years ago has been estimated basedonthesuccessivemutationsinmitochondrialDNA, whichare passed down from mother to child. A similar analy-sis can be performed on the strictly paternal lines of suc-cession, using the Y chromosome, which is passed downfrom father to son, to determine the approximate date of ‘Y Chromosome Adam.This date was originally esti-mated very loosely to fall between 27,000 and 270,000
Work in progress. Do not cite.
years ago (Dorit, Akashi, & Gilbert, 1995), a range thatwas more recently narrowed to 35,000–89,000 years ago(Ke et al., 2001).Nevertheless, an individual’s strictly maternal andstrictly paternal lines are just two of a vast numberof pos-sible paths back through his or her ancestors. What if weadopt a more common-sense notion of ancestry that in-cludes ancestors reachable along any path of succession,using both mothers and fathers? It seems likely that ourmostrecentcommonancestor(MRCA)underthisbroaderdefinitionwill be muchmorerecent than eitherMitochon-drial Eve or Y Chromosome Adam. Unfortunately, theage of our MRCA cannot as easily be estimated on thebasis of genetic information because the relevant genesare not passed from parent to child with only occasionalmutations but are, rather, the product of recombination.As a result of recombination, a given gene may not passfrom parent to child. In fact, an individual’s DNA mayretain none of the genes specific to a particular ancestorwho lived many generations in the past. These additionalcomplications make accurately dating the MRCA or re-constructing other details of population history on the ba-sis ofour genesextremelydifficult, if not impossible(Hey& Machado, 2003).However, alternative methods may be able to answerthis question. Some researchers have produced estimatesof the age of our MRCA by means of the theoreticalanalysis of mathematical models. Building on work byammerle (1991) and M¨ohle (1994), Chang (1999) ana-lyzed a model that assumes a fixed-size population, withdiscrete and non-overlapping generations, and randommating. That is, each child is the product of two parents,randomly selected from all members of the previous gen-eration. Chang showed that, in this model, the mixing of genes occurs quite rapidly. In fact, the number of gener-ations back to the MRCA is expected to be about
log
2
of the populationsize. With a populationof 6 billion people,this model predicts that the MRCA is likely to occur in just over 32 generations, or 800–975 years. This suggeststhat our all-paths MRCA may be exceptionally recent.But Chang was well aware of the limitations of thesimple model he analyzed. “What are the significance
 
R
OHDE
C
OMMON
A
NCESTORS OF
A
LL
L
IVING
H
UMANS
of these results? An application to the world populationof humans would be an obvious misuse...An importantsource of inapplicability of the model to this situation isthe obviousnon-randomnature of mating in the history of mankind.(pg. 1005) There are many factors that limitthe randomness of human mating. First of all, clearly, aresex differences, but Chang did address this, noting thatadding distinct sexes to the model would not cause a sub-stantial change in the estimate. Another factor is mar-riage. Once a couple has one child, they are likely to re-main together as they produce more children. Moreover,broadersociologicalandgeographicfactorsmayhavestillmore profound effects. In short, although we are becom-ing increasingly panmictic, humans groups have tendedtowards a high rate of endogamy, finding mates almostexclusively within the local population and social class,onlyoccasionallytranscendingbarriersofgeography,lan-guage, race, and culture.It seems likely that these restrictions on the random-ness of human mating may dramatically decrease the rateof ancestral mixing in the model. As a result, the true dateof the MRCA could be thousands or tens of thousands of years ago, rather than just hundreds. Thus, an obviousnext step is to test this possibility by expandingthe modelto include some or all of these constraints. Unfortunately,conducting a theoretical analysis of a more complicatedmathematical model would be very difficult. An alterna-tive approachis to implement a computersimulation. Theprincipal advantageof a computersimulation is that it canbe arbitrarily complex. However, even given the speed of today’s computer, efficiently simulating the ancestral his-tory of a population whose size is even close to the scaleof humanity is non-trivial. Furthermore, because a non-random model will necessarily involve numerous param-eters that cannot be adequately constrained by availabledata, the simulation must typically be run many times toexplore the consequences of various parameter settings.This study involves the implementation and analysis of several large-scale computer models of recent human his-tory. The models simulate individual human lives, includ-ing life span, birth rate, choice of mates, and migration,and the data they produce is analyzed to obtain more ac-curate estimates of the date of our most recent commonancestor. Given what seem to be reasonable parameterchoices, the final, most detailedmodel presentedhere pre-dictsthatourmostrecentcommonancestorprobablylivedbetween 2000 and 5000 years ago and that nearly every-one alive prior to a few thousand years before that are theancestors of either no one or of everyone alive today.
1.1 Modeling human genealogy
Mathematical models of human genealogy must be quitesimple if their analysis is to be possible. Most, likeChang’s, are some variant of a Wright-Fisher model, withdiscrete generations and parents selected at random fromthe preceding generation (Nordborg, 2001). Becausecomputer simulations are tested empirically rather thanthrough theoretical analysis, they are not subject to suchconstraints.However, there are practical limits to the complexityof a computer simulation. One is the matter of computa-tional efficiency. A model cannot be so complex that run-ning it requires an unreasonable amount of time or space.A more significant limitation, from a scientific perspec-tive, must be placed on the number of free parameters inthe model. Ideally, for the results of a model to be reli-able, any free parameters should be constrained by his-torical data, such as statistics on actual birth or migrationrates. However, much of the relevant data for the currentmodels concern events occurring thousands of years agoand cannot be obtained with any accuracy. In this case,the parameters must be varied within the range of plau-sible values to obtain bounds on the model’s predictions.A model with too many free parameters, especially onesunconstrainedby empirical data, will have reduced powerand will be difficult to study. Therefore, a good modelmust be complex enough to include relevant factors, butnot overly burdened by irrelevant ones.This study explores a progression of three models. Thefirst extends Chang’s results to a world consisting of fivemore or less panmictic islands, or continents, with onlyoccasional migrants between any pair of continents. Thesecond model, discussed in Section 3 arranges the islandsin a graph that roughly reflects the topology of the ma- jor world continents. The final model, discussed in Sec-tion 4, is a more detailed simulation of the actual world,with migration routes and dates based on historical dataor prehistoric estimates.
2 Model A: Fully-connectedcontinents
The first model, A, is quite abstract but incorporates sev-eral levels of detail beyond those found in most Wright-Fisher models. The model typically starts between 5000and 20000 BC and runs to the present day, which is takento be the year 2000 AD. As the model runs, it simu-lates important details in the lives of individual people,known as
sims
, including their lifespans, possible migra-tions, choice of mate, and production of offspring. Asthe model runs, it records this information in a series of largecomputerfiles. A secondprogram,discussed in Sec-tion 2.2, traces ancestral lines through this data to find thecommon ancestors.2
 
R
OHDE
C
OMMON
A
NCESTORS OF
A
LL
L
IVING
H
UMANS
2.1 Details of the model
2.1.1 Life span
The present models do not assume discrete, uniform gen-erations. Each sim is born in a certain year and has aparticular life span. The maximum age of any sim wasset to 100, as it seems highly unlikely that anyone wouldlive, let alone father children, beyond that age. The ageof sexual maturity was taken to be 16 years for both menandwomen. Anyonewho wouldhavedied beforethat agecould not have produced offspring and is thus not a factorfor the purposes of this study. Therefore, only the lives of those destined to at least reach adulthood were simulated.As a result, the populationsizes discussed throughoutthispaper are effectively somewhat larger than stated becausethey do not include any children.Otherwise, the probabilitythat an individual dies at age
s
, conditional on not having died before age
s
, is assumedto follow a discrete Gompertz-Makeham form (Pletcher,1999):
 p
(
s
) =
α
+ (1
α
)
e
(
s
100)
In this equation,
β 
is the
deathrate
. A higherdeath rateresults in shorter life spans on average, althoughthe resultis not linear. The
α
parameter is the
accident rate
, whichcan be adjusted to reflect the probability that an individ-ual of any age dies of unnatural causes. With an accidentrate of 0.01 and a death rate of 10.5, this formula quiteclosely models the life span data for the U.S. between1900 and 1930 (U.S. National Office of Vital Statistics,1956). To account for historically shorter life spans dueto poor nutrition, medicine, and so forth, the death rate,
β 
, was raised to 12.5 for the purposes of the model. Thisproduces an average life span of 51.8 for those who reachmaturity.The percentage of males born into the population wasset at 50%. It is true that the actual percentage of malesand females reaching adulthoodmay differ somewhat dueto infanticide coupled with the fact that a slightly higherpercentageofnewbornsaremalethanfemale(Davis, Got-tlieb, & Stampnitzky, 1998). But this probably does nothave much bearingon the results of the model. And whileit is true that women tend to live longer, the life span of women past child-bearing age is also not relevant to theoutcome of the model. Therefore, for simplicity, the lifespansofmalesandfemalesweregeneratedusingthesamedistribution.
2.1.2 Migration
Themodelsareorganizedintothreestructurallevels: con-tinents, countries, and towns. The continents representphysically separated land masses that are likely to havevery low rates of inter-migration. Europe and Asia arecontiguous, with no substantial geographical barrier tomigration, so they will be considered a single continent,along with Africa, North America, and South America.Indonesia, Australia, and Oceania are, taken together,somewhat more difficult to model, as there is clearly sub-stantial internal structure. For the purposesof the first twomodels, they will be considered a single continent, butwill be dealt with more appropriately in the third model.The models’ continents are divided into
countries
, ar-ranged in a grid. These reflect major tribal, ethnic, orlanguage groups, with both geographic and cultural bar-riers to intermarriage. Countries are, in turn, divided intotowns. These do not necessarily represent towns per se,but the relevant social unit from within which most peo-ple find mates. Thus, a town may actually reflect a clan,a rural county, or even a particular social class within alarger group. The towns within each country are assumedto be in relatively frequent contact with one another andare not in any particular geographic arrangement.Not all humans confine themselves to a single locationthroughout their lives and a critical factor in the model isthe rate at which people migrate to different places in theworld. Although it seems likely that many people, andperhaps the vast majority historically, live out their livesclose to where they were born, variousforms of migrationlead to the gradual spread of ancestral lineages over longdistances. When men and women from different groupsmarry, one of them, often the wife but sometimes the hus-band, moves to the others community. Merchants, sol-diers, and bureaucrats, who are typically male, sometimestravel widely, potentially fathering children far from theirplace of birth. And, occasionally, large groups of peoplehave conquered or colonized new areas.In terms of realism, it would certainly be desirable todistinguish between these and other specific types of mi-gration in the model. However, doing so would introducemany new parameters, for which we are unlikely to findsufficient data. Therefore,the model uses a simplified mi-gration system, in which each person can move only oncein his or her life. Each sim is born in the town in whichhis or her parents, or at least mother, lives, but then hasa chance to migrate to a different continent, country, ortown prior to adulthood. Henceforth, that person can pro-ducechildrenonlywithotherinhabitantsofhis or hernewtown, provided it contains potential mates.As is the case in human mating patterns (Fix, 1979),the rate of exogamy decreases substantially with largergroup size in the models. Adams and Kasakoff (1976)found that, across a variety of human societies, there wasa recognizable threshold in group size at around a 20%exogamy rate, although the sizes of these groups differedas a function of population density. This “natural” groupsize is taken here to be that of the town. The
Change-TownProb
parameter controls the percentage of sims who3

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->