UFDC Home  myUFDC Home  Help 
Material Information
Thesis/Dissertation Information
Subjects
Notes
Record Information

Material Information
Thesis/Dissertation Information
Subjects
Notes
Record Information

Full Text  
PAGE 1 1 GENETIC AND GEOGRAPHIC PERSPECTIVES ON HUMAN MIGRATION AND IMPLICATIONS FOR PREHISTORIC DEMOGRAPHIC RECONSTRUCTIONS By AIDA T MIR A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF TH E REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013 PAGE 2 2 2013 Aida T. Mir PAGE 3 3 Para Titi Letty PAGE 4 4 ACKNOWLEDGMENTS I would first like to thank my mom, dad and sister for their unconditional support and love in all of my endeavors. I would like to thank Professor Connie Mulligan for her encouragement, mentorship, and guidance throughout my graduate career. I would also like to thank Professors David Reed, Michael Miyamoto, and Steve Brandt for their insight and guidance throughout this learning process. I thank current and former postdoctoral researchers and graduate students in the Mulligan lab, Dr. David H ughes, Dr. Laurel Pearson, Dr. Andrew Kitchen, Dr. Amy Non, and Tamar Carter for their support and hel pful insight during the past six years. I thank all my undergraduate assistants, especially Alex Wang, Shannon McNulty, Timothy Scott, and Nubiana Todd, whose tireless work has greatly contributed to my dissertation research and with whom I have learned how to be a mentor. Im grateful for the Yemenite individuals who participated in the study that is part of this dissertation. I thank my extended family for always rooting for me, especially, Titi Letty, Jenny, mamamama, Tit and abuelitita. I thank my fri ends for their continuous support, for making me laugh, and for helping me keep a balance during graduate school, particularly Teresa Szakos and Dr. Yaraim Coln Cales. PAGE 5 5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 8 A BSTRACT ..................................................................................................................... 9 CHAPTER 1 INTRODUCTION .................................................................................................... 11 2 HUMAN DEMOGRAPHIC PROCESSES AND GENETIC VARIATION AS REVEALED BY MTDNA SIMULATIONS ................................................................ 24 Materials and Methods ............................................................................................ 27 Models .............................................................................................................. 27 Simulations ....................................................................................................... 28 Summary Statistics ........................................................................................... 28 Statistical Analysis ............................................................................................ 29 Results .................................................................................................................... 29 Partitioning of Genetic Variation by Demographic Parameters ......................... 29 Comparison of Summary Statistics .................................................................. 30 Comparison of Evolutionary Scenarios ............................................................ 31 Discussion .............................................................................................................. 33 Demographic Parameters ................................................................................. 33 Summary Statistics ........................................................................................... 34 Relevance to Human Evolution ........................................................................ 35 PAGE 6 6 3 HUMAN MIGRATION PATTERNS IN YEMEN AND IMPLICATIONS FOR RECONSTRUCTING PREHISTORIC POPULATION MOVEMENTS .................. 107 Methods ................................................................................................................ 109 Samples and Data .......................................................................................... 109 Estimation of Migration ................................................................................... 110 Results .................................................................................................................. 112 Discussion ............................................................................................................ 115 Patrilocality and Genetic Signals .................................................................... 115 Patterns of Migration ...................................................................................... 117 Empirical Estimates of Migration .................................................................... 118 Application of Migration Estimates in Prehistoric Demographic Modeling ...... 121 4 CONCLUSION ...................................................................................................... 134 LIST OF REFERENCES ............................................................................................. 147 BIOGRAPHICAL SKETCH .......................................................................................... 159 PAGE 7 7 LIST OF TABLES Table page 2 1 Summary statistics analyzed and their definitions .................................................. 40 2 2 Partitioning o f genetic variation by demographic parameter for each summary statistic. ................................................................................................................... 41 2 3 Recommendation of optimal summary statistic to use for each parameter of interest. ................................................................................................................... 42 2 4 Demographic scenarios compared in Tukeys tests and pvalues for each summary statistic. ................................................................................................... 43 2 5 Percent of simulated scenarios that agree with empiric al F st estimates separated by CS and GF categories. .................................................................... 102 3 1 Summary statistics for migration distances. .......................................................... 125 3 2 Best model to ex plain probability of migration. ...................................................... 126 3 3 Estimates for the direction of migration in each collection site across all three generation groups. ................................................................................................ 127 3 4 Directional means estimates for each group by collection site. ............................. 128 PAGE 8 8 LIST OF FIGURES Figure page 2 1 Alternativ e scenarios for initial colonization of modern humans out of Africa.. ...... 103 2 2 Box plots of estimates of 5 summary statistics that partition genetic variation more equally between CS, GF, and C SxGF than seen in the other summary statistics. ............................................................................................................... 104 2 3 Box plots of estimates of 4 summary statistics that partition genetic variation primarily by CS, compared to the other summary statistic s. ................................. 105 2 4 Box plots of estimates of 3 summary statistics that partition genetic variation similar to summary statistics in Figure 2 3.. .......................................................... 106 3 1 Proportion of migrants by sex for each generation group.. ................................... 129 3 2 Density plots combining migration distance and frequency of the distance for each group. ........................................................................................................... 130 3 3 Plot of migration distances for marital pairs. ......................................................... 131 3 4 Migration direction vectors and mean migration direction by collection site over all three generations. ............................................................................................ 132 3 5 Migration direction vectors and mean migration direction for each collection site by generation group.. ............................................................................................ 133 PAGE 9 9 Abstract of Dissertatio n Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy GENETIC AND GEOGRAPHIC PERSPECTIVES ON HUMAN MIGRATION AND IMPLICATIONS FOR PREHISTORIC DEMOGRAPHIC RECONSTRUCTIONS By Aida T Mir August 2013 Chair: Connie J. Mulligan Major: Genetics and Genomics This dissertation integrates simulated genetic data with empirical non genetic data to develop a framework for reconstructing human demographic processes. In the firs t study, I simulate mitochondrial DNA for demographic scenarios representing the initial dispersal of modern humans out of Africa. Summary statistics are estimated for the simulated datasets to calculate the percent of explained variation by each parameter and to identify which parameter combinations generate distinct differences in genetic variation. I also identify the informativeness of different summary statistics at summarizing genetic variation. The results show that colonization size and gene flow have the largest effects on genetic variation, suggesting that defining these migration parameters with more realistic values would allow a more accurate reconstruction of the migration out of Africa. In the second study, I analyze migration patterns through four generations in Yemen. I compare the proportion of migrants, and the distance and directionality of migration across the generations and identify factors that influence the migration PAGE 10 10 patterns. The results suggest that the proportion of migrants and mi gration distance is significantly lower in the grandparents generation and more likely represents realistic values f or prehistoric processes. I describe how the values for these migration parameters calculated for the grandparents generation can be used to generate more informative hypothesis models for prehistoric demographic reconstructions. The results from both studies are combined to develop a more realistic and geographically explicit model for the migration out of Africa that increases the possibil ity of accurately inferring the process and estimating values for parameters that have thus far been a challenge. This dissertation illustrates the importance of using interdisciplinary approaches from genetic and nongenetic disciplines in the continued e fforts of reconstructing evolutionary histories. PAGE 11 11 CHAPTER 1 INTRODUCTION Human evolutionary history has been characterized by a series of population growths and contractions, population fissions, and migration, in addition to mutation events. Each of these events has had an effect on patterns of human genetic variation. Mutations in the DNA of individuals increase the genetic variation of a population as, over generations, the mutations accumulate and become fixed or extinct in the population. However, diff erent demographic processes have varying effects on the fixation of mutations. Population growth increases genetic variation in a population, with each new individual increasing the opportunity for mutation and fixation. Population contractions and populat ion fissions tend to decrease the genetic variation as mutations can be lost from a population when individuals die off or leave the population. Population fissions can further create founder effects, where the individuals that leave a population carry onl y a subset of the genetic diversity of the original population ( Joblinget al. 2004) Over time, the original population and the new population accumulate mutations independently, increasing the genetic difference between the populations. This difference can be further increased by subsequent population growths or contractions. In contrast, migration can reduce the genetic difference between populations through the exchange of different genetic mutations (gene flow), while increasing the genetic diversity within each population. As can be seen, the effects of simple demographic processes on genetic variation, such as the effect of population fission, are well understood and can be accurately inferred. However, the effects of PAGE 12 12 more complex scenarios, as would occur from combinations of th ese processes, are less clear. Humans have expanded into almost every region of the world, creating a complex evolutionary history. Anatomically modern human arose in Africa around 100200 thousand years ago (kya) ( Forster 2004; Garrigan and Hammer 2006; Pakendorf and Stoneking 2005) Humans migrated out of Africa into Eurasia 6080kya from somewhere in northeast Africa ( Forster 2004; Macaulayet al. 2005; Soareset al. 2012) They colonized Europe 4050kya ( Forster 2004; Lell and Wallace 2000) The colonization of Asia occurred most likely through a coastal route to southern Asia and southeast Asia around 5066k ya ( Barkeret al. 2007; Macaulayet al. 2005) arriving in Oceania by 4863kya ( Macaulayet al. 2005; Turneyet al. 2001) The Americas were colonized by ~15kya from a group that migrated from southeast Asia ( Kitchenet al. 2008) After these initial colonizations, some areas of the world underwent major recolonization processes. For example, 3545 kya there was a major migration from a population in southern Asia back to northern Afr ica ( Gonzalezet al. 2007; Olivieriet al. 2006) Another notable example includes the recolonization of Europe around 10kya from a population somewhere in southwest Asia, introducing agriculturalists that admixed with the exi sting hunter gatherer populations ( Pinhasiet al. 2012; Rasteiro and Chikhi 2013) The combination of demographic processes that has occurred in each region has generated an elaborate pattern of genetic variation in the current population of the region. PAGE 13 13 Genetic anthropologists interpret the patterns of genetic variation within and between populations in order to reconstruct the demographic processes that have characterized human evolutionary history. Until recent advances in sequencing technolog y allowed for economical sequencing of autosomal DNA, mitochondrial (mtDNA) and Y chromosome (NRY) DNA have been the genetic markers primarily used to reconstruct demographic processes. Both markers are uniparentally inherited and have little or no recombi nation ( Heinet al. 2005; Pakendorf and Stoneking 2005) This allows researchers to directly trace genetic mutations (or combinations of mutations known as haplotypes) from parents to offspring over time. Using this framework, the presence of a mutation in two populations suggests that one population is descended from the other, both populations share a common ancestor, or gene flow has moved the mutation from one population to the other. Thus, the number of mutations and differences in mutations between individuals and populations can be used to infer estimates of population parameters, such as the timing of population fissons or levels of gene flow between populations. Genetic summary st st, which measures the proportion of genetic diversity due to haplotype differences among populations ( Excoffieret al. 1992) are used to infer the historic connectivity between populations ( Holsinger and Weir 2009) st values indicate that populations are st values indicate similarity between populations and suggest much more gene flow. Median networks and phylogenetic trees, which show the connectivi ty between individuals ( Bandeltet al. 1999 ; Fe lsenstein 1983) are also used to describe individual and population relationships. Individuals (or haplotypes) who lie close together in the network or tree are more PAGE 14 14 genetically similar (and more related) than individuals who lie far from each other. Fur thermore, rooted phylogenetic trees, in which the ancestral haplotype is known, are used to chronologically order and date the population fissions ( Felsenstein 1981) I ndividuals with shorter branch lengths indicate fewer mutations, suggesting population fissions occurred more recently, whereas individuals with long branch lengths suggest population fissions occurred longer ago. Traditionally, the values for individual p arameters in a demographic process are estimated from the genetic diversity and the estimates are jointly interpreted in a post hoc manner to describe a process that is consistent with other sources of evidence (e.g. archaeological or geological data). A n otable example is the reconstruction of the process of anatomically modern human out of Africa. Rooted phylogenetic trees from individuals from Africa, Europe, and Asia show that European and Asian populations are descendants of African populations ( Bowcocket al. 1991 ; Garrigan and Hammer 2006; Liet al. 2008) Furthermore, European populations lie intermediate to African and Asians in the trees ( Garrigan and Hammer 2006; Liet al. 2008) Fst estimates (which is st but for allele differences) are smaller between Africa and Europe (0.141) than between Africa and Asia (0.235) and smallest between Asia and Europe (0.093) ( Bowcocket al. 1991) suggesting that Europeans are most closely related to Asians and more closely related to Africans than Asians are related to Africans. Bowcock et al ( 1991) proposed a model to explain these findings, in which Europeans were the product of admixture between Africans and Asians, each contributing 35% and 65% respectively. However, the similarity between Africans and Europea ns could be caused by other demographic processes as well. Gene flow between African and European PAGE 15 15 populations after the European and Asian fission could lead Europeans to be more genetically similar to Africans ( Cerezoet al. 2012; McEvoyet al. 2011) Thus the confounding nature of demographic processes can lead to different processes having a similar effect on pa tterns of genetic variation. A major drawback of post hoc interpretation is that there is no possibility to test the interpretation against other alternative interpretations that may provide better explanations for the observed data. Model based methods of demographic reconstruction present a solution to post hoc interpretations by explicitly testing competing hypothesis through the simulation of genetic variation representing alternative models and comparisons of these hypothesis models to the empirical data ( Beaumontet al. 2002; Tavaret al. 1997) Approximation approaches (i.e. maximum likelihood approximations and Bayesian approximations) have been particularly successful model based methods for estimating parameter values ( Beaumont 2010; Beaumontet al. 2002; Garrigan and Hammer 2006; Marjoram and Tavar 2006) For these methods, different hypothesis models that can represent a demographic process are created to explain the observed patterns of genetic variation. The models are generated by defi ning exact values or ranges of values for different parameters of interest. These values are then used in simulations that generate DNA data with patterns of genetic variation reflecting the demographic processes of the models. Each model is represented by multiple (i.e. thousands) simulated datasets. Each simulated dataset reflects both the pattern of genetic variation generated by the specified demographic scenario and the stochasticity of the mutation process. Thus, PAGE 16 16 each simulated dataset for a specific model has a slightly different pattern of genetic variation that is still representative of the model. Each dataset is then compared to the empirical data to test which model best explains the observed data. In order to compare the simulated datasets to the empirical data, the genetic variation of each dataset must be summarized. Summary statistics provide quick and easy ways to calculate values to summarize genetic variation. The summary statistics calculated for each simulated dataset generate a distribution of summary statistic estimates for each model that represents all the possible patterns of genetic variation generated by the model. The summary statistics of each simulated dataset are compared to summary statistics of the empirical data. The summar y statistics that are most similar (i.e. that have the smallest absolute difference after being subtracted), represent the model(s) that best explain the empirical data. A cutoff value (e.g. 1%) is selected to determine the number of simulations that will be used to select the best model(s). The model(s) with the most support from the simulations within the cutoff value represents the best model(s). This model contains the values for each demographic parameter that best represents the empirical data. In the method described above, the values for each demographic parameter are jointly inferred from the best model. This presents an advantage of model based methods over traditional methods, where parameters are individually estimated. Explicitly accounting for parameter combinations when generating the simulated genetic variation and jointly inferring the values of the parameters from the best model, at least PAGE 17 17 partially, accounts for the confounding effects of combinations of parameters on genetic variation, whic h might not be detected through traditional approaches. Since different demographic processes (i.e. parameter combinations) can lead to similar patterns of genetic variation, the confounding effects of parameter combinations on genetic variation nonetheless limit inferences from model based methods, despite the improvement over traditional methods. Oftentimes more than one model can best explain the empirical data. Even within one model, such as the single origin out of Africa model explaining modern hum an expansion, there can be multiple parameter combinations that realistically represent the model. Parameter combinations with similar patterns of genetic variation are likely to all be selected as the best explanation to the empirical genetic data. Multiple best demographic scenarios decrease the probability of accurately inferring the values for the individual parameters because each parameter will comprise the range of values for all the scenarios included as best. Therefore, in order to generate inf ormative hypothesis models (or parameter combinations), it is important to determine if the parameter combinations produce patterns of genetic variation that are distinguishable from each other. Additionally, different summary statistics can summarize dif ferent aspects of the genetic variation. Thus, a specific summary statistic might more accurately reflect the portion of genetic variation generated by a specific parameter than another summary statistic. In order to select the summary statistics that most effectively identify the best model, it is necessary to ascertain which summary statistics are more informative for summarizing the genetic variation and differentiating the parameter combinations. PAGE 18 18 A further advantage of model based methods is that they allow the inclusion of data from multiple disciplines. Data from genetic and nongenetic sources, such as archaeological evidence, can be directly incorporated to inform the values for the demographic parameters. Incorporating data from across disciplines t o fix or set ranges on specific parameters offers multiple benefits. Prior knowledge is explicitly incorporated, instead of having to find post hoc explanations for the empirical genetic evidence that will be consistent with the other sources of evidence. The values for the parameters are more realistic and thus, more likely to be accurate, allowing for more informative hypothesis models. Realistic hypothesis models increase the overall probability of selecting the best model. Establishing values for spec ific parameters also offers the possibility of testing more values for parameters with unknown prior estimates. The inclusion of prior data from both genetic and nongenetic sources can greatly enhance a study that is using a model based approach. However, finding useful estimates to define parameter values can sometimes pose a challenge. For example, data are more readily available for some parameters to describe the migration of modern humans out of Africa, than for other parameters. The timing of modern humans out of Africa can be delimited by combining the dates established on archaeological remains out of Africa and the divergence dates of African and nonAfrican haplotypes estimated from previous genetic studies. Therm oluminescence a nd electron spin re sonance dating of archaic human remains found in Qafzeh and Skhul (modern day Israel) place archaic humans in the Levant 90120kya ( Grn and Stringer 1991; Mercieret al. 1993; Valladaset al. ) These data suggest that the successful migration of PAGE 19 19 modern humans out of Africa could not have happened before then. Alt ernatively, archaeological remains in Europe and Asia place modern humans arrival 45kya50kya (Mellars 2006). Analyses of mitochondrial DNA (mtDNA) suggest that the L3 haplogroup, which is ancestral to all nonAfrican haplogroups, diverged between 60 and 90 kya ( Forster 2004; Gonderet al. 2007 ; Soareset al. 2012 ) The most ancestral nonAfrican haplogroups, M and N, which gave rise to all other nonAfrican haplogroups, diverged as late as 50kya65kya ( Atkinsonet al. 2008 ; Beharet al. 2008; Macaulayet al. 2005) These data narrow the time during which modern humans could have migrated of Africa to a range of about 50 thousand years. Some parameters, such as the possible size of the population that migrated out of Africa, must rely primarily on genetic data because of the lack of nongenetic data. Values for the population size of nonAfricans have been estimated as low as 1% of the African population ( Atkinsonet al. 2008; Fagundeset al. 2007) and as high as 33% ( Relethford and Harpending 1995; Tenesaet al. 2007) Other studies have found more intermediate values of 1020% ( Gronauet al. 2011) Data on population movement acquired from a demographic approach would provide a more realistic value that could further delimit the possible size of the population migrating out of Africa. Still for other parameters, such as the migration rate between adjacent populations after the migration out of Africa, it is difficult to identify values that are informative for demographic reconstruction using model based methods. Current empirical estimates of migration based on ethnographic studies include short time frames of the seasonal movement of hunters and gatherers ( Hahnet al. 1966 ; Marlowe PAGE 20 20 2010) or migrant workers ( de H aan and Rogaly 2002) These data lack the information on movement across generations that is needed to define gene flow in the developed demographic scenarios. Demographic studies from birth certificates and census data provide data that allow us to trac e movement over longer periods of time (i.e. between individuals and their parents), but these studies have generally focused on a limited analysis of migration, such as distance moved or proportion moved, and have been performed in developed countries ( Boattiniet al. 2007; Gray and Bilsborrow 2013; Mielkeet al. 1994 ; Mortonet al. 1971) The focus on developed countries probably makes the values unrealistically high for prehistoric demographic processes. Genetic studies have estimated migration rates between Af rica and Eurasia ( Coxet al. 2008; Gravelet al. 2011) but the values were estimated from samples located very dist ant from each other, so it is uncertain how informative the values are to describe the movement between adjacent populations. Migration rates estimated in a developing country from a demographic approach could provide a source of more informative data to d efine the values for the migration rate that occurred immediately after the migration out of Africa. Additionally, spatial patterns of migration from demographic data could show whether there is a pattern in the direction that individuals, and the populati on as a whole, are moving. These type of data would also allow to determine whether geographic features, or other factors, have an effect on the patterns of migration (e.g. if people are moving to areas with particular geographic characteristics). Currentl y, it is very difficult to identify the exact geographic location where a demographic process occurred from DNA data ( Epperson 2003) For example, it is unclear whether the PAGE 21 21 divergence of mtD NA haplotypes L3 and M/N, which characterize African and nonAfrican populations respectively, occurred before the new population migrated out of Africa, or after (i.e.in the Middle East) and L3 was lost from the new population out of Africa due to genetic drift (Forester 2004). Knowledge on spatial patterns of migration could help elucidate the geographic area where a demographic process occurred by allowing more geographically explicit hypothesis models to be developed. In this dissertation, I attempt to fill some of the gaps of model based methods for their use in human demographic reconstruction. In Chapter 2, simulations characteristic of model based methods are generated under a simplified model of modern human out of Africa that includes multiple val ues for population size, gene flow and time of the population fission and movement out of Africa ( Mir Herrans and Mulligan 2013) I develop a framework to identify demographic scenarios that generate distinguishable differences in patterns of genetic variation and that offer informative hypothesis against which to compare empirical genetic data. M ultiple summary statistics a re calculated for each demographic scenario to identify the informativeness of different summary statistics in summarizing genetic variation. Based on the results, I provide specific recommendations about the summary statistics to use according to the spec ific demographic parameter under study. Additionally, I identify the contribution of each demographic parameter to the patterns of genetic variation. In Chapter 3, I use GPS coordinates from the place of residence of individuals sampled in Yemen, their bir thplaces and their parents and grandparents birthplaces to evaluate patterns of human migration. With the geographic coordinates, the direction of PAGE 22 22 migration is identified to detect whether geographic features, or other factors, have an effect on the patt erns of migration. I calculate the proportion of migrants and the distance of migration across generations. Comparisons of these migration parameters across generations offers the possibility of identifying if and how the migration parameters change over t ime and assess whether the values in one generation provide more informative estimates than another generation. The results indicate that the grandparents generation provides parameter values that more realistically represent estimates for prehistoric dem ographic processes. Specifically, my results provide empirical estimates for migration parameters that can be used to generate more informative and geographically explicit hypothesis models for prehistoric human processes. This dissertation illustrates the advantages of taking an interdisciplinary approach to address human evolution. Through the use of simulated genetic data, I identify the significant effect migration has had on human genetic variation and how our limited knowledge on human migration const rains the accurate reconstruction of demographic processes and estimation of other demographic parameters. I then address the challenges posed by the effects of migration by estimating empirical values for migration parameters from nongenetic data to generate more realistic hypothesis demographic scenarios and increase the probability of accurately reconstructing demographic processes. Specifically, by integrating data from genetic, geographic and demographic approaches, I provide a framework to develop more realistic and geographically explicit scenarios for the migration out of Africa that should allow for the accurate estimation of some of the demographic parameters that have thus far been a challenge to estimate. PAGE 23 23 More generally, this dissertation contri butes to the understanding of human evolution in various fields of study and provides a general framework for creating informative models that can be used for reconstructing the evolutionary history of many different species. PAGE 24 24 CHAPTER 2 HUMAN DEMOGRAPHIC PROCESSES AND GENETIC VARIATION AS REVEALED BY MTDNA SIMULATIONS Evolutionary history can be described as a series of sequential demographic events that created the genetic complexity observed in the organism under study. Reconstructing evolutionary histor y requires identifying the relevant demographic processes and understanding how these processes have affected patterns of existing genetic variation. To make inferences about human evolutionary processes, such as the first migration of humans out of Africa it is important to know whether different hypothesized demographic processes are distinguishable based on the genetic variation of the present human population. For instance, can we distinguish between a large and small colonizing population for the init ial migration out of Africa? Furthermore, different combinations of demographic parameters, such as population size and gene flow, can interact to generate similar patterns of genetic variation. By looking at small changes in demographic parameters, in com bination with each other, we can determine the influence of these parameters on patterns of genetic variation. With the growth in computational power, simulations now allow us to generate multiple sets of genetic data for complex evolutionary processes. We can compare the simulated datasets to each other to determine how genetic variation changes as demographic parameters change ( Carvajal Rodrguez 2008) and identify which parameter interactions cause detectabl e differences in genetic variation. Although many studies have compared simulations of evolutionary processes to empirical data to make inferences about the empirical data ( Deshpandeet al. 2009; Fagundeset al. 2007; PAGE 25 25 Gronauet al. 2011 ; Lohmuelleret al. 2009 ; Veeramahet al. 2012) few studies have used simulations to investigate the effects of demographic parameters and their interactions on genetic variation ( Calafellet al. 2001 ) Comparing simulated demographic scenarios can help us determine which demographic parameters merit more attention because of their increased effect on genetic variation and can direct the investigation to questions focused on the parameters with greatest effect. Comparisons of simulated scenarios also allow identification of the parameter combinations (and by inference, the d emographic scenarios) that can be distinguished from each other based on the genetic variation of each scenario. For instance, comparing simulated scenarios could improve our understanding of the critical period in human history when anatomically modern hu mans left Africa and colonized the rest of the planet. Although many studies have focused on estimating specific values for parameters of interest for the colonization of humans out of Africa, e.g. ( DeGiorgioet al. 2009; Fagundeset al. 2007; Gronauet al. 2011) the large variances for some of these estimat es suggest it would be useful to better understand how specific values for each parameter, and their interactions, affect genetic variation. Three parameters of particular interest for the colonization of humans out of Africa are i) the size of the coloniz ing population, ii) the timing of the event and iii) the amount of subsequent gene flow into and out of Africa. Examining the interaction of these primary parameters using estimates drawn from the literature should give insight into which of the parameters has a larger effect on genetic variation and whether increased efforts to refine the value will lead to increased resolution of other parameters of interest. PAGE 26 26 Demographic parameters have generally been inferred from statistics that summarize the patterns o f genetic variation. Although Lohse and Kelleher ( 2009) show that likelihood methods provide better estimates of demographic parameters, recent studies show that it is still common practice to use summary statistics ( Bustamante and Ramachandran 2009; Deshpandeet al. 2009 ; Keinanet al. 2008) The use of summary statistics has be come increasingly popular in methods of Bayesian inference, such as approximate Bayesian computation (ABC) ( Beaumont 2010 ; Beaumontet al. 2002; Marjoram and Tavar 2006) In brief, the ABC approach compares summary statistics calculated from an empirical data set with summary statistics calculated from sim ulated scenarios that serve as hypotheses to explain the empirical data. For ABC, and other approaches, it is necessary to know whether different demographic scenarios lead to different summary statistic values, thus reflecting the effect of different demographic parameter combinations on genetic variation. Furthermore, it is essential to determine, and is largely lacking in the current literature, which summary statistics are most informative for a parameter or evolutionary process of interest ( Hickersonet al. 2006) In this study, we simulate mitochondrial DNA (mtDNA) nucleotide sequences for 42 alternative demographic scenarios describing the colonization of modern humans out of Africa. The diverse set of parameter combinations allows us to evaluate the influence o f these parameters on genetic variation. Three parameters of primary interest were varied in these scenarios; colonization size (CS), rate of gene flow between African and nonAfrican populations (GF), and the time of the colonization event (TC). Values for these parameters were chosen from the literature to represent realistic demographic scenarios that could have produced the current human genetic variation. Twelve PAGE 27 27 summary statistics were calculated for each of 1,000 simulated datasets produced for 42 dif ferent demographic scenarios. The summary statistics were used to 1) detect differences between demographic scenarios and determine which scenarios could be distinguished from each other, 2) determine the effects of particular demographic parameters (CS, G F, and TC) on genetic variation, and 3) identify the informativeness of different summary statistics on genetic variation. Materials and Methods Models Forty two scenarios were modeled to describe the demographic process for the initial colonization of modern humans out of Africa. In this model (Figure 21), two populations split at one of two possible colonization times, with the colonizing population having one of three possible sizes, followed by seven possible proportions of bidirectional gene flow bet ween the populations. The values for colonization time (TC) were 100,000 and 50,000 years ago, the earliest and most recent estimates for movement of modern humans out of Africa, respectively ( Beharet al. 2008; Gonderet al. 2007; Klein 1998 ; Macaulayet al. 2005; Mellars 2006 ; Salaset al. 2002) The values for colonization size (CS) were estimated as a proportion of the African population and set at 1% th e lowest estimated value of migrants ( Atkinsonet al. 2008; Fagundeset al. 2007) 30%, the highest estimate d value ( Relethford and Harpending 1995; Tenesaet al. 2007) and 10% as an intermediate v alue ( Gronauet al. 2011) The values for gene flow (GF) were 106, 105, 104, 103, 102, 0.1, and 0.5 proportion of migrants from each population per effective individual per generation. These values for GF were selected to begin at Nem<<1 and increase by single orders of magnitude until Nem>>1 to reflect PAGE 28 28 highly structured populations and highly panmictic populations, respectively. The initial population size was fixed at 10,000 ( Atkinsonet al. 2008; Fagundeset al. 2007) with constant population size. Simulations One thousand coalescent simulations for each of the 42 demographic scenarios were generated using SIMCOAL2 ( Laval and Excoffier 2004) One hundred sequences of human mtDNA data wer e simulated for each population with a coding region of 15,446 nucleotides (nt) and a substitution rate of 1.7 x 108 substitutions per site per year ( Atkinsonet al. 2008; Ingmanet al. 2000) and a control region of 1123 nt and a substitution rate of 4.7 x 107 substitutions per site p er year ( Howellet al. 2003) Summary Statistics Twelve summary statistics (Table 21) were calculated for each of the 42,000 simulated mtDNA datasets to capture the genetic variation of the 42 different scenarios. Fst st were calculated between the two populations with ARLEQUIN 3.11 ( Excoffieret al. 2005) Tau hat ) ( was calculated from the mismatch distribution of the simulated mtDNA coding region with R ( R Development Core Team 2010) Number of W Onsins and Rozas R2, Tajimas D (TD), number of singleton sites (NSS), number of haplotypes (# Hap), number of singletons (# Single), and homozygosity (Hmzy) were calcu lated with Sample_stats3, a version of the Sample_stats utility distributed with Hudsons MS ( 2002) modified for DNA sequence data (available at http://github.com/ryanraaum/samplestats). The code to calculate Ramos Onsins and PAGE 29 29 Rozas R2 was incorporated with permission from Mlcoalsim ( Ramos Onsins and MitchellOlds 2007) Statistical Analysis A multi factorial analysis of variance (ANOVA) including all three parameters of inte rest (CS, GF, and TC) was performed for each summary statistic using R ( R Development Core Team 2010) The percent of variation explained by each parameter and parameter interaction was estimated from the additive component of variance, which was in turn calculated from the expected mean square value. Pair wise comparisons (Tukeys tests) were performed for each pair of the 42 demographic scenarios, for a total of 861 comparisons for each of the 12 summary statistics. Results Partitioning of Genetic Variation by Demographic Parameters ANOVAs were used to partition the genetic variation summed across all 42,000 simulated datasets into the tested parameters and their interactions and to determine which parameters and interactions had a significant effect in explaining differences between the 42 demographic scenarios as reflected in each summary statistic (Table 22). For parameters and interactions that were significant in explaining variation, the actual value of explained variance was then calculated to identify how variation was partitioned among the parameters and interactions within each summary statistic. All parameters an d interactions were significant (pvalue<1.0 x 106) in explaining the differences in genetic variation between the 42 demographic scenarios using the following summary statistics: Fstst, S, WD, R2, and (Table 22). TC and TC PAGE 30 30 interactions (i.e. TCxCS and TCxGF) were not significant for NSS, # Hap, # Single, and Hmzy. CS, GF and their interaction (i.e. CSxGF) were significant across all summary statistics and yielded the highest percent of explained variance. CS explained the most variation, ranging from 2.4% to 96.4%, depending on the summary statistic. This was followed by the interaction between CS and GF (3.143.5%) and then GF (0.886.8%). TC and it s interactions with GF and CS explained only a small percent of the variation (TC: 0.1 1.9%, TCxGF: 0.4 5.3%, and TCxCS: 0.10.8%). Comparison of Summary Statistics Percent of explained variation was also compared across summary statistics to determine how each summary statistic partitioned variation into the investigated parameters and interactions (Table 22). Fst st partitioned more variation in GF relative to the other summary statistics, with percent of explained variance of 85.6% and 86.8% respec tively. # Hap, # Single, and Hmzy partitioned the most variation for CS relative to the other summary statistics, with values of 84.3%, 96.4%, 96.1%, and 77.5%, respectively. While TC contributed little to the explained variation, Fs t st reflected more variation due to TC (1.0% and 1.9%) than any other summary statistic. S, WD, R2, and NSS had more similar patterns of partitioning the variation among each parameter and their interactions, with the variation more equally di stributed between CS, GF, and CSxGF than in the other summary statistics. PAGE 31 31 Comparison of Evolutionary Scenarios Pair wise comparisons were performed on all 42 parameter combinations for a total of 861 t tests to determine which of the 42 demographic scenari os could be distinguished from each other (Table 24). Significant differences in the summary statistic estimates indicate a distinguishable difference in genetic variation between the parameter combinations and suggest we could distinguish between the represented demographic scenarios. Four summary statistics ( # Hap, # Single, Hmzy) partitioned the variation primarily into a single parameter (CS) and are not discussed later, as the pair wise comparisons and box plots mainly showed a p attern where the demographic scenarios differed by CS category (Table 24 and Figure 23). It is worth noting, however, that the pair wise comparisons and box plots of have similar trends to those of the summary statistics discussed between demographic scenarios are less distinct, despite being developed as a measure to estimate time of a demographic event ( Rogers 1995) st and Fst also partition the majority of variation into a single category (GF), these summary statistics show more representation by TC and TC interactions (i.e. TCxCS and TCxGF) than the other summary statistics; box plots depicting the range of estimates of eight summary statistics for all 42 demographic scenarios are discussed later (Figures 22 and 24). Fst st, which measure similar aspects of genetic variation, have similar st has smaller differences between the demographic scenarios. Interestingly, estimates of both Fst st, significantly decrease in value as PAGE 32 32 GF increases within each CS category (Figures 22a and 24a and Table 24). GF categories of moderate levels of GF (104 and 103) are signi ficantly different from each other, whereas TC is significantly different for low GF categories (106 and 105). S and W are related measurements that also show similar trends in the box plots. Within the CS category of 1%, high and low levels of GF are c learly differentiated (Figures 22b and 24b and Table 24), but the difference between GF categories diminishes with increasing CS. With respect to TC, 50kya and 100kya produce significantly different estimates from each other when CS=10% for GF>103 and when CS=1% + GF=103. W estimates (Figure 22c), with smaller of R2 and TD show similar patterns in the box plots (Figure 22d and 24c). For CS=10%, categories of GF>103 produce significantly greater estimates than categories of GF of 103 4). Estimates of NSS are significantly different for high and low levels of GF for CS=1% and CS=10% (Figure 22e and Table 24). TC only produces significantly different estimates for CS=1% for GF=103. Most of the discussed summary statistics (except Fst and categories (<104) for CS=1% and CS=10% have similar summary statistics relative to high GF categories for CS=30%. Fst st show a different pattern, thus providing insight into the effects of gene flow by reflecting the diversity between two migrating populations instead of the overall diversity among the two populations, which the other summary statistics reflect. Fst st show a similar pattern for the three CS categories PAGE 33 33 where Fst st estimates decrease as GF category values increase. The discussed summary statistics also show that, in general, TC produced significantly different estimates for GF categories equal or less than 105, but not for high GF categories. The demographic scenarios with CS=1% and GF=103 were gener ally distinct from each other and all other scenarios. Discussion Demographic Parameters Our results illustrate that migration, whether represented as colonization size (CS) or gene flow (GF), shows the largest effect on human genetic variation over the ti me period in which humans colonized the planet. Specifically, our simulations indicate that CS has the largest influence on patterns of genetic variation (percent of variance explained averaged over all summary statistics) and CS, GF and CS x GF explain most of the genetic variation that has arisen in humans, as simulated in this study. Interactions between CS and GF have varying effects on patterns of genetic variation. For example, the majority of box plots (Figures 22 and 24) reveal that low to moderat e levels of GF (< 103) create similar patterns of genetic variation across all CS categories. This supports findings such as those of Kitchen et al. ( 2008 ) where reducing gene flow as low as zero produced a much larger colonization size for peopling of the Americas, in contrast to Heys ( 2005) results where colonization size was one hundred fold smaller with much higher levels of gene flow. However, the extreme values tested here show that large CS with low GF (less than 104) creates a very different effect relative to sm all CS and high GF. This can be explained because large CS with low GF leads to high genetic variation as it increases the difference between the populations, whereas small PAGE 34 34 CS and high GF lowers genetic variation as it decreases the difference between the migrating populations. Summary Statistics This is one of the first studies to investigate the informativeness of specific summary statistics in the inference and comparison of demographic processes ( Hickersonet al. 2006; Sefcet al. 2007) Specifically, we were interested in determining the extent to which different summary statistics can distinguish between the modeled scenarios. Although one summary statistic alone cannot distinguish between all of the scenarios, our results suggest that combining several summary statistics, selected based on the parameters of interest, will allow better resolution when comparing empirical data with simulated data. Our results clearly show that the summary statistics differentially explain variance depending on the demographic parameter (Table 22). In general, summary statistics in which percent of explained variation is distributed across multiple parameters and interactions (such as S, WD, R2, and NSS) are able to distinguish between more demographic scenarios. In contrast, summary statistics where the majority of variation is concentrated in one main parameter (such as # Hap, # Single, and Hmzy) are useful for studies that focus on one parameter, but have more limited utility to distinguish between different demographic scenarios. It is important to note that some demographic scenarios produce genetically similar results that cannot be differentiated regardless of the summary statistic used. This similarity suggests that some evolutionary processes PAGE 35 35 and specific questions, such as the timing of the first human migration out of Africa, may not be resolved based on mitochondrial DNA data. A limitation in meth ods that compare empirical data with simulated data, such as ABC approaches, is that the analyses require a small number of summary statistics to avoid the situation in which so many simulated scenarios are rejected that it is impossible to make any conclusions about the empirical data ( Beaumontet al. 2002; Hamilton 2005; Wegmannet al. 2009) Our results show similarity in box plot profiles across some pairs of summary statistics, most likely because they reflect a similar aspect of the genetic variation, for example, S and W. This similarity offers the p ossibility to reduce the number of summary statistics used when comparing empirical and simulation data. We make specific recommendations on the optimal summary statistics to use based on the parameters and questions of interest for studies of evolutionary history where CS and GF have played a dominant role, such as in the current analysis (Table 23). Relevance to Human Evolution A goal in evolutionary studies is to identify changes in demographic parameters given the observed genetic variation. In most c ases, it is unclear how small changes in the parameters of interest will influence genetic variation, for example, how will small changes in gene flow influence observed levels of genetic variation. We use a model of modern human migration out of Africa, w ith changes in colonization size (CS), time of colonization (TC) and gene flow (GF), to investigate the effect of demographic changes on genetic variation. PAGE 36 36 To address the extent of gene flow between African and nonAfrican populations after the initial m igration out of Africa, values for GF were selected such that GF<104 represents population substructure, GF=104 represents migration equilibrium (Nem and GF>104 represents panmixia. The box plots of the demographic scenarios (Figures 2 2 and 24) i llustrate that when CS=1% (and in some cases when CS=10%), GF=103 appears as a transition point where genetic variation decreases significantly (in a sigmoidal curve) as GF increases. Our pair wise comparisons (Table S 4) show that summary statistic estim ates when GF<103 are similar to each other, but have significantly greater estimates when GF>103, which also have similar estimates to each other. The sharp transition in genetic variation from GF of 104 to 103 reveals a rapid breakdown of population s ubstructure to panmixia within only an order of magnitude increase in gene flow. Thus, it should be possible to distinguish between scenarios on either side of the transition point, i.e. GF=103, but much more difficult to distinguish scenarios within high or low GF, i.e. GF>103 or GF<103. Fst has been commonly used to measure genetic differences between populations as a means to reflect gene flow. Our results allow us to assess the amount of gene flow required to generate different Fst estimates. When we compare our simulated Fst estimates with Fst estimates of African populations versus European populations (0.141) or versus Asian populations (0.235) ( Bowcocket al. 1991) we obs erve areas of overlap between the simulated and empirical estimates of Fst (0.1410.235) (Figures 22a and Table 25). For CS categories of 1% and 10%, we observe the most overlap with empirical Fst estimates at GF=103, whereas for CS=30%, the maximum ove rlap occurs at GF=104. Simulations of bottlenecks show that a CS of 30% PAGE 37 37 is not accompanied by a reduction in genetic diversity as we see empirically between African and nonAfrican populations ( Ramachandranet al. 2005; Relethford 2001 ) suggesting that CS=1% and CS=10% are better estimates of the size of the colonizing population. Our results show that there is almost twice as much overlap with empirical Fst estimates when CS=1% (Figures 22a and Table 25), providing the strongest support for a scenario in which the migrating population carried 1% of exi sting African mitochondrial genetic variation (i.e. CS=1%) and both African and nonAfrican populations experienced bidirectional GF of 103. It is interesting to speculate on what these values mean in terms of actual individuals, particularly with respect to GF and CS. A GF of 103 represents 10 individuals moving per generation in both directions. Although this value seems large, it may not be unreasonable that an average of ten individuals per generation migrated between African and nonAfrican populatio ns, particularly immediately after migration out of Africa when the populations were still geographically close. Deshpande et al. ( 2009) modeled a serial founder population history for migration out of Africa and colonization of Eurasia and they identified 0.01 (100 individuals) as the maximum bidirectional exchange rate between adjoining populations. Furthermore, 1% of the population leaving Africa (CS=1%) to colonize the world seems reasonable under our model of a panmictic population. However, Africa was potentially highly structured before the exit of modern humans out of Africa ( Campbell and Tishkoff 2008; Veeramahet al. 2012) A high degree of structure could imply that East Africa, where humans first emigrated to the rest of the world, was a distinct subpopulat ion, and the colonization size leaving from this area could be larger than 1%, reflecting published PAGE 38 38 colonization sizes of 918% ( Deshpandeet al. 2009; Gronauet al. 2011) Future studies should address the effect of African population substructure on the genetic variation of nonAfrican populations, specifically how African substructure affects our ability to accurately model the evolutionary processes that gave rise to nonAfrican genet ic diversity. Although changes in colonization size (CS) and gene flow (GF) create detectable differences in genetic variation in our study, time of colonization (TC) shows relatively little effect on patterns of genetic variation. For most of the simul ations, demographic scenarios of 50kya are not distinguishable from those of 100kya.This result is probably due to the relatively recent times chosen in the current analysis, i.e. 50kya and 100kya, which were chosen to reflect relevant times for human migr ation. Although time is an important factor affecting genetic diversity in general, anatomically modern humans short existence likely explains why colonization time has not played an important role in human genetic diversity. Notably, for values that may most accurately reflect human demographic history (CS=1% and GF=103), six summary statistics can distinguish between TC of 50kya and 100kya (Figures 2 2be and 24bc), suggesting time is only distinguishable under particular conditions. Our findings demonstrate the utility of comparing simulated demographic scenarios to understand the effect of demographic parameters on genetic variation. Our results show that different demographic parameters have varying effects on contemporary genetic variation. The par ameters that generate a larger difference in genetic variation obscure the differences in genetic variation caused by other PAGE 39 39 parameters, such that scenarios with differences in the lesser effect parameters cannot be distinguished from one another. In the ca se of humans, our comparisons reveal that migration (CS or GF) has such a large effect on genetic variation that scenarios with different times for an event are less likely to be distinguishable, particularly with mtDNA. A better understanding of the three particular parameters addressed in this study (CS, GF, and TC) allows other demographic parameters to be addressed. For example, by narrowing potential values of colonization size to 1% and gene flow to 103, new scenarios can be generated to assess the e ffect of other parameters of interest, such as population growth or the occurrence of gene flow at specific times. In a similar fashion, comparisons of simulated scenarios can provide insight into the evolutionary history of any system. With increased understanding of the effects of demographic parameters on genetic variation, more accurate inferences about evolutionary histories will be possible. PAGE 40 40 Table 21. Summary statistics analyzed and their definitions Summary statistic Reference Notation Definition F st ( Wright 1951 ) F st Proportion of genetic diversity due to allele differences among populations st ( Excoffieret al. 1992) st Proportion of genetic diversity due to haplotype differences among populations Segregating sites ( Fu 1995 ) S Number of polymorphic sites Wattersons theta ( Watterson 1975 ) W S corrected for number of samples Nucleotide diversity ( Nei 1987 ) Average number of nucleotide differences per site Ramos Onsins and Rozas R2 ( Ramos Onsins and Rozas 2002) R2 Test of neutrality based on difference between number of Tajimas D ( Tajima 1989 ) T D Test of neutrality based on difference between S and pair wise differences Number of singleton sites ( Baldinget al. 2003) NSS Number of sites where only one individual has a different allele Tau hat ( Rogers 1995 ) Estimate of time measured in mutational units Number of haplotypes ( Baldinget al. 2003) # Hap Number of different unique allele combinations Number of singletons ( Baldinget al. 2003) # Single Number of haplotypes that appear only once in the sample Homozygosity ( Baldinget al. 2003) Hmzy Probability of two samples in a population having the same h aplotype PAGE 41 41 Table 22. Partitioning of genetic variation by demographic parameter for each summary statistic. For each summary statistic, variation is partitioned by the parameters of int erest (CS,GF,TC) and their interactions and is presented as percent of variation explained. a Interactions between parameters *pvalue <1.0x106Parameter Fst st S W R2 TD NSS # Hap # Singl e Hmz y Colonization size (CS) 6.5* 2.4* 38.3* 38.3* 52.3* 44.4* 44.4* 10.1* 84.3* 96.4* 96.1* 77.5* Gene flow (GF) 85.6* 86.8* 28.8* 28.8* 19.1* 14.4* 14.6* 46.2* 5.4* 1.4* 0.8* 6.1* Time of colonization (TC) 1.0* 1.9* 0.1* 0.1* 0.1* 0.2* 0.2* 0 0.1* 0 0 0 CSxGFa 4.1* 3.1* 31.9* 31.9* 26.4* 36.9* 36.7* 43.5* 8.2* 2.2* 3.1* 16.4* CSxTCa 0.3* 0.3* 0.1* 0.1* 0.2* 0.3* 0.3* 0 0.8* 0 0 0 GFxTCa 2.0* 5.3* 0.4* 0.4* 1.3* 2.4* 2.5* 0.1 0.8* 0 0 0 CSxGFxTC a 0.5 0.4 0.4 0.4 0.6 1.4 1.4 0 0.4 0 0.4 0 PAGE 42 42 Table 23. Recommendation of optimal summary statistic to use for each parameter of interest. aSummary statistics were chosen based on an assessment that identified the statistic with the highest percent of variation explained (Table 22 ) and most extreme differentiation of parameter combinations (Fig ures 2 2, 2 3, and 24 ). Parameter of interest Optimal summary statistica Colonization size # Single Gene flow Fst Time Fst Coloni zation size x gene flow S, W R 2, T D Gene flow x time F st, R 2, T D PAGE 43 43 Table 24. Demographic scenarios compared in Tukeys tests and pvalues for each summary statistic Parameter combination comparison ( GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:100:1 vs 0.001:50:1 0.88663 0.00029 0 0 0 0 0 0.00218 1 1 1 1 0.001:100:1 vs 0.01:50:1 0 1.15E 11 0 0 0 0.02903 0.5498 0 0.99986 0 0 1 0.001:100:1 vs 0.1:50:1 0 4.68E 13 0 0 0 0 0 0 0 0 0 1 0.001:100:1 vs 0.5:50:1 0 4.61E 13 0 0 0.00014 0 0 0 0 0 0 0.4 9813 0.001:100:1 vs 1e 04:100:1 0 0 0 0 0 0 0 0 5.06E 11 0.50251 0 0 0.001:100:1 vs 1e 04:50:1 0 0 0 0 0 0 0 0 5.79E 13 0.11624 0 0 0.001:100:1 vs 1e 05:100:1 0 0 0 0 0 0 0 0 3.82E 13 0.00923 0 0 0.001:100:1 vs 1e 05:50:1 0 0 0 0 0 0 0 0 9.95E 11 0.201 24 3.63E 13 0 0.001:100:1 vs 1e 06:100:1 0 0 0 0 0 0 0 0 5.43E 13 0.08961 0 0 0.001:100:1 vs 1e 06:50:1 0 0 0 0 0 0 0 0 4.41E 13 0.01389 0 0 0.001:100:10 vs 0.001:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.001:100:10 vs 0.001:50:1 0 3.08E 05 0 0 0 0 0 5.24E 13 0 0 0 0 0.001:100:10 vs 0.001:50:10 1 0.99999 0 0 0 4.02E 13 4.02E 13 0.00325 1 1 1 0 0.001:100:10 vs 0.01:100:1 0.00141 3.68E 11 0 0 0 0 0 0 0 0 0 0 PAGE 44 44 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:100:1 vs 0.001:50:1 0.88663 0.00029 0 0 0 0 0 0.00218 1 1 1 1 0.001:100:1 vs 0.01:50:1 0 1.15E 11 0 0 0 0.02903 0.5498 0 0.99986 0 0 1 0.001:100:1 vs 0.1:50:1 0 4.68E 13 0 0 0 0 0 0 0 0 0 1 0.001:100:1 vs 0.5:50:1 0 4.61E 13 0 0 0.00014 0 0 0 0 0 0 0.4 9813 0.001:100:1 vs 1e 04:100:1 0 0 0 0 0 0 0 0 5.06E 11 0.50251 0 0 0.001:100:1 vs 1e 04:50:1 0 0 0 0 0 0 0 0 5.79E 13 0.11624 0 0 0.001:100:1 vs 1e 05:100:1 0 0 0 0 0 0 0 0 3.82E 13 0.00923 0 0 0.001:100:1 vs 1e 05:50:1 0 0 0 0 0 0 0 0 9.95E 11 0.201 24 3.63E 13 0 0.001:100:1 vs 1e 06:100:1 0 0 0 0 0 0 0 0 5.43E 13 0.08961 0 0 0.001:100:1 vs 1e 06:50:1 0 0 0 0 0 0 0 0 4.41E 13 0.01389 0 0 0.001:100:10 vs 0.001:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.001:100:10 vs 0.001:50:1 0 3.08E 05 0 0 0 0 0 5.24E 13 0 0 0 0 0.001:100:10 vs 0.001:50:10 1 0.99999 0 0 0 4.02E 13 4.02E 13 0.00325 1 1 1 0 0.001:100:10 vs 0.01:100:1 0.00141 3.68E 11 0 0 0 0 0 0 0 0 0 0 0.001:100:10 vs 0.01:50:1 0.00013 2.78E 10 0 0 0 0 0 0 0 0 0 0 PAGE 45 45 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:100:10 vs 0.01:50:10 0 1.21E 09 0 0 0.00015 0 0 0 0 0 1 6.45E 12 0.001:100:10 vs 0.1:100:1 0 1.74E 12 0 0 0 0 0 0 0 0 0 0 0.001:100:10 vs 0.1:50:1 0 1.96E 12 0 0 0 0 0 0 0 0 0 0 0.001:100:10 vs 0.1:50:10 0 5.67E 12 0 0 1.70E 11 0 0 0 0 0 1 0 0.001:100:10 vs 0.5:100:1 0 1.51E 12 0 0 0 0. 99977 0.99743 0 0 0 4.75E 13 0 0.001:100:10 vs 0.5:50:1 0 1.83E 12 0 0 0 0 0 0 0 0 4.29E 13 0 0.001:100:10 vs 0.5:50:10 0 3.84E 12 0.99998 0.99998 0 0 0 0 0 0 1 0 0.001:100:10 vs 1e 04:100:1 0 0 0 0 4.39E 13 0.98808 0.99314 8.52E 08 0 0 0 0 0.001:100:1 0 vs 1e 04:100:10 0 0 0 0 0 3.16E 13 3.42E 13 6.72E 12 1.96E 12 0 1 0 0.001:100:10 vs 1e 04:50:1 0 0 0 0 4.17E 13 0.99586 0.9978 0.00288 0 0 0 0 0.001:100:10 vs 1e 04:50:10 0 0 0 0 0 5.00E 13 5.66E 13 1.55E 11 4.12E 13 0 1 0 0.001:100:10 vs 1e 05:100:1 0 0 0 0 0 0 0 3.85E 13 0 0 0 2.16E 09 0.001:100:10 vs 1e 05:100:10 0 0 0 0 0 0 0 0 4.36E 13 0 1 0 0.001:100:10 vs 1e 05:50:1 0 0 0 0 0 5.35E 13 5.89E 13 4.27E 13 0 0 0 0.9404 0.001:100:10 vs 1e 05:50:10 0 0 0 0 0 0 0 4.08E 13 3.02E 13 0 1 0 PAGE 46 4 6 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:100:1 0 vs 1e 06:100:1 0 0 0 0 0 0 0 4.15E 13 0 0 0 3.44E 13 0.001:100:10 vs 1e 06:100:10 0 0 0 0 0 0 0 3.98E 13 3.70E 13 0 1 0 0.001:100:10 vs 1e 06:50:1 0 0 0 0 0 4.88E 13 5.24E 13 1.47E 08 0 0 0 0.99994 0.001:100:10 vs 1e 06:50:10 0 0 0 0 0 0 0 1.94E 12 3. 21E 13 0 1 0 0.001:100:30 vs 0.001:100:1 0 0.78442 0 0 0 0 0 0 0 0 0 0 0.001:100:30 vs 0.001:100:10 0 0.97029 0 0 0 0 0 4.90E 13 0 0 1 0 0.001:100:30 vs 0.001:50:1 0 6.05E 12 0 0 0 0 0 0 0 0 0 0 0.001:100:30 vs 0.001:50:10 0 0.06322 0 0 0 2.36E 06 2.91 E 06 0.00024 0 0 1 0 0.001:100:30 vs 0.001:50:30 1 1 0.99993 0.99993 0.29182 0.56511 0.5592 0.99728 1 1 1 0.00412 0.001:100:30 vs 0.01:100:1 5.28E 13 0.00011 0 0 0 0 0 0 0 0 0 0 0.001:100:30 vs 0.01:100:10 6.53E 14 0.00054 0 0 0 0.95964 0.98616 0 0 0 1 0 0.001:100:30 vs 0.01:50:1 3.72E 13 0.00045 0 0 0 0 0 0 0 0 0 0 0.001:100:30 vs 0.01:50:10 5.60E 14 0.00122 0 0 0 1.24E 07 7.69E 08 0 0 0 1 0 0.001:100:30 vs 0.01:50:30 0 0.00016 0.00167 0.00167 1 0.03597 0.03653 7.65E 08 0.84174 0.00167 1 0.48128 0.001:100:30 vs 0.1:100:1 0 9.90E 06 0 0 0 0 0 0 0 0 0 0 PAGE 47 47 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:100:30 vs 0.1:100:10 0 1.89E 05 0 0 0 4.93E 13 5.03E 13 0 1.49E 13 0 1 0 0.001:100:30 vs 0.1:50:1 0 1.11E 05 0 0 0 0 0 0 0 0 0 0 0.001:100:30 vs 0.1:50:10 0 2.94E 05 0 0 0 0 0 0 4.07E 13 0 1 0 0.001:100:30 vs 0.1:50:30 0 2.36E 05 0.00167 0.00167 0.91289 4.37E 05 4.99E 05 4.22E 13 0.21063 4.47E 07 1 0.08323 0.001:100:30 vs 0.5:100:1 0 8.51E 06 0 0 0 0 0 0 0 0 1.21E 13 0 0.001:100:30 vs 0.5:100:10 0 2.01E 05 0 0 3.84E 13 7.62E 13 9.18E 13 0 1 1 1 0 0.001:100:30 vs 0.5:50:1 0 1.04E 05 0 0 0 0.98564 0.99112 0 0 0 0 0 0.001:100:30 vs 0.5:50:10 0 2.03E 05 0 0 6.52E 05 0 0 0 1 1 1 5.07E 13 0.001:100:30 vs 0.5:50:30 0 2.18E 05 0.00102 0.00102 0.06706 0.97502 0.97359 0.87869 0.01178 6.72E 12 1 4.24E 13 0.001:100:30 vs 1e 04:100:1 0 0 0 0 0 0 0 0.34141 0 0 0 0 0.001:100:30 vs 1e 04:100:10 0 0 0.50244 0.50244 2.12E 08 4.51E 07 4.13E 07 0.99832 0 0 0.9989 0 0.001:100:30 vs 1e 04:100:30 0 0 7.90E 05 7.90E 05 0.0265 0.94783 0.95041 0.9091 1 0.88561 1 0 .00315 0.001:100:30 vs 1e 04:50:1 0 0 3.98E 13 3.98E 13 0 0 0 0.00027 0 0 0 0 0.001:100:30 vs 1e 04:50:10 0 0 0.00034 0.00034 4.03E 13 7.48E 11 6.04E 11 0.99516 0 0 0.99823 0 0.001:100:30 vs 1e 04:50:30 0 0 0.00399 0.00399 0.80317 1 1 0.99986 1 0.79828 1 0.80178 PAGE 48 48 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:100:30 vs 1e 05:100:1 0 0 0.02396 0.02396 0.68619 1 1 1 0 0 0 0 0.001:100:30 vs 1e 05:100:10 0 0 1 1 1 1 1 0.99996 0 0 0.99678 0 0.001:100:30 vs 1e 05:100:30 0 0 3.33E 06 3.33E 06 8.85E 06 0.02366 0.02465 0.86836 0.99996 0.63273 1 0.0866 7 0.001:100:30 vs 1e 05:50:1 0 0 4.98E 13 4.95E 13 5.02E 13 5.73E 11 5.36E 11 1 0 0 0 0 0.001:100:30 vs 1e 05:50:10 0 0 0.98819 0.98819 0.02793 0.12307 0.1148 1 0 0 0.99819 0 0.001:100:30 vs 1e 05:50:30 0 0 0.00044 0.00044 0.10254 0.99558 0.99606 0.9754 2 0.99991 0.50527 1 0.93678 0.001:100:30 vs 1e 06:100:1 0 0 0.3622 0.3622 0.99992 1 1 1 0 0 0 0 0.001:100:30 vs 1e 06:100:10 0 0 1 1 0.99991 0.99974 0.99973 1 0 0 0.99488 0 0.001:100:30 vs 1e 06:100:30 0 0 2.57E 12 2.57E 12 7.36E 13 3.54E 06 3.69E 06 0. 46882 0.9981 0.17981 1 7.54E 06 0.001:100:30 vs 1e 06:50:1 0 0 3.49E 13 3.48E 13 4.48E 13 8.74E 11 8.41E 11 0.55738 0 0 0 0 0.001:100:30 vs 1e 06:50:10 0 0 0.99977 0.99977 0.00327 0.0186 0.01744 0.99983 0 0 0.99729 0 0.001:100:30 vs 1e 06:50:30 0 0 5.37 E 07 5.37E 07 0.00074 0.66763 0.68201 0.80593 0.99997 0.62177 1 0.001 0.001:50:1 vs 1e 04:50:1 0 0 0 0 0 0 0 0 2.56E 11 0.21171 1.21E 13 0 0.001:50:1 vs 1e 05:50:1 0 0 0 0 0 0 0 0 1.24E 08 0.33648 3.33E 13 0 0.001:50:1 vs 1e 06:50:1 0 0 0 0 0 0 0 0 3.74E 13 0.03156 0 0 PAGE 49 49 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:50:10 vs 0.001:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.001:50:10 vs 0.001:50:1 0 0.04412 0 0 0 0 0 0 0 0 0 0 0.001:50:10 vs 0.01:100:1 0.00264 5.28E 13 0 0 0 0 0 0 0 0 0 0 0.001:50:10 vs 0.01:50:1 0.00026 3.17E 13 0 0 0 0 0 0 0 0 0 0 0.001:50:10 vs 0.1:100:1 0 3.39E 13 0 0 0 0 0 0 0 0 0 0 0.001:50:10 vs 0.1:50:1 0 3.47E 13 0 0 0 0 0 0 0 0 0 0 0.001:50:10 vs 0.5:100:1 0 3.28E 13 0 0 0 1.01E 12 3.42E 12 0 0 0 4.74E 13 0 0.001:50:10 vs 0.5:50:1 0 3.43E 13 0 0 0 0.08214 0.0751 8 0 0 0 4.25E 13 0 0.001:50:10 vs 1e 04:100:1 0 0 3.82E 13 3.84E 13 0.55455 1.12E 11 6.88E 12 0.99909 0 0 0 0 0.001:50:10 vs 1e 04:50:1 0 0 3.56E 13 3.56E 13 0.61015 4.67E 12 3.09E 12 1 0 0 0 0 0.001:50:10 vs 1e 04:50:10 0 0 0 0 0.01133 0.99999 0.99998 0.47556 3.66E 13 0 1 1 0.001:50:10 vs 1e 05:100:1 0 0 0 0 3.41E 13 0.00144 0.00167 0.00107 0 0 0 4.29E 08 0.001:50:10 vs 1e 05:50:1 0 0 0 0 0.44136 0.99999 0.99998 0.05665 0 0 0 0 0.001:50:10 vs 1e 05:50:10 0 0 0 0 3.36E 13 0.96551 0.97624 0.00137 2.77E 13 0 1 0.89561 0.001:50:10 vs 1e 06:100:1 0 0 0 0 1.40E 13 6.34E 06 7.74E 06 0.00147 0 0 0 0.00096 PAGE 50 50 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:50:10 vs 1e 06:50:1 0 0 0 0 0.51228 1 0.99999 0.9893 0 0 0 0 0.001:50:10 vs 1e 06:50:10 0 0 0 0 6.30E 13 0.99976 0.99988 0.24781 2.93E 13 0 1 0.869 79 0.001:50:30 vs 0.001:100:1 0 0.77075 0 0 0 0 0 0 0 0 0 0 0.001:50:30 vs 0.001:100:10 0 0.96651 0 0 0 0 0 0 0 0 1 0 0.001:50:30 vs 0.001:50:1 0 5.24E 12 0 0 0 0 0 0 0 0 0 0 0.001:50:30 vs 0.001:50:10 0 0.05924 0 0 0 2.92E 13 3.05E 13 1.00E 09 0 0 1 0 0.001:50:30 vs 0.01:100:1 4.69E 13 0.00012 0 0 0 0 0 0 0 0 0 0 0.001:50:30 vs 0.01:100:10 3.85E 13 0.00059 0 0 0 8.99E 05 0.0002 0 0 0 1 0 0.001:50:30 vs 0.01:50:1 4.43E 13 0.0005 0 0 0 0 0 0 0 0 0 0 0.001:50:30 vs 0.01:50:10 4.01E 13 0.00133 0 0 0 0. 21534 0.18082 0 0 0 1 0 0.001:50:30 vs 0.1:100:1 0 1.11E 05 0 0 0 0 0 0 0 0 0 0 0.001:50:30 vs 0.1:100:10 0 2.10E 05 0 0 0 5.97E 05 6.68E 05 0 0 0 1 0 0.001:50:30 vs 0.1:50:1 0 1.24E 05 0 0 0 0 0 0 0 0 0 0 0.001:50:30 vs 0.1:50:10 0 3.27E 05 0 0 0 0 0 0 3.93E 13 0 1 0 0.001:50:30 vs 0.5:100:1 0 9.51E 06 0 0 0 0 0 0 0 0 1.12E 13 0 PAGE 51 51 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:50:30 vs 0.5:100:10 0 2.24E 05 0 0 0 0.00017 0.00022 0 1 1 1 0 0.001:50:30 vs 0.5:50:1 0 1.16E 05 0 0 0 0.0002 0.00027 0 0 0 0 0 0.001:50:30 vs 0.5:50:10 0 2.26E 05 0 0 3.51E 13 4.14E 13 4.51E 13 0 1 1 1 0 0.001:50:30 vs 1e 04:100:1 0 0 0 0 0 0 0 0.00014 0 0 0 0 0.001:50:30 vs 1e 04:100:10 0 0 0.00171 0.00171 3.36E 13 4.88E 13 4.80E 13 0.04806 0 0 0.99878 0 0.001:50:30 vs 1e 04:50:1 0 0 0 0 0 0 0 1.21E 09 0 0 0 0 0 .001:50:30 vs 1e 04:50:10 0 0 1.41E 08 1.41E 08 0 4.20E 13 3.86E 13 0.03232 0 0 0.99805 0 0.001:50:30 vs 1e 04:50:30 0 0 0.66255 0.66255 1 0.99934 0.9992 1 0.99987 0.48876 1 0.99932 0.001:50:30 vs 1e 05:100:1 0 0 5.91E 06 5.91E 06 6.74E 07 0.02012 0.0198 3 0.97037 0 0 0 0 0.001:50:30 vs 1e 05:100:10 0 0 1 1 0.82373 0.99667 0.99628 1 0 0 0.99647 0 0.001:50:30 vs 1e 05:50:1 0 0 3.15E 13 3.15E 13 0 3.88E 13 3.80E 13 0.34895 0 0 0 0 0.001:50:30 vs 1e 05:50:10 0 0 0.06514 0.06514 1.32E 10 3.30E 08 2.70E 08 0 .95922 0 0 0.99801 0 0.001:50:30 vs 1e 05:50:30 0 0 0.28421 0.28421 1 1 1 1 0.99841 0.2244 1 0.99098 0.001:50:30 vs 1e 06:100:1 0 0 0.00075 0.00075 0.00045 0.42371 0.41897 0.95552 0 0 0 0 0.001:50:30 vs 1e 06:100:10 0 0 1 1 0.99997 1 1 0.99999 0 0 0.994 43 0 PAGE 52 52 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.001:50:30 vs 1e 06:50:1 0 0 3.58E 13 3.58E 13 0 4.47E 13 4.35E 13 0.00053 0 0 0 0 0.001:50:30 vs 1e 06:50:10 0 0 0.20982 0.20982 3.00E 12 7.45E 10 6.29E 10 0.09233 0 0 0.99703 0 0.001:50:30 vs 1e 06:50:30 0 0 0.00474 0.00474 0.99998 1 1 1 0.99928 0.31098 1 1 0.01:100:1 vs 0.001:100:1 0 1.90E 12 0 0 0 4.52E 13 4.76E 09 0 1 0 0 1 0.01:100:1 vs 0.001:50:1 0 0 0 0 0 0 0 0 0.99852 0 0 1 0.01:100:1 vs 0.01:50:1 1 1 0.01902 0.01902 0.99959 0.00859 0.05214 1 0.55582 0.99979 0.98176 1 0.01:100:1 vs 0.1 :50:1 0 1 0 0 1 0 0 0 0 0 1.16E 11 1 0.01:100:1 vs 0.5:50:1 0 1 0.99243 0.99243 1.21E 13 0 0 0 0 0 0 0.94464 0.01:100:1 vs 1e 04:100:1 0 0 0 0 0 0 0 0 3.34E 13 0 0 0 0.01:100:1 vs 1e 04:50:1 0 0 0 0 0 0 0 0 3.46E 13 0 0 0 0.01:100:1 vs 1e 0 5:100:1 0 0 0 0 0 0 0 0 4.24E 13 0 0 0 0.01:100:1 vs 1e 05:50:1 0 0 0 0 0 0 0 0 3.58E 13 0 0 0 0.01:100:1 vs 1e 06:100:1 0 0 0 0 0 0 0 0 3.36E 13 0 0 0 0.01:100:1 vs 1e 06:50:1 0 0 0 0 0 0 0 0 4.04E 13 0 0 0 0.01:100:10 vs 0.001:100:1 0 1.50 E 11 3.48E 13 3.48E 13 0 0 0 0 0 0 0 0 PAGE 53 53 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.01:100:10 vs 0.001:100:10 0 3.60E 10 0 0 6.50E 06 0 0 0 0 0 1 0.0133 0.01:100:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.01:100:10 vs 0.001:50:10 0 3.18E 13 0 0 0 0.13299 0.09045 0 0 0 1 0 0.01:100:10 vs 0.01:10 0:1 0 1 0 0 0 0 0 0 0 0 0 0 0.01:100:10 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 0.01:100:10 vs 0.01:50:10 1 1 0 0 3.17E 13 3.35E 13 3.59E 13 1 1 1 1 0 0.01:100:10 vs 0.1:100:1 1 1 0 0 0 0 0 0 0 0 0 0 0.01:100:10 vs 0.1:50:1 1 1 0 0 0 0 0 0 0 0 0 0 0.01:100:10 vs 0.1:50:10 0.73425 1 7.46E 08 7.46E 08 0 0 0 0 1.59E 05 5.29E 13 1 0 0.01:100:10 vs 0.5:50:1 0.62743 1 0 0 0 1 1 0 0 0 3.77E 13 0 0.01:100:10 vs 0.5:50:10 0.45014 1 0 0 0 0 0 0 0 0 1 0 0.01:100:10 vs 1e 04:100:1 0 0 0 0 0 0 0 0 0 0 0 6.09E 10 0 .01:100:10 vs 1e 04:100:10 0 0 0 0 0 0.0565 0.03021 0 0 0 1 0 0.01:100:10 vs 1e 04:50:1 0 0 0 0 0 0 0 0 0 0 0 6.97E 13 0.01:100:10 vs 1e 04:50:10 0 0 0 0 0 0.00024 8.80E 05 0 0 0 1 0 PAGE 54 54 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.01:100:10 vs 1e 05:100:1 0 0 0 0 0 1 1 0 0 0 0 7.46E 14 0.01:100:10 vs 1e 05:100:10 0 0 0 0 0 0.29932 0.4251 0 0 0 1 0 0.01:100:10 vs 1e 05:50:1 0 0 0 0 0 0.0002 8.09E 05 0 0 0 0 0.99931 0.01:100:10 vs 1e 05:50:10 0 0 0 0 0 1 1 0 0 0 1 0 0.01:100:10 vs 1e 06:100:1 0 0 0 0 0 0.98673 0.99653 0 0 0 0 0 0.01:100:10 vs 1e 06:100:10 0 0 0 0 0 0.02396 0.0432 0 0 0 1 0 0.01:100:10 vs 1e 06:50:1 0 0 0 0 0 0.00026 0.00011 0 0 0 0 0.85815 0.01:100:10 vs 1e 06:50:10 0 0 0 0 0 0.99942 0.9963 0 0 0 1 0 0.01:100:30 vs 0.001:100:1 0 3.14E 12 0 0 0 0 0 0 0 0 0 0 0.01:100:30 vs 0.00 1:100:10 0 6.79E 11 0 0 0 0 0 0 0 0 0.99999 0 0.01:100:30 vs 0.001:100:30 0 0.00017 5.10E 12 5.09E 12 1 0.01462 0.01397 8.06E 09 0.75022 0.00056 1 1 0.01:100:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.01:100:30 vs 0.001:50:10 0 2.76E 13 0 0 0 3.81E 13 3 .97E 13 0 0 0 0.99999 0 0.01:100:30 vs 0.001:50:30 0 0.00019 3.59E 13 3.59E 13 0.64956 1 1 0.00104 0.89517 0.00334 1 0.09071 0.01:100:30 vs 0.01:100:1 0 1 0 0 0 0 0 0 0 0 0 0 PAGE 55 55 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.01:100:30 vs 0.01:100:10 0.99749 1 0 0 0 4.29E 08 1.15E 07 9.59E 13 0 0 1 0 0.01:100:30 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 0.01:100:30 vs 0.01:50:10 0.998 1 0 0 0 0.98613 0.97856 3.41E 13 0 0 1 0 0.01:100:30 vs 0.01:50:30 1 1 0.47422 0.47422 1 1 1 1 1 1 1 0.97299 0.01:100:30 vs 0.1:100:1 1 1 0 0 0 0 0 0 0 0 0 0 0.01:100:30 vs 0.1:100:10 1 1 0 0 0 0.02283 0.02516 0 0 0 1 0 0.01:100:30 vs 0.1:50:1 1 1 0 0 0 0 0 0 0 0 0 0 0.01:100:30 vs 0.1:50:10 1 1 0 0 0 0 0 0 0 0 1 0 0.01:100:30 vs 0.1:50:30 1 1 0.47387 0.47387 0.99539 1 1 0.07003 1 1 1 0.57872 0.01:100:30 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 6.53E 14 0 0.01:100:30 vs 0.5:100:10 1 1 0 0 1.87E 13 0.04614 0.05717 0 0.59462 7.66E 05 1 0 0.01:100:30 vs 0.5:50:1 1 1 0 0 0 1.20E 07 1.67E 07 0 0 0 0 0 0.01:100:30 vs 0.5:50:10 1 1 0 0 5.40E 06 4.20E 10 7.14E 10 0 0.42396 1.00E 05 1 3.59E 13 0.01:100:30 vs 0.5:50:30 1 1 0 0 0.2438 0.99787 0.99739 0.01308 0.99999 0.6995 1 4.03E 13 0.01:100:30 vs 1e 04:100:1 0 0 0.05191 0.05191 0 0 0 4.70E 13 0 0 0 0 PAGE 56 56 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.01:100:30 vs 1e 04:100:10 0 0 0.00143 0.00143 9.79E 10 4.07E 13 3.87E 13 3.34E 1 3 0 0 0.99696 0 0.01:100:30 vs 1e 04:100:30 0 0 0 0 0.11881 0.9993 0.99912 0.01003 0.02212 5.65E 11 1 0.07466 0.01:100:30 vs 1e 04:50:1 0 0 0.56006 0.56006 0 0 0 0 0 0 0 0 0.01:100:30 vs 1e 04:50:10 0 0 0.74643 0.74643 2.84E 13 0 0 3.13E 13 0 0 0.99541 0 0.01:100:30 vs 1e 04:50:30 0 0 0 0 0.97779 0.44606 0.4283 0.00029 0.01432 1.84E 11 1 0.9991 0.01:100:30 vs 1e 05:100:1 0 0 0.10057 0.10057 0.3227 4.99E 05 4.77E 05 9.60E 10 0 0 0 0 0.01:100:30 vs 1e 05:100:10 0 0 1.21E 12 1.20E 12 1 0.3245 0.31283 0.0 0019 0 0 0.99224 0 0.01:100:30 vs 1e 05:100:30 0 0 0 0 0.0001 1 1 0.01416 0.00788 4.05E 12 1 0.5896 0.01:100:30 vs 1e 05:50:1 0 0 1 1 4.12E 13 0 0 1.16E 12 0 0 0 0 0.01:100:30 vs 1e 05:50:10 0 0 1.29E 05 1.29E 05 0.00464 3.62E 12 2.88E 12 6.59E 10 0 0 0 .99532 0 0.01:100:30 vs 1e 05:50:30 0 0 0 0 0.33198 0.98339 0.98034 0.00368 0.00596 1.61E 12 1 0.99997 0.01:100:30 vs 1e 06:100:1 0 0 0.00315 0.00315 0.99081 0.00736 0.00704 5.91E 10 0 0 0 0 0.01:100:30 vs 1e 06:100:10 0 0 3.05E 13 3.04E 13 1 0.92032 0. 9163 1.65E 07 0 0 0.98837 0 0.01:100:30 vs 1e 06:100:30 0 0 0 0 8.67E 12 0.99996 0.99997 0.09187 0.00193 4.34E 13 1 0.00059 0.01:100:30 vs 1e 06:50:1 0 0 1 1 3.55E 13 0 0 4.40E 13 0 0 0 0 PAGE 57 57 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.01:100:30 vs 1e 06:50:10 0 0 1.27E 06 1.27E 06 0.0004 4.60E 13 4.42E 13 3.93E 13 0 0 0.99333 0 0.01:100:30 vs 1e 06:50:30 0 0 0 0 0.00564 1 1 0.02128 0.00812 3.71E 12 1 0.03147 0.01:50:1 vs 0.001:50:1 0 0 0 0 0 0 0 0 1 0 0 1 0.01:50:1 vs 1e 04:50:1 0 0 0 0 0 0 0 0 5.27E 08 0 0 0 0.01:50:1 vs 1e 05:50:1 0 0 0 0 0 0 0 0 1.05E 05 0 0 0 0.01:50:1 vs 1e 06:50:1 0 0 0 0 0 0 0 0 3.90E 11 0 0 0 0.01:50:10 vs 0.001:100:1 0 5.38E 11 0 0 0 0 0 0 0 0 0 0 0.01:50:10 vs 0.001:50:1 0 0 7.11E 05 7.11E 05 0 0 0 0 0 0 0 0 0.01:50:10 vs 0.001:50:10 0 3.66E 13 0 0 4.72 E 07 0 0 0 0 0 1 5.84E 06 0.01:50:10 vs 0.01:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.01:50:10 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 0.01:50:10 vs 0.1:100:1 1 1 0 0 0 0 0 0 0 0 0 0 0.01:50:10 vs 0.1:50:1 1 1 0 0 0 0 0 0 0 0 0 0 0.01:50:10 vs 0.5:100:1 0.55749 1 0 0 0 0 0 0 0 0 4.19E 13 0 0.01:50:10 vs 0.5:50:1 0.64697 1 0 0 0 3.77E 13 3.78E 13 0 0 0 3.96E 13 0 PAGE 58 58 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.01:50:10 vs 1e 04:100:1 0 0 0 0 0.36286 0 0 0 0 0 0 0 0.01:50:10 vs 1e 04:50:1 0 0 0 0 0.31532 0 0 0 0 0 0 0 0.01:50:10 vs 1e 04:50:10 0 0 0 0 3.65E 13 0 0 0 0 0 1 1.29E 09 0.01:50:10 vs 1e 05:100:1 0 0 0 0 0 2.73E 11 1.61E 11 0 0 0 0 1 0.01:50:10 vs 1e 05:50:1 0 0 0 0 4.25E 13 0 0 0 0 0 0 3.84E 13 0.01:50:10 vs 1e 05:50:10 0 0 0 0 0 4.88E 13 4.11E 13 0 0 0 1 5.37E 13 0.01:50:10 vs 1e 06:100:1 0 0 0 0 0 4.13E 08 2.56E 08 0 0 0 0 1 0.01:50:10 vs 1e 06:50:1 0 0 0 0 4.62E 13 0 0 0 0 0 0 3.73E 13 0.01:50:10 vs 1e 06:50:10 0 0 0 0 0 3.84E 13 4.19E 13 0 0 0 1 6.11E 13 0.01:50:30 vs 0.001:100:1 0 2.98E 12 0 0 0 0 0 0 0 0 0 0 0.01:50:30 vs 0.001:100:10 0 6.40E 11 0 0 0 0 0 0 0 0 0.99999 0 0.01:50:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.01:50:30 vs 0.001:50:10 0 2.76E 13 0 0 0 3.60E 13 3.79E 13 0 0 0 0.99999 0 0.01:50:30 vs 0.001:50:30 0 0.00018 1.23E 07 1.23E 07 0. 99657 1 1 0.00481 0.9462 0.00899 1 1 0.01:50:30 vs 0.01:100:1 0 1 0 0 0 0 0 0 0 0 0 0 PAGE 59 59 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.01:50:30 vs 0.01:100:10 0.99801 1 0 0 0 2.00E 07 5.71E 07 4.43E 13 0 0 1 0 0.01:50:30 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 0.01:50:30 vs 0.01:50:10 0.99843 1 0 0 0 0.93872 0.91017 3.94E 13 0 0 1 0 0.01:50:30 vs 0.1:100:1 1 1 0 0 0 0 0 0 0 0 0 0 0.01:50:30 vs 0.1:100:10 1 1 0 0 0 0.00889 0.00926 0 0 0 1 0 0.01:50:30 vs 0.1:50:1 1 1 0 0 0 0 0 0 0 0 0 0 0.01:50:30 vs 0.1:50:10 1 1 0 0 0 0 0 0 0 0 1 0 0.01:50:30 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 7.46E 14 0 0.01:50:30 vs 0.5:100:10 1 1 0 0 0 0.01923 0.02298 0 0.70731 0.00025 1 0 0.01:50:30 vs 0.5:50:1 1 1 0 0 0 5.39E 07 8.16E 07 0 0 0 0 0 0.01:50:30 vs 0.5:50:10 1 1 0 0 2.15E 08 7.34E 11 1.13E 10 0 0.5379 3.66E 05 1 0 0.01:50:30 vs 1e 04:100:1 0 0 2.54E 09 2.54E 09 0 0 0 4.79E 13 0 0 0 0 0.01:50:30 vs 1e 04:100:10 0 0 0.99992 0.99992 1.82E 12 3.67E 13 3.60E 13 4.54E 13 0 0 0.99712 0 0.01:50:30 vs 1e 04:50:1 0 0 1.16E 06 1.16E 06 0 0 0 0 0 0 0 0 0.01:50:30 vs 1e 04:50:10 0 0 1 1 4.22E 13 0 0 3.99E 13 0 0 0.99563 0 PAGE 60 60 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.01:50:30 vs 1e 04:50:30 0 0 4.48E 13 4.48E 13 1 0.64731 0.64235 0.0015 0.02426 1.03E 10 1 1 0.01:50:30 vs 1e 05:100:1 0 0 1 1 0.0216 0.00018 0.00018 1.00E 08 0 0 0 0 0.01:50:30 vs 1e 05:100:10 0 0 0.0005 0.0005 1 0.51356 0.51293 0.00101 0 0 0.99259 0 0.01:50:30 vs 1e 05:50:1 0 0 0.12725 0.12725 2.98E 13 0 0 9.21E 12 0 0 0 0 0.01:50:30 vs 1e 05:50:10 0 0 0.86499 0.86499 6.29E 05 1.93E 11 1.67E 11 7.01E 09 0 0 0.995 55 0 0.01:50:30 vs 1e 05:50:30 0 0 4.79E 13 4.79E 13 0.94463 0.99775 0.99755 0.0151 0.01052 6.72E 12 1 1 0.01:50:30 vs 1e 06:100:1 0 0 0.99999 0.99999 0.5521 0.01919 0.01958 6.32E 09 0 0 0 0 0.01:50:30 vs 1e 06:100:10 0 0 5.10E 06 5.10E 06 1 0.9799 0.98065 1.34E 06 0 0 0.98886 0 0.01:50:30 vs 1e 06:50:1 0 0 0.03902 0.03902 1.87E 13 0 0 3.64E 13 0 0 0 0 0.01:50:30 vs 1e 06:50:10 0 0 0.5732 0.5732 3.26E 06 7.44E 13 7.13E 13 6.43E 13 0 0 0.99364 0 0.01:50:30 vs 1e 06:50:30 0 0 4.15 E 13 4.15E 13 0.13839 1 1 0.07065 0.01413 1.88E 11 1 0.99976 0.1:100:1 vs 0.001:100:1 0 4.60E 13 0 0 0 0 0 0 0 0 0 1 0.1:100:1 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 1 0.1:100:1 vs 0.01:100:1 0 1 0 0 1 0 0 0 0 0 1.83E 08 1 0.1:100:1 vs 0.01:50:1 0 1 0 0 0 .90829 0 0 0 0 0 3.00E 13 1 PAGE 61 61 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.1:100:1 vs 0.1:50:1 1 1 1 1 1 0.99802 0.99846 1 1 1 1 1 0.1:100:1 vs 0.5:50:1 0.99936 1 0 0 0 0 0 0.86402 0 0 0 0.93151 0.1:100:1 vs 1e 04:100:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:100:1 vs 1e 04:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:100:1 vs 1e 05:100:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:100:1 vs 1e 05:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:100:1 vs 1e 06:100:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:100:1 vs 1e 06:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:100:10 vs 0.001:100: 1 0 5.50E 13 0.99929 0.99929 0 0 0 0 0 0 0 0 0.1:100:10 vs 0.001:100:10 0 3.53E 12 0 0 0.28411 0 0 0 0 0 1 1 0.1:100:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:100:10 vs 0.001:50:10 0 3.81E 13 0 0 0 0 0 0 0 0 1 0 0.1:100:10 vs 0.01:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.1:100:10 vs 0.01:100:10 0.65153 1 1.58E 11 1.58E 11 0.92459 3.96E 13 4.10E 13 0 0.00027 7.38E 13 1 0.65483 0.1:100:10 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 PAGE 62 62 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.1:100:10 vs 0.01:50:10 0.67078 1 0 0 3.96E 13 0.99812 0.99926 0 0.00017 6.24E 13 1 4.98E 13 0.1:100:10 vs 0.1:100:1 0.99954 1 0 0 0 0 0 0 0 0 0 0 0.1:100:10 vs 0.1:50:1 0.99459 1 0 0 0 0 0 0 0 0 0 0 0.1:100:10 vs 0.1:50:10 1 1 0 0 3.45E 13 2.84E 13 2.93E 13 0.98952 1 1 1 0 0.1:100:10 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 3.74E 13 0 0.1: 100:10 vs 0.5:50:1 1 1 0 0 0 3.93E 13 3.92E 13 0 0 0 1.40E 13 0 0.1:100:10 vs 0.5:50:10 1 1 0 0 0 0.45254 0.4927 0 4.08E 13 0 1 0 0.1:100:10 vs 1e 04:100:1 0 0 0 0 0 0 0 0 0 0 0 4.80E 13 0.1:100:10 vs 1e 04:100:10 0 0 0 0 0 0 0 0 0 0 1 0 0.1:100:10 vs 1e 04:50:1 0 0 0 0 0 0 0 0 0 0 0 2.05E 13 0.1:100:10 vs 1e 04:50:10 0 0 0 0 0 0 0 0 0 0 0.99999 0 0.1:100:10 vs 1e 05:100:1 0 0 0 0 0 3.35E 13 3.43E 13 0 0 0 0 4.73E 13 0.1:100:10 vs 1e 05:100:10 0 0 0 0 0 1.26E 10 1.35E 10 0 0 0 0.99998 0 0.1:100:10 vs 1e 05:50:1 0 0 0 0 0 0 0 0 0 0 0 1 0.1:100:10 vs 1e 05:50:10 0 0 0 0 0 0 0 0 0 0 0.99999 0 PAGE 63 63 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.1:100:10 vs 1e 06:100:1 0 0 0 0 0 4.00E 13 4.10E 13 0 0 0 0 4.33E 13 0.1:100:10 vs 1e 06:100:10 0 0 0 0 0 3.94E 08 4.37E 08 0 0 0 0.99996 0 0.1:100:10 vs 1e 06:50:1 0 0 0 0 0 0 0 0 0 0 0 1 0.1:100:10 vs 1e 06:50:10 0 0 0 0 0 0 0 0 0 0 0.99999 0 0.1:100:30 vs 0.001:100:1 0 6.53E 13 0 0 0 0 0 0 0 0 0 0 0.1:100:30 vs 0.001:100:10 0 5.51E 12 0 0 0 0 0 0 0 0 0.99999 0 0.1:100:30 vs 0.001:100:30 0 2.88E 05 3.24E 13 3.24E 13 1 0.05056 0.05307 4.33E 13 0.28668 1.43E 06 1 1 0.1:100:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:100:30 vs 0.001:50:10 0 4.06E 13 0 0 0 4.10E 13 4.38E 13 0 0 0 0.99999 0 0.1:100:30 vs 0.001:50 :30 0 3.20E 05 3.89E 13 3.89E 13 0.12009 1 1 4.43E 13 0.46698 1.35E 05 1 0.21712 0.1:100:30 vs 0.01:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.1:100:30 vs 0.01:100:10 0.6213 1 0 0 0 3.68E 07 1.10E 06 0.00481 0 0 1 0 0.1:100:30 vs 0.01:100:30 1 1 1 1 1 1 1 0.13295 1 1 1 1 0.1:100:30 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 0.1:100:30 vs 0.01:50:10 0.64091 1 0 0 0 0.90204 0.85846 2.60E 06 0 0 1 0 PAGE 64 64 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.1:100:30 vs 0.01:50:30 1 1 0.0294 0.0294 0.99996 1 1 0.04492 1 1 1 0.99764 0.1:100:30 vs 0.1:100:1 0.9993 1 0 0 0 0 0 0 0 0 0 0 0.1:100:30 vs 0.1:100:10 1 1 0 0 0 0.00593 0.00593 0 0 0 1 0 0.1:100:30 vs 0.1:50:1 0.99267 1 0 0 0 0 0 0 0 0 0 0 0.1:100:30 vs 0.1:50:10 1 1 0 0 0 0 0 0 0 0 1 0 0.1:100:30 vs 0.1:50:30 1 1 0.02936 0.02936 0.70639 0.99999 0.99999 1 1 1 1 0.8160 1 0.1:100:30 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 6.53E 14 0 0.1:100:30 vs 0.5:100:10 1 1 0 0 4.10E 13 0.01316 0.01523 0 0.17666 1.27E 07 1 0 0.1:100:30 vs 0.5:50:1 1 1 0 0 0 9.77E 07 1.56E 06 0 0 0 0 0 0.1:100:30 vs 0.5:50:10 1 1 0 0 0.00037 3.57E 11 5.12 E 11 0 0.09795 1.12E 08 1 3.41E 13 0.1:100:30 vs 0.5:50:30 1 1 0 0 0.02018 0.99997 0.99997 3.58E 12 1 0.99982 1 6.61E 13 0.1:100:30 vs 1e 04:100:1 0 0 0.6099 0.6099 0 0 0 0 0 0 0 0 0.1:100:30 vs 1e 04:100:10 0 0 7.88E 06 7.88E 06 1.89E 07 4.33E 13 4.31E 13 0 0 0 0.99638 0 0.1:100:30 vs 1e 04:100:30 0 0 0 0 0.00698 0.99999 0.99999 2.36E 12 0.00193 3.90E 13 1 0.1851 0.1:100:30 vs 1e 04:50:1 0 0 0.99675 0.99675 0 0 0 0 0 0 0 0 PAGE 65 65 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.1:100:30 vs 1e 04:50:10 0 0 0.09057 0.09057 7.06E 13 0 0 0 0 0 0.99459 0 0.1:100:30 vs 1e 04:50:30 0 0 0 0 0.5363 0.7251 0.72807 3.53E 13 0.00115 3.44E 13 1 0.99998 0.1:100:30 vs 1e 05:100:1 0 0 0.00215 0.00215 0.90216 0.00029 0.00031 4.21E 13 0 0 0 0 0.1:100:30 vs 1e 05:100:10 0 0 2.75E 13 2.75E 13 1 0.59599 0.60371 3.29E 13 0 0 0.99097 0 0.1:100:30 vs 1e 05:100:30 0 0 0 0 1.22E 06 1 1 4.09E 12 0.00057 2.97E 13 1 0.82411 0.1:100:30 vs 1e 05:50:1 0 0 1 1 4.01E 13 0 0 0 0 0 0 0 0.1:100:30 vs 1e 05:50:10 0 0 2.79E 08 2.79E 08 0.08855 3.98E 11 3.71E 11 4.21E 13 0 0 0.99449 0 0.1:100:30 vs 1e 05:50:30 0 0 0 0 0.0332 0.99918 0.99919 7.61E 13 0.00041 5.49E 13 1 1 0.1:100:30 vs 1e 06:100:1 0 0 2.11E 05 2.11E 05 1 0.02766 0.02925 3.96E 13 0 0 0 0 0.1:100:30 vs 1e 06:100:10 0 0 3.99E 13 3.99E 13 0.996 0.99009 0.99122 3.48E 13 0 0 0.98661 0 0.1:100:30 vs 1e 06:100:30 0 0 0 0 4.07E 13 0.99767 0.99751 1.85E 10 0.00011 3.46E 13 1 0.00246 0.1:100:30 vs 1e 06:50:1 0 0 1 1 3.73E 13 0 0 0 0 0 0 0 0.1:100:30 vs 1e 06:50:10 0 0 1.88E 09 1.88E 09 0.01347 1.06E 12 1.06E 12 0 0 0 0.99222 0 0.1:100:30 vs 1e 06:50:30 0 0 0 0 0.00014 1 1 8.09E 12 0.00059 2.93E 13 1 0.08923 0.1:50:1 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 1 PAGE 66 66 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.1:50:1 vs 0.01:50:1 0 1 0 0 0.99998 0 0 0 0 0 3.74E 13 1 0.1:50: 1 vs 1e 04:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:50:1 vs 1e 05:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:50:1 vs 1e 06:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:50:10 vs 0.001:100:1 0 6.60E 13 0 0 0 0 0 0 0 0 0 0 0.1:50:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:50:10 vs 0.001:50:10 0 4.13E 13 0 0 0.08099 0 0 0 0 0 1 0.99431 0.1:50:10 vs 0.01:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.1:50:10 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 0.1:50:10 vs 0.01:50:10 0.7518 1 3.25E 13 3.27E 13 0.92991 3.94E 13 4.11E 13 0 9.6 9E 06 5.02E 13 1 0.09243 0.1:50:10 vs 0.1:100:1 0.99988 1 0 0 0 0 0 0 0 0 0 0 0.1:50:10 vs 0.1:50:1 0.99792 1 0 0 0 0 0 0 0 0 0 0 0.1:50:10 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 4.18E 13 0 0.1:50:10 vs 0.5:50:1 1 1 0 0 0 0 0 0 0 0 1.21E 13 0 0.1:50:10 vs 1e 04:100:1 0 0 0 0 1 0 0 0 0 0 0 0 PAGE 67 67 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.1:50:10 vs 1e 04:50:1 0 0 0 0 1 0 0 0 0 0 0 0 0.1:50:10 vs 1e 04:50:10 0 0 0 0 1.15E 12 0 0 0 0 0 0.99999 0.26954 0.1:50:10 vs 1e 05:100:1 0 0 0 0 0 0 0 0 0 0 0 0.00468 0.1:50:10 vs 1e 05:50:1 0 0 0 0 4.89E 09 0 0 0 0 0 0 0 0.1:50:10 vs 1e 05:50:10 0 0 0 0 0 0 0 0 0 0 0.99999 0.00332 0.1:50:10 vs 1e 06:100:1 0 0 0 0 0 0 0 0 0 0 0 0.72667 0.1:50:10 vs 1e 06:50:1 0 0 0 0 8.74E 09 0 0 0 0 0 0 0 0.1:50:10 vs 1e 06:50:10 0 0 0 0 0 0 0 0 0 0 0.99998 0.00316 0.1:50:30 vs 0.001:100:1 0 6.01E 13 0 0 0 0 0 0 0 0 0 0 0.1:50:30 vs 0.001:100:10 0 4.60E 12 0 0 0 0 0 0 0 0 0.99999 0 0.1:50:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.1:50:30 vs 0.001:50:10 0 3.90E 13 0 0 0 0 0 0 0 0 0.99999 0 0.1:50:30 vs 0.001:50:30 0 2.62E 05 1.23E 07 1.23E 07 1 0.9178 0.92943 3.61E 13 0.36642 4.56E 06 1 1 0.1:50:30 vs 0.01:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.1:50:30 vs 0.01:100:10 0.49755 1 0 0 0 6.69E 12 2.73E 11 0.01144 0 0 1 0 PAGE 68 68 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.1:50:30 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 0.1:50:30 vs 0.01:50:10 0.51734 1 0 0 0 1 1 8.70E 06 0 0 1 0 0.1:50:30 vs 0.01:50:30 1 1 1 1 1 1 1 0.02106 1 1 1 1 0.1:50:30 vs 0.1:100:1 0.99688 1 0 0 0 0 0 0 0 0 0 0 0.1:50:30 vs 0.1:100:10 1 1 0 0 0 0. 61412 0.60556 0 0 0 1 0 0.1:50:30 vs 0.1:50:1 0.97833 1 0 0 0 0 0 0 0 0 0 0 0.1:50:30 vs 0.1:50:10 1 1 0 0 0 0 0 0 0 0 1 0 0.1:50:30 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 5.60E 14 0 0.1:50:30 vs 0.5:100:10 1 1 0 0 0 0.76772 0.78652 0 0.12363 3.72E 08 1 0 0 .1:50:30 vs 0.5:50:1 1 1 0 0 0 2.19E 11 4.22E 11 0 0 0 0 0 0.1:50:30 vs 0.5:50:10 1 1 0 0 4.27E 12 1.45E 06 1.82E 06 0 0.06541 3.06E 09 1 0 0.1:50:30 vs 1e 04:100:1 0 0 2.53E 09 2.53E 09 0 0 0 0 0 0 0 0 0.1:50:30 vs 1e 04:100:10 0 0 0.99992 0.99992 5.05E 13 0 0 0 0 0 0.99617 0 0.1:50:30 vs 1e 04:50:1 0 0 1.15E 06 1.15E 06 0 0 0 0 0 0 0 0 0.1:50:30 vs 1e 04:50:10 0 0 1 1 3.08E 13 0 0 0 0 0 0.99428 0 PAGE 69 69 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.1:50:30 vs 1e 04:50:30 0 0 4.45E 13 4.45E 13 1 0.01045 0.01107 2.99E 13 0.00063 2.98E 13 1 1 0.1:50:30 vs 1e 05:100:1 0 0 1 1 8.89E 05 3.04E 08 3.64E 08 3.95E 13 0 0 0 0 0.1:50:30 vs 1e 05:100:10 0 0 0.0005 0.0005 0.99959 0.00541 0.00587 2.85E 13 0 0 0.99051 0 0.1:50:30 vs 1e 05:50:1 0 0 0.1271 0.1271 0 0 0 0 0 0 0 0 0.1:50:30 vs 1e 05:50:10 0 0 0.86521 0.86521 4.97E 08 5.49E 13 5.43E 13 3.45E 13 0 0 0.99418 0 0.1:50:30 vs 1e 05:50:30 0 0 4.81E 13 4.81E 13 1 0.23389 0.24073 4.85E 13 0.00022 4.53E 13 1 1 0.1:50:30 vs 1e 06:100:1 0 0 1 1 0.01952 1.72E 05 1.99E 05 3.17E 13 0 0 0 0 0.1:50:30 vs 1e 06:100:10 0 0 5.11E 06 5.11E 06 1 0.11237 0.12129 3.83E 13 0 0 0.98597 0 0.1:50:30 vs 1e 06:50:1 0 0 0.03896 0.03896 0 0 0 0 0 0 0 0 0.1:50:30 vs 1e 06:50:10 0 0 0. 57356 0.57356 1.39E 09 3.11E 13 3.15E 13 0 0 0 0.99181 0 0.1:50:30 vs 1e 06:50:30 0 0 4.14E 13 4.14E 13 0.93431 0.86045 0.8641 2.13E 12 0.00031 5.23E 13 1 1 0.5:100:1 vs 0.001:100:1 0 4.48E 13 0 0 0 0 0 0 0 0 0 0.01013 0.5:100:1 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0.04094 0.5:100:1 vs 0.01:100:1 0 1 0 0 0.00744 0 0 0 0 0 0 0.11284 0.5:100:1 vs 0.01:50:1 0 1 0 0 0.85995 0 0 0 0 0 0 0.01282 PAGE 70 70 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:100:1 vs 0.1:100:1 0.99802 1 4.02E 13 4.02E 13 0.00031 0 0 0.97831 0 0 0 0.09962 0.5:100:1 vs 0.1:50:1 0.98438 1 3.25E 13 3.28E 13 0.01711 0 0 0.94921 0 0 0 0.06586 0.5:100:1 vs 0.5:50:1 1 1 3.39E 13 3.39E 13 1.17E 08 0 0 1 0.82029 0.63273 1 1 0.5:100:1 vs 1e 04:100:1 0 0 0 0 0 1 1 0 0 0 0 0 0.5:100:1 vs 1e 04:50:1 0 0 0 0 0 1 1 0 0 0 0 0 0.5:100:1 vs 1e 05:100:1 0 0 0 0 0 0 0 0 0 0 0 0 0.5:100:1 vs 1e 05:50:1 0 0 0 0 0 5.08E 08 2.80E 07 0 0 0 0 0 0.5:100:1 vs 1e 06:100:1 0 0 0 0 0 0 0 0 0 0 0 0 0.5:100:1 vs 1e 06:50:1 0 0 0 0 0 3.50E 08 1.91E 07 0 0 0 0 0 0.5:100:10 vs 0.001:100:1 0 5.59E 13 0 0 0 0 0 0 0 0 0 0 0.5:100:10 vs 0.001:100:10 0 3.79E 12 3.66E 13 3.65E 13 0 0 0 0 0 0 1 0 0.5:100:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.5:100:10 vs 0.001:50:10 0 3.85E 13 0 0 0.98721 0 0 0 0 0 1 3.75E 13 0 .5:100:10 vs 0.01:100:1 0 1 0 0 0 0 0 0.99997 0 0 0 0 0.5:100:10 vs 0.01:100:10 0.49807 1 0 0 0 4.32E 13 4.37E 13 0 0 0 1 0 PAGE 71 71 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:100:10 vs 0.01:50:1 0 1 0 0 0 0 0 0.97686 0 0 0 0 0.5:100:10 vs 0.01:50:10 0.51786 1 0 0 4.89E 13 0.99979 0.99996 0 0 0 1 0 0.5:100:10 vs 0.1:100:1 0.9969 1 0 0 0 0 0 0 0 0 0 0 0.5:100:10 vs 0.1:100:10 1 1 0 0 0 1 1 0 4.04E 13 0 1 0 0.5:100:10 vs 0.1:50:1 0.97842 1 0 0 0 0 0 0 0 0 0 0 0.5:100:10 vs 0.1:50:10 1 1 0 0 2.52E 06 4.97E 13 5.02E 13 0 4.45E 13 0 1 3.74E 13 0.5:100:10 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 1.21E 13 0 0.5:100:10 vs 0.5:50:1 1 1 0 0 0 3.92E 13 3.48E 13 0 0 0 0 0 0.5:100:10 vs 0.5:50:10 1 1 1.66E 09 1.66E 09 0.0001 0.30498 0.3129 1 1 1 1 3.28E 13 0.5:100:10 vs 1e 04:100:1 0 0 0 0 0.00021 0 0 0 0 0 0 0 0.5:100:10 vs 1e 04:100:10 0 0 0 0 0.03885 0 0 0 0 0 0.99896 4.93E 13 0.5:100:10 vs 1e 04:50:1 0 0 0 0 0.00029 0 0 0 0 0 0 0 0.5:100:10 vs 1e 04:50:10 0 0 0 0 0.98988 0 0 0 0 0 0.99834 2.20E 10 0.5:100:10 vs 1e 05:100:1 0 0 0 0 4.64E 12 4.0 9E 13 4.32E 13 0 0 0 0 0 0.5:100:10 vs 1e 05:100:10 0 0 0 0 0 5.14E 10 7.14E 10 0 0 0 0.99695 0.57067 PAGE 72 72 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:100:10 vs 1e 05:50:1 0 0 0 0 1 0 0 0 0 0 0 0 0.5:100:10 vs 1e 05:50:10 0 0 0 0 3.87E 08 0 0 0 0 0 0.9983 1.02E 06 0.5:100:10 vs 1e 06: 100:1 0 0 0 0 2.85E 13 5.15E 13 5.67E 13 0 0 0 0 0 0.5:100:10 vs 1e 06:100:10 0 0 0 0 0 1.39E 07 1.93E 07 0 0 0 0.99514 0.99969 0.5:100:10 vs 1e 06:50:1 0 0 0 0 1 0 0 0 0 0 0 0 0.5:100:10 vs 1e 06:50:10 0 0 0 0 1.08E 06 0 0 0 0 0 0.99744 3.85E 06 0.5:100:30 vs 0.001:100:1 0 5.81E 13 0 0 0 0 0 0 0 0 0 0 0.5:100:30 vs 0.001:100:10 0 4.26E 12 0 0 0 0 0 0 0 0 0.99998 0 0.5:100:30 vs 0.001:100:30 0 2.21E 05 0.00022 0.00022 0.00011 0.09797 0.09865 0.86226 0.0057 1.06E 12 1 3.73E 13 0.5:100:30 vs 0 .001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.5:100:30 vs 0.001:50:10 0 3.85E 13 0 0 0 3.00E 13 3.15E 13 1.13E 11 0 0 0.99998 0 0.5:100:30 vs 0.001:50:30 0 2.47E 05 0.20036 0.20036 0.99788 1 1 1 0.01514 1.12E 11 1 1.47E 06 0.5:100:30 vs 0.01:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.5:100:30 vs 0.01:100:10 0.45654 1 0 0 0 1.28E 06 3.43E 06 0 0 0 1 0 0.5:100:30 vs 0.01:100:30 1 1 0 0 0.00101 1 1 0.01481 0.99989 0.43475 1 4.70E 13 PAGE 73 73 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:100:30 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 0.5:100:30 vs 0.01:50:10 0.47596 1 0 0 0 0.7 8676 0.73159 0 0 0 1 0 0.5:100:30 vs 0.01:50:30 1 1 3.91E 13 3.91E 13 0.04062 1 1 0.05168 0.99941 0.26548 1 6.23E 11 0.5:100:30 vs 0.1:100:1 0.99513 1 0 0 0 0 0 0 0 0 0 0 0.5:100:30 vs 0.1:100:10 1 1 0 0 0 0.00244 0.00258 0 0 0 1 0 0.5:100:30 vs 0.1:10 0:30 1 1 0 0 1.76E 05 1 1 4.42E 12 1 0.99461 1 3.70E 13 0.5:100:30 vs 0.1:50:1 0.9702 1 0 0 0 0 0 0 0 0 0 0 0.5:100:30 vs 0.1:50:10 1 1 0 0 0 0 0 0 0 0 1 0 0.5:100:30 vs 0.1:50:30 1 1 3.94E 13 3.94E 13 0.71666 0.99985 0.99989 1.28E 12 1 0.99915 1 9.28E 09 0.5:100:30 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 0 0 0.5:100:30 vs 0.5:100:10 1 1 0 0 0 0.00571 0.00702 0 0.00245 4.28E 13 1 0 0.5:100:30 vs 0.5:50:1 1 1 0 0 0 3.29E 06 4.82E 06 0 0 0 0 0 0.5:100:30 vs 0.5:50:10 1 1 0 0 4.03E 13 7.82E 12 1.23E 11 0 0.000 95 3.12E 13 1 0 0.5:100:30 vs 0.5:50:30 1 1 1 1 1 1 1 1 1 1 1 1 0.5:100:30 vs 1e 04:100:1 0 0 0 0 0 0 0 4.75E 06 0 0 0 0 PAGE 74 74 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:100:30 vs 1e 04:100:10 0 0 7.13E 13 7.14E 13 0 3.71E 13 3.68E 13 0.00439 0 0 0.99475 0 0.5:100:30 vs 1e 04:100:30 0 0 1 1 1 1 1 1 4.40E 06 3.83E 13 1 2.13E 06 0.5:100:30 vs 1e 04:50:1 0 0 0 0 0 0 0 1.38E 11 0 0 0 0 0.5:100:30 vs 1e 04:50:10 0 0 3.73E 13 3.73E 13 0 1.21E 13 1.03E 13 0.00268 0 0 0.99231 0 0.5:100:30 vs 1e 04:50:30 0 0 1 1 0.85507 0.86036 0.8558 9 1 2.27E 06 4.32E 13 1 3.53E 12 0.5:100:30 vs 1e 05:100:1 0 0 4.56E 13 4.56E 13 9.71E 13 0.00078 0.0008 0.64461 0 0 0 0 0.5:100:30 vs 1e 05:100:10 0 0 0.00075 0.00075 0.00297 0.75865 0.75659 1 0 0 0.98753 0 0.5:100:30 vs 1e 05:100:30 0 0 1 1 1 1 1 1 9.37E 07 3.45E 13 1 8.51E 09 0.5:100:30 vs 1e 05:50:1 0 0 0 0 0 9.33E 14 9.33E 14 0.06546 0 0 0 0 0.5:100:30 vs 1e 05:50:10 0 0 2.51E 10 2.51E 10 4.18E 13 1.78E 10 1.52E 10 0.60012 0 0 0.99218 0 0.5:100:30 vs 1e 05:50:30 0 0 1 1 0.99998 0.99994 0.99993 1 6.23E 07 1.49E 13 1 8.49E 13 0.5:100:30 vs 1e 06:100:1 0 0 4.94E 13 4.89E 13 2.95E 09 0.0567 0.05733 0.58728 0 0 0 0 0.5:100:30 vs 1e 06:100:10 0 0 0.02902 0.02902 0.146 0.99838 0.99844 0.98986 0 0 0.98191 0 0.5:100:30 vs 1e 0 6:100:30 0 0 0.72921 0.72921 0.62517 0.98692 0.98736 1 1.25E 07 0 1 0.00117 0.5:100:30 vs 1e 06:50:1 0 0 0 0 0 1.31E 13 1.31E 13 2.17E 05 0 0 0 0 PAGE 75 75 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:100:30 vs 1e 06:50:10 0 0 4.15E 09 4.15E 09 4.03E 13 3.30E 12 3.01E 12 0.01012 0 0 0.98916 0 0.5: 100:30 vs 1e 06:50:30 0 0 1 1 1 1 1 1 9.77E 07 3.17E 13 1 9.18E 06 0.5:50:1 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0.79308 0.5:50:1 vs 0.01:50:1 0 1 3.11E 07 3.11E 07 4.13E 13 0 0 0 0 0 0 0.54774 0.5:50:1 vs 0.1:50:1 0.9931 1 0 0 3.92E 13 0 0 0.77643 0 0 0 0.87616 0.5:50:1 vs 1e 04:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.5:50:1 vs 1e 05:50:1 0 0 0 0 0 8.71E 05 5.93E 05 0 0 0 0 0 0.5:50:1 vs 1e 06:50:1 0 0 0 0 0 0.00012 8.20E 05 0 0 0 0 0 0.5:50:10 vs 0.001:100:1 0 5.61E 13 0 0 0 0 0 0 0 0 0 0 0.5:50:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.5:50:10 vs 0.001:50:10 0 3.80E 13 0 0 7.90E 11 0 0 0 0 0 1 0 0.5:50:10 vs 0.01:100:1 0 1 0 0 0 0 0 0.99541 0 0 0 0 0.5:50:10 vs 0.01:50:1 0 1 0 0 0 0 0 0.81507 0 0 0 0 0.5:50:10 vs 0.01:50:10 0. 46949 1 0 0 0 0.00034 0.00067 0 0 0 1 0 0.5:50:10 vs 0.1:100:1 0.99479 1 0 0 0 0 0 0 0 0 0 0 PAGE 76 76 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:50:10 vs 0.1:50:1 0.96872 1 0 0 0 0 0 0 0 0 0 0 0.5:50:10 vs 0.1:50:10 1 1 0 0 9.33E 14 3.59E 06 3.79E 06 0 3.61E 13 0 1 0 0.5:50:10 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 1.31E 13 0 0.5:50:10 vs 0.5:50:1 1 1 0 0 0 0 0 0 0 0 0 0 0.5:50:10 vs 1e 04:100:1 0 0 0 0 4.21E 13 0 0 0 0 0 0 0 0.5:50:10 vs 1e 04:50:1 0 0 0 0 3.74E 13 0 0 0 0 0 0 0 0.5:50:10 vs 1e 04:50:10 0 0 0 0 0.41232 0 0 0 0 0 0.9 9844 0 0.5:50:10 vs 1e 05:100:1 0 0 0 0 0.88695 0 0 0 0 0 0 0 0.5:50:10 vs 1e 05:50:1 0 0 0 0 0.00979 0 0 0 0 0 0 0 0.5:50:10 vs 1e 05:50:10 0 0 0 0 1 0 0 0 0 0 0.99841 0 0.5:50:10 vs 1e 06:100:1 0 0 0 0 0.10614 0 0 0 0 0 0 0 0.5:50:10 vs 1e 06:50:1 0 0 0 0 0.00688 0 0 0 0 0 0 0 0.5:50:10 vs 1e 06:50:10 0 0 0 0 1 0 0 0 0 0 0.9976 0 0.5:50:30 vs 0.001:100:1 0 5.77E 13 0 0 0 0 0 0 0 0 0 0 0.5:50:30 vs 0.001:100:10 0 4.18E 12 0 0 0 0 0 0 0 0 0.99998 0 PAGE 77 77 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:50:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 0.5:50:30 vs 0.001:50:10 0 3.91E 13 0 0 0 6.59E 13 7.48E 13 1.43E 11 0 0 0.99998 0 0.5:50:30 vs 0.001:50:30 0 2.43E 05 0.41239 0.41239 1 1 1 1 0.0295 1.19E 10 1 0.00131 0.5:50:30 vs 0.01:100:1 0 1 0 0 0 0 0 0 0 0 0 0 0.5:50:30 vs 0.01:100:10 0.43881 1 0 0 0 0.00293 0.00619 0 0 0 1 0 0.5:50:30 vs 0.01:50:1 0 1 0 0 0 0 0 0 0 0 0 0 0.5:50:30 vs 0.01:50:10 0.45801 1 0 0 0 0.02177 0.01582 0 0 0 1 0 0.5:50:30 vs 0.01:50:30 1 1 3.33E 13 3.33E 13 0.89416 0.99986 0.99985 0.0 4638 0.99994 0.50527 1 3.61E 07 0.5:50:30 vs 0.1:100:1 0.99413 1 0 0 0 0 0 0 0 0 0 0 0.5:50:30 vs 0.1:100:10 1 1 0 0 0 1.00E 06 1.03E 06 0 0 0 1 0 0.5:50:30 vs 0.1:50:1 0.96595 1 0 0 0 0 0 0 0 0 0 0 0.5:50:30 vs 0.1:50:10 1 1 0 0 0 0 0 0 0 0 1 0 0.5:5 0:30 vs 0.1:50:30 1 1 3.28E 13 3.28E 13 1 0.40302 0.41511 1.10E 12 1 0.99999 1 2.29E 05 0.5:50:30 vs 0.5:100:1 1 1 0 0 0 0 0 0 0 0 0 0 0.5:50:30 vs 0.5:100:10 1 1 0 0 0 3.19E 06 4.08E 06 0 0.00529 8.02E 13 1 0 PAGE 78 78 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:50:30 vs 0.5:50:1 1 1 0 0 0 0.00583 0.0 079 0 0 0 0 0 0.5:50:30 vs 0.5:50:10 1 1 0 0 4.83E 13 4.89E 13 5.14E 13 0 0.00216 4.09E 13 1 0 0.5:50:30 vs 1e 04:100:1 0 0 0 0 0 0 0 5.70E 06 0 0 0 0 0.5:50:30 vs 1e 04:100:10 0 0 3.40E 12 3.40E 12 3.62E 13 4.23E 13 4.21E 13 0.00503 0 0 0. 99503 0 0.5:50:30 vs 1e 04:50:1 0 0 0 0 0 0 0 1.75E 11 0 0 0 0 0.5:50:30 vs 1e 04:50:10 0 0 3.92E 13 3.92E 13 0 4.78E 13 4.70E 13 0.00309 0 0 0.9927 0 0.5:50:30 vs 1e 04:50:30 0 0 1 1 1 1 1 1 6.35E 06 4.01E 13 1 2.76E 08 0.5:50:30 v s 1e 05:100:1 0 0 2.86E 13 2.95E 13 2.38E 08 0.20421 0.21078 0.67043 0 0 0 0 0.5:50:30 vs 1e 05:100:10 0 0 0.00325 0.00325 0.40857 1 1 1 0 0 0.98811 0 0.5:50:30 vs 1e 05:50:1 0 0 0 0 0 4.52E 13 4.58E 13 0.07255 0 0 0 0 0.5:50:30 vs 1e 05:50:10 0 0 2.40E 09 2.40E 09 3.08E 12 2.88E 06 2.64E 06 0.62656 0 0 0.99257 0 0.5:50:30 vs 1e 05:50:30 0 0 1 1 1 1 1 1 1.81E 06 4.29E 13 1 4.41E 09 0.5:50:30 vs 1e 06:100:1 0 0 1.24E 12 1.24E 12 2.97E 05 0.92753 0.93112 0.61383 0 0 0 0 0.5:50:30 vs 1e 06:100:10 0 0 0.08597 0.08597 0.99027 1 1 0.99209 0 0 0.98268 0 0.5:50:30 vs 1e 06:50:1 0 0 0 0 0 4.93E 13 4.92E 13 2.58E 05 0 0 0 0 PAGE 79 79 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 0.5:50:30 vs 1e 06:50:10 0 0 3.52E 08 3.52E 08 4.55E 13 9.49E 08 9.04E 08 0. 01149 0 0 0.98967 0 0.5:50:30 vs 1e 06:50:30 0 0 1 1 1 1 1 1 2.80E 06 4.12E 13 1 0.00531 1e04:100:1 vs 0.001:50:1 0 0 0 0 0 0 0 0 6.70E 09 0.68381 0 0 1e04:100:1 vs 0.01:50:1 0 0 0 0 0 0 0 0 6.22E 06 0 0 0 1e04:100:1 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:1 vs 0.5:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:1 vs 1e 04:50:1 1 1 1 1 1 1 1 0.99932 1 1 1 1 1e04:100:1 vs 1e 05:100:1 0 0 2.55E 11 2.55E 11 0 0 0 0.59095 0.99881 1 1 0 1e04:100:1 vs 1e 05:50:1 0 0 0.26424 0.26424 8.73E 07 6.30E 0 7 5.55E 07 0.99701 1 1 0.57183 2.76E 13 1e04:100:1 vs 1e 06:100:1 0 0 4.18E 13 4.19E 13 0 0 0 0.6482 1 1 1 0 1e04:100:1 vs 1e 06:50:1 0 0 0.54336 0.54336 1.46E 06 4.43E 07 3.81E 07 1 0.99978 1 1 4.55E 13 1e04:100:10 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:10 vs 0.001:50:10 0 0 0 0 6.78E 07 1 1 0.38532 1.02E 12 0 1 1 1e04:100:10 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 PAGE 80 80 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e04:100:10 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:10 vs 0.01:5 0:10 0 0 0 0 0 0 0 0 0 0 1 1.39E 06 1e04:100:10 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:10 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:10 vs 0.1:50:10 0 0 0 0 3.79E 13 0 0 0 0 0 1 0.96688 1e04:100:10 vs 0.5:100:1 0 0 0 0 0 4.42E 12 3.37E 11 0 0 0 4.44E 13 0 1e04:100:10 vs 0.5:50:1 0 0 0 0 0 0.03243 0.02435 0 0 0 3.03E 13 0 1e04:100:10 vs 0.5:50:10 0 0 0 0 1 0 0 0 0 0 0.99904 0 1e04:100:10 vs 1e 04:100:1 0 9.42E 09 3.58E 13 3.55E 13 5.13E 13 8.55E 11 7.65E 11 1 0 0 0 0 1e04:100:10 vs 1 e 04:50:1 0 8.58E 05 8.45E 12 8.45E 12 5.53E 13 3.29E 11 2.99E 11 0.40647 0 0 0 0 1e04:100:10 vs 1e 04:50:10 0.00814 0.85967 0.99662 0.99662 0.99937 1 1 1 1 1 1 1 1e04:100:10 vs 1e 05:100:1 0 0 1 1 0.06328 0.00039 0.00036 0.99996 0 0 0 8.50E 09 1e04: 100:10 vs 1e 05:100:10 0 0 0.30629 0.30629 1.81E 10 6.05E 10 5.66E 10 0.14281 1 0.99641 1 5.70E 05 1e04:100:10 vs 1e 05:50:1 0 0 8.99E 05 8.99E 05 0.51958 1 1 1 0 0 0 0 1e04:100:10 vs 1e 05:50:10 0 0 1 1 0.84955 0.86653 0.87134 0.99998 1 0.99978 1 0.97 115 PAGE 81 81 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e04:100:10 vs 1e 06:100:1 0 0 1 1 0.00047 1.27E 06 1.16E 06 0.99999 0 0 0 0.00029 1e04:100:10 vs 1e 06:100:10 0 0 0.02219 0.02219 4.78E 13 1.76E 12 1.61E 12 0.94598 0.99789 0.53575 1 4.22E 08 1e04:100:10 vs 1e 06:50:1 0 0 1.23E 05 1.23E 05 0.448 41 1 1 1 0 0 0 0 1e04:100:10 vs 1e 06:50:10 0 0 1 1 0.99298 0.99566 0.99575 1 1 1 1 0.95923 1e04:100:30 vs 0.001:100:1 4.34E 13 0 0 0 0 0 0 0 0 0 0 0 1e04:100:30 vs 0.001:100:10 0 0 0 0 0 0 0 0 0 0 1 0 1e04:100:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:30 vs 0.001:50:10 0 0 0 0 0 5.20E 13 5.71E 13 2.34E 11 0 0 1 0 1e04:100:30 vs 0.001:50:30 0 0 0.11659 0.11659 1 1 1 1 0.99997 0.61627 1 1 1e04:100:30 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:30 vs 0.01:100:10 0 0 0 0 0 0.00181 0. 00393 0 0 0 1 0 1e04:100:30 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:30 vs 0.01:50:10 0 0 0 0 0 0.03225 0.02374 0 0 0 1 0 1e04:100:30 vs 0.01:50:30 0 0 4.69E 13 4.69E 13 0.72896 0.99997 0.99997 0.03676 0.03662 3.10E 10 1 0.99999 1e04:100:30 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 PAGE 82 82 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e04:100:30 vs 0.1:100:10 0 0 0 0 0 1.89E 06 1.95E 06 0 4.74E 13 0 1 0 1e04:100:30 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:100:30 vs 0.1:50:10 0 0 0 0 0 0 0 0 4.69E 13 0 1 0 1e04:100:30 vs 0.1:50:30 0 0 4.66E 13 4.66E 13 0.99998 0.4894 0.50214 8.44E 13 0.00107 3.39E 13 1 1 1e04:100:30 vs 0.5:100:1 0 0 0 0 0 0 0 0 0 0 1.59E 13 0 1e04:100:30 vs 0.5:100:10 0 0 0 0 0 5.92E 06 7.53E 06 0 1 0.98782 1 0 1e04:100:30 vs 0.5:50:1 0 0 0 0 0 0.00369 0.00506 0 0 0 0 0 1 e 04:100:30 vs 0.5:50:10 0 0 0 0 3.71E 13 5.36E 13 2.69E 13 0 1 0.99966 1 0 1e04:100:30 vs 0.5:50:30 0 0 1 1 1 1 1 1 1.21E 05 4.22E 13 1 0.00174 1e04:100:30 vs 1e 04:100:1 0 0 0 0 0 0 0 8.33E 06 0 0 0 0 1e04:100:30 vs 1e 04:100:10 0 6.26E 13 4.80E 13 4.78E 13 3.74E 13 3.81E 13 3.75E 13 0.00666 0 0 0.99936 0 1e04:100:30 vs 1e 04:50:1 0 0 0 0 0 0 0 2.87E 11 0 0 0 0 1e04:100:30 vs 1e 04:50:10 0 1.36E 05 3.67E 13 3.67E 13 0 4.14E 13 4.01E 13 0.00412 0 0 0.99893 0 1e04:100:30 vs 1e 04:50:30 0 2.24E 1 2 1 1 1 1 1 1 1 1 1 0.99868 1e04:100:30 vs 1e 05:100:1 0 0 3.78E 13 3.78E 13 3.96E 09 0.15347 0.15885 0.72285 0 0 0 0 PAGE 83 83 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e04:100:30 vs 1e 05:100:10 0 0 0.00029 0.00029 0.22637 1 1 1 0 0 0.99798 0 1e04:100:30 vs 1e 05:100:30 0 0 1 1 0.99996 0.99985 0.99 985 1 1 1 1 1 1e04:100:30 vs 1e 05:50:1 0 0 0 0 0 3.98E 13 4.00E 13 0.08956 0 0 0 0 1e04:100:30 vs 1e 05:50:10 0 0 6.08E 11 6.08E 11 8.06E 13 1.53E 06 1.41E 06 0.68097 0 0 0.99891 0 1e04:100:30 vs 1e 05:50:30 0.72414 1 1 1 1 1 1 1 1 1 1 0.98564 1e0 4:100:30 vs 1e 06:100:1 0 0 3.94E 13 3.90E 13 6.71E 06 0.8825 0.8875 0.66867 0 0 0 0 1e04:100:30 vs 1e 06:100:10 0 0 0.01385 0.01385 0.94522 1 1 0.99547 0 0 0.99669 0 1e04:100:30 vs 1e 06:100:30 0 0 0.85816 0.85816 0.02546 0.18308 0.18193 1 1 1 1 1 1e04:100:30 vs 1e 06:50:1 0 0 0 0 0 4.25E 13 4.30E 13 3.70E 05 0 0 0 0 1e04:100:30 vs 1e 06:50:10 0 0 1.08E 09 1.08E 09 3.56E 13 4.77E 08 4.55E 08 0.01494 0 0 0.99832 0 1e04:100:30 vs 1e 06:50:30 4.00E 05 0.30463 1 1 1 1 1 1 1 1 1 1 1e04:50:1 vs 1e 05:50:1 0 0 0.92943 0.92943 1.30E 06 2.83E 07 2.53E 07 0.06203 1 1 0.93972 4.82E 13 1e04:50:1 vs 1e 06:50:1 0 0 0.99352 0.99352 2.16E 06 1.98E 07 1.72E 07 0.99134 1 1 1 4.05E 13 1e04:50:10 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:50:10 vs 0.00 1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 PAGE 84 84 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e04:50:10 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:50:10 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:50:10 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:50:10 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:50:10 vs 0.5:100:1 0 0 0 0 0 4.02E 08 2.53E 07 0 0 0 4.64E 13 0 1e04:50:10 vs 0.5:50:1 0 0 0 0 0 0.00011 6.46E 05 0 0 0 3.18E 13 0 1e04:50:10 vs 1e 04:100:1 0 4.04E 13 2.40E 08 2.40E 08 2.76E 10 5.04E 07 5.03E 07 1 0 0 0 0 1e04:50:10 vs 1e 04:50:1 0 2.89E 12 8.08E 06 8.08E 06 4.43E 10 2.26E 07 2.29E 07 0.49821 0 0 0 0 1e04:50:10 vs 1e 05:100:1 0 0 1 1 1.07E 05 2.92E 07 2.38E 07 0.99982 0 0 0 4.10E 12 1e04:50:10 vs 1e 05:50:1 0 0 0.30049 0.30049 1 1 1 1 0 0 0 0 1e04:50:10 vs 1e 05:50:10 0 0 0.6292 0.6292 0.00592 0 .06729 0.06607 0.99991 1 1 1 1 1e04:50:10 vs 1e 06:100:1 0 0 0.99944 0.99944 6.92E 09 2.58E 10 2.08E 10 0.99993 0 0 0 7.49E 07 1e04:50:10 vs 1e 06:50:1 0 0 0.1151 0.1151 1 1 1 1 0 0 0 0 1e04:50:10 vs 1e 06:50:10 0 0 0.30985 0.30985 0.04563 0.3127 0.3 0421 1 1 1 1 1 1e04:50:30 vs 0.001:100:1 0.025 0 0 0 0 0 0 0 0 0 0 0 PAGE 85 85 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e04:50:30 vs 0.001:100:10 0 0 0 0 0 0 0 0 0 0 1 0 1e04:50:30 vs 0.001:50:1 0.99999 0 0 0 0 0 0 0 0 0 0 0 1e04:50:30 vs 0.001:50:10 0 0 0 0 0 1.36E 09 1.88E 09 6.16E 09 0 0 1 0 1 e 04:50:30 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:50:30 vs 0.01:100:10 0 0 0 0 0 0.20373 0.31006 0 0 0 1 0 1e04:50:30 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:50:30 vs 0.01:50:10 0 0 0 0 0 0.0001 6.46E 05 0 0 0 1 0 1e04:50:30 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:50:30 vs 0.1:100:10 0 0 0 0 0 3.97E 10 4.09E 10 0 3.10E 13 0 1 0 1e04:50:30 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e04:50:30 vs 0.1:50:10 0 0 0 0 0 0 0 0 5.26E 13 0 1 0 1e04:50:30 vs 0.5:100:1 0 0 0 0 0 0 0 0 0 0 1.59E 13 0 1e04:50:30 vs 0.5:100:10 0 0 0 0 0 1.58E 09 2.10E 09 0 1 0.96629 1 0 1e04:50:30 vs 0.5:50:1 0 0 0 0 0 0.29948 0.35173 0 0 0 0 0 1e04:50:30 vs 0.5:50:10 0 0 0 0 1.27E 12 3.98E 13 4.41E 13 0 1 0.9982 1 0 PAGE 86 86 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e04:50:30 vs 1e 04:100:1 0 0 0 0 0 0 0 0.00051 0 0 0 0 1e04:50:30 vs 1e 04:100:10 0 0 2.82E 11 2.82E 11 4.08E 13 1.93E 10 1.88E 10 0.11189 0 0 0.9994 0 1e04:50:30 vs 1e 04:50:1 0 0 0 0 0 0 0 7.41E 09 0 0 0 0 1e04:50:30 vs 1e 04:50:10 0 0 3.22E 13 3.22E 13 1.03E 13 3.83E 13 3.75E 13 0.07903 0 0 0 .999 0 1e04:50:30 vs 1e 05:100:1 0 0 3.65E 13 3.65E 13 2.96E 05 0.96647 0.96912 0.99587 0 0 0 0 1e04:50:30 vs 1e 05:100:10 0 0 0.0115 0.0115 0.9965 1 1 1 0 0 0.9981 0 1e04:50:30 vs 1e 05:50:1 0 0 0 0 0 3.67E 13 3.76E 13 0.56106 0 0 0 0 1e04:50:30 v s 1e 05:50:10 0 0 1.85E 08 1.85E 08 1.28E 08 0.00147 0.00138 0.99347 0 0 0.99898 0 1e04:50:30 vs 1e 05:50:30 0 4.08E 13 1 1 1 1 1 1 1 1 1 1 1e04:50:30 vs 1e 06:100:1 0 0 7.62E 12 7.63E 12 0.00866 1 1 0.9926 0 0 0 0 1e04:50:30 vs 1e 06:100:10 0 0 0.20 38 0.2038 1 1 1 1 0 0 0.99687 0 1e04:50:30 vs 1e 06:50:1 0 0 0 0 0 3.91E 13 3.89E 13 0.00182 0 0 0 0 1e04:50:30 vs 1e 06:50:10 0 0 2.42E 07 2.42E 07 3.23E 10 9.51E 05 9.20E 05 0.1958 0 0 0.99842 0 1e04:50:30 vs 1e 06:50:30 0 8.39E 14 0.99996 0.99996 0.98047 0.99985 0.99987 1 1 1 1 0.98769 1e05:100:1 vs 0.001:50:1 0 0 0 0 0 0 0 0 3.31E 13 0.02165 0 0 PAGE 87 87 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e05:100:1 vs 0.01:50:1 0 0 0 0 0 0 0 0 1.33E 11 0 0 0 1e05:100:1 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:1 vs 0.5:50:1 0 0 0 0 0 1 1 0 0 0 0 0 1e05:100:1 vs 1e 04:50:1 0 0 2.03E 08 2.03E 08 0 0 0 0.00121 1 1 0.99827 0 1e05:100:1 vs 1e 05:50:1 0 0 0.01408 0.01408 5.53E 09 2.33E 07 2.16E 07 1 0.99704 1 0.00932 3.86E 13 1e05:100:1 vs 1e 06:100:1 0.11212 0.12746 1 1 1 1 1 1 1 1 0.79573 0.999 87 1e05:100:1 vs 1e 06:50:1 0 0 0.00305 0.00305 3.08E 09 3.33E 07 3.16E 07 0.79947 1 1 1 3.54E 13 1e05:100:10 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:10 vs 0.001:50:10 0 0 0 0 0 4.09E 09 5.4 1E 09 1.09E 08 3.82E 13 0 1 1.51E 05 1e05:100:10 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:10 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:10 vs 0.01:50:10 0 0 0 0 0 4.33E 05 2.82E 05 0 0 0 1 0 1e05:100:10 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:10 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 PAGE 88 88 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e05:100:10 vs 0.1:50:10 0 0 0 0 0 0 0 0 0 0 0.99997 1.24E 11 1e05:100:10 vs 0.5:100:1 0 0 0 0 0 0 0 0 0 0 4.97E 13 0 1e05:100:10 vs 0.5:50:1 0 0 0 0 0 0.41692 0.4727 0 0 0 3.34E 13 0 1e05:100:10 v s 0.5:50:10 0 0 0 0 1.35E 06 3.67E 13 4.14E 13 0 0 0 0.99714 0 1e05:100:10 vs 1e 04:100:1 0 0 0 0 0 0 0 0.00077 0 0 0 0 1e05:100:10 vs 1e 04:50:1 0 0 2.52E 13 2.52E 13 0 0 0 1.30E 08 0 0 0 0 1e05:100:10 vs 1e 04:50:10 0 0 9.41E 05 9.41E 05 4.89E 13 4 .50E 13 4.37E 13 0.10258 1 1 1 0.00924 1e05:100:10 vs 1e 05:100:1 0 1.07E 05 0.00889 0.00889 0.18188 0.98859 0.98928 0.99812 0 0 0 0 1e05:100:10 vs 1e 05:50:1 1 4.10E 13 3.76E 13 3.74E 13 3.81E 13 4.27E 13 4.31E 13 0.63154 0 0 0 0 1e05:100:10 vs 1e 0 5:50:10 0 0 0.94145 0.94145 0.00163 0.00302 0.00278 0.99687 1 1 1 0.44653 1e05:100:10 vs 1e 06:100:1 0 4.30E 13 0.20058 0.20058 0.95537 1 1 0.9964 0 0 0 3.83E 13 1e05:100:10 vs 1e 06:100:10 4.34E 13 2.91E 09 1 1 1 1 1 1 1 1 1 1 1e05:100:10 vs 1e 06:5 0:1 1 5.48E 13 2.86E 13 2.86E 13 4.07E 13 4.59E 13 4.54E 13 0.00266 0 0 0 0 1e05:100:10 vs 1e 06:50:10 0 0 0.99645 0.99645 0.00012 0.00022 0.0002 0.24198 1 1 1 0.58482 1e05:100:30 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 PAGE 89 89 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e05:100:30 vs 0.001:100:10 0 0 0 0 0 0 0 0 0 0 1 0 1e05:100:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:30 vs 0.001:50:10 0 0 0 0 0 4.79E 13 3.25E 13 1.23E 11 0 0 1 0 1e05:100:30 vs 0.001:50:30 0 0 0.01646 0.01646 0.93901 1 1 1 0.99922 0.32013 1 1 1e05:100:30 vs 0.01:100: 1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:30 vs 0.01:100:10 0 0 0 0 0 9.67E 08 2.93E 07 0 0 0 1 0 1e05:100:30 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:30 vs 0.01:50:10 0 0 0 0 0 0.96787 0.94761 0 0 0 1 0 1e05:100:30 vs 0.01:50:30 0 0 3.94E 13 3.94E 1 3 0.00688 1 1 0.0497 0.01374 2.08E 11 1 1 1e05:100:30 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:30 vs 0.1:100:10 0 0 0 0 0 0.01408 0.01427 0 3.58E 13 0 1 0 1e05:100:30 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:100:30 vs 0.1:50:10 0 0 0 0 0 0 0 0 2.80E 13 0 1 0 1e05:100:30 vs 0.1:50:30 0 0 3.88E 13 3.88E 13 0.3424 1 1 1.21E 12 0.0003 5.31E 13 1 1 1e05:100:30 vs 0.5:100:1 0 0 0 0 0 0 0 0 0 0 1.68E 13 0 PAGE 90 90 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e05:100:30 vs 0.5:100:10 0 0 0 0 0 0.02951 0.03416 0 1 0.89639 1 0 1e05:100:30 vs 0.5:50 :1 0 0 0 0 0 2.65E 07 4.21E 07 0 0 0 0 0 1e05:100:30 vs 0.5:50:10 0 0 0 0 1.87E 13 1.70E 10 2.47E 10 0 1 0.98812 1 0 1e05:100:30 vs 0.5:50:30 0 0 1 1 0.99891 0.99944 0.99945 1 2.68E 06 4.23E 13 1 2.13E 05 1e05:100:30 vs 1e 04:100:1 0.05123 0 0 0 0 0 0 5.07E 06 0 0 0 0 1e05:100:30 vs 1e 04:100:10 0 0 2.88E 13 2.97E 13 0 4.17E 13 4.10E 13 0.00462 0 0 0.99945 0 1e05:100:30 vs 1e 04:50:1 0.52556 0 0 0 0 0 0 1.50E 11 0 0 0 0 1e05:100:30 vs 1e 04:50:10 0 0 3.96E 13 3.96E 13 0 0 0 0.00282 0 0 0.99908 0 1e05:100:30 vs 1e 04:50:30 0 0 1 1 0.50515 0.55099 0.5515 1 1 1 1 1 1e05:100:30 vs 1e 05:100:1 0 0 3.32E 13 3.32E 13 3.93E 13 9.71E 05 0.0001 0.65401 0 0 0 0 1e05:100:30 vs 1e 05:100:10 0 0 1.42E 05 1.42E 05 0.00034 0.41926 0.42395 1 0 0 0.99824 0 1e05:100:30 vs 1e 05:50:1 0 0.01392 0 0 0 0 0 0.06795 0 0 0 0 1e05:100:30 vs 1e 05:50:10 1 6.54E 10 1.31E 12 1.31E 12 4.46E 13 8.19E 12 7.46E 12 0.60972 0 0 0.99906 0 1e05:100:30 vs 1e 05:50:30 0 0 1 1 0.99594 0.99367 0.99357 1 1 1 1 1 1e05:100:30 v s 1e 06:100:1 0 0 5.27E 13 5.27E 13 1.12E 10 0.01227 0.01286 0.59692 0 0 0 0 PAGE 91 91 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e05:100:30 vs 1e 06:100:10 0 0 0.00119 0.00119 0.03301 0.95884 0.96165 0.99073 0 0 0.99709 0 1e05:100:30 vs 1e 06:100:30 0.00028 0.00599 0.99558 0.99558 0.92152 0.99978 0.999 77 1 1 1 1 0.99681 1e05:100:30 vs 1e 06:50:1 0 0.00042 0 0 0 0 0 2.31E 05 0 0 0 0 1e05:100:30 vs 1e 06:50:10 0.08667 5.69E 05 1.71E 11 1.71E 11 4.24E 13 5.56E 13 5.52E 13 0.0106 0 0 0.99855 0 1e05:100:30 vs 1e 06:50:30 0 0 1 1 1 1 1 1 1 1 1 1 1e05: 50:1 vs 1e 06:50:1 0.99999 1 1 1 1 1 1 0.99985 0.99933 1 0.53299 1 1e05:50:10 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:10 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:10 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:10 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:10 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:10 vs 0.5:100:1 0 0 0 0 0 3.52E 13 4.16E 13 0 0 0 4.71E 13 0 1e05:50:10 vs 0.5:50:1 0 0 0 0 0 1 0.99999 0 0 0 3.19E 13 0 1e05:50:10 vs 1e 04:100:1 0.40667 2.35E 07 3.37E 13 3.37E 13 0 3.20E 13 3.25E 13 0.63561 0 0 0 0 PAGE 92 92 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e05:50:10 vs 1e 04:50:1 0.97175 6.11E 12 3.61E 13 3.67E 13 0 4.09E 13 4.10E 13 0.00155 0 0 0 0 1e05:50:10 vs 1e 05:100:1 0 0 0.99795 0.99795 1 0.92261 0.91221 1 0 0 0 5.41E 13 1e05:50:10 vs 1e 05:50:1 0 0 4.76E 07 4.76E 07 1.49E 05 0.05971 0.06267 1 0 0 0 0 1e05:50:10 vs 1e 06:100:1 0 0 1 1 0.94598 0.19719 0.18483 1 0 0 0 1.53E 10 1e05:50:10 vs 1e 06:50:1 0 0 4.68E 08 4.68E 08 9.20E 06 0.07208 0.07647 0.8 333 0 0 0 0 1e05:50:10 vs 1e 06:50:10 0.5389 0.99974 1 1 1 1 1 1 1 1 1 1 1e05:50:30 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:30 vs 0.001:100:10 0 0 0 0 0 0 0 0 0 0 1 0 1e05:50:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:30 vs 0.001:50: 10 0 0 0 0 0 1.82E 12 2.41E 12 1.36E 10 0 0 1 0 1e05:50:30 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:30 vs 0.01:100:10 0 0 0 0 0 0.00837 0.01697 0 0 0 1 0 1e05:50:30 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:30 vs 0.01:50:10 0 0 0 0 0 0.0082 5 0.00574 0 0 0 1 0 1e05:50:30 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 PAGE 93 93 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e05:50:30 vs 0.1:100:10 0 0 0 0 0 2.20E 07 2.22E 07 0 3.80E 13 0 1 0 1e05:50:30 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e05:50:30 vs 0.1:50:10 0 0 0 0 0 0 0 0 3.01E 13 0 1 0 1e05:50:30 vs 0.5:100:1 0 0 0 0 0 0 0 0 0 0 1.68E 13 0 1e05:50:30 vs 0.5:100:10 0 0 0 0 0 7.34E 07 9.28E 07 0 1 0.81334 1 0 1e05:50:30 vs 0.5:50:1 0 0 0 0 0 0.01582 0.02122 0 0 0 0 0 1e05:50:30 vs 0.5:50:10 0 0 0 0 5.51E 13 3.77E 13 3.96E 13 0 1 0.96695 1 0 1e05:50:30 vs 1e 04:100:1 0 0 0 0 0 0 0 3.14E 05 0 0 0 0 1e05:50:30 vs 1e 04:100:10 0 7.38E 09 1.24E 12 1.24E 12 3.77E 13 6.14E 13 6.05E 13 0.01736 0 0 0.99946 0 1e05:50:30 vs 1e 04:50:1 0 0 0 0 0 0 0 1.65E 10 0 0 0 0 1e05:50:30 vs 1e 04:50:10 0 0.01434 4.70E 13 4.70E 13 0 3.56E 13 3.53E 13 0.01114 0 0 0.99909 0 1e05:50:30 vs 1e 05:100:1 0 0 5.21E 13 5.21E 13 5.73E 08 0.36189 0.37389 0.87678 0 0 0 0 1e05:50:30 vs 1e 05:100:10 0 0 0.00147 0.00147 0.51883 1 1 1 0 0 0.99826 0 1e05:50:30 vs 1e 0 5:50:1 0 0 0 0 0 3.42E 13 3.47E 13 0.17883 0 0 0 0 1e05:50:30 vs 1e 05:50:10 0 0 6.98E 10 6.98E 10 7.26E 12 1.18E 05 1.11E 05 0.84802 0 0 0.99907 0 PAGE 94 94 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e05:50:30 vs 1e 06:100:1 0 0 6.51E 13 6.48E 13 6.14E 05 0.9831 0.98468 0.83916 0 0 0 0 1e05:50:30 vs 1e 06:100:10 0 0 0.04813 0.04813 0.9969 1 1 0.99961 0 0 0.99712 0 1e05:50:30 vs 1e 06:50:1 0 0 0 0 0 3.64E 13 3.63E 13 0.00013 0 0 0 0 1e05:50:30 vs 1e 06:50:10 0 0 1.09E 08 1.09E 08 5.69E 13 4.45E 07 4.35E 07 0.0364 0 0 0.99856 0 1e05:50:30 vs 1e 06 :50:30 0.80769 0.999 1 1 1 1 1 1 1 1 1 0.93329 1e06:100:1 vs 0.001:50:1 0 0 0 0 0 0 0 0 2.02E 11 0.16899 4.46E 13 0 1e06:100:1 vs 0.01:50:1 0 0 0 0 0 0 0 0 4.28E 08 0 0 0 1e06:100:1 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:1 vs 0.5:50:1 0 0 0 0 0 0.9964 0.99801 0 0 0 0 0 1e06:100:1 vs 1e 04:50:1 0 0 3.12E 11 3.12E 11 0 0 0 0.00167 1 1 1 0 1e06:100:1 vs 1e 05:50:1 0 0 0.00022 0.00022 1.48E 12 1.99E 10 1.85E 10 1 1 1 0.99995 1.49E 13 1e06:100:1 vs 1e 06:50:1 0 0 3.25E 05 3.25E 05 9.88E 13 3. 01E 10 2.88E 10 0.84236 1 1 1 3.56E 13 1e06:100:10 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:10 vs 0.001:50:10 0 0 0 0 0 1.03E 11 1.33E 11 2.09E 05 4.81E 13 0 1 8.35E 09 PAGE 95 95 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e06:100:10 vs 0.01:10 0:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:10 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:10 vs 0.01:50:10 0 0 0 0 0 0.00261 0.00188 0 0 0 1 0 1e06:100:10 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:10 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:10 vs 0.1:50:10 0 0 0 0 0 0 0 0 0 0 0.99994 3.07E 13 1e06:100:10 vs 0.5:100:1 0 0 0 0 0 0 0 0 0 0 5.27E 13 0 1e06:100:10 vs 0.5:50:1 0 0 0 0 0 0.04261 0.05281 0 0 0 3.54E 13 0 1e06:100:10 vs 0.5:50:10 0 0 0 0 1.38E 09 5.13E 13 3.11E 13 0 0 0 0.99541 0 1e06:100:10 vs 1e 04:100:1 0 0 0 0 0 0 0 0.10309 0 0 0 0 1e06:100:10 vs 1e 04:50:1 0 0 0 0 0 0 0 2.42E 05 0 0 0 0 1e06:100:10 vs 1e 04:50:10 0 0 7.10E 07 7.10E 07 3.56E 13 4.62E 13 4.48E 13 0.9059 1 0.99607 1 2.69E 05 1e06:100:10 vs 1e 05:100:1 0. 00039 1 0.00017 0.00017 0.00422 0.58043 0.58203 1 0 0 0 0 1e06:100:10 vs 1e 05:50:1 3.41E 13 0 3.74E 13 3.73E 13 0 4.50E 13 4.44E 13 0.99991 0 0 0 0 1e06:100:10 vs 1e 05:50:10 0 0 0.34659 0.34659 6.81E 06 5.16E 05 4.51E 05 1 1 1 1 0.0116 PAGE 96 96 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e06:100:10 vs 1e 06:100:1 3.49E 13 0.95548 0.01122 0.01122 0.24816 0.99824 0.99824 1 0 0 0 0 1e06:100:10 vs 1e 06:50:1 8.29E 10 0 4.40E 13 4.40E 13 0 4.72E 13 4.67E 13 0.21805 0 0 0 0 1e06:100:10 vs 1e 06:50:10 0 0 0.67176 0.67176 2.85E 07 2.25E 06 2.02E 06 0.984 34 1 0.99927 1 0.02407 1e06:100:30 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:30 vs 0.001:100:10 0 0 0 0 0 0 0 0 0 0 1 0 1e06:100:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:30 vs 0.001:50:10 0 0 0 0 0 0 0 6.69E 13 0 0 1 0 1e06:100:30 vs 0.001:50:30 0 0 3.66E 07 3.66E 07 0.00088 0.62804 0.6396 1 0.98536 0.05544 1 1 1e06:100:30 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:30 vs 0.01:100:10 0 0 0 0 0 6.73E 13 1.29E 12 0 0 0 1 0 1e06:100:30 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:30 vs 0.01:50:10 0 0 0 0 0 1 1 0 0 0 1 0 1e06:100:30 vs 0.01:50:30 0 0 0 0 5.35E 09 0.99915 0.99917 0.23782 0.00356 6.30E 13 1 0.80355 1e06:100:30 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:30 vs 0.1:100:10 0 0 0 0 0 0.91114 0.91415 0 4.93E 13 0 1 0 PAGE 97 97 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e06:100:30 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:100:30 vs 0.1:50:10 0 0 0 0 0 0 0 0 3.57E 13 0 1 0 1e06:100:30 vs 0.1:50:30 0 0 0 0 1.00E 05 1 1 4.01E 11 5.67E 05 4.91E 13 1 0.99717 1e06:100:30 vs 0.5:100:1 0 0 0 0 0 0 0 0 0 0 1.77E 13 0 1e06:100:30 vs 0.5:100:10 0 0 0 0 0 0.96871 0.9764 0 0.99979 0.4374 1 0 1e06:100:30 vs 0.5:50:1 0 0 0 0 0 1.22E 12 1.74E 12 0 0 0 0 0 1e06:100:30 vs 0.5:50:10 0 0 0 0 0 1.92E 05 2.61E 05 0 0.99999 0.74243 1 0 1e06:100:30 vs 0.5:50:30 0 0 0.466 02 0.46602 0.00901 0.13642 0.13553 1 3.78E 07 6.53E 14 1 0.15031 1e06:100:30 vs 1e 04:100:1 1 0 0 0 0 0 0 2.09E 07 0 0 0 0 1e06:100:30 vs 1e 04:100:10 0 0 3.64E 13 3.64E 13 0 0 0 0.0004 0 0 0.99951 0 1e06:100:30 vs 1e 04:50:1 0.9937 0 0 0 0 0 0 7.32E 13 0 0 0 0 1e06:100:30 vs 1e 04:50:10 0 0 0 0 0 0 0 0.00023 0 0 0.99917 0 1e06:100:30 vs 1e 04:50:30 0 0 0.2476 0.2476 3.14E 05 0.00151 0.00148 1 1 1 1 0.48351 1e06:100:30 vs 1e 05:100:1 0 0 0 0 2.89E 13 1.46E 09 1.56E 09 0.2452 0 0 0 0 1e06:100:30 vs 1e 05:100:10 0 3.07E 13 1.43E 11 1.43E 11 5.03E 11 0.00071 0.00072 1 0 0 0.99841 0 PAGE 98 98 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e06:100:30 vs 1e 05:50:1 0 1 0 0 0 0 0 0.0097 0 0 0 0 1e06:100:30 vs 1e 05:50:10 0.009 0 4.62E 13 4.62E 13 0 3.53E 13 3.39E 13 0.21477 0 0 0.99915 0 1e06:100:30 vs 1e 05:50:30 0 0 0.61424 0.61424 0.00514 0.06385 0.06262 1 1 1 1 0.27804 1e06:100:30 vs 1e 06:100:1 0 0 1.40E 13 1.40E 13 4.04E 13 1.29E 06 1.36E 06 0.20657 0 0 0 0 1e06:100:30 vs 1e 06:100:10 0 0 6.83E 09 6.83E 09 7.71E 08 0.0249 0.02567 0.82967 0 0 0.99735 0 1e06:100:30 vs 1e 06:50:1 0 1 0 0 0 0 0 1.09E 06 0 0 0 0 1e06:100:30 vs 1e 06:50:10 1 4.14E 13 5.06E 13 5.06E 13 0 4.76E 13 4.68E 13 0.00105 0 0 0.99868 0 1e06:100:30 vs 1e 06:50:30 0 0 0.99983 0.99983 0.31672 0.52445 0.5153 1 1 1 1 1 1e06:50:10 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:10 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:10 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:10 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:10 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:10 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:10 vs 0.5:100:1 0 0 0 0 0 4.01E 13 3.24E 13 0 0 0 4.84E 13 0 PAGE 99 99 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e06:50:10 vs 0.5:50:1 0 0 0 0 0 0.99698 0.99385 0 0 0 3.27E 13 0 1e06:50:10 vs 1e 04:100:1 1 1.06E 12 3.56E 13 3.56E 13 8.39E 14 3.63E 13 3.53E 13 1 0 0 0 0 1e06:50:10 vs 1e 04:50:1 1 3.35E 13 2.72E 13 2.72E 13 1.31E 13 3.14E 13 3.13E 13 0.26448 0 0 0 0 1e06:50:10 vs 1e 05:100:1 0 0 0.95731 0.95731 0.9998 0.54766 0.53086 1 0 0 0 2.79E 13 1e06:50:10 vs 1e 05:50:1 0 3.07E 13 3.78E 08 3. 78E 08 0.00025 0.28826 0.29341 1 0 0 0 0 1e06:50:10 vs 1e 06:100:1 0 0 1 1 0.62887 0.03494 0.03278 1 0 0 0 2.10E 10 1e06:50:10 vs 1e 06:50:1 0 3.47E 13 3.24E 09 3.24E 09 0.00016 0.32753 0.33578 1 0 0 0 0 1e06:50:30 vs 0.001:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:30 vs 0.001:100:10 0 0 0 0 0 0 0 0 0 0 1 0 1e06:50:30 vs 0.001:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:30 vs 0.001:50:10 0 0 0 0 0 3.18E 13 3.39E 13 5.60E 12 0 0 1 0 1e06:50:30 vs 0.01:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:30 vs 0.01:100 :10 0 0 0 0 0 0.00017 0.00042 0 0 0 1 0 1e06:50:30 vs 0.01:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:30 vs 0.01:50:10 0 0 0 0 0 0.15533 0.11939 0 0 0 1 0 PAGE 100 100 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e06:50:30 vs 0.1:100:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:30 vs 0.1:100:10 0 0 0 0 0 3.08E 05 3.02E 0 5 0 3.51E 13 0 1 0 1e06:50:30 vs 0.1:50:1 0 0 0 0 0 0 0 0 0 0 0 0 1e06:50:30 vs 0.1:50:10 0 0 0 0 0 0 0 0 2.86E 13 0 1 0 1e06:50:30 vs 0.5:100:1 0 0 0 0 0 0 0 0 0 0 1.68E 13 0 1e06:50:30 vs 0.5:100:10 0 0 0 0 0 8.74E 05 0.0001 0 1 0.89031 1 0 1e0 6:50:30 vs 0.5:50:1 0 0 0 0 0 0.00038 0.00056 0 0 0 0 0 1e06:50:30 vs 0.5:50:10 0 0 0 0 4.67E 13 3.77E 13 3.87E 13 0 1 0.9869 1 0 1e06:50:30 vs 1e 04:100:1 0 0 0 0 0 0 0 2.74E 06 0 0 0 0 1e06:50:30 vs 1e 04:100:10 0 0.00061 4.70E 13 4.70E 13 0 5.47E 13 5.50E 13 0.00291 0 0 0.99944 0 1e06:50:30 vs 1e 04:50:1 0 3.36E 13 0 0 0 0 0 6.85E 12 0 0 0 0 1e06:50:30 vs 1e 04:50:10 0 0.95528 8.39E 14 8.39E 14 0 4.20E 13 4.13E 13 0.00176 0 0 0.99907 0 1e06:50:30 vs 1e 05:100:1 0 0 4.45E 13 4.45E 13 9.24E 12 0.03175 0.03423 0.56567 0 0 0 0 1e06:50:30 vs 1e 05:100:10 0 0 2.47E 06 2.47E 06 0.01476 0.99902 0.99915 1 0 0 0.99822 0 1e06:50:30 vs 1e 05:50:1 0 0 0 0 0 3.92E 13 3.99E 13 0.0477 0 0 0 0 PAGE 101 101 Table 2 4. Continued. Parameter combination comparison (GF:TC:CS) Fst p value st p value S p value W p value p value R 2 p value TD p value NSS p value # Hap p value # Single p value Hmzy p value Tau hat p value 1e06:50:30 vs 1e 05:50:10 0 0 5.07E 13 5.10E 13 2.85E 13 7.33 E 08 7.06E 08 0.5206 0 0 0.99905 0 1e06:50:30 vs 1e 06:100:1 0 0 3.91E 13 3.91E 13 3.89E 08 0.52511 0.54102 0.50779 0 0 0 0 1e06:50:30 vs 1e 06:100:10 0 0 0.00027 0.00027 0.37211 1 1 0.97998 0 0 0.99706 0 1e06:50:30 vs 1e 06:50:1 0 0 0 0 0 4.41E 13 4 .51E 13 1.28E 05 0 0 0 0 1e06:50:30 vs 1e 06:50:10 0 0 2.19E 12 2.19E 12 3.59E 13 1.76E 09 1.78E 09 0.00687 0 0 0.99853 0 PAGE 102 102 T able 25. Percent of simulated scenarios that agree with empirical Fst estimates separated by CS and GF categories. Fst estimates for simulations were compared to empirical Fst estimates for Africa vs European (0.141) or Asian (0.235) populations (Bowcock et al. 1991). % overlap reflects simulated Fst estimates that fall between 0.141 and 0.235 for each GF x CS category. % Total overlap shows simulated Fst estimates that fall between 0.141 and 0.235 for each CS category for 42,000 simulations. The small values for % total overlap indicate that the majority of tested par ameter combinations do not fit the empirical data well. CS=1% CS= 10% CS=30% GF category % overlap GF category % overlap GF category % overlap 10 06 0 10 06 3.1 10 06 26.8 10 05 0 10 05 4.2 10 05 29.5 10 04 1.4 10 04 20.0 10 04 63.5 10 03 82.0 10 03 35.6 10 03 6.6 0.01 23.8 0.01 0.1 0.01 0 0.1 0 0.1 0 0.1 0 0. 5 0 0.5 0 0.5 0 % Total overlap 2.6 1.5 3.0 PAGE 103 103 Figure 21. Alternative scenarios for initial colonization of modern humans out of Africa. Scenarios include all combinations for time of colonization (TC) and colonization size (CS). Eac h scenario is modeled with seven values of bidirectional gene flow (where GF=106, 105, 104, 103, 102, 0.1, and 0.5 proportion of migrants per generation). PAGE 104 104 Figure 22. Box plots of estimates of 5 summary statistics that partition genetic variation more equally between CS, GF, and CSxGF than seen in the other summary statistics. P anels A) Fst. B) S. C D ) R2 E) NSS. Each panel shows the distribution of 42,000 estimates of the summary statistic calculated for each of 1,000 datasets simulated under 42 parameter combinations as listed at the bottom of the figure. The gray bar in panel A ) shows the range of empirical Fst values (0.1410.235) between African populations vs European and Asian populations ( Bowcocket al. 1991) PAGE 105 105 Figure 23. Box plots of es timates of 4 summary statistics that partition genetic variation primarily by CS, compared to the other summary statistics. A ) B ) #Hap. C ) #Single D ) Hmzy. Each box plot represents a demographic scenario described by the parameter values at the bot tom of the figure. A B C D PAGE 106 106 Figure 24. Box plots of estimates of 3 summary statistics that partition genetic variation similar to summ ary statistics in Figure 2 3. A ) st. B W. C ) TD. have similar box plot profiles to Fst, S, and R2, respectively Each box plot represents a demographic scenario described by the parameter values at the bottom of the figure. PAGE 107 107 CHAPTER 3 HUMAN MIGRATION PATTERNS IN YEMEN AND IMPLICATION S FOR RECONSTRUCTING PREHISTORIC POPULATION MOVEMENTS Humans facility for dispersal has played a large role in our evolutionary history, yet how and why humans have moved throughout history is unclear. Most data on human movement come from ethnographic st udies, comparisons of birthplaces from birth certificates, and census data. While ethnographic studies offer insight into social and environmental factors that influence human movement, they generally involve seasonal or temporary movements, as in the case of migrant workers ( de Haan and Rogaly 2002) or hunters and gatherers ( Hahnet al. 1966; Marlowe 2010) In order to understand how migration has influenced our evolutionary history, it is neces sary to address migration as the movement to a new location for permanent settlement. Birth certificate and census data allow us to trace movement across longer periods of time (i.e. between generations), but studies using these data generally focus either on the proportion of migrants or the distance moved, do not usually use multi generational families, and can typically only be studied in developed countries ( Bo attiniet al. 2007; Daviset al. 2013; Gray and Bilsborrow 2013; Levy 2010; Mielkeet al. 1994; Mortonet al. 1971) A deeper understanding of migration over multiple generations in a developing country offers the possibility of describing more general patterns of human migration and of identifying factors that may have influenced migration throughout human evolution. Since human migration has had the largest effect on genetic variation ( Mir Herrans and Mulligan 2013) a better understanding of human migration patterns would allow more accurate reconstructions of demographic processes. Comparisons of PAGE 108 108 empirical genetic data to s imulated genetic variation generated from models that realistically represent the demographic process under study offer the possibility of reconstructing prehistoric demographic processes ( Beaumontet al. 2002) Values for migration parameters estimated from human migration patterns, such as the proportion of the population that is moving, co uld define some model parameters to generate more realistic demographic scenarios. The ability to include empirically informed values to fix or set ranges on migration parameters increases the probability of identifying the best model to explain the data. Evaluating migration patterns in a developing country could provide migration estimates that are similar to prehistoric population movements. Yemen is a developing country ( Malik 2013) that has a heterogeneous landscape with coastal plains on the west and south, mountain ranges in the west and desert in the north, thus providing a fertile setting in which to investigate environmental factors that may hav e influenced prehist oric population movements. It also has a patrilocal and patrilineal society with a primarily shared language and religion ( Dresch 1989) which are social factors that could play a role in migration as well. The migration within a population of mostly agriculturalist s and pastoralist s could provide more realistic values of distance and proportion of migrati on for prehistoric movements since the advent of agriculture. The values could also provide informative lower limits for describing the migration of prehistoric hunter gatherers, who typically exhibit more movement than agriculturalists ( Hazelwood and Steele 2004) In this study, we use GPS coordinates from birthplaces and places of residence across four generations in Yemen to calculate the proportion, the distance and the PAGE 109 109 direction of migration between each generation. We test for differences in these values between the groups, we identify factors that influence migration patterns, and we di scuss possible effects of the migration patterns on genetic variation. Based on our results, we provide estimates for the proportion and distance of migration in a developing country, which can serve to define parameter values for demographic models agains t which to test genetic data to reconstruct prehistoric demographic processes. Our use of empirical data on population movements over four generations in Yemen provides knowledge that will allow for more accurate reconstruction of prehistoric processes of migration. Methods Samples and Data In 2007, saliva samples were collected throughout mainland Yemen for genetic analysis. Data were also collected from each study participant on place of residence, place of birth, parents place of birth and grandparents place of birth. Since all sampled individuals were adults, their current residence was assumed to be a proxy for the location of the next generation, i.e. their children, therefore providing data for residence patterns for a fourth generation in the study. For the purposes of this study, the individuals in each generation were considered independent samples. Location names for all birthplaces (and place of residence) were translated from Arabic and GPS coordinates were obtained using Geonames.org. In inst ances where a town name was not identifiable in the Geonames database, but the larger district could be identified, a GPS coordinate was obtained for the centroid of the district. Samples for which town or district locations could not be determined were removed. Ultimately, the resulting PAGE 110 110 dataset contained GPS coordinates for the sampled individuals place of residence and place of birth, mothers and fathers places of birth, and maternal grandmothers, maternal grandfathers, paternal grandmothers, and pa ternal grandfathers places of birth for 351 sampled individuals. Estimation of Migration The occurrence of migration was determined by the difference in birthplace or residence location between generations. A migration event occurred in the sampled indiv iduals generation (G1) if the place of residence was different from the birthplace. A migration in the parental generation (G2) occurred if the parents offspring was born in a different location than the parents birthplace (i.e. if the sampled individuals birthplace was different from their mothers or fathers birthplace). Similarly, a migration event in the grandparental generation (G3) occurred if the parents birthplace was different from the grandparents birthplace. Migration events were determined for eight different groups: female sampled individuals (G1fem), male sampled individuals (G1male), mothers (G2fem), fathers (G2male), maternal grandmothers (G3mfem), maternal grandfathers (G3mmale), paternal grandmothers (G3pfem), paternal grandfathers (G3pmale). The frequency of migration events was calculated for each of the eight groups (sample sizes were 70 in G1fem, 281 in G1male, and 351 in each group in G2 and in G3. The observed frequencies were compared through goodness of fit tests. The age o f the sampled individuals ranged from 13 to 69, which meant that each generation group (G1, G2, G3) essentially included two generation time periods. To account for the possibility of migration events occurring over different generation time periods withi n each generation group, the eight groups were further divided into two age PAGE 111 111 groups with a 25 year generation time between them, based on the ages of the sampled individuals (under and over 40 years old). Only 10% of the samples in any generation were in the over 40 years old subgroup, suggesting that any difference in migration event frequencies could be due instead, to the unbalanced sample size; thus no further analyses were performed with the groups partitioned by age over and under 40 years. Migration distance was calculated from the geographic distance between birthplaces/residences in two different generations using the GPS coordinates. G1 migration distances were calculated as the geographic distance between the sampled individuals birthplace and place of residence. G2 migration distances were calculated from the parents birthplace and the sampled individuals birthplace. Migration distances were calculated for G3 from the difference in grandparents birthplace and parents birthplace. The migration distances were compared between sex in each generation and between generations using Wilcoxon Rank tests and Kruskal Wallis analysis of variance. Different models including generation group, sex, birthplace location (latitude and longitude) and residence location were tested in logistic regressions to see which model (and parameters) best explained migration. AIC ( Akaike information criterion) were used to select the best model. Additionally, the migration events were plotted geographically and the mean direction of the migrations was calculated for each collection site (to account for sampling) using ESRI ArcMap10. PAGE 112 112 Results The proportion of migrants was calculated from the frequency of migration events for females and males in three generations (G1fem= 0 .314, G1male= 0.267,G2fem= 0.376, G2male= 0.311, G3mfem= 0.120, G3mmale= 0.111, G3pfem= 0.097,G3pmale= 0.080). Within each generation, the proportion of migrants between male and female groups was not significantly different (Figure 31). However, more rec ent generations G1 and G2 had a significantly larger proportion of migrants than G3 (p=0.0005). The proportion of migrants for each generation group (males and females combined) was G1=0.276, G2=0.343, G3=0.102. We calculated a multi generation proportion of migrants for G3 to correct for back migration events by determining the number of migration events in which the grandparents birthplace was different than the residence location. This produced a multi generation proportion of migrants for G3 of 0.086. The distance of migration was also calculated for each of the eight groups. G1 and G2 migration distances were significantly larger than G3 (p<2.2x1016). Density plots combining the migration distance (including nonmigrants) and the frequency of these di stances revealed that G1fem not only had the largest migration distance, but had more migrations at longer distances (>250km), than the other groups (Figure 32). However, when compared by sex within generations, female distances were not significantly dif ferent from male distances. Summary statistics on migration distances were calculated on all the individuals and on only migrating individuals (Table 31). Correlation analyses were performed on marital pairs in G2 and G3 to determine whether marital pair s were moving together and should be considered each a single group (instead of female and male groups) in further analyses. A low correlation PAGE 113 113 coefficient (<0.1) would suggest the marital pair migrations were completely independent from each other and a hi gh correlation coefficient (>0.9) would suggest that the marital pairs were moving together and could be treated as one group. G2 had a significant (p= 2.2x1016) Spearmans rho correlation coefficient of 0.589. Maternal grandparents (G3M) ha d a rho coeffi cient of 0.782 (p= 2.2x1016) and paternal grandparents (G3P) ha d a rho coefficient of 0.623 (p= 2.2x1016). The results showed there was a moderate and significant correlation between all the marital pairs. These coefficients suggest that a portion of th e marital pairs are moving together, but the correlations are not high enough to consider the marital pairs as a single group. Furthermore, the moderate correlation coefficients suggest these values could be due to post marital residence dynamics (i.e. fem ales moving with their husbands). Female and male marital pair distances were plotted and showed that correlated migrations were of the same distance, which is consistent with marital pairs moving to the same place (Figure 3 3). These results suggest that many of the marital pairs are moving together. Out of the 121 migration events in G3, 56% are of marital pairs moving together. Logistic regression models including different combinations of generation, sex, birthplace coordinates and residence location coordinates were performed to explain presence or absence of migration. The model with the lowest AIC included generation, sex, birth latitude and longitude and residence latitude (Table 32). This best model demonstrated that the probability of migration decreased in G3, decreased in males (consistent with females moving with their husbands family) and decreased with a more eastern birthplace, in comparison to G1fem. The probability of migration increased in G2 and increased with more northern birthplac es and places of residence. However, of PAGE 114 114 these factors, only G3 had a coefficient above one, suggesting that G3 contributes the most to the probability of migration, and specifically, belonging to the G3 generation decreases the probability of migration. Wh ile birthplace latitude, birth place longitude and residence location latitude had small coefficients, their statistically significant contribution to migration probability suggests that there could be factors pushing and pulling individuals to move ( Lee 1966) The birthplace and residence coordinates were used to plot the directionality of migration and assess whether or not there was a pattern in directionality that could explain the pushing and pulling effects (Figure 34). The mean migration direction was calculated from these migration vectors for each sample collection site (to account for the effect of sampling). While the mean migration di rections seem to have a southbound tendency, the circular variance (which describes the variation associated with the directional mean, where values close to 0 represent a similar direction for all migration vectors and values close to 1 correspond to vect ors in all compass directions) was moderate to high for all collection sites, ranging from 0.675 to 0.867 (Table 33), suggesting movement in all directions. The mean migration directions were further calculated by collection site for each generation group (Figure 3 5 and Table 34). Within generation groups G2 and G3, female and male migration directions were similar in many collection sites, supporting the idea that marital pairs moved together. The mean migration lengths were generally larger for G1 and G2 than for G3, reflecting the decreased migration distance in G3. For each collection site, the mean migration directions varied greatly between generation groups, suggesting a level of stochasticity to the migration directions. When the mean PAGE 115 115 migration directions were spatially compared to geographic features (i.e. elevation, land use/land cover, and watershed), no pattern arose (data not shown), further supporting stochasticity in the directionality of migrations. Discussion Our study helps elucidate hu man migration patterns using empirical population movement data across multiple generations in Yemen. Our results show that the proportion and distance of migration increased in recent generations. While movement in the recent generations may reflect social and political changes that have occurred in the last 50 years ( Federal Research Division 2008 ) the reduced movement i n the oldest generation most likely reflects a lack of technology and associated mobility ( Lee 1966) suggesting that this generation may be most representative of prehistoric movements. The correlated distance and directionality of migrations within marital pairs illustrate the prevalence of post marital residence dynamics. The significance of birthplace and residence locations in the probability of migration, but lack of pattern in the direction of migration, suggest a degree of stochasticity in terms of human movements. These cultural factors affecting modern movement have most likely played important roles in prehistoric migrations as well, suggesting that the migration patterns and estimates described in our results provide information to make more accurate prehist oric inferences. Patrilocality and Genetic Signals Moderate correlation coefficients for G2 and G3 marital pairs and the plot of migration distances in marital pairs suggests that pairs are moving together and the correlation seems to strengthen with incr easing distance (Figure 34). Our best fit PAGE 116 116 model, which shows that females are more likely to move than males when accounting for other contributing variables, suggests that patrilocality (females moving to their husbands family) may be driving the moveme nt. This is supported by ethnographic accounts that ~90% of the Yemeni population is patrilocal ( Weir 2007 ) However, the coefficient of the effect that being male has on the probability of migration is low ( 0.240) and within each generation the migration distance is not significantly different between females and males. This suggests that males ar e only slightly less likely to migrate than females and that males are travelling similar distances compared to females. In a perfect patrilocal post marital residence dynamic, males move short distances and stay close to their family, while females move l onger distances to be near their husbands family. The similar migration distances between females and males suggest there is not strict patrilocality in Yemen and that other factors are influencing male movement. This interpretation is supported by ethnog raphic data showing that males may occasionally migrate large distances from their birthplace for socioeconomic or political reasons ( Dresch 1989; Weir 2007 ) Our data show that male migration has occurred more often in the last 50 years (as shown by the increase in dispersal in G 1 and G2 relative to G3). The similar migration distances between females and males and consequent imperfect patrilocality may be the principal contributor to the lack of association observed between geographic and genetic distance in male lineages (i.e. Y chromosome) in Yemen ( Raaumet al. 2013) Females moving with their husbands (Figure 3 3) may also explain why shared mitochondrial DNA (mtDNA) haplotypes have been found between east and west Yemen, over 750km apart ( Cernyet al. 2008) PAGE 117 117 Patterns of Migration Logistic regressions were used to test the effect of birthplace and residence locations on the probability of migration in order to asses s whether there were factors pushing or pulling, respectively, individuals to a new location. Birthplace latitude and longitude and residence latitude were significant parameters in explaining the probability of migration. Given this result, birthplace and residence coordinates were used to plot migration directions and determine whether a pattern could be observed that could account for the effects of birthplace and residence locations. Mean migration directions were calculated by collection site (to account for sampling bias) to summarize the overall migration direction patterns (Figures 34 and 35 and Table 3 3). While the mean migration directions had a southbound trend, the circular variances were large, suggesting overall dispersal in multiple dir ections. Additionally, mean migration directions calculated by collection site for each generation showed that the collection sites had different mean directions between generations, further supporting migration in multiple directions. We also spatially co mpared the migration directions with different geographic features (i.e. elevation, land use/land cover, and watershed) to identify environmental factors that may influence migration direction. We found no pattern associated with the migration directions and the geographic features. These results suggest that while there may be factors pushing and pulling individuals to move, the overall direction of migration has little or no pattern. These results contrast with island migration patterns (e.g. Polynes ia) where migration direction has a pattern from larger islands to smaller islands ( Clarket al. 2006; Joblinget al. 2004; Kirch 1980 ) Given that PAGE 118 118 continental migrations are less limited by the carrying capacity of new colonization sites than islands, this result is not surprising. While island migrations have be en well described by ethnographic and archaeological data ( Clarket al. 2006; Kirch 1980) continental migration patterns have been primarily addressed through genetic data. Genetic evidence has suggested that overall continental migrations have a linear pattern, such that incr easing distance from Africa is correlated with decreasing genetic diversity ( Liet al. 2008; Ramachandranet al. 2005) Our data suggest that the smaller scale migrations (Figures 3 4 and 35) that led to this continental pattern may have been less directed. Our results are consistent with the idea that smaller migrations, which consider the movement of individual s, tend to be more random, while larger scale movements focused on populations have more directionality associated with them ( Hazelwood and Steele 2004; Skellam 1951) Empirical Estimates of Migration Comparisons of proportion of migrants and migration distanc es across four generations showed that migration was significantly lower around fifty years ago (G3). Furthermore, the best fit model to explain the probability of migration shows that G3 has not only the biggest effect, but a negative effect on the probability of migration (i.e. belonging to G3 decreases the probability of a migration event). Spatial patterns of migration in G3 (Figure 35c) show, that while there are some long migration distances, on average, the distances are short. Yemens less develope d state and poor transportation infrastructure ( Federal Research Division 2008 ) combined with the significantly reduced migration in G3, suggests that our data from the G3 generation PAGE 119 119 can provide empirically based estimates of migration frequency and dist ance that are reflective of prehistoric movements. We calculated the mean and median migration distances for G3 (Table 31). The mean migration distance for all individuals (i.e. including both individuals who migrated and those who did not) was 10km. The mean and median distances for migrating individuals were 96km and 26km, respectively. The shorter migration distance values (10km and 26km) are within the range of previously reported average migration distances ( Ammerman and Cavalli Sforza 1984; Markset al. 2012; Wijsman and Cavalli Sforza 1984) These mean m igration distances potentially demarcate the distances within which post marital residence patterns (patrilocality in the case of Yemen) have a distinguishable effect on genetic structure ( Markset al. 2012 ; Raaumet al. 2013) Beyond the mean distance is probably where sex biased migration is less detectable. The median value (26km) is within the range of 1030km that Ammerman and CavalliSforza ( 1984) believe is plausible for migration distance in agriculturalist societies. Dividing 26km by a generation time of 25 years results in a migration speed of 1.04km/year. This value is comparable to the 1km/year migration speed for the Neolit hic transition estimated from archeological data ( Hazelwood and Steele 2004; Pinhasiet al. 2005) These si milarities suggest that the median distance is representative of migration distances of agriculturalist groups. A migration speed (3.84km/year) from the mean value for migrating individuals (96km) falls within the broad range of hunter gatherer migration speeds calculated from archeological evidence. Fort et al ( 2004) estimated the speed of the hunter gatherers PAGE 120 120 recolonization of northern Europe after the last glacial maxima between 0.7 and 1.4km/year. Hamilton and Buchanan ( 2007) estimated a speed of 5 8km/year for the colonization of North America, while Hazelwood and Steele ( 2004) obtained estimates of 6 10km/year. Because our value is intermediate from the values of these studies, it provides a distance that may be more generally applicable to other migration processes, particularly de novo colonization migration distances by hunter gatherers. This can be seen when we compare our estimate with Macaulay et als ( 2005) inferred migration speed for the colonization of Southeast Asia. Based on founder time estimates from Eurasian and Australasian mtDNAs and the distance between India and Australasia, Macaulay et al infer a migration speed of 4km/year. Our empirical estimate of 3.84 km/year supports that the migration process they proposed, is i n fact plausible. While migration distance has been estimated through different approaches, few studies have estimated the proportion of migrants ( Boattiniet al. 2007 ; Markset al. 2012 ; Mortonet al. 1971; Woodet al. 1985) We calculated the proportion of migrants for G3 to be 0.102 (or 0.086 when adjusting for back migration in the four generations). These values are smaller than the 0.4 proportion of migrants that we calculated from Wood et als ( 1985) dataset from migration between parishes in Papua N ew Guinea and the 0.366 estimate obtained from the calculation of individuals that were not born in the same parishes as their parents in La Cabrera, Spain ( Boattiniet al. 2007) These differences from our estimates seem reasonable as Wood et als estimates are from a more recent population (and are closer to our G1 and G2 estimates) and Boatinni et als estimates are from a more developed country. Our estimates are somewhat larger than the 0.032 proportion of migrants into the island of Pingelap in Micronesia by Morton et al PAGE 121 121 ( 1971) However Morton et als value is close to our adjusted proportion of migrants (0.087). We also calculated the maximum and average number of individuals moving between the same locations, for a proportion of migrants of 0.0036 and 0.0011, respectively. These lower values are consistent with findings by Deshpande et al ( 2009) where the genetic estimates of proportion of migrants (i.e. migration rates) for a world wide colonization model is less than 0.01. Our values are similar to findings by Mir Herrans and Mulligan ( 2013) where the proportion of migrants exchanged between African and nonAfricans populations was 0.001 and similar to the migration rate for nonAfrican populations (1.5x103) obtained by Cox et al ( 2008) The similarity of our estimates with those of other migration studies, suggests that our values can be used to generate testable models for prehistoric reconstruction. Application of M igration Estimates in Prehistoric Demographic Modeling Model based approaches for inferring prehistoric processes from genetic variation are becoming increasingly popular ( Marjoram and Tavar 2006) These approaches, such as approximate Bayesian computation ( Beaumontet al. 2002) require the generation of explicit demographic models to compare to empirical data. Including specific values for known parameters and informative ranges of values for unknown parameters increases the probability of identifying the best model to explain the data. The results from our study provide estimates that can be used to fix or set ranges on parameters related to migration, such as gene flow or founding population size, so that other parameters of interest can be addressed in greater depth, e.g. time of a demographic event. For example, the maximum and average proportion of individuals moving between the same locations (0.0036 and 0.0011) can be used to define gene PAGE 122 122 flow (or migration rates) between populations stretching fro m southern Asia to northern Africa to create simulated DNA for models that address the back migration into Africa. The larger migration values (0.102 or 0.086) could be used to define the founding population sizes for each new population. Defining these parameters would allow for an in depth exploration of the timing of the back migration. Additionally, our results provide estimates to generate more geographically explicit models. Our mean and median migration distances (96km and 26km) provide estimates for the distance between populations, particularly for large scale movements, such as the back migration from southern Asia. The migration distance between each population would define the number of populations to be simulated for the region under study. For example, a distance of 100km between each population would require ~70 populations between southern Asia and northern Africa (approx. 7,000km). Understanding the possible distances involved in large scale movements also helps us determine how rapidly a mig ration could have occurred and how levels of gene flow may have been affected between the populations. The lack of migration directionality in our results suggests that explicitly including stochasticity or multidirectionality when describing the movement between populations might more accurately reflect the largescale migration process. For example, the back migration to Africa probably included movement through established populations, where the migrants settled in some of the established populations, but not in others. Therefore, a lattice stepping stone migration model, that includes some randomness in when a migration occurs and between which populations, might better reflect this migration process. PAGE 123 123 Our results show there is over a 58% correlation between female and male movement in marital pairs, in which more pairs move together with increasing distance. Additionally, we show that 56% of migration events in G3 were by marital pairs. This means that at least 50% of the migrants have a 1:1 female to m ale ratio. Even if the remaining 50% of migrants are only female or male, the ratio is at most 3:1. These results argue for, at most, a 3:1 ratio (for either sex) of sex biased migration for migrations at short distances, where post marital residence has a larger effect on population structuring ( Markset al. 2012; Raaumet al. 2013 ) Alternatively, for longer migrati ons, such as the migration from southern Asia to northern Africa, a female to male ratio closer to 1:1 may more accurately model demographically balanced populations that would have been reproductively self sustaining. In this study, we have analyzed empirical data on migration patterns over four generations of human populations in Yemen in order to gain insight into the factors that influence migration, and specifically may have affected prehistoric movements throughout human evolution. We have addressed the effect of these factors on genetic variation and provided empirical estimates for migrationrelated parameters that can be used to generate demographic models in model based methods of prehistoric reconstruction. Our empirical estimates of generation G3 have provided values for proportion of migrants, with values ranging from 0.102 or 0.086 proportion of overall migration, to 0.0036 or 0.0011 proportion of migrants between two specific populations. We have also provided migration distances (96km and 26km, mean and median, respectively) that can be used define the distance between populations and therefore the number of populations for the area under study. The findings from this study shed PAGE 124 124 light on human migration patterns and are intended to enable more accurate reconstruction of the demographic processes that characterized human evolution. PAGE 125 125 Table 31. Summary statistics for migration distances. Median and mode for All individuals was zero for all groups. Valu es in parenthesis represent sample size All individuals Mean SD Median Mode G1 (351) 69 249 0 0 G1Female (70) 156 405 . G1Male (251) 48 186 . G2 (702) 72 265 . G2Female (351) 73 269 . G2Male (351) 72 262 . G3 (1404) 10 66 . G3Female (702) 9 61 . G3Male (702) 10 71 . Migrating individuals Mean SD Median Mode G1 (97) 251 424 81 103 G1Female (22) 497 601 103 103 G1M ale (75) 179 328 75 103 G2 (241) 211 419 29 103 G2Female (132) 193 411 23 103 G2Male (109) 232 430 44 103 G3 (143) 96 188 26 26 G3Female (76) 82 169 24 17 G3Male (67) 111 208 28 26 PAGE 126 126 Table 32. Best model to explain probability of migration. p<0.04 for all factors. The probability of migration decreases in G3, decreases in males, and decreased with a more eastern birthplace, in comparison to G1fem. The probability of migration increases in G2 and increases with more northern birthplaces and places of residence. Factor Coefficient Intercept 1.924 Generation:G2 0.256 Generation:G3 1.244 Sex:Male 0.240 Birth Latitude 0.221 Birth Longitude 0.121 Residence Latitude 0.225 PAGE 127 127 Table 33. Estimates for the direction of migration in each collection site across all three generation groups. aMean directional angle is measured clockwise from due North. bMean distance is measured in decimal degrees.cCircular variance describes the variation associated with the directional mean, where values close to 0 represent a similar direction for all migration vectors and values close to 1 correspond to vectors in all compass directions Collection site Mean directional anglea Mean d istanceb Circular variancec Abyan 103.19 1.659 0.676 Al B ayda 107.87 1.405 0.806 Al Hudaydah 321.06 1.814 0.867 Al Mahra 161.96 2.282 0.833 Amran 120.51 0.995 0.704 Dhamar 62.78 0.370 0.758 Hadramout 171.57 3.953 0.675 PAGE 128 128 Table 34. Directional means estimates for each group by collection site. aMean directional angle is measured clockwise from due North. bMean distance is measured in decimal degrees. Group Generation Collection site Mean directional angle a Circular variance Mean distance b G1female G1 Al Bayda 20.676 0.045 0.174 G1female G1 Al Hudaydah 171.328 0.471 5.307 G1 female G1 Al Mahra 149.404 0.714 3.869 G1female G1 Hadramout 153.945 0.023 9.225 G1male G1 Abyan 93.105 0.513 3.115 G1male G1 Al Bayda 282.164 0.914 3.400 G1male G1 Al Hudaydah 98.593 0.603 1.960 G1male G1 Al Mahra 213.058 0.494 0.956 G1male G1 Amran 151.418 0.640 1.382 G1male G1 Dhamar 79.936 0.522 0.515 G1male G1 Hadramout 155.123 0.430 2.441 G2female G2 Abyan 93.264 0.675 1.525 G2female G2 Al Bayda 94.257 0.773 1.335 G2female G2 Al Hudaydah 329.910 0.795 1.620 G2female G2 Al Mahra 230.602 0.8 53 3.961 G2female G2 Amran 112.443 0.645 0.931 G2female G2 Dhamar 251.241 0.781 0.290 G2female G2 Hadramout 179.180 0.895 4.442 G2male G2 Abyan 116.708 0.539 2.298 G2male G2 Al Bayda 83.530 0.827 2.226 G2male G2 Al Hudaydah 6.450 0.765 2.635 G2male G2 Al Mahra 285.254 0.926 3.355 G2male G2 Amran 109.280 0.400 0.675 G2male G2 Dhamar 243.264 0.895 0.540 G2male G2 Hadramout 284.876 0.952 3.493 G3female G3 Abyan 106.306 0.522 0.200 G3female G3 Al Bayda 135.958 0.631 0.255 G3female G3 Al Hudaydah 26 4.634 0.643 0.148 G3female G3 Al Mahra 114.916 0.698 0.973 G3female G3 Amran 294.408 0.765 1.258 G3female G3 Dhamar 48.563 0.530 0.232 G3female G3 Hadramout 181.400 0.006 3.917 G3male G3 Abyan 281.966 0.846 0.479 G3male G3 Al Bayda 141.926 0.633 0.37 1 G3male G3 Al Hudaydah 303.292 0.583 0.339 G3male G3 Al Mahra 126.886 0.757 1.427 G3male G3 Amran 252.032 0.689 1.334 G3male G3 Dhamar 61.057 0.557 0.292 G3male G3 Hadramout 181.400 0.006 3.917 PAGE 129 129 Figure 31. Proportion of migrants by sex for each generation group. P values are shown for goodness of fit tests between groups. PAGE 130 130 Figure 32. Density plots combining migration distance and frequency of the distance for each group. Wilcoxon Rank tests were performed for G1 and G2 within generation comparisons and Kruskal Wallis tests were performed for G3 within generation comparison and between generation comparisons. P values are shown for the r espective tests. PAGE 131 131 Figure 33. Plot of migration distances for marital pairs. G2 (red circle), G3M (green triangle), G3P (blue cross). The solid line shows a theoretical 1:1 relationship, where females and males have the same dispersal dist ance. The inn er box shows a close up of the relationship for distances less than 250km. PAGE 132 132 Figure 34. Migration direction vectors and mean migration direction (large arrows) by collection site over all three generations. PAGE 133 133 Figur e 35. Migration direction vectors and mean migration direction (females light gray arrows, males dark gray arrows) for each collec tion site by generation group. A ) G1 B ) G2 C ) G3 A B C PAGE 134 134 CHAPTER 4 CONCLUSION Humans facility for dispersal and continuous movement has generated elaborate evolutionary histories throughout different regions of the planet. So me regions have experienced greater amounts and more intensive migration than others. The region from northeast Africa, ranging from the Horn of Africa (includes Ethiopia, Somalia, Djibouti, and Eritrea) to Egypt, through the Arabian Peninsula to western Asia (somewhere around Iran) has been the center of various major migrations in human evolutionary history. East Africa (including the Horn of Africa) is argued to have been a refugia ~70kya ( Compton 2011 ; Scholzet al. 2007) and the area of origin of mtDNA haplogroup L3 ( Atkinsonet al. 2009; Soareset al. 2012) L3 arose in east Africa 6080kya and was quickly followed by population growth ( Atkinsonet al. 2009) and migrations out of Africa and within Africa ( Soareset al. 2012) Haplogroups M and N, which are nonAfrican branches of L3, arose during the process out of Africa, or immediately thereafter somewhere in western Asia, 4070kya ( Forster 2004) The presence of haplogroup M1 in northern Africa supports a mig ration back to Africa from southern Asia 3545kya ( Gonzalezet al. 2007 ; Olivieriet al. 2006 ) This out o f Africa region (OAR) is posited to have had various local refugia during the latter part of the Last Glacial Maxima (LGM) from which a local expansion occurred starting 1215kya ( Al Abriet al. 2012; Cernyet al. 2011 ; Roseet al. 2013) Commercial exchange may have led to local migrations between Ethiopia and the southern part of the Arabian Peninsula since ~9kya, as well as between Ethiopia and the Levant (which includes countries off the east coast of the Mediterranean Sea) since ~3kya ( Boattiniet al. 2013; Paganiet al. PAGE 135 135 2012) A more recent migration occurred in the 7th century AD, with the expansion of Islam in the Arabian Peninsula and North Africa ( Hennet al. 2012; Nebelet al. 2002) Each of these demographic processes generated a unique pattern of genetic variation in the existing population. Every new migration wave introduced the new genetic variation into the OAR. After about 70ky of modern human occupation of this region, the combination of these demographic processes has created a very complex pattern of genetic varia tion. Reconstructing human evolutionary history in this region from the genetic data presents the challenge of inferring different ancient processes from the genetic variation generated by the combined demographic processes. The challenge is greatest when trying to reconstruct older processes, such as the migration out of Africa, from current patterns of genetic variation. The complexity of the processes that have occurred in the region can be disentangled, in part, by generating simulated patterns of genetic variation for demographic scenarios that might realistically represent the demographic processes under study and testing genetic data against these models (i.e. approximation model based methods). Hypothesis models can be developed to describe a singl e process, such as the migration of modern humans out of Africa, or multiple processes (e.g. including numerous migrations). To generate informative hypothesis models that will allow clear interpretation of the empirical data, its critical to identify whi ch hypothesis models generate distinguishable differences in the genetic variation and how each demographic parameter is contributing to the genetic variation. PAGE 136 136 The proportion of genetic variation explained by each parameter informs on how accurately the parameters can be estimated. A parameter that explains a larger portion of the genetic variation is more likely to be accurately estimated, than a parameter that explains much less of the variation. A parameter that explains a large portion of the genetic v ariation can cause many of the demographic scenarios to generate similar patterns of genetic variation, reducing the probability of accurately estimating the value for the parameters that explain less of the genetic variation. The individual contribution o f each demographic parameter to the total genetic variation can be calculated through the comparison of demographic scenarios. It is important to identify the parameters that explain more of the variation to determine the type of additional data (i.e. genetic or nongenetic) that would improve the estimation of the less contributing parameters. Defining a range of realistic values, calculated from these additional sources of data, for parameters that explain a larger proportion of the genetic variation can reduce the number of demographic scenarios tested against the empirical genetic data and lead to the selection of only one hypothesis model that best explains the data. This, in turn, leads to the improved estimation of parameters that explain less of t he genetic variation. My results show that for the migration of modern humans out of Africa the colonization size and rate of gene flow explain most of the genetic variation ( Mir Herrans and Mulligan 2013) Therefore, changes in the values of these parameters can cause significant changes on the patterns of genetic variation. The range of values that has been proposed for the size of the migrating population, from 1% to 33% ( Atkinsonet al. 2008; Deshpandeet al. 2009 ; Tenesaet al. 2007) can have vastly different effects on genetic variation. A population that is 1% of the original population can have a PAGE 137 137 significant reduction in genetic variation and can quickly become genetica lly dissimilar to the original population. A population that is 30% of the original population can maintain similar levels of genetic variation as the original population. While the effects of the extreme values for colonization size are fairly well unders tood, the effects of more intermediate, and probably more realistic values on genetic variation are less obvious. My results show that scenarios with colonization size greater than 10% usually dont generate distinguishable patterns of genetic variation, so the values for other parameters cant be accurately estimated. Similarly, scenarios with migration rates greater than 103 or less than 103 generate similar patterns of genetic variation, limiting the inference of other parameter values. Therefore, it becomes a priority to identify whether a realistic value for colonization size is around 10% and for migration rate is around 103, in order to generate hypothesis scenarios in which the values for the other parameters, such as the time for the migration out of Africa, can be estimated. The comparison of patterns of migration across four generations in Yemen (Chapter 3), show that the grandparents generation had the lowest levels and most restricted patterns of migration. This suggests that values for migr ation parameters estimated for this generation are most representative of migration parameter values for prehistoric demographic processes. Therefore, the empirical values calculated for the proportion of migrants can serve to narrow migration related parameters in hypothesis models of prehistoric processes, and more specifically for the model of migration out of Africa. The overall proportion of migrants for the grandparents generation was 0.102 (or 0.086 when adjusting for back migration in the four generations) and the maximum and average number of individuals moving between the same locations was 0.0036 and PAGE 138 138 0.0011, respectively. The overall proportion of migrants (0.102 or 0.086) could reflect the number of individuals moving at a given time, such as mi ght occur during a colonization event. Although in my study the most supported colonization size was 1%, when comparing Fst values for the simulated scenarios with Fst values of African versus European and African versus Asian, a colonization size of 10% also had strong support. Furthermore, using model based methods with a serial founder event model for out of Africa and nuclear data, Deshpande et al ( 2009) found t he colonization size to be 0.09 to 0.18 of the population. The number of individuals moving between the same locations (0.0036 and 0.0011) could represent the migration rate between adjacent populations. Cox et al ( 2008) report a rate of 1.5x103 for nonAfrican populations and 2.7x104 for African populations and Gravel et al ( 2011) suggest a rate of 1.5x104 between the African population and the nonAfrican population that gave rise to European and Asian populations. In contrast, Deshpande et al report a higher rate of migration between adjacent populations of up to 102. The results from my simulated demographic scenarios support a rate of 103, which was also shown to be a transition point from which it should be possible to distinguish between scenarios on either side (GF>103 or GF<103). Together the results from my studies suggest, that for human demographic reconstructions, migration parameters can be narrowed to a colonization size of (or around) 0.080.1 and migration rate of 104102 between adjacent populations. These values for migration parameters offer the best possibility of accurately accounting for the effects of migration and identifying the values for other parameters of interest. The demographic scenarios for the model out of Africa included two values for the timing o f the migration out of Africa (50kya and 100kya), representing the most PAGE 139 139 extreme values reflected in the literature, with the largest difference between the values ( Mir Herrans and Mulligan 2013) The timing of the population fission and migration out of Africa is of great interest in many fields. The wide range of values for when humans left Africa estimated from mtDNA and NRY overlap with multiple climate stages. These stages, as defined by oxygen isotopes have alternating cool and warm climates ( McDermott 2004) The state of the climate when humans left Africa has implications for human behavior, suggesting, for example, whether humans migrated to new areas while the climate made movement easier or w hether they migrated in search of new areas because the climate was unsuitable. Accurately identifying the timing for the migration out of Africa, thus, has implications for understanding human behavior. The result of the comparisons of the demographic sc enarios showed that scenarios of 50kya could not generally be distinguished from scenarios of 100kya based on the genetic variation of mtDNA. This inability to distinguish between 50kya and 100kya occurs most likely because of the small effects of time on the genetic variation of human mtDNA ( Mir Herrans and Mulligan 2013) These results suggest that additional genetic and nongenetic data could improve the ability to estimate the time of the migration of humans out of Africa. In a recent study of the population history of the KhoeSan in Africa, Veeramah et al ( 2012 ) calculated the time of population fission between nonPygmies, Eastern Pygmies and Western Pygmies and found the nonPygmy/Pygmy population split occurred at ~49kya and the Eastern/Western split occurred at ~32kya. The identification of times of two population fissions within a 17ky window of time demonstrates that time can be more accurately estimated with the use of multiple genetic markers. Advances in sequencing technology are now allowing us to PAGE 140 140 generate large quantities of genetic markers for multiple individuals at affordable prices. The increase in genetic data, along with the incorporation of demographic data, such as the migration estimates previously described, increase the possibility of more accurate estimates of realistic t ime values from demographic reconstructions. This will allow the identification of critical time points in human evolution and contribute to the overall understanding of human behavior. The identification of the most informative summary statistics also contributes to the improvement of demographic reconstructions. Approximation approaches using summary statistics require a small number of informative summary statistics to avoid the situation where the comparison of simulated datasets to the empirical datase t for each summary statistic excludes so many simulated scenarios that it is impossible to make inferences about the empirical data ( Beaumontet al. 2002; Hamilton 2005 ; Wegmannet al. 2009) Through the comparison of the developed demographic scenarios by multiple summary statistics, I have identified summ ary statistics that are informative at summarizing genetic variation of specific parameters ( Mir Herrans and Mulligan 2013) Fst summarizes the individual effects of gene flow and time and the combined effect of gene flow and time, while number of singletons optimally summarizes the effect of colonization time (Table 23). Tajimas D summarizes the combined effects of coloni zation size and gene flow, as well as the combined effects of gene flow and time. These results suggest that the combined use of these three summary statistics (i.e. Fst, number of singletons and Tajimas D) offers the possibility of more accurate reconstr uctions of human demographic processes. PAGE 141 141 The results presented suggest that the out of Africa model developed for this dissertation, while simple, provides a useful model that can be refined to further investigate the process of modern humans out of Africa. By incorporating the results from the empirical patterns of migration, a more geographically explicit model can be generated that can serve to address the specific route humans took during their migration out of Africa. Using the framework presented in this dissertation, different demographic scenarios can be compared to determine whether demographic scenarios with different routes can be distinguished and which scenarios present informative hypotheses to compare to empirical data. In addition to the compl exity described thus far, the expansion out of Africa is further debated as to whether it occurred in one migration wave or two waves and which route was used. Using traditional approaches of demographic inference from uni parentally inherited markers (i.e. mtDNA and NRY), those who propose two routes suggest one wave went South through the Horn of Africa, across the Red Sea, and across the Arabian Peninsula 5969kya, possibly giving rise to the Asian population, and one wave went North through the Levant 39 53kya, possibly giving rise to the European population ( Luiset al. 2004 ; Maca Meyeret al. 2003) Those w ho propose one migration posit either a Southern route 6080kya ( Forster 2004; Metspaluet al. 2004) or a Northern route 4050kya ( Rowoldet al. 2007) Additionally, it is unclear whether the haplotypes that characterize non African populations, and are lacking in African populations, diverged in Africa right at the time of departure or after the new population migrated out of Africa ( Forster 2004) PAGE 142 142 Multiple other studies have addressed the migration out of Africa using approximation model based methods ( Deshpandeet al. 2009 ; Gravelet al. 2011; Gutenkunstet al. 2009 ; Liuet al. 2006; Ramachandranet al. 2005) Two main models have been used among these studies, a serial founder event model ( Deshpandeet al. 2009; Liuet al. 2006 ; Ramachandranet al. 2005) and a three population model ( Gravelet al. 2011; Gutenkunstet al. 2009) The serial founder event models have described a onedimensional stepping stone model for the movement from Eastern Africa. They assume a landbased movement, leading to a coloniz ation through the Levant and then linearly across Asia and to the Americas. An advantage of this approach is that it offers the possibility of incorporating some ecological parameters such as the size of each population and the carrying capacity for each population (i.e. the size of the population before a new colonization occurs). These studies demonstrate that a serial founder event model best explains the migration out of Africa. The three population models assumes a population fission between African an d nonAfrican populations and a second fission between European and Asian populations. An advantage of this approach is that it offers the possibility of two dimensions of movement, thus the migration between the three populations can be estimated. The goal of these migration studies has been to describe models for humans worldwide colonization, not the specific process of humans moving from Eastern Africa to Western Asia. Therefore, they assume a single migration wave out of Africa and do not consider th e demographic scenarios that test the one wave and two wave models for the migration out of Africa. By integrating the findings from these studies and the results from my analyses, a geographically explicit model can be developed that allows the comparison of a single PAGE 143 143 northern migration (NM), a single southern migration (SM), and a two wave northern and southern migration (NSM) out of Africa. The use of a twodimensional stepping stone model (Kimura and Weiss 1964) allows for serial founder events to occur in two dimensions and offers the advantages of the two out of Africa models previously described. This twodimensional model creates a lattice of populations that can be overlayed onto geographic boundaries extending from eastern Africa (from the Horn of Africa to Egypt), across the Arabian Peninsula, and to Western Asia. The empirical distance calculated for mean distance of migrating individuals (96km) was identified as a reasonable distance for migration of hunter gatherers in Chapter 3. Thus, if hunter gatherers move ~100km to a new population, the size of each population can be defined as being 100x100km. The distance also defines the number of populations that might have realistically been colonized across the geographic boundaries. Starting the migr ation from Addis Ababa, Ethopia ( Deshpandeet al. 2009; Liuet al. 2006 ; Ramachandranet al. 2005) and ending somewhere in Iran, a northern migration would require more population colonizations, than a southern migration. The number of populations can serve as a parameter that distinguishes the NM from the SM. Programs such as SPLATCHE ( Curratet al. 2004) allow a cost to be defined across the geographic boundaries to define the difficulty, and therefore the probability, of a specific area to be traversed during migration. Alternative demographic scenarios with different geograp hic costs would allow the same geographic model to be tested for the different hypothesized routes. For example, to define the NM, a high cost can be assigned to the area of the Red Sea, such that the migration occurs through the North. PAGE 144 144 Alternatively, the populations North of Addis Ababa can be assigned a cost that requires the migration to cross the Babel Mandeb Strait, which was the narrowest part of the Red Sea ( Baileyet al. 2007) to design an SM route. Additionally, costs could be assigned to the northern populations and the Red Sea area to generate patterns of genetic variation for the NSM two wave demographic scenarios. The cost map can also be used to describe the Arabian Peninsula to define the more likely path. The results of the migration patterns in Chapter 3 illustrate that there is some stochastic ity in the direction of migration. This stochasticity can be incorporated in the geographic model to account for movement across large continental areas, such as the Arabian Peninsula. By defining a number of migrants, based on realistic migration rates id entified as optimal in this dissertation (104102), in a migration model that draws from a multinomial distribution to assign those migrants to the adjacent populations (also available in SPLATCHE), each adjacent population receives a different number of migrants per generation that is representative of the realistic migration rates. This approach, along with the cost map, would generate demographic scenarios that account for the uncertainty of the migration process across the Arabian Peninsula. The abi lity to define additional demographic and ecological parameters increases the possibility of estimating the time of the migration out of Africa, particularly as it applies to the two migration waves. The realistic colonization rates identified in this dis sertation (0.080.10) can be incorporated. The population growth rate can be defined as a logistic growth rate with a rate of 1.8 ( Deshpandeet al. 2009; Liuet al. 2006 ; Ramachandranet al. 2005) The optimal carrying capacity has been estimated to be between 600 and 1200 ( Deshpandeet al. 2009; Liuet al. 2006) PAGE 145 145 The time of migration(s) out of Africa is of great interest, as the timing primarily defines the two migra tion waves. Analyses of nuclear data have estimated the migration out of Africa at ~50kya, with a 95% confidence interval of 4070kya ( Fagundeset al. 2007; Gravelet al. 2011; Gronauet al. 2011) Hypothesis scenarios with different time intervals for each route can be defined in which the NM occurs between 4055kya, the SM occurs 50 80kya, and the NSM has two time intervals of 4055kya and 5080kya (50kya is incorporated into all scenarios to include the fact that 50kya is the most likely time that has been estimated for the migration out of Africa). The overlap in time intervals suggests that estimating the time of the migration out of Africa, particularly if there were two migrations at different times, presents the biggest challenge for reconstructing the migration out of Africa. The combination of different time intervals, geographically distinct scenarios, and realistic demographic parameters offers the greatest possibility of generating distinct hypothesis scenarios for the NM, SM, and NSM routes that are informative for accurately reconstructing the migrat ion out of Africa. The considerations described above offer the possibility of generating demographic scenarios that most accurately represent the different scenarios for the migration out of Africa. Future studies comparing these scenarios will allow us to identify whether the patterns of genetic variation generated from migration through the Northern and Southern routes can be distinguished, as well as identify which scenarios are distinguishable and most informative. Following the approaches in Liu et al ( 2006) where they compared data from simulated populations with data from empirical populations that corresponded to the same geographical locati ons, the informative PAGE 146 146 hypothesis scenarios can then be compared to the empirical data to test which migration scenario (NM, SM, or NSM) best explains the migration out of Africa. The revised model presented for the migration out of Africa illustrates how ge netic and nongenetic data can be incorporated to improve human demographic reconstruction. Overall, this dissertation demonstrates how integrating data from nongenetic disciplines can enhance our ability to make inferences from genetic data and improve our interpretations of prehistoric demographic processes. Interdisciplinary approaches, such as I have described in this dissertation, will be essential as we continue to move forward to unravel the evolutionary history of different species. PAGE 147 147 LIST OF REFERE NCES Al Abri A, Podgorna E, Rose JI, Pereira L, Mulligan CJ, Silva NM, Bayoumi R, Soares P, Cerny V 2012. PleistoceneHolocene Boundary in Southern Arabia from the Perspective of Human Mtdna Variation. Am J Phys Anthropol 149: 291298. Ammerman AJ, Cavalli Sforza LL. 1984. The Neolithic Transition and the Genetics of Populations in Europe. Princeton, New Jersey: Princeton University Press. Atkinson QD, Gray RD, Drummond AJ 2009. Bayesian Coalescent Inference of Major Human Mitochondrial DNA Haplogroup Expansions in Africa. Proc Biol Sci 276: 367373. Atkinson QD, Gray RD, Drummond AJ 2008. Mtdna Variation Predicts Population Size in Humans and Reveals a Major Southern Asian Chapter in Human Prehistory. Molecular Biology and Evolution 2 5: 468474. Bailey GN, Flemming NC, King GCP, Lambeck K, Momber G, Moran LJ, Al Sharekh A, Vita Finzi C 2007. Coastlines, Submerged Landscapes, and Human Evolution: The Red Sea Basin and the Farasan Islands. The Journal of Island and Coastal Archaeology 2: 127160. Balding DJ, Bishop M, Cannings C. 2003. Handbook of Statistical Genetics. In. West Sussex, England: John Wiley & Sons, Ltd. Bandelt H J, Forster P, Rhl A 1999. MedianJoining Networks for Inferring Intraspecific Phylogenies. Molecular Biology and Evolution 16: 37 48. Barker G, Barton H, Bird M, Daly P, Datan I, Dykes A, Farr L, Gilbertson D, Harrisson B, Hunt C, Higham T, Kealhofer L, Krigbaum J, Lewis H, McLaren S, Paz V, Pike A, Piper P, Pyatt B, Rabett R, Reynolds T, Rose J, Rushworth G, S tephens M, Stringer C, Thompson J, Turney C 2007. The 'Human Revolution' in Lowland Tropical Southeast Asia: The Antiquity and Behavior of Anatomically Modern Humans at Niah Cave (Sarawak, Borneo). J Hum Evol 52: 243261. Beaumont MA 2010. Approximate Bay esian Computation in Evolution and Ecology. Annual Review of Ecology, Evolution, and Systematics 41: 379406. Beaumont MA, Wenyang Z, Balding DJ 2002. Approximate Bayesian Computation in Population Genetics. Genetics 162: 2025 2035. PAGE 148 148 Behar DM, Villems R, Soodyall H, Blue Smith J, Pereira L, Metspalu E, Scozzari R, Makkan H, Tzur S, Comas D, Bertranpetit J, QuintanaMurci L, Tyler Smith C, Wells RS, Rosset S 2008. The Dawn of Human Matrilineal Diversity. The American Journal of Human Genetics 82: 11301140. Boattini A, Blanco Villegas MJ, Pattener D 2007. Genetic Structure of La Cabrera,Spain, from Surnames and Migration Matrices. Human Biology 79: 649666. Boattini A, Castri L, Sarno S, Useli A, Cioffi M, Sazzini M, Garagnani P, De Fanti S, Pettener D, Luiselli D 2013. Mtdna Variation in East Africa Unravels the History of AfroAsiatic Groups. Am J Phys Anthropol 150: 375385. Bowcock AM, Kidd JR, Mountain JL, Hebert JM, Carotenuto L, Kidd KK, Cavalli Sforza LL 1991. Drift, Admixture, and Selection in Hum an Evolution: A Study with DNA Polymorphisms. Proc Natl Acad Sci U S A 88: 839843. Bustamante CD, Ramachandran S 2009. Evaluating Signatures of Sex Specific Processes in the Human Genome. Nature Genetics 41: 810. Calafell F, Grigorenkob EL, Chikani AA, Kidd KK 2001. Haplotype Evolution and Linkage Disequilibrium: A Simulation Study. Human Heredity 51: 8596. Campbell MC, Tishkoff SA 2008. African Genetic Diversity: Implications for Human Demographic History, Modern Human Origins, and Complex Disease Mapping. Annual Review of Genomics and Human Genetics 9: 403433. Carvajal Rodrguez A 2008. Simulation of Genomes: A Review. Current Genomics 9: 155159. Cerezo M, Achilli A, Olivieri A, Perego UA, Gomez Carballa A, Brisighelli F, Lancioni H, Woodward SR, Lopez Soto M, Carracedo A, Capelli C, Torroni A, Salas A 2012. Reconstructing Ancient Mitochondrial DNA Links between Africa and Europe. Genome Res 22: 821826. Cerny V, Mulligan CJ, Fernandes V, Silva NM, Alshamali F, Non A, Harich N, Cherni L, El Gaaie d AB, Al Meeri A, Pereira L 2011. Internal Diversification of Mitochondrial Haplogroup R0a Reveals Post Last Glacial Maximum Demographic Expansions in South Arabia. Mol Biol Evol 28: 7178. PAGE 149 149 Cerny V, Mulligan CJ, Ridl J, Zaloudkova M, Edens CM, Hajek M, Pereira L 2008. Regional Differences in the Distribution of the SubSaharan, West Eurasian, and South Asian Mtdna Lineages in Yemen. Am J Phys Anthropol 136: 128137. Clark G, Anderson A, Wright D 2006. Human Colonization of the Palau Islands, Western Micro nesia. The Journal of Island and Coastal Archaeology 1: 215232. Compton JS 2011. Pleistocene SeaLevel Fluctuations and Human Evolution on the Southern Coastal Plain of South Africa. Quaternary Science Reviews 30: 506527. Cox MP, Woerner AE, Wall JD, H ammer MF 2008. Intergenic DNA Sequences from the Human X Chromosome Reveal High Rates of Global Gene Flow. BMC Genet 9: 76. Currat M, Ray N, Excoffier L 2004. Splatche: A Program to Simulate Genetic Diversity Taking into Account Environmental Heterogeneit y. Molecular Ecology Notes 4: 139142. Davis KF, D'Odorico P, Laio F, Ridolfi L 2013. Global SpatioTemporal Patterns in Human Migration: A Complex Network Perspective. PLoS One 8: e53723. de Haan A, Rogaly B 2002. Introduction: Migrant Workers and Their Role in Rural Change. Journal of Development Studies 38: 114. DeGiorgio M, Jakobsson M, Rosenberg NA 2009. Out of Africa: Modern Human Origins Special Feature: Explaining Worldwide Patterns of Human Genetic Variation Using a Coalescent Based Serial Foun der Model of Migration Outward from Africa. Proceedings of the National Academy of Sciences 106: 1605716062. Deshpande O, Batzoglou S, Feldman M, Cavalli Sforza L 2009. A Serial Founder Effect Model for Human Settlement out of Africa. Proceedings of the Royal Society B: Biological Sciences 276: 291300. Dresch P. 1989. Tribes, Government, and History in Yemen. New York: Clarendon Press. Epperson BK. 2003. Geographical Genetics. Princeton, New Jersey: Princeton University Press. PAGE 150 150 Excoffier L, Laval G, Sc hneider S 2005. Arlequin (Version 3.0): An Integrated Software Package for Population Genetics Data Analysis. Evolutionary Bioinformatics Online 2005: 4750. Excoffier L, Smouse PE, Quattro JM 1992. Analysis of Molecular Variance Inferred from Metric Dist ances among DNA Haplotypes: Application to Human Mitochondrial DNA Restriction Data. Genetics 131: 479491. Fagundes NJR, Ray N, Beaumont M, Neuenschwander S, Salzano FM, Bonatto SL, Excoffier L 2007. Statistical Evaluation of Alternative Models of Human Evolution. Proceedings of the National Academy of Sciences 104: 1761417619. Federal Research Division. 2008. Country Profile: Yemen. Published:Congress Lo. http://lcweb2.loc.gov/frd/cs/profi les/Yemen.pdf Felsenstein J 1981. Evolutionary Trees from DNA Sequences: A Maximum Likelihood Approach. Journal of Molecular Evolution 17: 368376. Felsenstein J 1983. Statistical Inference of Phylogenies. Journal of the Royal Statistical Society. Series A (General) 146. Forster P 2004. Ice Ages and the Mitochondrial DNA Chronology of Human Dispersals: A Review. Philosophical Transactions of the Royal Society B: Biological Sciences 359: 255264. Fort J, Pujol T, CavalliSforza LL 2004. Palaeolithic Populations and Waves of Advance. Cambridge Archaeological Journal 14: 5361. Fu YX 1995. Statistical Properties of Segregating Sites. Theor Popul Biol 48: 172197. Garrigan D, Hammer MF 2006. Reconstructing Human Origins in the Genomic Era. Nat Rev Genet 7: 669 680. Gonder MK, Mortensen HM, Reed FA, de Sousa A, Tishkoff SA 2007. WholeMtdna Genome Sequence Analysis of Ancient African Lineages. Molecular Biology and Evolution 24: 757768. PAGE 151 151 Gonzalez AM, Larruga JM, AbuAmero KK, Shi Y, Pestano J, Cabrera VM 2007. Mitochondrial Lineage M1 Traces an Early Human Backflow to Africa. BMC Genomics 8: 223. Gravel S, Henn BM, Gutenkunst RN, Indap AR, Marth GT, Clark AG, Yu F, Gibbs RA, Genomes P, Bustamante CD 2011. Demographic History and Rare Allele Sharing among Human Populations. Proc Natl Acad Sci U S A 108: 1198311988. Gray C, Bilsborrow R 2013. Environmental Influences on Human Migration in Rural Ecuador. Demography. Gronau I, Hubisz MJ, Gulko B, Danko CG, Siepel A 2011. Bayesian Inference of Ancient Human Demography from Individual Genome Sequences. Nature Genetics 43: 10311034. Grn R, Stringer CB 1991. Electron Spin Resonance Dating and the Evolution of Modern Humans. Archaeometry 33: 153199. Gutenkunst RN, Hernandez RD, Williamson SH, Bustamante CD 2009. Inferring the Joint Demographic History of Multiple Populations from Multidimensional Snp Frequency Data. PLoS Genet 5: e1000695. Hahn CHL, Vedder H, Fourie L. 1966. Native Tribes of South West Africa: Frank Cass and Company. Hamilton G 2005. Bayesi an Estimation of Recent Migration Rates after a Spatial Expansion. Genetics 170: 409417. Hamilton MJ, Buchanan B 2007. Spatial Gradients in Clovis Age Radiocarbon Dates across North America Suggest Rapid Colonization from the North. Proc Natl Acad Sci U S A 104: 1562515630. Hazelwood L, Steele J 2004. Spatial Dynamics of Human Dispersals. Journal of Archaeological Science 31: 669679. Hein J, Schierup MH, Wiuf C. 2005. Gene Genealogies, Variation and Evolution: A Primer in Coalescent Theory. New York: Oxford University Press. PAGE 152 152 Henn BM, Botigue LR, Gravel S, Wang W, Brisbin A, Byrnes JK, Fadhlaoui Zid K, Zalloua PA, MorenoEstrada A, Bertranpetit J, Bustamante CD, Comas D 2012. Genomic Ancestry of North Africans Supports Back to Africa Migrations. PLoS Genet 8: e1002397. Hey J 2005. On the Number of New World Founders: A Population Genetic Portrait of the Peopling of the Americas. PLOS Biology 3: 09650975. Hickerson MJ, Dolman G, Moritz C 2006. Comparative Phylogeographic Summary Statistics for Testin g Simultaneous Vicariance. Molecular Ecology 15: 209223. Holsinger KE, Weir BS 2009. Genetics in Geographically Structured Populations: Defining, Estimating and Interpreting Fst. Nature Reviews Genetics 10: 639650. Howell N, Smejkal CB, Mackey DA, Chin nery PF, Turnbull DM, Herrnstadt C 2003. The Pedigree Rate of Sequence Divergence in the Human Mitochondrial Genome. There Is a Difference between Phylogenetic and Pedigree Rates. Am J Hum Genet 72: 659670. Hudson RR 2002. Generating Samples under a Wrig ht Fisher Neutral Model of Genetic Variation Bioinformatics 18: 337338. Ingman M, Kaessmann H, Pbo S, Gyllensten U 2000. Mitochondrial Genome Variation and the Origin of Modern Humans. Nature 408: 708713. Jobling MA, Hurles ME, Tyler Smith C. 2004. H uman Evolutionary Genetics: Origins, Peoples & Disease. New York: Garland Science. Keinan A, Mullikin JC, Patterson N, Reich D 2008. Accelerated Genetic Drift on Chromosome X During the Human Dispersal out of Africa. Nature Genetics 41: 6670. Kirch P 19 80. Society Polynesian Prehistory: Cultural Adaptation in Island Ecosystems: Oceanic Islands Serve Asarchaeological Laboratories for Studying the Complex Dialectic between Human Populations and Their Environments. American Scientist 68: 3948. Kitchen A, Miyamoto MM, Mulligan CJ 2008. A ThreeStage Colonization Model for the Peopling of the Americas. PLoS One 3: e1596. PAGE 153 153 Klein RG. 1998. Why Anatomically Modern People Did Not Disperse from Africa 100,000 Years Ago. In: Akazawa T, Aoki K, Bar Yosef O, editor s. Neandertals and Modern Humans in Western Asia. New York: Plenum Press. p. 509521. Laval G, Excoffier L 2004. Bioinformatics 20: 2485 2487. Lee E 1966. A Theory of Migration. Demography 3: 4757. Lell JT, Wallace DC 2000. The Peopling of Europe from the Maternal and Paternal Perspectives. American Journal of Human Genetics 67: 1376 1381. Levy M 2010. ScaleFree Human Migration and the Geography of Social Networks. Physica A: Statistical Mechanics and its Applications 389: 49134917. Li JZ, Absher DM, Tang H, Southwick AM, Casto AM, Ramachandran S, Cann HM, Barsh GS, Feldman M, Cavalli Sforza LL, Myers RM 2008. Worldwide Human Relationships Inferred from GenomeWide Patterns of Variation. Science 319: 11001104. Liu H, Prugnolle F, Ma nica A, Balloux F 2006. A Geographically Explicit Genetic Model of Worldwide HumanSettlement History. Am J Hum Genet 79: 230237. Lohmueller KE, Bustamante CD, Clark AG 2009. Methods for Human Demographic Inference Using Haplotype Patterns from Genomewide Single Nucleotide Polymorphism Data. Genetics 182: 217231. Lohse K, Kelleher J 2009. Measuring the Degree of Starshape in Genealogies Summary Statistics and Demographic Inference. Genetics Research 91: 281. Luis JR, Rowold DJ, Regueiro M, Caeiro B, Cinnioglu C, Roseman C, Underhill PA, CavalliSforza LL, Herrera RJ 2004. The Levant Versus the Horn of Africa: Evidence for Bidirectional Corridors of Human Migrations. Am J Hum Genet 74: 532544. Maca Meyer N, Gonzlez AM, Pestano J, Flores C, Larruga J M, Cabrera VM 2003. Mitochondrial DNA Transit between West Asia and North Africa Inferred from U6 Phylogeography. BMC Genetics 4. PAGE 154 154 Macaulay V, Hill C, Achilli A, Rengo C, Clarke D, Meehan W, Blackburn J, Semino O, Scozzari R, Cruciani F, Taha A, Shaari NK, Raja JM, Ismail P, Zainuddin Z, Goodwin W, Bulbeck D, Bandelt HJ, Oppenheimer S, Torroni A, Richards M 2005. Single, Rapid Coastal Settlement of Asia Revealed by Analysis of Complete Mitochondrial Genomes. Science 308: 10341036. Malik K. 2013. Human Dev elopment Report 2013. In: United Nations Development Program. Marjoram P, Tavar S 2006. Modern Computational Approaches for Analysing Molecular Genetic Variation Data. Nature Reviews Genetics 7: 759770. Marks SJ, Levy H, Martinez Cadenas C, Montinaro F Capelli C 2012. Migration Distance Rather Than Migration Rate Explains Genetic Diversity in Human Patrilocal Groups. Mol Ecol 21: 49584969. Marlowe FW. 2010. The Hadza: Hunter Gatherers of Tanzania. Berkeley, CA: University of California Press. McDerm ott F 2004. PalaeoClimate Reconstruction from Stable Isotope Variations in Speleothems: A Review. Quaternary Science Reviews 23: 901918. McEvoy BP, Powell JE, Goddard ME, Visscher PM 2011. Human Population Dispersal "out of Africa" Estimated from Linkag e Disequilibrium and Allele Frequencies of Snps. Genome Res 21: 821829. Mellars P 2006. Going East: New Genetic and Archaeological Perspectives on the Modern Human Colonization of Eurasia. Science 313: 796800. Mercier N, Valladas H, Bar Yosef O, Vander meersch B, Stringer C, Joron JL 1993. Thermoluminescence Date for the Mousterian Burial Site of Es Skhul, Mt. Carmel. Journal of Archaeological Science 20: 169174. Metspalu M, Kivisild T, Metspalu E, Parik J, Hudjashov G, Kaldma K, Serk P, Karmin M, Behar DM, Gilbert MT, Endicott P, Mastana S, Papiha SS, Skorecki K, Torroni A, Villems R 2004. Most of the Extant Mtdna Boundaries in South and Southwest Asia Were Likely Shaped During the Initial Settlement of Eurasia by Anatomically Modern Humans. BMC Genet 5: 26. PAGE 155 155 Mielke J, Relethford JH, Eriksson A 1994. Temporal Trends in Migration in the Aland Islands: Effects of Population Size and Geographic Distance. Human Biology 66: 399410. Mir Herrans AT, Mulligan CJ 2013. Human Demographic Processes and Genetic Variation as Revealed by Mtdna Simulations. Mol Biol Evol 30: 244252. Morton N, Harris D, Yee S, Lew R 1971. Pingelap and Mokil Atolls: Migration. American Journal of Human Genetics 23: 339349. Nebel A, LandauTasseron E, Filon D, Oppenheim A, Faerman M 2002. Genetic Evidence for the Expansion of Arabian Tribes into the Southern Levant and North Africa. American Journal of Human Genetics 70: 1594 1596. Nei M. 1987. Molecular Evolutionary Genetics. New York: Columbia Univ. Press. Olivieri A, Achilli A, Pala M, Battaglia V, Fornarino S, Al Zahery N, Scozzari R, Cruciani F, Behar DM, Dugoujon JM, Coudray C, SantachiaraBenerecetti AS, Semino O, Bandelt HJ, Torroni A 2006. The Mtdna Legacy of the Levantine Early Upper Palaeolithic in Africa. Science 314: 1767 1770. Pagani L, Kivisild T, Tarekegn A, Ekong R, Plaster C, Gallego Romero I, Ayub Q, Mehdi SQ, Thomas MG, Luiselli D, Bekele E, Bradman N, Balding DJ, Tyler Smith C 2012. Ethiopian Genetic Diversity Reveals Linguistic Stratification and Complex Influences on the Ethiopian Gene Pool. Am J Hum Genet 91: 8396. Pakendorf B, Stoneking M 2005. Mitochondrial DNA and Human Evolution. Annu Rev Genomics Hum Genet 6: 165183. Pinhasi R, Fort J, Ammerman AJ 2005. Tracing the Origin and Spread of Agriculture in Europe. PLoS Biol 3: e410. Pinhasi R, Thomas MG, Hofreiter M, Currat M, Burger J 2012. The Genetic History of Europeans. Trends Genet 28: 496505. R Development Core Team. 2010. R: A Language and Environment for Statistical Computing. Version 2.11.1. Vi enna, Austria: R Development Core Team. PAGE 156 156 Raaum RL, Al Meeri A, Mulligan CJ 2013. Culture Modifies Expectations of Kinship and Sex Biased Dispersal Patterns: A Case Study of Patrilineality and Patrilocality in Tribal Yemen. Am J Phys Anthropol 150: 526538. Ramachandran S, Deshpande O, Roseman CC, Rosenberg NA, Feldman MW, Cavalli Sforza LL 2005. Support from the Relationship of Genetic and Geographic Distance in Human Populations for a Serial Founder Effect Originating in Africa. Proceedings of the National Academy of Sciences 102: 1594215947. Ramos Onsins SE, Mitchell Olds T 2007. Mlcoalsim: Multilocus Coalescent Simulations. Evolutionary Bioinformatics Online 3: 4144. Ramos Onsins SE, Rozas J 2002. Statistical Properties of New Neutrality Tests agains t Population Growth. Molecular Biology and Evolution 19: 20922100. Rasteiro R, Chikhi L 2013. Female and Male Perspectives on the Neolithic Transition in Europe: Clues from Ancient and Modern Genetic Data. PLoS One 8: e60944. Relethford J. 2001. Genetic s and the Search for Modern Human Origins. New York: Wiley Liss. Relethford JH, Harpending HC 1995. Ancient Differences in Population Size Can Mimic a Recent African Origin of Modern Humans. Current Anthropology 36: 667674. Rogers A 1995. Genetic Evidence for a Pleistocene Population Explosion. Evolution 49: 608615. Rose JI, Cerny V, Bayoumi R 2013. Tabula Rasa or Refugia? Using Genetic Data to Assess the Peopling of Arabia. Arabian archaeology and epigraphy 2013 95 101. Rowold DJ, Luis JR, Terreros M C, Herrera RJ 2007. Mitochondrial DNA Geneflow Indicates Preferred Usage of the Levant Corridor over the Horn of Africa Passageway. J Hum Genet 52: 436447. Salas A, Richards M, De la Fe T, Lareu M V, Sobrino B, Snchez Diz P, Macaulay V, Carracedo 2002 The Making of the African Mtdna Landscape. The American Journal of Human Genetics 71: 10821111. PAGE 157 157 Scholz CA, Johnson TC, Cohen AS, King JW, Peck JA, Overpeck JT, Talbot MR, Brown ET, Kalindekafe L, Amoako PY, Lyons RP, Shanahan TM, Castaneda IS, Heil CW, Forman SL, McHargue LR, Beuning KR, Gomez J, Pierson J 2007. East African Megadroughts between 135 and 75 Thousand Years Ago and Bearing on Early Modern Human Origins. Proc Natl Acad Sci U S A 104: 1641616421. Sefc KM, Payne RB, Sorenson MD 2007. Geneti c Differentiation after Founder Events: An Evaluation of Fst Estimators with Empirical and Simulated Data. Evolutionary Ecology Research 9: 2139. Skellam JG 1951. Random Dispersal in Theoretical Populations. Biometrika 38: 196218. Soares P, Alshamali F Pereira JB, Fernandes V, Silva NM, Afonso C, Costa MD, Musilova E, Macaulay V, Richards MB, Cerny V, Pereira L 2012. The Expansion of Mtdna Haplogroup L3 within and out of Africa. Mol Biol Evol 29: 915 927. Tajima F 1989. Statistical Method for Testing the Neutral Mutation Hypothesis by DNA Polymorphism. Genetics 123: 585595. Tavar S, Balding DJ, Griffiths RC, Donnlly P 1997. Inferring Coalescence Times from DNA Sequence Data. Genetics 145 505518. Tenesa A, Navarro P, Hayes BJ, Duffy DL, Clarke GM, Goddard ME, Visscher PM 2007. Recent Human Effective Population Size Estimated from Linkage Disequilibrium. Genome Research 17: 520526. Turney CSM, Kershaw AP, Moss P, Bird MI, Fifield LK, Cresswell RG, Santos GM, Di Tada ML, Hausladen PA, Zhou Y 2001. R edating the Onset of Burning at Lynch's Crater (North Queensland): Implications for Human Settlement in Australia. Journal of Quaternary Science 16: 767771. Valladas H, Reyss JL, Joron JL, Valladas G, Bar Yosef O, Vandermeersch B Thermoluminescence Dating of Mousterian TrotoCro Magnon'remains from Israel and the Origin of Modern Man. Nature 331: 614 616. Veeramah KR, Wegmann D, Woerner A, Mendez FL, Watkins JC, DestroBisol G, Soodyall H, Louie L, Hammer MF 2012. An Early Divergence of Khoesan Ancestors from Those of Other Modern Humans Is Supported by an Abc Based Analysis of Autosomal Resequencing Data. Molecular Biology and Evolution 29: 617630. PAGE 158 158 Watterson GA 1975. On the Number of Segregating Sites in Genetical Models without Recombination. Theoretical Population Biology 7: 256276. Wegmann D, Leuenberger C, Excoffier L 2009. Efficient Approximate Bayesian Computation Coupled with Markov Chain Monte Carlo without Likelihood. Genetics 182: 12071218. Weir B. 2007. A Tribal Order: Politics and Law in the Mountains of Yemen. Austin, TX: The University of Texas Press. Wijsman EM, CavalliSforza LL 1984. Migration and Genetic Popoulation Structure with Special Reference to Humans. Annual Review of Ecology and Systematics 15: 279301. Wood JW, Smouse PE, Long JC 1985. Sex Specific Dispersal Patterns in Two Human Populations of Highland New Guinea. The American Naturalist 125: 747768. Wright S 1951. The Genetical Structure of Populations. Annals of Eugenics 15: 323 354. PAGE 159 159 BIOGRAPHICAL SKETCH Aida T Mir graduated from Colegio Rosa Bell in Guaynabo, Puerto Rico in Spring 2001. She then attended the University of Puerto Rico, Rio Piedras from Fall 2001 to Fall 2006 and graduated in December 2006 with a Bachelor of Science degree in biology and a Bachelor of Science degree in anthropology. She interned at Walt Disney World during Spring 2007. She then began graduate school at the University of Florida in August 2007 and received a Doctor of Philosophy degree in genetics and genomics in August 2013. She began a postdoctorate fellowship at the University of Texas, Austin in January 2014. 