lable at ScienceDirect

Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184

Contents lists avai

Studies in History and Philosophy of Biological andBiomedical Sciences

journal homepage: www.elsevier .com/locate/shpsc

Biogeographical ancestry and race

Lisa GannettDepartment of Philosophy, Saint Mary’s University, 923 Robie Street, Halifax B3H 3C3, Canada

a r t i c l e i n f o

Article history:Available online 29 June 2014

Keywords:Ancestry-informative marker (AIM)Biogeographical ancestry (BGA)GeneticsGenomicsPopulation structureRace

E-mail address: lisa.gannett@smu.ca.

http://dx.doi.org/10.1016/j.shpsc.2014.05.0171369-8486/� 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

The use of racial and ethnic categories in biological and biomedical research is controversialdforexample, in the comparison of disease risk in different groups or as a means of making use of or con-trolling for population structure in the mapping of genes to chromosomes. Biogeographical ancestry(BGA) has been recommended as a more accurate and appropriate category. BGA is a product of thecollaboration between biological anthropologist Mark Shriver from Pennsylvania State University andmolecular biologist Tony Frudakis from the now-defunct biotechnology start-up company DNAPrintgenomics, Inc. Shriver and Frudakis portray BGA as a measure of the ‘biological’, ‘genetic’, ‘natural’, and‘objective’ components of race and ethnicity, what philosophers of science would call a natural kind. Thispaper argues that BGA is not a natural kind that escapes social and political connotations of race andethnicity, as Shriver and Frudakis and other proponents believe, but a construction that is built uponracedas race has been socially constructed in the European scientific and philosophical traditions. Morespecifically, BGA is not a global category of biological and anthropological classification but a localcategory shaped by the U.S. context of its production, especially the forensic aim of being able to predictthe race or ethnicity of an unknown suspect based on DNA found at the crime scene. Therefore, cautionneeds to be exercised in the embrace of BGA as an alternative to the use of racial and ethnic categories inbiological and biomedical research.

� 2014 Elsevier Ltd. All rights reserved.

When citing this paper, please use the full journal title Studies in History and Philosophy of Biological and Biomedical Sciences

1. Introduction: categories of race and ethnicity in biologicaland biomedical research

Completion of the Human Genome Project (HGP), and theconsequent shift of resources to investigating patterns of variabilityacross genomes that are of possible biological and biomedical in-terest, have contributed to an increased use of categories of raceand ethnicity to study group-based genetic and genomicdifferences.

‘Single-gene’ diseases that appear at different frequencies indifferent racial and ethnic groups are familiar to us: for example, inthe United States, sickle-cell anaemia is more common in AfricanAmericans and cystic fibrosis is more common in EuropeanAmericans. Mutations implicated in these relatively rare diseaseshave been mapped to locations on the chromosomes. The focus is

now on mapping genes that are involved in more common ‘com-plex’ diseases such as hypertension, asthma, cancer, and dementia,but the task has proved more challenging than the rhetoric sellingthe benefits of the HGP promised, and than geneticists themselvesenvisioned. In complex diseases, multiple genetic, epigenetic, andenvironmental causal factors need to be identified and their sepa-rate roles and interaction effects unravelled. Some complex dis-eases occur at increased frequencies in particular racial and ethnicgroups: for example, hypertension in African Americans and dia-betes in Native Americans. Although they recognize the importanceof epigenetic and environmental factors and discount geneticdeterminism, many biomedical geneticists believe that geneticvariants found at different frequencies in different racial and ethnicgroups are implicated in complex diseases, just as in single-genediseases.

The use of racial and ethnic categories in biological andbiomedical research is controversial, however. In the United States,for example, criticisms are directed against the double duty that is

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184174

asked of the Office of Management and Budget (OMB) categories ofrace and ethnicity. The OMB regulates the collection of racial andethnic data by federal agencies. When OMB Directive 15 was issuedin 1977, its aim was to standardize the racial and ethnic categoriesused to collect data so that compliance with recently passed civilrights legislation that prohibited discrimination in areas likehousing, education, and employment could bemonitored. The OMBsystem of classification, as it was updated most recently in 1997,asks individuals to self-identify, first, by ‘ethnicity’ (their choicesare ‘Hispanic or Latino’ and ‘Not Hispanic or Latino’) and, second, by‘race’ (their choices are ‘American Indian or Alaska Native’, ‘Asian’,‘Black or African American’, ‘Native Hawaiian or Other PacificIslander’, and ‘White’, and more than one racial class can be cho-sen). According to the OMB (1997), ‘The categories represent asocial-political construct designed for collecting data on the raceand ethnicity of broad population groups in this country, and arenot anthropologically or scientifically based’ (‘Supplementary In-formation’, Sec. A). Consistent with this mission is use of the OMBcategories in clinical research where the variables of interest aresocial and related to the history of racial and ethnic discriminationin the United States: for example, access to health care, exposure toracism, income level, educational attainment, proximity to envi-ronmental toxicity, etc. But, as critics point out, use of the OMBcategories in biologically-based clinical research goes beyond this,and risks suggesting that ‘there are fundamental biological andbehavioral differences among racial groups’ (e.g., Institute ofMedicine, 1999, p. 82).

Given existing controversy concerning the use of racial andethnic categories in biological and biomedical research, theircooption in service of the genetic mapping of complex traits hasbeen a topic of debate. Mapping genes implicated in complex traitsis amore complicated endeavour thanmapping genes implicated insingle-gene or Mendelian traits. An early strategy extended thelinkage mapping used for Mendelian diseases that segregate infamilies to populations supposed to be relatively homogeneous,such as the Icelandic population. In more heterogeneous pop-ulations, different strategies are required, several of which haveintroduced the expertise of population geneticists, evolutionarygeneticists, and anthropological geneticists into biomedical ge-netics. One such strategy is admixture mapping. Admixture map-ping (AM) makes use of linkage in a different way: specifically, thelinkage disequilibrium that arises when historically separatedpopulations combine. Correlations are expected in the ‘admixedpopulation’ between incidence of the disease and regions of thegenome derived from the ancestral population inwhich the diseaseis most prevalent. Another strategy is made possible by the abilityto screen for many mutations at the same time. Genome-wide as-sociation studies (GWAS) use large numbers of individuals dividedinto cases and controls to seek correlations between disease inci-dence and SNP (single nucleotide polymorphism) markers orhaplotypes (characteristic patterns of neighbouring markers)across the genome.

Progress has been slow using AM and GWAS to map genesassociated with complex traits, and ‘population structure’ has bothhelped and hindered. ‘Population structure’ refers to the patternby which genetic variability is distributed across a species orsubpopulation of a species. Genetic variability is patterned in waysthat reflect the past and present operation of factors of evolu-tionary importance such as selection, drift, migration, and matingstructure (random, assortative, etc.). Population genetic structureis a help in mapping genes associated with complex traits insofaras it is the very basis of the linkage disequilibrium used in AM.Population genetic structure is a hindrance in both AM and GWASinsofar as it functions as a confounding variable when ‘false pos-itives’ are yielded by genetic variants that are more frequent in the

population in which the trait of interest is more frequent but playno causal role. Racial and ethnic categories are employed in thedelineation of ancestral populations in AM and in order to controlfor population structure in AM and GWAS. Scientists recognizethat doing so is problematic. They realize that it is difficult toextricate race and ethnicity from their social and political conno-tations when these categories are used in biological andbiomedical research. They are also concerned about whether racialand ethnic self-identification by subjects is appropriate when thegoals are scientific. Consequently, there have been efforts to comeup with alternative group concepts that gain in scientific objec-tivity by ostensibly foregoing the social and political connotationsof race and ethnicity.

In this paper, I focus on the newly invented concept ‘biogeo-graphical ancestry’. Biogeographical ancestry has met with apositive reception by knowledgeable critics and policy-makers. Acritical examination of the use of race as a category of classifica-tion in the mapping of complex traits by a multidisciplinary groupof U.S. biological and social scientists accepts that biogeographicalancestry provides an objective method of controlling for popula-tion structure (Shields et al., 2005). The authors consider self-identified race (i.e., the OMB classification) to be the appropriatecategory to use for monitoring health disparities and an acceptablecategory to use for recruitment for genetic studies. But theyrecommend that genetic studies themselves assess populationstructure empirically by genotyping individuals to determine their‘continental ancestry’ proportions. Because of the social harmsthat attach to using race as a variable in genetic research, theyurge the National Institutes of Health (NIH) to provide financialsupport for developing and facilitating the use of such tools. Citingthis work, a 2005 review article by the National Human GenomeResearch Institute’s (NHGRI’s) Race, Ethnicity, and GeneticsWorking Group, titled ‘The Use of Racial, Ethnic, and AncestralCategories in Human Genetics Research’, recognizes that the U.S.census categories were not designed for genetic research andsuggests that instead, ‘labels based on biogeographical ancestrymay be suited for many genetics studies’ (p. 526), thus providing ameans of moving beyond self-identified race, ethnicity, andancestry as proxies.

I am more sceptical than these critics and policy-makers aboutbiogeographical ancestry as an alternative to the use of categoriesof race and ethnicity in biological and biomedical research.Biogeographical ancestry emerged as a product of the collaborationbetween biological anthropologist Mark Shriver and members ofhis Pennsylvania State University laboratory and molecular biolo-gist Tony Frudakis’s now-defunct biotechnology start-up companyDNAPrint genomics. We will see that the research context sur-rounding the emergence of the concepts and techniques associatedwith biogeographical ancestry was shaped by diverse inter-estsdsocial and commercial as well as scientificdin DNA forensics,gene mapping, pharmaceutical development, and direct-to-consumer genealogy testing. Close attention to this researchcontext reveals that BGA is itself a construction built upon race asrace has been socially constructed in the European scientific andphilosophical traditions, but especially in the United States. As such,BGA does not provide a means of moving beyond ‘proxy’ socialcategories like race, ethnicity, and ancestry, as the NHGRI’s Race,Ethnicity, and Genetic Working Group assumes, and BGA does notprovide a wholly empirical method of assessing population struc-ture, as is the hope of the multidisciplinary group of critics.

In the next section, I outline the steps leading to the invention ofbiogeographical ancestry through the collaborative efforts ofShriver and Frudakis. To say that BGA is an invention is toemphasize not only its invention as a concept but as a technologyfor which patent protection was sought.

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184 175

2. Inventing biogeographical ancestry

The term ‘biogeographical ancestry’ was introduced in 2000, ina poster presentation by Carrie Pfaff, Esteban Parra, and MarkShriver at the meeting of the American Society of Human Genetics(at the time, Pfaff was a doctoral student and Parra was a post-doctoral researcher in Shriver’s lab). The ASHG abstract notes that‘Ethnicity is comprised of both biological and cultural components’and defines ‘biogeographical ancestry’, abbreviated as ‘BGA’, as ‘thecomponent of ethnicity that is biologically determined and can beestimated using genetic markers that have distinctive allele fre-quencies for the populations in question (referred to as population-associated alleles PAAs)’ (Pfaff, Parra, & Shriver, 2000). The authorsnote that BGA as a measure of ‘individual admixture’ or ‘propor-tional ancestry’ would be useful in ‘the identification of unknownsamples in forensic cases’ (Pfaff et al., 2000). BGA was also envi-sioned as being useful for mapping genes to chromosomes: anaccompanying poster from the Shriver lab (Parra et al., 2000)concerns use of a panel of 10 ‘ancestry informative markers’ tomeasure admixture in African Americans living in different parts ofthe United States with a view to carrying out admixture mapping inAfrican American populations. These poster presentations were thecontinuation of previous forensic research conducted by Shriver asa postdoctoral fellow in the Department of Human Genetics atUniversity of Pittsburgh and funded by the National Institute ofJustice, which sought to identify the ‘ethnicity’ of unknown per-petrators of crimes based on DNA left at the scene. In this research,Shriver et al. (1997) isolated ‘population-specific alleles’, defined as‘genetic markers with large allele-frequency differentials’, specif-ically, for biallelic markers, a frequency differential more than 50percent between ‘any two major geographically or ethnicallydefined populations’ (pp. 957e958). ‘European Americans’, ‘AfricanAmericans’, and ‘Hispanic Americans’ were the populations tar-geted for the reason that ‘together they constitute 95% of U.S. res-idents,’ but the authors note that similar sets of markers foridentifying ‘other populations common in the United States, such asChinese Americans, Native Americans, and Polynesian Americans’could be produced (p. 963).

Tony Frudakis, at DNAPrint genomics, shared Shriver’s interestin DNA forensics. DNAPrint’s report to the Securities ExchangeCommission (SEC) for 2001 (filed April 9, 2002) provides infor-mation on several forensic products under development: Retinome,a ‘classifier’ panel of single nucleotide polymorphisms (SNPs) forthe prediction of eye colour shade; Retinome HA, a ‘classifier’ panelof SNPs for the prediction of hair colour; and Verity, a ‘classifier’panel of SNPs for ‘the inference of race from DNA’. For Verity,DNAPrint planned ‘to tap into specific regions of the genome thatare especially informative for racial determination’ in order toisolate SNPs that ‘encode’ physical traits such as skin colour, eyecolour, and hair colour (DNAPrint, 2002a, pp. 6e7). In the patentapplication he filed on May 28, 2002, titled ‘Compositions andmethods for detecting polymorphisms associated with pigmenta-tion’, Frudakis credits Shriver et al. (1997) for first describing ‘howSTR [single tandem repeat] markers could be used for racialprofiling’ (Frudakis, 2002, Sec. 0805). Frudakis promotes his owninvention for racial profiling (called Verity in the SEC report) asdeparting from the STR markers being used in DNA forensics toprovide the ‘first SNP base [sic] method for reliably distinguishingbetween the world’s major racial groups’ (Frudakis, 2002, Sec.0811). Frudakis describes the ‘classifier’ as a panel of 60 ‘race-related SNPs’ or ‘significant markers of race’ found in pigmentationand xenobiotic metabolism genes with different frequencies be-tween ‘racial classes’dspecifically, ‘African’, ‘Caucasian’, and ‘Asian’(Frudakis, 2002, Sec. 0247 & 0805). While the attention to genesinvolved in the metabolism of drugs and other substances foreign

to the body reflects DNAPrint’s interest in pharmacogenomics,Frudakis emphasizes the forensic importance of being able to inferrace on the basis of correlation with markers for ‘differentpigmentation phenotypes’di.e., eye and hair colour and shade. Hesuggests several ways in which the ‘classifier’ could be used in fo-rensics: to direct the choice of the appropriate reference data-base(s) to calculate the probability that DNA found at a crime sceneis a match to that of a suspect in custody or on trial, to guidecriminal investigations by predicting race and other physicalcharacteristics for unknown suspects based on crime-scene DNA,and to provide just cause for obtaining DNA from suspects who‘cannot be racially excluded’.

Shriver signed a consulting contract with DNAPrint dated June12, 2002. The contract recognizes Shriver as the ‘inventor’ of atechnique that allows ‘ancestral proportions’ to be estimated,which was useful for the company because its ‘classifier’ could beused to infer ‘major ancestral affiliation’ only (Frudakis et al., 2003,p. 777). By this time, Shriver had switched from STRs to SNPs, andthe markers called ‘population-specific alleles’ or ‘PSAs’ by Shriveret al. in 1997 and ‘population-associated alleles’ or ‘PAAs’ by Pfaffet al. in 2000 had been renamed ‘ancestry informative markers’ or‘AIMs’. The project outlined in the contract was to create a panelcontaining a ‘minimum and optimum set’ of ‘Ancestry InformativeMarkers (AIMs)’ that would support development of a ‘kit productthat could be used to infer Ancestry Admixture Ratios in individualhuman beings’ for ‘target ancestral groups’ (possible groupsmentioned include those of ‘African, African-American, European,European-American, Native American, Hispanic, East Asian, andSouth Asian heritage’). Shriver was to receive compensation in cashand shares for his services as both a consultant and a member ofDNAPrint’s board of directors and based on net sales of the productsdeveloped (DNAPrint, 2002b).

Scientific validation for the BGAeAIMs technology followed theJune 2002 contract. With Shriver’s assistance, biomedical re-searchers in several other laboratories used the technology. Onegroup sought to ‘explain’ racial differences in insulin sensitivityamong African-American and European-American children byappealing to ‘genetic admixture’ as determined by 20 AIMs thatShriver’s team had identified on the basis of frequency differentialsbetween ‘European and African parental populations’ (Gower,Fernández, Beasley, Shriver, & Goran, 2003, p. 1048). Anothergroup used 31 AIMs from DNAPrint’s pilot test, ANCESTRY 1.0,which proportioned BGA among ‘Native American, West African,and European’, to confirm that a family self-identifying as ‘white’,but possessing a genetic variant associated with cardiac arrhythmiaand sudden death previously found only in African Americans, wasindeed of ‘European descent’ (Chen et al., 2002, p. 913). Articles bymembers of the Shriver laboratory that were published in peer-reviewed journals during 2003 prominently display and offer def-initions for ‘biogeographical ancestry’ and ‘ancestry informativemarkers’ in titles, abstracts, and keywords (Halder & Shriver, 2003;Shriver et al., 2003), and extol the ‘immense potential’ of admixturemapping (Halder & Shriver, 2003, p. 60). Admixture mapping’spotential was demonstrated using ANCESTRY 1.0’s panel of 31e34AIMs with skin pigmentation providing a model phenotype(Shriver et al., 2003). Federal funding was received for these studiesfrom the National Human Genome Research Institute, NationalCenter for Research Resources, National Institute of Mental Health,and National Institute of Diabetes and Digestive and KidneyDiseases.

Commercial developments proceeded alongside scientific ones.On September 19, 2002, DNAPrint announced plans to add ‘rec-reational genomics’ to its forensic and pharmacogenomic interests,with the goal of raising cash for the company, which had lost $5.46million since late 1998 (Mulle, 2002). For the genealogically

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184176

curious, at an initial cost of $160 and planned increase to $290, theproduct ANCESTRYbyDNA� 2.0 would determine a person’s ‘racialmix’ using DNA obtained from a mailed-in cheek swab (Wade,2002). Specifically, when introduced in December 2002, ANCES-TRYbyDNA� 2.0’s panel of 75 markers assigned customers to oneof four ‘major bio-geographical ancestry groups’ (‘East Asian’,‘European’ or ‘IndoEuropean’, ‘Native American’, and ‘sub-SaharanAfrican’), and estimated the proportions that other groups hadcontributed to their ancestry. In a Sarasota Herald-Tribune article,Shriver seems to embrace wholeheartedly the new direction takenby DNAPrint: ‘I don’t think anybody else has even used the phrase“recreational genomics”. I see (research and development)extending into physical features and personality traits. There’sgoing to be a blurry line between medical information and rec-reational information’ (Mulle, 2002, p. 8D). DNAWitness� 2.0, theforensic counterpart to the direct-to-consumer genealogy testANCESTRYbyDNA� 2.0, accompanied by a digital photo databaseof individuals with specific ancestral proportions, became avail-able in May 2003 at a cost of $1000. On June 5, 2003, DNAPrintissued a press release that made an African-American male theface of its advertising campaign for DNAWitness� (DNAPrint,2003). Police in Baton Rouge, Louisiana had failed to find theperpetrator of a series of rapes/murders that occurred from 2001to 2003. A Federal Bureau of Investigation (FBI) psychologicalprofile and an eyewitness report led police to suspect a Whiteman, 25e35 years of age. Over 1000 men fitting the profile wereasked to provide DNA samples to eliminate themselves as sus-pects. In March 2003, after volunteering its services, DNAPrintestimated the suspect’s ancestral origins to be 85 percent ‘sub-Saharan African’ and 15 percent ‘Native American’, and policewidened their search. In May 2003, a 34 year-old African-Amer-ican man named Derrick Todd Lee was arrested and charged. Bypublicizing its involvement in the case, DNAPrint hoped to attractthe attention of investors and other law enforcement agencies;subsequently, in 2006, in order to promote sales of its forensicoffering, the company published a ‘Forensic Flyer’ that included‘The Louisiana Serial Killer Case’ as a ‘case study’ (Fig. 1).

Having achieved scientific validation of the BGAeAIMs tech-nology and brought genealogical and forensic products to themarket, Frudakis and Shriver sought to protect their investment. InAugust 2003, they submitted the patent application ‘Compositionsand Methods for Inferring Ancestry’, which claims as inventionspanels of 32, 71, and 331 ‘Ancestry Informative Markers’ that pre-dict ‘BioGeographical Ancestry’di.e., a person’s ancestral pro-portions of ‘East Asian’, ‘IndoEuropean’, ‘Native American’, and‘sub-Saharan African’. Here and elsewhere, the Shriver lab andDNAPrint portray biogeographical ancestry as a means ofconstruing race and ethnicity in appropriately scientificd‘bio-logical’, ‘genetic’, ‘natural’, ‘heritable’, or ‘objective’dways. As Fru-dakis’s original name ‘Verity’ implies, there is presumed to be atruth about race. Recall the definition of biogeographical ancestryin Pfaff et al.’s 2000 poster abstract: ‘Ethnicity is comprised of bothbiological and cultural components. Biogeographical ancestry(BGA) refers to the component of ethnicity that is biologicallydetermined and can be estimated using genetic markers that havedistinctive allele frequencies for the populations in question.’ TheFrudakiseShriver patent application elaborates further:

BioGeographical Ancestry (BGA) is the heritable component ofrace. Because socio-cultural and geo-political metrics formeasuring human race are human, not natural, constructs, theiruse in genetics research makes it difficult to control for popu-lation genetic structure, and may obscure important correla-tions between BGA and human biology. (Frudakis & Shriver,2003, Sec. 0233).

The same document characterizes BGA as a basic representation ofhuman population structure: ‘On a basic level, human populationstructure can be represented in terms of BioGeographical Ancestry(BGA), which is the heritable component of “race” or heritage, andwhich is relevant on any scale of resolution’ (Frudakis & Shriver,2003, Sec. 0007). Responses to FAQs for ANCESTRYbyDNA�(DNAPrint website, retrieved 4 July 2007) include the definition ofbiogeographical ancestry as a ‘term given to the biological or ge-netic component of race’ and ‘a simple and objective description ofthe Ancestral origins of a person, in terms of the major populationgroups’. Similarly, DNAPrint’s 2006 ‘Forensic Flyer’ promotesDNAWitness� as providing law enforcement with ‘a simple andobjective description of your subject’s ancestral origins’ that elim-inates the need to rely on subjective psychological profiles oreyewitness accounts (Fig. 1).

In the next two sections, I argue that BGA is not, as its pro-ponents claim, a ‘biologically determined component of ethnicity’,a ‘heritable’, ‘biological’, or ‘genetic component of race’, and an‘objective description of ancestral origins’ that provides a globalcategory of biological and anthropological classification. BGA doesnot escape social and political connotations of race and ethnicitybut is itself a construction that is built upon race as race has beensocially constructed, not only in the European scientific and phil-osophical traditions but also in the United States.

3. Biogeographical ancestry as a global category of biologicaland anthropological classification

In this section, I show how BGA’s inventors and merchandisersre-enact ‘race’ as it has been traditionally conceived and con-structed as a global category of biological and anthropologicalclassification within the European scientific and philosophical tra-ditions, what philosophers often call a natural kind (Gannett, 2010).Traditional conceptions and constructions of race associated with18th-century natural history, 19th-century racial biology, and early20th-century physical anthropology tend to be essentialist, taxo-nomic, and typological (in the senses specified below). BGA in-corporates similar assumptions, and hence, just as race fails as aglobal category of biological and anthropological classification, sodoes BGA.

Racial classification has traditionally been carried out using anessentialist approach whereby individuals are assigned to racialclasses on the basis of properties that are considered to be essentialto their membership in that class. Such necessary and/or sufficientproperties, or racial ‘essences’, are usually intrinsic and may bephenotypic and/or genetic. Frudakis takes this essentialistapproach to BGA: recall that in the patent application he filed inMay 2002, just before his consulting agreement with Shriver, Fru-dakis identifies ‘Asian, African American, and Caucasian’ as ‘racialclasses’ that differ in their frequencies of various ‘race-relatedSNPs’, which he assumes to be ‘significant markers of race’ becausethey lie in functional genes implicated in the development ofphenotypic traits that are widely regarded as ‘racial’di.e., differ-ences in skin colour (pigmentation genes) and differences in drugmetabolism (xenobiotic metabolism genes). The invention, char-acterized by Frudakis as a ‘racial profiling tool’, involves the use ofone SNP or a combination of SNPs to ‘allow an inference to bedrawn as to a genetic pigmentation trait such as hair shade, haircolor, eye shade, or eye color, and further allow an inference to bedrawn as to race’ (Frudakis, 2002, Sec. 0009)din forensics, forexample, ‘an inference . of a human Caucasian having a particulareye color or eye shade’ based on crime-scene DNA (Frudakis, 2002,Sec. 0015). As the language ‘and further allow’ suggests, Frudakisdoes not treat race and eye/hair pigmentation as equivalent: hedistinguishes ‘race-related SNPs’ and ‘race-related genes’ from

Fig. 1. ‘Forensic Flyer’, 2006, retrieved from DNAPrint genomics, Inc.’s website, 17 July 2007.(Permission to reproduce the image could not be obtained because the copyright holder, DNAPrint genomics, no longer exists. The image is reproduced under provisions of ‘fair dealing’ for purposes of research, criticism, and review.)

L.Gannett

/Studies

inHistory

andPhilosophy

ofBiological

andBiom

edicalSciences

47(2014)

173e184

177

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184178

‘pigmentation-related SNPs’ and ‘pigmentation-related genes’. Butthey are closely connected, with ‘racial classes’ used to identify‘pigmentation-related SNPs’, and pigmentation classes used toidentify ‘race-related SNPs’. To identify a ‘race-related SNP’, onebegins with a candidate SNP of a pigmentation or xenobioticmetabolism gene and finds out whether ‘genotype distributionsand allele frequencies between racial classes’ differ significantly.SNP alleles are considered to be ‘racially informative’ if they are‘preferentially present’ in ‘the Caucasian population’, ‘the AfricanAmerican population’, or ‘the Asian population’. The ‘genetic clas-sifier for racial inference’ included in Frudakis’s patent applicationcomprised 60 such SNPs (Frudakis, 2002, Sec. 0786).

Frudakis’s racial essentialism has long been discounted withinbiology, however. Since the modern evolutionary synthesis of the1930se40s, evolutionary biologists and population geneticists haverejected essentialism for species and races alikedit is relationsamong organisms (genealogy, breeding propensity, competition,etc.), not intrinsic properties, that determine inclusion in thesegroups, and there is no longer any expectation that each and everymember of the group shares a species- or race-specific character-istic, whether genetic or phenotypic. Shriver, as a biological an-thropologist, appears more versed in the theoretical perspective ofpopulation genetics than does Frudakis, as a molecular biologist.Shriver conceives of biological race in terms of statistical differ-ences in allele frequencies among geographically distinct pop-ulations regardless of the functional effects of the alleles onphenotype.1 Ultimately, the product resulting from the FrudakiseShriver collaboration that was released to the forensic market inMay 2003 was not Frudakis’s ‘racial classifier’ but a forensicsversion of ANCESTRYbyDNA� 2.0, which used a panel of 71 SNPs todetermine ‘proportional ancestries’ of ‘East Asian’, ‘IndoEuropean’,‘Native American’, and ‘sub-Saharan African’. Of the 71 SNPs, only afew overlapped with the 60 SNPs in pigmentation and xenobioticmetabolism genes that Frudakis had included in his ‘classifier’: i.e.,two SNPs in OCA2 (the oculocutaneous albinism II gene that codesfor a protein located in melanocytes), one SNP in HMGCR (a genefor an enzyme involved in cholesterol synthesis), and one SNP inCYP3A4 (a gene for an enzyme involved in drug metabolism andthe synthesis of cholesterol, steroids, and other lipids). Theremaining SNPs were located in noncoding regions of genes (in-trons) and therefore of no known functional significance, asidefrom the Duffy blood group FY-null allele, which was includedbecause it is seldom found outside African populations.

Racial taxonomies, although traditionally based in essentialism,need not be. Construing race as a taxonomic concept involves twobasic assumptions (Gannett, 2010). One assumption is that all in-dividuals belonging to the relevant domain (in this case, Homosapiens) can be sorted into racial groupsdi.e., the category isexhaustive. The other assumption is that the category’s classes aremutually exclusivedi.e., individuals are definitively sorted intoeither unique classes or multiple classes hierarchically arranged.The FrudakiseShriver patent application’s characterization of BGAas a ‘basic’ representation of ‘human population structure’ thatcaptures ‘the heritable component of “race”’ is, like traditionalconcepts of race, taxonomic. The application reads as follows:

1 By the time Molecular Photofitting was published in 2008, Frudakis appears tohave incorporated more of a population-genetic approach, in the preface referringto ‘panels of good AIMsdnot those found in genes, but bonafide AIMs of neutralevolutionary character’ (p. xiii) and characterizing the theoretical background to thebook as represented by such influential works as Cavalli-Sforza and Bodmer’s TheGenetics of Human Populations, Balding et al.’s The Handbook of Statistical Genetics,and publications of Chakraborty’s (p. xiv).

On a basic level, human population structure can be representedin terms of BioGeographical Ancestry (BGA), which is the heri-table component of ‘race’ or heritage, and which is relevant onany scale of resolution. For example, on a crude level, BGA can bedetermined for 2 groups (e.g., European vs. others); or on a finelevel, e.g., it can refer to ‘race’ in terms of 4 groups such asIndoEuropeans, East Asians, sub-Saharan African and NativeAmerican; or on a finer level, e.g., it can refer to ethnicity withinthe European group (for example, Mediterranean or Scandina-vian); or on a still finer level, e.g., it can even refer to groups offamilies within ethnic groups, such as groups of O’Reilly’s [sic]descendent from a set of common ancestors within the Irishgroup. (Frudakis & Shriver, 2003, Sec. 0007)

What Frudakis and Shriver’s mention of ‘any scale of resolution’refers to is a hierarchical, or tree-like, taxonomy of mutuallyexclusive classes in which family groups are embedded withinethnic groups, and ethnic groups are embedded within continentalgroups. At the first level of resolution, there is the ‘IndoEuropean’continental group; at the second level of resolution, there are Eu-ropean ethnic groups such as the Irish; and at the third level ofresolution, there are Irish families such as the O’Reillys.

Consistent with the approach brought about by the modernevolutionary synthesis, whereby relations among organismsdnotintrinsic propertiesddetermine inclusion in species and races, it isgenealogical ties that determine BGA for individuals. However, theexample of the European/Irish O’Reillys points to the challengesfaced in determining individual membership in the hierarchicaltaxonomy of ancestrally-defined BGA groups, from the level offamily, to ethnic, to racial group. Who counts as an O’Reilly? Thosewith the O’Reilly surname, whether gained by marriage, by adop-tion, by birth? Those without the O’Reilly surname, since lost bymarriage, by adoption, by illegitimacy? And who is only anO’Reilly? Traditionally, by convention, the O’Reilly surname travelswith the Y-chromosome along the paternal line. And so, other thanby marriage, the name O’Reilly is likely to grace only one of twoparents, one of four grandparents, one of eight great-grandparents,one of sixteen great-great-grandparents, and so forthdthough atsome point, surnames begin to be shared among ancestors of thesame generation. Based on genealogy alone, those with surnameslike Braun, Patel, Wong, etc. who share a great-great-grandfatherwith someone named O’Reilly are no less O’Reillys. The indeter-minacy associated with who counts as a member of the O’Reillyfamily and BGA’s failure to define mutually exclusive classes at thelevel of ‘family’ occurs at each of the higher levels for similar rea-sons: at the level of ‘ethnicity’ and who counts as Irish, or Polish, orMacedonian, etc. and at the level of ‘race’ and who counts as Eu-ropean, or Asian, etc.

While racial taxonomies make an allowance for a relativelysmall number of mixed-race individuals in whose cases theexception serves to prove the rule, as mixed-raced numbersmultiply, taxonomies tend to be modified in response, becomingdefined by hypodescent (e.g., the ‘one-drop’ rule in the UnitedStates) or a proliferation of classes (e.g., blood quantum rules).2 BGAtakes an approach similar to blood quantum. In the news media,countering the impression that the DNAPrint tests involve cate-gorical thinking about race, Shriver argues just the opposite: ‘By

2 Hypodescent assigns persons of ‘mixed race’ to the socio-economically subor-dinate group. The one-drop rule considers persons Black if they have any knownBlack ancestors (a single drop of ‘black blood’). Blood quantum uses fractional‘blood’ (i.e., ancestral) proportions: e.g., in the United States, one-quarter NativeAmerican ancestry is required to receive most government benefits. Blood quantummay support the proliferation of racial classes: e.g., ‘white’, ‘black’, ‘mulatto’,‘quadroon’, and ‘octoroon’ on the 1890 U.S. census.

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184 179

showing the continuum of genetic variation among people, our testdispels race as a scientific way of categorizing people’ (HawkinsSimons, 2003). Shriver is not denying that the category of BGA isexhaustive: he and his colleagues assume that everyone (i.e., allexisting members of Homo sapiens) can be classified by BGA.Shriver is instead claiming that the four ‘major’ BGA groupsd‘EastAsian’, ‘IndoEuropean’, ‘Native American’, and ‘sub-Saharan Afri-can’dare notmutually exclusive: individuals belong to one or moreof these in determinate proportions, and hence, a ‘continuum ofgenetic variation’ is exhibited. It is important to emphasize thatBGA is not intended to portray human population structure inpresent-day, wholly geographical terms; rather, BGA assigns in-dividuals in determinate proportions to geographically definedancestral groups. The ‘User Manual’ DNAPrint’s genealogical cus-tomers received with their ANCESTRYbyDNA� test results char-acterizes the four ‘major’ BGA groups as ‘founder groups’ that aroseaftermigration out of Africa and gave rise to lineages that terminatein ‘present-day Europeans, Native Americans, Africans, and EastAsians’:

Human beings migrated out of central, sub-Saharan Africa some200,000 years ago to inhabit various regions of our globe. Thesemigrants established founder groups that gave rise to present-day Europeans, Native Americans, Africans, and East Asians..If your heritage has been derived from more than one of thesegroups the test results tell you what your mixture ratios are.(DNAPrint website, retrieved 4 July 2007)

The implication is that each of us, no matter where in the world welive, is in some sense either an unmixed descendant of one of thefour ‘founder groups’ or a mixed descendant of two or more of thefour ‘founder groups’ defined by migrations within and out ofAfrica.

However, for BGA classification to be exhaustive, there can onlyhave been these four ‘founder groups’, for were there five or six ortwenty or fifty such groups, some present-day individuals mighthave ancestry from one of them (unless, of course, these all gaverise to lineages that eventually went extinct, without mixing withany of the four BGA ‘founder groups’). The ability to assign present-day individuals determinate BGA proportions assumes that at somepoint in the past, these four ‘founder groups’ or ‘ancestral line-ages’d‘East Asian’, ‘IndoEuropean’, ‘Native American’, and ‘Sub-Saharan African’dwere well-circumscribed and bounded entities,separate and unmixed with one another, into which each and everymember of Homo sapiens living at that time belonged fullydthat is,if only in the past, BGA classes were once mutually exclusive. Butwhen might this have been? From ‘a genomic perspective’, SvantePääbo (2003) reminds us, ‘we are all Africans, either living in Af-rica or in quite recent exile outside Africa’ (p. 410). What reasons dowe have for privileging any particular time slice of an evolutionaryhistory that took place mostly in Africa? Andwhenever did humanslive only in the Americas, East Asia, Europe, and sub-Saharan Africa,but not in northern Africa, the Middle East, central Asia, orAustralia?3

BGA shares with blood quantum the need to constitute pre-existing racial wholes in order to attach determinate fractional-race values to present-day individuals. The inherent logic of thisenterprise makes types out of these pre-existing racial wholes:types that are primordial, pure, static, and/or ideal, and bring tomind the ‘typological thinking’ that the modern evolutionary syn-thesis and rise of population genetics during the 1930se40s weresupposed to have eliminated (Gannett, 2001). In 1951, in the third

3 This discussion assumes a predominantly branching view of human evolu-tionary history, but many biologists discount this.

edition of Genetics and the Origin of Species, Theodosius Dobzhan-sky, so-called architect of the synthesis and founder of populationgenetics, criticized typological thinking that treats races as typesrather than as ‘living populations’:

Platonic philosophy considers the elusively multiform, alwayschanging natural phenomena to be mere shadows of theimmutable ideas, of the eternally fixed essences of things. Thisphilosophy has appealed to many scientists. Individual organ-isms and living populations are often supposed to representimperfect incarnations of ideas, patterns, or types of theirrespective races, species, genera, etc. .. Acceptance of thebiological evolution theory did not completely overcome thenotion that the annoying variability of individuals is somehow afalse front which conceals slowly changing racial or speciestypes. (p. 108)

The definition of the four ‘major’ BGA groups as ‘ancestral groups’that gave rise to lineages leading to present-day East Asians, Eu-ropeans, Native Americans, and sub-Saharan Africans is couchedempirically in terms of historical populations and theirmigrationsdi.e., Dobzhansky’s ‘living populations’dbut typolog-ical assumptions are central.

Assumptions about primordial types are reflected in thedistinction Frudakis and Shriver (2003) draw between ‘recentadmixture’ and ‘ancient admixture’. Although BGA is an exhaustivecategory, individuals are treated differently depending on where inthe world they live and what their ancestry is consideredda pri-oridto be. Making sense of BGA requires paying attention to a di-vision between ‘Old World’ and ‘New World’ populations asdemarcated historically by Columbus’s voyage in 1492 (Fullwiley,2008). Whereas New World populations such as ‘African Ameri-cans’, ‘European Americans’, and ‘Hispanics’ are considered to berelatively recent (since 1492) admixtures of the four ‘foundergroups’, ‘admixture’ is handled differently for Old World pop-ulations outside the Americas. Present-day ‘East Asians’, ‘Euro-peans’, and ‘sub-Saharan Africans’ are treated as unmixed groupswith well-delineated ancestral ties to the ‘founder groups’ andstraightforward histories of migration as peoples. Elsewhere in the‘Old World’, in contrast, groups located in geographical regionsintermediate to East Asia, Europe, and sub-Saharan Africa aretreated as racially mixed groups, caught up in complicated patternsof historical migrations, with difficult-to-untangle ancestral ties.The ANCESTRYbyDNA� User Manual instructs DNAPrint’s genea-logical customers:

It is notable that some regions of the world have morecomplicated histories than other regions making the concept ofancestry more complicated and even tedious.. [F]or millenniapeople have moved about and particular regions of the worldshow traces of these migrations back and forth into and out ofcontinental and sub-continental regions. Some examples ofsuch regions are East Africa, North Africa, Central Asia, SouthAsia, and Insular Southeast Asia. (DNAPrint website, retrieved 4July 2007; italics mine)

But what justification exists for the distinction between ‘recentadmixture’ in ‘New World’ populations and ‘ancient admixture’ in‘Old World’ populations? The Americas are not the only region ofthe world to which migrants have arrived from afar during the past500 years. The designation ‘ancient admixture’ arises because ofthe mistaken assumption that at some point in our evolutionarypast, there were four distinct ‘ancestral groups’ in existence.Obviously, where patterns of genetic variability are clinal, that is,allelic frequencies vary continuously across an area, sampling at theperipheries (e.g., in Japan and Ireland) will produce groups that are

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184180

more distinct genetically than when samples are taken more cen-trally (e.g., in Tunisia and Turkey), but this does not imply that theperipherally located groups are relatively pure descendants ofprimordial lineages and the centrally located groups are admix-tures of these same primordial lineages.

The logic that governs admixture mapping supports theconceptualization of BGA groups as pure and static, as well as pri-mordial, types. Admixture mapping assumes the relative homo-geneity of ancestral or parental populations and the relativeheterogeneity of ‘admixed’ descendant populations, which en-courages (false) assumptions about the purity of BGA groups. In alaboratory-based ethnographic study, anthropologist DuanaFullwiley (2008) insightfully captures a scientist exercising thislogic in discussing the limits of admixture mapping over periodslonger than 10 generations: ‘it’s like freshly mixed paint. You cansee the different colors for a while, until they are too mixed up tosee any more’ (p. 719). False assumptions about purity are furtherencouraged when researchers try to ensure as homogeneous asample as possible when determining parental group allele fre-quencies. For example, Bonilla et al. (2004) eliminated from theirNative American sample any ‘recently admixed individuals’ withmore than 10 percent non-native ancestry (p. 141). The FrudakiseShriver patent application discusses assigning to parental groupsonly individuals with ‘relatively homogeneous affiliation’ to theregion (based on the racial identity of parents and grandparents),who exhibit a ‘strong physical appearance associated with de-scendants of each group’ (Frudakis & Shriver, 2003, Sec. 0236). Theancestral group, ostensibly wholly biological, is thereby con-structed in conformity to the social conception of a pure racial‘type’. The assumption of a static, as well as primordial and puretype, is also supported by the logic of admixture mapping. Since theoriginal ‘founder groups’ no longer exist, the ‘present-day Euro-peans, Native Americans, Africans, and East Asians’ presumed to betheir direct descendants serve as proxies or modern-day stand-insfor determining the SNP allelic frequencies used to select AIMs anddefine their statistical power. This wrongly suggests that the ge-netic composition of population lineages remains unchanged overtime.

The theoretical conceptualization of BGA groups as ideal types isimplied by the computational approach that is used to assign BGAproportions and distinguish between ‘real’ and ‘artificial’ admix-ture. Whether individuals are found to belong partially to multiplegroups (e.g., 85 percent ‘sub-Saharan African’ and 15 percent‘Native American’) or wholly to a single group (e.g., 100 percent‘East Asian’), the approach assumes the discreteness of the four‘major’ BGA groups and, therefore, at least in a theoretical sense(consistent with a reality in which the groups do not have anyactual members), the mutual exclusivity of membership in thosegroups. As explained in the ANCESTRYbyDNA� User Manual, ‘real’admixture refers to genuine BGA proportions, whereas ‘artificial’admixture is attributed to ‘statistical noise’ (DNAPrint website,retrieved 4 July 2007). Statistical noise arises because the differentalleles at the various loci tested are not fixed (at 100 percent) ineach BGA group but differ in frequency (recall that AIMs areselected on the basis of their frequency differential between twogroups). Hence, an individual who has ancestry from only one BGAgroup will nonetheless possess alleles that are more characteristicof (i.e., more likely to be found in) other BGA groups. On the tele-vision mini-series African American Lives, the analogy Frudakis usesto explain the distinction between ‘real’ and ‘artificial’ admixtureshows the idealizing theoretical assumption of discreteness for thefour ‘major’ BGA groups:

Imagine a town which has 4 different clothing stores. Store 1specializes in yellow clothes, but also sells red, green and blue

clothes. Store 2 specializes in red, but also sells yellow, greenand blue clothes. Store 3 specializes in green clothes but alsosells red, yellow, and blue clothes. Store 4 specializes in blueclothes but also sells clothes of the other three colored [sic]. Let’sassume every person in town shops at all these stores. Now if wego out on the street and look at you, an individual who lives inthis town, based on what you are wearing, we can form aprobability statement on howmuch time you have spent in eachof the 4 stores. The clothes you arewearing are analagous [sic] toyour unique gene sequences; and the clothing stores are thepopulation groups to which you may trace your ancestry. (PBS,2006)

‘Real’ admixture presumably reflects the time each person spendsin each of the four stores, whereas ‘artificial’ admixture presumablyreflects the fact that each store, though specializing in clothes of aparticular colour, sells clothes of the other three colours as well.Consequently, the time a person spends, say, in Store 2 cannot bemeasured solely by the red clothes that are being worn since someof that person’s red clothes may have been purchased at one of theother three stores and some of the clothes that person purchasedfrom Store 2 may not be red. The analogy works well to explain thisdistinction. However, it also assumes that the four BGA groups arediscrete (like the walls of a store) and spatially separated (likestores at different street addresses or around the block from oneanother), that the allele frequencies characteristic of each BGAgroup remain constant over time (like fashion trends not disruptingthe combination of colours each store stocks), and that individualsbelong to multiple BGA groups only in virtue of having mixedancestry and not because BGA group membership is itself inde-terminate, including at the time level at which the four ‘foundergroups’ are said to have been established (like the impossibility ofshopping at more than one store at the same time). Hence,although the theoretical representation of discrete groups thatenables ‘real’ and ‘artificial’ admixture to be distinguished recog-nizes that genetic variants are distributed continuously across thespecies (analogous to the four colours of clothing), it fails torecognize that the genealogical relations that are constitutive ofancestrally defined groups are also distributed continuously acrossthe species (disanalogous to the four stores).

It has been argued in this section that BGA incorporates as-sumptions that are similar to the essentialist, taxonomic, andtypological underpinnings of the traditional conceptions and con-structions of race associated with 18th-century natural history,19th-century racial biology, and early 20th-century physical an-thropology. BGA’s inherent flaws and failure as a global category ofbiological and anthropological classification are best understood byseeing it as a local categorydbuilt in a specific place, at a specifictime, for specific purposes. This is to see BGA as race made in theU.S.A.

4. BGA as race made in the U.S.A.

BGA fails as a global category of biological and anthropologicalclassification. As Fullwiley (2008) points out, the paucity of groupssampled in order to identify and validate AIMs for ‘ancestral pop-ulations’ has resulted in a ‘drastically simplified’ representation ofhuman genome diversity across space and time:

The complexity of how both ‘geography’ and ‘time’ have bornout human variation has been drastically simplified.. While thegirth of the globe has been flattened to a small area of WestAfrica, sporadic points in North and South America, and evensparser points of Europe, time has been collapsed into a worldhistory that pivots on the year 1492 with Columbus’s arrival. (p.702)

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184 181

Even so, it may make sense to consider the four ‘major’ BGA classesas ‘founder groups’ if BGA is considered to be a local category,specific to the United States. ‘East Asian’, ‘European’ or ‘IndoEur-opean’, ‘sub-Saharan African’, and ‘Native American’ identify re-gions of the world to which many people from the United Statestrace their ancestry. The makeup of the U.S. population in large partreflects a history that includes the conquest and colonization of theAmericas by the Spanish, French, and English, and the genocidaleffects on the Aboriginal peoples; the forcible removal of peopleliving in western parts of Africa for enslavement in the Americas;and patterns of immigration during the 19th and 20th centuries,which although initially concentrated on northwestern Europeancountries eventually included significant numbers from thesoutheast of Europe and China and Japan.

The idea behind BGAdthat DNAmarkers that are found to differin frequency in different populations can be used to identify anindividual’s population of origindis nothing new. As Dobzhanskywrote in 1950,

The probability that an individual taken from a given populationwill carry a given gene may be either greater or smaller than itwould be for an individual from another population.. By andlarge, the more traits examined in an individual . the moreprecisely can be inferred the part of the world fromwhich theseindividuals came. (pp. 116e117)

And the expression ‘biogeographical ancestry’ certainly has anappropriately scientific ring to it. The first term draws lexically onthe venerable 19th-century scientific tradition of biogeography, its20th-century successors, phenogeography and genogeography,and, of course, the acceptability of the concept of geographical racein zoological species. The second term recognizes the importance ofancestry to our genetic makeup: who our ancestors are, where inthe world they lived, and when they lived there, do matter. When itcomes to the 0.1 percent of DNA bases inwhich Homo sapiens differ,we tend to share more DNA in common with our closest relativesthan we do with our distant relatives.

It is for such reasons, I believe, that biogeographical ancestry hasbeen embraced as a replacement for race as a category of classifi-cation in genetic and genomic research by some tough-minded,socially aware, politically leftist critics. For instance, Britishneurobiologist Steven Rose, well known for his anti-racist socialprogressivism, lends support to BGA. In a 2009 commentary inNaturewhere he argues against conducting research on race and IQ,Rose portrays as an ‘increasing consensus’ among geneticists thatbiogeographical ancestry does better than race at capturing popu-lation differences of interest to geneticists, while managing to avertthe problem race has concerning ‘whether it is a biologically, asopposed to socially, meaningful category’ (p. 787). Rose alsocoauthored a 2005 commentary in the Guardian newspaper withsociologist Hilary Rose that recommends that biogeographicalancestry replace race for the ‘study of biological difference’: race,they argue, is ‘politically and socially constructed’. Only a smallportion of human genetic diversity is ‘racial’, and much of this, theypoint out, is ‘readily visible genetic variation’ that is associated withskin colour, hair form, etc.dthat is, ‘everyday’ differences of littlescientific interest. Questions that matter scientifically, about dis-ease risk and genealogy, require attention to ‘deeper’, less visiblegenetic variation, they say.

The Roses ignore that some scientists are interested in the smallportion of human genetic diversity that is ‘racial’despecially, the‘readily visible genetic variation’ that is associated with skin colour,hair form, etc. Shriver and Frudakis want to predict race in an‘everyday’ sense, and they believe that BGAeAIMs technologyprovides a means of doing so. DNAPrint foresaw a future whereby,with a DNAWitness� test kit in ‘every patrol car’ (DNAPrint,

2004a), felony suspects could immediately be assigned a racial-ized physical appearance. A database of individual BGA admixtureresults matched to ‘ethnic’ self-definitions and digital photographswould allow ‘better physical profiling’ when police officers areconfronted with significant admixture and left wondering whatrace their suspect will appear to be: ‘By querying the database usinga specific BGA admixture result, investigators can see for them-selves what the range of variability is corresponding to that resultfor various features, such as skin shade, hair texture, nose shape,epicanthal eye folding, etc.’ (DNAPrint, 2004b). Given the centralityof Shriver and Frudakis’s forensic aims to the development of BGAeAIMs technology, the Roses are mistaken to assume that BGA andrace are autonomous categories, with BGA an appropriately scien-tific measure of biological diversity and race a social and politicalconstruct. BGAeAIMs technology is the product of race as race hasbeen socially and politically constructed in the UnitedStatesdalthough, as Fullwiley (2008) notes, by repackaging race as‘continental ancestry’, the ‘politically charged historical baggage’ ofthe word ‘race’ is avoided (p. 699).

Evidence of this repackaging is found in the deracialization ofthe language of Frudakis, the industry-based molecular biologist,over the course of his collaboration with Shriver, the university-based biological anthropologist. In the May 28, 2002 patentapplication, which preceded the consulting contract with Shriverby about two weeks, Frudakis describes his invention as a ‘racialclassification tool that genetically defines a person’s racial andethnic background’ (Frudakis, 2002, Sec. 0242). He refers to‘Caucasian’ hair and eye colour; ‘Caucasian racial group’, ‘Africanracial group’, and ‘Asian racial group’; and even ‘light (blue, green,or hazel) iris colored race’ versus ‘dark (black and brown) iriscolored race’ (Frudakis, 2002, Sec. 0903). During 2002, the sup-porting scientific paper was under review by Journal of ForensicSciences with the title ‘A Classifier for SNP-Based Racial Inference’(DNAPrint website, retrieved 6 July 2007), but when it was finallypublished in 2003 (Frudakis et al., 2003), after being revised andresubmitted three times, the title was ‘A Classifier for the SNP-Based Inference of Ancestry’. Language in the body of the articleand patent application also differ: the ‘classifier’ is described in thepatent application as the ‘first SNP base [sic] method for reliablydistinguishing between the world’s major racial groups’ (Frudakis,2002, Sec. 0811), but in the article as the ‘first SNP-based methodfor inferring the ethnic origin of a DNA specimen’ (Frudakis et al.,2003, p. 772); similarly, whereas the application mentions‘Asians, African Americans or Caucasians’ (Frudakis, 2002, Sec.0252), the article refers to ‘individuals of Asian, African, and Eu-ropean descent’ (Frudakis et al., 2003, p. 771). That terminologyassociated with ‘ethnicity’, ‘ancestry’, and ‘descent’ so readily sub-stitutes for ‘race’ suggests that race thinking itself has not beeneliminated in the process but merely disguised.

BGA’s origins lie within a U.S. research context driven not onlyby the forensic aims of predicting the racialized physical appear-ance of suspects from crime-scene DNA but by the post-HGPchallenges of mapping genes for complex traits and the promisesof profits in pharmacogenomics. Given this context, it is not sur-prising that the ‘major’ BGA groups coincide with racial classes asthey have been defined traditionally in the United States ascontinental-level differences and institutionalized by governmentagencies. In DNA forensics, as part of CODIS (Combined DNA IndexSystem), the FBI maintains ‘African American’, ‘Asian’, ‘Caucasian’,‘Hispanic’, and ‘Native American’ population databases for thedetermination of match probabilities. In the mapping of genes forcomplex traits, NIH-funded scientists must use the OMB categoriesof race and ethnicity to satisfy the mandate of The National In-stitutes of Health Revitalization Act of 1993 that ‘minorities’ beincluded in clinical research. In pharmacogenomics, the Food and

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184182

Drug Administration has required since 1998 that new drug ap-plications present data on effectiveness and safety ‘by gender, age,and racial subgroups’ (FDA,1998), and it recommends that the OMBcategories of race and ethnicity be used in clinical trials to collectthese data (FDA, 2005). As outlined in the introduction, the OMBsystem asks individuals to self-identify, first, by ‘ethnicity’ (theirchoices are ‘Hispanic or Latino’ and ‘Not Hispanic or Latino’) and,second, by ‘race’ (their choices are ‘American Indian or AlaskaNative’, ‘Asian’, ‘Black or African American’, ‘Native Hawaiian orOther Pacific Islander’, and ‘White’). Note that the OMB considersthe categories to be ‘social-political construct[s]’ that ‘are not an-thropologically or scientifically based’ (OMB, 1997, ‘SupplementaryInformation’, Sec. A).

DNAPrint’s decision to market the ANCESTRYbyDNA� line ofdirect-to-consumer genealogy tests in order to improve cash flowfor the company also shaped the development of BGAeAIMs tech-nology. Among non-Aboriginal North Americans, genealogy haslong been a prevalent pastime, especially for ‘European Americans’,but as a result of the popularity of Alex Haley’s book Roots andsubsequent TV mini-series in the late-1970s, ‘African Americans’ aswell. This curiosity about ancestry seems to be conceived over-whelmingly in ways that coincide with BGA’s hierarchical embed-ding of ethnic and national identities within continental-level racialdifferences. Emphasis is placed on oversea ties to distantdthoughnot too distantdcousins and places alike. For some, a visit to Ire-land’s County Tyrone may be deemed meaningful in a way that avisit to Toronto’s Cabbagetown (less distant kin) or Kenya’s TurkanaBasin (more distant kin) is not. For others, whose ancestors werekidnapped and sold as slaves, the return to Africa completes thejourneydspecifically to places associated with this dislocation,where one’s closest kin on that continent might live. Henry LouisGates (2007) writes: ‘Like 130 million other people, I watchedevery episode of Alex Haley’s Roots when it first aired in 1977. Andlike many other African-Americans, I have yearned ever since totrace my own roots, to identify where in Africa my own ancestorscame from, what tribe they were part of’ (p. 133).

To the extent that BGA manages to facilitate the arrest of asuspect by predicting a racialized physical appearance on the basisof crime-scene DNA, or to eliminate false positives in mappingvariants associated with complex traits to the genome, it is notbecause BGA is a wholly ‘biological’, ‘genetic’, ‘natural’, ‘heritable’,or ‘objective’measure of race and ethnicity, as its proponents claim.In a media interview (Willing, 2005), Shriver acknowledges that‘there’s a huge sensitivity about race in our society’ about which it isimportant to be sensitive. But he continues: ‘that doesn’t take awaythe reality that people often describe each other in terms of race.We’re saying: Let DNA be the witness.’ What is important toemphasize is that the ‘reality’ to which Shriver refers is a social andpolitical reality that is sustained by shared discursive practiceswhereby ancestral lineages and continental shores becomeconfigured inways that are constitutive of our identities, of whowetake ourselves and others to be. When BGAeAIMs technologysucceeds in predicting racial and ethnic identities, it is becausepatterns of DNA difference have been successfully mobilized in theservice of existing social and political constructions of race andethnicity.

In ‘Biologistical Construction of Race: “Admixture” Technologyand the New Genetic Medicine’, Fullwiley provides an account ofBGAeAIMs technology as the product of race as race has been so-cially and politically constructed in the United States. By ‘bio-logistical construction’, Fullwiley means that two domains havebeen brought into ‘correspondence’: one, that of ‘body traits mademeaningful through conceptions of race (given certain social andpolitical contexts)’; the other, that of ‘supposedly politically andsocially neutral DNA’ (2008, p. 698). Fullwiley denies that an a

priori distinction of nature versus culture exists between the twodomains. Instead, she describes BGAeAIMs technology as devel-oped thus far as a ‘tautological product’ wholly embedded in‘American racial taxa’ traditionally construed as continental ‘racial’stocks, i.e., ‘Black’/African, ‘White’/European, and ‘Red’/NativeAmerican (2008, pp. 697e698). DNA differences come to incorpo-rate meanings that reflect the social construction of race in theUnited States as continental ‘stocks’ and attach already to differ-ences in skin colour, hair form, etc.

Presumably, Fullwiley’s use of the term ‘tautological’ is intendedto convey the idea that the genome tells us nothingmore about racethan what scientists bring to it, their claims about continentalancestry and the genetic basis of racial identity mere re-enactmentsof insidious, longstanding, and widespread cultural beliefs aboutcontinentally-defined racial differences at the level of phenotype.Fullwiley’s description of BGAeAIMs technology as a ‘tautologousproduct’ should not be taken to imply, however, that operations oflogic alone mediate the transfer of racial meanings from body traitsto DNA. The construction of identities based on group affiliation, atthe level of the genome as elsewhere, is a dynamic narrative pro-cess. The scientist-inventors of BGAeAIMs technology, as activeparticipants in such a process, are responsible for the contingentchoices they make.

In a discussion of genetic testing as a basis for proving NativeAmerican identity and tribal membership, anthropologist KimberlyTallBear (2008) emphasizes the limits of DNA and centrality ofnarrative construction in establishing who we take ourselves to be:

Native DNA should not be understood simply as an objectivemolecular ‘thing’. It is simultaneously a conceptual apparatusthrough which humans constitute and deploy life-organizingnarratives: historical, national, and racial narratives; narrativesabout family and tribe and the origins of peopledwho in-dividuals ‘really are’. (p. 236)

Sociologist Alondra Nelson (2008) describes how consumers‘deploy’ results from genetic genealogy tests to construct their‘individual and collective biographies’ in ways that they findmeaningful, using nongenetic evidence about ancestry to interpretresults, such that the ‘geneticization’ of racial and ethnic identitiesabout which critics have worried does not arise (pp. 761e762).Indeed, although it is widely assumed that ‘ancestry’ and ‘geogra-phy’ are more objective categories than ‘race’ and ‘ethnicity’, andthat DNA alone determines the geographically-defined ancestralgroups to which people belong, it is difficult to imagine, even inprinciple, how this is possible. We have countless ancestors, each ofwhom lived in a certain place or places during their lives. Narrativeis required for conceiving identity as a function of so many placeswhere so many ancestors lived. Which lineages matter? Whathistorical periods matter? How can geographical locations benamed without recourse to socially and politically salient cate-gories (e.g., nation, tribe, ethnic group) that fix people to place?

Although the narrative construction of identity precludes aninevitable geneticization of existing social and political categoriesof race and ethnicity, BGAeAIMs technology, as it was developed,satisfies culturally dominant narratives: the ‘major’ BGA groupscoincide with continental races belonging to the European scien-tific and philosophical traditions, and BGA as a measure of ‘recentadmixture’ accommodates U.S. racial groups defined by hypo-descent and blood quantum. As Fullwiley (2008) notes, since BGAdetermines racial admixture based on genetic contributions fromall ancestral lines, it departs from the hypodescent concept of racethat has been traditional in the United States, whereby ‘one drop ofblood’ from an African or African-American ancestor assigns aperson to that group: ‘Today, admixture theories rely less on anotion of hypo-descent to solidify one’s identity than on a notion of

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184 183

racialized genomic fractions, or parts’ (p. 726). However, for manyof DNAPrint’s genealogical customers, ‘genomic fractions’ becomemeaningful only within the parameters of the one-drop rule. TheANCESTRYbyDNA� test was not helpful to many African Ameri-cans, more interested in tracing their ancestry back to particularplaces in Africa than in ascertaining how ‘racially mixed’ they are.As one of Nelson’s African-American informants told her, ‘I don’tneed to take that test. We’re all mixed up. We know that already’(2008, p. 766). In contrast, European Americans who take theirracial purity (no drops of non-White blood) for granted are likely tobe surprised by, though may welcome, results that suggest theyhave African or Native American ancestry.

The scientist-inventors of BGAeAIMs technology ignored itspotential to contribute to racial injustice. In the United States, theracialization of crime and tenacity of race as a social and politicalcategory combine to make the description of a suspect’s appear-ance seemingly unimaginable apart from race. In the release of theDNAWitness� line, little attention seems to have been paid byDNAPrint to the social and political ramifications of promoting atechnology that embeds race in the genome by using crime sceneDNA to racially profile suspects, despite well-known concernsabout the racialization of crime and criminalization of racial mi-norities in the United States. Indeed, the face of the company’sadvertising campaign for the product was that of an African-American male. In framing their research, scientists are facedwith choices, and they are responsiblednot just epistemically, butethically, socially, and politicallydfor the choices they make. Thedevelopment of BGAeAIMs technology for forensic use offers ma-terial support to institutional racism in the U.S. criminal justicesystem that sees a disproportionate number of men of colour,particularly Black men, stopped and frisked by police, incarceratedfor minor offences, and sentenced to death for murder (especiallywhen the victim is White). There were alternative choices. Scien-tists might have decided to avoid contributing to institutionalracism by foregoing research on forensic DNA phenotyping entirelyor delaying the implementation of such research. Or scientistsmight have entered the field of forensic DNA phenotyping bydeveloping panels of SNPs that predict suspects’ appearanceswithout racialization, whether by investigating the genetic basis oftraits not usually considered racial (e.g., height) or reconceivingdifferences in traits that tend to be considered racial (e.g., dark andlight skin colour, straight and curly hair) as individual differences.

5. Conclusion: hold the hoorays for BGA

Today, DNAPrint genomics, Inc. is defunct. The company neverdidmakemoney. Bymid-2007, after losing $8.7 million in 2005 and$12.3 million in 2006, DNAPrint was trading for less than a penny ashare and forced to issue shares in its subsidiary DNAPrint Phar-maceuticals as payment of $6 million to Dutchess Private EquitiesFund Ltd. In January 2008, Nanobac Pharmaceuticals signed a letterof intent to acquire DNAPrint, but the deal did not go through, andin May 2008, the company moved from a NASDAQ to pink-sheetlisting. By early 2009, the company had ceased operations alto-gether. In his 2008 book, Molecular Photofitting, Frudakis blames‘misinformation’ and ‘politics’ for some of the ‘resistance’ to thetechnology, and claims that taking a genetic approach to the racialand ethnic stratification of populations is subject to more stringentstandards for its adoption by the legal system than are social sci-entific approaches such as eyewitness testimony (pp. 609e610).Frudakis thanks the ‘average, every-day citizens’ who invested inDNAPrint genomics and made the work possible despite rejectedgrant applications, and points to the Louisiana Task Force as evi-dence that notwithstanding their financial losses, they had made a‘positive difference in the world’ (p. xv).

Nevertheless, biogeographical ancestry is still with us. For thefive-year period from 2008 through 2012, there are 319 referencesto ‘biogeographical ancestry’ on Google Scholar. One article predictspigment phenotype (‘typical European’ with blue or green eyes,light hair and skin) and biogeographical ancestry (likely European)for 25 human skeletons found in Siberia dating from Bronze andIron ages (Bouakaze, Keyser, Crubézy, Montagnon, & Ludes, 2009);another article establishes a panel of ancestry-informative SNPs todifferentiate major populations in Australia for forensic purposes(Daniel, Sanchez, Nassif, Hernandez, & Walsh, 2008); biogeo-graphical ancestry is also used to study the genetic contribution toindividual differences in pain sensitivity (Herbstman et al., 2008);and in Brazil, researchers calibrate self-reported ‘color’ (a censuscategory), skin pigmentation measured by reflectometry, and‘genomic ancestry’ to support efforts to control for populationstratification in epidemiological studies (Leite, Fonseca, França,Parra, & Pereira, 2011). Shriver has maintained an interest in fo-rensics and the ability to predict racial or ethnic identity from crimescene DNA, which includes, in recent years, the relationship be-tween facial features and admixture proportions (Klimentidis &Shriver, 2009).

The Roses close their Guardian commentary with these words:

[A]s a scientific concept, race is well past its sell-by date. Thestudy of biological difference needs a new conceptdone thatmakes possible sensitive recognition of the diversity of healthrisk, and of tracing our own ancestral roots through DNA anal-ysis. And we need it to be a concept fit for the longing for a post-racist 21st century. So let’s hear it for the clunky, let’s hear it forthe precisedhooray for biogeographical ancestry. (Rose & Rose,2005)

I would urge more caution in embracing biogeographical ancestryas an alternative to race. Unfortunately, what we find in consideringthe history of this ‘new concept’ is that what Shriver called‘ethnicity’ in 1997 and Frudakis called ‘race’ in 2002 was subse-quently simply renamed ‘ancestry’ and then ‘biogeographicalancestry’, thus coming to seemmore scientific and less contentious,its origins in race and ethnicity as socially and politically con-structed categories hidden from view. We have seen that theresearch context surrounding the invention of BGAeAIMS tech-nology was shaped by diverse interestsdsocial and commercial aswell as scientificdin DNA forensics, gene mapping, pharmaceuticaldevelopment, and direct-to-consumer genealogy testing. Closeattention to this research context reveals that biogeographicalancestry is itself a construction built upon race as race has beensocially constructed in the European scientific and philosophicaltraditions, and especially in the United States. As such, we ought tohold off ‘hoorays’ for BGA.

Acknowledgements

I presented material from this paper at a number of conferences,workshops, and colloquia over several years, and I am indebted toall of the organizers, participants, and audience members involved.I am especially grateful to Jenny Bangham and Soraya de Cha-darevian for their helpful editorial input and for inviting me to bepart of this important project on post-WWII human heredity, whichbrought together such a wonderful group of scholars.

References

Bonilla, C., Parra, E. J., Pfaff, C. L., Dios, S., Marshall, J. A., Hamman, R. F., et al. (2004).Admixture in the Hispanics of the San Luis Valley, Colorado, and its implicationsfor complex trait gene mapping. Annals of Human Genetics, 68, 139e153.

Bouakaze, C., Keyser, C., Crubézy, E., Montagnon, D., & Ludes, B. (2009). Pigmentphenotype and biogeographical ancestry from ancient skeletal remains:

L. Gannett / Studies in History and Philosophy of Biological and Biomedical Sciences 47 (2014) 173e184184

Inferences from multiplexed autosomal SNP analysis. International Journal ofLegal Medicine, 123, 315e325.

Chen, S., Chung, M. K., Martin, D., Rozich, R., Tchou, P. J., & Wang, Q. (2002). SNPS1103Y in the cardiac sodium channel gene SCN5A is associated with cardiacarrhythmias and sudden death in a white family. Journal of Medical Genetics, 39,913e915.

Daniel, R., Sanchez, J. J., Nassif, N. T., Hernandez, A., & Walsh, S. (2008). SNPsassociated with physical traits: A valuable tool for the inference of biogeo-graphical ancestry. Forensic Science International: Genetics Supplement Series, 1,538e540.

DNAPrint genomics. (2002a). Annual report: U.S. Securities and Exchange Com-mission Form 10-KSB [for 2001]. Retrieved from http://www.secinfo.com/d1zWxj.3h.htm.

DNAPrint genomics. (2002b). DNAPrint/Mark Shriver Consulting Contract.Retrieved from http://www.secinfo.com/d1zWxj.328.d.htm#1stPage.

DNAPrint genomics. (2003). DNAPrint’s DNAWitness test provided break in theLouisiana Multi-Agency Homicide Task Force serial killer case (5 June) [pressrelease]. PR Newswire.

DNAPrint genomics. (2004a). DNAPrint’s DNAWitness Technology Tightens Net onUK Rapist (28 April) [press release]. Retrieved from http://www.prnewswire.co.uk/news-releases/dnaprints-dnawitness-technology-tightens-net-on-uk-rapist-154560635.html.

DNAPrint genomics. (2004b). DNAPrint upgrades DNAWitness� 2.5 (22 July) [pressrelease]. Retrieved from http://www.prnewswire.com/news-releases/dnaprint-upgrades-dnawitnesstm-25-new-information-technology-for-better-physical-profiling-71376087.html.

Dobzhansky, T. (1950). The genetic nature of differences among men. In S. Persons(Ed.), Evolutionary thought in America (pp. 86e155). New Haven, CT: Yale Uni-versity Press.

Dobzhansky, T. (1951). Genetics and the origin of species (3rd ed.). New York: Uni-versity of Columbia Press.

FDA [Food and Drug Administration]. (1998). Investigational new drug applicationsand new drug applications. Retrieved from http://www.gpo.gov/fdsys/pkg/FR-1998-02-11/html/98-3422.htm.

FDA [Food and Drug Administration]. (2005). Guidance for industry: Collection ofrace and ethnicity data in clinical trials. Retrieved from http://www.fda.gov/downloads/RegulatoryInformation/Guidances/ucm126396.pdf.

Frudakis, T. N. (2002). Compositions and methods for detecting polymorphismsassociated with pigmentation. U.S. Patent and Trademark Office PublicationNumber 20030211486. Retrieved from http://portal.uspto.gov/pair/PublicPair.

Frudakis, T. N. (2008).Molecular photofitting: Predicting ancestry and phenotype usingDNA. Boston: Elsevier/Academic Press.

Frudakis, T. N., & Shriver, M. D. (2003). Compositions and methods for inferringancestry. U.S. Patent and Trademark Office Publication Number 20040229231.Retrieved from http://portal.uspto.gov/pair/PublicPair.

Frudakis, T., Venkateswarlu, K., Thomas, M. J., Gaskin, Z., Ginjupalli, S., Gunturi, S.,et al. (2003). A classifier for the SNP-based inference of ancestry. Journal ofForensic Sciences, 48, 771e778.

Fullwiley, D. (2008). The biologistical construction of race: ‘Admixture’ technologyand the new genetic medicine. Social Studies of Science, 38, 695e735.

Gannett, L. (2001). Racism and human genome diversity research: The ethical limitsof ‘population thinking’. Philosophy of Science, 68, S479eS492.

Gannett, L. (2010). Questions asked and unasked: How by worrying less about the‘really real’ philosophers of science might better contribute to debates aboutgenetics and race. Synthese, 177, 363e385.

Gates, H. L. (Dec 2007). ‘We are all Africans’: Genealogical research and DNA testingcan reveal your ethnic connection to Africa. Ebony, 132e136.

Gower, B. A., Fernández, J. R., Beasley, T. M., Shriver, M. D., & Goran, M. I. (2003).Using genetic admixture to explain racial differences in insulin-related phe-notypes. Diabetes, 52, 1047e1051.

Halder, I., & Shriver, M. D. (2003). Measuring and using admixture to study thegenetics of complex diseases. Human Genomics, 1, 52e62.

Hawkins Simons, D. (15 Jun 2003). Getting DNA to bear witness: Genetic tests canreveal ancestry, giving police a new source of clues. US News & World Report.Retrieved from http://www.usnews.com/usnews/culture/articles/030623/23dna.htm.

Herbstman, D., Mao, X., Shriver, M., Herrera, D., Rahim-Williams, B., Wallace, M.,et al. (2008). The association of ancestry informative markers with experi-mental pain sensitivity. The Journal of Pain, 9(4 Suppl. 2), 62.

Institute of Medicine. (1999). The unequal burden of cancer: An assessment of NIHresearch and programs for ethnic minorities and the medically underserved.Retrieved from http://www.nap.edu/catalog.php?record_id¼6377.

Klimentidis, Y. C., & Shriver, M. D. (2009). Estimating genetic ancestry proportionsfrom faces. PLoS ONE, 4(2), e4460.

Leite, T. K. M., Fonseca, R. M. C., França, N. M., Parra, E. J., & Pereira, R. W. (2011).Genomic ancestry, self-reported ‘color’ and quantitative measures of skinpigmentation in Brazilian admixed siblings. PLoS ONE, 6(11), e27162.

Mulle, M. A. (19 Sep 2002). DNAPrint sells racial test to public. Sarasota Herald-Tribune, 1D, 8D.

National Institutes of Health Revitalization Act of 1993. Retrieved from http://history.nih.gov/research/downloads/PL103-43.pdf.

Nelson, A. (2008). Bio science: Genetic genealogy testing and the pursuit of Africanancestry. Social Studies of Science, 38, 759e783.

OMB [Office of Management and Budget]. (1997). Revisions to the standards for theclassification of federal data on race and ethnicity. Retrieved from http://www.whitehouse.gov/omb/fedreg_1997standards.

Pääbo, S. (2003). The mosaic that is our genome. Nature, 421, 409e412.Parra, E. J., Pfaff, C. L., Hiester, K., Sosnoski, D., Jien, Y., Gulden, F., et al. (2000).

Pattern and dynamics of admixture in US African-American populations: Aglobal perspective. American Society of Human Genetics Annual Meeting.Retrieved from http://www.ashg.org/genetics/abstracts/abs00/f1196.htm.

PBS [Public Broadcasting Service]. (2006). What is admixture testing? AfricanAmerican Lives. Retrieved from http://www.pbs.org/wnet/aalives/2006/science_tests3.html.

Pfaff, C. L., Parra, E. J., & Shriver, M. D. (2000). Genetic estimation of biogeographicalancestry. American Society of Human Genetics Annual Meeting. Retrieved fromhttp://www.ashg.org/genetics/abstracts/abs00/f1195.htm.

Race, Ethnicity, and Genetics Working Group. (2005). The use of racial, ethnic, andancestral categories in human genetics. American Journal of Human Genetics, 77,519e532.

Rose, S. (2009). Should scientists study race and IQ? No: Science and society do notbenefit. Nature, 457, 786e788.

Rose, S., & Rose, H. (9 April 2005). Why we should give up on race. Guardian.Retrieved from http://www.guardian.co.uk/world/2005/apr/09/race.science.

Shields, A. E., Fortun, M., Hammonds, E. M., King, P. A., Lerman, C., Rapp, R., et al.(2005). The use of race variables in genetic studies of complex traits and thegoal of reducing health disparities: A transdisciplinary perspective. AmericanPsychologist, 60, 77e103.

Shriver, M. D., Smith, M. W., Jin, L., Marcini, A., Akey, J. M., Deka, R., et al. (1997).Ethnic-affiliation estimation by use of population-specific DNA markers.American Journal of Human Genetics, 60, 957e964.

Shriver, M. D., Parra, E. J., Dios, S., Bonilla, C., Norton, H., Jovel, C., et al. (2003). Skinpigmentation, biogeographical ancestry, and admixture mapping. Human Ge-netics, 112, 387e399.

TallBear, K. (2008). Native-American-DNA.com: In search of Native American raceand tribe. In B. A. Koenig, S. S.-J. Lee, & S. S. Richardson (Eds.), Revisiting race in agenomic age (pp. 235e252). Rutgers University Press.

Wade, N. (1 Oct 2002). For sale: A DNA test to measure racial mix. New York Times.Retrieved from http://www.nytimes.com/2002/10/01/science/for-sale-a-dna-test-to-measure-racial-mix.html?ref¼nicholaswade.

Willing, R. (16 Aug 2005). DNA tests offer clues to suspect’s race. USA Today.Retrieved from http://usatoday30.usatoday.com/news/nation/2005-08-16-dna_x.htm?csp¼15.