AMERICAN JOURNAL OF HUMAN BIOLOGY 8:609414 (1996)

Genetic Structure of the Utah Mormons: A Comparison of Kinship Estimates From DNA, Blood Groups, Genealogies, and Ancestral Arrays

ELIZAEiETH O’BRIEN, REBEKHA ZENGER, AKD LYNN R. JORDE Department of Human Genetics, Eccles Institute of Human Genetics, Uniuersily of Utah Health Sciences Center, Salt Lake City, Ulah 84112

ABSTRACT Kinship estimates based on shared proportions of European ancestry are constructed for 284 Utah males born in eight geographic subdivi- sions. These “ancestral kinship,’ estimates are compared with kinship coeffi- cients based on DNA polymorphisms, blood groups, genealogies, isonymy, and migration matrices. At the subdivision level, a high correlation is observed between ancestral kinship and kinship based on isonymy. Comparing individuals, a significant correlation is obtained between ancestral kinship and genealogy derived kinship. Kinship estimates based on DNA and blood groups do not correlate significantly with ancestral kinship. This is most likely due to the effects of several generations of random mating in this pOpUk3tiOn. 0 1996 Wiley-Liss. Inc.

Investigators often rely on a single type of data to infer the evolutionary history of a population (e.g., gene frequencies, isonymy, migration matrices, anthropometrics, or ge- nealogies). Each type of data has strengths and weaknesses, and a richer interpretation of evolutionary genetics is usually obtained by comparing multiple types of data in the same population. Now that DNA polymor- phisms are commonly used in evolutionary studies (Bamshad et al., 1995; Bowcock et al., 1991, 1994; Cann et al., 1987; Deka et al., 1991, 1995; Di Rienzo et al., 1994; Ed- wards et al., 1992; Harpending et al., 1993; Horai et al., 1993; Jorde et al., 1995; Kidd et al., 1991; Martinson et al., 1993; Mountain and Cavalli-Sforza, 1994; Rogers and Jorde, 1995; Spurdle et al., 1994; Stoneking et al., 1990; Torroni et al., 1990, 1994; Watkins et al., 1995), it is especially important to com- pare estimates of genetic structure based on these polymorphisms with estimates based on other types of data.

A number of different data types have been used to infer the genetic structure of the Utah Mormon population, including mi- gration matrices (Jorde, 1952), isonymy (Jorde and Morgan, 1987), pedigrees (Jorde, 19891, gene frequencies estimated from blood groups and protein electrophoresis (McLellan et al., 19841, and restriction frag-

ment length polymorphisms (RFLPs) (O’Brien et a]., 1994bj. These studies have shown that this population has a low in- breeding rate and is genetically quite homo- geneous. Gene frequency analyses demon- strated that the Utah Mormon population is genetically similar to the northern European populations from which it was derived, with little indication of genetic drift. A recent study revealed no evidence of a founder ef- fect (OBrien et al., 1994a).

These results are all consistent with the demographic history of this population. It was founded in the mid-19th century by > 100,000 individuals, most of whom were derived from northern Europe. High birth rates and a high rate of immigration contrib- uted to rapid growth and a high degree of population mobility.

The genetic structure ofthe Utah Mormon population has recently been investigated using kinship estimates based on RFLPs (O’Brien et al., 1994b). In the present study, these results are compared with a measure of kinship based on the degree of shared Eu-

Recrivcd February 9, 1YY4; accepted June 23, 1995. Address reprint requests ki Dr. L.B. Jorde, Dept. of Human

Genetics, Eccles lnstitute of Human Genetics, Univ. of Utah Health Sciences Centcr, Salt Lake City, UT 84112.

0 1996 Wiley-Liss, Inc.

61 0 E. O'BRIEN ET AL

Fig. 1. Map of Utah showing locations of the eight counties from which subjects were sampled.

ropean ancestry in each pair of study sub- jects. Results indicate that this measure does not correlate well with kinship based on blood groups and RFLPs. The lack of cor- relation is likely to be due to several genera- tions of random mating in this highly mo- bile population.

MATERIALS AND METHODS The study sample consisted of 284 Utah

were chosen because previous studies indi- cated that they provided a maximum level of between-groups genetic diversity (OBrien et al., 1994b). Males were chosen as subjects because Y-chromosome variation is being an- alyzed in ongoing studies. Eight blood group systems were typed for each subject (ABO, Rh, MNS, Lewis, P, Kell, Duffy, and Edd) , yielding 25 alleles. Twenty-two restriction fragment length (RFLP) systems ~~ were also

males born between 1941 and 1955 in eight different counties (Fig. 1). These counties

typed, yielding 47 alleles. Further details on subject ascertainment and laboratory meth-

GENETIC STRUCTURE OF UTAH MORMONS 61 1

ods are given in O’Brien et al. (1994b3. Com- parative kinship measures derived from ge- nealogies, isonymy, and migration matrices have been published previously (Jorde, 1982, 1989; Jorde and Morgan, 1987).

All study subjects are members of the Utah Population Data Base (UPDP), a com- puterized genealogical data base that con- sists of 1.1 million individuals linked into large pedigrees. This made i t possible to as- semble ascending pedigrees for each of the 442 males reported in a previous study (OBrien et al., 1994b). Paternal and mater- nal lines of ascent were both included, and there were, on average, four generations be- tween a subject and terminal ancestor. The birthplace of each terminal ancestor (i.e., an ancestor with no parents in the genealogy) was ascertained, and a subject was included in this study if 50% or more of his terminal ancestors were born in Europe. Since a large proportion of these ancestors were born in the United States, this limited the study population to 284 males. Terminal ancestors born in the United States were not included in subsequent calculations. Virtually all of the terminal ancestors born outside of the United States originated in nine European countries: Denmark, England, Ireland, Ger- many, Norway, Scotland, Sweden, Switzer- land, and Wales.

The following procedure was used to esti- mate the “ancestral kinship’’ of each pair of study subjects. First, the country of origin of each terminal ancestor was examined to create an array, a, of proportionate ancestral contributions for each individual. For exam- ple, an individual with grandparents from Denmark, England, Sweden, and Switzer- land would have entries, ak, of 0.25 corres- ponding to each of these countries and zero for all other countries. A matrix, R, of genetic covariances (Harpending and Jenkins, 1973) was estimated for the nine European coun- tries using published gene frequency data (Mourant et al., 1976). Then the ancestral kinship, sg, between each pair of individuals was estimated as:

9 9

k=l I-1

In this equation, the genetic covariance be- tween each European country was used to weight the degree of ancestral kinship be- tween each pair of individuals. Equation 1

was used to form a matrix of ancestral kin- ship coefficients for all possible pairs of 284 subjects. This matrix was compared with a genealogical kinship matrix estimated from known pedigree relationships between indi- viduals and with matrices of proportions of shared alleles estimated from RFLPs and blood groups (see O’Brien et al., 1994b, for details on these measures).

In addition t o estimating ancestral kin- ship between all possible pairs of individu- als, an 8x8 matrix of ancestral kinship was estimated using geographic subdivisions as the unit of analysis. Here, s , ~ values were averaged for all possible pairs of subjects born in the same subdivision to form the diagonal of the matrix. The off-diagonal ele- ments of the matrix consisted of the average of the s , ~ values for all pairs of individuals born in different subdivisions (e.g., the rn,nth entry of the matrix would be obtained by estimating kinship values between all indi- viduals in m against all individuals in n and then averaging these values).

Correlations between matrices and associ- ated significance levels were estimated us- ing the Mantel matrix comparison technique (Mantel, 1967; Smouse et al., 1986). An em- pirical distribution of correlation coefficients was formed by performing 10,000 random permutations of each matrix.

RESULTS The distribution of European ancestry for

each of the eight subdivisions and for the total sample is given in Table 1. Nearly one- half of the ancestry of this population origi- nates in England, another 21% originates in Denmark, and another 10% originates in Sweden and Norway. These figures are con- sistent with historical descriptions of Utah‘s 19th-century immigrants, which indicate that approximately one-half of them came from England and another one-third from Scandinavia (Arrington and Bitton, 1979; Bitton and Irving, 1976; Mulder, 1976).

There is considerable diversity among subdivisions with regard to proportionate ancestral contributions. Again, these differ- ences are consistent with the historical liter- ature. For example, a high concentration of Scandinavians settled in the rural central part of the state (Sanpete County), whereas a large proportion of the immigrants to mountainous Wasatch County came from Switzerland.

Table 2 shows the results of Mantel matrix

61 2 E. OBRIEN ET AL.

TABLE 1. Distribution of ancestral origins b.y subdivision

Beaver Box Elder Cache Davis Salt Lake Sanpete Wasatch Washington Total

0.1046 0.3408 0.1709 0.1270 0.1995 0.4650 0.1318 0.1039 0.2119

County Denmark England Ireland Germany Norway Scotland Sweden Switzerland Wales

-

0.6769 0.4236 0.4936 0.6397 0.4754 0.2654 0.4327 0.5223 0.4855

0.0277 0.0191 0.0128 0.0046 0.0082

0.0516 0.0148 0.0176

o.0106

0.0031 0.0223

0.0069 0.0273 0.0234 0.0287 0.0059 0.0186

0.0278

0.0000 0.0382 0.0491 0.0577 0.0164 0.0488 0.0000 0.0208 0.0313

0.0585 0.0573 0.0491 0.0577 0.0710 0.0361 0,1089 0.0593 0.0607

0.0308 0.0605 0.0812 0.0462 0.1475 0.0913 0.0602 0.0208 0.0692

0.0062 0.0000 0.0470 0.0185 0.0109 0.0149 0.1461 0.2166 0.0545

0.0923 0.0382 0.0684 0.0416 0.0437 0.0446 0.0401 0.0356 0.0506

TABLE 2. Mantel matrix correlations betiueen nnmstral kinship arid other types of dala a t the subdivision leuel

Data type Correlation Simificance level

Blood groups RFLPs

0.013 0.4 0.001 0.5

Genealogies 0.285 0.09 Isonymy 0.672 0.001 Mieration matrix 0.327 0.07

L

Geographic distance 0.072 0.4

TABLE 3. Mantel matrix correlationn for kinship matrices nt the individual leuel'

Ancestral Blood kinship moups RFLPs

Blood groups -0.020 (0.07)

RFLPs -0.011 0.016 (0.19) (0.06)

Genealogies 0.0422 0.009 0.023 (0.001) (0.03 (0.001)

'Significance levels are in parentheses below each correlation. ?Ths value increases to 0.141 when only pairs with non-zerogenralogi- cal kinship values are used.

comparisons of different types of data at the geographic subdivision level (i.e., 8x8 matri- ces were compared in each case). Low corre- lations are apparent between ancestral kin- ship and the blood group and DNA data. There is also little correlation between inter- subdivision geographic distance and ances- tral kinship. Moderate and nearly signifi- cant correlations are seen between ancestral kinship and kinship estimates based on ped- igrees and migration matrices. A highly sig- nificant correlation is evident between ancestral kinship and kinship estimated from isonymy.

Table 3 presents a matrix of correlation coefficients of kmship estimates a t the indi- vidual level (i.e., all possible pairs of the 284 subjects). Although none of the correlations is large, some are statistically significant be-

cause of the large number of pairs. Correla- tions between ancestral kinship and kinship based on blood groups and DNA polymor- phisms are not significant, but there is a significant correlation between ancestral and genealogical kinship (roughly consistent with the subdivision-level results presented above). This correlation increases to 0.141 when only pairs of individuals with nonzero genealogical kinship coefficients are in- cluded.

DISCUSSION A previous study using these RFLP and

blood group polymorphisms showed that the Utah Mormon population is genetically quite homogeneous, with only -1% of its genetic variation attributable to the effects of subdi- vision (O'Brien et al., 1994b). This study also showed that subdivision-level comparisons yielded nonsignificant correlations between kinship measured by RFLPs, blood groups, and genealogies. However, at the individual level, genealogical kinship and kinship based on blood groups and DNA polymor- phisms were significantly correlated. Genea- logical kinship was more highly correlated with kinship based on RFLPs than with kin- ship based on blood groups, indicating that RFLPs are likely t o provide better measures of population relationships than are blood groups.

In the present study, kinship based on Eu- ropean ancestry is added to the battery of kinship measures previously obtained for this population. Ancestral kinship is highly correlated with kinship estimated from ison- ymy. This is probably because certain ethnic groups, with specific surnames, tended to cluster in geographically dispersed locations in the state (Table 1). As would be expected, this non-uniform set,tlement pattern pro- duces a nonsignificant correlation between geographic distance and ancestral kinship.

GENETIC STRUCTURE OF UTAH MORMONS 61 3

These patterns of con-elation are consistent with an earlier study showing little correla- tion between isonymy and geographic dis- tance (Jorde and Morgan, 1987). The correla- tions between ancestral kinship and kinship derived from migration matrices and geneal- ogies are substantially lower than those with isonymy, and the correlations between an- cestral kinship and the DNA and blood group polymorphisms are lower still.

At the individual level, significant corre- lations were obtained between ancestral kinship and genealogical kinship. As in a previous study(0Brien et al., 1994b), genea- logical kinship correlated significantly with kinship based on blood groups and DNA polymorphisms. But, as with the subdivi- sion-level comparisons, there was no correla- tion between ancestral kinship and kinship measured by blood groups and DNA poly- morphisms.

Several factors may explain this lack of correlation. First, although studies of Euro- pean populations do reveal isolation-by-dis- tance effects (McLellan et al., 1984; Morton et al., 1977), the genetic differences among western European populations are relatively small (Nei and Roychoudhury, 1982). Thus significant differences in gene frequencies of DNA polymorphisms and blood groups among their descendants may be difficult to detect.

Second, the sample of each individual’s ge- nome studied here, 25 blood group alleles and 47 RFLP alleles, is still relatively small. A larger number of loci may yield higher correlations. Indeed, this was shown to be the case in a previous study comparing RFLP and blood group variation with genea- logical kinship (OBrien et al., 1994b).

Third, a complete profile of European an- cestry was not available for most subject.s, since some terminal ancestors were born in the United States.

Fourth, it could be argued that the use of an R matrix based on modern European gene frequencies in equation 1 is inappropriate, since most of the European ancestors under consideration were born in the 19th century. It is obviously impossible to gauge the effects of recent changes in gene frequencies di- rectly, although many studies have indicated that these gene frequencies reveal historical relationships among European populations (Cavalli-Sforza et al., 1993, 1994; Sokal et al., 1991, 1993). An analysis in which the R matrix was not used as a weighting factor

(i.e., the term rkl was removed from equation 1) yielded uniformly lower correlations be- tween ancestral kinship and all other types of kinship. Thus inclusion of the R matrix does improve the observed correlations.

Finally, and probably most importantly, this population has been mating at random, with some avoidance of consanguinity, for several generations (Jorde, 1989). Intersub- division mobility has been very high (Jorde, 1982, 1984). The result is that nearly all study subjects have a highly mixed profile of European ancestry. This will tend to blur the ethnic differences among individuals.

In a study of immunoglobulin haplotypes in European-Americans, Stevenson and Schanfield (1981) found a high level of ge- netic similarity between indigenous Europe- ans and Americans whose ancestors origi- nated in a European country. An important characteristic of their study design was that they sampled only European Americans whose four grandparents came from the same European country. This avoids the ran- dom mating effects observed in the pres- ent study.

In summary, this study demonstrated a high correlation between ancestral kinship and isonymy-derived kinship, reflecting the influence of non-uniform settlement pat- terns of Utah by different ethnic groups. The lack of correlation between ancestral kinship and kinship based on DNA and blood groups is probably due to several factors, the most important of which is the effect of several generations of random mating.

ACKNOWLEDGMENTS We are grateful for comments and assis-

tance from Alan Rogers and Scott Watkins. This work was supported by NSF grants BNS-8720330 and DBS-9209262 and by NIH grant CA42014 (Utah Cancer Center).

LITERATURE CITED Arrin@on L.J, Ritton D (1979) The Mormon Experience:

A History ofthe Latter-Day Saints. New York: Knopf. Bamshad M, Fraley AE, Crawford MH, Cann RL, Busi

HR, Naidu JM, Jorde LB (1996) MtDNA variation in caste populations of Andhra Pradesh, India. Hum. Biol. (in press).

Bitton D, Irving G (1976) The continental inheritance. In Papanikolas HZ fed): The Peoples of Utah. Salt Lake City: Utah State Historical Society, pp. 221-250.

Bowcock AM, Kidd JR, Mountain JL, Hebert JM, Caro- tenuto I,, Kidd KK, Cavalli-Sforza LL (1991) Drift, admixture. and selection in human evolution: A studv with DNA pulymorphisms. Proc. Natl. Acad. Sci. USA 88t839-843.

61 4 E. OBRIEN ET AL

Bowcock AM, Ruiz-Linares A, Tomfohrde J , Minch E, Kidd JR, Cavalli-Sforza LL (1994) High rcsolution of human evolutionary trees with polymorphic microsa- tellites. Nature 368:455-457.

Cann RL, Stoneking M, Wilson AC (1987) Mitochondria1 DNA and human evolution. Nature 325t31-36.

Cavalli-Sforza LL, Menozzi P, Piazza A (1993) Demic expansions and human evolution. Science 25.9t639-646.

Cavalli-Sforza LL, Menozzi P, Piazza A (1994) The His- tory and Geography of Human Gcnes. Princeton: Princeton University Press.

Deka R, Chakraborty R, Ferrell RE (1991) A population genetic study of six VNTR loci in three ethnically de- fined populations. Genomics 11.83-92.

Deka R, Jin L, Shriver MD, Yu LM, DeCroo S, Hun- drieser J , Bunker CH, Ferrel RE, Chakraborty R (1995) Population genetics of dinucleotide (dC-dA)n. (dG-dT)n polymorphisms in world populations. Am. J . Hum. Gcnet. 56:461-474.

Di Rienzo A, Peterson AC, Garza JC, Valdes AM, Slatkin M, Freimer NB (1994) Mutational processes of simple- sequence repeat loci in human populations. Proc. Natl. Acad. Sci. USA 91:3166-3170.

Edwards A, Hammond HA, Jin L, Caskey CT, Chakra- borty R (7992) Genetic variation at fivc trimeric and tetrameric tandem repeat loci in four human popula- tion groups. Genomics 12t241-253.

Harpending H, Jenkins T i 1973) Genetic distance among Southern African populations. In Crawford MH, Workman PL (eds.1: Methods and Theories of Anthro- pological Genctics. Albuquerque: University of New Mexico Press, pp. 177-199.

Harpending HC, Sherry ST, Rogers AR, Stoneking M (1993) The genetic structure of ancient human popula- tions. Curr. Anthropol. 343483-496.

Horai S, Kondo R, Nakagawa-Hattori Y, Hayashi W, Sonoda S, Tajima K (1993) Pcopling of the Amcricas, founded by four major lineages ofmitochondrial DNA. Mol. Bid. Evol. 10:23-47.

.Jorde LB (1982) The genetic structure of the Utah Mor- mons: Migration analysis. Hum. Biol. 54~583-597.

Jorde LB (1984) A comparison of parent-offspring and marital migration data as measures of gene flow. In Boyce AJ (ed.): Migration and Mobility: Biosocial As- pects of Human Movcment. London: Taylor and Fran- cis, pp. 83-96.

Jorde LR (1989) Inbreeding in the Utah Mormons: An evaluation of estimates based on pedigrees, isonymy, and migration matrices. Ann. Hum. Genet. 53:339-355.

Jorde LB, Morgan I< (1987) Genetic structure of the Utah Mormons: isonymy analysis. Am. J. Phys. An- thropol. 72:403-412.

Jorde LB, Bamshad MJ, Watkins WS, Zenger R, Fraley AE, Krakowiak PA, Carpenter KD, Soodyall H, Jen- kins T, Rogers AR (19951 Origins and affinities ofmod- ern humans: a comparison of mitochondrial and nu- clear genetic data. Am. J. Hum. Genet. 57.523-538.

Kidd JR, Black FL, Weiss KM, Balazs I, Kidd KK(1991) Studies of three Amerindian populations using nu- clear DNA polymorphisms. Hum. Biol. 63t775-794.

Mantel N (1967) The detection of disease clustering and a generalized regression approach. Cancer Res. 27.209-220.

Martinson JJ, Harding RM, Philippon G, Sainte-Marie

FF, Roux J, Boyce AJ, Clegg J B (1993) Demographic reductions and genetic bottlenecks in humans: Minis- atellite distributions in Oceania. Hum. Genet. 91:445450.

McLcllan T, Jorde LB, Skolnick MH (1984) Genetic dis- tances between thc Utah Mormons and related popu- lations. Am. J . Hum. Genet. 36t836-857.

Morton NE, Dick HM, Allan NC, Izatt MM, Hill R, Yee S (1977) Bioassay of kinship in northwestern Europe. AM. Hum. Genct. 47.249-255.

Mountain JL, Cavalli-Sforza LL (1994) Inference of hu- man evolution throueh cladstic analvsis of nuclear DNA restriction poly&rphisms. Proc. Natl. Acad. Sci. USA YIt6515-6519.

Mourant AE, Kopec AC, Domeniewska-Sobczak K(1976) The Distribution of Human Blood Groups. London: Oxford University Press.

Mulder W (1976) Scandinavian saga. In Papanikolas HZ (ed): The Peoples of Utah. Salt Lake City: Utah State Historical Society, pp. 141-185.

Nei M, Roychoudhury AK (1982) Genetic relationship and evolution of human races. Evol. Biol. 14:1-59.

O’Brien E, Kerber RA, Jorde LB, Rogcrs AR (1994a) Founder effect: An assessmcnt of variation in genetic contribution among founders. Hum. Biol. 66:185-204.

OBrien E, Rogers AR, Beesley J, Jorde LB (199413) Ge- netic structure of the Utah Mormons: A comparison of results based on DNA, blood groups, migration matrices, isonymy, and pcdigrees. Hum. Biol. 66: 743-759.

Rogers AR, Jorde LB 11995) Genetic evidcnce on the origin of modern humans. Hum. Bid. 67:l-36.

Smousc PE, Long JC, Sokal RR (1986) Multiple regres- sion and correlation extensions of thc Mantel test 01 matrix correspondence. Syst. Zool. 35~627-632.

Sokal RR, Jacyucz GM, Oden NL, DiGiovanni D, Falsetti AB, McGee E, Thomson BA (1993) Genetic relation- ships of European populations reflect their ethnohist- orical affinities. Am. J . Phys. Anthropol. 91r55-70.

Snkal RR, Oden NL, Wilson C (1991) Genetic evidence for the sprcad of agriculture in Europe by demic diffu- sion. Nature 351t143-145.

Spurdle AB, Hammer MF, Jenkins T 11994) The Y Alu polymorphism in Southern African populations and its relationship to other Y-specific polymorphisms. Am. J. Hum. Genct. 54:319-330.

Stevenson JC, Schanfield MS (1981) Immunoglobulin allotypes in European populations: 111. Gm, Am, and Km allotypes in people of European ancestry in the United States. Hum. Biol. 53521-542.

Stoneking M, Jorde LB, Bhatia K, Wilson AC (1990) Geographic variation of human mitochondrial DNA from Papua New Guinea. Genetics 124:717-733.

Torroni A, Chen Y, Semino 0, Santachiara-Bcneceretti AS, Scott CR, Lott MT, Winter M, Wallace DC (1994) mtDNA and Y-chromosome polymorphisms in four Native American populations from southern Mexico. Am. J. Hum. Genet. 54303-318.

Torroni A, Semino 0, Scozzari R, Sirugo G, Spedini G, Abbas N, Fellow M, Benecerreti ASS (199O)Y chromo- some DNA polyrnorphisms in human populations: dif- ferences between Caucasoids and Africans detected by 49a and 49f probes. Ann. Hum. Genet. 54:287-296.

Watkins WS, Ramshad MJ, Jorde LB (1995) Population genetics of trinucleotide repeat polymorphisms. Hum. Molec. Genet. 4t1485-1491.