Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Isonymy and Isolation by Distance in Italy

Author(s): I. BARRAI, A. RODRIGUEZ-LARRALDE, E. MAMOLINI and C. SCAPOLI


Source: Human Biology, Vol. 71, No. 6 (December 1999), pp. 947-961
Published by: Wayne State University Press
Stable URL: http://www.jstor.org/stable/41465794 .
Accessed: 21/06/2014 12:24

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Wayne State University Press is collaborating with JSTOR to digitize, preserve and extend access to Human
Biology.

http://www.jstor.org

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
Isonymy and Isolation by Distance in Italy

I. BARRAI,1
A. RODRIGUEZ-LARRALDE,2
E. MAMOLINI,1
ANDC. SCAPOLI1

Abstract The isonymy structure of Italywas studiedusingthesur-


namedistribution of 5,043,580privatetelephone usersselectedfroma
1996commercial CD-ROMthatcontains all 24 millionusersinthecoun-
try.The users were distributed
in 123 towns selected on a geographic
basis.The 123 townswereeitheron themaincommunication roadsof
thecountry or at theendsof suchroads.The shortest distance between
nearest neighbor townswas 5.3 km(Carrara andMassa),andthelargest
distancewas 1,136km(AostaandCastrignano del Capo). Thenumber
ofdifferent surnames foundinthewholeanalysiswas215,623.Lasker's
distance, thenegative valueofthelogarithm ofrandom isonymy between
localities,was linearlyandsignificantly correlated withthelogarithm of
geographic distance, withr = 0.63 ± 0.008.A dendrogram was built
fromthematrix of isonymy distances, usingUPGMA.It separates the
Italiantownsinto5 mainclusters: 1 inthesouthern portionofthecountry,
a secondclustertowardthecenter, and 3 in thenorthern areaof Italy.
Within eachclustersmallsubclusters withspecific geographic distribu-
tionscouldbe relatedto regional borders. Comparisons with the results
ofa previous analysis ofSwitzerland andGermany's structuresaregiven.
Fromthepresent analysisisolationbydistance emerges clearly,although
itis less strongthanin Switzerland andstronger thanin Germany. The
random of
component inbreeding estimated from isonymy indicatesthat
thesouthern areaof Italyis on averagemoreinbredthanthenorthern
area.In fact,theheterogeneityis greater inthenorthern area,particularly
in theplainofthePo River,thananywhere else in Italy.

In thesystemsof surnameattribution through thepaternalline surnamessim-


ulate neutralalleles of a gene transmitted only throughthe Y chromosome
(Yasuda and Morton 1967; Yasuda and Furusho 1971; Yasuda et al. 1974;
Zei et al. 1983, 1984) and therefore satisfy expectationsof the neutral
the
theory of evolution (Cavalli-Sforzaand Bodmer1971; Crow 1980), whichis
entirelydescribed by random geneticdrift,mutation,and migration(Kimura

1 ofBiology, ofFerrara,
ViaL.Borsari
46,1-44100
Ferrara,
2Department
Center
University
ofExperimental
Medicine, ofHuman
Laboratory Instituto
Genetics,
Italy.
Venezolano
deInves-
Científicas,
tigaciones Venezuela.
Caracas,
Human , December
Biology v.71,no.6,pp.947-961.
1999,
© 1999
Copyright State
Wayne Press,
University Detroit, 48201-1309
Michigan
KEYWORDS: ISOLATION
ISONYMY, BY DISTANCE, GER-
ITALY,SWITZERLAND,
SURNAME
MANY, DISTRIBUTION

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
948 /BARRAI
ET AL.

1983). This property of surnamestogetherwiththeirpromptavailabilityhas


madethemusefulforthestudyofpopulationstructure since 1965,whenCrow
and Mange publishedthequantitative relationbetweenisonymyand inbreed-
ing [forreferencessee Gottlieb(1983), Mascie-Tayloret al. (1985), and
Rodriguez-Larralde, Barraiet al. (1998)].
In the last 3 yearsunexpecteddevelopmentsin cd-rom and PC tech-
nologyhave made available enormousnumbersof surnamesand the capa-
bilityto analyzelargedata setsin a reasonableperiodof time.In some coun-
triesmillionsof surnamesof telephoneusers available on cd-roms can be
efficiently analyzed withthe new processors.For example,we studiedthe
isonymy structure of Switzerland(Barraiet al. 1996) using 1.7 millionsur-
namesand were able to describetheinbreedinghierarchicalstructure of that
confederation. We also describedisolationby distancein Germanyusingthe
surnames of 5.2 million private telephone users (Barrai et al. 1997;
Rodriguez-Larralde, Barraiet al. 1998). We now focusourattention on Italy.
Thereis a largeamountof literature on theisonymystructure of local
in
populations Italy. Piazza et al. (1987) studiedtherelationbetween Karlin-
McGregor'sv and thedemographicrateof immigration 103 in Italian prov-
inces usingthesurnamesof morethan9 millionprivatetelephoneusersand
foundsignificant correlation.We describehere an analysis of isonymyin
Italy.We presentthestructure parametersderivedfromthedistribution of 5
millionindividualsin 123 towns.The parameters ofgeneticstructureobtained
fromthe surnamedistribution were isonymy/, namely,4FST, Fisher's a
(Fisher1943), and Karlin-McGregor's v (Karlinand McGregor1967). How-
ever,our main purposein the presentstudyis to reporthow the isonymy
distancevarieswithgeographicdistance,as we observedin Switzerlandand
Germany(Barrai et al. 1996; Rodriguez-Larralde,Scapoli et al. 1998;
Rodriguez-Larralde, Barraiet al. 1998).

Materials and Methods


Surnames. In 1997 we obtaineda copyof thetelephoneand addressbook
of Italyin thecd-rom I-Info2.0 producedby Topware,a commercialfirm.
The cd-rom containsthefilesof about24 milliontelephoneusersregistered
by thetelephonecompanyin Italyfor1996. To assess isonymystructure as
a functionof geography,we selected123 townson themaincommunication
roadsof Italy.The averagedistancebetweennearestneighborswas 34.1 km,
witha standarddeviationof 13.7 km.
For each townthe surnamesof the individualusers were downloaded
fromthecd-rom to Database files.Using thespecificsoftwareprovidedfor
thatpurposewiththe cd-rom and witha programof extractionof private
userswritten byus, we selected5,043,580privatetelephoneusers,an average
of 41,005 userspertown.The numberof privateuserspertownrangedfrom

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
Isonymyand IsolationbyDistance in Italy/949

629 (Dobbiaco) to 858,318 (Rome) and had from293 (Castrignanodel Capo)


to 84,583 (Rome) different surnames.The totalnumberof surnamesforthe
whole sample was 215,623. Users were not distinguished by sex, and they
constituted a 9.8% samplingof theresidentsof Italy,out of a populationof
57 millionresidentsin 1994 (AnnuarioStatisticoItaliano 1995). The iden-
tification, surnamesin the 123 townsis
samplesize, and numberof different
given in Table 1. The geographicdistribution
of the townsappears in Fig-
ure 1.

Sources of Error. The commercialcd-roms are generatedby scanning


thetelephonebooks of entirenations.Spellingerrorstherefore maybe com-
mon.We haveprepared3 mainprogramsfortheselectionofprivatetelephone
users.The firstexcludes companiesand similarenterprises based on a few
words.This programis relativelyefficient;
itreducesthesetof usersto about
60% of thetotal.A secondprogramexcludesfromthesetthoserecordswith
1 of about 1,700 commercialor public names,and thisreducesthe set by
another10%. At thispointpresumedsurnamesare orderedby length,and a
visual scan permitsone to eliminateor correctforsuch prefixesas А/,De,
El Mac , Van, etc. The prefixeswere joined to the body of the name; for
example,Al Bebawi became AlBebawi,De Luca became DeLuca and so on.
This thirdprogramgivesthefinalproduct.The recordsare sortedbysurname,
and theisonymyanalysiscan proceed.
We also investigatedhow largeour erroris aftertheprivatetelephone
usersare accepted.We selectedforall Italythe 11,763 surnamesbeginning
withtheletterA and wentover themall. We found62 obvious errors.The
totalentitiesconcernedwere 80, mostof thembelongingto theclass of fre-
quency 1. There are 210,145 individualswithsurnamesbeginningwithA;
thustheerrorrateis 0.000381 or about400 errorsper 1 millionsurnames.
Thereforewe
The effectof thiserrorrateon parametersis irrelevant.
acceptedthedata producedby theselectionprogramswithoutfurther refine-
ment.We assume thatthe errorrateobservedforsurnamesbeginningwith
A is representative
of all surnames.

randomisonymybetween
Statistics. Based on the surnamedistribution,
townsi andj (/y)was estimatedas

= 2 (i)
h PkiPkß
к

wherepkiand pkjare therelativefrequenciesof surnameк in townsi andj,


respectively;the sum is over all surnames.For an orderedreview of the
quantitiesthatcan be derivedfromthe surnamedistribution, see Relethford
(1988). Note,however, that1ц is twiceLasker's coefficient
of R¿
relationship
(Lasker 1977). We note also that is a functionof thekinshipфц between

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
950 /BARRAI
ET AL.

Table 1. SampleSize N, Numberof Different


Surnames S, Fisher'sa, Karlin-
McGregor's v,Random Horizontal
Isonymy, X,andVertical
Coordinate Yfor
Coordinate
123ItalianTowns

Random
City N S a v Isonymy X Y
1 Alessandria 30,240 9,420 1,827 0.05697 0.0005471 37.0 275.5
2 Ancona 31,306 7,585 1,489 0.04540 0.0006713142.0 233.5
3 Aosta 11,878 5,382 2,691 0.18466 0.0003715 10.5 302.0
4 Arezzo 25,462 5,441 728 0.02777 0.0013730106.0 229.0
5 AscoliPiceno 13,003 2,779 513 0.03792 0.0019474143.0 212.0
6 Assisi 4,265 1,520 583 0.12022 0.0017126121.0 217.0
7 Asti 23,983 7,171 1,414 0.05565 0.0007070 28.0 276.0
8 Avellino 11,790 3,055 485 0.03949 0.0020588173.0 152.0
9 Avezzano 9,296 2,556 724 0.07217 0.0013810139.5 187.0
10 Bari 79,914 11,545 1,204 0.01484 0.0008302219.0 161.0
11 Belluno 10,590 3,438 451 0.04084 0.0022126113.0 310.0
12Benevento 10,440 2,545 615 0.05556 0.0016254171.0 159.0
13 Bergamo 36,298 9,770 1,112 0.02972 0.0008986 60.0 298.0
14 Biella 15,174 6,353 1,394 0.08411 0.0007170 25.5 295.5
15 Bologna 148,306 25,387 1,926 0.01281 0.0005193 94.5 260.5
16 Bolzano 29,383 10,236 3,368 0.10284 0.0002969 95.0 321.0
17 Brescia 59,773 13,917 2,113 0.03414 0.0004730 71.0 293.0
18 Brindisi 22,086 4,655 1,066 0.04604 0.0009373244.0 147.0
19 Campobasso 12,122 3,232 778 0.06028 0.0012843168.0 172.0
20 Canosa diPuglia 7,435 1,375 272 0.03524 0.0036684201.0 163.0
21 Carrara 16,954 4,110 581 0.03308 0.0017209 67.5 249.0
22 CasaleMonferrato 12,397 4,599 1,163 0.08575 0.0008592 33.0 282.0
23 Caserta 10,570 3,617 718 0.06357 0.0013917161.0 157.0
24 Castellamare
diStabia 14,990 2,576 200 0.01316 0.0049773165.0 145.5
delCapo
25 Castrignano 1,333 293 73 0.05138 0.0136520254.5 125.0
26 Catanzaro 18,608 3,936 605 0.03146 0.0016517216.0 93.5
27 Cesena 23,731 3,665 412 0.01794 0.0024249114.0 250.0
28 Chieti 14,482 3,952 668 0.04404 0.0014970156.5 196.0
29 Chioggia 11,692 1,842 86 0.00725 0.0115750115.0 282.0
30 Chivasso 7,401 3,749 1,170 0.13645 0.0008540 21.5 285.0
31 Civitavecchia 14,814 5,060 1,954 0.11648 0.0005117103.0 188.5
32 Comacchio 9,044 3,418 181 0.01962 0.0054950112.0 266.0
33 Como 25,904 9,108 1,031 0.03826 0.0009693 48.0 302.0
34 Cosenza 13,371 3,418 638 0.04551 0.0015662207.5 105.0
35 Cremona 24,100 6,020 1,011 0.04025 0.0009883 67.5 281.0
36 Crotone 7,533 2,139 633 0.07746 0.0015783229.0 99.0
37 Cuneo 14,879 4,782 523 0.03391 0.0019110 12.5 261.0
38 Dobbiaco 629 337 145 0.18630 0.0068850114.0 328.5
39 Fabriano 8,840 2,282 629 0.06632 0.0015898127.0 225.0
40 Fasano 8,867 1,351 230 0.02528 0.0043297231.0 153.0
41 Ferrara 43,534 7,569 1,050 0.02356 0.0009577100.5 271.0
42 Firenze 136,235 25,922 2,409 0.01737 0.0004150 89.0 239.0
43 Foggia 32,258 6,087 1,114 0.03336 0.0008975189.0 169.5
44 Foligno 14,127 3,554 960 0.06359 0.0010412123.5 213.5
45 Forli 31,926 5,849 611 0.01877 0.0016346109.5 252.0
46 Formia 6,308 2,309 294 0.04438 0.0034012144.5 163.0

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
Isonymyand IsolationbyDistance in Italy/951

Table1. Continued
Random
City N S a Isonymy X v Y
47 Fresinone 10,849 3,148 311 0.02786 0.0032061139.0 174.0
48 Genova 243,430 39,002 2,333 0.00949 0.0004285 42.0 260.0
49 GioiaTauro 3,140 1,043 414 0.11625 0.0024145201.5 79.5
50 Gorizia 12,302 5,331 1,846 0.13041 0.0005417143.5 305.0
51 Grosseto 22,460 6,685 1,629 0.06758 0.0006139 88.0 209.0
52 Imperia 13,502 5,500 1,262 0.08543 0.0007922 22.5 245.0
53 Isernia 3,540 1,319 409 0.10351 0.0024399159.0 174.0
54 Ivrea 8,152 4,354 2,622 0.24328 0.0003814 21.6 293.5
55 La Spezia 35,019 10,341 1,774 0.04820 0.0005634 61.5 250.0
56 LameziaTerme 8,842 1,794 250 0.02742 0.0039957207.5 93.0
57 Latina 21,079 8,053 3,545 0.14396 0.0002820127.5 169.0
58 Lecce 22,322 4,751 686 0.02980 0.0014563250.5 140.0
59 Lecco 14,590 4,056 317 0.02124 0.0031478 54.5 303.0
60 Livorno 52,869 11,556 1,989 0.03625 0.0005022 72.0 232.0
61 Lucca 14,659 3,872 715 0.04646 0.0013979 76.0 241.0
62 L'Aquila 12,688 4,078 942 0.06908 0.0010669140.5 196.0
63 Macerata 11,874 3,428 905 0.07078 0.0011043139.5 223.5
64 Mantova 15,519 5,086 1,565 0.09158 0.0006387 73.0 280.0
65 MassaCarrara 18,168 4,529 456 0.02443 0.0021928 68.5 248.0
66 Matera 12,572 2,486 333 0.02576 0.0029991215.0 147.0
67 Milano 513,552 75,523 4,657 0.00899 0.0002147 49.5 291.5
68 Modena 56,147 11,388 940 0.01645 0.0010637 85.5 265.0
69 Monfalcone 9,815 4,636 1,916 0.16330 0.0005217141.5 300.5
70 Monza 39,187 12,643 932 0.02322 0.0010727 51.0 295.0
71 Napoli 235,971 26,905 891 0.00376 0.0011222160.0 150.0
72 Novara 31,989 11,510 3,178 0.09034 0.0003146 38.0 291.5
73 NoviLigure 10,115 3,922 812 0.07429 0.0012303 39.5 270.5
74 Orvieto 5,694 1,958 748 0.11603 0.0013360109.0 208.0
75 OstiaLido 26,739 12,243 4,618 0.14724 0.0002165114.0 177.0
76 Padova 66,167 16,100 1,617 0.02384 0.0006184106.0 288.0
77 Parma 56,791 11,394 1,196 0.02061 0.0008360 73.0 270.0
78 Pavia 26,595 8,962 1,733 0.06117 0.0005768 48.0 283.0
79 Perugia 37,464 10,249 1,725 0.04401 0.0005894116.0 219.0
80 Pesaro 25,790 5,570 885 0.03315 0.0011298127.5 243.0
81 Pescara 33,149 8,472 1,594 0.04585 0.0006273157.5 199.0
82 Piacenza 32,775 7,772 934 0.02769 0.0010700 60.0 278.0
83 Piombino 12,402 3,731 1,304 0.09508 0.0007667 76.0 215.0
84 Pisa 28,632 9,617 1,996 0.06515 0.0005009 73.0 238.0
85 Pordenone 15,696 5,990 1,255 0.07401 0.0007964122.0 305.0
86 Potenza 22,711 4,393 543 0.02335 0.0018384196.0 145.0
87 Prato 49,295 10,857 1,271 0.02512 0.0007866 89.0 242.0
88 Ravenna 44,622 9,536 1,037 0.02270 0.0009641113.0 258.0
89 ReggioCalabria 35,534 5,545 475 0.01318 0.0021022195.0 68.0
90 ReggioEmilia 42,181 8,330 671 0.01565 0.0014905 79.5 267.0
91 Rieti 12,229 3,118 516 0.04047 0.0019347 127.0 198.0
92 Rimini 34,738 7,899 742 0.02090 0.0013467120.5 247.0
93 Roma 858,318 84,583 4,545 0.00527 0.0002200118.5 182.0
94 RossanoCalabro 3,820 1,048 276 0.06718 0.0036209216.0 113.6

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
952 /BARRAI
ET AL.

Table1. Continued
Random
City N S a Isonymy X v Y
95 Rovigo 14,035 3,612 854 0.05730 0.0011707104.5 278.0
96 Salerno 33,620 6,534 868 0.02516 0.0011509172.0 145.0
97 SanBenedetto 12,888 3,318 597 0.04420 0.0016749150.0 214.0
98 Sapri 1,324 731 548 0.02925 0.0018220191.5 128.0
99 Savona 24,273 8,371 1,674 0.06450 0.0005974 32.0 257.5
100Siena 19,614 5,385 1,108 0.05343 0.0009023 94.0 224.0
101Sondrio 6,630 2,557 646 0.08876 0.0015458 65.0 312.5
102Susa 1,847 1,125 806 0.30346 0.0012405 3.5 284.0
103Taranto 51,291 9,442 1,449 0.02747 0.0006897229.0 142.0
104Tarvisio 1,721 1,112 906 0.34306 0.0011108142.5 322.0
105Teramo 9,194 2,907 847 0.08432 0.0011796146.0 205.0
106Terni 33,194 7,365 1,419 0.04098 0.0007044122.5 203.0
107Tolmezzo 2,983 1,081 279 0.08547 0.0035726130.5 318.0
108Torino 308,090 51,235 5,922 0.01886 0.0001689 16.5 281.0
109Trento 30,773 7,367 875 0.02762 0.0011428 91.0 308.5
110Treviso 24,802 7,925 1,398 0.05334 0.0007150114.0 296.0
111Trieste 90,055 22,529 4,098 0.04352 0.0002440147.0 295.0
112Udine 30,054 10,202 2,454 0.07548 0.0004075135.0 308.0
113Urbino 4,417 1,650 612 0.12150 0.0016339122.0 237.0
114Varese 27,041 10,571 1,662 0.05788 0.0006016 42.0 302.5
115Vasto 8,639 2,509 534 0.05817 0.0018704169.0 189.0
116Venezia 92,754 17,574 1,026 0.01093 0.0009747115.5 289.0
117Ventimiglia 9,078 4,361 1,241 0.12022 0.0008054 13.0 243.0
118Verbania 10,208 4,614 1,944 0.15992 0.0005143 37.0 306.0
119Vercelli 15,787 6,068 1,858 0.10528 0.0005380 32.5 288.0
120Verona 77,966 16,669 1,932 0.02418 0.0005175 87.5 289.5
121Vicenza 32,922 9,223 1,852 0.05323 0.0005399 99.0 293.0
122Viterbo 16,825 5,225 1,239 0.06858 0.0008065110.0 198.0
123Vittorio
Veneto 8,590 2,393 238 0.02695 0.0041852115.0 305.0

townsi and y, as shownby Rodriguez-Larralde, Scapoli et al. (1998). We


defineLasker's distance as -log (/y) (Rodriguez-Larralde,Scapoli et al.
1998).
Isonymydistancebetweentownsi andj was also estimatedusingNei' s
standardizeddistance(Nei 1973) and Cavalli-Sforzaand Edwards's (1969)
Euclideandistance.
The lineargeographicdistancebetweentowns i and j was estimated
usingthecoordinatesof each townas the referencepoint.Productmoment
correlationcoefficients
were estimatedbetweenthelogarithmic transforma-
tionsof geographicdistanceand the 3 different geneticdistances,and their
significancewas assessed using Mantel's testto comparematrices(Smouse

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
I sonymyand IsolationbyDistance in Italy/953

1. MapofItaly
Figure withthesampling Alltowns
points. areonmaincommunication
roads
or
attheendsofsuchroads.
Thedashedlinesenclose
theclusters
defined
bythedendro-
gram 4).SeeTable1fortown
(seeFigure
analysis names.

et al. 1986), with 1,000 bootstraps,each of size 7,000, randomlyextracted


fromthedistancematrices.
A dendrogramwas constructed fromthe matrixof Lasker's distances
usingtheunweightedpair groupmethodwitharithmetic mean (upgma) for
a graphicalrepresentation ofthesurnamerelationship betweendifferent
towns
in Italy.We underlinethatthedendrogram indicatesonlysurnamesimilarity,
and no phylogenetic amongtownsis implied.Most of thetowns
relationship
wereextantwiththeirpresentnamelong beforesurnameswereinventedand
used.

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
954 /BARRAI
ET AL.

2. Log-log
Figure distribution
(natural oftheoccurrence
logarithms) inItaly
ofsurnames ina
of5 million
sample individuals
in1996.

Results and Discussion

Most Frequent Surnames. The log-logfrequencydistribution of theoc-


currenceof surnames(Fox and Lasker 1983) is givenin Figure2. In Table 2
we give the 40 mostfrequentsurnamesin the sample. The tradesurnames
are practicallythesame as thoseobservedin Germanyand Switzerland.The
mostfrequentsurnameis Rossi (22,637), whichis graphicallyuniformand
ubiquitouswiththe southernvariantRusso (10,713) and a constellationof
derivativessuch as Rossetti,Rossini,and Rossotto.This groupof surnames
probablyoriginatesfromskin and hair traits,such as Bianchi (9,025) and
Ricci [8,200,curly;southernvariants,Rizzi (2,673) and Rizzo (3,461)]. Next
comes Ferrari(11,594, blacksmith),probablyas polyphyleticas the trade.
Surelypolyphyletic is Esposito (10,351), whichis correlatedwithsouthern
geography.Esposito is a surnamethattraditionally was given to newborns
abandonedto charities.A detailedgeographicdistribution of the mostfre-
quent Italiansurnames is givenby De Felice (1980).

IsonymyParameters in ItaUan Towns. In Table 1 we give the sample


size and theisonymyparametersin the 123 townsstudiedhere.In thesame
linearscale and a numericalcode are
table linearcoordinatesin an arbitrary
also given foridentification.The minimumsample size was in Dobbiaco,
with629 individualsand 337 surnames,and themaximumsample size was

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
Isonymyand IsolationbyDistance in Italy/955

Table 2. Forty
MostFrequent in a Sampleof 5,031,000
Surnames ItalianTelephone
Users
Surname Frequency
Villa 3,375
Montanari 3,443
Martinelli 3,447
Grassi 3,458
Rizzo 3,461
Leone 3,489
Valentini 3,504
Marchetti 3,596
Ferri 3,723
Santoro 3,723
Morelli 3,733
Marini 3,741
Rinaldi 3,952
Fontana 3,981
Galli 4,149
Deangelis3 4,208
Desantisa 4,264
Martini 4,269
Mariani 4,334
Greco 4,338
Fabbri 4,449
Giordano 4,535
Marino 4,602
Colombo 4,684
Bruno 4,694
Costa 4,810
Gallo 4,907
Barbieri 4,955
Moretti 4,988
Lombardi 5,013
Deluca8 5,066
Conti 6,010
Mancini 6,262
Romano 6,366
Ricci 8,200
Bianchi 9,025
Esposito 10,351
Russo 10,713
Ferrari 11,594
Rossi 22,637
a. Pooled
with
De Santis,
De Angelis,
De Luca,respectively.

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
956 /BARRAI
ET AL.

the megalopolisof Rome with858,318 privatetelephoneusers and 84,583


different surnames.We have alreadydiscussedelsewherethecorrelationof
sample size and surnamenumberwithisonymyparameters(Scapoli et al.
1997). These werecomputedalso forItaly,and theyconfirmthoseobserved
in Sicily (Rodriguez-Larraldeet al. 1994), Switzerland(Barraiet al. 1996),
and Germany(Barraiet al. 1997; Rodriguez-Larralde, Barraiet al. 1998).
We believe thatsufficientindicators of structure are a and its inverse,
in
unbiasedisonymy.Again, a descriptivestudy likethe presentone we prefer
a because it indicatespromptlythe amountof surnamedifferentiation in a
group.We recallthata is equivalentto theeffective in
allele number a genetic
system,and we have called it the effectivesurnamenumber,namely,the
numberof surnameshavingthe same frequencyand resultingin the same
isonymylevel as theone observed.
Minimuma was computedin Castrignanodel Capo (73), locatedat the
tipof theheel of thepeninsula,and maximuma was computedin theindus-
trialtownof Torinoin thenorthwest (5,922). These values can be compared
witha forthewhole country, whichwas 5,855. The variationseemsconsid-
erable comparedwithothercases, wherelarge sample sizes were studied,
such as Sicily and Switzerland,indicatingconsiderableheterogeneity in the
Italianisonymystructure. The large value of a forthecountryis indicative
of recentimmigration to Italyfromabroad.

Isolation by Distance. The correlationcoefficient and its standarderror


betweenthelogarithmic transformationof geographicdistanceand Lasker's
distancewas 0.63 ± 0.008, and the correspondingvalues for Nei' s and
Cavalli-Sforzaand Edwards'sdistanceswere0.51 ± 0.009 and 0.57 ± 0.01,
respectively. Because in previousstudieswe used Lasker's distance,therest
of the analysis was performedwiththis measureof isonymydistancefor
further comparisons.Figure3 shows a graphicrepresentation of therelation-
ship between the of
logarithm geographic distance and Lasker's distance
based on 7,503 points.It clearlydescribesthe isolationbetweentownsas
geographicdistanceincreases.
The smallestisonymydistancefoundin thisanalysiswas thatobserved
betweenNapoli and Castellamaredi Stabia,witha distanceof 6.63 unitsand
just 26 km apart.Cesena is the townmostrelatedto all theothers,withan
averageLasker's distanceof 8.67 units( ± 0.681 standarddeviation),whereas
the least relatedtown is Dobbiaco, withan average distanceof 9.89 units
( ± 0.436), locatedtowardthenortheastern portionof Italy,close to German-
speakingAustria.The highestdistanceobservedwas 11.61 units,between
townsCastrignanodel Capo and Tolmezzo, whichare 865 km apart.

Cladistic of Italian Towns. The associationof isonymydistancewith


geographic distance prompted to construct
us a dendrogramof Italiantowns

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
Isonymyand IsolationbyDistance in Italy/957

3. Isolation
Figure bydistance 5 million
inItaly(123towns, Scatter
individuals). diagramof
Lasker's
distance
overthelogarithmofgeographic inkilometers.
distance Notethe
Lasker's
diminishing atlarge
distance distances,
geographic internal
indicating migra-
which
tion, inItaly
takes from
placemostly thesouthtothenorth.

based on thematrixof Lasker's distance.The dendrogram is givenin Figure


4. Five main clustersare clearlyidentifiedin the figure.The firstclusteris
formedby 33 towns,whichare localized towardthesouthernportionof Italy
(see Figure1). Napoli, Reggio Calabria,Lecce, and Bari are includedin this
group.A secondclusterjoins thefirstclusterat a distanceof 8.97 units,which
is localized towardthe centralportionof Italy and is formedby 23 towns.
Rome, Ravenna, Rimini,Perugia,and Viterboare includedhere. A third
clusterjoins these 2 clustersat 9.25 units,formedby 33 townsdistributed
fromcentralItaly to the borderwith Switzerland.Pisa, Firenze,Bologna,
Ferrara,Como, Milano, Sondrio,and Verbania are includedin this group.
The nextclusterto join the previousclusters,at a distanceof 9.30 units,is
formedby 16 townsat the northwestern partof Italy.Torino,Genova, and
Aosta are representatives of thiscluster.A fifth
clusterformedby 15 towns
in thenortheastern portionofItalyjoins thedendrogram at 9.55 units.Rovigo,
Padova, Venice, Trieste,and Tarvisioare includedin thisgroup.Finally,at
a distanceof 9.66 units,Bolzano, Dobbiaco, and Trento,join thedendrogram.
The firstfiveclustersare identifiedby dashedlines and RomannumeralsI to
V in Figure 1. The separationscorrespondfairlywell to thegeographicdis-
tribution of geneticmarkers(Barbujaniand Sokal 1991; Zei et al. 1993). In

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
958 /BARRAI
ET AL.

4. Mainclusters
Figure onthebasisofthematrix
identified ofLasker's Thehighest
distances.
isinnorthern
heterogeneity Italy.

general,townsthatare at shortgeographicdistancestendto clusterin the


same subgroups;such is the case forMilano and Monza, Alessandriaand
Casale Monferrato, Cesena and Rimini,and Pordenoneand VittorioVeneto,
all pairs at distancesbelow 30 km. Nevertheless,thereare borderregions
towardthe centralportionof Italy (Ascoli Piceno and Teramo),towardthe
north(Ferraraand Rovigo,Comacchio and Ravenna),towardthenorthwest-
ernportionof Italy(Novara and Vercelli),or towardthenortheasternportion
(Verona and Vicenza) whereeach of the 2 townsis includedin a different
cluster,even if theyare in geographicproximity.
As alreadynoted,theclus-
terscorrelatewiththegenetic,historical,and geographicbordersof Italy.

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
Isonymyand IsolationbyDistance in Italy/959

Conclusions

An equivalentsource of information (cd-rom of telephoneusers) and


thesame methodology describedherewas used to analyzetheisonymystruc-
ture of Switzerlandand Germany(Barrai et al. 1996, 1997; Rodriguez-
Larralde,Scapoli et al. 1998; Rodriguez-Larralde, Barraiet al. 1998). In those
countriesthe correlationsbetweenLasker's distanceand the logarithmof
geographicdistancewere 0.72 ± 0.028 and 0.51 ± 0.010, respectively,
whereasthecorrelationfoundin Italyis 0.63. Also, the smallestand largest
average Lasker's distances,when all possible comparisonsare made, are
widerapartin Italy(smallest,8.67; largest,9.89) and Switzerland(smallest,
7.54; largest,8.85) thanin Germany(smallest,7.42; largest,8.37). These
differences can be explainedby the alpine topographyof Switzerlandand
Switzerland's4 different languagesandbytheelongatedgeographicstructure
ofItalycomparedwiththerelativehomogeneity ofGermanyfroma linguistic
and geographicviewpoint.The fitting of a surfaceto the values of a, the
inverseof randomisonymy,indicatesthaton averagethesouthern regionsof
Italyare moreinbredthanthenorthern regions(see Figure5).
The findingthatLasker's distanceis positivelyand significantly cor-
relatedwith geographicdistanceis easily explained if isonymydecreases
accordingto Malécoťs exponentialmodel of kinshipdecay with distance
(Malécot 1948, 1959, 1966).
The dendrogrambuiltfromthe matrixof isonymydistanceseparates
Italian townsinto 5 main clusters:1 towardthe south,anothertowardthe
center,and 3 towardthenorthern portionof thecountry.Withineach cluster
subclusterscan be identified. Lasker's distancesin Italyare on averagelarger
thanthoseobservedin Germany,whichis fairlyuniform, and also largerthan
in Switzerland,wherelanguagesand geographytendto favorisolation.
The isonymystructure of Italy shows considerableheterogeneity. It is
of some interestto notethattheheterogeneity is largerin thenorthern area,
particularlyin theplain of thePo River,thananywhereelse in Italy.

AcknowledgmentsThis researchwas supportedby grantsfromMinistero


e RicercaScientifica
dell'Università e Tecnologica(MURST), by Agreements
132.36.1(Italy)and PI-117 (Venezuela),by ConsiglioNazionaledelle Ricerche/
ConsejoNacionalde InvestigacionesCientificay Tecnólogicas(CNR/CONICIT)
1998-2000,andbyCNR through grants95.02119.CT04and97.04001. CT04

Received13 October1998; revision 8 February


received 1999.

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
960 /BARRAI
ET AL.

5. Three-dimensional
Figure ofFisher's
representation in1996.Thesouthern
a inItaly of
part
Italyis relatively
more than
inbred thenorthern which
part, haslarger
values
ofa.
Sample size,5.2million.

Literature Cited

Annuario Italiano.
Statistico 1995.Rome, Istituto
Italy: CentralediStatistica.
Barbujani,G.,andR.R.Sokal.1991.Genetic population ofItaly.II. Physical
structure and
cultural togeneflow.
barriers Am.J.Hum.Genet. 48:398-411.
I.,C. Scapoli,
Barrai, M.Berettaetal. 1996.Isonymy andthegenetic ofSwitzerland.
structure
I. Thedistributionsofsurnames. Ann.Hum.Biol.23:431^4-55.
I.,C. Scapoli,
Barrai, M. Berettaetal. 1997.Isolationbydistance inGermany. Hum.Genet.
100:684.
L.L.,andW.Bodmer.
Cavalli-Sforza, 1971.Human Population Genetics. CA:
SanFrancisco,
W.H.Freeman.
L.L.,andA.W.F.Edwards.
Cavalli-Sforza, 1969.Phylogenetic analysismodels andestimation
procedures.Am.J.Hum.Genet. 19:233-257.
Crow,J.F.1980.Theestimation ofinbreeding from isonymy. Hum.Biol.52:1-4.
Crow,J.F.,andA.P.Mange. 1965.Measurements ofinbreeding fromthefrequency ofmarriages
between persons ofthesamesurnames. Eugen. Q. 12:199-203.
De Felice,E. 1980.1 CognomiItaliani.Bologna,Italy:SocietàEditriceII Mulino.
R.A.1943.Therelation
Fisher, between thenumber ofspecies andthenumber ofindividuals
ina random sample ofanimal population.J.Anim. Ecol.12:42-58.

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions
Isonymyand IsolationbyDistance in Italy/961

Fox,W.R.,andG.W.Lasker.1983.Thedistribution ofsurname frequencies. Int.Stat.Rev.


51:81-87.
K.,ed.1983.Surnames
Gottlieb, inmodels ofinbreeding andmigration. Hum.Biol.55:1-512.
S.,andJ.McGregor.
Karlin, 1967.Thenumber ofmutant forms maintained ina population. In
Proceedings oftheFifth Berkeley Symposium onMathematics , Statistics
, andProbabil-
ity, v.4,415-438.
Kimura,M. 1983.TheNeutral Theory ofMolecular Evolution.Cambridge, England: Cambridge
University Press.
Lasker,G.W.1977.A coefficient ofrelationship byisonymy: A method forestimating the
genetic relationship between populations. Hum.Biol.49:489-493.
Malécot,G. 1948.Lesmathématiques del'hérédité. Paris,France: Masson.
Malécot,G. 1959.Lesmodeles stocastiquesengenetique depopulation. Pubi.Inst.Stat.(Univ.
Paris)8:173-210.
Malécot,G. 1966.Identical lociandrelationship. InProceedings oftheFifth Berkeley Sym-
posium onMathematics , Statistics
, andProbability, v.4,317-332.
Mascie-Taylor, C.G.N.,A.J.Boyce, andG.Brush. 1985.Mapsanddiagrams of100surnames
inEngland andWales:Anappendix. InSurnames andGenetic Structure, G.W.Lasker,
ed.Cambridge, England: Cambridge University Press.
Nei,M. 1973.Thetheory andestimation ofgenetic InGenetic
distance. Structure ofPopula-
tions, N.E.Morton, ed.Honolulu, HI:University PressofHawaii, 45-54.
Piazza,A.,S. Rendine, G.Zeietal. 1987.Migration ratesofhuman populations from surname
distributions. Nature 329:714-716.
Relethford,J.H.1988.Estimation ofkinship andgenetic distance from surnames. Hum.Biol.
60:475-492.
Rodriguez-Larralde, A.,I. Barrai, С. Nestietal. 1998.Isonymy andisolation bydistance in
Germany. Hum.Biol.70:1041-1056.
Rodriguez-Larralde, A.,A. Pavesi, С. Scapolietal. 1994.Isonymy andthegenetic structureof
Sicily. J.Biosoc.Sci.26:9-24.
Rodriguez-Larralde, A.,C. Scapoli, M. Beretta etal. 1998.Isonymy andthegenetic structure
ofSwitzerland. II. Isolation bydistance. Ann.Hum.Biol.25:533-590.
С.,A. Rodriguez-Larralde,
Scapoli, M. Beretta etal. 1997.Correlations between isonymy pa-
rameters. Int.J.Anthropol. 12:17-37.
Smouse, P.E.,J.C.Long,andR.R.Sokal.1986.Multiple regressionandcorrelation extensions
oftheMantel testofmatrix correspondence. Syst.Zool.35:627-632.
Yasuda,N.,andT. Furusho. 1971.Random andnonrandom inbreeding revealed from isonymy
study. I. Smallcities inJapan. Am.J.Hum.Genet. 23:303-316.
Yasuda,N.,andN.E.Morton. 1967.Studies onhuman populationstructure.InThird Interna-
tional Congress ofHuman Genetics, J.F.CrowandJ.V.Neel,eds.Baltimore, MD:Johns
Hopkins University Press, 249-265.
Yasuda,N.,L.L.Cavalli-Sforza, M. Skolnick etal. 1974.Theevolution ofsurnames: Ananal-
ysisoftheir distribution andextinction. Theor. Popul.Biol.5:123-142.
Zei,G.,A.Piazza,andL.L.Cavalli-Sforza. 1984.Geographic analysis ofsurname distributions
inSardinia: A testforneutrality. Atti Assoc.Genet. Ital.30:247.
Zei,G.,G. Barbujani, A. Lisaetal. 1993.Barriers togeneflowestimated bysurname distri-
bution inItaly.Ann.Hum.Genet. 57:123-140.
Zei,G.,R.G.Matessi, E. Sirietal. 1983.Surnames inSardinia.I. Fitoffrequency distributions
forneutral allelesandgenetic population structure.
Ann.Hum.Genet. 47:329-352.

This content downloaded from 195.78.108.174 on Sat, 21 Jun 2014 12:24:06 PM


All use subject to JSTOR Terms and Conditions

You might also like