Nous sommes tous des Toscans...

Si les éventuels forumistes généticiens, ethnologues ou autres scientifiques pouvaient nous livrer la substantifique moëlle de cet article que je viens de dénicher... (bon courage)

Citation :

Genetic History of the Population of Corsica (Western Mediterranean) as Inferred from Autosomal STR Analysis

Tofanelli, Sergio; Taglioli, Luca; Varesi, Laurent; Paoli, Giorgio
8276 mots
1 avril 2004
Human Biology
229
Volume 76; Issue 2; ISSN: 00187143
Anglais
Copyright (c) 2004 Bell & Howell Information and Learning Company. All rights reserved.

Abstract

To genetically reconstruct the demographic history of the human population of Corsica (western Mediterranean), we analyzed the variability at eight autosomal STR loci (FES, VWA, CSF1PO, TH01, F13A1, TPOX, CD4, and D3S1358) in a sample of 179 native blood donors from 4 out of the 5 administrative districts. The main line of genetic discontinuity inferred from the spatial distribution of STR variability overlapped the linguistic and geographic boundaries. In the innermost areas (Corte district) several estimators had larger stochastic effects on allele frequencies. Genetic distance measures underlying different evolutionary models all pointed to a higher variability within Corsicans than within the rest of the Mediterranean reference populations. All Corsican subsamples showed the highest distance with a pooled sample from central Sardinia, thus making recent gene flow between the two neighboring islands unlikely. Hierarchical AMOVA and distance-based multivariate genetic spaces stressed the closeness of Tuscan and Corsican frequency distributions, which could reflect peopling events with different time depths. Anyway, estimated separation times well support the linguistic hypothesis that Neolithic/Chalcolithic events have been far more important than Paleolithic or historical processes in the shaping of present Corsican variability.

Population History, Language, and Genetics of Corsica

The present population of Corsica, the third largest island in the western Mediterranean, consists of 260,149 individuals (INSEE, 1999) dispersed over a territory of 8,680 km^sup 2^ (30 inhabitants/km^sup 2^). Corsica is administratively subdivided into two departments, Haute Corse and Corse du Sud, and five arrondissements, related to the cities of Bastia, Calvi, Corte, Ajaccio, and Sartene (Figure Ia). Administrative boundaries are based on different historical, geographic, and linguistic backgrounds (Figures 1b and 1c).

The uniqueness of Corsican genetics is intimately tied to the geography of the island: The distance from continental Europe is large enough to have caused long-term isolation; its strategic position along the main Mediterranean routes led to its coasts being occupied by people of heterogeneous origin since prehistorical times; and the altitude and harshness of the landscape have facilitated the fragmentation of the inland population into small, self-sustaining nuclei.

Population History. The first peopling of Corsica is still a debated issue. Although the crossing of human groups through the Tuscan archipelago and Corsica during the Last Glacial Maximum has been suggested to explain the human phalanx found in Sardinia (20,000 yr B.P.) (Sondaar et al. 1995), Corsican evidence of Paleolithic humans is still to be discovered. The oldest anthropological remains date to the pre-Neolithic (9th millennium B.p.) and presumably belonged to scattered groups of hunter-gatherers with unspecialized cultures (Weiss 1999). They were replaced by larger groups showing the influence of the Western Cardial (8th-7th millennium B.p.), the Neolithic pottery style common to the Tyrrhenian continental area and found also in the Tuscan archipelago (Pianosa island) and in Sardinia (Tozzi and Weiss 2000).

Megaliths (dolmen, statue-menhir) and fortified architectures (torre, castelli a torra) from the Chalcolithic (5,000-4,000 yr B.P.) and the Bronze Age (4,000-3,000 yr B.P.) suggest population growth, more stable settlements, a more intense exploitation of inner lands, and a rural elite-based society with cultural links to Gallura (northern Sardinia) and continental Italy.

In historical times Corsica saw a long series of sporadic occupations (by, among others, Phocaeans, Phoenicians, Carthaginians, Syracusans, Etruscans, Vandals, Saracens, Aragonese, Austrians, and British) and more stable political or military colonization (by Romans, Byzantines, Pisans, and Genovese). Pisa, which ruled the island from A.D. 1077 to 1288, spread the Catholic religion over the whole island with the consequent neo-Latinization of the language. Corsica was definitively annexed to France in 1789.

In every historical period the orography and poor communications conditioned the distribution and relationships of the inhabitants and hampered a true demographic development, finally producing a fiercely proud and dispersed native population. Most of the hinterland has been dominated over a long time by a self-sustaining economy based on a patriarchal feudal system.

The Language. The ability to trace the evolution of the Corsican idiom, or Corsu, has been limited by a lack of written sources. The Roman historian Seneca wrote in A.D. 40-41 that "Corsicans speak a rough and incomprehensible idiom," thus suggesting the presence of a pre-Latin spoken language of uncertain origin. A recent interpretation of this archaic substratum (Alinei 2000) claims a linguistic continuity with ancient Indo-European Italid variants of Tuscan-Ligurian origin with later influences of Sardinian (for southern Corsica), Central Italian (for the southeast), and Celtic (for the whole island). As soon as the first massive wave of Christianization came to the island in the 11th century, the Tuscan language was rapidly and widely absorbed but, according to the same theory, it would have never overwhelmed the Corsican prehistoric background. With few localized exceptions subsequent governments exerted little influence on the structure of the medieval Corsu.

Present Corsican is classified into the Southern Romantic subbranch of the Indo-European family, in the Tuscan group of Italian varieties (Grimes 1996). Independent studies have identified two main lexical clusters corresponding to the northern and southern halves of the island: the Cismontano, with clear Tuscan influence, and the Oltremontano, with more conservative features (Melillo 1971; Calafell et al. 1996) (see Figure 1c). Four dialects have been recognized: Northern Corsican, Venachese, Ajaccio, and Sartenese. The geographic dispersal of these dialects coincides approximately with the arrondissements of Bastia, Corte, Ajaccio, and Sartene, respectively.

Previous Genetic Studies. The genetic structure and affinities of Corsicans have only recently been investigated. Nevertheless, the level of within-population heterogeneity and the genetic relationships with other Mediterranean populations, in particular with the closest Sardinians, still remain highly controversial matters.

Different levels of heterogeneity among Corsican geographic districts have been identified using different genetic systems. A remarkable differentiation, compared to other continental Mediterranean populations, has been observed by means of traditional markers (Vona et al. 2002), NRY haplogroups (Scozzari et al. 2001), HLA antigens (Grimaldi et al. 2002), and isonymy (Morelli et al. 2002). Conversely, genetic homogeneity was found for 3 blood groups in 10 regions (Calafell et al. 1996).

Several researchers claim a common genetic legacy for Corsicans and Sardinians. Phylogenetic trees based on blood proteins (Varesi et al. 1996; Memmi et al. 1998; Vona et al. 2002), HLA class I markers (Grimaldi et al. 2002), and mitochondrial HVS-I sequences (Varesi et al. 2000) place the two island populations on the same branch, separated from other continental Mediterranean populations. According to some investigators, the results can be reasonably explained by the persistence in the two islands of the same Paleolithic genetic background. However, studies based on NRY haplogroup variability (Scozzari et al. 2001; Francalacci et al. 2003), Alu insertions (Moral et al. 1999), and mtDNA coding regions (Morelli et al. 2000) suggest much earlier origins, almost no recent gene How from Corsica to Sardinia, and shared genetic features with Tuscans or Catalans.

Aims of This Study. The analysis of hypervariable DNA polymorphisms, such as autosomal short tandem repeats (STRs), has become a powerful tool to elucidate the recent demographic histories of human populations (Jorde et al. 1997; Rosenberg et al. 2002; Zhivotovsky et al. 2003). The parameters estimated from unlinked STR loci more closely depict the genetic portrait of human populations than parameters estimated from uniparentally transmitted genes, because they reflect the average contribution of several independently segregating genes, each one having undergone its own historical pathway. Moreover, absolute dates of divergence between populations can be estimated from measures of genetic distances specifically constructed for STR data (Goldstein et al. 1995a). At least, the reliability of demographic reconstructions based on STR data, on the evolutionary time scale presumed for the human colonization of the Mediterranean (< 1,500 generations), is much less affected by the confounding effect of homoplasy, because of the high-rate stepwiselike mutational mechanism and the existence of range constraints in allele size variation.

In this paper we analyzed the allele frequency distributions at eight autosomal STR loci in a widely distributed Corsican autochthonous sample that included all the recognized dialect areas. Then, we compared the observed variability with that of historically and geographically related populations. We aimed to provide a better evaluation of two still debated issues about Corsicans' genetic history: (1) the level of within-island heterogeneity and, in particular, how well the genetic structure of extant Corsicans fits the inner subdivisions distinguished by linguistic and geomorphologic data; and (2) the position of the Corsican gene pool within the chronological and spatial framework of Mediterranean peopling.

Citation :

Materials and Methods

Sampling. Peripheral blood samples were collected from 179 healthy unrelated native donors (whose ancestry has been traced for three generations) from both Corsican departments and from 4 arrondissements: 96 individuals from Haute Corse (51 from Bastia and 45 from Corte) and 83 individuals from Corse du Sud (30 from Ajaccio and 53 from Sartene). all donors gave their informed consent beforehand, and personal data have been treated anonymously.

The blood samples were frozen at - 20°C for shipment to the Laboratory of Molecular Anthropology at the University of Pisa.

DNA Analysis. Genomic DNA was isolated from 200 µl whole blood by the QIAamp Blood Mini Kit (Qiagen) following the manufacturer's instructions, and then quantified against standards on agarose minigels. Diluted DNA was PCR-amplified using 5' fluoresceinated primers. The primers were specific for the tetranucleotide STRs FES (GenBank accession number M14589), VWA (GenBank accession number M25716), CSF1PO (GenBank accession number X14720), TH01 (GenBank accession number D00269), F13A1 (GenBank accession number M21986), TPOX (GenBank accession number M68651), D3S1358 (GenBank accession number NT005997), and the pentanucleotide STR CD4 (GenBank accession number M86525). Details of PCR protocols are described elsewhere (Tofanelli et al. 2002).

PCR products were separated by capillary electrophoresis using the ABI Prism 310 Genetic Analyzer (Applied Biosystems). Fragment sizes were automatically detected using the Genescan PCR Analysis software (Applied Biosystems) and were typed by comparison with an internal size standard (Genescan 500 TAMRA, Applied Biosystems) and with allele ladders of known size. The level of accuracy of computer-generated allele sizing was estimated to be high (the coefficient of variation ranged from 0.06 for FES to 0.12 for VWA) to allow the unambiguous detection of alleles differing by 1 bp (i.e., alleles 9.3 and 10 at TH01). Allele nomenclature followed ISFG recommendations (Olaisen et al. 1998). Individual genotypes are available from the authors upon request.

Statistical Analysis. Indexes of population genetic structure were estimated by the analysis of molecular variance (AMOVA) using the Arlequin 2.0 software package (Schneider et al. 2000).

We used the Genepop 3.1d software package (Raymond and Rousset, 1995) to estimate departures from Hardy-Weinberg equilibrium using the two-tailed (probability test) and one-tailed (heterozygote excess and deficiency) exact tests with a Markov chain simulation method. For multiple testing of the same hypothesis, we applied a Bonferroni correction.

The unbiased estimate of heterozygosity averaged across loci (H) and its sampling variance were calculated according to Eqs. (8.6) and (8.7), respectively, of Nei (1987). Gene and genotype differentiation over all loci was evaluated using the G test (FSTAT program; Goudet 2001). In contrast to Fisher's (1951) method, this test allows one to combine individual p values with loci being weighted according to their level of polymorphism: The higher the polymorphism, the larger the weight (Petit et al. 2001).

Pairwise genetic distances were calculated using D^sub c^ (Cavalli Sforza and Edwards 1967), F^sub ST^ (Reynolds et al. 1983), ([delta]µ)^sup 2^ (Goldstein et al. 1995b), and D, (Feldman et al. 1996) using Microsat 1.5d (Minch et al. 1997). Statistical significance was assessed using 100 bootstraps. Distance-based patterns of genetic differentiation among populations were analyzed using principal-components analysis (PCA) and the Statistica 6.0 software package (StatSoft Italia 1997).

Reference Data. Data on STR allele frequencies from 12 Mediterranean population samples were taken as follows: (1) Spain, Aragon (Martinez Jarreta et al. 1999); (2) Spain, Basque Country (Iriondo et al. 1997; Garcia et al. 1998; ALFRED 2001 ; Perez-Lezaun et al. 2000); (3) Spain, Catalonia (Perez-Lezaun et al. 2000; Crespillo et al. 1997; Gené et al. 1997); (4) France (Pfitzinger et al. 1995; Rousselet et al. 1996; Busque et al. 1997); (5) Greece (Kondopoulou et al. 1999; Greenhalgh et al. 1994; Wall et al. 1993); (6) continental Italy (Maviglia et al. 2001); (7) Italy, Liguria (Gruppo Ematologi Forensi Italiani, personal communication, 1999; Casarino et al. 1996); ( Italy, Pisa (Domenici et al. 1998a, 1998b); (9) Italy, Etruscan Tuscany (Tofanelli et al. 2002); (10) Italy, Sardinia (Gruppo Ematologi Forensi Italiani, personal communication, 1999; Cessa et al. 1998; Vona et al. 1996; I. Boschi, personal communication, 2000); (11) Italy, Sicily (I. Boschi, personal communication, 2000; Asmundo and Crino 1998); and (12) Turkey (Alper et al. 1995; Akbasak et al. 2001; Iwasa et al. 1997).

Results

Within-Population Analyses. The allele frequency distributions at the eight STR loci obtained for each subdivision unit are shown in Table 1. As for most European populations, they all follow a unimodal pattern, with the exception of the systems that combine a simple repeat structure with a low mutation rate, TH01, TPOX, and CD4, which are slightly bimodal.

Gene diversity was higher in Corse du Sud than in Haute Corse (H = 0.744 ± 0.019 vs. H = 0.731 ± 0.019) and was higher in Ajaccio (H = 0.760 ± 0.018) than in the other arrondissements (Bastia, H = 0.730 ± 0.019; Corte, H = 0.734 ± 0.022; Sartene, H = 0.737 ± 0.021).

AMOVA failed to demonstrate a significant apportionment of the overall genetic diversity among groups of populations defined as arrondissements or departments. However, a differentiation between departments was detected by G-based tests for both allele (p = 0.030) and genotype (p = 0.033) frequency distributions.

To more thoroughly test the hypothesis of random mating within the island, we verified Hardy-Weinberg equilibrium by means of the standard probability test (P) and one-tailed tests for heterozygote excess (EE) and deficiency (DE) (Table 2). By applying the Bonferroni correction for multiple tests under the same hypothesis, we found that any locus in any subpopulation fell below the significance level (0.05/24 = 0.0021). However, if we consider the Corsican sample as a whole, the probability for the multilocus P test was at the conventional limit of confidence (p = 0.049) and the deviations were mainly due to heterozygote deficiency in Haute Corse (DE test, p = 0.035). A deeper analysis of the innermost district of Corte (data not shown) demonstrated that the major contribution to deviations from Hardy-Weinberg equilibrium was the high rate of homozygotes in the regions of Bozio and Venachese, which are reasonably considered independent cndogamous units.

To better understand whether the amount of genetic heterogeneity observed in the Corsican gene pool was higher or lower than average within the Mediterranean basin and to reduce some intrinsic statistical bias, we applied genetic distance measures from different mutation models with different sensitivity to the demographic parameters to a wide data set of populations potentially involved in the history of Corsican peopling (see the Materials and Methods section).

D^sub c^ (Cavalli-Sforza and Edwards 1967) is a model-free distance that is most efficient in reconstructing correct phylogenetic relationships under various evolutionary conditions because it shows a good compromise between time linearity and error size (Takezaki and Nei 1996). F^sub ST^ (Reynolds et al. 1983) is based on the infmite-allele mutation model (IAM) of Kimura and Crow (1964) and considers genetic drift as the main driving force in population differentiation, ([delta]µ)^sup 2^ (Goldstein et al. 1995b) is a measure specifically designed for microsatellite data. It is based on a strict single-step mutation model (SMM; Ohta and Kimura 1973) and is independent of population size under a mutation-drift equilibrium. D, (Feldman et al. 1996) is a measure derived from ([delta]µ)^sup 2^, which takes into account lower and higher range constraints to allele size variation and is quite robust to population fluctuations.

As reported in Table 3, all the distance methods gave the same hierarchical pattern of differentiation, with Corsican samples (arrondissements) having the highest mean distance values and non-Corsican samples the lowest values.

Interpopulation Analyses. When analyzing pairwise genetic distances between Corsicans and the other reference populations from the Mediterranean basin (average D^sub c^ values), the highest distance was with Sardinians and the lowest value occurred in continental Italians. In single-locus analyses the same order of relationships was obtained at FES and TH01. Similar hierarchies (second highest distance with Sardinians, lowest with Italians) were also obtained at CD4, CSFlPO, and F13A1. At the remaining loci (D3S1358, TPOX, VWA) the highest distance was with Basques or Catalans. When considering the single population subunits, multilocus comparisons again gave the highest D^sub c^ values with Sardinians and the lowest D^sub c^ values with Italians. When any subdivision criterion of the Corsican sample was followed, Haute Corse always appeared to be genetically closest to Tuscan samples (data not shown).

The genetic relationships among populations are better synthesized in Figure 2. The first three principal components computed on the D^sub c^ matrix account for 60.5% of the variation. Sardinians are confined at one extreme of the graph and clearly emerged as outliers within Mediterranean STR variability. The more clear-cut cluster joins Corsicans and Tuscans, whereas the other samples plot according to a geographically oriented order, with western Mediterraneans (France, Basque Country, and Aragon) on the upper right, eastern Mediterraneans (Greece, Turkey, Sicily) approaching the center, and continental Italians and Ligurians falling in between.

Hierarchical AMOVA was also used to assess Corsican genetic affinities. First, Mediterranean STR diversity was apportioned among individuals within populations and between populations. Single-locus FST values ranged from 0.00043 (p = 0.295) at CSF1PO to 0.00577 (p<0.001) at TH01. Mean FST was low (0.0256) but highly significant (Fisher's [chi]^sup 2^ = 100.17, df = 16, p< 0.00001). Multilocus pairwise F57-values between Corsicans and reference populations are reported in Table 4. If we ignore the comparisons involving continental Italians (Tuscans, Ligurians) and Catalans, the probability was always below the 0.05 confidence limit and sometimes (for Greeks, Turks, Basques, and Sardinians) even below 0.001.

Second, we performed an AMOVA by apportioning the overall diversity within and between groups of populations. Four groups were defined following criteria of geographic proximity and considering the patterns of genetic legacy inferred from principal-components and FST analyses (Figure 2; Table 4): group A, continental Italians (Pisa, Etruscan Tuscany, continental Italy, Liguria); group B, western Mediterraneans (Aragon, Catalonia, Basque Country, France); group C, eastern Mediterraneans (Sicily, Greece, Turkey); and group D, Sardinia. As expected when analyzing autosomal variability among Europeans, the betweengroup variance component ([Phi]^sub CT^) was lower than the between-population withingroup component ([Phi]^sub SC^). A significant [Phi]^sub CT^ value (Fisher's [chi]^sup 2^ = 31.52, df = 16, p < 0.001) was obtained only when Corsica was included in group A (continental Italians). When Corsica was moved to another group, [Phi]^sub CT^ decreased by 55.669.4% and became statistically nonsignificant, whereas [Phi]^sub SC^ was 17.8-20.9% higher.

To assess the most reliable time estimates of population splits from STR diversity, we calculated separation times between pairs of populations using the widely used distance for microsatellite loci ([delta]µ)^sup 2^ (Goldstein et al. 1995a). Under a mutation-drift equilibrium and a strict single-step mutation model, ([delta]µ)^sup 2^ values are expected to increase linearly with time according to the formula ([delta]µ)^sup 2^ = 2µ[tau], where µ is the mutation rate (mutations per locus per generation) and [tau] is the time in generations since separation. There are three sources of statistical error underlying this estimating procedure: (1) the between-locus error caused by sampling loci, most likely the major component because hundreds of loci are required to give stable evolutionary estimates (Zhivotovsky and Feldman 1995); (2) the within-locus error caused by multinomial sampling of individuals [in our case this type of error is considered negligible because our analysis is based on many observations (179 individuals × 8 loci = 1,432)]; and (3) the uncertainty of mutation rate estimates.

To better calibrate the mutational bias, we calculated [tau] using three estimated values of µ: 0.00165, the arithmetic mean of previously published direct estimates for tetranucleotide autosomal STRs (Weber and Wong 1993; Brinkmann et al. 1998; Henke and Henke 1999; Sajantila et al. 1999; Leopoldino and Pena 2003); 0.00127, the average mutation rate for our set of loci observed in genealogical germ-line transfers (Brinkmann et al. 1998; Henke and Henke 1999; Sajantila et al. 1999); and 0.00093, an indirect estimate of the effective mutation rate for tetrameric STRs (Zhivotovsky et al. 2000).

By assuming 25 years as the generation time, we calculated point estimates for the separation between Corsicans (pooled sample) and non-Corsicans ranging from 4,929 to 8,746 yr (Table 5), and lower bounds (taking into consideration 95% confidence limits of the earliest estimate) never exceeded 19,000 yr B.P.

The distributions of divergence estimates between pairs of populations (averaged across loci) were similarly shaped for Corsicans and non-Corsicans (Figure 3). The highest peak when ad hoc intermediate µ values (µ = 0.00127, Table 5, second row) were considered somewhat coincided with the archeological evidence of demographic expansion in Corsica: the former colonization being dated to the Early-Middle Neolithic (8th-6th millennium B.P.) and the largest growth rate to the Chalcolithic (around the 4th millennium B.P.). The oldest and largest peak in the distribution of divergence times among non-Corsicans closely agreed with the estimated time range of the waves of expansion advanced in the model of the Neolithic demie diffusion (10th-5th millennium B.P.; Ammerman and Cavalli-Sforza 1984) and in supporting genetic surveys (Sokal et al. 1991; Barbujani and Chikhi 2000; Rosser et al. 2000; Malaspina et al. 2001; Scozzari et al. 2001).

Citation :

Discussion

Genetic Structure. The analysis of autosomal STRs yielded conflicting results about the genetic structure of the Corsican gene pool. Multilocus tests used to check the hypothesis of random mating and population differentiation showed significant unpaired distribution of frequencies, but the AMOVA and single-locus Hardy-Weinberg equilibrium tests did not show this. A possible explanation is that the present pattern of genetic diversity in the Mediterranean basin has been most likely shaped by recent demographic processes rather than by the progressive accumulation of mutations on the chromosomes of isolated populations (Barbujani and Chikhi 2000). Accordingly, the differences among gene pools are more faithfully estimated by differences in allele frequencies (G test for population differentiation) than by measures of molecular distance between genotypes (AMOVA).

Nonetheless, Corsica has a level of heterogeneity higher than those in the populations currently living along the Mediterranean basin. This finding is in agreement with [Phi]^sub ST^ values previously calculated from Y-chromosome haplogroup variability (Scozzari et al. 2001), which were found to be higher in Corsicans (0.138, p < 0.0001) than in other insular Mediterraneans, either Sardinians (0.045, p < 0.0001) or Sicilians (0.036, p = 0.02).

The main genetic discontinuity was found between northern and southern districts. The northwest-southeast inner guideline where linguistic and geographic barriers largely overlap (See Figure 1) most likely acted as the main constraint to the exchange of people across the two sides of the island, ultimately unpairing gene flow and favoring a more rapid change of genetic frequencies than elsewhere.

The historical practice of transhumance (the seasonal moving of shepherds from coastal villages to mountain grazing areas) and navigation are factors that could have circumvented this main natural barrier. However, ancient sheep pastures, with their relative pathways, have been shared only among close villages on the same side (De Lanfranchi 1990), and navigation, owing to the long-term foreign occupation of the coasts, never did represent the transport of choice.

Previous analyses of the marriage structure in Corsica (Ranque et al. 1961) confirmed that a genetic unpairing persisted until recent times, with exogamy rates being the lowest and even zero between microregions belonging to different sides or arrondissements.

The higher genetic homogeneity observed in Haute Corse could be derived from geomorphological features. The southwestern side is harsher, with a system of parallel deep valleys, whereas the northeastern side is formed by a more open landscape and by marsh plains, which remained almost uninhabited until recently because of malaria (see Figure 1b).

STR results largely agree with previous studies in considering the area of Corte a peculiar genetic environment. Morelli et al. (2000) found a novel mutation that defines a subbranch of the mitochondrial haplogroup T in about 7% of the sample. Varesi et al. (2000) found 8.5% of mitochondrial D-loop sequences with characteristic transitions. Grimaldi et al. (2002) observed a haplotype frequency of 4.3% for the A*30-Cw*7-B*5801 haplotype, a rare HLA haplotype in the rest of Europe. These findings were more likely caused by local mutations that underwent a rapid increase because of drift or inbreeding than by the retaining of more conservative genetic features. The population living around Corte is, in fact, scattered in 51 villages with an average size of 180 inhabitants.

As a whole, therefore, drift seems to be the most likely cause of the excess genetic variance observed in Corsica with respect to the Mediterranean mainland.

Population Relationships. It is generally agreed that the stratification of the population structure of European populations resulted from past demographic processes with different origins, predating, contemporary with, and subsequent to the Last Glacial Maximum, around 20,000 years ago. However, the contribution of the various components to the current gene pools is still being debated. A Neolithic component has been associated with clinal patterns of genetic variation and should have consisted mainly in migration waves from the Levant or the Black Sea with associated population growth (Cavalli-Sforza et al. 1994; Excoffier and Schneider 1999; Barbujani and Chikhi 2000; Rosser et al. 2000).

As far as mtDNA is concerned, European haplogroups J, T1, and U3 have been assumed to originate from Neolithic founders and to show an east-west dine (Richards et al. 2002). In this perspective it is worth noting that central Sardinia seems to retain a mitochondrial background with prevalent Upper Paleolithic origins, sharing with Iberian populations the presence of haplogroup V and a higher incidence of H haplogroups. Conversely, Corsica and Gallura show a more recent pattern of variability, sharing with Tuscany the absence of haplogroup V and higher frequencies of haplogroup J (Morelli et al. 2000).

As regards Y-chromosome variability, a clear east-west frequency distribution has been observed for haplogroup HG9 [(classified as haplogroup J by the Y Chromosome Consortium (2002)J (Semino et al. 2000; Scozzari et al. 2001). One of the microsatellite-defined sublineages, HG9.2/YCC J2, has been associated with a coastal dispersal of Neolithic farmers and/or with historical male displacements, both from the Near East (Scozzari et al. 2001). Corsicans, especially individuals from Balagna (Calvi arrondissement, Haute Corse), showed frequencies of 9.2/J2 chromosomes higher than the frequencies for Sardinians but comparable to the frequencies found in central-southern Italians. Another Y-chromosome haplogroup, HG2.2/YCC I, which is rare or absent in the rest of Europe, was rarely observed in Corsica (1.6-3.6%) and the European mainland (1.5-6.7% in Spain, France, and Italy), whereas it always reached high frequencies (27.0-55.6%) in different samples from central and southern Sardinia. For both haplogroups the distributions found in Gallura have intermediate values.

Similarly, chromosomes I bearing the M26 mutation (haplogroup YCC Ilb2) were never observed in Corsica, whereas they are present in 36% of inner Sardinians (Francalacci et al. 2003).

Sardinia and Corsica therefore seem to have had a rather different genetic history, with Corsica being more closely associated with Tuscans and continental Italians and Sardinia showing more marked features of isolation, with some possible ancient contact with the Iberian Peninsula. Moreover, the fact that maternally and paternally transmitted genetic systems are in substantial agreement suggests that there was no sex-differential contribution in the population pathways.

The divergence between Sardinians and Corsicans that appears from autosomal STR profiles is unambiguous and consistent across loci. It substantially confirms the evolutionary pattern supported by uniparental markers; the archaic genetic background formerly shared with Sardinians, if any, has been smoothed (in Corse du Sud and Gallura) or even obscured (in Haute Corse) by the supply of novel genes from the Mediterranean mainland.

The fact that this evolutionary pattern differs from patterns found in previous studies on nuclear [blood proteins (Memmì et al. 1998) and HLA class 1 alleles (Grimaldi et al. 2002)] and mitochondrial (HVS-I) (Varesi et al. 2000) markers could be due to sampling bias, because most investigations genotyped individuals only from the innermost Corsican region: Corte and surroundings. A source of bias in the present analysis could be the underestimation of the diversity among Sardinians, here represented by a pooled sample from three different geographic and linguistic districts (Logodurese, Sassarese, Campidanese).

Anyway, what conflicts is not only the mutual position of the two island populations and their relationships with other Mediterranean populations but also the chronological window of their demographic evolution. The most likely STRbased estimates (estimate B of Table 5) place the major events of the Corsican peopling within a post-Last Glacial Maximum time span (mean value 6,400 yr, latest 95% confidence interval bound = 13,800 yr). Mitochondrial-based estimates date the first demographic expansion of Corsicans to Paleolithic times (mean value 27,600 yr and 95% confidence interval 14,400-41,600 yr; Varesi et al. 2000).

At this point we wondered whether the earlier ancestry inferred from STR variability could be a reliable estimate or whether it could be biased by some demographic process. On one hand, time estimates derived from Goldstein's distance can be affected by demographic dynamics when a mutation-drift equilibrium is not verified (Goldstein et al. 1995b; Zhivotovsky et al. 2000). In particular, ([delta]µ)^sup 2^ assumes constant population size over time and no migration from outside. As a consequence, it underestimates divergence times if populations have suffered rapid expansions or extensive gene flow. On the other hand, an overestimation of separation times could derive from stochastic effects caused by severe bottlenecks or population substructuring.

Historical sources and the genetic evidence suggest that, among these factors, size contractions coupled with deme isolation could have played a major role to distort estimated separation times. Census size might have suffered stronger fluctuations in Corsica than elsewhere (e.g., the present population living in Corsica is one-sixth that living in Sardinia), but population density had mild fluctuations during the last 2,000 years. However, the fluctuations never exceeded 31 inhabitants/km^sup 2^, and the lowest values occurred in the Middle Ages (about 12 inhabitants/km^sup 2^) because of the effects of epidemic diseases and Saracen incursions (Francalacci et al. 2003). Most likely, losses and additions of alleles caused by size variations eventually counterbalanced.

Both stochastic and sampling bias might be higher for H VS-I-derived inferences than for STR-derived inferences, given the following issues: (1) An unbalanced drift effect, which acted in the innermost communities of Corsica, could have led to larger overestimating effects for mitochondrial-based reconstructions, given the lower size and narrower geographic coverage of mtDNA-typed samples; (2) HVS-I sequences behave as a single evolutionary unit, mutate faster, and have a lower effective size, thus accumulating genetic differences between isolated populations much easier than nuclear recombining sequences; and (3) the consistency between results from estimators underlying different evolutionary conditions makes STR-based estimates better statistically supported.

The genetic affinities between Tuscan and Corsican STR profiles are much more apparent and could reflect peopling events from different time periods, because the occurrence of cultural links with Tuscany can be inferred from the Corsican archeological record since the pre-Neolithic. However, STR diversity and derived separation estimates seem to support recent linguistic hypotheses (Alinei 2000), suggesting that Neolithization would have been far more important than neo-Latinization in the differentiation of present Corsican variability.

Received 24 October 2002; revision received 31 October 2003.

Copyright Wayne State University Press Apr 2004 | SERGIO TOFANELLI,1 LUCA TAGLIOLI,1 LAURENT VARESI,2 AND GIORGIO PAOLI1 | 1 Dipartimento di Etologia, Ecologia ed Evoluzione, Unità di Antropologia, Universilà di Pisa, Via S. Maria 55, 56126 Pisa, Italy. | 2 Faculté des Sciences et Techniques, Université de Corse, 20250 Corte, France. | Human Biology, April 2004, v. 76, no. 2, pp. 229-251. | Copyright © 2004 Wayne State University Press, Detroit, Michigan 48201-1309

© 2006 Dow Jones Reuters Business Interactive, LLC (exerçant son activité sous la dénomination commerciale Factiva). Tous droits réservés.

Quand j'ai 5 min je m'y colle mais là j'ai pas trop envie de

trois pages en anglais ô fou!

Antone a écrit:

trois pages en anglais ô fou!

c'est imbuvable à lire, je dois même l'avoir archivé quelque part, j'ai pas compris grand chose ...

disons qu'il y a un paquet d'infos et d'en faire la synthèse est un peu difficile...

C'est une étude qui vise à deteminer l'origine de la population corse (qui au passage tord le coup à la notion de peuple corse) à travers des corrélations d'odre historiques et géographiques. Pour ce faire, les auteurs cherche à déterminer les distances entre differents groupuscules géographiques corses (Haute corse, Venachese, Ajaccio, Sartè) et méditerranéens Toscans, Sardes de l'interieurs et du littoral (Galluresi) au moyen d'études de leur ADN.

Première idée c'est que la diversité génétique en Corse est plus élévée que dans les autres iles méditerrannéenes... donc la consanguinité on nous la rejouera au violoncelle...

Citation :

Nonetheless, Corsica has a level of heterogeneity higher than those in the populations currently living along the Mediterranean basin. This finding is in agreement with [Phi]^sub ST^ values previously calculated from Y-chromosome haplogroup variability (Scozzari et al. 2001), which were found to be higher in Corsicans (0.138, p < 0.0001) than in other insular Mediterraneans, either Sardinians (0.045, p < 0.0001) or Sicilians (0.036, p = 0.02).

Il est dit que les populations corses et sardes ont un fonds génétique commun qui remonterait au néolithique quand les iles ont été peuplées...

Mais après évolution dûe au brassage, il en ressort que d'une manière générale, les corses sont plus proches des toscans que ne le sont les Sardes qui, eux, ont dans leur patrimoine génétique des influences ibériques...

Citation :

Sardinia and Corsica therefore seem to have had a rather different genetic history, with Corsica being more closely associated with Tuscans and continental Italians and Sardinia showing more marked features of isolation, with some possible ancient contact with the Iberian Peninsula.

ils observent également des différences entre groupuscules de poulations en Corse... la Haute Corse est génétiquement plus homogène que la Corse du Sud où les frontières naturelles limitaient le brassage, d'où la subsistance de groupuscules assez bien définis géographiquement...

Citation :

The divergence between Sardinians and Corsicans that appears from autosomal STR profiles is unambiguous and consistent across loci. It substantially confirms the evolutionary pattern supported by uniparental markers; the archaic genetic background formerly shared with Sardinians, if any, has been smoothed (in Corse du Sud and Gallura) or even obscured (in Haute Corse) by the supply of novel genes from the Mediterranean mainland.

voilà en résumé, mais la biologie moléculaire et la génétique c'est pas trop mon domaine...

Si j'ai bien compris donc, les aiatchins seraient donc fondés à nous "traiter" de gênois et nous de faire de même en les assimilant à des sardes...

on ne peut rien te cacher...

Toute l'etude ADN et proteines HLA machin-chose, les stats et tout le tintouin on s'en fout un peu... avec ton anglais, qui doit etre meilleur que le mien, tu as pas besoin d'etre scientifique pour comprendre...

bon je prends un coup là, ma vie vient de changer.

tu prends tes désirs pour la réalite.... loin de moi l'idée de vouloir te mettre un coup...

pas de malentendu grivois, depuis que je me sais vaguement Toscan, je ne me sens plus d'aise.

si tu me croyais toscan tu t'plantais

tu te fais des films mon p'tit...

Une traduction se serait possible???? c'est bien de mettre des sujets en Anglais, c'est mieux de les traduire.
Sinon qui pouvait douter de notre cousinage avec les Toscans??? , una di le più belle regione d'europa.

d'accord sur la Toscane!

Je ne vois pas en quoi cet article remet en cause la notion de peuple corse, à moins de rester sur une base ethnique, mais le concept ethnique de nation ou de peuple en sciences humaines est très largement dépassé, très très largement.

c'est une information qui étaient déja connus des historiens, sans bases scientifiques prouvées certes, dans les bons livres sur la question il est en général considéré que les corses sont issus d'une branche celto-ligure assez vagues mais plus généralement de l'italie du nord.
De même les anciennes légendes parlent aussi d'une population du nord de l'itale (qui connait le coup de la bergère "corsa" qui suit son taureau nageur ?)
D'ailleurs la taille des corses a chuté après la conquete romaine, les romains étant un peu plus petit que les liguriens et les celtes.
Tiens ça me fait penser que nos ancetres sont finalement un peu des gaulois en fait puisqu'en partie celte.
Enfin la langue corse est très proche du toscan (on dit qeu c'est un toscan du XV eme siècle qui a évolué différement), rien d'étonnant donc dans cette analyse génétique.
ça tombe bien en plus la toscane est considéré comme une des, voir la, plus belle région du monde....encore un point commun.

sans compter que les Toscans perdent aussi les adresses internet à ce qu'on raconte

il vaut mieux pas trop aller chercher ce que l'ont a dans le sang, ont pourrait être surpris, parce que les origines Toscane ne peuvent correspondre qu'à une petite partie de notre patrimoine génétique ...

L'intérêt est historique pour comprendre les mouvements de population. On ne sait pas vraiment les mouvements des temps pré et proto historiques. De même on connait mal l'impact réel des invasions. Par exemple, en France, au début du moyen âge, l'impact des invasions germaniques a dû être faible car on estime maintenant le nombre des francs, vandales et autres goths à quelques dizaines de milliers. On nes ait pas vraiment l'impact génétique (donc quantitatif) des romains en Gaule. Il est peut être faible. Pas l'impact culturel et civilisationnel bien sûr, mais celà remet en cause les thèses de romanisation par absorption. On aurait plus certainement une romanisation par croisement dans les élites et par diffusion et mobilité sociale.

je partage l'avisde Ghjattu. L'intérêt de ces études est évident mais un peuple ne se construit pas sur des données génétiques. D'autres dans le temps, on tenté cette approche redoutable et on sait ce qu'il en est advenu.

je trouve ça très intéressant de connaittre la proximité génétique des peuples, parce qu'on apprend aussi comment les "invasions" ont été vécue : peu ou pas d'influence des barbares d'outre rhin en gaule comme dit ghjattu, ce qui veut dire peu ou pas de naissances issues de coucheries avec l'envahisseur. Par contre, similitude entre les toscans pour les corses, ce qui peut être liée à une "invasion" plus douce, culturelle et économique, plutot que militaire.

Ca va très loin cette analyse des peuples, et d'ailleurs, ces méthodes d'analyses génétiques sont à l'origine d'un dépoussiérage de la notion de race dans l'espèce humaine. D'après ce que j'ai lu, les différents foyers humains connus (pour schématiser : blanc, noir, jaune, rouge) n'ont pas été séparés assez longtemps pour conduire à l'emergence de races réellement distinctes. On parle plutôt de types raciaux, ce qui n'est pas juste jouer sur les mots. Il y a quelques decennies, l'anthropologie (fortement post-colonialiste) utilisait des critères morphologiques pour ce genre d'études, comme la forme de la machoire ou des arcades sourcillières, d'où les égarements qu'on connait.

Antone a écrit:

L'intérêt de ces études est évident mais un peuple ne se construit pas sur des données génétiques. D'autres dans le temps, on tenté cette approche redoutable et on sait ce qu'il en est advenu.

Tout à fait !

POur revenir sur le point de la romanisation de la gaule et de la différence avec les francs et les goths, elle est certainement très importante pour plusieurs raisons.

1) D'abord la période très longue de brassage en gaule (5 siècles au moins)
2) le type d'échange et le mode de fonctionnement des romains, qui à la différence des grecs n'exterminent pas leur vaincus mais dans le concept d'impérium souhaite "romanisé". Ce qui se traduit par une latinisation culturel des élites mais aussi des mariages (seul alexandre chez les grecs avait fait parel en egypte), la place du commerce qui facilite les échanges même humains, les fait que le sud de la gaule était déja très latinisé (la gaule en toge à la différence de la gaule chevelu du nord), l'abscence de vison "raciale" des romains (ils s'en foutent carrément, ils ont des généraux noirs), l'étalement de leurs armées sur le continent européen ou de nombreux soldats font souches (une légion c'est 6 000 hommes, il y a eu jusqu'a 50 légions dont au moins 15 composé de pures romains), la place primordiale du sexe chez les romains (le phallus est partout ce sont des gros baiseurs surtout que les religions monothéistes ne sont pas encore trop implanté). D'ailleurs la france est considéré comme un pays latin, bien que la moitié de son térritoire soit plutot de type nord européens.