Origins of Greek Civilization: Minoans and Mycenaeans

Genetic history of Calabrian Greeks reveals ancient events and long term isolation in the Aspromonte area of Southern Italy

Stefania Sarno,
Rosalba Petrilli,
Paolo Abondio,
Andrea De Giovanni,
Alessio Boattini,
Marco Sazzini,
Sara De Fanti,
Elisabetta Cilli,
Graziella Ciani,
Davide Gentilini,
Davide Pettener,
Giovanni Romeo,
Cristina Giuliani &
Donata Luiselli

Scientific Reports volume 11, Article number: 3045 (2021)

Abstract

Calabrian Greeks are an enigmatic population that have preserved and evolved a unique variety of language, Greco, survived in the isolated Aspromonte mountain area of Southern Italy. To understand their genetic ancestry and explore possible effects of geographic and cultural isolation, we genome-wide genotyped a large set of South Italian samples including both communities that still speak Greco nowadays and those that lost the use of this language earlier in time. Comparisons with modern and ancient populations highlighted ancient, long-lasting genetic links with Eastern Mediterranean and Caucasian/Near-Eastern groups as ancestral sources of Southern Italians. Our results suggest that the Aspromonte communities might be interpreted as genetically drifted remnants that departed from such ancient genetic background as a consequence of long-term isolation. Specific patterns of population structuring and higher levels of genetic drift were indeed observed in these populations, reflecting geographic isolation amplified by cultural differences in the groups that still conserve the Greco language. Isolation and drift also affected the current genetic differentiation at specific gene pathways, prompting for future genome-wide association studies aimed at exploring trait-related loci that have drifted up in frequency in these isolated groups.
Download PDF
Introduction

The Italian Peninsula represents a key area of investigation to explore population and demographic processes that characterized the peopling history of Europe and the Mediterranean, and to reconstruct patterns of genetic diversity at different geographical scales1. The genomic variability among present-day Italians is indeed due to the multi-layered mosaic of pre-historical and historical processes of migration and admixture that interested the continent throughout time, but also to its valuable diversity and richness in terms of ecological complexity and cultural heterogeneity. The existence of distinct environmental pressures and climate conditions were proved to have forced different patterns of local adaptations between the Northern and Southern Italy, contributing to the observed genetic structure2,3. In the same way, geographic constraints or cultural factors (e.g. language, ethnicity, socio-economic structure) were suggested to have additionally influenced human population variability along the Peninsula, being responsible for different paths of isolation and differentiation or population mobility at more fine-grained local levels4,5,6.

The distinctive genomic variability of Italy has been largely investigated by using both uniparental markers7,8,9,10, autosomal SNP-chip data2,11,12,13 and, more recently, through one of the first study on whole genome sequencing3. In this context, the analysis of geographically isolated populations or of cultural enclaves and ethno-linguistic minorities can provide a simplified observatory for exploring population relationships within and among human groups. Indeed, the condition of isolation might have helped reducing the confounding effects of admixture14,15. Furthermore, the founding event and the limited external gene flow, ultimately resulting in smaller effective population size (Ne) and increased homozygosity and linkage disequilibrium (LD), may help the study of potentially trait-associated alleles found at higher frequency in these groups, thus making population isolates key models for genome-wide association studies16,17,18,19,20. In line with this viewpoint, some of the geographic and cultural isolates today settled in the Italian territory have long attracted the attention of population genetic studies based on both uniparental and autosomal markers4,5,21,22,23,24,25,26.


Figure 1

(a) Sampling map showing the approximate geographic location of analyzed populations. Sampling points are color-coded according to the province of origin: Benevento (blue); Castrovillari (purple); Catanzaro (magenta); previously collected samples from Reggio Calabria (orange); newly collected samples from Reggio Calabria (gold). The two enlarged boxes detail the sampling locations of villages in the province of Reggio Calabria (left) and in the province of Catanzaro (right), respectively. (b) Historical map showing the approximate extension of the National Park of the Aspromonte mountain area (in pink) as well as the range of the Greek-speaking area at different time periods as reported in the legend at the top-left. Geographical map has been generated with the package RgoogleMaps [v. 1.4.1]92 (URL: www.jstatsoft.org/v63/i04/) of the software R [v. 3.2.4] (https://www.r-project.org/).

Recently, a genomic survey on patterns of ancient and recent admixture in Southern Italy and the Mediterranean27 brought specific interest on the enigmatic Calabrian Greek-speaking communities residing in the Aspromonte mountain area of Bovesia, in the territory of Reggio Calabria (Southern Italy). These ethno-linguistic groups represent extant Hellenic islands in Southern Italy that still preserve a unique variety of Greek, also known as Greco or Calabrian Greek. Linguistic studies have supplied interesting information about the possible origins of these communities, historically counterpoising two main antithetic hypotheses. The first hypothesis leads back to the Medieval period and suggests that this language might derive from the descendants of Byzantines who settled Southern Italy between the fifth and eleventh centuries28,29,30,31,32. The second hypothesis instead brings to the Magna Graecia colonization of Southern Italy in the eighth century BC to trace the origin of the language. Thereafter, the Greek of Calabria would have been uninterruptedly spoken during the centuries until the present, with local developments33,34,35. More recently, reconciling scenarios tried to mitigate the initial controversy between the “Hellenic” vs. “Byzantine” dichotomy, mostly in virtue of two considerations: (i) the impact of intense Greek-Romance linguistic contacts into a dynamic model of coexistence and cohabitation between Greek-speaking Latins and Latin-speaking Greeks; and (ii) the implications of archaic lexical elements shared with peripheral Greek dialects such as those of Cyprus and the Dodecanese, to assess the importance of linguistic contributes from the Greek of different periods36,37. The new moderated hypotheses in fact consider the Greek of Magna Graecia to have been survived in diglossia with Latin during the Roman Empire and then to have been rekindled in the Byzantine era and subsequent periods. Importantly, multiple strata which repeatedly brought waves of Greek speakers onto the Calabrian coasts were suggested to have overall contributed to the Greek heritage of the region38. Historically, it is known that the Greek presence in Calabria was continuous since ancient times and that the area of Greek-influence in Southern Italy was originally more largely extended with respect to the enclaves present nowadays39. In fact, the number of Greco-speaking people is rather limited at present and today the language survives mainly in few communities residing in the Aspromonte mountain area of Reggio Calabria40,41.

Our previous characterization of Calabrian Greek communities for approximately 150,000 genome-wide SNPs with the Illumina GenoChip 2.0 DNA Ancestry chip suggested possible signs of genetic drift experienced by these groups27. In the present study, we further address issues of geographic and cultural isolation by using a higher in-depth level of analysis, which was achieved both by increasing the number of analyzed markers (720 K) and by genotyping also other communities from the same geographic area of Southern Italy (Fig. 1a). In particular, we significantly expanded the population samples from the Aspromonte mountain area by including six additional communities (Amendolea, Africo, San Lorenzo, Cardeto, Samo, San Luca) to the five already collected previously (Bova, Roghudi, Roccaforte del Greco, Gallicianò, Condofuri), therefore doubling the representativeness of the genetic structure of the area. The sampling strategy specifically covered both the communities that still preserve the Greco language as well as those from the same geographically isolated area of the Aspromonte that lost the use of this language earlier in time (Fig. 1b). This more extended sampling should therefore mirror the progressive restriction of the area of Greek-influence in the Southern Calabrian territory of Reggio Calabria. The groups from the Aspromonte mountain area were finally compared with newly-collected samples coming from a similar, but less isolated geographical context, which encompasses four villages from the Calabrian province of Catanzaro (Girifalco, Jacurso, Pentone, Tiriolo), as well as with ‘open’ (i.e. not isolated) Southern Italian groups from Castrovillari (Northern Calabria, Southern Italy) and Benevento (Campania, Southern Italy) (Fig. 1a). In this context, the aim of this study is to investigate the past population events and the local demographic factors that significantly contributed to the current genetic differentiation of Calabrian Greeks. In particular, by comparing their allelic architecture to the more general Southern Italian population we looked for the effect of geographic and cultural isolation on the detected genetic structure.

Local patterns of isolation and genetic differentiation within Southern Italy

With the aim of directly exploring local patterns of genetic differentiation, we focused more specifically on the genetic structure observed within Southern Italy. Consistently with the global analysis, the PCA applied exclusively on our Southern Italian “local” dataset replicates the distinctiveness observed for the Aspromonte group (Suppl. Figure S5a). Most of the populations from that area indeed depart from the rest of Southern Italy along the first PC, by forming a scattered pattern in which the communities that still preserve the Greco language appear as the most differentiated (i.e. occupying more peripheral positions in the PCA plot). Similarly, ADMIXTURE results for the best value of K = 2 (Suppl. Figure S5b) identify an ancestral genetic component which is maximized in almost all the current Greco-speaking communities (100% Roghudi and Gallicianò, 88% Roccaforte Del Greco, 73% Condofuri) and accounts for 30–60% of ancestry also in the core of the other communities from the same Aspromonte group.

To formally test signals of isolation and drift, we analyzed patterns of within-population genetic variation by calculating the inbreeding coefficient (Fin) and the genome-wide homozygosity (Fhom) indexes, and by analyzing the number and the extension of genomic runs of homozygosity (ROH). Both Fin and Fhom values are averagely higher for the populations from the Aspromonte area compared to the rest of Southern Italy (Suppl. Table S3), and accordingly the distribution of genome-wide homozygosity (Fhom) shows higher variance in inbreeding for the Aspromonte group than for the other Southern Italian populations (Fig. 3a), thus reflecting higher isolation patterns differentiating these communities. Similarly, a much higher number (NSEG) and length (KB) of ROH has been observed on average for the Aspromonte samples from Reggio Calabria, with respect to the tendency to lower number and length of homozygous segments exhibited by the other Southern Italian populations (Fig. 3b).

Figure 3



Intra-population patterns of genetic variation in the newly analyzed Southern Italian populations. (a) Distribution of genome-wide homozygosity index Fhom and (b) total length of ROHs (KB) plotted against number of ROHs (NSEG). Color-code as follow: Benevento (blue), Castrovillari (purple), Catanzaro (magenta) and Aspromonte area of Reggio Calabria (orange).

Since levels of isolation correlate also with the degree of relatedness between and within groups46, we further estimated the extent of genome shared identically by descent (IBD) at different classes of length using the fastIBD pipeline implemented in the BEAGLE software. Overall, patterns of genetic relatedness across populations reveal higher within-population compared to between-population sharing for longer bin classes. In particular, for bins ≥ 27 cM almost all the connections are within-population and the few links between-populations appear almost exclusively limited to the communities belonging to the ASPR cluster, consistently with these groups being more closely related to each other. For shorter classes of length among-population connections then extend to all the rest of Southern Italy (Suppl. Figure S6).

To further explore signals of genetic drift and population differentiation and to link this peculiar genetic background with biological functions, we finally used the Southern Italian genetic clusters identified by FineSTRUCTURE to detect loci that may have drifted up in frequency in the isolated Aspromonte group. In particular, we compared the allele frequencies of all variants within the populations belonging to the ASPR cluster against their respective frequencies in the rest of South Italy. More precisely, as suggested by FineSTRUCTURE results (Fig. 2c, Suppl. Figure S3), we considered both the Benevento + Castrovillari (BN + CS) cluster and the one grouping individuals from Catanzaro (CZ) as representative of the “not-isolated” Southern Italian background. We then computed pairwise FST values between the above-mentioned three clusters (i.e. ASPR, CZ and BN + CS) for all the 621,755 SNPs included in our “local” dataset. Finally, we retained the 797 loci scoring in the top 1% of FST distribution in both ASPR vs. CZ and ASPR vs. BN + CS comparisons (Suppl. Figure S7, Suppl. Table S4). Enrichment analysis on the list of corresponding top genes (Suppl. Table S5) shows that the most significantly enriched Gene Ontology (GO) terms are associated with processes of “nervous system development” and with “neuron part”, “cell periphery” and “plasma membrane” of the cellular components (Suppl. Table S6).