Population structure analysis

We assessed population structure using PLINK v. 1.9 [19] to carry out a principal component analysis (PCA) and the R software v.4.1.3. was employed for visualizing the resulting plot. Considering the large number of breeds and samples, the same software was used to plot the centroids of the principal components 1 and 2 for each breed, and such values were used to construct the PCA presented in the main and Additional Figures. Moreover, population structure was investigated with the ADMIXTURE v.1.3.0 package [21] with number of clusters (K) varying from 2 to 15. To assess the quality of the clustering process and thus infer the most likely K-value, we estimated the cross-validation error for each K-value. To visualize the results of the ADMIXTURE analysis, we used the Pophelper R package [22].

Diversity of European goat populations is not correlated with distance to Ganj Dareh

We investigated whether H_o and H_e values of African, European, and Asian populations show significant correlations (r) with distance from their sampling location to Ganj Dareh. When analyzing goat populations from Europe (Fig. 3a), we obtained negative and significant correlations (H_o: r= − 0.47, P=0.002, Fig. 3a; He: r=− 0.40, P=0.01, Fig. 3a) for both heterozygosity values. However, these two correlations became non-significant (H_o: r=− 0.22, P=0.24, Fig. 3a; H_e: r=− 0.22, P=0.22, Fig. 3a) when British and Irish populations were removed from the European data set. Indeed, the majority of European breeds displayed moderate to high heterozygosity values (Fig. 3a), with the exception of the populations from United Kingdom (H_o=0.29; H_e=0.32) and Ireland (H_o=0.35; H_e=0.37). Even the Spanish Bermeya and Malagueña breeds, which are located very far apart from Ganj Dareh, displayed high heterozygosities (H_o=0.41; H_e=0.40 in Bermeya and H_o=0.42; H_e=0.42 in Malagueña). On the other hand, correlations between F_ST values and distance to Ganj Dareh were positive and significant when insular populations were included in the analysis (r=0.37, P=0.02), but became non-significant (r=0.28, P=0.12) when such populations were removed from the analysis (see Additional file 1: Figure S5a). Moreover, correlations between F_ROH values and distance to Ganj Dareh with (r=0.13, P-value=0.52) or without (r=0.06, P-value=0.79) islands were non-significant (see Additional file 1: Figure S6a), and the interpolation map (see Additional file 1: Figure S7) and list (see Table 1) of F_is values evidenced that they are, in general, weak and negative.

Open in a separate window

Fig. 3

Graphs depicting the relationships between observed and expected heterozygosities of European, African and Asian goat populations and distance between their sampling locations and Ganj Dareh. Graphs depicting the relationships (expressed as Pearson correlations and their P-values) between observed heterozygosity and expected heterozygosity and distance from Ganj Dareh (early Neolithic settlement in the Zagros Mountains representative of the geographic coordinates of the areas of early goat management in the Fertile Crescent) to sampling locations of a European breeds, including and not including insular populations, b African breeds, including and not including insular populations, c Asian populations. In all plots, country of origin is indicated with specific colours. Breed acronyms are listed in Tables ​Tables1,1, ​,22 and ​and33

Such results do not fully match those of Cañón et al. [7], who described a decrease in caprine genetic diversity from the south-east to the north-west of Europe. This could be due to the limited number of microsatellite markers used by Cañón et al. [7], but also to the fact that Cañón et al. [7] had a much broader collection of Eastern European goat breeds than us. The significant gradient that we observe when British and Irish populations are included in the analysis might be due to their strong demographic recession [37], which is reflected by their high levels of homozygosity [29]. However, we cannot rule out the possibility that the low diversity of British and Irish cattle is partly explained by one or more founder effects associated with the arrival of livestock to the United Kingdom and Ireland 5800–6000 YBP, as suggested for British cattle [38].

The lack of a significant gradient of diversity in European goat breeds could be due to post-domestication migratory movements associated with trading and herding. Throughout the millennia, the Mediterranean Sea has facilitated the exchange of goods and livestock via a dense network of commercial maritime routes connecting distant port cities within and outside Europe. Indeed, Cardoso et al. [29] reported that goats from Mediterranean islands have lower levels of homozygosity than those from remote islands as Iceland, La Palma or Madagascar. In addition, the Great European Plain, which is one of the largest continuous expanses of plain on the Earth's surface, may have facilitated the exchange of goats and other livestock amongst distant locations within Europe. This interpretation is supported by the mostly negative F_is values shown in the corresponding interpolation map (see Additional file 1 Figure S7), which are compatible with a slight excess of heterozygosity. Even more, in recent times the widespread use of improved breeds (e.g., Saanen, Toggenburg and Alpine), and artificial insemination might also have contributed to increasing gene flow between distant European populations. Besides, there is evidence that these highly productive cosmopolitan breeds have introgressed many local breeds in Europe [30].

Detection of a significant gradient of diversity associated to distance to Ganj Dareh in African goats

In contrast with European goats, significant negative correlations between the diversity of African caprine populations and distances to Ganj Dareh have been observed in the data sets with (Madagascar and La Palma) and without islands (Fig. 3b). Indeed, we obtained correlation coefficients of − 0.46 (H_o, P=0.0044) and − 0.49 (H_e, P=0.0023) in the data set with no islands (Fig. 3b) and correlation coefficients of − 0.51 (H_o, P=0.00079) and − 0.53 (H_e, P=0.00043) in the data set with islands (Fig. 3b). Consistently, the magnitude of F_ST coefficients was highly correlated with distance from the African sampling sites to Ganj Dareh for both data sets with (r=0.57, P=0.00011) and without (r=0.62, P=0.000045) islands (see Additional file 1: Figure S5b). The correlation between F_ROH and distance to Ganj Dareh was not significant (r=0.23, P=0.21) when insular populations were excluded from the analysis, while it became significant (r=0.38, P=0.025) when Malagasy goats were taken into consideration (see Additional file 1: Figure S6b). This result could be anticipated because Malagasy goats have high F_ROH coefficients, probably because of the occurrence of a strong founder effect [29]. Based on these results and the interpolation map (see Additional file 1: Figure S7) and list (See Table 2) displaying F_is values, which are mostly close to zero and negative (except in North Africa), we conclude that the decrease of diversity associated to distance to Ganj Dareh observed in African breeds is not caused by a parallel augment of inbreeding.

We have observed that the Egyptian, Algerian, and Sudanese populations, which are closest to the Fertile Crescent, show the highest heterozygosity values (see Table 2). When proceeding southwards and particularly south-eastwards, diversity decreases, as evidenced in goat breeds from Mozambique (H_o=0.33; H_e=0.34) and Malawi (H_o=0.35; H_e=0.37), and particularly in the island of Madagascar (H_o=0.31; H_e=0.33). With regard to indigenous South African breeds, their diversity is high (H_o=0.39; H_e=0.42), probably because many of these breeds have been introgressed by Boer goats. The Boer breed has a mixed Asian and African ancestry [30], and there is evidence that Anglo-Nubian bucks contributed to its foundation [4].

The dispersal of livestock by land is expected to take place through a series of founder effects, thus generating gradients of decreasing diversity and increasing genetic differentiation as the ones observed in our work. In contrast, when domestic animals are transported by sea it is more likely to observe a leap-frog pattern of diffusion that does not necessarily result in genetic clines of differentiation or diversity. In consequence, the detection of a gradient of diversity (H_o and H_e) and genetic differentiation (F_ST) associated with distance to Ganj Dareh in African goats is consistent with an overland rather than maritime post-domestication dispersal of goats throughout the African continent, with the only exception of the North African shoreline where maritime diffusion throughout the Mediterranean Sea was important [39] as attested by remains of impressed pottery, crop plants and sheep, goats, and cattle remains found in archaeological sites in Lybia, Algeria and Morocco [40]. The predominant overland spread of domesticates in Africa (when compared to Europe) might be explained by the fact that the surfaces of Europe and Africa are about 10 million km² and 30 million km², respectively, while their coastal lines are 30,000 km (Africa) and 143,000 km (Europe) long [41, 42]. Besides, in Africa there is a relative scarcity of natural harbors and long navigable river systems, the latter due to the ruggedness of the terrain, with rapids and waterfalls as well as shallow river points, strong seasonal fluctuations in water flow, siltation, and sedimentation in lower reaches [41]. This means that the inner parts of the African continent are less easily accessible by navigation than European inland, making transportation of livestock difficult.

The early entry of goats in Africa probably took place in North Africa through the Sinai Peninsula as well as through the Mediterranean Sea [6], coinciding with the opening of a grassland niche in the Sahara that was gradually occupied by pastoral communities [6]. The increasing aridity of the Sahara around 4500 YBP and the consequent southward retreat of the Tsetse fly belt favored the migration of herders towards the Sahel. However, the entry of livestock into West and East Africa took place not before than 3500 YBP or even later [40], possibly because of a lack of immunity to endemic diseases. Goat and sheep remains dating back to 2400 YBP and 2100 YBP have been found at the sites of Salumano (Zambia) and Bamba (Zimbabwe), proving that the arrival of small ruminants to Southern Africa is quite recent [6]. This might have involved migrations through and along the coastal areas of the Congo Basin or facilitated by the opening of Tsetse corridors along the highland of the Rift Valley [40, 43]

As shown in Fig. 4, goats from Central and East Africa are less diverse than their Northern counterparts, possibly because the Sahara Desert, which covers 9.1 million km², constitutes a formidable geographical barrier to the southwards spread of pastoral communities and their livestock [44]. Moreover, Central Africa overlaps with the Tsetse fly belt, which covers a geographic area of 10 million km², between latitudes 14° N and 20° S, representing about one third of the African continent. Trypanosomiasis is a protozoan disease which causes anemia, fever, and weight loss and sometimes can be fatal, representing a heavy economic burden to African countries in which this infection is endemic [45]. Susceptibility to this parasite may have limited the diffusion and exchange of caprine stocks in Tsetse fly infested areas. Interestingly, Traorè and coworkers showed that the presence of the Tsetse fly influences the genetic variability of goats from Burkina-Faso, and they demonstrated that trypanosomiasis might have acted as a landscape boundary both for the spread of trypanosensitive goats and for strong selection pressure on trypanotolerant goats in infested areas [46].

Open in a separate window

Fig. 4

Interpolation maps showing the geographic distribution of observed and expected heterozygosities in African, European and Asian breeds. Interpolation maps showing the distribution of genetic diversity in African, European and Asian breeds. a Observed heterozygosity, H_o. b Expected heterozygosity, H_e. Blue points represent sampling localities in a and b, respectively. In Europe, a reduction of diversity is evident in goats from the United Kingdom and Ireland, while in Africa low diversity coincides with the Tsetse fly belt (a geographic area comprised between latitudes 14° N and 20° S) and Madagascar. In Asia, low variation is detected in Pakistan and Southern China

We have detected a high variability of several South African indigenous breeds even though this region remained considerably isolated from Asia and Europe [47]. We excluded from the gradient analysis South African commercial goats (Boer, Kalahari Red and Savanna) because it is well known that Boer goats have a mixed African and Asian ancestry [16, 30], and that Kalahari and Savanna goats have a strong Boer component. We kept in our analysis indigenous communal populations sampled in the main goat-producing provinces of South Africa (Limpopo, Freestate, Gauteng, Northwest), which happened to have high levels of heterozygosity. This could be due to the fact that these South African populations have been also introgressed to some extent by Boer goats as well as by goats of European origin. Indeed, the establishment, in South Africa, of British and Dutch farmers, during the seventeenth-nineteenth centuries, promoted the development or importation of highly productive breeds to improve the local stocks [4].

Thanks to Dr. Hans Lenstra (Utrecht University, Netherlands) for his valuable contribution to obtain, analyze and discuss the data. We also acknowledge the sample providers of the AdaptMap project: James Kijas, CSIRO, Australia; Bernt Guldbrandtsen, Aarhus University, Denmark; Juha Kantanen, Luke, Finland; Dylan Duby, Museum National d’Histoire Naturelle, France; Pierre Martin, Capgenes, France; Coralie Danchin, Delphine Duclos, Institut de l’Elevage, France; Daniel Allain, Rémy Arquet, Nathalie Mandonnet, Michel Naves, Isabelle Palhière, Rachel Rupp, INRAE, France and CABRICOOP breeders; François Pompanon, LECA, France; Hamid R. Rezaei, Gorgan University of Agricultural Sciences and Natural Resources, Iran; Sean Carolan and Maeve Foran, Old Irish Goats Society, Ireland; Alessandra Stella, IBBA-CNR, Italy; Paolo Ajmone-Marsan, Licia Colli, BioDNA, Università Cattolica del Sacro Cuore, Italian Goat Consortium, Italy; Alessandra Crisà, Council for Agricultural Research and Economics (CREA), Italian Goat Consortium, Italy; Donata Marletta, University of Catania, Italian Goat Consortium, Italy; Paola Crepaldi, University of Milan, Italian Goat Consortium, Italy; Michele Ottino, Parco Nazionale del Gran Paradiso, Italy; Ettore Randi, ISPRA Istituto Superiore per la Protezione e la Ricerca Ambientale, Italy; Badr Benjelloun, INRA Maroc, Morocco; Hans Lenstra, University of Utrecht, The Netherlands; Muhammad Moaeen-ud-Din and Jim Reecy, PMAS-Arid Agriculture University, Pakistan; Isabel Alvarez, Área de Genética y Reproducción Animal del Serida, Spain; Marcel Amills and Armand Sànchez, Centre for Research in Agricultural Genomics (CRAG), Spain; Juan Capote, Instituto Canario de Investigaciones Agrarias (ICIA), Spain; Jordi Jordana, Universitat Autònoma de Barcelona, Spain; Agueda Pons, Serveis de Millora Agrària i Pesquera (SEMILLA), Illes Balears, Spain; Amparo Martínez and Antonio Molina, University of Córdoba, Spain; Benjamin Rosen, USDA/ARS, USA; Carina Visser, Faculty of Natural and Agricultural Sciences, South Africa; Cord Drögemüller, University of Bern, Switzerland; Gordon Luikart, Montana Conservation Genomics lab (MCGL), Division of Biological Sciences, University of Montana, Missoula, USA; Clet Wandui Masiga, Tropical Institute of Development Innovations (TRIDI), Uganda; Denis Fidalis Mujibi, International Livestock Research Institute (ILRI), Nairobi Kenya; Hassan Ally Mruttu, Ministry of Livestock and Fisheries Development, Tanzania; Timothy Gondwe, Department of Animal Science, Lilongwe University of Agriculture and Natural Resources, Malawi; Joseph Sikosana, Department of Research and Specialist Services, Division of Livestock Research, Zimbabwe; Maria Da Gloria Taela, Animal Production Institute, Ministry of Agriculture, Maputo, Mozambique; Oyekan Nash, National Biotechnology Development Agency, Nigeria.