The Genomic Legacy of Human Management and sex-biased Aurochs hybridization in Iberian Cattle
Curation statements for this article:-
Curated by eLife
eLife assessment
Using genomic data from ancient and modern samples, this important study investigates the genomic history of cattle in Iberia, focusing on the admixture between domestic cattle and their wild ancestors, aurochs. The authors present solid evidence for interbreeding between domestic cattle and wild aurochs since the Neolithic period, although the extent, sex bias, and directionality of genetic flow over time remain highly unclear. The authors also show that the aurochs ancestry in cattle stabilized at ~20% since ~4000 years ago and continues into modern breeds, including the Lidia breed that is bred for aggressiveness and used in bullfighting. The work will be of interest to evolutionary biologists and quantitative geneticists who seek to understand the genomic history and genetic basis of trait variation of domesticated animals.
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
- Genetics and Genomics (eLife)
Abstract
Cattle have been a valuable economic resource and cultural icon since prehistory. From the initial expansion of domestic cattle into Europe during the Neolithic period, taurine cattle (Bos taurus) and their wild ancestor, the aurochs (B. primigenius), had overlapping ranges leading to ample opportunities for intentional and unintentional hybridization. We performed a bioarchaeological analysis of 24 Bos remains from Iberia dating from the Mesolithic to the Roman period. The archaeogenomic dataset allows us to investigate the extent of domestic-wild hybridization over time, providing insight into the species’ behavior and human management by aligning changes with cultural and genomic transitions in the archaeological record. Our results show frequent hybridization during the Neolithic and Chalcolithic, likely reflecting a mix of hunting and herding or relatively unmanaged herds, with mostly male aurochs and female domestic cattle involved in hybridization. This is supported by isotopic evidence of ecological niche sharing, with only a few domestic cattle possibly being managed. The proportion of aurochs ancestry in domestic cattle remains relatively constant from about 4000 years ago, probably due to herd management and selection against hybrids, coinciding with other cultural transitions. The constant level of wild ancestry (~20%) continues into modern western European breeds including the Spanish Lidia breed which is bred for its aggressiveness and fighting ability, but does not display elevated levels of aurochs ancestry. This study takes a genomic glance at the impact of human actions and wild introgression in the establishment of cattle as one of the most important domestic species today.
Article activity feed
-
-
-
eLife assessment
Using genomic data from ancient and modern samples, this important study investigates the genomic history of cattle in Iberia, focusing on the admixture between domestic cattle and their wild ancestors, aurochs. The authors present solid evidence for interbreeding between domestic cattle and wild aurochs since the Neolithic period, although the extent, sex bias, and directionality of genetic flow over time remain highly unclear. The authors also show that the aurochs ancestry in cattle stabilized at ~20% since ~4000 years ago and continues into modern breeds, including the Lidia breed that is bred for aggressiveness and used in bullfighting. The work will be of interest to evolutionary biologists and quantitative geneticists who seek to understand the genomic history and genetic basis of trait variation of domesticated …
eLife assessment
Using genomic data from ancient and modern samples, this important study investigates the genomic history of cattle in Iberia, focusing on the admixture between domestic cattle and their wild ancestors, aurochs. The authors present solid evidence for interbreeding between domestic cattle and wild aurochs since the Neolithic period, although the extent, sex bias, and directionality of genetic flow over time remain highly unclear. The authors also show that the aurochs ancestry in cattle stabilized at ~20% since ~4000 years ago and continues into modern breeds, including the Lidia breed that is bred for aggressiveness and used in bullfighting. The work will be of interest to evolutionary biologists and quantitative geneticists who seek to understand the genomic history and genetic basis of trait variation of domesticated animals.
-
Reviewer #1 (Public Review):
My main concern is the use of the 700K SNP dataset. This set of SNPs suffers from a heavy ascertainment bias, which can be seen in the PCA in the supplementary material where all the aurochs cluster in the center within the variation of cattle. Given the coverage of some of the samples, multiple individuals would have less than 10K SNP covered. The majority of these are unlikely to be informative here given that they would just represent fixed positions between taurine and indicine or SNPs mostly variable in milk cattle breeds. The authors would get a much better resolution (i.e. many more SNPs to work with their very low genome coverage data) using the 1000 bull genome project VCF data set: https://www.ebi.ac.uk/ena/browser/view/PRJEB42783 which based on whole genome resequencing data from many cattle. This …
Reviewer #1 (Public Review):
My main concern is the use of the 700K SNP dataset. This set of SNPs suffers from a heavy ascertainment bias, which can be seen in the PCA in the supplementary material where all the aurochs cluster in the center within the variation of cattle. Given the coverage of some of the samples, multiple individuals would have less than 10K SNP covered. The majority of these are unlikely to be informative here given that they would just represent fixed positions between taurine and indicine or SNPs mostly variable in milk cattle breeds. The authors would get a much better resolution (i.e. many more SNPs to work with their very low genome coverage data) using the 1000 bull genome project VCF data set: https://www.ebi.ac.uk/ena/browser/view/PRJEB42783 which based on whole genome resequencing data from many cattle. This will certainly help with improving the resolution of qpAdm and f4 analysis, which have huge confidence intervals in most cases. Right now some individuals have huge confidence intervals ranging from 0 to 80% auroch ancestry...
I agree with the authors that qpAdm is likely to give quite a noisy estimate of ancestry here (likely explain part of the issue I mentioned above). Although qpAdm is good for model testing here for ancestry proportion the authors instead could use an explicit f4 ratio - this would allow them to specify a model which would make the result easier to interpret.
The interpretation of the different levels of allele sharing on X vs autosome being the result of sex-bias admixture is not very convincing. Could these differences simply be due to a low recombination rate on the X chromosome and/or lower effective population size, which would lead to less efficient purifying selection?
The authors suggest that 2 pop model rejection in some domestic population might be due to indicine ancestry, this seems relatively straightforward to test.
The first sentence of the paper is a bit long-winded, also dogs were domesticated before the emergence of farming societies.
It would be good to be specific about the number of genomes and coverage info in the last paragraph of the intro.
-
Reviewer #2 (Public Review):
Summary:
In this paper, the authors investigated the admixture history of domestic cattle since they were introduced into Iberia, by studying genomic data from 24 ancient samples dated to ~2000-8000 years ago and comparing them to modern breeds. They aimed to (1) test for introgression from (local) wild aurochs into domestic cattle; (2) characterize the pattern of admixture (frequency, extent, sex bias, directionality) over time; (3) test for correlation between genetic ancestry and stable isotope levels (which are indicative of ecological niche); and (4) test for the hypothesized higher aurochs ancestry in a modern breed of fighting bulls.Strengths:
Overall, this study collects valuable new data that are useful for testing interesting hypotheses, such as admixture between domestic and wild populations, and …Reviewer #2 (Public Review):
Summary:
In this paper, the authors investigated the admixture history of domestic cattle since they were introduced into Iberia, by studying genomic data from 24 ancient samples dated to ~2000-8000 years ago and comparing them to modern breeds. They aimed to (1) test for introgression from (local) wild aurochs into domestic cattle; (2) characterize the pattern of admixture (frequency, extent, sex bias, directionality) over time; (3) test for correlation between genetic ancestry and stable isotope levels (which are indicative of ecological niche); and (4) test for the hypothesized higher aurochs ancestry in a modern breed of fighting bulls.Strengths:
Overall, this study collects valuable new data that are useful for testing interesting hypotheses, such as admixture between domestic and wild populations, and correlation between genome-wide aurochs ancestry and aggressiveness.Weaknesses:
Most conclusions are partially supported by the data presented. The presence of admixed individuals in prehistorical periods supports the hypothesized introgression, although this conclusion needs to be strengthened with an analysis of potential contamination. The frequency, sex-bias, and directionality of admixture remain highly uncertain due to limitations of the data or issues with the analysis. There is considerable overlap in stable isotope values between domestic and wild groups, indicating a shared ecological niche, but variation in classification criteria for domestic vs wild groups and in skeletal elements sampled for measurements significantly weakens this claim. Lastly, the authors presented convincing evidence for relatively constant aurochs ancestry across all modern breeds, including the Lidia breed which has been bred for aggressiveness for centuries. My specific concerns are outlined below.Contamination is a common concern for all ancient DNA studies. Contamination by modern samples is perhaps unlikely for this specific study of ancient cattle, but there is still the possibility of cross-sample contamination. The authors should estimate and report contamination estimates for each sample (based on coverage of autosomes and sex chromosomes, or heterozygosity of Y or MT DNA). Such contamination estimates are particularly important to support the presence of individuals with admixed ancestry, as a domestic sample contaminated with a wild sample (or vice versa) could appear as an admixed individual.
A major limitation of this study is uncertainty in the "population identity" for most sampled individuals (i.e., whether an individual belonged to the domesticated or wild herd when they were alive). Based on chronology, morphology, and genetic data, it is clear the Mesolithic samples from the Artusia and Mendandia sites are bona fide aurochs, but the identities of individuals from the other two sites are much less certain. Indeed, archeological and morphological evidence from El Portalon supports the presence of both domestic animals and wild aurochs, which is echoed by the inter-individual heterogeneity in genetic ancestry. Based on results shown in Fig 1C and Fig 2 it seems that individuals moo017, moo020, and possibly moo012a are likely wild aurochs that had been hunted and brought back to the site by humans. Although the presence of individuals (e.g., moo050, moo019) that can only be explained by two-source models strongly supports that interbreeding happened (if cross-contamination is ruled out), it is unclear whether these admixed individuals were raised in the domestic population or lived in the wild population and hunted.
Such uncertainty in "population identity" limits the authors' ability to make conclusions regarding the frequency, sex bias, and directionality of gene flow between domestic and wild populations. For instance, the wide range of ancestry estimates in Neolithic and Chalcolithic samples could be interpreted as evidence of (1) frequent recent gene flow or (2) mixed practices of herding and hunting and less frequent gene flow. Similarly, the statement about "bidirection introgression" (on pages 8 and 11) is not directly supported by data. As the genomic, morphological, and isotope data cannot confidently classify an individual as belonging to the domesticated or wild population, it seems impossible to conclude the direction of gene flow (if by "bidirection introgression" the authors mean something other than "bidirectional gene flow", they need to clearly explain this before reaching the conclusion.)
The f4 statistics shown in Fig 3B are insufficient to support the claim regarding sex-biased hybridization, as the f4 statistic values are not directly comparable between the X chromosome and autosomes. Because the effective population size is different for the X chromosome and autosomes (roughly 3:4 for populations with equal numbers of males and females), the expected amount of drift is different, hence the fraction of allele sharing (f4) is expected to be different. In fact, the observation that moo004 whose autosomal genome can be modeled as 100% domestic ancestry still shows a higher f4 value for the X chromosome than autosomes hints at this issue. A more robust metric to test for sex-biased admixture is the admixture proportion itself, which can be estimated by qpAdm or f4-ratio (see Patterson et al 2012). However, even with this method, criticism has been raised (e.g., Lazaridis and Reich 2017; Pfennig and Lachance, 2023). In general, detecting sex-bias admixture is a tough problem.
In general, the stable isotope analysis seems to be very underpowered, due to the issues of variation in classification criteria and skeletal sampling location discussed by the authors in supplementary material. The authors claimed a significant difference in stable nitrogen isotope between (inconsistently defined) domestic cattle and wild aurochs, but no figures or statistics are presented to support this claim. Please describe the statistical method used and the corresponding p-values. The authors can consider including a figure to better show the stable isotope results.
-
Reviewer #3 (Public Review):
Summary:
Günther and colleagues leverage ancient DNA data to track the genomic history of one of the most important farm animals (cattle) in Iberia, a region showing peculiarities both in terms of cultural practices as well as a climatic refugium during the LGM, the latter of which could have allowed the survival of endemic lineages. They document interesting trends of hybridisation with wild aurochs over the last 8-9 millennia, including a stabilisation of auroch ancestry ~4000 years ago, at ~20%, a time coincidental with the arrival of domestic horses from the Pontic steppe. Modern breeds such as the iconic Lidia used in bullfighting or bull running retain a comparable level of auroch ancestry.Strengths:
The generation of ancient DNA data has been proven crucial to unravel the domestication history of …Reviewer #3 (Public Review):
Summary:
Günther and colleagues leverage ancient DNA data to track the genomic history of one of the most important farm animals (cattle) in Iberia, a region showing peculiarities both in terms of cultural practices as well as a climatic refugium during the LGM, the latter of which could have allowed the survival of endemic lineages. They document interesting trends of hybridisation with wild aurochs over the last 8-9 millennia, including a stabilisation of auroch ancestry ~4000 years ago, at ~20%, a time coincidental with the arrival of domestic horses from the Pontic steppe. Modern breeds such as the iconic Lidia used in bullfighting or bull running retain a comparable level of auroch ancestry.Strengths:
The generation of ancient DNA data has been proven crucial to unravel the domestication history of traditional livestock, and this is challenging due to the environmental conditions of the Iberian peninsula, less favourable to DNA preservation. The authors leverage samples unearthed from key archaeological sites in Spain, including the karstic system of Atapuerca. Their results provide fresher insights into past management practices, and permit characterisation of significant shifts in hybridization with wild aurochs.Weaknesses:
- Treatment of post-mortem damage: the base quality of nucleotide transitions was recalibrated down to a quality score of 2, but for 5bp from the read termini only. In some specimens (e.g. moo022), the damage seems to extend further. Why not use dedicated tools (e.g. mapDamage), or check the robustness by conditioning on nucleotide transversions?- Their more solid analyses are based on qpAdm, but rely on two single-sample donor populations. As the authors openly discuss, it is unclear whether CPC98 is a good proxy for Iberian aurochs despite possibly forming a monophyletic clade (the number of analysed sites is simply too low to assess this monophyly; Supplementary Table S2). Additionally, it is also unclear whether Sub1 was a fully unadmixed domestic specimen, depleted of auroch ancestry. The authors seem to suggest themselves that sex-biased introgression may have already taken place in Anatolia ("suggesting that sex-biased processes already took place prior to the arrival of cattle to Iberia").
Alternatively, I recommend using Struct-f4 as it can model the ancestry of all individuals together based on their f4 permutations, including outgroups and modern data, and without the need to define pure "right" and "left" populations such as CPC98 and Sub1. It should work with low-coverage data, and allows us to do f4-based MDS plots as well as to estimate ancestry proportions (including from ghost populations).
- In the admixture graph analyses (supplementary results), the authors use population groups based on a single sample. If these samples are pseudohaploidised (or if coverage is insufficient to estimate heterozygosity - and it is at least for moo004 and moo014), f3 values are biased, implying that the fitted graph may be wrong. The graph shown in Fig S7 is in fact hard to interpret. For example, the auroch Gyu2 from Anatolia but not the auroch CPC98 also from Anatolia received 62% of ancestry from North Africa? The Neolithic samples moo004 and moo014 also show the same shocking disparity. I would consider re-doing this analysis with more than a sample per population group.
-
-