Genomic analysis of the domestication and post-Spanish conquest evolution of the llama and alpaca

Picture the Andes in the sixteenth century. The Inca state had spent generations managing llamas and alpacas as two distinct populations — pack animals on one side and fine-wool producers on the other — separated by careful husbandry and centuries of selective breeding. Then the Spanish arrived, and within a generation, that entire agricultural order collapsed. Fan and colleagues found that collapse written into the DNA of every living llama and alpaca. The signal is impossible to miss. To understand what they found, hold four animals in mind. The guanaco and the vicuña are the two surviving wild species of South American camelid. The llama — historically used as a pack animal — is the domesticated form derived from the guanaco. The alpaca, selected for its extraordinarily fine fleece, descends from the vicuña. This much had been suspected, but earlier studies using mitochondrial DNA, microsatellites, and Y-chromosome markers covered only a tiny fraction of the genome and couldn't cleanly distinguish ancient shared ancestry from recent hybridization. The picture stayed muddy for decades. Fan and colleagues cleared it up by generating de novo reference genomes for the llama, guanaco, and vicuña — each assembly around 2.6 gigabases, with coverage ranging from about one hundred and two-fold to one hundred and twenty-nine-fold — and comparing them to an existing alpaca reference. They then resequenced additional individuals across all four species and analyzed nearly five point nine million genome-wide single-nucleotide polymorphisms, or SNPs. Neighbor-joining trees and ADMIXTURE clustering, both run at that scale, placed llama with guanaco and alpaca with vicuña at one hundred percent bootstrap support. When the researchers stripped non-vicuña segments out of the alpaca genome, alpacas clustered cleanly with the northern vicuña lineage. The domestication origins, long debated, were settled. The methods they used to detect introgression — the flow of genomic segments between formerly separate lineages — are worth understanding briefly, because the numbers they produce are striking. The ABBA-BABA framework asks whether two species share more uncommon mutations with a third than a simple family tree would predict. A statistic called fd maps that excess in windows across the genome. A separate technique called local ancestry inference uses a hidden Markov model to assign each stretch of an individual's genome to one ancestral population or another. Windowed FST and extended haplotype tests then flagged regions under selection. These approaches are independent; when they converge on the same answer, you can trust the result. And converge they did. The ABBA-BABA D statistic for the test placing alpaca against vicuña, llama, and the bactrian camel came out at zero point fifty-seven — a large, statistically clear signal of introgression. The fd estimate put the fraction of the alpaca genome derived from the llama lineage at roughly thirty-nine percent. Local ancestry inference agreed: about thirty-six percent of the alpaca genome traces to guanaco and llama ancestry. For the llama, the signal ran the other way but was far smaller — only about five percent of the llama genome derives from vicuña or alpaca ancestry. That asymmetry is the heart of the story. Fan and colleagues dated the pulse of admixture using tract lengths from local ancestry inference. The estimate: one hundred and twenty-one generations for alpacas and one hundred and fifteen for llamas, with overlapping confidence intervals. Assuming a four-year generation, those figures place the mixing event around four hundred and eighty-five and four hundred and sixty-one years ago respectively — which lands squarely at the Spanish conquest, beginning in fifteen twenty-nine. The Inca system had kept these animals separate. Once that system broke down, the lineages mixed, and they mixed heavily, especially into the alpaca. The X chromosome tells a sharper version of the same story. Across the autosomes, the D and fd statistics for llama-into-alpaca introgression averaged zero point fifty-six and zero point seventeen. On the X chromosome, those values dropped to zero point twenty-eight and essentially zero. Selection was actively limiting gene flow on the X — which makes sense, because X-linked variants are exposed to selection in males without a second copy to buffer their effects. Fan and colleagues identified four X regions with almost zero introgression, covering one hundred and sixteen genes, and four with high introgression, covering seventy-nine genes. Across those eight regions, fifty-three genes have known deleterious or lethal X-linked effects in humans. The genome is not a passive record of history. It bears the marks of selection deciding, region by region, what could be tolerated and what could not. The autosomal introgression, by contrast, appears to have carried some genuinely useful cargo. One of the most striking examples is a block of sequence syntenic — meaning structurally equivalent — to a region on human chromosome four called HSA4q21. This block contains four genes: ANTXR2, PRDM8, FGF5, and C4orf22. Local ancestry inference gave this region a score of zero point ninety-one, and the fd statistic was zero point eighty-seven — both among the highest values in the genome, indicating it moved from the llama lineage into alpaca. Nucleotide diversity across this block dropped from zero point seventeen in guanaco to zero point thirteen in llama to just zero point zero two in alpaca. That's the kind of serial narrowing you see when selection has swept a haplotype through a population. Why would this particular haplotype be favored? The HSA4q21 region has human genome-wide association study links to blood pressure homeostasis, and C4orf22 has been associated with hypertension susceptibility. Fan and colleagues tested what these genes do under low oxygen. They cultured alpaca and sheep melanocytes for seventy-two hours at normal oxygen — twenty-one percent — and at hypoxic conditions mimicking high altitude — thirteen percent. FGF5, ANTXR2, and C4orf22 all showed higher expression under hypoxia. PRDM8, a histone methyltransferase that typically represses transcription, showed significantly lower expression at thirteen percent oxygen in both cell lines. HIF1α and SOX6 also fell. Fan and colleagues interpret PRDM8 downregulation as a potential mechanism by which repression is relieved under hypoxia, allowing other genes to ramp up — functioning, in their words, analogously to the hypoxia-inducible factor pathway already known to operate at high altitude. PRDM8 appears to be a novel node in that network, previously undocumented in this context. FGF5, sitting in the same introgressed block, connects the physiology directly to fleece. Sequencing FGF5 in Suri alpacas — which have long silky wool — and Huacaya alpacas — which have shorter crimped fiber — revealed nearly identical coding sequences, just two nucleotide substitutions. But expression told a different story entirely. FGF5 mRNA and protein were significantly elevated in Suri relative to Huacaya animals, with a two-way analysis of variance giving a p-value below zero point zero zero one for both measures. The introgressed haplotype carries variation that shapes the very fiber trait for which alpacas were domesticated. Olfactory receptor complexes complete the picture of functional exchange. OR5 family receptors show introgression from llama into alpaca; OR2 family haplotypes moved from alpaca into llama. South American camelids have far fewer functional olfactory receptor genes than cattle or horses — around five hundred and sixty-six to six hundred and two genes, roughly half the count — but those families show a mosaic of ancestry consistent with bidirectional exchange between the two domestic species. Underlying all of this is a population that has been shrinking. PSMC and MSMC reconstructions — both methods for inferring historical population size from genomic data — show a long decline from around one hundred thousand to twenty-five thousand years before present. Finer-scale analyses document continuing decline within the last thousand years, with current effective population sizes inferred below one thousand individuals and an acceleration in the rate of decline at approximately one hundred and ten generations ago. The Spanish conquest compressed these populations severely. What introgression then did was complicated: it introduced deleterious variants on the X chromosome but also shuffled in haplotypes conferring high-altitude tolerance and improved fiber production on the autosomes. Fan and colleagues close with an argument worth sitting with. Introgression in domestic species is often treated as a problem — contamination of a pure lineage. But the genomic record here suggests something more like an accident with mixed consequences. The breakdown of Inca management practices was catastrophic for population size. Yet in the chaos, some adaptive variation moved between species that might otherwise never have exchanged genes. These genomes now give breeders and conservationists a detailed map of what was lost and what was, inadvertently, preserved. Five hundred years of history, readable at single-base resolution. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.