Population genomics finds veritas in the demographic history of vino
One of the more, hah, fruitful applications of genomic data has been in crop and livestock improvement. Biologists know that domesticating plants and animals for human use has involved powerful artificial selection — usually inadvertent at first, then intensive and deliberate. Compared to their wild ancestors, domesticated populations usually have more cultivation-friendly phenology and mating systems, produce more of the whatever feature it is that humans use, and even show behavioral changes. Genome sequencing lets us find the actual changes in the genetic code that underly those selected changes.
A nice new example of this work is online as a preprint at bioRxiv, which reports analysis of population genomic samples of cultivated and wild grapes. The paper’s coauthors, led by Yongfeng Zhou, are particularly interested in the fact that domestic grapes are perennials, propagated by cloning from cuttings. Clonal propagation is far and away the easiest route to domestication, especially of a perennial plant, because it skips over multi-year or multi-decade generation times, and it lets cultivators and breeders rapidly access useful traits in individual lines of the plant. But it also means that the cultivated population can rapidly lose genetic diversity — this is the reason bananas are particularly vulnerable to disease — and ongoing clonal propagation may allow a buildup of deleterious mutations. Comparing population samples from both cultivated and wild grapes lets Zhou et al. examine that “cost of domestication”.
The authors collected whole-genome sequence data from nine lines of wild grape, 18 lines of domestic grape, and one individual from another species in the genus Vitis as an outgroup. They inferred the demographic history of the wild and domestic populations using the multiple sequentially Markovian coalescent, and applied two related tests for positive selection based on the shape of the site frequency spectrum (the distribution of allele frequencies at variable sites) to identify genome regions that have experienced selective sweeps within the populations of domestic and wild grapes, and in domestic grapes in comparison to wild grapes. The former test, SweeD, conducts a fairly standard test for skewing of the site frequency spectrum in individual genome regions relative to the rest of the genome, or in comparison to expectations from a model of past population history. The latter test, XP-CLR, identifies regions at which allele frequencies in two populations differ more than expected due to drift since the time of their differentiation. (How robust this is to ongoing gene flow — which definitely occurs between wild and domestic grapes, and in fact makes the identification of “pure” wild grapes a challenge — doesn’t seem to be addressed in the XP-CLR paper.)
The coalescent demographic reconstruction found evidence for a decline in the effective population size of domestic grape starting about 22 thousand years ago and bottoming out between seven and 11 thousand years ago, about when grapes are thought to have been domesticated. The domestic grape population was estimated to have diverged from the wild population at the start of that population decline, which suggests either that the population that eventually gave rise to domestic grapes was naturally in decline before domestication, or that humans began managing natural populations of grapevines long before grapes were actually domesticated. Wild grapes also showed signs of population decline in the same analysis, but that population increased again with the end of the last glacial maximum, while the domestic population doesn’t show signs of recovering its pre-decline diversity.
The tests for selective sweeps in the domestic population found signs of selection near genes with roles in flower development, sugar transport, and berry development. In wild grapes, SweeD found fewer regions with significant signs of selection, and none in common with those that have been selected in domestic grapes — genes under selection in the wild grapes were associated with biotic and abiotic stress resistance pathways, instead of fruit development.
Finally, the authors used SIFT to predict deleterious variants in the wild and domesticated grape sequences. This identified a somewhat elevated rate of deleterious mutation accumulation in the domestic population. Deleterious variants most commonly found as heterozygotes in both wild and domestic grapes, consistent with their being, usually, recessive. Deleterious variants hitchhiking in selective sweeps doesn’t seem to explain this; though regions that had experienced selective sweeps in the domestic population showed an elevated frequency of deleterious mutations relative to synonymous mutations, they didn’t contain more deleterious mutations than expected for a given length of genome sequence. Clonal propagation can maintain deleterious mutations masked in heterozygous genotypes for generations — and as the authors point out, crossing grape vines from the same clonal lineage usually results in bad inbreeding depression, as expected if the cross creates a lot of homozygous pairings of recessive deleterious variants.
Grape cultivation has, to be sure, maintained a certain degree of diversity in domestic populations, particularly for traits that connect to human aesthetic preferences — just check out this collection of photo-realistic paintings of grape varieties from U.P. Hedrick’s 1908 book The Grapes of New York to get a glimpse of that diversity. But it’s clearly not the diversity that would have been maintained in nature, as shown by the comparison Zhou et al make to wild grapes. Thousands of generations of clonal propagation has taken a toll on the genetic robustness of the domesticated grape population, and breeders have already been coping with that toll in the form of inbreeding depression. With genome sequence data in hand, it may be possible to plan new grape crosses that can do what clonal propagation doesn’t, improving grapes for the future by cleaning out those accumulated mutations.
Chen H, N Patterson, and D Reich. 2010. Population differentiation as a test for selective sweeps. Genome Research, 20(3), 393-402. doi: 10.1101/gr.100545.109
Larson G and DQ Fuller. 2014. The evolution of animal domestication. Annual Review of Ecology, Evolution, and Systematics, 45, 115-136. doi: 10.1146/annurev-ecolsys-110512-135813
Myles S, AR Boyko, CL Owens, PJ Brown, F Grassi, MK Aradhya, B Prins, A Reynolds, JM Chia, D Ware, CD Bustamante, and ES Buckler. 2011. Genetic structure and domestication history of the grape. Proceedings of the National Academy of Sciences, 108(9), 3530-3535. doi: 10.1073/pnas.1009363108
Ng PC and S Henikoff. 2003. SIFT: Predicting amino acid changes that affect protein function. Nucleic Acids Research, 31(13), 3812-3814. doi: 10.1093/nar/gkg509
Pavlidis P, D Živković, A Stamatakis, and N Alachiotis. 2013. SweeD: likelihood-based detection of selective sweeps in thousands of genomes. Molecular Biology and Evolution, 30(9), 2224-2234. doi: 10.1093/molbev/mst112
Schiffels S and R Durbin. 2014. Inferring human population size and separation history from multiple genome sequences. Nature Genetics, 46(8), 919-925. doi: 10.1038/ng.3015
Zhou Y, M Massonnet, J Sanjak, D Cantu, and BS Gaut, 2017. The evolutionary genomics of grape (Vitis vinifera ssp. vinifera) domestication. bioRxiv, 146373. doi: 10.1101/146373