LAUSR

1Fujian Provincial Key Laboratory of Plant Molecular and Cell Biology, Oil Crops Research Institute, College of Plant Protection, Fujian Agriculture and Forestry University, Fuzhou, China. 2School of Life Sciences, North China University of Science and Technology, Tangshan, China. 3School of Genomics and Bio-Big-Data, Chengdu University of Traditional Chinese Medicine, Chengdu, China. 4Plant Genome Mapping Laboratory, University of Georgia, Athens, GA, USA. 5Nextomics Biosciences Institute, Wuhan, China. 6College of Agriculure and Biology, Zhongkai University of Agriculture and Engineering, Guangzhou, China. 7Center of Excellence in Genomics and Systems Biology, International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Hyderabad, India. 8Institute of Agriculture, The University of Western Australia, Perth, Western Australia, Australia. 9These authors contributed equally: Weijian Zhuang, Xiyin Wang. ✉e-mail: [email protected]; [email protected]; [email protected] The origin of peanut (Arachis hypogaea) has been the focal point of peanut studies1–3. In our previous research, by calculating Ks values for collinear gene pairs and the expansion time of transposable elements, we inferred that peanut polyploidization occurred ~450,000 years ago and that the sequenced Arachis duranensis was not a direct descendent of the ancestor of the A-subgenome donor1. However, Bertioli et al. suggest that polyploidization happened <10,000 years ago, aided by human activities, and that the sequenced A. duranensis was a close relative of the A-subgenome donor2. In replying to Bertioli et al.4, we performed multiple analyses using diverse evolutionary models and obtained new lines of evidence that support our former inference1 and suggest likely ongoing DNA flow between different lines of peanuts, diploid or tetraploid. We first performed Ks analysis to date the origin of tetraploid peanut and the divergence of the Ad (A. duranensis) and Bd (Arachis ipaensis) diploid genomes5, At and Bt (the A and B subgenomes of A. hypogaea), Ad and At, and Bd and Bt (Supplementary Data 1). This approach can distinguish recent evolutionary events such as the formation of Brassica napus ~7,500–12,500 years ago (ref. 6). If tetraploid peanut originated as recently as a few thousand years ago, Ks analysis should have the power to discern this. Surely, if two homeologous or orthologous genes are very similar, Ks may lack resolution: for example, PAML outputs 0 when the Ks value is <0.0001 (ref. 7). Thanks to comments from Bertioli et al.4, we reinvestigated whether the removal of genes with Ks = 0 is reasonable. The Ks histograms for Ad–Bd and At–Bt (Shitouqi) comparisons (Fig. 1a,b) have bimodal distributions with peaks at Ks of ~0.03 and 0 (95% confidence interval, 0.0301–0.0310 and 0.0303–0.0312, respectively; Supplementary Table 1). A bimodal distribution is not consistent with the hypothesis that gene pairs at 0 are merely highly conserved, but instead suggests a different phenomenon. It is clear that Ks = 0 cannot be used to infer the split time of the A and B genomes and subgenomes, ignoring the major peak at ~0.030–0.031. The average genome sequence similarity between the diploid A and B genomes is 90.45%, and the tetraploid A and B subgenomes have 90.28% identity1,2,5 (Fig. 2 and Supplementary Data 2a,b). Accordingly, we could only use the second peak for correct inference, which indicated that the two diploids diverged 3.7 million years ago, which is similar to a previous estimate1 based on a mutation rate of 8.12 × 10−9 Ks per year. The same bimodal distribution of Ks for the At–Bt comparison is evident from the Tifrunner genome, with a much higher peak at Ks = 0 (Fig. 1e), again with a divergence time of ~3.7 million years ago for the A and B genomes. We further used EMMIX8 to perform time estimation and characterized the Ks mean and covariance for each pair of subgenomes, with very similar results (Supplementary Table 1). Bimodal distributions also occurred for the Ad–At and Bd–Bt Ks values, only with sharper peaks at Ks = 0 (Fig. 1c,d). If ignoring the peaks at Ks = 0, we infer the split times to be approximately 366,869–457,612 years ago (Ks of ~0.003 for the Bd–Bt comparison and ~0.0037 for the Ad–At comparison), as inferred in our previous publication1 and in line with analysis of the Tifrunner genome (Fig. 1f,g). These times were also confirmed with peak identities of ~98.72% (average 95.84%) and 99.48% (average 97.70%) obtained by BLAST analysis, corresponding to 128 and 52 differences per 10,000 nucleotides, dating divergence to 800,000 and 325,000 years ago for the Ad–At and Bd–Bt pairs, respectively1 (Fig. 2c,d and Supplementary Data 2; see also Fig. 2c in ref. 1). The modal peaks calculated by MUMmer were lower but still conformed (Fig. 2e,g). Are the Arachis genes with Ks = 0 highly conserved in evolution? If peaks at Ks = 0 represent conserved genes, then their low mutation rates are part of a distribution that reflects the overall divergence of paralogous and homoeologous genes. To investigate this, five alignments of randomly selected genes were linked and Ks values were inferred with the Nei–Gojobori approach implemented in PAML7. The peaks in the Ad–At and Bd–Bt Ks distributions continued to indicate that divergence occurred several million years ago, albeit with differing origins in cultivars Shitouqi and Tifrunner for the Ad–At and Bd–Bt pairs, respectively (Extended Data Fig. 1), implying that genes in peaks at Ks = 0 are not highly conserved in Arachis. Furthermore, we compared inferred Medicago orthologous genes9 to peanut homologs with Ks = 0 or Ks > 0 (Extended Data Fig. 2), by calculating Ks. The genes in these two groups had similar distribution patterns, indicating that genes with Ks = 0 in tetraploid peanut are not highly conserved in legume evolution. Reply to: Evaluating two different models of peanut’s origin