Correction to: Nature Communications https://doi.org/10.1038/s41467-019-10017-2, published online 30 April 2019.

In the original article, we analyzed 38 genomes from pygmy hog and related suid species. Our analysis not only identified a signal of introgression between Sus scrofa and pygmy hog but also reinforced the idea that there was gene flow between Sus scrofa and an extinct (ghost) Suidae linage1. In our original analysis, however, for the D-statistics equation in Admixtools2, we mistakenly interpreted as \(D = \frac{{{\rm{ABBA}} - {\rm{BABA}}}}{{{\rm{ABBA}} + {\rm{BABA}}}}\), while it was in fact \(D = \frac{{{\rm{BABA}} - {\rm{ABBA}}}}{{{\rm{ABBA}} + {\rm{BABA}}}}\). This led to an inversion of the direction of admixture shown in Fig. 2b in the original version of the article. To remain consistent with the rest of the analyses in the article, we rectified the formula of D-statistics as \(D = \frac{{{\rm{ABBA}} - {\rm{BABA}}}}{{{\rm{ABBA}} + {\rm{BABA}}}}\) and updated Fig. 2b (Fig. 1; this addendum). The results after correcting the formula indicate that in fact there is an excess of shared derived alleles between the pygmy hog and Island of Southeast Asian (ISEA) Sus (ABBA) and not between the pygmy hog and Sus scrofa (BABA). The excess of ABBA can be the result of admixture either between pygmy hog and ISEA Sus or between Sus scrofa and the archaic ghost linage or both. Methods to detect hybridization such as Patterson’s D, fd, or Twisst3,4,5, however, are inadequate to distinguish gene-flow events between P1 and P3 (Fig. 2, orange arrow) from those between P2 and an archaic ghost lineage basal to P3 (Fig. 2, blue arrow).

To address this issue an alternative method was used, based on relative nucleotide distance, as implemented in RNDmin6 to calculate the relative node depth between each taxon. To do so, we first used BEAGLE v4.17 to perform haplotype phasing on genotypes of all individuals (excluding Babyrousa babyrussa, Potomochoerus larvatus, and Potomochoerus porcus) with default parameters. Then we calculated RNDmin in 100-kb sliding windows along autosomes with <50% missing sites using PopGenome8. Comparisons were carried out between pygmy hog, ISEA Sus, and Sus scrofa, using the genome of a warthog as an outgroup. This shows that, for the genomic regions supporting topoA (Fig. 3a), the distribution of RNDmin computed between the pygmy hog and Sus scrofa is significantly lower (16.2%, Welch’s t test p value <1.8e−14) than RNDmin between pygmy hog and ISEA Sus, indicating that the first two species share more similarities. We also note that this pattern is more apparent in regions supporting topoA than in the overall genomic background (gray dots in Fig. 3a). Altogether, this indicates that the average genomic distance between the pygmy hog and S. scrofa is smaller than that between the pygmy hog and the ISEA Sus clade, which supports the existence of gene flow between pygmy hog and S. scrofa.

If the result of our D-statistics were affected by an admixture between pygmy hog and ISEA Sus, we would expect that, in regions that display topoB (Fig. 3b), the RNDmin computed between the pygmy hog and ISEA Sus should be lower than in the overall genomic background. Our analysis, however, shows that this is not the case. In fact, we found these distributions to be very similar (Fig. 3b), especially when compared to the result shown in Fig. 3a. In addition, we found the distance between ISEA Sus and Sus scrofa to be higher (23.2%, Welch’s t test p value <2.2e−16) than the overall genomic background in the regions that show topoB, suggesting that topoB was caused by the admixture with an unsampled taxon.

Altogether, our analyses indicate that the findings and interpretations presented in the original article are correct. First, there was no admixture between ISEA Sus and pygmy hog. Instead, the combined results of the D-statistics and RNDmin are more consistent with a ghost admixture between a distantly related taxon and S. scrofa, and this inflated the number of shared derived alleles between ISEA Sus and the pygmy hog. Second, the shorter distance between Sus scrofa and pygmy hog suggests that there was also admixture between these two species.

Fig. 1: A diagram depicting the excess derived allele sharing when comparing sister taxa and outgroups.
figure 1

Each column contains the fraction of excess allele sharing by a taxon (up/down) with the pygmy hog/outgroup compared to its sister taxon (up/down). We computed D statistics of the form D (X, Y, Pygmy hog, warthog). Error bars correspond to three standard errors. (SBSB Sus barbatus, SCEB Sus cebrifons, SCEL Sus celebensis, SVSV Sus verrucosus, EUD European domesticated pig, EUWB European wild boar, JWB Japanese wild boar, MS Meishan, NCWB Northern China wild boar, SCWB Southern China wild boar, XI Xiang).

Full size image
Fig. 2: Schematic representation of the potential gene-flow scenarios discussed in the text.
figure 2

Two-way arrows pointing the two hybridized species.

Full size image
Fig. 3: Phylogenetic tree indicated the alternative topologies (topoA and topoB) to the main species tree.
figure 3

Scatter and density plots below show the distribution of RNDmin between different comparisons. a Blue circles represent RNDmin distribution within windows supporting topoA. b Red circles represent RNDmin distribution within windows supporting topoB. Gray circles represent RNDmin distribution of the 100-kb windows among all autosomes. Populations used to calculate RNDmin is shown on axis labels. Warthog was used as outgroup.

Full size image