The present erratum is in regards to our article entitled ‘Ancient DNA and the tropics: a rodent's tale’. We were made aware of problems with some of the ancient sequences submitted to GenBank and conducted a systematic review of all the files used in our study. We discovered that, unfortunately, an incorrect file was sent to GenBank and was also used in some of our downstream analyses. We immediately contacted GenBank, explained the situation and corrected the file. We have redone some analyses with the correct file and describe these changes below.
The following analyses were performed with the same parameters as described in the article. The new estimated best-fit substitution model under the Akaike Information Criterion (AIC) was HKY + I + G (A = 0.3298, C = 0.3080, G = 0.0981 and T = 0.2641: I = 0.2970 and γ = 0.7880). The ML (data not shown) and Bayesian trees (figure 1a) showed a concordant topology, with high support for the O. phyllotis major clades. All ancient samples were grouped within the Yucatan peninsula lineage (posterior probability = 0.82, aLRT-SH = 97). Divergence times remain the same for the complete phylogeny (figure 1a); the new results show that the divergence of the ancient haplotype basal to the Yucatan lineage occurred within the Pliocene (from 5 Ma onwards) and that of the rest of the ancient haplotypes occurred much later, coinciding with the diversification of modern haplotypes (1.9–0.33 Ma).
We obtained nine ancient haplotypes (figure 1b; accession numbers KJ751487–KJ751498). The entire ancient dataset did not deviate from neutrality (p > 0.10) (Tajima's D = −1.388; Fu and Li's D = −1.341; F = −1.542) and showed high haplotype (h = 0.939 ± 0.058) and low nucleotide diversity (π = 0.0093 ± 0.0018). Genetic differentiation between stratigraphic layers was low (0.1–0.5%), indicating continuity between samples within the Loltún ancient samples irrespective of layer. The haplotype network was constructed for the Yucatan peninsula lineage in accord with the estimated phylogenetic trees, given that the ancient set is no longer a unique clade and the ancient sequences are within this lineage. The resulting haplotype network shows that, with one exception (5AH), the ancient samples are not dispersed throughout the modern haplotypes. Instead, they are joined in a central star-like shape, with no more than nine mutational steps between them. The ancient set connects with other haplotypes from the Yucatan peninsula that are mostly located basal to subclades with recent diversifications (figure 1a,b).
To infer past population size changes, demographic indices and a Bayesian Skyline Plot were calculated for the Yucatan peninsula lineage. The skyline plot results remained consistent (relatively constant growth from 1.99 Ma and an Ne of less than 100 that decreases towards the present). Demographic statistics are in agreement with a recent population expansion signal: negative Fu's Fs index (−30.083) and low R2 and raggedness indices (R2 = 0.0321; r = 0.0059). This historical demographic pattern was also confirmed by the neutrality test results, in which this lineage shows a signal of recent expansion indicated by significant (p < 0.02) and negative values of Tajima's D and Fu's D and F (D = −2.267, D = −5.466 and F = −4.949, respectively).
Importantly, despite the corrections to our sequences as well as the new analyses we performed, two of our three main results did not change: (i) the DNA signal is indeed real and tropical caves can be a good source of ancient DNA regardless of theoretical predictions for these environmental conditions. We have successfully extracted DNA and amplified a fragment significantly longer than other reported from Quaternary rodents (666 bp for 12 samples). (ii) Genetic information obtained from unique fossil samples can reveal relationships that could not be assessed by morphology due to the limited availability of complete fossil samples. We were able to assess the phylogenetic ancestry of the ancient samples, in which our study shows that the jawbones from Loltún, identified as O. phyllotis, belong in fact to this species.
Our conclusion in the paper that the ancient samples formed a distinctive ancestral clade to modern O. phyllotis lineages is no longer valid. The corrected results show that the ancient samples fall within the lineage from the Yucatan peninsula. This result is consistent with our original hypothesis, which stated that the ancient samples were likely to be related to the Yucatan peninsula lineage. Our results remain congruent with the main processes of the Holocene, in that ancient and modern examples of O. phyllotis arose during the Miocene–Pliocene and diversified during the Pleistocene. To date we have not found any ancient haplotypes within extant populations, suggesting drift (extinction of ancient haplotypes) during the Holocene. This argues for remarkable genetic continuity within the Yucatan lineage over at least the last 500 000 years. Further genetic analyses on the well preserved remains from Loltún cave will allow for a more refined demographic analysis in the near future.
- © 2015 The Author(s) Published by the Royal Society. All rights reserved.