More than 200 cortical bone samples were obtained from museum specimens of modern, Pleistocene, and American lions (Table S1, Supporting information). Specimens ranged in age from modern to > 62 000 years.
To prevent duplicate sampling within a site, either the same skeletal element or samples associated with widely separated radiocarbon dates were chosen. DNA extraction was performed as described in Barnett et al. (2006) in dedicated ancient DNA(aDNA) facilities at the University of Oxford.
Four European lions were also extracted and amplified at Mainz University under similarly stringent conditions (Burger et al. 2004). One of the cave lions from Mainz (sample 8 in Table 1) was previously analysed for cytochrome b in an earlier study showing the phylogenetic position of the cave lion among the Panthera cats (Burger et al. 2004). A ~215-bp fragment of the mitochondrial hypervariable region 1 (HVR1) and a 143-bp fragment of ATP8 were polymerase chain reaction (PCR) amplified, and cloned, purified, and sequenced following Barnett et al. (2006).
Strict aDNA protocols were followed, including the use of multiple PCRs and negative controls, overlapping fragments and independent replication. Nuclear mitochondrial inserts, which are known to be widespread among felids (e.g. Kim et al. 2006), were identified and excluded from the analysis. All specimens possessing unique haplotypes, along with a subset of those with shared haplotypes, were amplified several times and cloned to check for the presence of contamination (details are given in Supporting Information: supporting text, Tables S1–S3).
Thirty-three samples were submitted to the Oxford Radiocarbon Accelerator Unit (University of Oxford) for radiocarbon analysis. Analyses were performed using 0.2 g of bone taken from a site adjacent to the sample used for DNA extraction. An additional sample (sample 8 in Table 1) was radiocarbon dated at the Leibniz laboratory of the University of Kiel. For all samples, total bone collagen was extracted, graphitized, and dated by accelerator mass spectrometry. Dates are presented as uncalibrated radiocarbon values.
Two of the oldest specimens (samples 31 and 34 in Table 1) were dated twice, to test the reproducibility of dates in the > 50 000-year time frame. Phylogenetic analyses were performed on the HVR1 and ATP8 data sets separately, due to the differing number of specimens that yielded sequences from the two regions (Supporting Information: supporting text, Table S3). Medianjoining networks were produced for both data sets using the program Network version 4.1.0.3 (Bandelt et al. 1999).
Bayesian Markov chain Monte Carlo (MCMC) analyses were then performed using beast (Drummond & Rambaut 2007), first for HVR1 and ATP8 data sets separately, and then for a smaller, combined analysis consisting of only those samples from which both sequence fragments could be amplified and which were associated with finite radiocarbon dates. For all three data sets, comparison of Akaike information criterion scores suggested the HKY85 model of nucleotide substitution. Each beast analysis assumed this substitution model as well as a constant population size and a strict molecular clock calibrated using the age of the split between the spelaea and leo groups (with a normal prior mean of 550 000 years, standard deviation 25 000years), based on the first appearance of Panthera leo fossilis(Garcia Garcia 2001; Burger et al. 2004). Demographic and evolutionary model parameters were then estimated simultaneously along with the phylogeny, with samples drawn from the posterior every 5000 MCMC steps over a total of 5 000 000 steps. The first 500 000 steps were discarded as burn-in. Acceptable mixing and convergence to stationarity were checked using the program Tracer version 1.4(Rambaut & Drummond 2007).
此主題相關圖片如下：
