2014-05-05_REVISION - C1996083 - Laserfiche WebLink

Home Browse Search

6 J. L. METCALF ET AL. samples based on the 430 -bp mitochondrial DNA data set using TCS (Clement et al. 2000). Finally, we inferred a phylogeny of extant and historical haplotypes based on the same 430 -bp sequence data set. In the phyloge- netic analysis, we included sequence data for all major subspecies of cutthroat trout (O. c. bouvieri, O. c. utah, O. c. henshawi, O. c. lewisi and O. c. clarki) as well as rainbow trout (Oncorhynchus mykiss) to clarify whether museum samples represent a trout stocked from another region in North America. These additional sequence data were originally published in Metcalf et al. (2007) or Loxterman & Keeley (2012) (GenBank No's EF673223- EF73232; EF673250- EF673259; EF673233- EF673249; EF673260- EF673276; JQ747557 JQ747623). We used a nucleotide substitution model of GTR +I (Metcalf et al. 2007). We inferred a phylogeny using maximum - likelihood methods with the software PhyML 3.0 (Guin- don et al. 2010). We used NNI for tree improvement and computed branch support using the approximate likelihood -ratio test for branches described in Anisim- ova et al. (2011). We also constructed phylogenies using BEAST v1.6.1 (Drummond & Rambaut 2007). As our analysis spans the boundary between populations and species, we used two different tree priors (i) a coales- cent model assuming a constant population size and (ii) a Yule speciation process model. Each analysis was run for 100 000 000 MCMC generations, sampling every 1000 trees with a burn -in of 10 000 000 generations. Stationarity of the posterior probabilities distribution and the ESS values for the priors was examined in Tracer v1.4 (Rambaut & Drummond 2007) to confirm the burn -in was appropriate. Maximum clade credibility consensus trees were generated from posterior distribu- tions of 90 000 trees using TreeAnnotator v1.6.1 (Drummond & Rambaut 2007). Additionally, we investigated the geographical distri- bution of ND2 molecular variation for both museum samples and a broad survey of modern populations. We performed an analysis of molecular variance among drainages of historical (1857 -1889 A.D.) and modern (2000 A.D. to present) populations separately to under- stand how the distribution of variance has changed over time. For the modern data, we included only the portion of ND2 sequenced for museum samples (378 bp) in the analysis. We generated distance matrices and performed AMovA analyses using tools included in the ape and pegas packages in the R statistical envi- ronment ( Paradis et al. 2004; Paradis 2010). Distance matrices were generated using default parameters, except loci with missing data were not excluded from the analysis. Drainage delineations included South Platte, Arkansas, Rio Grande, San Juan, Gunnison and Colorado River basins. Samples from the Yampa River drainage were not included in the analysis because the museum data set had only one sample from this drain- age. To account for the difference in sample size between the museum and modern data set, we randomly subsampled the modern data set 1000 times using the same number of samples per drainage as the museum data set. Subsampled data were generated using a custom Python v2.7.2 script (www.python.org) in conjunction with the NumPy v1.6.1 (www.numpy. org) numerical Python library. Using R 2.14.2 with pegas v0.4 -1, we performed the AMOVA analysis on each subsampled data set to generate a range of possible modern values to compare with the historical estimate. Results Museum fish DNA Skin, gill, muscle and bone were sampled from 44 cutthroat trout specimens stored in ethanol. Several samples that yielded DNA were extracted multiple times (up to four times) to confirm authenticity of endogenous DNA (Table S1). Sixty -one museum sample extractions were performed (not including mock extrac- tion controls). Extracted DNA was highly degraded. PCR success declined with increasing target fragment size (Fig. S1) and 14 of the 44 individuals (32 %) failed to yield DNA that could be amplified using PCR. None- theless, 30 individuals (Tables 1 and S1) yielded a total of 430 bps of mtDNA (two ND2 gene fragments and one COI gene fragment) that permitted robust inference of the phylogeny and regional biogeography of native cutthroat trout (GenBank No's JX195398- JX195487). Furthermore, sequence data were reproducible —data generated from independent extractions of museum samples were identical. In particular, the consensus sequence from cloned amplicons (at CU) matched amplicons directly sequenced (at ACRD) at >99 % of nucleotides. Most mismatches were probably due to deamination (postmortem chemical damage). In a few cases in which mismatches were not resolved, bases were left ambiguous. Phylogenetic analyses reveal historical diversity of Colorado's cutthroat trout Individual museum samples were assigned to one of the four extant subspecies depending on the clade- defining mutations (new modern haplotypes include Genbank No's JX195488- JX195497) (Table S4). Twenty -one museum samples assigned to modern clades and nine did not (Supporting information). A sta- tistical parsimony haplotype network including museum sequences and contemporary haplotypes revealed six distinct clusters, each of which was separated by at C 2012 Blackwell Publishing Ltd