4 and 5), and appropriate amounts (0.5–4 ng) of template DNA were added to the PCR master mixture following the manufacturer's protocol.Portions of the PCR products were run on an 1.5% agarose gel with a Lambda/RI size marker to verify the size and amount of the PCR products (Fig. Single band PCR products were purified with Qiagen PCR purification kits.The final data set included 40 accessions of model.Bayesian probabilities were obtained under the GTR Gamma model, with four Markov chain Monte Carlo chains run for 556,700 generations, using random trees as starting points, sampling every 10th generation, and discarding the first 16,300 trees as burn-in. For the age estimates, combined chloroplast share five nucleotide substitutions with angiosperms that are not found in other seed plants.
Separated and purified PCR products were PCR-sequenced, using ABI Prism Big Dye Terminator Cycle Sequencing Ready Reaction kits (Perkin–Elmer) with the amplification primers. Length and GC content of each domain were calculated from the aligned sequences (Tables 3 and 4, which are published as supporting information on the PNAS web site).1), with domains I and V most crucial for the introns' enzymatic activity (12, 14) and ribozymic function (11, 12).Because of these characteristics and the presence of reverse transcriptase ORFs, they are believed to be the ancestors of spliceosomal introns and non-long-terminal repeat retroelements (13, 15).Heuristic searching used 100 random taxon addition replicates, holding 100 trees at each step, tree bisection–reconnection branch swapping, Mul Trees, Collapse, and Steepest Descent options, and no upper limit for trees held in memory.The partial exons/intron sequence tree and two full-length intron sequence trees had basically identical topologies, with the intron trees showing more phylogenetic structure. Filled squares indicate intron presence; hatched squares represent partially sequenced or PCR-verified accessions; open squares indicate lack of intron.Dots above the topmost sequence indicate the posttranscriptionally edited sites. Nonparametric bootstrap support was obtained by resampling the data 1,000 times with the same search options and model.