A laboratory guide for generating DNA barcodes in grasses: a case study of Leptochloa s.l. (Poaceae: Chloridoideae)

There is no easy way to identify to species, a small, vegetative leaf or culm sample of a grass and there are more than 12,000 species in this large, important family. The long-range aim of our study is to produce a standard DNA barcode library available to the public for all grasses (±1960 species) in North America (includes all Canada, Mexico and USA) that will facilitate the easy identification of these morphologically cryptic species. We provide a detailed protocol of the laboratory procedures for DNA extraction in grasses and the DNA-specific primers used for the polymerase chain reaction (PCR) enabling the laboratory work to be performed in any well-supplied molecular laboratory. In this paper we present a test of four barcodes [matK, rbcL, rpl32-trnL and internal transcribed spacer (ITS)] to discriminate among 50 taxa of grasses (55 samples), predominately in the subfamily Chloridoideae, and we used a tree-based method to identify relationships among species of Leptochloa sensu lato. The sequence divergence or discriminatory power based on uncorrected p-value, among the four DNA sequence markers was greatest in ITS (96%), followed by rpl32-trnL (25.6%), matK (3.0%) and rbcL (0.0%). matK was twice as effective in discriminating among the species compared with rbcL; rpl32-trnL was nearly 3.4 times better than rbcL; and nuclear rDNA ITS was 14 times better than rbcL. There are significant threshold levels of 0.0682 for ITS and 0.0732 for ITS + rpl32-trnL between intrageneric and intergeneric sequence divergences within the 16 species of Dinebra and between Dinebra and Diplachne, Disakisperma and Leptochloa sensu stricto. In our tree-based analyses of Leptochloa s.l. the following number of nodes with strong support (PP = 0.95−1.00) were successfully recovered (in descending order): combined ITS + rpl32-trnL, 43; ITS, 34; rp32-trnL, 27; matK, 19; and rbcL, 3.