Metabolomic analysis applied to chemosystematics and evolution of megadiverse Brazilian Vernonieae (Asteraceae) Marília Elias Gallon a , Marcelo Monge b , Rosana Casoti a , Fernando Batista Da Costa a , Jo ~ ao Semir b , Leonardo Gobbo-Neto a, * a University of S~ ao Paulo (USP), School of Pharmaceutical Sciences of Ribeir~ ao Preto, Av. do Caf e s/n  , 14040-903, Ribeir~ ao Preto, SP, Brazil b University of Campinas (UNICAMP), Institute of Biology, Cidade Universit aria “Zeferino Vaz”, 13083-970, Campinas, SP Brazil article info Article history: Received 1 November 2017 Received in revised form 6 March 2018 Accepted 12 March 2018 Keywords: Vernonia Schreb Vernoniinae Lepidaploinae Rolandrinae Compositae LC-MS Multivariate statistical analyses Chemotaxonomy abstract Vernonia sensu lato is the largest and most complex genus of the tribe Vernonieae (Asteraceae). The tribe is chemically characterized by the presence of sesquiterpene lactones and flavonoids. Over the years, several taxonomic classifications have been proposed for Vernonia s.l. and for the tribe; however, there has been no consensus among the researches. According to traditional classification, Vernonia s.l. com- prises more than 1000 species divided into sections, subsections and series (sensu Bentham). In a more recent classification, these species have been segregated into other genera and some subtribes were proposed, while the genus Vernonia sensu stricto was restricted to 22 species distributed mainly in North America (sensu Robinson). In this study, species from the subtribes Vernoniinae, Lepidaploinae and Rolandrinae were analyzed by UHPLC-UV-HRMS followed by multivariate statistical analysis. Data mining was performed using unsupervised (HCA and PCA) and supervised methods (OPLS-DA). The HCA showed the segregation of the species into four main groups. Comparing the HCA with taxonomical classifications of Vernonieae, we observed that the groups of the dendogram, based on metabolic profiling, were in accordance with the generic classification proposed by Robinson and with previous phylogenetic studies. The species of the genera Stenocephalum, Stilpnopappus, Strophopappus and Rolandra (Group 1) were revealed to be more related to the species of the genus Vernonanthura (Group 2), while the genera Cyrtocymura, Chrysolaena and Echinocoryne (Group 3) were chemically more similar to the genera Lessingianthus and Lepidaploa (Group 4). These findings indicated that the subtribes Verno- niinae and Lepidaploinae are non-chemically homogeneous groups and highlighted the application of untargeted metabolomic tools for taxonomy and as indicators of species evolution. Discriminant com- pounds for the groups obtained by OPLS-DA were determined. Groups 1 and 2 were characterized by the presence of 3 0 ,4 0 -dimethoxyluteolin, glaucolide A and 8-tigloyloxyglaucolide A. The species of Groups 3 and 4 were characterized by the presence of putative acacetin 7-O-rutinoside and glaucolide B. There- fore, untargeted metabolomic approach combined with multivariate statistical analysis, as proposed herein, allowed the identification of potential chemotaxonomic markers, helping in the taxonomic classifications. © 2018 Elsevier Ltd. All rights reserved. 1. Introduction Current metabolomic approaches have been applied as a po- tential chemotaxonomic tool (Kim et al., 2016; Martucci et al., 2014; Sandasi et al., 2012) and as a tool for understanding species evo- lution (Dos Santos et al., 2017) and biogeographic trends in plant diversification (Padilla-Gonz alez et al., 2017). Furthermore, obtaining metabolomic fingerprints associated with multivariate statistical analysis has had wide applications in natural products research, such as quality control of medicinal plants, solving phylogenetic problems and discovery of bioactive compounds (Chagas-Paula et al., 2015; Coxet al., 2014; Yuliana et al., 2011). Vernonia Schreb. sensu lato (Asteraceae) is the largest and most complex genus of the tribe Vernonieae Cass. (Keeley et al., 2007). The “Vernonia problem” (Bremer, 1994) arose because several species from the tribe Vernonieae, 1000 of the 1500 species, were * Corresponding author. E-mail address: gobbo@fcfrp.usp.br (L. Gobbo-Neto). Contents lists available at ScienceDirect Phytochemistry journal homepage: www.elsevier.com/locate/phytochem https://doi.org/10.1016/j.phytochem.2018.03.007 0031-9422/© 2018 Elsevier Ltd. All rights reserved. Phytochemistry 150 (2018) 93e105