More than 12,000 species have been listed under the category of berries, and most of them belong to the orders Ericales and Rosales. Recent phylogenetic studies using molecular data have revealed disagreements with morphological approaches mainly due to diverse floral arrangements, which has proven to be a problem when recognizing species. Therefore, the use of multilocus sequence data is essential to establish robust species boundaries. Although berries are common in Andean cloud forests, diversity of these taxa has not been extensively evaluated in the current context of DNA-based techniques. In this regard, this study characterized morphologically and constructed multilocus phylogenies using four molecular markers, two chloroplast markers (matK and rbcL) and two nuclear markers (ITS and GBSSI-2). Specimens did not show diagnostic features to delimit species of berries. A total of 125 DNA-barcodes of andean berries were newly generated for the four molecular markers. The multilocus phylogenies constructed from these markers allowed the identification of 24 species grouped into the order Ericales (Cavendishia ¼ 1, Clethra ¼ 2, Disterigma ¼ 2, Gaultheria ¼ 4, Thibaudia ¼ 4, Vaccinium ¼ 3) and Rosales (Rubus ¼ 8), incorporating into the Peruvian flora four new records (Disterigma ecuadorense, Disterigma synanthum, Vaccinium meridionale and Rubus glabratus) and revealing the genus Rubus as the most diverse group of berries in the Amazonas region. The results of this study showed congruence in all the multilocus phylogenies, with internal transcribed spacer (ITS) showing the best resolution to distinguish the species. These species were found in coniferous forests, dry and humid forests, rocky slopes, and grasslands at 2,506-3,019 masl from Amazonas region. The integration of morphological and DNA-based methods is recommended to understand the diversity of berries along the Peruvian Andean cloud forest.
