The comparison of the molecular analysis with morphological and anatomical data presented here represents an important basis for a new formal classification for the Araceae and for the understanding of the evolution of this ancient family, a monocot group known in the fossil record from the early Cretaceous.

Familial, subfamilial, and tribal monophyly and relationships of aroids and duckweeds were assessed by parsimony and Bayesian phylogenetic analyses of five regions of coding (rbcL, matK) and noncoding plastid DNA (partial trnK intron, trnL intron, trnL-trnF spacer) for exemplars of nearly all aroid and duckweed genera. Our analyses confirm the position of Lemna and its allies (formerly Lemnaceae) within Araceae as the well-supported sister group of all aroids except Gymnostachydoideae and Orontioideae. The last two subfamilies form the sister clade of the rest of the family. Monophyly of subfamilies Orontioideae, Pothoideae, Monsteroideae, and Lasioideae is supported, but Aroideae are paraphyletic if Calla is maintained in its own subfamily (Calloideae). Our results suggest expansion of the recently proposed subfamily Zamioculcadoideae (Zamioculcas, Gonatopus) to include Stylochaeton and identify problems in the current delimitation of tribes Anadendreae, Heteropsideae, and Monstereae (Monsteroideae), Caladieae/Zomicarpeae, and Colocasieae (Aroideae). Canalization of traits of the spathe and spadix considered typical of Araceae evolved after the split of Gymnostachydoideae, Orontioideae, and Lemnoideae. An association with aquatic habitats is a plesiomorphic attribute in Araceae, occurring in the helophytic Orontioideae and free-floating Lemnoideae, but evolving independently in various derived aroid lineages including free-floating Pistia (Aroideae).

Isolated granitic rock outcrops or 'inselbergs' may provide a window into the molecular ecology and genetics of continental radiations under simplified conditions, in analogy to the use of oceanic islands in studies of species radiations. Patterns of variability and gene flow in inselberg species have never been thoroughly evaluated in comparison to related taxa with more continuous distribution ranges, or to other species in the same kingdom in general. We use nuclear microsatellites to study population differentiation and gene flow in two diploid, perennial plants adapted to high-altitude neotropical inselbergs, Alcantarea imperialis and Alcantarea geniculata (Bromeliaceae). Population differentiation is pronounced in both taxa, especially in A. imperialis. Gene flow in this species is considerably lower than expected from the literature on plants in general and Bromeliaceae in particular, and too low to prevent differentiation due to drift (N(e)m< 1), unless selection coefficients/effect sizes of favourable alleles are great enough to maintain species cohesion. Low gene flow in A. imperialis indicates that the ability of pollinating bats to promote gene exchange between inselbergs is smaller than previously assumed. Population subdivision in one inselberg population of A. imperialis appears to be associated with the presence of two colour morphs that differ in the coloration of rosettes and bracts. Our results indicate a high potential for inselbergs as venues for studies of the molecular ecology and genetics of continental radiations, such as the one that gave rise to the extraordinary diversity of adaptive strategies and phenotypes seen in Bromeliaceae.

Highlights• A deep learning approach to quantify discriminatory leaf is proposed.• Shape is not a dominant feature for leaf but rather the different orders of venation.• Deep learning reveals transformation of leaf features from general to specific types.• Findings archived fit with the hierarchical botanical definitions of leaf characters• Features learned using deep learning can improve plant recognition performance. AbstractPlant identification systems developed by computer vision researchers have helped botanists to recognize and identify unknown plant species more rapidly. Hitherto, numerous studies have focused on procedures or algorithms that maximize the use of leaf databases for plant predictive modeling, but this results in leaf features which are liable to change with different leaf data and feature extraction techniques. In this paper, we learn useful leaf features directly from the raw representations of input data using Convolutional Neural Networks (CNN), and gain intuition of the chosen features based on a Deconvolutional Network (DN) approach. We report somewhat unexpected results: (1) different orders of venation are the best representative features compared to those of outline shape, and (2) we observe multi-level representation in leaf data, demonstrating the hierarchical transformation of features from lower-level to higher-level abstraction, corresponding to species classes. We show that these findings fit with the hierarchical botanical definitions of leaf characters. Through these findings, we gained insights into the design of new hybrid feature extraction models which are able to further improve the discriminative power of plant classification systems. The source code and models are available at: https://github.com/cs-chan/Deep-Plant.

No abstract