Identification of plants is a challenging task which aims to identify the family, genus, and species level according to morphological features. Automated deep learning-based computer vision algorithms are widely used for identifying plants and can help users to narrow down the possibilities. However, numerous morphological similarities between and within species make the classification difficult. In this paper, we tested a custom convolution neural network (CNN) and vision transformer (ViT) based models using the PyTorch framework to classify plants. We used a large dataset of 88K and 16K images for classifying plants at genus and species levels respectively. Our results show that for classifying plants at the genus level, ViT models perform better compared to CNN-based models ResNet50 and ResNet-RS-420, and other state-of-the-art CNN-based models suggested in previous studies on a similar dataset. The ViT model achieved top accuracy of 83.3% for classifying plants at the genus level. ViT models also perform better for classifying plants at the species level compared to CNN-based models ResNet50 and ResNet-RS-420, with a top accuracy of 92.5%. We show that the correct set of augmentation techniques plays an important role in classification success.

In tropical Africa, forests and savannas are the two most widespread biomes and potentially represent alternative stable states with divergent species composition. A classic, but untested, hypothesis posited by White (1983) suggests that the transition zones between forests and savannas contain a floristically impoverished assemblage with few representatives from each biome. Further, the evolutionary dimension of diversity has received limited attention, despite its importance for understanding the biogeographic history of biomes. Here, we quantify species richness and several measures of evolutionary diversity in 1° grid cells, using c. 300K occurrence records of trees and shrubs combined with biome affiliation data for 3,125 species. We find that assemblages in transition zones hold fewer woody species than assemblages in forest and savanna zones, as posited by White. However, transition zones hold more phylogenetic diversity than expected given their species richness, whether one considers forest and savanna assemblages separately or together. We also show that the Congo basin forest has low levels of phylogenetic diversity given the number of species and highlight south-eastern African savannas as a centre of savanna woody species richness and phylogenetic diversity. Regions with high phylogenetic diversity given the number of both forest and savanna species were centred around the Dahomey Gap and Cameroon, mainly in transition zones. Overall, our study shows that even if floristically impoverished, transition zones lead to unexpectedly high evolutionary diversity, suggesting they are important centres of evolutionary innovation and diversification.