2025/04/09 by Luke Mander, Adele C. M. Julier, Carlos Jaramillo · 1 voice · 1 citation
Agricultural and Biological Sciences · Biochemistry, Genetics and Molecular Biology · #Plant and animal studies #Biochemical and biochemical processes #Botanical Research and Applications
paper · doi:10.1093/botlinnean/boaf032
Abstract Flower morphology varies greatly between different plant taxa, and the variety of flower types has stimulated a number of different approaches to the analysis of floral form. In this paper we have thought of flower morphology as a network, with individual morphological characters and taxa linked as a system. We scored the flowers of 951 species of plants in a Neotropical rainforest (Barro Colorado Island, Panama) for nine discrete characters that have a total of 35 character states and constructed a bipartite network connecting nodes representing species to nodes representing morphological characters. The basic unit of this bipartite network is a single node that is connected to nine other nodes: a star. The creation of larger networks results in variation in network architecture, which arises as a consequence of the patterns of connectivity between taxa and morphological characters. We find that the network of BCI flower morphology is characterized by disassortative mixing in which nodes with high degree are preferentially connected to nodes with low degree. From an architectural perspective this is a consequence of the prevalence of stars in our networks. From an ecological perspective this means that the most common individual characters are only rarely combined into a single flower. The floral characters we have analysed here are distributed non-randomly among five plant groups on BCI: herbs, lianas, epiphytes, shrubs, and trees. Trees show a negative association with sympetaly, zygomorphy, and epigyny while herbs show the opposite patterns of association.