Spectral signatures as evidence to test hypotheses in plant species complexes

Abstract

Traditional methods for species definition, based on macromorphological characteristics, face limitations due to the high phenotypic plasticity observed in plants, which makes it challenging to accurately identify species complexes. Fourier transform near-infrared spectroscopy emerges as a promising, non-destructive technique for analyzing plant material, enabling the distinction of species. This study focuses on the Scaly clade of the fern genus Microgramma, characterized by complex taxonomic boundaries and morphological variations. A total of 94 samples from eight species, including fertile and sterile leaves of dimorphic and monomorphic species, were evaluated to test the effectiveness of FT-NIR in distinguishing these lineages. The average identification accuracy ranged from 86% to 91%, depending on the models and validation employed. Species with better-defined morphological characteristics, such as Microgramma percussa, achieved an accuracy of 100%. Conversely, species with higher morphological overlap showed lower accuracy, which may be associated with hybridization, introgression, or cryptic variation. Dimorphic species, with morphologically distinct fertile and sterile fronds, exhibited higher intraspecific spectral variation compared to monomorphic species, which may explain their lower accuracy rates. Fertile fronds, in some cases, provided more informative data, possibly due to the presence of sori increasing the complexity of spectra. This study highlights the potential of FT-NIR as a complementary tool in plant systematics. However, further research is needed to understand the influence of processes such as hybridization and features such as the indumentum on spectral readings. Overall, FT-NIR presents itself as a promising method to elucidate species limits in ferns and improve knowledge about their diversity.