A comparison of interspecific and intraspecific phenotypic variation in spectral signatures of ferns with robust versus uncertain species boundaries.

Species delimitation methods based on macromorphology are often limited by phenotypic plasticity in plants. Fourier Transform near-infrared spectroscopy (FT-NIR) provides a promising alternative as a non-destructive technique that measures molecular vibrations (overtone and combination bands of C-H, N-H, and O-H bonds) from plant tissue exposed to near-infrared light (780-2,500 nm). We applied FT-NIR to the taxonomically challenging Scaly clade of Microgramma ferns (94 samples, eight species), including dimorphic and monomorphic taxa, to evaluate its diagnostic potential. Using multivariate models and cross-validation, we achieved 81-100% average identification accuracy. Well-defined species (e.g., M. percussa) reached 100% accuracy, while morphologically overlapping taxa showed lower accuracy, likely due to hybridization, introgression, or cryptic variation. Dimorphic species exhibited higher intraspecific spectral variability and lower accuracy linked to differences between fertile/sterile fronds than monomorphic species. FT-NIR proves effective as a complementary tool for fern systematics, elucidating species limits and diversity patterns. Further studies should address how hybridization, introgression, and indumentum affect spectral data.