Spectral network analysis illuminates coordinated trait adaptation across plant populations

Understanding how plant populations respond to environmental variation through functional leaf traits remains challenging due to limitations of traditional phenotyping approaches. Hyperspectral reflectance offers a rapid, non-destructive and high-throughput method to capture functional trait variation and detect signatures of local adaptation across populations.

We combined hyperspectral data, inverse modeling, and network analysis to investigate population-level adaptations in Streptanthus tortuosus . Using a common garden experiment with four populations, we applied partial least square discriminant analysis (PLS-DA) and ridge regression for population discrimination, inverse PROSPECT modeling to estimate leaf biochemical traits, and canonical correlation analysis to examine trait-climate relationships across historical (1900-1994) and recent (1995-2024) periods. We developed a novel spectral network approach treating wavelength correlations as biologically meaningful trait networks.

Populations showed distinct, heritable spectral signatures with high classification accuracy. Significant population differences emerged in anthocyanins, carotenoids, chlorophyll, and water content. Trait-climate correlations shifted between time periods, consistent with historical climate adaptation. Network analysis revealed population- specific integration patterns, with more variable environments displaying greater spectral modularity.

Hyperspectral signatures provide a high-throughput tool for detecting population-level adaptation and trait coordination. Our findings suggest plant populations respond to climate change through evolved shifts in trait networks rather than isolated traits alone.