Desarrollo y selección de un índice espectral basado en el intervalo visible para diferenciar posiciones de muestreo en Stenocereus queretaroensis

Resumen

Spectral characterization of vegetation using reflectance indices commonly relies on relationships between spectral bands selected according to their ability to discriminate conditions or characteristics of interest. However, formulations operating exclusively within the visible range are relevant when spectral acquisition is restricted to 400–700 nm. This study developed and evaluated a procedure for selecting a spectral relationship capable of differentiating sampling positions P0 and P150 in Stenocereus queretaroensis. Thirty-one spectral bands at 10 nm intervals were analyzed, and 1,860 candidates were generated from four formulations: difference, ratio, normalized difference, and spectral slope. Candidates were evaluated using paired P0–P150 comparisons by campaign, Benjamini–Hochberg false discovery rate correction, generalized estimating equation models, and plant-grouped cross-validation. The 520–540 nm relationship showed one of the most consistent behaviors among the candidate set and was significant in all 13 campaigns. The normalized ND520–540 formulation, defined as (R520 − R540)/(R520 + R540), showed a mean P150−P0 difference of −0.0271 and a standardized paired effect size of d_z = −0.850; the GEE model yielded β = −0.0308 (p = 2.59 Å~ 10⁻..). Five-fold plant-grouped cross-validation showed consistent selection of the proposed formulation across held-out plant partitions. The results indicate that the 520–540 nm relationship represents a consistent member of a central spectral family, without demonstrating unequivocal statistical superiority over all nearby formulations. ND520–540 is therefore proposed as the primary normalized formulation for differentiating P0 and P150 under the experimental conditions studied. Independent validation is required before assessing its generalization to new plants, campaigns, or acquisition conditions.