Numerical Ordination and Permutation-Based Inference

Abstract

This study used Redundancy Analysis (RDA) via a Partial Least Squares Regression (PLSR) surrogate with Monte Carlo permutation tests to examine how environmental gradients shape the biochemical and antioxidant traits of Moroccan carob (Ceratonia siliqua L.). Six geo-environmental predictors (altitude, latitude, longitude, mean annual temperature, soil pH, rainfall) were modeled against bioactive compounds (polyphenols, flavonoids, tannins, proteins, minerals) and antioxidant indices (FRAP, IC50 DPPH, IC50 ABTS). The RDA biplot showed clear ecological patterns: altitude, pH, and rainfall favored polyphenolic and lipid accumulation, while magnesium and tannins reflected xeric conditions. Antioxidant indices responded differently, with FRAP linked to altitude and magnesium, and IC50 DPPH/ABTS clustering with low moisture and higher fat content. Permutation tests (n = 999) confirmed model robustness (global R2 = 0.454, p < 0.0001), highlighting constrained ordination as a reproducible tool for quantifying ecological structuring of molecular traits.