Abstract
Introduction and objectives
– Effective monitoring of threatened plant populations requires non-destructive methods to track individual biomass accumulation. For procumbent Cactaceae associated with ironstone outcrops, we found no validated non-destructive allometric models. We developed and calibrated a non-destructive allometric model for
Arthrocereus glaziovii
, an Endangered microendemic of the ironstone campo rupestre of the Quadrilátero Ferrífero, Brazil, targeted by rescue and reintroduction programmes.
Material and methods
– Seventeen individuals were destructively sampled to calibrate six candidate models combining zenithal image morphometry (Projected Area) with structural variables measured directly on the individuals (Height, Number of Branches). Predictors were screened for collinearity using the variance inflation factor, and the model was selected using the Akaike information criterion corrected for small samples. Generalisation was assessed by leave-one-out cross-validation, reproducibility by the intraclass correlation coefficient and Bland-Altman analysis, and robustness by sensitivity analysis.
Key results
– The resulting model integrates Projected Area and Number of Branches (Ŷ = 0.1611 × Area^0.7058 × Number of Branches^0.4027), explaining 81.7% of biomass variation on the logarithmic scale (mean absolute percentage error of 30.2%; 34.8% under cross-validation). Interobserver reproducibility was high (intraclass correlation coefficient = 0.9952; 95% confidence interval: 0.9856–0.9983). Height contributed no explanatory power in any formulation (the Height-only model had a corrected Akaike information criterion 26.28 units higher than the selected model), consistent with the procumbent habit on open-canopy ironstone outcrops. Population-level monitoring requires n ≥ 29 individuals for ±15% precision at 95% confidence (residual error only).
Conclusion
– The calibrated model indicates that Projected Area and Number of Branches, rather than Height, explain most of the biomass variation in this procumbent cactus. The model requires no specialised equipment for reintroduction programmes, and its functional form provides a testable starting point for other rupicolous Cactaceae of similar habit, pending independent recalibration and validation.