Allometric Equations for Predicting <i>Agave lechuguilla</i> Torr. Aboveground Biomass in Mexico

Abstract

Quantifying biomass is important for determining the carbon stores in land ecosystems. The objective of this study was to predict aboveground biomass (AGB) of Agave lechuguilla Torr., in the states of Coahuila (Coah), San Luis Potosí (SLP) and Zacatecas (Zac), Mexico. To quantify AGB, we applied the direct method, selecting and harvesting representative plants from 32 sampling sites. To predict AGB, the potential and the Schumacher–Hall equations were tested using the ordinary least squares method using the average crown diameter (Cd) and total plant height (Ht) as predictors. Selection of the best model was based on coefficient of determination (R2 adj.), standard error (Sxy), and the Akaike information criterion (AIC). Studentized residues, atypical observations, influential data, normality, variance homogeneity, and independence of errors were also analyzed. To validate the models, the statistic prediction error sum of squares (PRESS) was used. Moreover, dummy variables were included to define the existence of a global model. A total of 533 A. lechuguilla plants were sampled. The highest AGB was 8.17 kg; the plant heights varied from 3.50 cm to 118.00 cm. The Schumacher–Hall equation had the best statistics (R2 adj. = 0.77, Sxy = 0.418, PRESS = 102.25, AIC = 632.2), but the dummy variables revealed different populations of this species, that is, an equation for each state. Satisfying the regression model assumptions assures that the predictions of A. lechuguilla AGB are robust and efficient, and thus able to quantify carbon reserves of the arid and semiarid regions of Mexico.

Keywords

• allometric equations
• aboveground biomass
• Schumacher–Hall
• dummy variables
• robust regression