Dynamic growth models for Caragana korshinskii shrub biomass in China. (1st September 2020)

• Record Type: Journal Article
• Authors: Xu, Hao; Wang, Zhanjun; Li, Ying; He, Jianlong; Wu, Xudong
• Is Part Of: Journal of environmental management. Volume 269(2020)
• Publication Date: 2020-09-01
• Subjects: Dynamic growth models -- Fixed and mixed effects -- Heteroscedasticity -- Caragana korshinskii shrub; Environmental policy -- Periodicals; Environmental management -- Periodicals; Environment -- Periodicals; Ecology -- Periodicals; 363.705
• DOI: 10.1016/j.jenvman.2020.110675
• Languages: English
• ISSNs: 0301-4797

Abstract

Abstract: The Caragana korshinskii shrub is a widely distributed plant found in arid regions and plays an important role in ecological environment protection. Accurate estimations of shrub biomass are particularly important for natural resource management decision making. 114 individual C. korshinskii shrubs from three regions were collected in northwest China in this study. With regions as fixed (dummy variables) and random effects, the nonlinear least square ( NLS ) regression approach, nonlinear fixed effects ( NLFE ) approach and nonlinear mixed effects ( NLME ) approach were developed to predict dynamic growth of total, aboveground, stem, foliage, and root biomass values of C. korshinskii shrub based on logistic function. Results revealed that both NLFE and NLME models performed better than NLS, which indicated that regions were important factors influenced shrub biomass dynamic growth. Additionally, NLME models had a smaller Bayesian information criterion (BIC) than NLFE models. For NLME models, the random effects of regions mainly influenced the growth rate and asymptotic value of the dynamic growth curve, and there was no significant influence on the values associated with the curve shape. Moreover, the modified NLME models with heteroscedasticity exhibited extremely significant differences ( p < 0.0001) when compared to NLME models by the likelihood ratio (LR) test. The NLME models were proved to be an efficient approach for considering the random effects on shrub biomass dynamic growth and accounted for the heteroscedasticity of shrub biomass data. Highlights: Shrub components biomass growth varied from region to region. Nonlinear fixed ( NLFE ) and mixed effects ( NLME ) models of shrub dynamic biomass growth were developed. The heteroscedasticity could be accounted for NLME models with constant power function. Random effect coefficients are documented for future use in estimating shrub biomass. Regions as random effects showed significantly influence on the growth rate and asymptotic value of function.