Figure 3: An SR model for the overstoryunderstory relationship in mature Douglasfir forests. The exogenous latent variables (and associated observed variables) are (1) aspect (ASPECT × 10), (2) overstory cover (TREE), (3) tree size (QMD), (4) stand age (AGE), and (5) stand density (ln(TPH) × 10). The endogenous latent variables (and associated observed variables) are (1) fine litter (FLITTER), (2) coarse litter (CLITTER), (3) understory cover (UNDER), and (4) lateseral herb cover (LSHERB). All factor loadings are fixed at a value of 1.0. Measurement errors are fixed with a priori values as depicted by the numerical values with arrows pointing towards the observed variables. Arrows pointing towards the endogenous latent variables depict errors of the variables. Single arrowheads linking two latent variables depict direct effects and the double arrowheads depict correlation between errors of fine litter and coarse litter latent variables. 
