Fig 1.
Schematic representation of the deadwood C cycle and variables that affect snag fall rates.
C pools are surrounded by rounded boxes and C fluxes by bold arrows. Controlling or predictor variables are in square boxes and their hypothesized influence on fluxes represented by thin solid arrows pointing to control symbols associated with fluxes. The hypothesized direction of the effect is indicated by symbol in the open circle.
Fig 2.
Factors that influence wood decay—temperature, wood durability and initial diameter—strongly affect snag persistence compared to stand density, which primarily influences snag exposure to breaking forces.
Predictors include mean annual temperature, species wood durability, initial diameter at 1.37 m height, and number of trees per hectare. Horizontal lines show average effect sizes for all snags and filled circles show effect sizes for subsets representing different progressive decay classes. Outer horizontal lines represent 95% credible intervals (CIs) for all snags. Thin vertical lines (whiskers) with filled circles represent 95% CIs based on 5000 independent draws from the posterior distribution of parameters and imputed species wood durability. All predictors increased the probability of snag persistence except mean annual temperature, which decreased the probability of persistence. Other factors were examined but were not significant.
Fig 3.
Increasing temperature reduces snag persistence, while species wood durability, diameter and stand density increase snag persistence.
The thick central curves correspond to the mean of the posterior distribution for each effect, and the transparent curve overlay represents uncertainty by showing 100 curves drawn from the posterior distribution of the relevant parameters. The open circles in panel B correspond to the posterior means for all 205 species in the dataset with thin vertical bars representing 95% credibility intervals (CIs) for species-level effects and thin horizontal bars representing the 95% CI for imputed species-level wood durability. The vertical spread of the violin plots in panels A, C and D represent the distributions of the predictor values for observed standing (S = 1) versus fallen snags (S = 0).
Fig 4.
Residual species differences, spatial locations and physiographic classes have decreasing influence on variation in snag fall rates.
Horizontal lines represent random effect standard deviations for all snags and filled circles represent random effect standard deviations for different progressive decay classes. Outer horizontal lines represent 95% credible intervals (CIs) for all snags. Thin vertical lines (whiskers) with filled circles represent 95% CIs based on 5000 independent draws from the posterior distribution of parameters. The standard deviations correspond to σS (species, Eq (5)), σG (space, Eq (4)), and σP (physiography, Eq (3)).
Fig 5.
Snags in the upper Midwest have a higher residual probability of falling that snags in the southeasten United States.
The surface is a bilinear interpolation over values estimated at the centers of 151 square (1.7° x 1.7°) grid cells. The bar at the right indicates the color scale denoting probability of snag persistence, with green colors representing slower than expected rates of snag fall after controlling for other drivers (predictor variables) that vary spatially, including tree species composition, stand density, physiographic class, and temperature.