Table 2: Results from bootstrapped boosted tree regression models for measures of forest productivity.

From: Intertidal resource use over millennia enhances forest productivity

Table 2: Results from bootstrapped boosted tree regression models for measures of forest productivity.
Response variableExplanatory variables in order of relative importance in explaining varianceTraining data correlationNumber of treesPercent deviance explained
A) Forest canopy heightASP (28.0), DIST (24.6), SLO (15.7) ELE (13.0), CLA (8.9), UPS (7.6), EXP (2.2), MOR (0.1)0.881,35076.0
B) Forest widthDIST (38.0), ASP (22.5), ELE (10.6), SLO (8.2), UPS (8.0), CLA (7.5), EXP (4.2), MOR (0.8)0.8195064.2
C) Vegetation greennessASP (37.5), SLO (27.4), DIST (15.1), CLA (8.5), UPS (5.2), ELE (3.7), EXP (2.7), MOR (0.02)0.8270065.7
D) Forest canopy coverSLO (20.5), ASP (19.3), ELEV (19.0), DIST (18.9), CLA (11.6), UPS (7.4), EXP (2.6), MOR (0.5)0.791,00060.0
  1. For each response variable (forest canopy height, forest width, vegetation greenness, and forest canopy cover), 8 explanatory variables were used to explain model deviance [aspect (ASP), distance from habitation site (DIST), mean slope (SLO), mean elevation (ELE), shoreline classification (CLA), upstream area (UPS), shoreline exposure (EXP) and surface material and coastal morphology (MOR)].