Table 2.
Results of comparative models across drivers.
| Driver | Full Model 1 (n = 1406) | Full Model 2 (n = 1,096) | Full Model 3 (n = 640) | |||
|---|---|---|---|---|---|---|
| Num. DF | F-Value | Num. DF | F-Value | Num. DF | F-Value | |
| Stand age | 1 | 1760.9c | 1 | 1493.0c | 1 | 650.3c |
| Genus | 9 | 7.8c | 9 | 8.3c | 9 | 6.7c |
| Endemism | 2 | 3.1a | 2 | 3.3a | 1 | 0.2 |
| Prior land use | 3 | 8.8c | 3 | 24.3c | 3 | 18.1c |
| Biome | 10 | 2.2a | 9 | 1.2 | 9 | 3.1b |
| Leaf type, phenology, & N Fixation | 1 | 21.7c | 1 | 20.3c | 1 | 7.3b |
| Wood density | 1 | 0.8 | 1 | 1.2 | 1 | 5.7a |
| Planting density | – | – | 1 | 0.2 | – | – |
| Use of fertilizer | – | – | – | – | 1 | 10.8b |
aIndicates significance at the 0.05 level,
bindicates significance at the 0.01 level, and
cindicates significance at the 0.001 level.
Full Models 1–3 testing the relative effects across different types of potential drivers. Significance was determined using two-sided F-tests with Satterthwaite approximations of degrees of freedom. For each model, square root transformed aboveground carbon is modeled as a linear combination of the listed drivers with site ID included as a random intercept. Stand age was square root transformed to linearize its relationship with aboveground carbon.