Table 1.
Models’ explanatory power and partitioning of variation in biomass, growth and probability of occurrence to different components.
| Species | Data source (model) | Explanatory power | Variation partitioning (occurrence/biomass) | ||||
|---|---|---|---|---|---|---|---|
| Tjur-R2 (occurrence) | R2 (biomass) | Joint effect of temperature and salinity | Effect of depth | Effect of Fucus biomass | Spatial random effect | ||
| Fucus | Experiments | 0.47 | 0.50 | ||||
| Distribution | 0.42 | 0.52 | 0.62/0.20 | 0.18/0.35 | 0.20/0.45 | ||
| Distribution + Experiments | 0.39 | 0.52 | 0.30/0.19 | 0.34/0.39 | 0.37/0.42 | ||
| Idotea | Experiments | 0.09 | |||||
| Distribution | 0.59 | 0.63/ | 0.08/ | 0.29/ | |||
| Distribution + Experiments | 0.66 | 0.58/ | 0.09/ | 0.33/ | |||
Explanatory power is measured by Tjur-R2 (occurrence models) and R2 (biomass and growth models) statistics, which measure how well the models explain the training data (n = 2000). The variation partitioning summarizes what proportion of the total variation in expected biomass, growth and expected occurrence (in log odds ratio scale) over the data points is explained by different model components.