Table 2.
Multiple-predictor regression models to explain global species richness of climbing (n = 534) and non-climbing (n = 1911) palms.
| Palm species richness |
||||||
|---|---|---|---|---|---|---|
| Climbers |
Non-climbers |
|||||
| Standard coefficient | P | Standard coefficient | P | |||
| Intercept | 2.798 | *** | 2.413 | *** | ||
| CANOPY | 0.403 | ** | 0.227 | * | ||
| CANOPY2 | NA | 0.063 | n.s. | |||
| PREC | - | 0.271 | n.s. | |||
| TEMP | 0.421 | *** | 0.539 | *** | ||
| PREC SEAS | -0.356 | * | 0.219 | n.s. | ||
| PREC SEAS2 | NA | 0.044 | n.s. | |||
| TEMP SEAS | - | -0.507 | ** | |||
| TEMP SEAS2 | - | NA | ||||
| PREC DRY | - | NA | ||||
| PREC DRY2 | NA | NA | ||||
| TEMP COLD | - | NA | ||||
| TEMP COLD2 | - | NA | ||||
| LGMPREC | - | - | ||||
| LGMPREC2 | - | NA | ||||
| LGMTEMP | - | 0.220 | n.s. | |||
| PLIOPREC | -0.274 | * | -0.070 | n.s. | ||
| PLIOPREC2 | NA | -0.201 | * | |||
| PLIOTEMP | - | - | ||||
| MIOPREC | -0.183 | n.s. | -0.001 | n.s. | ||
| MIOPREC2 | NA | -0.190 | * | |||
| MIOTEMP | - | - | ||||
| REGION | ||||||
| Afrotropics | -1.359 | *** | -1.098 | *** | ||
| Australasia | -1.786 | *** | 0.078 | n.s. | ||
| Neotropics | -1.773 | *** | 0.658 | n.s. | ||
| Oceania | NA | -0.057 | n.s. | |||
| Nearctic | NA | -0.854 | n.s. | |||
| Palearctic | NA | -0.855 | * | |||
| AIC | 200 | 443 | ||||
| R2 | 0.69 | 0.70 | ||||
| Moran’s I | -0.13 | n.s. | 0.11 | n.s. | ||
Based on the AIC, a minimum adequate model was selected from the full set of predictor variables after removing those variables that showed high multicollinearity. Standardized coefficients are given for continuous predictor variables (scaled before the analysis to mean = 0 and SD = 1). The effect of the categorical variable REGION is relative to Indomalaya. Statistically significant effects of continuous and categorical variables are highlighted in bold. Sampling units are International Working Group on Taxonomic Databases (TDWG) level 3 units (n = 82 for climbers; n = 164 for non-climbers). ‘–’ indicates not selected variables, ‘NA’ not available variables or categorical levels. Residual spatial autocorrelation was tested using Moran’s I values based on the closest neighbor of each TDWG level 3 unit. Species richness and several continuous predictor variables (TEMP, PREC SEAS, TEMP SEAS, PREC DRY, LGMTEMP) were log10 transformed. Abbreviations of predictor variables: CANOPY, canopy height; PREC, annual precipitation; TEMP, annual mean temperature; PREC SEAS, precipitation seasonality; TEMP SEAS, temperature seasonality; PREC DRY, precipitation of driest quarter; TEMP COLD, Mean temperature of coldest quarter; LGMPREC, Last Glacial Maximum precipitation anomaly; LGMTEMP, Last Glacial Maximum temperature anomaly; PLIOPREC, Pliocene precipitation anomaly; PLIOTEMP, Pliocene temperature anomaly; MIOPREC, Miocene precipitation anomaly; MIOTEMP, Miocene temperature anomaly; REGION, biogeographic region (categorical variable, effects are relative to the Indomalayan region). Significance levels: ***P < 0.001; **P < 0.01; *P < 0.05; n.s., not significant. R2, explained variance.