Table 1 Model comparisons of QuaSSE analyses of continuous drivers with significant predictive ability in the XGBoost model, with AIC scores of the best-fitting and null models

From: Identifying the multiple drivers of cactus diversification

Variable and sample size

Best-fitting model

ΔAIC of best fitting model versus constant model (null)

Midpoint value (back-transformed to the original scale if they were log-transformed), and drift parameter

Mean diurnal air temperature range (n = 850)

Modal with drift

233.98

10.20 °C, 0.024

Soil sand content (n = 850)

Modal with drift

65.53

48.38%, 0.00045

Plant size (height or length) (n = 750)

Modal with drift

76.72

45.74 cm, −0.060

Isothermality (n = 850)

Modal with drift

237.10

0.55, 0.0012

Geographic range size (n = 850)

Modal with drift

307.86

0.23 AOO, −0.45

  1. Delta AIC is >4 for all best-fitting models versus second best-fitting models with p < 0.0001 for each of the best-fitting models.