Table 2 Summary statistics for the top-ranked generalised least-squares models.

	Area	Variable	Coefficient		Correlation
			value	std error	P_a	EminP_a	DF_a	P_a × EminP_a	DF_a × EminP_a	NPP_b	H_b	P_b	EminP_b
Ext_t ~ P_a + EminP_a + DF_a + (P_a × EminP_a) + (DF_a × EminP_a)	n/coexist	P_a	−4098.61	555.92	1	0.01	0.63	−0.66	−0.28
		EminP_a	378.48	784.24		1	0.29	0.42	−0.18	–	–	–	–
		DF_a	−513.70	1044.20			1	−0.39	−0.69	–	–	–	–
		P_a × EminP_a	3592.36	1340.64				1	−0.29	–	–	–	–
		DF_a × EminP_a	−10519.01	3238.69					1	–	–	–	–
Ext_b ~ NPP_b	n/coexist	NPP_b	−0.38	0.09						1	–	–	–
Ext_b ~ H_b + P_b + EminP_b	coexist	H_b	0.34	0.09							1	0.04	0.33
		P_b	−0.18	0.09								1	−0.18
		EminP_b	−0.13	0.07									1

Coefficients estimated for each predictor variable of the top-ranked generalised least-squares model, standard error, and correlation between predictor variables to describe the timing of megafauna extirpation (Ext_t) and its bearing (Ext_b) in human-megafauna non-coexistence (n/coexist) and coexistence (coexist) areas. Climate variables are mean annual temperature (T), mean annual precipitation (P), mean annual freshwater availability (EminP), mean annual net primary production (NPP), and the fraction of desert within the grid cell (DF), and the human predictor is the estimated timing of initial human arrival (H). Predictor variables subscripted a (P_a, EminP_a, DF_a) indicate that we used the mean annual anomaly relative to the period 50–30 ka for these variables, whereas predictor variables subscripted b (P_b, NPP_b, EminP_b, H_b) indicate that we used the directional bearing. For clarity, we did not present the results of the model best describing Ext_t in the areas with human and megafauna coexistence because we did not have any model with a percentage of variance explained > 0% in those areas

Quick links

Search