Table 3.

Parameter values, standard errors and associated probability levels of ordinary least‐squares (OLS, a, b) and spatial autoregressive (SR, c, d) best‐fit models for nonstygobiotic and noncold stenotherm species richness, and log‐transformed environmental variables

OLS models
Nonstygobiotic species				Noncold stenotherm species
Variable	Coefficient	Standard error	p(t)	Variable	Coefficient	Standard error	p(t)
a				b
Constant	29.306	6.969	<.001	Constant	42.789	5.703	<.001
Area	1.085	0.625	.095	Area	1.396	0.512	.011
Elevation	−0.810	2.211	.001	Elevation	−12.698	1.809	<.001

Autoregressive models
Nonstygobiotic species				Noncold stenotherm species
Variable	Coefficient	Standard error	p(t)	Variable	Coefficient	Standard error	p(t)
c				d
Constant	28.863	14.097	.051	Constant	43.007	11.438	<.001
Area	−0.003	0.819	.997	Area	0.886	0.665	.194
Elevation	−7.762	4.763	.115	Elevation	−12.252	3.865	.004

OLS regression statistics: a) n = 30, R ² = .624; F = 22.443, p < .001, AICc = 142.361; b) n = 30, R ² = .846; F = 74.085, p < .001, AICc = 130.328. SR regression statistics: c) n = 30, R ² (pseudo) = .540 (AICc = 152.363), R ² (predictors + space) = .379 (AICc = 161.376); rho = .987, F = 15.844, p < .001; d) n = 30, R ² (pseudo) = .832 (AICc = 136.895), R ² (predictors + space) = .689 (AICc = 155.261); rho = .987, F = 63.680, p < .001.