Table 2.

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

OLS models
Total species richness				Stygobiotic species				Cold stenotherm species
Variable	Coefficient	Standard error	p(t)	Variable	Coefficient	Standard error	p(t)	Variable	Coefficient	Standard error	p(t)
a				b				c
Constant	52.085	7.177	<.001	Constant	1.394	0.589	0.042	Constant	1.237	0.251	<0.001
Area	2.315	0.644	0.001	Area	2.500	0.346	<0.001	Area	0.880	0.265	0.003
Elevation	−15.431	2.277	<.001					Discharge	0.408	0.146	0.010

Autoregressive models
Total species richness				Stygobiotic species				Cold stenotherm species
Variable	Coefficient	Standard error	p(t)	Variable	Coefficient	Standard error	p(t)	Variable	Coefficient	Standard error	p(t)
d				e				f
Constant	51.602	14.363	0.001	Constant	1.697	1.845	.382	Constant	1.299	0.884	.155
Area	1.444	0.835	0.096	Area	2.858	0.789	.006	Area	0.451	0.388	.260
Elevation	−14.736	4.853	0.005					Discharge	0.337	0.292	.260

OLS regression statistics: a) n = 30, R ² = .859; F = 82.282, p < .001, AICc = 144.128; b) n = 12, R ² = .839; F = 52.242, p < .001, AICc = 48.844; c) n = 27, R ² = .762; F = 38.368, p < .001, AICc = 74.552. SR regression statistics: d) n = 30, R ² (pseudo) = .836 (AICc = 152.630), R ² (predictors + space) = .621 (AICc = 177.751); rho = .987, F = 68.707, p < .001; e) n = 12, R ² (pseudo) = .822 (AICc = 55.824), R ² (predictors + space) = .958 (AICc = 38.501); rho = .916, F = 46.217, p < .001; f) n = 27, R ² (pseudo) = .668 (AICc = 87.548), R ² (predictors + space) = .551 (AICc = 95.696); rho = .983, F = 24.173, p < .001.