Table 1. Results of the survival analysis (Cox regression model).

The effects on LT (seconds) of the explanatory variables, species, distance to the MPA and their interaction and the density of fish were retained by the AIC-based model selection. The regression coefficients (coef) of the logarithm of the hazard ratio and its standard error (s.e.), the exponent of the coefficient for interpretation, the z-value and p-value of the minimal adequate Cox regression model are shown.

Variable tested	coef	exp(coef)	s.e.(coef)	z-value	Pr(>|z|)
C. julis	3.5360	34.34	0.9145	3.87	<0.001
D. annularis	2.3700	10.70	0.9192	2.58	<0.05
S. scriba	4.3680	78.90	0.9176	4.76	<0.001
Distance (m)	-0.0001	0.9999	0.0001	-0.55	0.581
Other fish	0.0091	1.0090	0.0056	1.63	0.104
C. julis × D	0.0001	1.0000	0.0001	0.46	0.648
D. annularis × D	0.0001	1.0000	0.0001	1.01	0.314
S. scriba × D	-0.0003	0.9997	0.0002	-2.00	<0.05

Concordance = 0.724 (s.e. = 0.026). R-square = 0.331 (max possible = 0.991).