Table 4.

Generalized additive model of location, shape and scale (GAMLSS) fits for mean abundance of Gyrodactylus cichlidarum (MAG; dependent variable) in Nile tilapia farms in Yucatán state. The best model was selected using a stepwise procedure and the lowest values of Akaike information criterion (AIC) and global deviance. The final model was: MAG ~ cs(Do) + cs(Tc) + Qa + Pt where cs is a cubic spline smooth function, and those independent variables without smoother had a linear relationship with the dependent variable (MAG)

Model	Degrees of freedom (df)	Global deviance	Deviance explained (%)	AIC	P (0.05)
(Null model) MAG ~ 1	-	130.82	-	132.82	5.4e-6*
MAG ~ cs(Do)	df =16	100.52	23.16	110.51	0.0002*
MAG ~ cs(Do) + cs(Tc)	df =16	92.65	29.17	110.65	0.0059*
MAG ~ cs(Do) + cs(Tc) + Qa	df =19	86.29	34.03	106.29	0.0208*
MAG ~ cs(Do) + cs(Tc) + Qa + Pt	df =17	48.05	63.27	74.05	0.0001*

Abbreviations: Do Dissolved oxygen (mg/l), Tc Tank capacity (m³), Qa Use of quarantine area, Pt Use of prophylactic treatments

*P value < 0.05