Table 2. Results of the PLS regression for the whole data set.

	ANOVA				Model Selection and Validation			Disease Predictors
All data	P-value	F	D.F.	Component	Global R2	Predicted Global R2	Salinity	POC	Tmax 7d
PLS ANOVA	<0.0001	111.88	167	1	0.65	0.62	−0.6
PLS Model				2	0.67	0.63		0.32
				3	0.67	0.63			0.08

Results of the final model for the combined data set of AN versus the most important disease predictors: salinity, particulate organic carbon (POC) and maximum temperature 7 days preceding and including the sampling date (Tmax 7d). The PLS model was highly significant with its first three components explaining 74% of the variation. Following cross-validation the model still explained 36% of the variation.