Results obtained for the validation set employing a five-component model.

Table 2: Results obtained for the validation set employing a five-component model.

	B[a]P (ng mL^-1)			B[b]F (ng mL^-1)
	Nominal	Predicted	%R	Nominal	Predicted	%R
1	12.3	14.0	115	60	51.5	86
1	12.3	13.4	109	40	30.8	77
3	6.5	6.8	105	30	25.4	85
4	6.5	6.2	95	50	47.5	95
5	18.0	19.2	107	30	25.9	86
6	12.3	13.7	112	40	38.0	95
RMSEP (1)	1.2			5.9
REP (%) (2)	9.6			16
LOD (ng mL^-1) (3)	2			10
LOQ (ng mL^-1) (4)	6.7			33
	B[a]A (ng mL^-1)			Chry (ng mL^-1)
	Nominal	Predicted	%R	Nominal	Predicted	%R
1	105	105.3	100	135	133.57	99
2	15	16.8	112	135	162	120
3	150	172.7	115	80	90.4	113
4	60	59.4	99	190	194.9	103
5	150	158.8	106	190	204.2	107
6	105	117.2	112	135	162.1	120
RMSEP (1)	11			17
REP (%) (2)	11			11
LOD (ng mL^-1) (3)	18			32
LOQ (ng mL^-1) (4)	60			107

(1) $R E P = \frac{100 * R M S E P}{\bar{y}},$ , $\bar{y}$ is the mean concentration of the prediction set.

(2) $R M S E P = \sqrt{\frac{\sum_{i = 1}^{I} {(y_{i} - \hat{y_{l}})}^{2}}{N}},$ , $y_{i}$ is the added concentration, $\hat{y_{l}}$ is the predicted concentration and N is the number of samples in the test set.

(3) Limits of detection (LODs) were calculated with the mvc2 interface according to [23,24].

(4) Limits of quantification (LOQs) were calculated with the mvc2 interface according to [23,24], multiplying LOD^*3.33.