software was used and the chromatog

software was used and the chromatograms were monitored at
285 nm, 350 and 450 nm.
The identification was made by comparison of their chromatographic and UV–vis spectroscopic characteristics with those of
standards.
The quantification was carried out by external calibration from
the areas of the chromatographic peaks obtained by DAD detection at the following wavelengths: 285 nm for phytoene and TOCS,
350 nm for phytofluene and 450 nm for CARS and CHLS.
2.5. Method validation
Linearity of the method was evaluated by considering of the
detector response (area units) to different amounts (g) of CARS,
CHLS and TOCS by means of linear regression. In order to satisfy
basic requirements such as homoscedasticity and linearity, the Ftest and the residual plot were performed at the 95% significance
level [35,36].
Stock solutions of the individual compounds were prepared separately. The concentrations of the working standards
were determined spectrophotometrically using published molar
absortivity values (εmol) as indicated in Supplementary material.
The desired concentration range for the preparation of standard
curves was obtained by serial dilution. The calibration curves
were constructed by plotting the response area against the corresponding concentration injected. The limits of detection (LOD) and
quantification (LOQ) were calculated from the calibration curves,
using the Microcal Origin ver. 3.5 software (OriginLab Corporation,
Northampton, MA, USA). The LOD were calculated as three times
the relative standard deviation of the analytical blank values calculated from the calibration curve. The LOQ were calculated as ten
times the relative standard deviation of the analytical blank values
calculated from the calibration curve.
To evaluate the precision and recovery of the methods, four samples were selected considering its composition; chard as a source
of both CHLS and CARS (mainly 9

-cis-neoxanthin + violaxanthin,
lutein, and -carotene), carrot as a source of carotenes (mainly, -and -carotene) and TOCS, red peach as a source of -cryptoxanthin
and grapefruit as a good source of lycopene, -carotene and colorless carotenoids (phytoene and phytofluene).
The within-laboratory repeatability (within-day precision) was
developed according to UNE 82009 standards [37]. It was ascertained by analyzing the CARS, TOCS and CHLS content in samples,
under the same analytical conditions. Three replicates from each
sample were extracted and all the samples and standards were
injected three times.
Within-laboratory reproducibility (day-to-day precision) was
assessed by extracting and analyzing the CARS at 2-day intervals
during 3 days.
A recovery study was performed to validate the accuracy of the
developed method. It was expressed as the percentage recovery,
which was calculated as:
%Re =
observed concentration − theoretical concentration
theoretical concentration
× 100
The accuracy of the method was evaluated spiking with known
quantities of different standards (trans--apo-8

-carotenal (APO),
chlorophyll a (CHLA), -carotene (BCAR) and -tocopherol (ATOC).
The spiked samples were then extracted and analyzed with the
proposed RRLC method.
To evaluate the recovery of carotenoids, the samples of chard,
carrot and grapefruit were spiked with 100 l of trans--apo-8

-carotenal (dissolved in acetone at a concentration of 432 mg/l) and
then extracted by the methodology described above (Section 2.3).
Trans--apo-8

-carotenal is a synthetic carotenoid that has been
traditionally used as internal standard in some laboratories [38].
Likewise, to evaluate the recovery of CHLS and TOCS, samples
of banana were spiked with 20 l of CHLA (dissolved in acetone at
a concentration of 910 mg/l) and with 50 l of ATOC (dissolved in
acetone at a concentration of 1870 mg/l) and then extracted by the
methodology described above (Section 2.3).
On the other hand, the effect of saponification in the recovery of
CARS and TOCS was evaluated. For this purpose, the same samples
of banana were spiked with 200 l of BCAR (dissolved in acetone
at a concentration of 540 mg/l) and with 50 l of ATOC (dissolved
in acetone at a concentration of 1870 mg/l). Finally, these samples
were extracted and saponified by the methodology described above
(Section 2.3).
To reduce to the minimum sources of errors, recoveries were
evaluated in banana because this sample had the least complex
carotenoid profile among the products analyzed.
3. Results and discussion
Different assays were carried out to optimize the chromatographic conditions in order to obtain a suitable separation of the
isoprenoids in the extracts. For this, a mixture of standard was
used; Figs. 1 and 2 show the chromatograms of the standard mixture using the optimized conditions, detailed in Section 2.4. The
developed method allows the separation seventeen compounds in
the fruit and vegetables in a 12-min run. Specifically a mixture of
nine CARS (violaxanthin, zeaxanthin, zeinoxanthin -cryptoxathin,
-carotene, -carotene, lycopene, phytoene, phytofluene), four
tocopherols and four chlorophylls and derivatives (chlorophylls
and pheophytins) was separated.
3.1. Analytical characteristics
3.1.1. Linearity and limits of detection and quantification
The validating parameters of each calibration curve (slope,
intercept, coefficients of determination, LOD and LOQ) are shown
in Table 1. The linear calibration range used was tested for
homoscedasticity to confirm the application of the linear leastsquares method (constant variance).
All curves showed good linearity (R
2
> 0.996) in the range of concentrations studied. As regards to carotenoids, LODs ranged from
0.001 g in phytofluene to 0.070 g in lycopene. These limits are
lower than those recently reported [39].
Concerning the tocopherols, LODs ranged between 0.007 g for
-tocopherol and 0.067 g for -tocopherol, while those of chlorophylls ranged from 0.004 g (for pheophytin b) to 0.080 g (for
chlorophyll b).
Depending on the compound, LOQs ranged from 0.002 g to
0.268 g (for phytofluene and chlorophyll b, respectively). The
results obtained for the quantitation limits show that the proposed
method is sensitive enough for the determination of dietary isoprenoid compounds of interest.
3.2. Precision and accuracy
The repeatability and reproducibility were evaluated by considering the relative standard deviation (Table 2). Concerning
repeatability, the RSD values of the method for unsaponified
samples were under 7.2%. The highest values corresponded to -carotene (7.18%) and the lowest ones to violaxanthin (0.58%). Both
values were obtained in the analyses of carrot. The highest RSD
observed in the reproducibility corresponded to -tocopherol in
the carrot sample (11.87%) and the lowest ones to lutein (4.66%)
in the chard sample. Nonetheless, most of the RSD values obtained
were below 12%, which confirmed the high reproducibility of the
method