Effect of agronomical practices on carpology, fruit and oil composition, and oil sensory properties, in olive (Olea europaea L.)

Food Chemistry 159 (2014) 236–243

Contents lists available at ScienceDirect

Food Chemistry

journal homepage: www.elsevier .com/locate / foodchem

Effect of agronomical practices on carpology, fruit and oil composition,and oil sensory properties, in olive (Olea europaea L.)

http://dx.doi.org/10.1016/j.foodchem.2014.03.0140308-8146/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +39 074349743; fax: +39 074343634.E-mail address: [email protected] (A. Rosati).

Adolfo Rosati a,⇑, Caterina Cafiero b, Andrea Paoletti a, Barbara Alfei c, Silvia Caporali a, Lorena Casciani b,Massimiliano Valentini d

a Consiglio per la Ricerca e la sperimentazione in Agricoltura – Centro di Ricerca per l’olivicoltura e l’industria olearia, CRA-OLI, Via Nursina 2, 06049 Spoleto, PG, Italyb Consiglio per la Ricerca e sperimentazione in Agricoltura – Centro di ricerca per lo studio delle relazioni fra pianta e suolo, CRA-RPS, Strada della Neve, S.P. Pascolare Km 1,Monterotondo, 00015 Roma, Italyc Agenzia Servizi Settore Agroalimentare Marche, ASSAM, Via dell’Industria 1, 60027 Osimo, AN, Italyd Consiglio per la Ricerca e sperimentazione in Agricoltura – Centro di ricerca per gli alimenti e la nutrizione, CRA-NUT, Via Ardeatina 546, 00178 Roma, Italy

a r t i c l e i n f o

Article history:Received 31 October 2013Received in revised form 26 February 2014Accepted 4 March 2014Available online 13 March 2014

Keywords:HRMAS-NMRPLS-DAFrantoioLeccinoOrganicConventionalOliveOlea europaea

a b s t r a c t

We examined whether some agronomical practices (i.e. organic vs. conventional) affect olive fruit and oilcomposition, and oil sensory properties. Fruit characteristics (i.e. fresh and dry weight of pulp and pit, oilcontent on a fresh and dry weight basis) did not differ. Oil chemical traits did not differ except forincreased content of polyphenols in the organic treatments, and some changes in the acidic composition.Sensory analysis revealed increased bitterness (both cultivars) and pungency (Frantoio) and decreasedsweetness (Frantoio) in the organic treatment. Fruit metabolomic analysis with HRMAS-NMR indicatedsignificant changes in some compounds including glycocholate, fatty acids, NADPH, NADP+, some aminoacids, thymidine, trigonelline, nicotinic acid, 5,6-dihydrouracil, hesanal, cis-olefin, b-D-glucose, propanaland some unassigned species. The results suggest that agronomical practices may have effects on fruitcomposition that may be difficult to detect unless a broad-spectrum analysis is used.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Organic agriculture has been shown to have lower environmen-tal impact, due to lower pollution with chemicals and syntheticfertilizers (Reganold, Glover, Andrews, & Hinman, 2001), and lowerpresence of chemical residues in organic products (Bourne & Pres-cott, 2002; Reganold et al., 2001). Whether organic fruits are ofhigher quality than conventional ones remains controversial. High-er quality of organic fruit is often detected by chemical and sensoryanalyses. For instance, organic produce is often found to have ahigher content of vitamin C and dry matter while nitrate levelsare usually lower (D’Evoli et al., 2013; Leclerc, Miller, Joliet, & Roc-quelin, 1991). Minerals are often more concentrated in organicproduce (Smith, 1993), while protein content is often lower, butof superior quality (Magkos, Arvaniti, & Zampelas, 2003; Worthing-ton, 2001). Better flavour is sometimes found in organic than inconventional foods, but in other studies conventional productsare preferred by the tasters (Bourne & Prescott, 2002). This is due

to the fact that the flavour and the related content in minor com-pounds depend on many genetic and environmental factors (Ase-njo, 1962; Hornick, 1992) and it is extremely difficult to separatethe effects of the agronomic practices alone. In many other studies,no differences were found in the quality of organic vs. conventionalfruits (e.g. Bourne & Prescott, 2002; Magkos et al., 2003).

The role of the agronomic practices involved in the organic vs.conventional management (i.e. fertilisation, soil management, dis-ease and pest control) in olive oil quality is controversial. Organicolive oils may potentially differ in quality from conventional ones,but studies on this subject are extremely scarce in the literature.Data from Gutièrrez, Arnaud, Miguel, and Albi (1999) support thehypothesis that organic olive oils have better intrinsic qualitiesthan conventional oils: these authors reported lower acidity andperoxide index, higher rancimat induction time and concentrationof tocopherols, polyphenols, o-diphenols and oleic acid. This workwas carried out during 1 year and with one olive cultivar and theresults may not be widely applicable. In a 3-year study on the qual-ity of extra virgin olive oils from organic vs. conventional oils (cvsLeccino and Frantoio), many nutritional and merceological param-eters (i.e. acidity, UV-extinction coefficients, peroxide index,

A. Rosati et al. / Food Chemistry 159 (2014) 236–243 237

phenols, o-diphenols, tocopherols, antioxidant capacity, volatilecompounds, and sensory analysis) differed occasionally within ayear, but no consistent trends across the 3 years were found(Ninfali et al., 2008). The authors concluded that genotype andyear-to-year variability in weather conditions, have moremarked effects than agronomic practices.

The lack of quality differences between organic and conven-tional produce may be explained in many ways. One possibilityis that the definition of organic and conventional practices is toobroad, including very different techniques under the same system.Organic systems, for instance, may include water soluble fertilizersthat act in ways more similar to chemical fertilizers, rapidly releas-ing nutrients in proportions different than from the mineralisationof compost or humus, while conventional systems might includesoils rich in organic matter. Another problem is that the two sys-tems are often compared with farms/fields in different locations,with differences in environmental parameters (i.e. soil, microcli-mate, management) that might override the effect of the agro-nomic systems. Furthermore, the parameters measured are oftenlimited to few chemical compounds (e.g. vitamin C) or broadgroups of compounds (i.e. proteins, carbohydrates, etc.), while dif-ferences in quality might lie in subtle changes in many, thoughperhaps minor, compounds, or their proportions. Finally, in thecase of olive oil, it must be considered that the extracted oil ismostly the fatty part of the fruit, while in the fruit, fat representsabout half of the dry weight. Therefore, large changes in the non-fat composition of the fruit might be difficult or impossible to de-tect in the extracted oil. To the best of our knowledge, no data existon the effect of agronomic practices (i.e. organic vs. conventional)on the olive fruit composition.

In the present work we attempted to overcome, at least in part,the above limitations, by reducing the number of variables, in or-der to investigate possible differences in olive fruit and oil qualityfrom organic vs. conventional systems. To achieve this, we com-pared fruits and oil samples from trees grown in the same fieldand with the same agronomic practices, except for chemical fertil-isation and soil management, and in the same year. The metabolicprofile of fresh olives was studied by means of HRMAS-NMR spec-troscopy. The latter is a relatively novel tool in food science, whichcombines the advantages of HR-NMR and solid-state NMR. Semi-solid samples, as suspended food fragments, can be easily mea-sured without any chemical and/or physical treatment of the sam-ple, thus preventing any chemical modification of the composition.While HRMAS-NMR is largely used in pharmacology and medicine(Martinez-Bisbal, Esteve, Martinez-Granados, & Celda, 2011), itsuse in food science is still in its infancy (Ritota, Marini, Sequi, &Valentini, 2010; Valentini, Ritota, Cafiero, Cozzolino, Leita & Sequi,2011).

The objectives of this work were to assess possible differencesbetween the two agronomic managements in carpology, oil com-position and sensory properties, and fruit metabolomics.

2. Materials and methods

2.1. Plant material

The study was carried out in an organic olive orchard planted in1992 with a spacing of 6 � 5 m, located in central Italy at 500 mabove the sea level. Trees of Leccino and Frantoio cultivars werealternated along the rows. The orchard received 30 t ha�1 of sheepmanure at planting and no subsequent fertilisation. The soil wasmanaged with green mulch with naturally occurring vegetation,mowed twice a year in May and June, leaving the mowed vegeta-tion in situ as a form of organic fertilisation. Given the relativelyhigh altitude and longitude, there was no need to spray for the

olive fly, and no other chemical treatments were applied for dis-ease and pest control.

In 2009, three random trees of each cultivar were selected andfertilised with the equivalent of 180 kg of urea-N (based on 333trees per hectare), split in three equal doses in April, before thenew flash of vegetative growth, in June, during fruit set and in Sep-tember. In 2010, the same trees were fertilised with the equivalentof 150 kg of N per hectare, in three equal doses in June, July andSeptember, using urea in the first application and ammonium ni-trate in the second and third. In the first date of fertilizer applica-tion of each year, K2O fertilizer (potassium sulphate) was alsoapplied for an equivalent amount of 100 kg of K2O per hectare.

Immediately after the application of fertilizers, the soil wastilled under each tree and up to the neighbouring trees.

Fruit samples were harvested on 15 November 2010, from eachof the three trees per cultivar which received the chemical fertilisa-tion (i.e. conventional treatment) and from three more trees percultivar (i.e. organic treatment), randomly chosen among thosenot bordering trees receiving chemical fertilisation (to avoid bor-der effects). Samples of 50 fruits per tree (randomly chosen fromthe outside of the canopy at about 2 m of height) were used tomeasure fresh fruit, and fresh and dry pit weights. In addition, be-fore de-pitting the fruits, the same sample was used to assess thefruit maturity index (Uceda & Hermoso, 1998), and pulp firmness(g), using a penetrometer with a point of 1 mm in diameter. A sec-ond sample of 50 fruits per tree was used to measure fruit moisturecontent (i.e. fresh weight minus dry weight after drying in an ovenat 70 �C until constant weight), fruit dry weight and fruit oil con-tent, using nuclear magnetic resonance (the MINISPEC mq 20NMR Analyzer, Bruker Corporation, Billerica, Massachusetts, USA).

A third sample of fruits was taken for the metabolomic analysisdescribed in Session 2.3.

2.2. Oil extraction and analysis

On the same date that the other samples were taken (15November 2010) an additional sample of 3 kg of fruits per treewas harvested (adopting the same criterion as for the 50 fruit sam-ples) and the oil was immediately extracted using a small experi-mental mill. The olives were crushed and malaxated for 40 minat room temperature. The paste was then pressed for 4 min in alaboratory press and the liquid obtained was then centrifuged for5–7 min. The non-filtered oils were kept in the dark at room tem-perature until the chemical and sensory analyses were performed.

Analysis for chemical attributes included acidity, peroxide in-dex, spectrophotometric constants and acidic composition, andwere carried out according to the official methods defined byReg. EC 796/02. The total polyphenols content was determinedby colorimetry using the Folin Ciocalteau test, calibrated with gal-lic acid. A sensory analysis was carried out by the taste panel of‘‘Assam-Marche’’ (Ancona, Italy), to verify the absence of defectsand to define the sensorial profile, using the form specified inAppendix XII of EEC Reg. 2568/91, expressing intensity on a nu-meric scale of 0–5. The global score was instead expressed on scaleof 1–10.

2.3. NMR analysis and measurements

For the metabolomic analysis either 6 or 8 fruits per repetitiontree were analysed, with a total of 78 one-fruit samples; 36 fromFrantoio, half conventional and half organic, and 42 from Leccino,24 conventional and 18 organic. Samples were collected togetherwith all other samples, transported at +4 �C to the laboratory andstored in a refrigerator at �80 �C. 1H- and 13C-HRMAS-NMR spectraof both varieties were recorded by sampling the central part of thepulp directly with a spatula. In order to increase the signal-to-noise

238 A. Rosati et al. / Food Chemistry 159 (2014) 236–243

ratio, by reducing the intense water residual signal at ca. 4.7 ppm,2D-NMR spectra were acquired on freeze dried samples. To ensurethat the freeze-dried operation did not alter the metabolic profilewe compared the 1H-HRMAS-NMR spectra of freeze-dried sampleswith those obtained directly taking a piece of fresh olive and nodifference were found in the statistical results (data not shown)in agreement with what previously found for other vegetables (Ri-tota et al., 2010). We also measured the same olive three times bysampling in three different zones, and the multivariate model wasunable to discriminate between them (data not shown).

Samples were prepared by inserting about 5 mg of olive (freshfor 1D spectra and freeze-dried for multidimensional) in a10 mm HRMAS rotor with a 50 ll spherical insert. Ca. 40 lL ofD2O phosphate buffer, 0.01 M concentration and pH value equalto 7.2, with 0.01% TSP, i.e. 3-(trimethylsilyl)-propionic-2,2,3,3-d4

acid sodium salt, were then added. HRMAS-NMR spectra were re-corded at 298 K with a Bruker AVANCE spectrometer operating at a1H frequency of 400.13 MHz, equipped with a 4 mm HRMAS dualchannel probehead and spinning the samples at 7 kHz. 1H-NMRspectra were referenced to the methyl groups signal at d0.00 ppm of TSP and 13C-NMR spectra were referenced to the TSPd 0.00 ppm.

1H-HRMAS-NMR spectra were acquired by using a water sup-pression pulse sequence, noesypr1D (Bruker library), using 32 Kdata points over a 4807 Hz spectral width and adding 128 tran-sients. A recycle delay of 2 s and a delay for allowing efficientNOE effect equal to 95 ms were used, the 90� pulse length was5.2 ls, saturation of water residual signal was achieved by irradiat-ing during recycle delay at d equal to 4.70 ppm. Each spectrum wasFT transformed with 64 K data points and a line broadening factorequal to 0.3 Hz, i.e. exponential function, was applied to the FID.Spectra were manually corrected for phase and baseline distortionswith ACDlab 8.0 software (Advanced Chemistry Development, Inc.,Canada).

13C-HRMAS-NMR spectra were acquired with the power-gateddecoupling sequence, zgpg30 (Bruker library), using a 30� flip anglepulse of 5.0 ls. Experiments were carried out using 64 K datapoints over a 22,123 Hz (�220 ppm) spectral width by adding64 K transients with a recycle delay of 3 s. Each spectrum was FTtransformed with 128 K data points and manually phased andbase-lined, and a line broadening factor of 0.5 Hz was applied tothe FID prior FT.

The 1H–1H TOCSY experiment was acquired in the TPPI phase-sensitive mode, with a 4807 Hz spectral width in both dimensions,100 ms of spin-lock time, 1 K data points in f2 and 256 incrementsin f1, each with 64 scans. The 1H–13C HSQC spectra were acquiredin TPPI phase-sensitive mode, with a 4807 Hz spectral width in f2

dimension and a 15,083 Hz spectral width in f1. 2 K data pointsin f2 and 256 increments in f1, each with 64 scans, were used.

1H–13C-HMQC spectra were acquired in TPPI phase-sensitivemode, with a 4807 Hz spectral width in f2 dimension and a15083 Hz spectral width in f1. 1 K data points in f2 and 256 incre-ments in f1, each with 32 scans, were used. HMQC was preferred tothe HSQC since the latter is more sensible to the optimisation ofthe acquisition parameters.

2.4. NMR data reduction and preprocessing and multivariate dataanalysis

All 1H HRMAS-NMR spectra were manually phased, baselinecorrected and aligned by ACDlab 8.0 software (Advanced Chemis-try Development, Inc., Canada). Spectra alignment was performedreferring to TSP signal. Water residue signal and random noise re-gions were removed from the 1H spectra applying the dark regionmethod by ACDlab 8.0 software. To prepare NMR data for multivar-iate modelling the collected spectra were reduced into spectral

bins of 0.04 ppm widths (thus obtaining 197 buckets) by usingthe ACD bucketing method within ACDlab 8.0 software. The areaunder each bin was integrated and normalised with respect tothe sum of all integrals, which was set equal to 100. The resultingdata matrix, consisting of rows that reflect observations/samplesand columns that represent variables, was used as input variablesfor the statistical analysis. The reduced and normalised NMR spec-tral data were than imported into MATLAB (version 7.1, The Math-works, Natick, USA), mean centred and unit variance scaled, andanalysed using in-house routines and PLS_Toolbox (version 4.2,Eigenvector Research, 2007) by MATLAB software.

NMR data were analysed by means of the Partial Least Squaresprojections to latent structures-Discriminant Analysis, PLS-DA. Thedetermination of the optimal number of Latent Variables, i.e. LVs,was obtained by evaluating the Variance Captured and Statistics,in particular considering parameters such as the REsidual Sum ofSquares (RESS2), the PREdicted Sum of Squares (PRESS), Q2 andcumulative Sum of Squares X and Y (SSQX and SSQY, respectively).PRESS is a measurement of the predictive ability of the model,while SSQ is related to the model’s fitting goodness. Q2 representsthe default parameter used in PLS-DA discriminations and focuseson how well the class label can be predicted from new data (Cuny,Vigneau, Le Gall, Colquhoun, Lees, & Rutledge, 2008). By using theSSQ and PRESS parameters and the Q2 criterion as a guideline(Q2 > 0.05) it is possible to evaluate the optimal number of compo-nents for a model with a good fitting, an high predictive ability andwithout the over-fitting.

The model obtained was found to be based on four LVs and wasvalidated by the Venetian blinds cross-validation method.

3. Results

3.1. Carpology

Fruit characteristics within cultivars were not significantly dif-ferent among treatments (Table 1). Organic and conventional fruitshad similar fresh and dry fruit, pulp and pit weights (and conse-quently similar water content, data not shown); similar oil contenton both fresh and dry matter basis. Fruit maturation was also sim-ilar and not statistically different across treatments in Leccino,while in Frantoio the organic fruits had significantly but slightlyhigher coloration, and lower firmness, suggesting that the organicFrantoio fruits were slightly more ripe at the same sampling date.Even though we did not perform statistical tests for the cultivardifferences, since it was not the scope of the work, it may be ob-served that cultivar differences were always much larger thantreatment differences within cultivars.

3.2. Oil chemical composition and sensory analysis

No differences among treatments were found within each culti-var in acidity, peroxide index and spectrophotometric constants(DK, K232, K270), while polyphenols content was significantlyand strongly reduced in the conventional treatment in both culti-vars (Table 2). The acidic composition was similar among treat-ments (within cultivars) with the exception of a significant butslight reduction in palmitoleic, stearic (significant only in Leccino)and linoleic (significant only in Frantoio) acids, and a slight in-crease in heptadecenoic and linolenic (significant only in Frantoio)acids, in the conventional treatment (Table 3). The most importantand abundant fatty acid, oleic acid, did not differ significantly withthe agronomic treatment.

The changes in polyphenolic content was associated with thechange in the sensory evaluation, with the organic oil havingslightly but significantly higher bitterness, pungency and lower

Table 1Fruit characteristics of Frantoio and Leccino olives under organic and conventional cultivation.

Frantoio Leccino

Conventional Organic Conventional Organic

Average SE Average SE Average SE Average SE

Fruit fresh weight (g) 2.00 0.06 1.94 0.03 ns 1.78 0.70 1.68 0.05 nsFruit dry weight (g) 0.99 0.03 0.99 0.01 ns 0.82 0.01 0.84 0.02 nsPit fresh weight (g) 0.50 0.02 0.50 0.01 ns 0.41 0.02 0.44 0.02 nsPit dry weight (g) 0.41 0.02 0.41 0.01 ns 0.34 0.02 0.36 0.01 nsPulp fresh weight (g) 1.50 0.04 1.44 0.03 ns 1.37 0.68 1.24 0.04 nsPulp dry weight (g) 0.59 0.02 0.58 0.02 ns 0.48 0.01 0.48 0.02 nsOil in fruit (% fresh matter) 34.8 1.5 34.1 1.8 ns 31.9 1.2 31.0 0.9 nsOil in fruit (% dry matter) 17.3 0.7 17.3 0.9 ns 14.7 0.7 15.5 0.6 nsMaturation index (n) 0.95 0.10 1.65 0.39 ⁄ 2.79 0.09 2.89 0.02 nsFirmness (g) 262 3 234 7 ⁄⁄ 228 7 214 8 ns

ns = not significant; ⁄ and ⁄⁄ = significant at P < 0.05 and 0.01, respectively.

Table 2Chemical parameters of oils from Frantoio and Leccino under organic and conventional cultivation.

Acidity (% oleic) SE DK SE K232 SE K270 SE Peroxide index SE Polyphenolsmg kg�1

SE

Frantoio org 0.19 0.04 �0.0010 0.0000 1.79 0.03 0.14 0.01 13.9 5.23 528 3Frantoio conv 0.21 0.05 �0.0008 0.0004 1.09 0.81 0.12 0.01 11.1 0.00 358 3Significance ns ns ns ns ns ⁄⁄⁄

Leccino org 0.15 0.02 �0.0005 0.0000 1.52 0.09 0.12 0.01 11.4 5.94 386 25Leccino conv 0.16 0.00 �0.0003 0.0004 1.62 0.08 0.10 0.01 12.1 3.18 227 57Significance ns ns ns ns ns ⁄⁄⁄

ns = not significant; ⁄,⁄⁄ and ⁄⁄⁄ = significant at P < 0.05, 0.01 and 0.001, respectively.

A. Rosati et al. / Food Chemistry 159 (2014) 236–243 239

green fruit sensation, and, conversely, slightly lower sweetnessthan the conventional oil, in Frantoio (Table 4). Leccino oil had sim-ilar trends (except for green fruit sensation which was equal be-tween treatments), but the differences where statisticallysignificant only for bitterness. The overall score of the oil wasnearly identical between treatments.

3.3. Multivariate NMR data analysis

The 1H-HR-NMR spectrum of olive oil has been already assignedin details (Mannina et al., 2003). Even though olive oil and fresh ol-ive fruits are expected to have a similar composition, and thus 1H-HR-NMR and 1H-HRMAS-NMR signals should display almost thesame chemical shifts, we acquired 1D and 2D HRMAS-NMR exper-iments to ensure the correct assignment. We found only small dif-ferences, most likely due to the barely measurable modifications ofacquisition condition parameters, such as medium viscosity, mag-netic susceptibility, pH, etc. Since differences were very small, wereport below only metabolites relevant for the multivariate analy-ses and not the full assignment. Readers interested in the patternrecognition procedure may refer to the existing literature (Mannin-a et al., 2003).

The metabolic profile determined by means of HRMAS-NMRwas used for multivariate statistical analysis in order to buildPLS-DA models capable of classifying samples according to thefarming technique employed (details on the PREdicted Sum ofSquares (PRESS), Q2 and cumulative Sum of Squares X and Y, SSQXand SSQY, respectively, for the 3 LVs PLS-DA models for cv. Leccinoand Frantoio are shown in Table S1). The first LV explained 21% and13% of the captured variance for cv. Frantoio and Leccino, respec-tively, providing a reliable discrimination between the farmingtechniques, with organic olives mostly lying on the negative partof LV1 axis (Fig. 1). To further investigate the reliability of thePLS-DA models obtained, we also performed an OPLS-DA analysis

(Figure S1 in the Supplementary materials). Both orthogonal mod-els are obtained with only 2 Latent Variables, but no gain wasfound in sample separation.

To assess which metabolites were mostly responsible for thisdiscrimination, both the VIP scores and the regression coefficientsof the PLS-DA model were inspected. In particular, the VariableImportance in Projection (VIP) score of a predictor is a value thatexpresses the contribution of the individual variable in the defini-tion of the F-latent vector model. The VIP scores of the model arereported in Figure S2 in Supplementary material, while Table 5 re-ports the 20 most relevant intervals, with the associated metabo-lites, in discriminating buckets. Table 5 reports also theimportant weight factors (W) of each bucket, which allow to assesswhich treatment (i.e. organic vs. conventional) contains largeramounts of the associated metabolite. In our case, all metaboliteswith a positive W value are more abundant in conventional fruits,while those with negative values are more abundant in organicsamples. Based in these values, conventional samples of Leccinohad larger amounts of glycocholate, fatty acids, NADPH, some ami-no acids (i.e. phenylalanine, asparagine and GABA), thymidine,trigonelline and nicotinic acid, while organic samples had highercontents of 5,6-dihydrouracil and NADP+, which is the oxidisedform of nicotinamide-adenine-dinucleotide-phosphate (i.e.NADPH). Similarly, organic Frantoio had larger amounts of NADP+,polyunsaturated fatty acids (i.e. those with the ACH@CHACH2-

ACH@CHA sequence, like linoleic acid), some amino acids (i.e.tryptophan, tyrosine and phenylalanine), hesanal and some unas-signed species. Conventional Frantoio olives had larger quantitiesof cis-olefin, b-D-glucose and propanal.

4. Discussion

Aside from differences in the metabolic profile of the fruits, themain effect of the different agronomic practices was to increase

Tabl

e3

Aci

dic

com

posi

tion

ofoi

lsfr

omFr

anto

ioan

dLe

ccin

oun

der

orga

nic

and

conv

enti

onal

cult

ivat

ion.

Palm

itic

(%)

SEPa

lmit

olei

c(%

)SE

Hep

tade

can

oic

(%)

SEH

epta

dece

noi

c(%

)SE

Stea

ric

(%)

SEO

leic

(%)

SELi

nol

eic

(%)

SELi

nol

enic

(%)

SEEi

cosa

noi

c(%

)SE

Eico

sen

oic

(%)

SE

Fran

toio

org

10.9

0.16

0.64

0.01

0.04

80.

003

0.10

60.

011.

870.

2179

.80.

525.

390.

430.

810.

050.

350.

014

0.15

0.03

Fran

toio

con

v10

.10.

710.

570.

010.

052

0.00

10.

122

0.00

1.63

0.08

81.1

0.64

4.85

0.00

0.95

0.00

0.37

0.00

70.

140.

01

Sign

ifica

nce

ns

⁄⁄n

s⁄⁄⁄

ns

ns

⁄⁄

ns

ns

Lecc

ino

org

12.6

0.13

0.71

0.00

0.05

10.

001

0.10

40.

011.

830.

1178

.90.

064.

400.

040.

850.

090.

340.

007

0.12

0.02

Lecc

ino

con

v12

.80.

100.

610.

080.

053

0.00

10.

123

0.00

1.56

0.08

78.9

0.44

4.58

0.32

0.90

0.11

0.32

0.03

50.

110.

02

Sign

ifica

nce

ns

⁄n

s⁄⁄⁄

ns

ns

ns

ns

ns

ns

=n

otsi

gnifi

can

t;⁄,⁄⁄

and⁄⁄⁄

=si

gnifi

can

tat

P<

0.05

,0.0

1an

d0.

001,

resp

ecti

vely

.

240 A. Rosati et al. / Food Chemistry 159 (2014) 236–243

polyphenols content in the organic oil (Table 2), which reflected inthe slight differences in some sensory parameters, especially bit-terness (Table 4), while other fruit characteristics were not af-fected, except for a slightly more advanced maturation in organicFrantoio (Table 1). The differences in polyphenols content wereprobably related to greater soil nitrogen availability in the conven-tional treatment. It is well known that increasing N fertilisation de-creases polyphenols content in olive fruits and oil (Fernández-Escobar, Beltrán, Sánchez-Zamora, García-Novelo, Aguilera, & Uce-da, 2006; Tognetti et al., 2008). Urbanczyk-Wochniak and Fernie(2005) found that nitrate assimilation strongly modified the aminoacid metabolism of the plant with consequent changes in the levelof the most representative compounds of the secondary metabo-lism. In fact, this is a general phenomenon as discussed by Brandt,Leifert, Sanderson, and Seal (2011) who carried out a meta-analysisof differences in contents of secondary metabolites in fruits andvegetable grown with organic vs. conventional practices, and con-cluded that, due to higher N fertilisation, conventional treatmentsreduce the accumulation of defence-related secondary metabo-lites, particularly polyphenols, which are among the compoundsthat provide the most health benefits to consumers. Higher con-tents of phenolic compounds in organic food were found also inother reviews of the published literature (e.g. Lima & Vianello,2011).

The greater N availability in our conventional treatment deriveddirectly from the N fertilisation and, probably, also from soil tilling,which likely promoted mineralisation from the soil organic matter,compared to the untilled control, where no nitrogen fixing legumi-nous weeds were present, most weed represented by grasses andplants of the Asteraceae family (data not shown). Furthermore,the soil tilling in our conventional system reduced weed competi-tion and, although the weeds were mowed multiple times per year,they probably competed more with the olive trees, for nutrients, inthe organic system. When commenting the greater quality of oliveoil with organic farming found by Gutièrrez et al. (1999), Ninfaliet al. (2008) concluded that the greater levels of mineral N fertilisa-tion were probably at the base of some of the results, including thelower polyphenols content in the conventional treatment. There-fore, nutrient, particularly N, availability was probably at the baseof the observed differences. This is further supported by consider-ing another analogy between the N fertilisation experiment ofFernández-Escobar et al. (2006) and our results: increasing N fertil-isation in the former experiment not only decreased polyphenolscontent, as in the present work, but also effected the fatty acidcomposition in a similar way, with significant but slight variationin the same few fatty acids (increasing heptadecenoic and linolenicacids and decreasing linoleic and stearic acids, with increasing N).Of such significant variations, the most quantitatively important inthe present study was the decrease in linoleic acid (i.e. higher inthe organic treatment), which was significant only in Frantoio,increasing by about a half point percent over an average contentof 5% (i.e. 10% variation, Table 3), again in agreement with datafrom Fernández-Escobar et al. (2006). The other fatty acids whichchanged significantly, represent such small fractions of the totalcontent that their variations, though significant, implies very min-or changes in the oil composition. This agrees well with the meta-bolomic analysis (Table 5) showing significant increases inpolyunsaturated fatty acids with the ACH@CHACH2ACH@CHA se-quence in organic Frantoio, where linoleic acid (i.e. a polyunsatu-rated acid with that sequence) increased in the organictreatment, representing, as mentioned above, the greatest quanti-tative variation (Table 3).

In our experiment, in addition to nitrogen, potassium was alsosupplied to the conventional treatment. The few works publishedon the effects of potassium applications suggests that fatty acidcomposition and polyphenol concentration in the olive oil is not af-

Table 4Sensory evaluation of oils from Frantoio and Leccino under organic and conventional cultivation.

Green fruit SE Green grass/leaf SE Bitterness SE Pungency SE Sweetness SE Almond SE Herbs SE Overall score SE

Frantoio org 2.3 0.1 1.0 0.0 2.5 0.0 2.0 0.0 1.8 0.1 2.0 0.0 0.5 0.5 7.2 0.1Frantoio conv 2.5 0.0 1.0 0.0 1.8 0.4 1.5 0.3 2.3 0.4 2.0 0.0 0.0 0.0 7.2 0.1Significance ns ns * * * ns ns nsLeccino org 2.0 0.6 1.0 0.6 1.8 0.1 1.5 0.3 2.8 0.4 1.5 0.3 0.5 0.3 7.2 0.3Leccino conv 2.0 0.6 1.0 0.6 1.3 0.1 1.3 0.1 3.3 0.1 1.5 0.3 0.5 0.3 7.0 0.3Significance ns ns * ns ns ns ns ns

ns = not significant.* Significant at P < 0.05.

A. Rosati et al. / Food Chemistry 159 (2014) 236–243 241

fected by K levels (Simões, Pinheiro-Alves, Cordeiro, & Marcelo,2002). It may therefore be concluded that the differences in oiland fruit quality here reported are more likely due to N availabilitythan to K.

The differences in the polyphenol concentration in the oil wereprobably not related to fruit ripeness, since both maturationparameters (maturity index and pulp firmness) were either not dif-ferent (Leccino), or fruits were slightly more ripe (i.e. more col-oured and less firm) in the organic treatment (Frantoio). Sincepolyphenol concentration naturally decreases with increasing ripe-

Fig. 1. PLS-DA score plots of fresh olives Leccino (top panel) and Frantoio (bottom)grown under conventional (black circles) and organic (black triangles) system.

ness (Brenes, García, García, Rios, & Garrido, 1999; Salvador, Aran-da, Gómez-Alonso, & Fregapane, 2001), this would have led to adecrease in polyphenols, while the opposite was true, suggestingthat changes in polyphenol concentration were directly related tothe agronomical practices, possibly via changes in nutrientavailability.

Polyphenols are among the minor components of food with po-tential beneficial effect on human health and olive oil is consideredone of the main food sources of polyphenols with beneficial effectson human health (Covas, 2008; Fabiani et al., 2006). However, thevariation in phenolic concentration with conventional vs. organic

Table 5Metabolites relevant for discrimination between organic and conventional olives forcv. Leccino and cv. Frantoio, top and bottom, respectively. For simplicity we report thefirst 20 signals.

ppm VIP Metabolite W

cv. Leccino9.25 4.1778 NADP+ �0.02610.72 3.989 Glycocholate 0.02159.57 3.5867 5,6-Dihydrouracil �0.07511.12 3.3408 CH3 of fatty acid 0.02470.76 3.2866 Glycocholate 0.00266.59 2.8088 NADPH 0.12299.13 2.7937 Trigonelline 0.04451.16 2.7691 CH3 of fatty acid 0.01659.53 2.6651 5,6-Dihydrouracil 0.02259.21 2.5826 NADP+ �0.03837.39 2.4077 Phenylalanine 0.08566.63 2.3909 NADPH 0.13297.43 2.387 Phenylalanine 0.10392.96 2.3018 Asparagine 0.1126.55 2.2449 NADPH 0.11963.00 2.2006 GABA 0.10421.08 2.1852 CH3 of fatty acid 0.01899.01 2.1703 Nicotinic acid 0.04536.27 2.1323 Thymidine 0.12958.85 2.0988 Trigonelline 0.0914

cv. Frantoio2.80 3.116 ACH@CHACH2ACH@CHA �0.12446.75 2.7452 Unknown �0.16917.27 2.6313 Trp, Phe �0.12254.44 2.4192 cis-Olefin 0.12916.87 2.3333 Tyrosine �0.15047.23 2.3001 Tryptophan �0.12866.67 2.2814 NADP+ �0.1386.91 2.219 CH3 hesanal �0.14485.95 2.2102 Unknown �0.12656.11 2.1255 NADP+ �0.12836.07 2.1092 NADP+ �0.13055.71 2.0664 Unknown �0.11847.19 2.0621 Tyr, Trp �0.12736.03 2.0387 NADP+ �0.12595.99 2.028 NADP+ �0.12237.55 1.9805 Tryptophan �0.12616.71 1.9731 Unknown �0.13565.75 1.9477 Unknown �0.10643.20 1.9475 b-D-Glu 0.09741.04 1.9439 CH3 propanal 0.1290

242 A. Rosati et al. / Food Chemistry 159 (2014) 236–243

farming has not been studied much. The few existing studies re-port contrasting results (Gutièrrez et al., 1999; Ninfali et al.,2008). The present results support the hypothesis that agronomicpractices may effect polyphenol concentration in olive fruit andoil, at least via different nutrient (i.e. nitrogen) availability whenthis differs substancially.

Other factors affecting fruit ripeness and polyphenols concen-trations are fruit load and light availability. Reducing fruit load inolive leads to earlier coloration and higher polyphenols content(Barone, Gullo, Zappia, & Inglese, 1994). In our study fruit loadwas not statistically different between organic and conventionalLeccino with an average of about 8 kg of fruits per plant (datanot shown), therefore the difference in polyphenols content werenot related to fruit load. However, in Frantoio, the organic plantshad a significantly reduced fruit load (3.5 vs. 7 kg per plant), possi-bly justifying, at least in part, the higher polyphenols content.However, reducing fruit load artificially increases resource avail-ability per fruit, while when the tree naturally adjusts its produc-tivity to varying resource availabilities, the effects might differ,making it difficult to conclude whether natural variation in fruitsload has the same effect of artificial fruit thinning. Light availabilityalso affects polyphenols content which increases at increasing light(Proietti, Nasini, Famiani, Guelfi, & Standardi, 2012). Although wehave not measured light availability, we have no reason to believethere were any important differences because the field practiceswere differentiated for only 2 years, not enough time to affect treesize of adult trees, and fruit samples were collected on the outsideof the canopy. Additionally, the light environment is also affectedby nearby trees and the different treatments were applied in indi-vidual, randomly chosen, trees, therefore any selected tree wassurrounded by a similar mixture of organic and conventional trees.

The effect of the agronomic practices was very evident in themetabolic profile of the fruits, as highlighted by the PLS-DA modelof the 1H-HRMAS-NMR data (Fig. 1). These effects were large andcomparable to differences among cultivars, supporting the ideathat it is more difficult to find differences in the oil, which is mostlyfat, compared to the fruit, as discussed in the introduction. Signif-icant variation in fruit metabolomics occurred for many com-pounds that are absent (or present in very low amounts) in theoil, such as amino acids and few more species listed in Table 5. Thisimplies that the lack of difference in oil parameters between differ-ent agronomic treatments, does not necessarily rule out possibledifferences in the fruit.

In both cultivars, the conventional treatment had higher con-tents of NADPH and lower NADP+. This two species are involvedin many processes, including nitrate reduction and the Calvin cy-cle. Given that plants pick up nitrogen mostly as nitrate, whichneeds to be promptly reduced (via nitrate reductase), and thatnitrogen availability affects photosynthetic rates (Field & Mooney,1986), it seems likely that the different contents of NADPH/NADP+between treatments are consequent to the different nitrogen avail-ability in the two treatments. Understanding such biochemicalmechanisms, however, is probably not possible with our dataand is beyond the scope of this paper, but would be worth pursuingin future research.

Many studies found that organic management decreases theprotein content in the produce, but proteins are often of greaterquality (Magkos et al., 2003; Worthington, 2001). Our results,showing differential contents in some amino acids, support thehypothesis that protein composition might be affected by agro-nomic management.

In this work, we deliberately compared different agronomicalpractices using replication of individual trees, randomly chosenin the same field, and data were collected in the same year andwith the same timing. Therefore, environmental (i.e. soil, climate,etc.) effects on the results can probably be ruled out, and differ-

ences in the metabolic profile of the fruits and in oil parameterswere likely due to the agronomic practices alone. In many previousstudies, differences in nutrient content and sensory properties be-tween organic and conventional foods were not found or wereinconsistent over time, as reviewed by many authors (e.g. Bourne& Prescott, 2002; Magkos et al., 2003). This is probably becausethe nutrient and sensory qualities of foods depends on a varietyof factors, including cultivar, climate, soil type, nutrient and wateravailability, duration and conditions of storage (Asenjo, 1962; Hor-nick, 1992). This is true also in olive oil (D’Imperio, Dugo, Alfa,Mannina, & Segre, 2007; Ripa et al., 2008). Hence, the differencesrelated to the cultivation methods may be difficult to isolate. How-ever, lack of observable difference does not imply that organicallyproduced foods are in no way different from conventional ones.Our results suggest that when environmental variations can be ru-led out, agronomic practices can have clear effects on the plantmetabolism, especially if investigated in whole fruits (rather thanjust oil), and when evaluating a large number of specific metabo-lites rather than general groups of compounds. However, our datais limited to the specific field conditions, and the effects observedare probably related to some particular practices, such as mineralfertilisation, and may not be generalised to broader agronomic def-initions, such as organic or conventional, which may include verydifferent combination of different practices.

Feeding trials indicate that, when given a choice, animals tendto choose organically (or biodynamically) produced feeds over con-ventional ones, often receiving health benefits (Hodges & Scofield,1983; Velimirov, Huber, Lauridsen, Rembiałkowska, Seidel, &Bügel, 2010; Woese, Lange, Boess, & Bögl, 1997), especially with re-gard to the reproductive function. Some similar benefits have beenalso observed in humans (Larsen, Spano, Giwercman, & Bonde,1999), where consumption of organic foods with higher antioxi-dant activity results in higher total antioxidant capacity in the hu-man plasma (Di Renzo et al., 2007). Our results suggests thatparameters other than those usually measured might differ be-tween organic and conventional products and that it is less likelythat clear differences will be demonstrated using traditional ana-lytical approaches, since it is impossible to measure enough foodcomponents and their bioavailability (Bourne & Prescott, 2002).Brandt et al. (2011) also speculated that the production system islikely to affect unknown compounds that are responsible for somehealth benefits. Our results suggest that the metabolomic ap-proach, including the HRMAS-NMR, with its broad range of ana-lysed compounds, but also other technologies, might contributeto unravel the effects of agronomical practices, that might remainhidden when using different and simpler analyses.

5. Conclusions

Possible superior nutritional properties of organically producedfood represent only one of the consumers’ reasons for choosingsuch foods. There is enough evidence that these foods may containlower pesticide content (Bourne & Prescott, 2002; Reganold et al.,2001). These reasons alone are sufficient for some consumers toprefer organic food, irrespective of their nutritional properties (By-rne, Toensmeyer, German, & Muller, 1992). Nonetheless, healthbenefits are usually the first reason motivating consumer’s choice(Woodward & Meier-Ploeger, 1999). Our data suggests that, atleast under certain conditions, there might be differences in thecomposition of fruits grown with different agronomical practices,some of which readily measurable, like polyphenols content, andothers that might be difficult to decipher unless a broad-spectrumanalysis, such as HRMAS-NMR for instance, is used. However,whether these differences are relevant compared to differencesarising from cultivar, climate or other variables, remains an open

A. Rosati et al. / Food Chemistry 159 (2014) 236–243 243

question. Additionally, while for polyphenols and antioxidant com-pounds the beneficial effect on human health have been verified,further research will have to assess whether differences in othercompounds are important for the health of the consumers.

Acknowledgements

The study was funded by the Italian Ministry for Agricultural,Food and Forestry Politics (MiPAAF), BIODATI Project. We thankcolleagues of CRA-CIN, Osimo, for the use of laboratory equipment.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.foodchem.2014.03.014.

References

Asenjo, C. F. (1962). Variations in the nutritive values of foods. American JournalClinical Nutrition, 11, 368–376.

Barone, E., Gullo, G., Zappia, R., & Inglese, P. (1994). Effect of crop load on fruitripening and olive oil (Olea europea L.). Journal of Horticultural Science, 69,67–73.

Bourne, D., & Prescott, J. (2002). A comparison of the nutritional value, sensoryqualities, and food safety of organically and conventionally produced foods.Critical Reviews in Food Science and Nutrition, 42, 1–34.

Brandt, K., Leifert, C., Sanderson, R., & Seal, C. J. (2011). Agroecosystem managementand nutritional quality of plant foods: the case of organic fruits and vegetables.Critical Reviews in Plant Sciences, 30, 177–197.

Brenes, M., García, A., García, P., Rios, J. J., & Garrido, A. (1999). Phenolic compoundsin Spanish olive oil. Journal of Agricultural and Food Chemistry, 47, 3535–3540.

Byrne, P. J., Toensmeyer, U. C., German, C. L., & Muller, H. R. (1992). Evaluation ofconsumer attitudes towards organic produce in Delaware and the Delmarvaregion. Journal of Food Distribution Research, 23, 29–44.

Covas, M. I. (2008). Bioactive effects of olive oil phenolic compounds in humans:reduction of heart disease factors and oxidative damage.Inflammopharmacology, 16, 216–218.

Cuny, M., Vigneau, E., Le Gall, G., Colquhoun, I., Lees, M., & Rutledge, D. N. (2008).Fruit juice authentication by H-1 NMR spectroscopy in combination withdifferent chemometrics tools. Analytical and Bioanalytical Chemistry, 390,419–427.

D’Evoli, L., Moscatello, S., Baldicchi, A., Licarini, M., Cruz-Castillo, J. G., Aguzi, A., et al.(2013). Post-harvest quality, phytochemicals and antioxidant activity in organicand conventional kiwifruit (Actinidia deliciosa, CV Hayward). Italian Journal ofFood Science, 25(3), 362–368.

Di Renzo, L., Di Pierro, D., Bigioni, M., Sodi, V., Galvano, F., Cianci, R., et al. (2007). Isantioxidant plasma status in humans a consequence of the antioxidant foodcontent influence? European Review for Medical and Pharmacological Sciences, 11,185–192.

D’Imperio, M., Dugo, G., Alfa, M., Mannina, L., & Segre, A. L. (2007). Statisticalanalysis on Sicilian olive oils. Food Chemistry, 102, 956–965.

Fabiani, R., De Bartolomeo, A., Rosignoli, P., Servili, M., Selvaggini, R., Montedoro, G.F., et al. (2006). Virgin olive oil phenols inhibit proliferation of humanpromyelocytic leukemia cells (HL60) by inducing apoptosis anddifferentiation. Journal of Nutrition, 136, 614–619.

Fernández-Escobar, R., Beltrán, G., Sánchez-Zamora, M. A., García-Novelo, J.,Aguilera, M. P., & Uceda, M. (2006). Olive oil quality decreases with nitrogenover-fertilization. HortScience, 41, 215–219.

Field, C., & Mooney, H. A. (1986). The photosynthesis–nitrogen relationship in wildplants. In T. J. Givinsh (Ed.), On the economy of form and function (pp. 25–55).Cambridge: Cambridge University Press.

Gutièrrez, F., Arnaud, T., Miguel, A., & Albi, M. A. (1999). Influence of ecologicalcultivation on virgin olive oil quality. Journal of the American Oil Chemists’Society, 76, 617–621.

Hodges, R. D., & Scofield, A. M. (1983). Effect of agricultural practices on the healthof plants and animals produced: A review. In W. Lockeretz (Ed.),Environmentally Sound Agriculture (pp. 3–34). New York: Praeger Scientific.

Hornick, S. B. (1992). Factors affecting the nutritional quality of crops. AmericanJournal of Alternative Agriculture, 7, 63–68.

Larsen, S. B., Spano, M., Giwercman, A., & Bonde, J. P. (1999). Semen quality and sexhormones among organic and traditional Danish farmers. ASCLEPIOS studygroup. Occupational and Environmental Medicine, 56, 139–144.

Leclerc, J., Miller, M. L., Joliet, E., & Rocquelin, G. (1991). Vitamin and mineralcontents of carrot and celeriac grown under mineral or organic fertilization.Biological Agriculture And Horticulture, 7, 339–348.

Lima, G. P. P., & Vianello, F. (2011). Review on the main differences between organicand conventional plant-based foods. International Journal of Food Science &Technology, 46, 1–13.

Magkos, F., Arvaniti, F., & Zampelas, A. (2003). Organic food: Nutritious food forthought? A review of the evidence. International Journal of Food Sciences andNutrition, 54, 357–371.

Mannina, L., Dugo, G., Salvo, F., Cicero, L., Ansanelli, G., Calcagni, C., et al. (2003).Study of the cultivar-composition relationship in Sicilian olive oils by GC, NMR,and statistical methods. Journal of Agricultural and Food Chemistry, 51, 120–127.

Martinez-Bisbal, M. C., Esteve, V., Martinez-Granados, B., & Celda, B. (2011).Magnetic resonance microscopy contribution to interpret high-resolutionmagic angle spinning metabolomic data of human tumor tissue. Journal ofBiomedicine and Biotechnology, 2011 [doi: 763684].

Ninfali, P., Bacchiocca, M., Biagiotti, E., Esposto, S., Servili, M., Rosati, A., et al. (2008).A 3-year study on quality, nutritional and organoleptic evaluation of organicand conventional extra-virgin olive oils. Journal of the American Oil ChemistrySociety, 85, 151–158.

Proietti, P., Nasini, L., Famiani, F., Guelfi, P., & Standardi, A. (2012). Influence of lightavailability on fruit and oil characteristics in Olea europaea L.. Acta Horticulturae,949, 243–249.

Reganold, J. P., Glover, J. D., Andrews, P. K., & Hinman, H. R. (2001). Sustainability ofthree apple production systems. Nature, 410, 926–930.

Ripa, V., De Rose, F., Caravita, M. L., Parise, M. R., Perri, E., Rosati, A., et al. (2008).Qualitative evaluation of olive oils from new olive selections and effects ofgenotype and environment on oil quality. Advances in Horticultural Science, 22,95–103.

Ritota, M., Marini, F., Sequi, P., & Valentini, M. (2010). Metabolomic characterizationof Italian sweet pepper (Capsicum annum L.) by means of HRMAS-NMRspectroscopy and multivariate analysis. Journal of Agricultural and FoodChemistry, 58, 9675–9684.

Salvador, M. D., Aranda, F., Gómez-Alonso, S., & Fregapane, G. (2001). Cornicabravirgin olive oil: a study of five crops seasons. Composition, quality and oxidativestability. Food Chemistry, 74, 267–274.

Simões, P., Pinheiro-Alves, C., Cordeiro, A. M., & Marcelo, M. E. (2002). Effect of thenitrogen and potassium fertilization on fatty acids composition and oxidativestability for ‘Carrrasquenha’ cultivar olive oil at different harvest periods –Preliminary study. Acta Horticulturae, 586, 337–340.

Smith, B. (1993). Organic foods versus supermarket food: element levels. Journal ofApplied Nutrition, 45, 35–39.

Tognetti, R., Morales-Sillero, A., D’Andria, R., Fernandez, J. E., Lavini, A., Sebastiani, L.,et al. (2008). Deficit irrigation and fertigation practices in olive growing:convergences and divergences in two case studies. Plant Biosystems, 142,138–148.

Uceda, M., & Hermoso, M. (1998). La calidad del aceite de oliva. In D. Barranco, R.Fernandez-Escobar, & L. Rallo (Eds.), El Cultivo del Olivo (pp. 547–572). Madrid,Spain: Junta de Andalucia Ediciones Mundi-Prensa.

Urbanczyk-Wochniak, E., & Fernie, A. R. (2005). Metabolic profiling reveals alterednitrogen nutrient regimes have diverse effects on the metabolism ofhydroponically-grown tomato (Solanum lycopersicum) plants. Journal ofExperimental Botany, 56, 309–321.

Valentini, M., Ritota, M., Cafiero, C., Cozzolino, S., Leita, L., & Sequi, P. (2011). TheHRMAS-NMR tool in foodstuff characterisation. Magnetic Resonance inChemistry, 49, S121–S125.

Velimirov, A., Huber, M., Lauridsen, C., Rembiałkowska, E., Seidel, K., & Bügel, S. G.(2010). Feeding trials in organic food quality and health research. Journal of theScience of Food and Agriculture, 90, 175–182.

Woese, K., Lange, D., Boess, C., & Bögl, K. W. (1997). A comparison of organically andconventionally grown foods – results of a review of the relevant literature.Journal of the Science of Food and Agriculture, 74, 281–293.

Woodward, L., & Meier-Ploeger, A. (1999). Raindrops on roses and whiskers onkittens: consumers’ perceptions of organic food quality. Proceedings of the 12thInternational IFOAM Scientific Conference (pp. 81–88), Argentina.

Worthington, V. (2001). Nutritional quality of organic versus conventional fruits,vegetables, and grains. Journal of Alternative and Complementary Medicine, 7,161–173.