PAPER 236 Ital. J. Food Sci., vol. 27 - 2015 - Keywords: moulds, sensory defects, virgin olive oil, volatile compounds, yeasts - INVESTIGATION ON MICROBIOLOGY OF OLIVE OIL EXTRACTION PROCESS S. GUERRINI, E. MARI, M. MIGLIORINI1, C. CHERUBINI1, S. TRAPANI, B. ZANONI*, M. VINCENZINI Department of Agricultural, Food and Forestry Management Systems - GESAAF - Food Science and Technology and Microbiology Section, University of Florence, Via Donizetti 6, 50144 Florence, Italy 1PromoFirenze - Laboratorio Chimico Merceologico, Special Agency of the Florence Chamber of Commerce, Via Orcagna 70, 50121 Florence, Italy *Corresponding author: bruno.zanoni@unifi.it ABSTRACT Several batches of approx. 200 kg olives from Frantoio and Moraiolo cultivars were processed in an oil mill at two dates of harvesting. Samples were collected in several steps of extraction pro- cess for sensory, chemical and microbial analyses. All extracted olive oil from the second olive harvesting date was affected by sensory defects and hence classified as being “non-extra virgin”. A distinction between extra virgin olive oil and non- extra virgin olive oil obtained from both harvesting dates was explained by the volatile compounds content of olive oil samples and by yeast and mould counts collected at different processing steps. mailto:bruno.zanoni@unifi.it Ital. J. Food Sci., vol. 27 - 2015 237 INTRODUCTION The absence of sensory defects is necessary for olive oil to be marketed as ‘‘extra virgin’’ in the EU. Extra virgin olive oil (EVOO) is character- ized by pleasant sensory notes. They are main- ly originated by aldehydes, esters, alcohols and ketones, which are responsible for oil sensory attributes such as “green” and “fruity” (APARI- CIO and MORALES, 1998; MORALES et al., 2005, BENDINI et al., 2012). However, several phenom- ena can alter the initial pleasant flavour, giving rise to unpleasant sensory notes. The current olive oil regulations (EU Reg. 1348/2013) classify the most frequent senso- ry defects into four groups as follows: “fusty”, “musty”, “winey–vinegary”, and “rancid”. Storage of olive fruits in piles before being pro- cessed is a cause of sensory alterations in EVOO. Olive transpiration during storage is known to increase pile temperature, enabling microbial cells to grow and to affect the chemical compo- sition of olives (MORALES et al., 2005). Both bio- genesis of volatile compounds and transforma- tion phenomena of phenolic compounds can be significantly influenced by microbial contamina- tion of olives. Effects of olive microbiota on oil characteristics are considered even greater than time-temperature conditions of malaxation (VI- CHI et al., 2011). Oil quality may be affected by microorgan- isms, according to their metabolic activities. During olive crushing, microorganisms might migrate into oil through both solid particles of ol- ive fruit and micro-drops of vegetation water (CI- AFARDINI and ZULLO, 2002). Some microorgan- isms do not survive a long time, but others may persist and become a typical microbiota of olive oil. For example, yeasts may remain metabolical- ly active during olive oil storage and thus mod- ify olive oil characteristics (ZULLO et al., 2010). Enzymatic activities of yeasts and moulds iso- lated from either olives or EVOO have been re- ported to include β-glucosidase, β-glucanase, polyphenoloxidases, peroxidase and, in some cases, lipase and cellulase activities (CIAFAR- DINI and ZULLO, 2002; CIAFARDINI et al., 2006; ZULLO and CIAFARDINI, 2008; ROMO-SANCHEZ et al., 2010). Enzymes such as β-glucosidase are known to improve oil quality by increasing phe- nolic compound extractability, while others such as lipase, polyphenoloxidases and peroxidase are known to cause detrimental effects (PALOMARES et al., 2003; ROMO-SANCHEZ et al., 2010; VICHI et al., 2011; MIGLIORINI et al., 2012). Penicilli- um and Fusarium spp. isolates have been shown to produce amounts of exogenous lipoxygenase (FAKAS et al., 2010) that, together with endog- enous lipoxygenase, is the key enzyme of LOX pathway (ANGEROSA et al., 2004). Extraction process control should include monitoring activities on microbial contamination in olives and EVOO, as associated with sensory and chemical analyses. The study of the micro- bial populations occurring at different steps of EVOO extraction process, as well as their role in affecting oil characteristics, appears to be in- creasingly useful. The aim of this work was to investigate both microbial ecology throughout olive oil process- ing and a possible relationship between EVOO volatile compound content and microbial con- tamination. MATERIALS AND METHODS Experimental design During 2011 crop season, several batches of approx. 200 kg olives from Frantoio and Moraio- lo cultivars were processed in an oil mill (Azien- da Agricola Buonamici, Fiesole, Florence, Italy). Plant for oil extraction (TEM, Florence, Ita- ly) consisted of a cleaning and water washing system, an olive grinding cutter crusher (mod. FR350), a controlled-temperature vertical axis malaxation equipment (500 kg capacity) (mod. V500), a “decanter” (two-step mod. D1500) with 1500 kg/h maximum capacity and a cardboard filter press (15 μm cut-off). Plastic residue or “alperujo” from decanter was subjected to sep- aration by centrifugation of stone fragments to obtain destoned pomace (Fig. 1). Olives were processed within 12 h from har- vest at two dates (HD): November 16, 2011 (HD1) and November 23, 2011 (HD2). Oil extraction tri- als were carried out in quadruple. Olives were crushed at 2,500 rpm (crusher holes 6.5 mm in diameter); malaxation was car- ried out at half capacity under vacuum (residual pressure of 20 kPa) at 22 ± 1°C for a mean time of 15 min to work under low oxidative stress im- pact conditions; decanter worked with a screw conveyor rotating at a slower speed than that of the bowl. Samples were collected in several steps of ex- traction process for sensory, chemical and mi- crobial analyses, as shown in Fig. 1. Chemical analyses Olives A homogeneous olive sample was crushed with a laboratory crusher, and resulting olive paste was used for chemical analyses. The water content (g kg−1 of dry matter) was measured on olive paste by gravimetric method (CHERUBINI et al., 2009). The total sugar content was determined by the UNI 22608 method, modified as described in a previous study (CHERUBINI et al., 2009). Re- sults for sugar content obtained from the equip- ment (Compact Titrator, Crison, Modena, Italy) were expressed as g kg−1 of dry matter. 238 Ital. J. Food Sci., vol. 27 - 2015 Fig. 1 - Overview of extraction process and analyses carried out. The total oil content was determined with hex- ane in an automatic Randall extractor (mod. 148, Velp Scientifica, Milan, Italy), following the ana- lytical technique described in a previous study (CHERUBINI et al., 2009). Results were expressed as g kg−1 of dry matter. The total phenolic compounds content was determined by weighing 4 g crushed olives and adding 80 mL Methanol:Water (60:40) solution; two series of stirring for 30 min and centrifuga- tion at 4,000 rpm for 15 min were performed, and the supernatant was collected. The pheno- lic extract was adjusted to volume of 200 mL by Methanol:Water (60:40) and placed in the freez- er for 2 h. After thawing, the phenolic extract was filtered. One mL filtered extract, 5 mL Folin Ciocalteu reagent, and 20 mL sodium carbon- ate were placed into a 100 mL flask and adjust- ed to volume with distilled water. Sixty minutes were waited for colour development; after one hour, UV reading (UV/VIS, Varian model Cary 1E, The Netherlands) was performed at 765 nm wavelength. The total phenolic compounds con- tent was expressed as mg kg−1 of dry matter on a calibration curve. Olive oil Acidity (% oleic acid), peroxide value (meq O 2 kg-1) and spectroscopic indices were meas- ured according to EU official method (EC Reg. 1989/2003). Extraction, identification and determination of phenolic compounds were performed in agree- ment with IOC Official Method (IOC, 2009) by an HPLC equipment consisting of a Hewlett Packard 1200 diode-array detector system and a Hewlett Packard model 1200 autosampler (Agilent Tech- nologies, Santa Clara, California, USA). Secoir- idoids, lignans, flavonoids and phenolic acids were quantified in mg tyrosol kg oil -1. The total pheno- lic compounds content (mg tyrosol kg oil -1) was deter- mined using the sum of the peak areas of phe- nols recorded at 280 nm. The tocopherol content was determined ac- cording to ISO 9936:2006 (ISO, 2006) us- ing a Hewlett Packard mod. 1050 liquid chro- matograph with quaternary pump and fluores- cence detector, provided with Hewlett Packard mod.1100 autosampler (Agilent Technologies, Santa Clara, California, USA). Quantitative anal- ysis was carried out using the external standard method. Results were expressed as mg of total tocopherols per kg of oil. The volatile compound content was deter- mined according to the literature (VICHI et al., 2003), using HS-SPME-GC-MS technique (solid phase microextraction of the head space, cou- Ital. J. Food Sci., vol. 27 - 2015 239 pled with a gas chromatograph with a mass spec- trometer as a detector). Analysis was performed using the Trace CG instrument combined with a Trace DSQ Thermo Finnigan instrument (Fish- er Scientific SAS, Illkirch, France). Quantitative analysis was performed using 4-methyl-2-pen- tanol as an internal standard. Results were ex- pressed as mg of volatile compound per kg of oil. Sensory analyses Sensory evaluation of olive oil was performed by a panel test according to the EU official meth- od (EU Reg. 1348/2013). Samples were ana- lyzed by a panel of professional tasters (8 tasters and a panel leader) of CCIAA (Chamber of Com- merce, Industry, Handcraft and Agriculture) of Florence. The panel has been recognized by MI- PAAF (Ministry of Agricultural Policies, Food and Forestry) since 2002. Intensity of both sensory defects and “fruity”, “bitter” and “pungent” attri- butes was assessed and expressed as the medi- an of tasters score on a scale range from 0 to 10. Microbiological analysis Paste and oil samples from each batch were sterilely withdrawn and then transported to the laboratory under refrigerated conditions (4°C). Ten g of olive paste or 10 mL of unfiltered oil were transferred into 90 mL of sterile saline and homogenized for 10 min with a Stomacher Lab Blender 400 (Seward Ltd, Worthing, West Sus- sex, UK) and a magnetic stirrer, respectively. Af- ter decimal dilutions, 100 µL suspension was plated on specific growth media for cell enumer- ation in triplicate using the spread plating tech- nique. Yeasts were counted on MYPG agar (ZUL- LO and CIAFARDINI, 2008) integrated with ampi- cillin and sodium propionate in order to inhib- it growth of bacteria and moulds, respectively (ROMO-SANCHEZ, 2010). The plates were incu- bated at 30°C for 48-72 h. Moulds were counted on MYPG agar without inhibitors (KAWAI et al., 1994) after incubation at 30°C for 24-48 h. Final- ly, total mesophilic microorganisms were count- ed on Plate Count Agar (Oxoid Ltd, Basingstoke, Hampshire, UK) after incubation at 30°C for 48 h. Filtered oil samples (100 mL) were microfil- tered through nitrocellulose filters with a poros- ity of 0.45 µm (Minisart NML-Sartorius, Göttin- gen, Germany), which was able to retain yeasts and moulds. Then the nitrocellulose filters con- taining the microorganisms were washed with 10 mL saline and placed onto the specific me- dia described above. Data processing Chemical, sensory and microbiological deter- minations were processed according to one-way ANOVA followed by Tukey’s test (significance lev- el: p = 0.05). Principal Component Analysis (PCA) was used to classify samples by Statistica 7.0 software package (Stasoft GmbH, Hamburg, Germany). Correlation studies between microbial cell den- sity and the volatile compounds content of oil samples were carried out by calculating both Pearson and Spearman rank correlation coeffi- cients (significance level: α = 0.05). RESULTS Characteristics of olive and olive oil samples Chemical characteristics of processed olives are given in Table 1. They show a slight increase in olive ripening level between the two dates of harvesting. As reported in the literature (RYAN et al., 2002; SERVILI et al., 2004; CHERUBINI et al., 2009), a significant decrease in phenolic com- pounds content occurred, and a decrease in sugar content, even if significant only for Fran- toio cultivar, was also observed. No significant variations were measured in both water and ol- ive oil contents during the harvesting interval. Sensory and chemical characteristics of ex- tracted olive oil are given in Table 2, while their volatile compounds content is reported in Table 3. Samples are encoded in relation to olive cul- tivar, harvesting date and batch. Table 2 shows that all olive oil samples ex- tracted from olives of the first harvesting date were extra virgin. They had much lower values Table 1 - Olive characteristics on two harvesting dates (HD1 and HD2). SD: standard deviation; different letters in the same row indicate significant differences (p < 0.05) for the same cultivar; dm: dry matter. Frantoio Cultivar Moraiolo Cultivar HD1 HD2 HD1 HD2 Mean value SD Mean value SD Mean value SD Mean value SD Phenolic Compounds (mg/kg dm) 33000 a 2285 26000 b 1811 37000 a 2579 30000 b 2097 Sugar Content (g/kg dm) 75 a 5 54 b 4 77 a 5 69 a 5 Water Content (g/kg) 391 b 20 437 a 22 411 a 21 430 a 22 Oil Content (g/kg dm) 440 a 31 460 a 32 500 a 35 450 a 31 240 Ital. J. Food Sci., vol. 27 - 2015 than EU legal chemical limits, no sensory defects and a value of “fruity” attribute with a me- dium intensity of perception, as reported in EU Reg. 1348/2013. Conversely, all olive oil sam- ples extracted from olives of the second harvesting date were not extra virgin, as they had sig- nificant sensory defects. De- spite this, they were in compli- ance with all legally established (EU Reg. 1348/2013) chemical characteristics and “fruity” at- tribute. As a result of malaxation op- erating conditions at low oxi- dative stress impact, olive oil resulted in high phenolic com- pounds content and a phenolic profile characterized by slightly degraded phenolic compounds (SERVILI et al., 2004; GOMEZ- RICO et al., 2009). The total phe- nolic compound content was approx. 670 mg/kg; the dialde- hydic form of decarboxymethyl oleuropein aglycone (3,4-DH- PEA-EDA) was the most abun- dant phenolic compound, and its content was approx. 150 mg/kg; low (approx. 0.7 mg/kg) hydroxytirosol content (3,4-DH- PEA) was found. No significant differences were observed be- tween samples at the two har- vesting dates; the medium in- tensity of “bitter” and “pungent” attribute perception can be ex- plained by phenolic compounds values (EU Reg. 1348/2013). Volatile compounds content of olive oil samples were sub- divided into chemical classes, as reported in Table 3. Com- pounds that have been shown (KALUA et al., 2007; DI GIACIN- TO et al., 2010; APARICIO et al., 2012) to be significantly relat- ed to oil defects are reported. Underlined volatile compounds are intermediate of LOX path- way and they are considered (DI GIACINTO et al., 2010; KOT- TI et al., 2011; APARICIO et al., 2012) to be responsible for ol- ive oil “fruity” positive attribute. A sum of underlined com- pound contents is reported in Table 2 as “Fruity volatile com- pounds”; “fruity” attribute, measured by panel test, can be explained by these values. Ta b le 2 - C h em ic a l a n d s en so ry c h a ra ct er is ti cs o f ex tr a ct ed o li ve o il . Fr an to io c ul tiv ar M or ai ol o cu lti va r H D 1 H D 2 H D 1 H D 2 Ba tc h co de (1 ) F1 a F1 b F1 c M ea n va lu e F2 a F2 b F2 c F2 d M ea n va lu e M 1a M 1b M 1c M 1d M ea n va lu e M 2a M 2b M 2c M 2d M ea n va lu e ± SD ± SD ± SD ± SD EU le ga l c ha ra ct er is tic s Fr ee a cid ity ( % o lei c a cid ) 0. 20 0. 20 0. 22 0. 21 a ± 0. 01 0. 19 0. 19 0. 23 0. 24 0 .2 1a ±0 .0 3 0. 20 0. 22 0. 22 0. 22 0. 22 b ± 0. 01 0. 25 0. 25 0. 27 0. 27 0. 26 a ± 0. 01 Pe ro xid e va lue (m eq O 2/k g oil ) 3. 6 3. 6 3. 5 3. 8b ±0 .1 6. 2 4. 9 5. 3 4. 6 5. 2a ±0 .7 3. 9 4. 0 4. 3 4. 6 4. 2b ±0 .3 5. 3 4. 5 5. 2 4. 9 5. 0a ±0 .4 K 2 32 1.7 6 1.7 1 1.7 4 1.7 4a ±0 .0 3 1.7 7 1.8 0 1.8 7 1.8 5 1.8 2a ±0 .0 5 1.8 0 1.7 4 1.8 2 1.8 0 1.7 9a ±0 .0 3 1.7 9 1.8 6 1.7 9 1.8 3 1.8 2a ±0 .0 3 K 2 70 0. 15 0. 15 0. 16 0. 15 b ± 0. 01 0. 19 0. 19 0. 23 0. 24 0 .2 1a ±0 .0 3 0. 17 0. 16 0. 17 0. 17 0. 17 a ± 0. 01 0. 15 0. 17 0. 17 0. 17 0. 17 a ± 0. 01 ∆K 0. 00 0. 00 0. 00 0. 00 a ± 0. 00 0. 00 0. 00 0. 00 0. 00 0 .0 0a ±0 .0 0 0. 00 0. 00 0. 00 0. 00 0. 00 a ± 0. 00 0. 00 0. 00 0. 00 0. 00 0. 00 a ± 0. 00 P.A .: F ru ity * ( 0- 10 ) 3. 5 3. 4 4. 1 3. 7a ±0 .4 n. d. 2. 9 3. 3 4. 0 3. 4a ±0 .6 3. 8 3. 9 4. 4 3. 7 4. 0a ±0 .3 2. 9 3. 2 3. 4 4. 1 3. 4a ±0 .5 P.A .: B itte r* (0 -1 0) 2. 6 3. 6 3. 6 3. 3a ±0 .6 n. d 3. 6 3. 3 5. 1 4. 0a ±1 .0 2. 6 3. 6 4. 6 3. 7 3. 6a ±0 .8 3. 8 3. 5 3. 5 3. 5 3. 6a ±0 .1 P.A .: P un ge nt * ( 0- 10 ) 3. 7 4. 5 4. 9 4. 4a ±0 .6 4. 1 4. 1 4. 2 5. 0 4. 4a ±0 .4 4. 0 4. 2 4. 7 4. 6 4. 4b ±0 .3 4. 9 5. 8 5. 6 5. 6 5. 5b ±0 .4 N. A. : R an cid * ( 0- 10 ) 0. 0 0. 0 0. 0 0. 0b ±0 .0 0. 0 1.9 1.6 1.6 1.3 a ± 0. 9 0. 0 0. 0 0. 0 0. 0 0. 0b ±0 1.2 1.4 1.4 1.1 1.3 a ± 0. 2 N. A. : F us ty* (0 -1 0) 0. 0 0. 0 0. 0 0. 0b ±0 .0 2. 1 1.6 1.3 1.0 1.5 a ± 0. 5 0. 0 0. 0 0. 0 0. 0 0. 0b ±0 1.1 1.2 0. 8 1.0 1.0 a ± 0. 2 N. A. : W ine y-V ine ga ry * ( 0- 10 ) 0. 0 0. 0 0. 0 0. 0b ±0 .0 1.7 1.7 0. 6 1.0 1.3 a ± 0. 5 0. 0 0. 0 0. 0 0. 0 0. 0b ±0 1.2 1.4 1.4 1.1 1.3 a ± 0. 2 Bi oc om po un ds (m g/ kg ) To ta l p he no l c on te nt 72 0 64 0 72 0 69 0a ±4 6 56 0 61 0 70 0 69 0 64 0a ±6 8 69 0 67 0 73 0 72 0 70 0a ±3 0 57 0 70 0 61 0 70 0 64 0a ±6 5 Ol eu ro pe in 39 32 32 34 a ± 4 30 40 50 50 40 a ± 11 40 50 70 60 55 a ± 11 40 50 60 60 50 a ± 10 3, 4 DH PE A- ED A (2 ) 23 0 17 0 17 0 19 0a ±3 2 13 0 13 0 15 0 14 0 14 0a ±1 0 15 3 14 2 15 2 15 7 15 1a ±6 10 0 14 0 10 0 11 0 11 0b ±1 9 p- HP EA -E DA (3) 92 84 84 83 a ± 4 11 0 90 60 50 80 a ± 27 67 59 46 47 55 a ± 10 40 38 34 26 34 b ± 6 3, 4 DH PE A- EA (4) 31 29 26 29 a ± 2 41 38 58 51 47 a ± 9 26 36 33 33 32 b ± 4 48 48 40 49 46 a ± 4 pH PE A- EA (5) 8. 4 8. 1 8. 0 8. 1a ±0 .2 8. 0 8. 0 19 .0 18 .0 13 a ± 6 8. 6 8. 2 7.7 7.4 8. 0a ±0 .5 17 15 11 12 14 b ± 2 3, 4 DH PE A (6 ) 0. 5 0. 5 0. 7 0. 6a ±0 .1 0. 4 0. 6 0. 8 0. 7 0. 6a ±0 .2 0. 8 0. 6 0. 7 0. 6 0. 7a ±0 .1 0. 6 0. 7 1.2 0. 9 0. 9a ±0 .3 To co ph er ols 29 0 25 0 24 0 26 0a ±2 5 29 2 29 2 28 7 28 8 29 0b ±3 24 9 25 6 25 2 25 7 25 4a ±4 28 0 28 0 25 0 25 0 27 0a ±1 7 Fr uit y v ola tile co m po un ds (7) (m g/ kg ) 13 .4 14 .0 13 .4 13 .6 a ± 0. 4 9. 8 10 .2 10 .9 10 .5 10 .3 b ± 0. 4 13 .6 13 .4 13 .1 14 .2 13 .6 a ± 0. 5 11 .7 11 .5 13 .0 10 .9 11 .8 b ± 0. 9 H D : h ar ve st in g da te ; S D : s ta nd ar d de vi at io n; a, b d iff er en t l et te rs in th e sa m e ro w in di ca te s ig ni fic an t d iff er en ce s (p < 0 .0 5) fo r t he s am e cu lti va r; n. d. : n ot d et er m in ed ; * m ed ia n of th e ta st er s sc or e; P .A . a nd N .A .: Po si tiv e an d N eg at iv e A ttr ib ut es . (1 ) F : F ra nt oi o; M : M or ai ol o; 1 : F irs t h ar ve st in g da te ; 2 : S ec on d ha rv es tin g da te ; a , b , c , d : o liv e ba tc he s. (2 ) D ia ld eh yd ic fo rm o f d ec ar bo xy m et hy l o le ur op ei n ag ly co ne ; ( 3) D ia ld eh yd ic fo rm o f d ec ar bo xy m et hy l l ig st ro si de a gl yc on e; (4 ) O le ur op ei n ag ly co ne ; ( 5) L ig st ro si de a gl yc on e; (6 ) H yd ro xy ty ro so l; (7 ) S um o f t he u nd er lin ed v ol at ile c om po un d co nt en ts in T ab le 3 . Ital. J. Food Sci., vol. 27 - 2015 241 T a b le 3 - V o la ti le c o m p o u n d s co n te n t o f ex tr a ct ed o li ve o il . H D : h a rv es ti n g d a te ; n .d . n o t d et er m in ed . A. C las s o f e ste rs , a cid s a nd h yd ro ca rb on s     M et hy l a ce ta te Et hy l a ce ta te Bu ty l a ce ta te Ci s- 3- he xe ny l Tr an s- 2- he xe ny l Bu ty ric a ci d Pe nt an oi c ac id He xa no ic a ci d O ct an oi c ac id H ep ta n O ct an ac et at e ac et at e   HD Ba tc h (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) co de Fr an to io 1 F1 a 0. 03 6 0. 01 9 0. 00 2 0. 14 4 0. 00 3 nd 0. 01 0 0. 27 4 0. 07 0 0. 00 7 0. 04 8 F1 b 0. 02 2 0. 01 7 0. 00 2 0. 13 1 0. 00 2 nd 0. 01 3 0. 33 3 0. 08 9 0. 00 6 0. 04 3 F1 c 0. 01 5 0. 02 0 0. 00 2 0. 13 4 0. 00 1 nd 0. 01 1 0. 25 5 0. 04 7 0. 00 4 0. 03 8 2 F2 a 0. 01 5 0. 04 2 0. 00 2 0. 07 0 nd 0. 01 1 0. 00 7 0. 21 0 0. 12 1 0. 00 5 0. 05 3 F2 b 0. 00 7 0. 03 5 0. 00 2 0. 40 9 nd 0. 01 0 0. 00 8 0. 24 3 0. 13 5 0. 00 2 0. 03 9 F2 c 0. 00 8 0. 02 7 0. 00 1 0. 61 4 nd 0. 01 2 0. 01 3 0. 23 7 0. 17 0 0. 00 3 0. 03 6 F2 d 0. 00 6 0. 02 3 0. 00 1 0. 61 9 nd 0. 01 0 0. 01 0 0. 23 5 0. 16 6 0. 00 3 0. 03 6 M or ai ol o 1 M 1a 0. 00 5 0. 01 8 0. 00 1 0. 74 7 0. 00 3 nd 0. 00 4 0. 23 6 0. 09 5 0. 00 3 0. 03 2 M 1b 0. 00 5 0. 01 6 0. 00 2 1.0 70 0. 02 7 nd 0. 00 7 0. 25 9 0. 08 3 0. 00 5 0. 03 1 M 1c 0. 00 6 0. 01 9 0. 00 3 1.1 73 0. 00 4 nd 0. 00 8 0. 27 7 0. 06 5 0. 00 3 0. 03 3 M 1d 0. 00 5 0. 01 5 0. 00 1 0. 48 0 0. 00 6 nd 0. 00 4 0. 21 4 0. 06 5 0. 00 5 0. 02 8 2 M 2a 0. 00 6 0. 02 1 0. 00 1 1.0 32 nd 0. 01 3 0. 01 2 0. 27 7 0. 18 5 0. 00 3 0. 03 5 M 2b 0. 00 5 0. 02 0 0. 00 2 0. 85 8 nd 0. 01 3 0. 01 1 0. 26 1 0. 16 0 0. 00 4 0. 04 0 M 2c 0. 00 8 0. 02 2 0. 00 1 0. 92 7 nd 0. 01 2 0. 00 8 0. 20 1 0. 11 9 0. 00 2 0. 03 6     M 2d 0. 00 6 0. 02 3 0. 00 1 0. 70 1 nd 0. 01 4 0. 01 5 0. 26 2 0. 17 1 0. 00 3 0. 03 7 B. C las s o f a ld eh yd es Va le ra ld he yd e He xa na l Tr an s- 2- Ci s- 3- He pt an al Tr an s- 2- Oc ta na l Tr an s- 2- 2. 4 Tr an s- 2- Be nz al de hy de Tr an s- 2- Tr an s- 2- Pe nt en al He xe na l He xe na l He pt en al He xa di en al Oc ta na l No ne na l De ce na l HD Ba tc h (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) co de Fr an to io 1 F1 a 0. 12 8 0. 58 6 0. 03 1 1.4 21 0. 02 5 10 .0 90 0. 13 1 0. 03 8 0. 29 2 0. 01 5 0. 03 1 0. 19 0 0. 22 9 F1 b 0. 10 6 0. 64 4 0. 03 5 1.8 22 0. 02 6 10 .2 17 0. 15 9 0. 03 3 0. 31 3 0. 01 0 0. 03 1 0. 19 6 0. 23 2 F1 c 0. 08 7 0. 59 6 0. 03 0 1.5 69 0. 02 5 9. 96 3 0. 13 4 0. 02 4 0. 27 0 0. 01 0 0. 03 0 0. 18 5 0. 23 7 2 F2 a 0. 09 4 0. 48 1 0. 02 3 1.0 85 0. 01 7 7.3 31 0. 06 3 0. 04 2 0. 19 8 0. 01 0 0. 03 0 nd 0. 13 2 F2 b 0. 07 9 0. 50 4 0. 02 8 1.0 84 0. 01 7 6. 93 8 0. 06 6 0. 03 4 0. 18 9 0. 01 5 0. 03 2 nd 0. 12 5 F2 c 0. 07 0 0. 54 7 0. 03 5 1.1 57 0. 02 3 6. 78 5 0. 09 0 0. 03 1 0. 17 7 0. 01 4 0. 03 3 nd 0. 07 8 F2 d 0. 07 1 0. 53 5 0. 03 5 1.0 50 0. 02 2 6. 54 5 0. 10 3 0. 03 0 0. 17 2 0. 01 6 0. 03 4 nd 0. 09 2 M or ai ol o 1 M 1a 0. 09 2 0. 55 6 0. 03 6 1.7 42 0. 02 8 8. 75 0 0. 14 8 0. 01 3 0. 27 1 0. 01 9 0. 02 8 0. 18 7 0. 20 3 M 1b 0. 07 9 0. 44 3 0. 03 6 2. 01 6 0. 02 2 7.6 84 0. 13 7 0. 00 9 0. 26 2 0. 01 6 0. 02 9 0. 19 7 0. 18 M 1c 0. 10 0 0. 48 9 0. 04 2 1.9 37 0. 02 9 7.3 71 0. 15 2 0. 02 0 0. 24 6 0. 01 4 0. 03 1 0. 19 4 0. 20 4 M 1d 0. 08 7 0. 56 3 0. 03 2 2. 04 5 0. 02 5 9. 43 5 0. 11 7 0. 02 3 0. 27 6 0. 01 9 0. 02 8 0. 20 2 0. 18 0 2 M 2a 0. 05 3 0. 53 8 0. 03 9 1.3 20 0. 02 7 6. 59 5 0. 11 7 0. 02 4 0. 20 6 0. 01 2 0. 03 6 nd 0. 08 1 M 2b 0. 04 4 0. 47 4 0. 04 3 1.8 03 0. 02 2 6. 17 2 0. 09 3 0. 02 1 0. 22 9 0. 00 9 0. 03 5 nd 0. 05 8 M 2c 0. 04 2 0. 41 6 0. 04 5 4. 05 1 0. 01 5 5. 12 1 0. 07 2 0. 01 0 0. 34 8 0. 02 2 0. 03 4 nd 0. 05 9    M 2d 0. 07 0 0. 58 5 0. 03 8 1.1 23 0. 02 5 6. 60 7 0. 12 6 0. 03 6 0. 19 2 0. 01 8 0. 03 9 nd 0. 09 9 242 Ital. J. Food Sci., vol. 27 - 2015 A multidimensional map of all samples relat- ed to volatile compounds was obtained by PCA. The relevant sample loading and score plots are reported in Fig. 2. The model explained 60% of data variability along the first (Factor 1) and sec- ond (Factor 2) principal components. A comparison between the score plot and the loading plot showed that olive oil samples ex- tracted from olives of the second harvesting date were all positioned on the left side of the plot. They were characterized by high values of ben- zaldehyde, 2-butanone, butyric acid, 2-heptanol, octanoic acid, 1-octen-3-ol, 1-octen-3-one and 2-octanone. All these compounds are related to olive oil defects: These compounds have been associ- ated with “musty”, “winey–vinegary” and “fus- ty” defects by some literature data (KALUA et al., 2007; APARICIO et al., 2012), whereas they have been associated with “rancid” defect by DI GIACINTO et al. (2010). Microbial ecology of oil extraction process Cell concentrations of dominant microbial populations at different steps of oil extraction process from Frantoio and Moraiolo cultivar ol- ives are shown in Tables 4 and 5, respectively. Yeasts and/or moulds were always the domi- nant populations, independently of the sampling point. Cell density of bacteria only accounted for 1% of the total microbial counts on PCA plates. The cell concentrations in olive paste after crushing (P) and in extracted olive oil (D) ranged between values below 10 and above 104 CFU/g or mL. These values were higher than that ob- tained from filtered olive oil (O), which, in most cases, was < 102 CFU/100 mL. Microbial counts of each olive batch were of- ten affected by high standard deviation values, as it typically occurs in manufacturing process- es of raw materials (such as olives) at industri- al scale. A rough general pattern for microbial evolution during olive processing could none- theless be drawn. Mould counts in olive paste after crushing (PM) were always significantly higher than those in extracted olive oil (DM), while yeast counts showed a different behaviour. In most olive batches (from both Frantoio and Moraiolo cultivars) of the first harvesting date, yeast counts decreased by about one order of magnitude from olive paste after crushing (PY) to extracted olive oil (DY), as expected on the basis of olive oil yield. At the second harvesting date, yeast counts remained almost unchanged from olive paste (PY) to olive oil (DY), or even in- creased in extracted olive oil (DY), suggesting a progressive yeast colonization of the malaxation equipment and/or “decanter”. Indeed, at the sec- ond harvesting date, olive paste (PY) harboured almost the same yeast concentration as that at the first harvesting date, with values ranging be-C . C las s o f a lco ho ls, ke to ne s a nd p he no ls 1- Pe nt en -3 -o l 2- He pt an ol Pe nt an ol 1- Oc te n- 3- ol Tr an s- 3- Ci s- 3- Ci s- 2- 2- Bu ta no ne 2 -O ct an on e 1- Oc te n- 1- Pe nt en - 6- m et hy l- Gu ai ac ol Ph en ol Et hy l- 4- Et hy l- He xe no l He xe no l Pe nt en ol 3- on e 3- on e 5- He pt en -2 -o ne gu ai ac ol ph en ol HD Ba tc h (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) (m g/ Kg ) co de Fr an to io 1 F1 a 0. 44 6 nd 0. 00 5 0. 00 2 0. 00 4 0. 24 2 0. 33 7 0. 43 3 0. 00 3 0. 00 1 0. 43 3 0. 00 4 0. 00 5 0. 26 1 0. 13 1 nd F1 b 0. 45 4 nd 0. 00 5 0. 00 2 0. 00 4 0. 26 0 0. 35 8 0. 47 7 0. 00 4 0. 00 1 0. 47 7 0. 00 5 0. 00 5 0. 27 5 0. 12 4 nd F1 c 0. 47 5 nd 0. 00 4 0. 00 2 0. 00 3 0. 20 0 0. 35 3 0. 48 4 0. 00 4 0. 00 1 0. 48 4 0. 00 4 0. 00 8 0. 25 3 0. 15 8 0. 08 0 2 F2 a 0. 40 3 0. 22 5 0. 00 4 0. 00 5 0. 00 4 0. 08 9 0. 28 8 0. 36 0 0. 01 4 0. 00 3 0. 36 0 nd 0. 00 2 0. 19 8 nd nd F2 b 0. 49 1 0. 29 4 0. 00 4 0. 00 7 0. 00 5 0. 25 9 0. 31 0 0. 48 1 0. 01 2 0. 00 2 0. 48 1 nd 0. 00 4 0. 20 0 nd nd F2 c 0. 59 7 0. 35 1 0. 00 5 0. 00 8 0. 00 6 0. 43 1 0. 38 2 0. 70 8 0. 01 0 0. 00 1 0. 70 8 nd 0. 00 5 0. 19 3 nd nd F2 d 0. 58 2 0. 34 6 0. 00 5 0. 01 0 0. 00 7 0. 45 4 0. 37 6 0. 71 2 0. 03 5 0. 00 4 0. 71 2 nd 0. 00 6 0. 20 3 nd nd M or ai ol o 1 M 1a 0. 55 4 nd 0. 00 6 0. 00 2 0. 00 6 0. 52 1 0. 42 8 0. 58 6 0. 00 8 0. 00 2 0. 58 6 0. 00 4 0. 00 4 0. 24 5 nd nd M 1b 0. 56 9 nd 0. 00 6 0. 00 2 0. 00 9 0. 69 5 0. 43 5 0. 59 0 0. 00 4 0. 00 2 0. 59 0 0. 00 2 0. 00 4 0. 25 1 0. 11 7 nd M 1c 0. 58 4 nd 0. 00 5 0. 00 2 0. 00 8 0. 70 9 0. 44 4 0. 60 6 0. 00 6 0. 00 2 0. 60 6 0. 00 4 0. 00 3 0. 35 2 0. 11 4 0. 06 7 M 1d 0. 57 1 nd 0. 00 6 0. 00 2 0. 00 6 0. 34 6 0. 44 7 0. 53 3 0. 00 5 0. 00 1 0. 53 3 0. 00 3 0. 00 4 0. 23 8 nd nd 2 M 2a 0. 65 8 0. 37 5 0. 00 5 0. 00 9 0. 00 8 0. 71 4 0. 42 0 0. 71 5 0. 00 9 0. 00 2 0. 71 5 nd 0. 00 6 0. 20 8 nd nd M 2b 0. 57 5 0. 36 9 0. 00 5 0. 00 7 0. 00 5 0. 74 5 0. 41 2 0. 71 1 0. 00 3 0. 00 2 0. 71 1 nd 0. 00 4 0. 21 0 nd nd M 2c 0. 60 3 0. 36 1 0. 00 4 0. 00 6 0. 00 6 0. 85 3 0. 40 2 0. 72 0 0. 00 3 0. 00 2 0. 72 0 nd 0. 00 2 0. 19 6 nd nd     M 2d 0. 63 6 0. 38 1 0. 00 6 0. 01 2 0. 00 7 0. 50 9 0. 39 4 0. 75 3 0. 01 1 0. 00 4 0. 75 3 nd 0. 00 6 0. 20 5 nd nd Ital. J. Food Sci., vol. 27 - 2015 243 Fig. 2 - Principal Component Analysis carried out on volatile compounds content of olive oil samples. A: similarity map de- termined by Principal Component (Factor) 1 and 2; B: projection of the variables on the factor plane. Samples are coded as reported in Table 2. tween 102 and above 103 CFU/g. On the contra- ry, the extracted olive oil of the second harvest- ing date (DY) harboured yeast concentrations, in most cases, of about one or two orders of mag- nitude higher than the extracted olive oil of the first harvesting date. Correlation studies demonstrated that mould counts in olive paste after crushing (PM) and in extracted olive oil (DM) were positively related to each other, suggesting that mould contami- nation of unfiltered oil could be affected by the hygienic level of olives (Table 6). On the contra- ry, yeast cell densities in olive paste (PY) and in olive oil (DY) were statistically unrelated, sug- 244 Ital. J. Food Sci., vol. 27 - 2015 Table 4 - Microbial cell counts at different steps of oil extraction process on two harvesting dates (HD) for Frantoio cultivar. P = olive paste after crushing; D = olive oil after extraction by “decanter”; O = olive oil after filtration; TMC = total microbial count; different letters indicate significant differences between different extractive steps of the same olive batch (p < 0.05); when no letter is reported, no significant difference was found. Yeasts Moulds TMC HD Batch Sampling code point Mean SD Mean SD Mean SD 1 F1a P (CFU/g) 1.60 x 103a 1.40 x 102 1.00 x 102 0 1.40 x 103a 1.40 x 102 D (CFU/mL) 4.50 x 101b 7.07 <10 - 4.00 x 101b 2.82 O (CFU/100mL) <1 - <1 - <1 - F1b P (CFU/g) 8.50 x 102a 2.12 x 102 <10 - 1.60 x 103a 5.66 x 102 D (CFU/mL) 1.00 x 102b 2.80 x 101 <10 - 4.00 x 101b 0 O (CFU/100mL) <1 - <1 - <1 - F1c P (CFU/g) 1.10 x 103a 1.41 x 102 8.00 x 102 0 2.00 x 103 1.41 x 103 D (CFU/ml) 3.25 x 102b 3.54 x 101 <10 - 5.00 x 102 2.83 x 102 O (CFU/100mL) <1 - <1 - <1 - 2 F2a P (CFU/g) 1.00 x 102a 1.40 x 101 4.20 x 104a 2.82 x 103 4.80 x 104a 2.83 x 103 D (CFU/mL) 3.00 x 102b 2.80 x 101 4.00 x 101b 2.82 3.00 x 102b 2.88 x 101 O (CFU/100mL) 5.00 x 101c 1.41 <1 - 5.00 x 101b 2.82 F2b P (CFU/g) 2.70 x 103 1.84 x 103 2.85 x 104 2.32 x 104 8.75 x 103a 3.18 x 103 D (CFU/ml) 2.92 x 103 1.99 x 103 3.33 x 101 3.27 x 101 1.00 x 102b 0 O (CFU/100mL) 5.50 x 101 1.41 <1 - 6.50 x 101b 1.41 F2c P (CFU/g) 2.30 x 103 9.90 x 102 2.50 x 104 2.25 x 104 1.10 x 103a 1.41 x 102 D (CFU/ml) 3.26 x 103 1.60 x 103 9.67 x 101 8.96 x 101 1.81 x 103a 7.66 x 102 O (CFU/100mL) 5.50 x 101 2.82 <1 - 1.00 x 101b 2.82 F2d P (CFU/g) 4.00 x 102a 2.83 x 101 3.45 x 104 3.32 x 104 7.00 x 103a 1.41 x 102 D (CFU/ml) 1.38 x 104b 6.36 x 102 1.20 x 102 2.83 x 101 1.35 x 104b 9.90 x 102 O (CFU/100mL) 1.50 x 101a 1.40 5.00 0 4.00 x 101c 2.82 Table 5 - Microbial cell counts at different steps of olive oil extraction process on two harvesting dates (HD) for cultivar Mo- raiolo. P = olive paste after crushing; D = olive oil after extraction by “decanter”; O = olive oil after filtration; TMC = total mi- crobial count; different letters indicate significant differences between different extraction steps of the same olive batch (p < 0.05); when no letter is reported, no significant difference was found. Yeasts Moulds TMC HD Batch Sampling code point Mean SD Mean SD Mean SD 1 M1a P (CFU/g) 1.10 x 103a 1.41 x 102 4.00 x 102a 0 1.45 x 103 6.36 x 102 D (CFU/mL) 4.50 x 101b 7.07 <10 - 4.00 x 101 1 O (CFU/100mL) <1 - 4.00 x 101b 1.41 1.00 x 101 1 M1b P (CFU/g) 3.75 x 103a 3.54 x 102 5.50 x 102 5.36 x 102 5.35 x 103 2.33 x 103 D (CFU/mL) 5.00 x 101b 1.40 <10 - <10 - O (CFU/100mL) <1 - 2.00 x 101 1.00 2.00 x 101 1.40 M1c P (CFU/g) 1.10 x 103a 1.41 x 102 4.00 x 102a 0 2.35 x 103a 9.19 x 102 D (CFU/mL) 6.90 x 103b 2.82 x 102 <10 - 1.50 x 104b 3.54 x 103 O (CFU/100mL) <1 - 1.00 x 101b 2.82 1.00 x 101a 1.40 M1d P (CFU/g) 1.10 x 103 1.41 x 102 4.00 x 102a 1.41 x 101 1.45 x 103 7.78 x 102 D (CFU/ml) 3.20 x 102 3.11 X 102 <10 - 3.50 x 101 2.12 x 101 O (CFU/100mL) <1 - 2.00 x 101b 0 2.00 x 101 1.40 2 M2a P (CFU/g) 1.70 x 103 2.83 x 102 2.60 x 103a 1.98 x 103 2.70 x 103a 4.24 x 102 D (CFU/mL) 1.04 x 103 7.45 x 102 6.00 x 101a 5.66 x 101 9.73 x 102ab 8.60 x 102 O (CFU/100mL) <1 - 5.50 x 101b 2.82 5.50 x 101a 1.41 M2b P (CFU/g) 2.45 x 103a 7.78 x 102 6.00 x 102 5.66 x 102 2.35 x 103a 9.19 x 102 D (CFU/ml) 3.27 x 102b 1.42 x 102 3.50 x 101 2.12 x 101 3.80 x 102b 2.31 x 102 O (CFU/100mL) 1.00 x 101c 0 1.60 x 102 2.82 x 101 7.50 x 101b 3.53 M2c P (CFU/g) 7.45 x 103 2.19 x 103 1.00 x 103 2.82 x 102 4.15 x 103a 2.12 x 102 D (CFU/mL) 9.72 x 103 5.04 x 103 4.00 x 101 3.66 x 101 1.16 x 103b 1.07 x 103 O (CFU/100mL) 8.00 x 101 14.00 <1 - 1.65 x 102b 3.00 M2d P (CFU/g) 1.65 x 103 1.61 x 103 6.75 x 103 1.77 x 103 5.80 x 103 3.11 x 103 D (CFU/mL) 3.08 x 103 3.02 x 103 6.00 x 101 5.66 x 101 2.53 x 103 2.04 x 103 O (CFU/100mL) 1.10 x 103 2.00 <1 - 5.50 x 101 1.41 Ital. J. Food Sci., vol. 27 - 2015 245 gesting that yeast growth could be encouraged by malaxation and/or “decanting” steps. Final- ly, no correlation was found between yeast and mould concentrations in both olive paste (PY and PM, respectively) and filtered oil (OY and OM, respectively). According to PCA of all microbiological data (Fig. 3), processed olive batches clustered into two different groups, independently of the olive culti- var: The samples of the first harvesting date, har- boring the lowest microbial cell densities, clus- tered in group a), while all batches of the second harvesting date resulted to be included in group b). It is worth noting that both the PCA resulting from all microbiological data (Fig. 3) and the PCA resulting from volatile compounds (Fig. 2) are in full agreement, as olive batches from both sta- tistical analyses are clustered in the same way. Finally, some statistically significant correla- tions were found between microbial cell densi- ties at the different steps of oil processing and some volatile compounds of olive oil. The signifi- cant correlations between yeast (Y) and mould (M) counts, in both extracted (D) and filtered olive oil (O), and volatile compounds content of the final olive oil samples are reported in Table 7. In par- ticular, correlation coefficients (i.e. Pearson and Spearman) agreed on indicating significant posi- tive correlations between yeast and mould counts in olive oil, both before and after filtration, and some volatile compounds; among the latter, the highest significance was related to ethyl acetate, 2-butanone, butyric acid, pentanol, 2-heptanol, octanoic acid and 1-octen-3-ol contents. Since most of these compounds were identi- cal to those correlated to olive oil batches with sensory defects, as described in the previous paragraph, yeast and mould contamination may have been responsible for those sensory defects. Which specific sensory defects were associated with the above-mentioned compounds could not be explained, as in the literature “rancid”, “fus- ty”, “winey–vinegary” and “musty” defects have been associated with both yeasts and moulds. As an example, a recent study demonstrated the capability of some oil born strains of Candida Fig. 3 - Principal Component Analysis of the various olive batches tested by considering as variables the microbial cell con- centrations during various extraction process steps. Samples are coded by combination of letters which identify both sam- ples at processing steps (P = olive paste after crushing; D = olive oil after centrifugation by “decanter”; O = olive oil after fil- tration) and microorganisms (TMC = total microbial count; Y = yeasts; M = moulds). A: similarity map determined by Princi- pal Component (Factor) 1 and 2; B: projection of the variables on the factor plane. Table 6 - Correlation coefficients calculated between microbial contaminations (Y = yeasts; M = moulds) of olive paste af- ter crushing (P) and microbial contaminations of extracted (D) and filtered olive oil (O). Statistically significant correlations (p<0.05) are underlined. DY DM OY OM   Spearman Pearson Spearman Pearson Spearman Pearson Spearman Pearson r r r r r r r r PM 0.8304 0.7347 -0.1575 -0.2485 PY 0.08641 0.05563 0.2841 0.1241 246 Ital. J. Food Sci., vol. 27 - 2015 Table 7 - Correlation coefficients calculated between yeast (Y) and mould (M) counts of extracted and filtered olive oil (D) and volatile compounds of the final olive oil samples (O). Statistically significant correlations (p<0.05) are underlined. DY DM OY OM   Spearman Pearson Spearman Pearson Spearman Pearson Spearman Pearson r r r r r r r r Esters, acids and hydrocarbons Methyl acetate -0.006737 -0.3253 -0.08564 -0.4042 0.01485 -0.2926 -0.8317 -0.5387 Ethyl acetate 0.4464 0.2619 0.6978 0.5603 0.7348 0.6665 -0.4953 -0.4413 Butyl acetate -0.3046 -0.27 -0.5824 -0.5091 -0.3481 -0.3366 -0.1901 -0.1524 Cis-3-hexenil acetate 0.3194 0.325 0.1589 0.2075 -0.0114 0.01915 0.652 0.6335 Trans-2-hexenil acetate -0.5621 -0.4382 -0.862 -0.4494 -0.7775 -0.3811 0.243 0.2902 Butyric acid 0.5532 0.5918 0.8694 0.9727 0.7434 0.8125 0.03821 -0.0028 Pentanoic acid -0.0685 -0.1613 -0.1662 -0.1045 -0.3671 -0.331 0.5344 0.5813 Hexanoic acid -0.1944 -0.3239 -0.337 -0.3943 -0.4902 -0.5172 -0.01623 - Octanoic acid 0.4624 0.4945 0.8818 0.9251 0.6282 0.6242 0.1469 0.1633 Heptan -0.6824 -0.5102 -0.5535 -0.4601 -0.5756 -0.4022 -0.0887 -0.218 Octan -0.1092 -0.2505 0.0174 -0.04 0.205 0.0699 -0.6035 -0.5001 Aldehydes Valeraldheyde -0.5583 -0.4607 -0.708 -0.7013 -0.5681 -0.5144 -0.291 -0.388 Hexanal -0.3341 -0.3271 -0.3477 -0.3659 -0.4185 -0.4153 -0.3585 -0.315 Trans-2-Pentenal 0.3599 0.3423 0.2134 0.1599 0.06842 0.02669 0.5254 0.464 Cis-3-Hexenal -0.2487 0.1017 -0.59 -0.1577 -0.4412 0.03461 0.3088 0.07831 Heptanal -0.3023 -0.3369 -0.4689 -0.5529 -0.7091 -0.7582 0.2273 0.2932 Trans-2-Hexenal -0.6784 -0.6681 -0.8182 -0.8242 -0.6991 -0.7192 -0.2628 -0.1959 Octanal -0.405 -0.4349 -0.6645 -0.7347 -0.7142 -0.7745 0.1745 0.2041 Trans-2-Heptenal -0.0552 -0.01303 0.1559 0.1205 0.2061 0.1596 -0.6411 -0.5767 2.4 Hexadienal -0.4503 -0.3867 -0.7002 -0.6627 -0.4882 -0.42 -0.0957 -0.0823 Trans-2-Octanal -0.0244 0.1057 -0.0918 -0.1199 0.08817 0.08237 -0.0718 -0.1306 Benzaldehyde -0.5041 0.3715 0.5785 0.4958 0.4283 0.3689 -0.1509 0.1584 Trans-2-Nonenal -0.6304 -0.6281 -0.8605 -0.9855 -0.7762 -0.8358 0.07103 0.04037 Trans-2-Decenal -0.5946 -0.615 -0.8119 -0.895 -0.6682 -0.7092 -0.2806 -0.2243 Alcohols, ketones and phenols 1-Penten-3-ol 0.6681 0.4847 0.5822 0.4303 0.3147 0.1095 0.5055 0.5417 2-Heptanol 0.6178 0.6498 0.8857 0.9774 0.7328 0.7846 0.06237 0.0213 Pentanol 0.4111 0.4182 0.8213 0.8118 0.7221 0.7422 -0.1372 -0.2154 Cis-3-Hexenol 0.3252 0.2959 0.2486 0.216 0.07623 0.05016 0.6277 0.5872 Trans-3-Hexenol 0.3032 0.2647 0.2875 0.1636 0.0247 - 0.5309 0.4904 1-Octen-3-ol 0.6212 0.6381 0.9304 0.9199 0.7286 0.7176 -0.0491 -0.0339 Cis-2-Pentenol 0.0486 0.08362 -0.1142 -0.16 -0.2467 -0.2997 0.685 0.618 2-Butanone 0.5204 0.5477 0.782 0.7111 0.5529 0.5283 0.00517 -0.1297 1-Penten-3-one 0.6461 0.5539 0.6247 0.5717 0.412 0.3304 0.3666 0.4029 2-Octanone 0.264 0.4949 0.5565 0.5658 0.4097 0.3878 -0.06146 -0.0623 1-Octen-3-one 0.2545 0.3958 0.5142 0.5573 0.5459 0.5804 0.1314 0.0085 6-methyl-5-Hepten-2-one -0.5882 -0.5785 -0.8678 -0.9186 -0.7828 -0.779 -0.1013 -0.1062 Guaiacol -0.0316 -0.1852 0.08201 -0.1711 -0.2404 -0.3518 -0.1757 -0.1935 Phenol -0.4724 -0.258 -0.8245 -0.8028 -0.8142 -0.7034 0.1395 0.0637 Ethyl-guaiacol -0.3957 -0.4097 -0.6943 -0.7573 -0.6262 -0.6423 -0.3215 -0.2771 4-Ethyl-phenol -0.0820 -0.1033 -0.4472 -0.4672 -0.3997 -0.3962 -0.1999 -0.2132 spp. to induce defects such as “musty” and/or “rancid” in oil (ZULLO et al., 2013). CONCLUSIONS This study was carried out on several olive oil samples extracted by olive batches from Fran- toio and Moraiolo cultivars, harvested on two dif- ferent dates. All extracted olive oil samples from the second olive harvesting date were classified as “non extra virgin”, as they were affected by sensory defects. By combining chemical, sensory, and micro- biological data, it can be assumed that the ol- ive oil samples with sensory defects were sig- nificantly correlated with specific volatile com- pounds (i.e., 2-butanone, butyric acid, 2-hep- tanol, octanoic acid, 1-octen-3-ol). The same vol- atile compounds were correlated to both yeast and mould counts. It could not be evidenced whether a specific sensory defect might result from specific volatile compounds, which in turn can be produced by specific yeasts and moulds. Different processing steps were also identi- fied, which resulted to be the most critical steps to cause the measured sensory defects: (i) the Ital. J. Food Sci., vol. 27 - 2015 247 mould contamination of olives; (ii) the two cen- tral steps of olive oil processing (i.e. malaxation and extraction by “decanter”), which were likely to have enabled some yeast species to grow. A study on identification of yeast isolates and determina- tion of their enzymatic properties is being carried out to further investigate the incidence of yeast populations during olive oil extraction process. ACKNOWLEDGEMENTS Research was supported by “Bando Misura 124 Program- ma di Sviluppo Rurale (PSR) 2007-2013” of Tuscany Region, which was published in the Gazzetta Ufficiale of Tuscany Region (BURT) No. 7 of 16/02/11. REFERENCES Angerosa F., Servili M., Selvaggini R., Taticchi A., Esposto S. and Monteodoro G. 2004. Volatile compounds in vir- gin olive oil: occurrence and their relationship with the quality. J. Chromatogr. A. 1054:17. Aparicio M., Morales M.T. and Garcìa-Gonzàles D.L. 2012. Towards new analyses of aroma and volatiles to under- stand sensory perception of olive oil. Eur. J. Lipid Sci Tech- nol. 114:1114. Aparicio R. and Morales M.T. 1998. Characterization of ol- ive ripeness by green aroma compounds of virgin olive oil. J Agric Food Chem. 46: 1116. Bendini A., Valli E., Barbieri S. and Gallina Toschi T. 2012. Sensory Analysis of Virgin Olive Oil. Ch. 6. In “Olive Oil - Constituents, Quality, Health Properties and Biocon- versions”. D. Boskou (Ed.), p. 109. InTech Open Access Publisher. Cherubini C., Migliorini M., Mugelli M., Viti P., Berti A., Cini E. and Zanoni B. 2009. Towards a technological ripening index for olive oil fruits. J. Sci Food Agric. 89:671. Ciafardini G. and Zullo B.A. 2002. Microbiological activity in stored olive oil. Int. J Food Microbiol. 75:111. Ciafardini G., Zullo B.A., Cioccia G. and Iride A. 2006. Li- polytic activity of Williopsis californica and Saccharomic- es cerevisiae in extra virgin olive oil. Int. J Food Micro- biol. 107: 27. Commission Regulation (EC) No 1989/2003. OJEU. L295:57. Commission Implementing Regulation (EU) No 1348/2013. OJEU. L338:31. Di Giacinto L., Di Loreto G., Di Natale C., Guianni G., Gua- sti S., Migliorini M., Pellegrino M., Perri E. and Santo- nico M. 2010. Caratterizzazione analitica degli attributi sensoriali degli oli vergini di olive (AROMOLIO). Camera di Commercio di Firenze. Florence. Fakas S., Kefalogianni I., Makri A., Tsoumpeli G., Rouni G., Gardeli C., Papanikolaou S. and Agelis G. 2010. Cha- racterization of olive fruit microflora and its effect on ol- ive oil volatile compounds biogenesis. Eur. J. Lipid Sci Technol. 112:1024. Gomez-Rico A., Salvador M.D. and Fregapane G. 2009. Vir- gin olive oil and olive fruit minor constituent as effect- ed by irrigation management based on SWP and TDF as compared to ETc in medium-density young olive orchards (Olea Europea L. cv. Cornicabra and Morisca). Food Res. Int. 42:1067. IOC/T.20/Doc No 29. Official methods of analysis. Deter- mination of biophenols in olive oils by HPLC. COI. Ma- drid (2009). ISO 9936:2006. Animal and vegetable fats and oils – De- termination of tocopherol and tocotrienol contents by high-performance liquid chromatography. ISO. Gene- va (2006). Kalua C.M., Allen M.S., Bedgood Jr. D.R., Bishop A.G, Pren- zler P.D. and Robards K. 2007. Olive oil volatile com- pounds, flavour development and quality: A critical re- view. Food Chem. 100:273. Kawai Y., Otaka M., Kaki M., Oeda Y., Inoue N. and Shina- no H. 1994. Screening of Antioxidant-Producing Fungi in Aspergillus niger Group for Liquid- and Solid-State Fer- mentation. Bull. Fac. Fish. Hokkaido Univ. 45:26. Migliorini M., Cecchi L., Cherubini C., Trapani S., Cini E. and Zanoni B., 2012. Understanding degradation of phe- nolic compounds during olive oil processing by inhibitor addition. Eur. J. Lipid Sci Technol. 114:942. Morales M.T., Luna G. and Aparicio R. 2005. Compara- tive study of virgin olive oil sensory defects. Food Chem. 91:293. Palomares O., Villalba M.M. and Rodriguez R. 2003. The C-terminal segment of the 1,3-�-glucanase Ole e 9 from olive (Olea europea) pollen is an indipendent domain with allergenic activity: expression in Pichia pastoris and char- acterization. Biochem. J. 369:593. Romo-Sànchez S., Alves-Baffi M., Arèvalo-Villena M., Ube- da-Iranzo J. and Briones-Pèrez A. 2010. Yeast biodiver- sity from oleic ecosystems: study of their biotechnologi- cal properties. Food Microbiol. 75:487. Ryan D., Antolovich M., Prenzler P., Robards K. and Lavee S. 2002. Biotranformations of phenolic compounds in Olea Europea L.. Hort. Sci. 92:147. Servili M., Selvaggini R., Esposto S., Taticchi A., Montedo- ro G.F. and Morozzi G. 2004. Health and sensory prop- erties of virgin olive oil hydrophilic phenols: agronomic and technological aspects of production that affect their occurence in the oil. . J. Chromatogr. A. 1054:113. Vichi S., Castellote A.I., Pizzale L., Conte L.S., Buxaderas S. and Lòpez-Tamames E. 2003. Analysis of virgin olive oil volatile compounds by headspace solid-phase microex- traction coupled to gas chromatography with mass spec- trometric and flame ionization detection. J. J. Chroma- togr. A. 983:19. Vichi S., Romero A., Tous J. and Caixach J. 2011. The ac- tivity on healthy olive microbiota during virgin olive oil extraction influences oil chemical composition. J. Agric. Food Chem. 59:4705. Zullo B.A. and Ciafardini G. 2008. Lypolitic yeast distribu- tion in commercial extra virgin olive oil. Food Microbi- ol. 25:970. Zullo B.A., Cioccia G. and Ciafardini G. 2010. Distribution of dimorphic yeast species in commercial extra virgin ol- ive oil. Food Microbiol. 27:1035. Zullo B.A., Cioccia G. and Ciafardini G. 2013. Effects of some oil borne yeasts on the sensory characteristics of Ital- ian virgin olive oil during storage. Food Microbiol. 36:70. Paper Received July 30, 2014 Accepted November 12, 2014 http://www.sciencedirect.com/science/article/pii/S0740002010001802 http://www.sciencedirect.com/science/article/pii/S0740002010001802 http://www.sciencedirect.com/science/article/pii/S0740002010001802