Quality losses in virgin olive oil due to washing and short-term storage before olive milling

To identify critical points during olive mill pre-processing operations, the effect of the closed circuit washing stage on olive microbiological contamination, and the influence of the successive short-term storage on olives and virgin olive oil (VOO) quality were evaluated. Microbiological, physical, and chemical parameters were assessed in olives and oils at three mill pre-processing stages: reception, washing, and short-term storage. Olive washing in closed loop systems was shown to be a critical control point at the olive mill due to microbiological cross-contamination and fruit physical damage. Moreover, when the olives were short-term stored before oil extraction, positive VOO sensory attributes decreased by as much as one point of intensity, as justified by the changes observed in phenolic and lipoxygenase derived volatile compounds. These results confirm the high risk of fruit cross-contamination due to the poor hygiene of the water used in olives mills to wash olives, and point out the effect on VOO quality of a common practice such as short term silo storage of olives. 
 
 
 
Practical applications: The identification of new critical control points within pre-processing operations in ordinary usage, and commonly accepted as good production practices, will contribute to enhance virgin olive oil quality. Preventive actions can be undertaken on the basis of the reported results, such as the control of water hygiene and short term storage conditions. 
 
 
 
 
 
 
 
The figure shows the microbiological contamination of olives during washing with recycled water and during short-term silo storage, and the loss of sensory quality in the correspondent virgin olive oils.


INTRODUCTION 33
The concept of critical production steps has recently been applied to virgin olive oil (VOO) 34 production as a tool to ensure the quality of the product [1]. Several critical points, which must 35 be monitored to allow control of the sensory attributes of the olive oil, have been identified 36 from harvesting to VOO storage. Among post-harvest operations prior to oil extraction, 37 storage of the olives is the step that has been most considered. In the past years, several 38 studies have been carried out to evaluate the effect of long time storage on olive oil quality on 39 the quality of the olives and the oils extracted from them [2][3][4][5][6][7]. The storage periods evaluated 40 range from three days to three weeks at temperatures from 4ºC to 20ºC. The conclusion to be 41 drawn is that storage conditions are crucial for the quality of VOO. However, in most cases, 42 storage for several days could not usually be considered an option; in order to preserve olive 43 quality until processing for oil extraction, it is recommended that storage be short-term (<24h) 44 [8], in keeping with the mill processing capacity. Although short-term silo storage is a common 45 practice adopted to optimize the processing capacity of mills, little information is available on 46 its effect on olive and oil quality. 47 In addition to the effects of storage conditions, recent reports indicate that there is a risk of 48 microbiological cross-contamination at olive mills during washing in closed circuits [9][10][11] and 49 Mac Conkey agar and Pseudomonas on Cetrimide agar supplemented with 100 mg/L 110 cycloheximide (Cetrimide-C). The plates were incubated at 30 °C during 3-5 days and viable 111 counts were expressed as log cfu/g olive. Analyses were performed in triplicate. 112

Virgin olive oils quality indices and sensory analysis 113
Free acidity, coefficients of specific extinction at 232 and 270 nm (K 232 and K 270 ), and peroxide 114 value (PV) of VOO samples obtained from the assay were determined in analytical duplicate 115 according to regulation (EU) No 1348/2013 [16].  Vichi et al. [18]. Briefly, 2 g of oil spiked with 4-methyl-2-pentanol (internal standard; final 126 concentration 1.5 mg/kg), was weighed into a 10 mL vial fitted with a silicone septum. The vial 127 was placed into a water or bath fixed at 40 ºC, where the sample was maintained under 128 magnetic stirring (700 rpm). After 10 minutes of sample conditioning, a DVB/CAR/PDMS fiber 129 was exposed during 30 min to the oil headspace and immediately desorbed in the gas 130 chromatograph injector. Each extraction was performed in duplicate.

Olive quality parameters 174
Olive mill pre-processing operations had a remarkable influence on the physical and hygienic 175 conditions of the olives. First, the integrity of the olives ( Table 2) was assessed by visual 176 examination (n=100 for each sampling) and computing bruised, squashed and fermented 177 fruits. The initial incidence of injured fruit, corresponding to real conditions of handpicking and 178 transport, is relatively high because it comprises also injuries of very low intensity. The is given by the initial differences between olive batches, and it is in turn explained by the 181 differences in the maturity of olives from the different batches. The incidence of damaged 182 fruits progressively increased through the pre-processing steps from reception to silo exit, 183 prior to milling. The loss of integrity due to blows during unloading and throughout the 184 washing circuit is especially important if the olives are stored before milling, because rupture 185 of the tissues provides a foothold for microbial growth. During silo storage, healthy olives 186 undergo further damage caused by the weight of olives in the silo and fermentation processes. 187 From the point of view of hygiene, microbiological assays showed that on delivery to the mill, 188 fresh olives intended for oil production presented spontaneous microbiota composed by fungi, 189 lactic bacteria, enterobacteria and Pseudomonas (Table 2), in agreement with previous reports 190 [5,11,12]. At this point, considerable batch-to-batch variability was observed in contamination 191 by Pseudomonas, enteric and acetic bacteria, as evidenced by the high standard deviation. 192 Despite the heterogeneous microbiological profile of the olive batches on reception, the stage 193 of passing through the olive mill washing tank resulted in a significant increase of 194 microbiological contamination, also as previously reported [11]. This additional contamination 195 was fairly similar for the different olive batches, and it remained after short-term silo storage. 196 During this last step, a further increase of lactic acid bacteria was observed. 197 It should be considered that these silo are usually not washed during the harvesting season, 198 with heavy risks for the hygienic aspects of stored fruits. The surfaces of silo can be covered 199 by molds, so the risk of cross-contamination with mycotoxins should be considered in future 200

research. 201
These results confirm the high risk of cross-contamination due to the use of recycled water to 202 wash the olives [9][10][11], and the need to establish critical hygiene control points in olive oil 203 production process 204 Finally, no significant differences in the VOO yield have been found after the distinct 205 treatments ( Table 2), so we can conclude that possible losses of quality would not be 206 compensated by an increase in the production of VOO. 207

Virgin olive oil quality parameters 208
Analysis of the VOOs obtained from olives collected at each pre-processing step did not 209 produce any evidence that olive deterioration had substantial effects on the official VOO 210 quality parameters ( Table 3): all the oils corresponded to the EVOO category, according to EU 211 regulations [16,17]. Indices of oxidative status such as K 270 and PV were lower in the oils 212 obtained after the olives were washed and stored in the silo; in the case of the stored olives, 213 this could be explained by the reducing anaerobic conditions in the silo. 214 Although no sensory defects arose after any of the pre-processing steps, VOO sensory 215 attributes were influenced by the different operations evaluated ( Table 2). In particular, short-216 term silo storage negatively influenced VOO sensory quality by reducing the intensity of the 217 positive attributes, as established in EU regulation [4,17]: fruity, bitter and pungent; as well as 218 other secondary attributes, such as astringency and greenness ( Table 3). In contrast, the ripe 219 fruit (banana, kiwi, strawberry) note significantly increased after this stage. It is worth 220 mentioning that pre-processing operations carried out according to overall accepted practices, 221 caused a decrease of one point of fruity note intensity, which represents a remarkable loss of 222 sensory quality. Although this loss did not determine the declassing the EVOO to lower 223 categories, it would have commercial repercussions. In fact, according to the EU and the IOC 224 Regulations [17,20], some samples of the study passed from a "intense fruity" (fruity>6) to a 225 "medium fruity" (3<fruity<6) classification after olives short-term storage. As far as we know, 226 this is the first report showing the effect of short term silo storage of olives on the quality of 227 VOO. The global fruity attribute, which is the sum of all the fruity notes perceived by the 228 panelists, not only became weaker after short-term storage, but also turned into a ripe fruit note, as evidenced by the increase of this secondary attribute ( Table 2). These results indicate 230 that during fruit storage, at the very beginning of the olive fruit degradation, and before 231 sensory defects or chemical alterations appear, the fruity note decrease and turns into a ripe 232 fruit note. This modification could be induced by several factors including microbiological 233 activity and the slight over-ripening caused by the storage conditions. 234 The reduction of VOO bitterness after olives storage had been previously described and 235 proposed to increase the acceptability of oils with high bitter intensities [4,21]. In the present 236 work, a slight but significant decrease of bitterness, as well as of puncency and astringency, 237 was observed even storing olives during less than 12h ( Table 3). In contrast to experimental 238 findings at the laboratory scale [11], the intensity of the fruity but not of the bitter descriptor 239 was reduced in oils obtained from olives contaminated during the washing step, due to the 240 activity of olive microbiota during the oil extraction process. This could be explained by the 241 fact that in the present study on reception at the mill the olive batches presented a higher 242 microbiological charge than in the assay cited above, so modifications in the microbiological 243 activity induced during the washing stage were less discernible in the extracted VOO. 244

Volatile and phenolic compounds in VOO 245
The alterations of the VOO sensory profile induced by the pre-processing steps can be 246 explained by modification in the VOO volatile and phenolic fractions. Figure 1 illustrates the 247 modifications induced by the pre-processing steps on C6 compounds from the lipoxygenase 248 pathway. It is worth mentioning that not only the short-term silo storage, but also the washing 249 of olives with contaminated water had a significant effect on VOO C6 volatiles, confirming that 250 the activity of olive microbiota influences VOO chemical composition even during the 251 extraction process [11], and justifying the loss of fruity note reported in VOOs from washed 252 olives [12,13]. In agreement with previous results [11], the C6 alcohols hexanol and (E)-2-253 hexenol were more abundant in the oils obtained after olive washing and silo storage, respectively, while (Z)-3-hexenol progressively increased over both stages. C6 acetate esters 255 showed behavior analogous to that of the corresponding C6 alcohols. In contrast, through the 256 pre-processing steps considered in the present study, and in particular after short-term silo 257 storage of the olives, C6 aldehydes hexanal, (Z)-3-hexenal and (E)-2-hexenal showed a 258 progressive and significant decrease. C5 compounds and pentene dimers from the 259 lipoxygenase pathway were also negatively affected both by microbiological contamination 260 during washing and by microbiological activity during storage (Figure 2). Out of these LOX 261 derivatives, 1-penten-3-one, (Z)-2-pentenol and all the C6 compounds were present at 262 concentrations above their perception thresholds [21], excepting (E)-2-hexanol, which was 263 always below the threshold of 5 mg/kg [22]. Interestingly, hexyl acetate, and (Z)-3-hexenol 264 reached their perception threshold (1 mg/kg) [22] just after the olive washing and storage 265 steps, respectively. On this basis, the changes in the proportion of C6 alcohols and esters 266 versus C6 aldehydes and C5 compounds could explain the change of VOO sensory notes 267 without the appearance of sensory defects. In fact, the green, herbaceous, leafy note has 268 previously been reported to be positively related to some LOX C5 compounds and negatively 269 related to LOX C6 alcohols such as (E)-2-hexenol [22]. Conversely, the ripe fruit note could be 270 associated to the increase of LOX esters (Figure 1), although no previous references about this 271 correlation are available. 272 Among the typical fermentative compounds (Table S1, supplementary information), acetoin 273 and methylbutyl acetate were observed to increase slightly during the storage stage; however, 274 the short duration of the storage meant that their concentrations did not reach those 275 necessary to cause a defect [24]. 276 Meanwhile, the changes in the phenolic fraction induced by the pre-processing operations 277 explained the observed decrease of the related sensory attributes such as bitter, astringent compounds were influenced by the pre-processing steps, including apigenin, the levels of 281 which dropped after the olive washing stage; while the concentration of simple phenol tyrosol 282 was observed to increase in oil after short-term olive storage, probably due to hydrolysis of 283 ligstroside aglycon promoted by microbiological activity, in agreement with previous results 284 [11]. Finally, the progressive decrease of VOO o-diphenols after each olive processing step 285 could explain the observed reduction of the VOO oxidative stability, as measured by the 286 rancimat test ( Table 2). 287 In conclusion, of the post-harvest operations olive washing in closed loop systems, where the 288 water is not renewed in a continuous process, and it is only periodically replaced, was shown 289 to be a critical control point at the olive mill due to microbiological cross-contamination. At the 290 olive mill scale, the volatile composition and the fruity attribute of VOOs were influenced by 291 olive microbiota during oil extraction, while the relatively high initial microbiological charge of 292 some batches on reception hindered the identification of further effects of contamination on 293 VOO sensory and phenolic profiles. Moreover, the common practice of short-term silo storage 294 of olives after washing was shown to influence VOO sensory quality. Although no sensory 295 defects arose from this step, some positive VOO sensory attributes decreased by as much as 296 one point of intensity. The reduction of the green and fruity attributes can be explained by the 297 changes observed in lipoxygenase derived compounds, specifically the reduction in C6 298 aldehydes, pentene dimers and C5 compounds, and the increase in C6 alcohols. Short-term silo 299 storage was also accompanied by the appraisal of a ripe fruit note. Moreover, bitter, pungent 300 and astringent attributes were reduced in oils after olive silo storage, due to the decrease in 301 phenolic compounds. 302 These results confirm the high risk of fruit cross-contamination due to the microbiologically control points for olive oil production process. Moreover, the effect of short term (<12h) olives 305 storage on VOO quality parameters was pointed out.     Table 2. Microbiological profile a , characteristics, and damage of olive fruits b through the preprocessing steps. Differences between groups were assessed by one-way ANOVA. Different letters in the same row indicate significant Fisher's LSDs (least significant differences) (p < 0.05).
Step c (n=5) short-term silo storage.  [16] (median of the intensity sensory attribute); c : secondary positive attributes (median of the intensity sensory attribute). Table 4. Concentration a (mg/kg) of phenols in virgin olive oils obtained from fruits collected after each pre-processing step. Differences between groups were assessed by one-way ANOVA.
Different letters in the same row indicate significant Fisher's LSDs (least significant differences) (p < 0.05).