3.1. Freshly Prepared Minimally-Processed Pineapple Characteristics
A sensory analysis was performed to assign descriptors to freshly minimally-processed samples. The major color descriptor was “yellow”, and olfactive descriptors were, in descending order of frequency of occurrence, “pineapple”, “fresh”, “fruity”, “pomegranate”, “red fruit”, “citrus”, and “sugared”. These olfactive descriptors are primarily related to fruity characteristics. This is consistent with previous studies showing the role of methyl esters and lactones in the typical fruity flavor of pineapple [
19,
20].
The 25 pineapple batches were independently minimally-processed. For each, pH, TA, TSS, firmness, L*, a*, and b*, color parameters were determined on the day of processing. Psychrotrophic bacteria, enterobacteria, as well as yeasts and molds were counted.
Table 3 shows mean values and data dispersion between batches.
The pH values varied between 3.09 and 4.20, and 64% of them ranged between 3.3 and 3.8. Similarly, for TA, 68% of values were in the range 0.73 and 0.98 g/100 mL. The values for pH, TA, and TSS were in accordance with previously published data of carbohydrate content of pineapple [
1,
2,
7,
10,
21,
22].
Firmness range was large, from 2.0 to 6.7 N. Among all parameters, firmness exhibited the highest variation coefficient, of 32%. This range is in accordance with literature data.
For the L* parameter, except for one extremely high value, all batches exhibited values below 41.6. On the opposite, a* values were spread on the whole range, whereas 85% of b* values were above 45. As a consequence, C* and h° exhibited a Normal distribution, with p-values, calculated with the Anderson-Darling test, of 0.53 and 0.23, respectively.
Counts of psychrotrophic bacteria were for most batches below the detection limit. However, five batches exhibited counts above 4 log CFU/g. For yeasts and molds, 72% of batches exhibited counts above 4 log CFU/g, and 16% above 5 log CFU/g. Eventually, for enterobacteria, 40% of batches exhibited counts above 4 log CFU/g, and 8% above 5 log CFU/g. The highest counts were thus observed for yeasts and molds.
Correlations between independent quantitative variables (pH, TA, TSS, microbial counts, L*, a*, b*, firmness) were searched with the non-parametric Kendall test. A positive correlation between L* and psychrotrophic bacteria enumeration was detected with a p-value of 0.028 and a Kendall’s tau coefficient of 0.345. Two negative correlations, between pH and L* and between yeast and mold counts and b*, were detected with p-values of 0.021 and 0.044 and coefficients of -0.336 and -0.291, respectively. Surprisingly, no correlation was pointed out between TA and TSS, which evolve in an opposite way during fruit ripening.
Eventually, correlations were searched between qualitative (season and location of sampling) and quantitative variables. Biserial correlation tool showed a correlation between season and pH, with a
p-value of 0.0002 and a coefficient of 0.60, and a correlation between season and b*, with a
p-value of 0.015 and a coefficient of 0.49. This grouping is visualized from pineapple batches plotted on a graph with
x-axis being b* and y-axis being pH (
Figure 2).
Harvesting season has considerable impacts on the post-harvest quality of pineapple, affecting internal browning and storage life [
19]. The correlation between season and pH is not surprising but could have been expected also with TA and TSS. Pineapple flesh color, especially b* and C*, TSS, TA, and pH were influenced by the season in Thailand (three seasons: summer, rainy and winter) for the Smooth Cayenne cultivar [
23]. The importance of pre-harvest factors on post-harvest quality was underlined by Chen et al. (2009) [
24]. Whereas a model has been proposed to predict TSS of ‘Queen Victoria’ pineapple flesh from agroclimatic conditions of Reunion Island [
21], no correlation has been previously pointed out between pH, b* color parameter and season, for this cultivar or crop location.
The relationship between pH and L* could be explicated by different pineapple flesh compositions, which are reflected on these two parameters. Pineapple flesh color depends on its composition in carotenoids and flavonoids, and the content in those compounds was showed to depend on agroclimatic conditions [
25]. Moreover, the color of flavonoids depends on the pH. By contrast, explaining the correlations between a color parameter and a microbial count would require extensive metabolomic analyses to find out which compounds would be implied.
3.2. Minimally-Processed Pineapple Changes over Refrigerated Storage
Sensory descriptive profiles (olfactory and aspect) were established on minimally-processed pineapple during refrigerated storage, after 3, 7, 10, and 14 days, and compared to freshly processed samples (
Figure 3).
The visual aspect of minimally-processed pineapple clearly turned slightly brown and shiny after 14 days at 4 °C (
Figure 3a). K-means classification positioned samples into four classes (
Figure 3b). Day 0 samples (fresh pineapple) were in class 1. Olfactive descriptors of samples from day 14 were “fermented”, “pungent”, “alcoholic”, “vegetable”, and “milky”. These descriptors indicate a negative evolution of sensory quality of the product over time. They can be related to a previously described increase in volatile organic compounds such as ethyl acetate, acetic acid, ethanol or palmitic acid during storage of pineapple cuts [
19,
26]. Significant differences were observed between freshly prepared and (class 4) 14-day stored samples. This accounts for “fresh”, “pineapple”, “pungent”, and “chemical” descriptors. Samples from day 3 and 7, respectively in classes 2 and 3, were mainly characterized by “acid”, “fermented” and “chemical”. PCA analysis showed clearly the differences depending on the storage time of samples determined from olfactive descriptors (
Figure 3c).
Depending on the batch, quicker spoilage could be observed (data not shown) and thus analyses were stopped when a spoilage was observed. For instance, batches TP1, TP2 and TP3, which came from the same place, were not acceptable after 7 days of storage. Batches BP, CP01, CP02, CP03, CP1, V0, V1, V2, and V3 were not acceptable after 14 days of storage. The common feature of the latter batches is that they were all sampled during the summer season. All winter batches, except TP3, were considered as acceptable after 14 days of storage.
Changes in physicochemical parameters were determined over the shelf-life of minimally-processed pineapple (
Figure 4 and
Figure S1). The values for pH, TA and TSS did not significantly vary according to storage time. On the opposite side, storage time influenced firmness and color parameters. A decrease of 26% of firmness was observed after 14 days, when compared to the determination performed the day of processing. The L* value was not modified according to storage time, but a* and b* decreased. The calculated dependent parameter C* decreased over storage time, but h° and ΔE did not vary significantly.
Some physicochemical parameter changes of minimally-processed pineapple during cold storage have been described [
5,
7,
8,
10,
12,
27,
28,
29,
30]. In most of these studies, pH appeared stable or varying by less than 0.2 units over storage time. For TSS, conflicting results are reported, decreasing [
28], stable [
27] or increasing [
10] during storage. A gradual decrease in firmness was observed previously [
5,
10,
30]. Changes of color, especially browning, have been described during the shelf-life of minimally-processed pineapple. A a* value increase was observed during the first 8 days of storage in several studies [
10,
11,
29], whereas a sharp L* and b* values decrease was observed in other studies [
7,
10,
12,
30]. The most reproducible changes are thus pH and TA stability, firmness decrease and b* decrease, indicating a loss of the yellow color because of either browning or translucency.
Counts of psychrotrophic bacteria did not increase during storage time, with mean values of 3.6 to 3.9 log CFU/g at each sampling time (
Table 4). This observation hides great differences between samples, most of samples exhibiting counts below the detection limit and some of them showing counts up to 6.9 log CFU/g after 10 days of storage. Consequently, a moderate increase of ca. 1.5 log CFU/g would not have been detected as significant in our experimental conditions. It was showed that psychrotrophic bacteria counts increased by less than 2 log CFU/g during 12 to 14 days of storage of pineapple cuts [
7,
11]. Counts of enterobacteria remained stable at 3.9 log CFU/g during the first 7 days and increased thereafter to reach 4.5 log CFU/g after 14 days of storage (
Table 4). The maximal observed value of enterobacteria counts gradually increased during storage. Lastly, yeast and mold counts gradually increased during storage time, from initially 4.4 log CFU/g to 5.1 log CFU/g after 7 days and 6.0 log CFU/g after 14 days. Both minimal and maximal yeast and molds counts changed during storage time to reach respectively 4.0 log CFU/g and 7.6 log CFU/g after 10 days, and 5.0 log CFU/g and 7.9 log CFU/g after 14 days (
Table 4). Significant differences depending on storage duration were observed for yeast and mold populations.
Fungal population was compared after 3 days of storage according to the visual spoilage observed. Batches TP1, TP2 and TP3, which appeared spoiled after 7 days of storage, exhibited a fungal population after 3 days of 5.6 log CFU/g. Batches spoiled after 14 days of storage (BP, CP01, CP02, CP03, CP1, V0, V1, V2, and V3) showed fungal counts of 4.8 log CFU/g after 3 days. The last group of batches, not spoiled after 14 days of storage, exhibited a mean fungal population after 3 days of 4.8 log CFU/g, and thus were identical to the latter group. Yeast and mold counts cannot be strictly correlated to shelf-life.
Mesophilic and psychrotrophic bacteria are possibly involved in minimally-processed pineapple spoilage, whereas enterobacteria enumerations were used as hygienic indicators of processing. In our study, the abundance of the two bacterial groups was monitored and did not increase significantly. The growth of mesophilic and psychrotrophic bacteria in minimally-processed pineapple has been reported, but consistently to a lesser extent than yeasts and molds [
7,
10,
27,
28]. A large and rapid increase of yeast and mold counts has been previously observed during cold storage of fresh-cut pineapple [
7,
10,
27,
29], confirming our observation. Yeasts and molds are reported as the main contaminant of fruit salads and fruit juices [
31,
32]. Yeasts and molds are favored by the high sugar content and the pH values, comprising between 3.09 and 4.20 for all batches, of minimally-processed pineapple. They can be responsible for spoilage by producing gas, ethanol and volatile compounds with off-odors.
Moreover, we showed that fungal counts cannot be solely used as a spoilage indicator, as they are not strictly correlated to shelf-life. For that reason, yeast and mold diversity and the relationship to spoilage was focused on.
3.3. Diversity of Yeasts and Molds and Modulation during Storage
The profile of yeast and mold communities during refrigerated storage of minimally processed pineapple was determined to see if differences between samples could be observed. To that aim, PCR-DGGE analysis was applied to eight batches (
Figure 5): CP1 sampled in East area in summer, VSA sampled in East area in winter, CP03 sampled in the North in summer, V1 sampled in the South area in summer, V4 and V6 sampled in the South area in winter, and TP1 and TP2 sampled from a local producer in summer. V1, V4 and V6 originated from the same producer.
We observed that samples were primarily gathered according to the pineapple batch, rather than according to storage time. Three groups were differentiated.
The first group contained TP1 and TP2 batches that presented only four bands per lane. For these two batches, a rapid spoilage occurred and both enterobacteria and yeast and mold counts were above 5.5 log CFU/g after 3 days of storage.
The second group gathered batches V4, CP1 and V6. For those batches, most of the bands were observed at the top of the gel. An evolution of the main fungal communities was observed during storage, with the appearance (black arrow) or disappearance (empty square) of DNA bands at the latter storage times (days 10 and 14).
Lastly, the third group gathered samples CP03, VSA and V1. Their profiles were similar for five bands, which were labelled “a”, “b”, “c”, “d”, and “g”. The “d” band disappeared from V1 profile after 14 days (
Figure 5). DNA retrieved from these bands and sequenced resulted in the identification of (a)
Resinicium saccharicola (mold), (b)
Cladosporium sphaerospermum (mold), (c)
Cladosporium cladosporioides (mold), (d)
Disporotrichum dimorphosporum (mold), and (g)
Rhizopycnis vagum (mold) respectively (
Table 5). The band labelled “h” was only present in V1 sample after 14 days of storage and identified as belonging to the
Rhodotorula glutinis species (pink yeast) (
Figure 5 and
Table 5). In VSA and V1 groups, two specific bands were labelled “e” and “f”, and were identified as
Galactomyces candidum (mold) and
Clavispora lusitaniae (yeast), respectively.
Except for TP1 and TP2 coming from the same producer and exhibiting a short shelf-life, PCR-DGGE grouping was not related to sampling season, neither to the location or producer, as it can be specifically seen from V1, V4 and V6.
R. saccharicola was isolated from sugar cane [
33].
Cladosporium spp. is widely present on plant material and can cause post-harvest spoilage of fruit [
34,
35,
36,
37,
38], even at a low temperature.
D. dimorphosporum is industrially used as a producer of plant cell wall lytic enzymes [
39,
40].
R. vagum, infecting roots and tubers, is known to contribute to vine decline and root necrosis [
41,
42,
43].
G. candidus, teleomorph of
Geotrichum candidum, is mainly derived from cheese [
44], but also from fruit tree phyllosphere [
45]. It was identified from necrotic lesions of pineapple [
46].
C. lusitaniae was identified in apple juice and on rotten fruit [
47,
48,
49].
Rhodotorula spp., corresponding to the “h” band that appeared in the V1 sample after 14 days of storage, contaminates at high levels fruit salads and juices made from cantaloupe, citrus, honeydew, strawberry, coconut water, grape, and apple [
31,
47,
48,
50]. All detected fungal species of this study have already been associated to different fruit or plant materials.
3.4. Identification of Fungal Isolates Involved in Spoilage of Minimally-Processed Pineapple
Nine fungal isolates were obtained from six minimally-processed pineapple batches at different storage times. Identification is proposed in
Table 2. “R” and “S” colony phenotypes were similar and identified as
Penicillium citrinum mold species.
Talaromyces amestolkiae was another isolated mold.
Rhodotorula mucilaginosa, Saccharomyces cerevisiae, and
Meyerozyma caribbica corresponded to yeast species isolated.
Rhodotorula was the only genus also identified from PCR-DGGE profiles.
P. citrinum has been identified from jujube (
Ziziphus mauritiana), acid food products from citrus, coco milk, coffee and cocoa beans, in which its strong polygalacturonase activity was detected [
38,
51,
52,
53,
54]. This fungus can produce the mycotoxin citrinin.
Talaromyces spp. can occasionally be isolated from low-pH juices [
55,
56]. On the opposite,
R. mucilaginosa and
S. cerevisiae are commonly involved in the spoilage of fruit products [
48,
51,
57,
58].
M. caribbica is mostly known as an endophyte yeast, but has been isolated from spoiled minimally-processed pineapple [
26,
59].
Five isolates were selected from DNA identification and colony morphology: P. citrinum (R), T. amestolkiae (20), R. mucilaginosa (A), S. cerevisiae (C), and M. caribbica (F). Their involvement in spoilage of minimally-processed pineapple during refrigerated storage was investigated. Pineapple cuts, dipped in the fungal cocktail composed of the five isolates, were analyzed and compared to control cuts (not dipped) after 7 days of storage, in triplicate.
Visually, treated cuts appeared darker (
Figure 6). Sensory triangle tests confirmed the difference, with a 99.99% confidence. The panel proposed descriptors of the treated samples: These descriptors were mainly negative adjectives, “fermented” and “putrid” being the most frequent. The descriptors “milky”, “Roquefort cheese”, then “sugared” and “toffee” were less frequently cited.
The comparison of physicochemical parameters pointed to large differences between the control and the treated samples after 7 days of storage (
Table 6). As expected, pH and TSS did not significantly change. On the opposite, TA decreased during cold-storage of control and increased for the treated samples after 7 days of storage. Color analysis showed that a* and Hue angle were significantly higher for the treated samples than for the control conditions. For L*, b*, C*, and ΔE, a bilateral Dunnet’s test pointed out differences between the treated samples and the control ones after 7 days of storage.
As expected, the fungal population increased for the control condition during storage, and the increase was much more marked for the treated sample (
Table 6). Colony phenotypes of the five isolates were observed on enumeration media for the 7-day stored treated samples.