Survey and sampling of select dried fruits

Peanut (Arachis hypogaea), pecan (Carya illinoinensis), walnut (Juglans regia), and squash seed (Cucurbita argyrosperma) samples were purchased in municipal produce markets of cities and towns across different regions of Mexico, where dried fruits can be purchased loose. The different regions were defined as Northwest, Northeast, West, Center, and Southeast based on climate and land use as reported by the National Institute of Statistics and Geography (INEGI), summarized in Table 1. Samples consisted of at least 250 g of the selected dried fruit dehulled, unsalted, and unroasted. A total of 111 samples (36 peanut samples, 30 pecan samples, 12 walnut samples, and 33 squash seed samples) were obtained and transferred to the Laboratory of Food Microbiology of the Universidad de las Americas Puebla for further analysis. Details about the samples can also be seen in Table 1.

Physicochemical characteristics of dried fruits

For the physicochemical analysis of dried fruits, approximately 10 g of each sample were manually ground using a ceramic mortar and pestle. Ground samples were immediately transferred to sealable plastic cuvettes for water activity analysis to prevent as much exposure to the environment as possible. Both moisture content and water activity determinations were carried out within a day of the sample having been received.

Moisture content determination

For moisture content determination, 2 g of ground samples were precisely weighed on constant-weight aluminum trays with a bed of silica sand using an analytical scale. The initial weight was recorded, and the samples were transferred to a 130 ° C oven. Samples were dried until constant weight was achieved (~24 h.). Samples were then removed from the oven and allowed to cool at ambient temperature (~25 °C) in a desiccator with silica gel until temperature equilibrium was achieved (15-20 min). Afterwards, dry weight was measured using an analytical scale and moisture content was calculated as the ratio of the difference between sample weight and dry weight to sample weight.

Water activity determination

For water activity (a_w) determination, enough ground sample was placed in sealable plastic cuvettes to cover the bottom of the container. Water activity was then measured in a 4TEV dew-point water activity meter (AquaLab, USA) at 25 °C. Assays were carried out in triplicates.

Yeast and mold count

Total yeast and mold counts were determined on dried fruit samples. 10 g of each sample were weighed and placed in sterile sampling bags. Samples were then diluted with 90 mL of 0.1 % peptone (Becton-Dickinson, Mexico) solution and homogenized using a Stomacher ® laboratory blender (Seward, United Kingdom.). Serial dilutions of the sample were prepared using the same 0.1 % peptone solution and plated in potato dextrose agar (PDA) (Becton-Dickinson, Mexico) plates acidified with 1.6 mL/100 mL of a 10 % tartaric acid solution to a final pH of 3.5. Colonies in inoculated plates were counted after being incubated at 25 °C over a period of 5 days. For samples in which Rhizopus species were abundant, Dicloran Chloramphenicol Rose Bengal agar (DRBC) (Becton-Dickinson, USA) was used instead of PDA to limit colony growth and allow for a more accurate quantification of yeasts and molds. Assays were carried out in triplicates.

Fungal isolation and identification

Individual mold colonies were taken from counting plates and inoculated in PDA plates, without tartaric acid, for isolation. Plates were incubated at 25 °C for 7 days, or 3 days when a Rhizopus colony was apparent, and inspected afterwards for contamination. Resampling and inoculation of the mold colonies was repeatedly carried out until only a single mold species was apparent in the agar plate. Afterwards, molds were point-inoculated in malt extract agar (MEA) (Becton-Dickinson, Mexico), Czapek Yeast Extract agar (CYA) (Becton-Dickinson, Mexico), Czapek Dox agar (CZD) (Becton-Dickinson, Mexico), and G25N agar, the latter of which was prepared according to the formula given by ^{Pitt and Hocking (2009)}. For identification, molds were incubated at 5, 25 and 36 °C for 7 days, or 3 days when a Rhizopus spp. colony was readily apparent.

After incubation of the fungal isolates, macro and microscopic colony morphology was observed and identification was carried out following dichotomous keys as outlined by ^{Pitt and Hocking (2009)} and ^{Samson et al. (1995)}.

For microscopic observation and measurement, a sample was taken from the outer edges of the colony using a sterile dissection needle, cutting a small (16 mm², approx.) and shallow (< 1 mm deep) square of media along with the desired colony portion. The colony sample was then placed in a microscope slide and a drop of aniline blue solution (0.1 % aniline blue; Química Meyer, Mexico) in 85 % lactic acid (Química Meyer, Mexico) was added, as mounting fluid. Afterwards, a drop of 70 % ethanol was added. A slide cover was then placed on top of the sample, and the sample was carefully heated with a Meker-Fisher burner to melt the culture medium. The sample was then observed under 100x magnification in an optical microscope (American Optical, USA.) equipped with an Axiocam ERc 5S camera (Zeiss Microscopy, Germany.) and associated software for microphotography and measurements.

Identification of Penicillium subgenus

Incidence of individual genera or species of mold

Incidence of individual genera or species of mold was calculated as the fraction of samples contaminated with a particular species or genus of mold, using the following equation:

Where I _s is the incidence (%) of a genus or species of mold in a subset of samples; n _c is the number of samples from which the genus or species in question was isolated at least once within a subset of samples, and n _t is the total number of samples in that subset. Samples were categorized by region and seed type and incidence was expressed thusly. A sample was considered contaminated when the genus or species in question was isolated and identified from the sample.

Data analysis

Principal component analysis (PCA)

To illustrate the relationships between variables, a principal component analysis (PCA) was carried out between the system variables of a_w, % H, yeast and mold count, and number of different identified species. Grouping was decided based on categorical variables (dried fruit type and region). The cophenetic correlation coefficient (CCC) was computed to measure the accuracy of the PCA, where a value closer to 1.0 corresponds to a more faithful preservation of correlations between variables. For these statistical analyses, XLSTAT 2019 addon for Microsoft Excel was used (Addinsoft LLC, Paris, France).

Physicochemical analysis of dried fruits

Both water activity and moisture content of dried fruit samples was analyzed. Equal variance tests showed significantly different variance (p<0.05) for several groups of samples. Variance was found to be significantly different for water activity, moisture content, and yeast and mold counts for all sample groupings except for walnuts, where no significant difference was found for variances among regions. For this reason, ANOVAs were run without assuming equal variances among samples and the Games-Howell post-ANOVA pairwise comparison method was used. The results of these analysis (Figures 1 and 2) show most dried fruits exhibited water activity levels below 0.6 and moisture content below 6 %. Only squash seeds from the South-East region showed moisture content higher than 6 %, while pecans from the same region showed water activity close to 0.6 (0.597±0.004). Peanuts had moisture levels ranging from 4.17±0.02 % to 2.40±0.80 %, which is considerably lower than the 8.79±0.07 % moisture content found by ^{Ghorbani and Hosseini (2017)}, but closer to the moisture content reported by ^{Liu et al. (2018)} of 3.65±0.05 to 4.05±0.09. On the other hand, water activity levels of peanuts are within the range considered by ^{Ghorbani and Hosseini (2017)} as ideal quality for peanut kernels (from 0.33 to 0.43 at 25 °C) with only peanuts from the north-east and north-west regions showing water activity values outside that range.

Figure 1 Moisture content of studied dried fruits from selected regions of Mexico. Different letters indicate significant (p<0.05) differences amongst dried fruits from different regions. 

Figure 2 Water activity of studied dried fruits from selected regions of Mexico. Different letters indicate significant (p<0.05) differences amongst dried fruits from different regions. 

Peanut samples from different regions showed significant differences (p<0.05) in water activity and moisture content. Peanuts from the Northeast and Northwest regions had significantly higher water activity than peanuts from the Western region. Peanuts from the Northwest also had significantly higher water activity than peanuts from the West and Center regions. This probably indicates a difference in composition in the peanuts, as water activity is not only affected by moisture content, but also by content of low-molecular-weight soluble solids, as well as other components. It is known that moisture content and water activity of peanuts can also vary depending on the state of maturity of the seed at the time of harvest ^{(Dorner 2008)}. It is unlikely that the water activity of the peanuts was influenced by storage conditions or local climate, as peanuts from the arid Northwest region had significantly higher water activity that peanuts from other regions, as well as no significant difference with peanuts from the more tropical Southeast region.

Pecan samples also showed significant differences in both water activity and moisture content, with nuts from the Southeast region having a significantly higher water activity (p<0.05) than samples from all other regions, except from those from the Northeast region. Nuts from the Southeast region were also close to the threshold for growth and development of mold spores (a_w > 0.6) which would make them vulnerable to spoilage and/or aflatoxin generation. Water activity levels were considerably lower than those reported by ^{Valle-Garcia et al. (2019)}, who found all pecan samples tested above the 0.6 threshold and up to 0.9. Moisture content levels appeared lower than those reported by ^{Reis Ribeiro et al. (2020)}, who found moisture levels ranging from 4.56 % to 5.50 %. In the case of pecans, regional climate appeared to have an influence on their water activity, as only nuts from the wetter Southeast region had water activity close to the mold growth threshold.

Walnuts from different regions showed significant differences (p<0.05) in both water activity and moisture content, with walnuts from the West region showing a significantly lower value than from the other two studied regions. Walnuts from this region also show a much smaller water activity value than what was reported by ^{Schmidt and Fontana (2019)} of 0.570.58. Walnuts from the other regions presented values closer to what was reported, yet still considerably smaller. Moisture content of all samples is within the range reported by ^{Peng et al. (2021)}, who found moisture content for walnuts ranged between 1.20 to 9.92 %. The water activity values found in this study for walnuts show a remarkable resistance to ambient conditions, as it has been reported that walnuts are highly hygroscopic and will readily absorb moisture at relative humidity levels as low as 1.5 % ^{(Boaghi et al. 2019)}.

Squash seeds showed significant differences (p<0.05) in water activity only between the Southeast region and the West and Norwest regions. On the other hand, most regions showed a significant difference in moisture content, with seeds from the West region having the highest moisture content, at 6.10±0.03 %. ^{Ardabili et al. (2011)} reported a moisture content for Cucurbita pepo seeds of 5.20±0.28 %, which is similar to some of the values found in this study. Cucurbita argyrosperma seeds remain poorly studied as a food source. To the authors’ knowledge, no study has reported their moisture content or water activity levels before, though they can be assumed to be similar to those of the seeds of other species in the Cucurbita genus.

Water activity value of samples was comfortably below the optimum water activity levels for aflatoxin production of 0.96, and below minimum water activity for aflatoxin production of 0.85 ^{(Liu et al. 2017)}. This alone is not enough to rule out the possibility of aflatoxin contamination, as cross-contamination is still a possibility. However, it is an indication that storage conditions were adequate to prevent the development of mold spores and the endogenous production of aflatoxins.

Microbiological analysis of dried fruits

Total yeast and mold count

Several dried fruit samples exceeded the permitted number of total molds and yeasts as outlined in Mexican official norm NOM-247-SSA1-2008, which lists a maximum of 300 UFC/g of yeasts and molds in edible seeds and their derivatives. There was great variability in the yeast and mold counts of individual dried fruit samples. Figure 3 shows the distribution of yeast and mold counts in dried fruit samples. It is notable that no peanut or walnut samples were above the maximum allowable limit; however, one pecan sample from the center region and three squash seed samples (one from center, one from west, and one from northeast) were, with a squash seed sample having the highest registered yeast and mold count of all samples at 4200 CFU/g. In some sample groupings, significant correlation (p<0.05) between either water activity or moisture content and yeast and mold count was found. Table 2 shows the correlation coefficients and p values for the correlation between yeast and mold counts and both moisture content and water activity. In the case of peanuts, yeast and mold count was found to be significantly positively correlated to moisture content, whereas correlation was not significant to water activity. The opposite was true for pecan, where a significant positive correlation was found between yeast and mold count and water activity, but not moisture content. In the case of walnut, a significant negative correlation was found between water activity and yeast and mold count, while no significant correlation between either parameter was found in the case of squash seed. In general, a positive correlation is expected between water activity and/or moisture content and yeast and mold counts; however, this might not always be the case since fungal development is limited at lower water activity and/or moisture content values. Lack of correlation between a_w/moisture content and yeast and mold counts would appear to indicate no endogenous mold growth was possible in dried fruit samples, and thus that the yeasts and molds detected and isolated come primarily from exogenous contamination.

Figure 3 Box plot of the distribution of yeast and mold content in the four studied dried fruits. 

Table 2 Correlation coefficients (r) and p values for yeast and mold counts versus water activity (aw) and moisture content (% H) for studied dried fruits 

Dried fruit	aw		%H
	r	p	r	p
Peanut	0.128	0.478	0.620	0.000
Pecan	0.546	0.003	0.403	0.037
Walnut	-0.975	0.000	-0.527	0.145
Squash seed	-0.975	0.986	-0.070	0.713

Despite conditions and practices that could be perceived as “unsanitary” in Mexican plantations, produce markets, and distribution centers ^{(García-Gómez et al. 2002}, ^{Cedillo Camarena 2013)}, seeds and nuts are generally in good microbiological condition, in respect to yeast and mold content, in accordance with national regulations. The low water activity and moisture content values found also indicate that the seeds were properly stored prior to sale, as even in regions with a tropical climate (center and southeast), seeds did not show a_w values that could be considered risky for mold development and all samples were below 7 % moisture content, which is considered optimal for long-term storage of dried fruits ^{(Norlia et al. 2019)}.

Fungal isolation and identification

From the many different genera and species of mold isolated and identified from dried fruit samples (Table 3), several spoilage and toxigenic molds stand out. Aspergillus species (including A. flavus) were found throughout the different samples. Penicillium and Rhizopus species were also ubiquitous. These three genera are important since many of the species contained in them are either responsible for spoilage of seeds and grains or capable of producing toxic metabolites. The presence of A. flavus, Fusarium spp., and Penicillium spp. also stand out as known producers of mycotoxins. Alternaria spp. was also found on some samples, as were other molds associated with field contamination, such as Phoma spp. and Cladosporium spp. that otherwise are not known to deteriorate dried fruits or produce mycotoxins.

Incidence of different genera and species of mold

A high incidence of relevant toxigenic and spoilage mold was found across all four tested dried fruits. A. flavus incidence was highest in squash seed, where around 80 % of the seeds tested showed contamination with this mold (Figure 4). In contrast, only 30 % of peanuts tested showed contamination with A. flavus. Peanut samples, in turn, showed a high proportion of contamination with A. niger, a mold not usually associated with mycotoxin production, but capable of spoiling dried fruits. Pecan nuts showed a similar incidence of contamination with A. flavus and A. niger. These two species of mold are known to be antagonistic to each other ^{(Aziz and Shahin 1997}, ^{Fendiyanto and Satrio 2020}, ^{Griffin et al. 2001)} and are therefore seldom found together in a sample. In this case, only six out of the 30 samples tested contained both mold species at the same time, while only five contained neither. Walnuts showed a similar trend, with a high incidence of A. niger, but no incidence of A. flavus.

Figure 4 Incidence of contamination with some important pathogenic and deteriorative molds in select dried fruits. 1: Aspergillus flavus. 2: Aspergillus niger. 3: Penicillium spp. 4: Rhizopus spp. 5: Fusarium spp. 6: Aspergillus spp. 7: Cladosporium spp. 

Incidence of species of Penicillium was above 50 % across all four kinds of dried fruits. Penicillium are a ubiquitous genus of molds, and so are expected to appear in high rates in dry foods. Some Penicillium species can produce mycotoxins such as ochratoxin, and therefore their presence is cause for concern ^{(Pitt and Hocking 2009)}.

Incidence of Rhizopus species, mostly represented by R. oryzae, was also above 50 % in all four of the tested dried fruits. R. oryzae is a common contaminant of grains in tropical locations and capable of spoiling improperly stored dried foods. It is, however, not known to produce toxins ^{(Pitt and Hocking 2009)}.

Mold species found in peanuts were similar to those reported by other authors: Aspergillus flavus has been reported in roasted peanuts from South Africa ^{(Adetunji et al. 2020)} and Bangladesh ^{(Khandaker et al. 2019a}, ^2019b), and in raw peanuts from Benin ^{(Adjou et al. 2012)}, Pakistan ^{(Abbas et al. 2019)}, and Turkey ^{(Gürses 2006)}. These same authors have also variously reported presence of A. niger, Fusarium spp., Penicillium spp., Alternaria alternata, Rhizopus spp., Cladosporium spp., Mucor spp., and other mold genera and species. In general, mycobiota of peanuts appears varied, though the most prevalent molds are from the Aspergillus genus, with a high incidence of A. flavus, and A. niger.

The mold species found in pecans were like those reported by ^{Valle-Garcia et al. (2019)} in pecans from Rio Grande do Sul State in Brazil, although the frequency at which the different species were found seems to differ. ^{Valle-García et al. (2019)} reported complete absence of A. flavus in the analyzed samples, which was not the case in the samples analyzed in this study. They also reported a predominance of Aspergillus species, which was also the case in the samples analyzed on this study. However, in these samples, Penicillium species were found at a higher frequency than in ^{Valle-García et al. (2019)} report. Finally, they also reported an incidence of Cladosporium spp. of 30 %, which was like the rate at which the genus was isolated from pecan samples in this study. Cladosporium is a known pathogen of pecan nuts that has been reported previously in Mexican pecan ^{(Walker et al. 2016)}.

On the other hand, mold species found in walnut samples are different than those reported in other studies, where Aspergillus species, such as A. flavus, A. terreus, A. niger, A. fumigatus and A. versicolor have been shown to predominate ^{(Abbas et al. 2019)}, although Penicillium species have also been reported in walnut previously ^{(Gürses 2006)}. Contamination with A. flavus has been demonstrated in Mexican walnuts before, although aflatoxin content was reportedly low ^{(Adaya-González et al. 2015)}.

As with many aspects of Cucurbita argyrosperma, the mycobiota of the seeds remains poorly studied. Phytopathogenic molds such as Phytophthora capsici, Rhizoctonia solani, and Sclerotium rolfsii have been reported to cause fruit rot ^{(Díaz-Nájera et al. 2015}, ²⁰¹⁸⁾; however, no studies have been made over the mycological ecology of the seeds of this gourd. It can be assumed that mycobiota of C. argyrosperma seeds is similar to the mycobiota of seeds of other cucurbits, such as that of C. pepo. In that regard, ^{Rahim et al. (2013)} have reported the presence of at least 100 different species of mold in the seeds of C. pepo, among which are those found in the present study. As was the case with the other dried fruits analyzed, the genus Aspergillus appeared with a high incidence, and the degree of contamination with Aspergillus flavus is of particular concern, being present in over 70 % of the samples tested, and on a higher degree than on peanuts and pecans.

Despite high incidence of contamination with molds, most of the samples tested were within the allowable limits for yeast and mold counts. This would seem to indicate that Mexican dried fruits are safe for consumers despite contamination with molds. However, care must be taken when transporting and storing dried fruits, since mold spores can develop and increase in number in the right conditions. Data is pertaining mostly to dried fruits from the center region of Mexico, which encompasses the most highly urbanized states in the country and contains the Mexico City megalopolis. It is known that rural areas are more vulnerable to contamination with toxigenic molds than urban areas, and the results of this study reflect this ^{(Zuki-Orozco et al. 2018)}.

Regional variation in dried fruit mycobiota

Variation was observed in mycobiota of dried fruits across different regions, as is shown in Figure 5. It is notable that A. flavus, A. niger, and Fusarium spp. were much more prevalent in samples from the Northeast and Southeast regions, whereas other species of mold showed a more evenly distributed incidence across all regions. Regional variations in mycobiota have been reported in other countries and have been shown to be related to regional climate; in particular, to relative humidity and temperature ^{(Fatahinia et al. 2018}, ^{Jayashree and Wesely 2019}, ^{Temperini et al. 2019}, ^{Aveling et al. 2020)}. Locations and seasons with high relative humidity show increased fungal load and diversity when compared to drier environments ^{(Fatahihia et al. 2018)}. This influence can also be observed in the present study: the incidence of mold species in the Northwest region, the most arid of the five regions under study, consistently shows lower incidence of all mold species, except for species of Aspergillus other than A. flavus and A. niger.

Figure 5 Incidence of contamination with some important pathogenic and deteriorative molds in studied dried fruits from selected regions of Mexico. 1: Aspergillus flavus. 2: Aspergillus niger. 3: Penicillium spp. 4: Rhizopus spp. 5: Fusarium spp. 6: Aspergillus spp. 7: Cladosporium spp. 

Principal component analysis

Figure 6 shows the scatter plot of the main three principal components: F₁, F₂ and F₃. The obtained cophenetic correlation coefficient was 0.98, indicating that the three main principal components represent an accurate reduction in the number of variables for this analysis. Together, F₁, F₂ and F₃ account for 84.4 % of variability, with each individual component accounting for 38.6, 28.6, and 17.2 % of variability, respectively.

Figure 6 Scatter plot for observations on principal component axes F₁, F₂, and F₃. Shapes indicate region as follows: center (sphere), northeast (cube), northwest (cylinder), southeast (octahedron), and west (icosahedron). 

From the results of the PCA, it is clear that seed type is a stronger predictor for characteristics than region. This is to be expected, as there is bound to be more similarity between the same dried fruits across different regions, as opposed to different dried fruits from the same region. Squash seeds and pecan nuts show distinct clustering in separate areas of the [F₁, F₂, F₃] space, along the F₂ axis and the [F₁, F₂] plane. This is due to the correlation of F₁ and F₂ with yeast and mold content (Table 4), where squash seeds have the highest and most widely distributed values of all tested dried fruits.