Fermented beverages constitute a major part of the diets of traditional African homes (Tafere, 2015). In Nigeria, traditional beverages are widely consumed and mostly preferred to commercial soft drinks by individuals from low income settings due to their relatively low cost of production and high nutritional benefits (Ezekiel et al., 2015).

Kunu is a traditional beverage produced principally from single or mixed cereals such as maize, millet, rice, or sorghum. In some cases, peanut is added to the cereals to make-up the raw material input. The grain (cereals and nuts) input could also be supplemented with additives such as cloves, pepper, ginger, sweet potato and tiger nut, which are added as homogenized mixtures just before the fermentation of the beverage. The sets of cereal and nut applied to the production of kunu determine its variety. For example, kunu-zaki comprises of millet, sorghum or maize; kunu-tsamiya, millet, sorghum or rice; kunu-gyada, rice, peanut, millet or sorghum; and kunu-gayamba, solely millet (Gaffa et al., 2002). Kunu-zaki is the commonest of the kunu varieties due to its nutritional and health benefits (Adelekan et al., 2013). Generally, kunu is consumed whilst in an active state of fermentation by both adults and children (Efiuvwevwere and Akona, 1995). The nutritional content of kunu includes 9.84–12% carbohydrate, 1.56–3% protein, 0.1–0.3% fat, and 0.61–075% dietary minerals (Adeyemi and Umar, 1994; Badifu et al., 1999), and its health benefits range from purging the bowels and relief of flatulent conditions (Omakwu, 1980), to the enhancement of lactation in nursing mothers (Efiuvwevwere and Akona, 1995).

The production process of kunu comprises six critical steps: steeping of the cereals in water to allow for softening and fermentation by autochthonous bacteria, wet milling, gelatinization of a large portion of milled grains by addition of boiling water, addition of a mix of milled additives and the remainder portion of the milled grains to the gelatinized gruel, fermentation of the mixture, and sieving of the fermented slurry to obtain kunu (Gaffa and Ayo, 2002). The pH of kunu is usually acidic (pH 3–5.46) (Efiuvwevwere and Akona, 1995; Gaffa et al., 2002; Adelekan et al., 2013). The steeping duration varies and is largely dependent on the type of cereal used (Gaffa et al., 2002). Similarly, the duration of the fermentation step of the gelatinized gruel mix varies from 8 to 24 h depending on the complexity of the food matrices used for kunu processing and the proportion of mixture of milled grains and additives to the gelatinized gruel (Gaffa and Ayo, 2002; Osuntogun and Aboaba, 2004; Oluwajoba et al., 2013; Olosunde et al., 2015).

The fermentation stages of kunu-zaki are driven by consortia of bacteria (mostly lactic acid bacteria) (Efiuvwevwere and Akona, 1995; Gaffa and Gaffa, 2004; Osuntogun and Aboaba, 2004; Oguntoyinbo et al., 2011; Ikpoh et al., 2013; Aboh and Oladosu, 2014) and a few yeasts (notably Saccharomyces cerevisiae) (Efiuvwevwere and Akona, 1995; Gaffa and Gaffa, 2004) that contribute to the breakdown of complex macromolecules into simpler compounds. Although, there is a recent culture-independent (sanger-based sequencing technology) microbiological study of a kunu variety (Oguntoyinbo et al., 2011), most of the previous microbiological studies on kunu are based on conventional isolation methods and classical identification techniques which are prone to biases, low taxonomic resolution, misidentification of species and underestimation of species richness and diversity (Cocolin and Ercolini, 2015; Ezeokoli et al., 2016). Currently, there are no high-throughput sequencing (HTS)-based studies on kunu microbial ecology. The application of HTS technologies (also referred to as next generation sequencing) may help unravel hitherto unidentified bacterial species associated with kunu processing and kunu products (Franzosa et al., 2015; Ezeokoli et al., 2018). In addition, there is a paucity of information on the microbial diversity of different formulations of kunu at different stages of processing. Consequently, an in-depth understanding of the microbiology of kunu will provide insight into the community structure and functional roles of microbes in the production of varieties of this beverage. Furthermore, the knowledge of kunu microbial community will facilitate the selection of starter cultures for improvement of the safety and quality of this widely consumed beverage.

Chemical food contaminants (e.g., mycotoxins) may, however, distort the safety and quality of kunu due to the use of diverse cereals and nuts that have been reported to be prone to several mycotoxins in the beverage formulation (Adetunji et al., 2014; Afolabi et al., 2015; Ezekiel et al., 2015, 2018; Oyedele et al., 2017). Thus, there is a need to evaluate the extent to which mycotoxins can be carried over into various kunu formulations, considering the diverse grain inputs into the production of this beverage. Previous studies have shown that a few mycotoxins can be present in kunu-zaki, albeit at reduced levels (Ezekiel et al., 2015; Olosunde et al., 2015). However, the influence of specific processing steps on the levels of mycotoxins during the processing of one or many formulations of kunu is yet to be elucidated. Such data coupled with information on the microbial community structure is highly relevant to food safety and protection of consumer health in view of the adverse health effects that may arise from dietary mycotoxin exposures (International Agency for Research on Cancer [IARC], 2015). Furthermore, it has been postulated that traditional beverages can contribute to increased mycotoxin exposures in high cereal -dependent regions such as sub-Saharan Africa (SSA) (Ezekiel et al., 2018). Thus, there is a need to determine the processing steps and grain combinations that are critical to the reduction of mycotoxin exposure through beverage consumption. This information may be useful for recommending safe kunu formulation(s) for consumer benefits.

Therefore, this study aimed to determine the bacterial community diversity and dynamics during the processing of three kunu formulations by using HTS-based technology, and to evaluate the effect of processing on the mycotoxin profiles at various stages of processing by using liquid chromatography tandem mass spectrometry (LC-MS/MS).

This study reports, for the first time, on the application of HTS in the delineation of bacterial community structure during processing of kunu vis-a-vis changes in mycotoxin concentrations. Overall, this study clearly indicates the core bacterial communities and succession during the various stages of processing of three formulations of kunu, and the possible role of the microbial communities in reducing the mycotoxin levels (from ingredients to the final kunu product) irrespective of formulations.

The high number of OTUs (species richness) obtained during the steeping stage of the various kunu formulations suggests a rich genetic diversity of bacteria in the grains. Although potable tap water used for the steeping was not evaluated independently for microbial diversity, it is also likely that potable water contributed to the species diversity at the steeping stage. Furthermore, this observation of high species richness may be due to the stepping environment and duration serving as a broth and permitting the proliferation, respectively, of the autochthonous microbial species. Similar observations of diverse bacteria occurring during the steeping of raw grains for kunu-zaki production were suggested in previous studies (Efiuvwevwere and Akona, 1995; Gaffa and Gaffa, 2004). The high number of OTUs in formulations A and C, and shared OTUs between these two kunu formulations in contrast to the relatively lower number of OTUs obtained for the formulation B could be attributed to the composition of ingredients used for the various formulations: only formulations A and C contained tiger nut. This observation suggests that tiger nut may serve as the source of the additional OTUs in formulations A and C. Tiger nut grows underground and has been reported to harbor a plethora of microbiota consequent on its direct interaction (not enclosed in pods) with the soil environment (Ayeh-Kumi et al., 2014; Gambo and Da’u, 2014; Ike et al., 2017).

Herein, we put forward some hypotheses to explain the observed changes in OTU richness between successive stages of kunu processing. The lower OTU richness observed in the milled grains in comparison to OTU richness in steep liquor could be attributed to the fact that during the overnight steeping, bacterial species associated with the grains and steep water proliferate, giving rise to a fermented broth—the steep liquor (usually with bubbles or foams due to CO₂ release observed). Since this fermented broth (or steep liquor) are decanted and discarded before subsequent washing of the steeped grains, a large number of species present in the steep liquor are not carried over to the milling stage. Furthermore, since the steep liquor is not used in subsequent processing step, the OTUs detected in the milled grains are likely those that are within the grains, OTUs not completely eliminated during washing of the steeped grains, and/or any OTUs introduced by chance contamination during milling of the grains. Hence, the milled grain is expected to contain significantly lower species than the steep liquor as observed in all formulations. Between the milled grains and onset of fermentation, a reduction in OTUs was observed. This relatively low OTU diversity observed at the 0 h fermentation time compared to the pre-gelatinization stage in all the kunu formulations is attributable to the hot water (100°C) treatment during the preceding gelatinization of the milled grains. Certainly, such heat treatment will have at least a bacteriostatic (if not significantly bactericidal) effect on most of the indigenous microbiota, consequently contributing to reduced species diversity (Russell, 2003). In a study by Akharaiyi and Omoya (2008), following heat treatment of fermented maize (ogi) at 100°C, the total viable count of microorganisms drastically reduced to minimal levels. Despite expected increase in species after the addition of the un-heated 1/3 portion, the restoration of species richness to the pre-gelatinization state may have not been detected due to dilution effect thereby resulting in a lower chance of species (or their DNA) recovery during DNA extraction. Bias in the efficiency of recovery and detection of rare species is a well-known limitation of metagenomics approaches (Wooley and Ye, 2010; Breitwieser et al., 2017).

During the fermentation stage, an increase in OTUs was observed between the onset and first 6 h of fermentation. Kunu fermentation is a lactic acid fermentation type, with initial slightly acidic pH of about 6.0 (Gaffa and Gaffa, 2004), which gradually reduces to an acidic pH of about 4.76 (Gaffa and Gaffa, 2004) at the end of 12 h and then to 3.0 in the kunu product (Efiuvwevwere and Akona, 1995; Gaffa et al., 2002; Adelekan et al., 2013). Increase in acidity is a selective pressure on the microbial community during fermentation (Wolfe and Dutton, 2015; Zabat et al., 2018). We hypothesize that the observed increase in OTUs between the first 6 h was due to the moderate pH and other favorable conditions (e.g., temperature, oxygen availability, absence of growth inhibition metabolites) during this period, permitting the growth of species which were suppressed by the heat treatment in the gelatinization stage (Yang et al., 2016). However, as the pH further decreases (selection pressure) toward the end of fermentation (at 12 h), only adapted species (mostly lactic acid bacteria) proliferate (Gaffa and Gaffa, 2004). Hence, a reduction (compared to at 6 h) in OTU diversity at the end of fermentation was observed in formulations A and C. In contrast, an increase in OTU richness was observed between the 6 and 12 h of fermentation in formulation B. The reason for this observation may be due to the presence of sweet potato and the absence of tiger nut in formulation B compared to formulations A and C. Difference in substrate types influences the genetic and functional diversity and dynamics of bacteria during fermentation, and also play a role in the equilibrium of biochemical reactions and conditions (e.g., pH, oxygen availability, redox reactions, enzymatic activities, and temperature) during fermentation (Giraffa, 2004; Ijabadeniyi, 2007; Van der Meulen et al., 2007). Furthermore, the observed general reduction in OTUs after the sieving step may be attributed to the removal of a large number of species in the pomace; the pomace contains fibrous material which provides a larger surface area for the adhesion of microorganisms compared to the liquid kunu fraction.

In view of the afore discussions on OTU diversity, the trend in numbers of OTUs obtained from the mid-stream (6 h) fermentation stage up to the final product stage for the formulations suggests the proliferation of adapted (after heat treatment) functional species (key drivers of biochemical conversions of macromolecules in the fermenting substrate) and development of a climax/streamlined bacterial community observed in the kunu product. The associations observed between the bacterial community structure (as indicated by the ordination plots) of processing stages and/or different formulations suggest that, indeed, ingredients and processing steps influence microbial community composition and dynamics.

Taxonomically, the Firmicutes was the dominant phylum during most of the kunu processing stages and in the final product. Also, similar to observations of several culture-dependent microbiological studies on kunu (Efiuvwevwere and Akona, 1995; Gaffa and Gaffa, 2004; Osuntogun and Aboaba, 2004) and a culture-independent study of retailed kunu (Oguntoyinbo et al., 2011), lactic acid bacteria (LAB) were generally dominant during various processing stages of the three kunu formulations. With this finding, the dominant role of LAB during kunu production becomes more obvious and explanations for the acidic pH of kunu reported in earlier studies become clearer (Efiuvwevwere and Akona, 1995; Gaffa et al., 2002; Adelekan et al., 2013). In particular, Lactobacillus spp. was the most dominant of the LAB species in most of the fermentation stages of all three formulations and in kunu of most formulations. Similar observation for the dominance of Lactobacillus spp. in kunu-zaki was reported by Inyang and Dabot (1997) and Oguntoyinbo and Narbad (2012). The high occurrence of Lactobacillus in these formulations suggests that Lactobacillus spp. play a vital role during kunu processing and further adds to the growing scientific evidence on the predominance of this bacteria during processing of cereal-based fermented products such as ogi (Oguntoyinbo et al., 2011; Okeke et al., 2015), boza (Gotcheva et al., 2000), and bushera (Muyanja et al., 2003). In addition, Lactobacillus spp. are multifunctionally diverse, including many species with probiotic properties (Sanni et al., 2013; Oh et al., 2018) and physiological capabilities for the breakdown of complex polysaccharides in human and animal diets (Barrangou et al., 2006); thus, they could be a reason for some of the nutritional and health benefits generally associated with kunu consumption (Omakwu, 1980; Efiuvwevwere and Akona, 1995).

In contrast to the general assertions that the microbial community of kunu is dominated by Lactobacillus, Clostridium sensu stricto (cluster 1) dominated bacterial community of kunu product in formulation C. Although species of Clostridium have been previously reported in some kunu variety (e.g., kunu-zaki) (Oguntoyinbo et al., 2011), this is the first report associating the Clostridium sensu stricto (cluster 1) group with kunu as well as the dominance of the cluster. Until recently, the phylogenetic differentiation (and hence taxonomic classification) of some species within the Clostridium sensu stricto from other Clostridium species has been unclear, making their functional role in human diet and health, as well as in food fermentations unclear (Collins et al., 1994; Wiegel et al., 2006; Gupta and Gao, 2009; Kaur et al., 2014). A few species in the Clostridium sensu stricto cluster are being reclassified and functionally annotated (Wiegel et al., 2006; Gupta and Gao, 2009; Lopetuso et al., 2013; Yutin and Galperin, 2013). In fact, there is growing scientific evidence for their association with the human gut (Magne et al., 2006; Jacquot et al., 2011) and potential applications as hetero-fermenters in food fermentations and industry (Wiegel et al., 2006). Based on the findings of the study, kunu is a suitable source for the targeted isolation of species of the Clostridium sensu stricto (cluster 1) group. Subsequently, studies may explore the functional and metabolic traits of these species in kunu as well as their suitability as starter cultures for kunu fermentation or other cereal-based fermented beverages. Similarly, Acetobacter, Gluconobacter, Gluconacetobacter and Propionibacterium genera were present during kunu processing, especially in the steeped grains. To the best of our knowledge these species have not been previously reported in the microbiology of kunu processing. Acetobacter, Gluconobacter and Gluconacetobacter are acetic acid fermenters that have relevant industrial usefulness due to their ability to convert several sugars and alcohols into industrially important organic acids (e.g., vinegar) (Gupta et al., 2001; Sengun and Karabiyikli, 2011; Mamlouk and Gullo, 2013). Acetobacter have been identified in other cereal-based fermented foods such as burukutu, a fermented traditional sorghum-based beer (Oguntoyinbo, 2014), while Gluconobacter and Gluconacetobacter are associated with acidic beers (Bokulich and Bamforth, 2013; Mamlouk and Gullo, 2013; Spitaels et al., 2014a). Conversely, Propionibacterium is mainly associated with dairy and fermented dairy products (Moslemi et al., 2016), with some species being able to improve the probiotic properties of lactic acid bacteria when incorporated into the vegetables during the production (fermentation) of sauerkraut and other vegetable salads (Babuchowski et al., 1999).

The succession in specific taxa during the processing of kunu formulations elucidate functional roles of the dominant species at specific time points as well as their adaptation to the prevailing fermentation conditions. The observation of a streamlined and relatively dominant community of lactic acid bacteria, including species of Lactobacillus, Lactococcus, Leuconostoc, Pediococcus, Weissella and possibly species within the Clostridium sensu stricto clusters, toward the end of fermentation and in the final kunu product is related to their adaptation to the acidity pH of kunu fermentation which is expected as they are the drivers of the biochemical reactions and generators of lactic acid during the process (Gaffa and Gaffa, 2004; Amadou et al., 2011; Oguntoyinbo et al., 2011; Oguntoyinbo, 2014). As discussed earlier, these species are the drivers of the fermentation and are involved in the metabolic interconversions during fermentation as well as instigators of the selective pressure predisposing succession patterns in species/OTU diversity observed during the fermentation stages of kunu processing.

In the present study, several phylotypes common to all processing stages and final product of all formulations were identified. These common (or shared) phylotypes may constitute the core bacterial diversity of kunu production. The occurrence of these phylotypes throughout the fermentation stages and among all formulations suggest their key roles in kunu fermentation and earmark them as potential starter cultures for kunu processing. Some of these shared phylotypes, including species of Enterobacter, Gluconacetobacter, Lactobacillus, Leuconostoc Pediococcus and Weissella, have been previously reported in kunu (Amusa and Odunbaku, 2009; Nwachukwu et al., 2010; Oguntoyinbo et al., 2011), lactic acid-fermented west African cereal-based beverages such as ogi, koko, and akasa (Adebayo-tayo and Onilude, 2008; Oguntoyinbo, 2014; Okeke et al., 2015), cocoa fermentation (Nielsen et al., 2007), Spanish farm cheese (Abriouel et al., 2008) and suan-tsai (fermented mustard) from Taiwan (Chao et al., 2009). The role of these core microbiota during the fermentation of kunu may include the breakdown complex polysaccharides into simpler useful monomers (e.g., mannitol, a low-calorie sugar) (Hemme and Foucaud-Scheunemann, 2004; Ozogul and Hamed, 2018) and generation of aroma compounds that impact characteristic flavor to fermented food products. Some of these species, for example Leuconostoc spp., may also synthesize antibacterial compounds such as bacteriocins (Sawa et al., 2010) that help to eliminate pathogenic bacteria during fermentation or in the fermented food product (Ozogul and Hamed, 2018). The possible production of these useful chemical compounds by these species and other known beneficial/probiotic species found in the kunu samples, further underlines the health benefits associated with kunu consumption. Similarly, Enterobacter hormaechei have been previously reported during processing of fermented products such as lambic beer (Spitaels et al., 2014b) and inyu, a fermented black bean sauce (Wei et al., 2013). Also, strains of E. hormaechei are known to produce food additives and stabilizers such as trehalose (Richards et al., 2002). However, some E. hormaechei strains can cause infections in humans (Davin-Regli et al., 1997). Certainly, isolation and whole genome sequencing of some E. hormaechei strains associated with kunu processing may help elucidate its direct role (beneficial or pathogenicity) in this traditional beverage. As earlier mentioned, Gluconacetobacter is an acetic acid bacterium; however, its role in food fermentation is not yet completely understood. It has been associated with traditional balsamic vinegar (Gullo et al., 2006), however, Gluconacetobacter is also regarded as a spoilage organism in acidic beer (Bokulich and Bamforth, 2013; Mamlouk and Gullo, 2013). Nonetheless, the role of species of Gluconacetobacter during kunu processing could be a possible involvement in the oxidation of ethanol to acetic acid (Gomez-Manzo et al., 2010).

It is obvious from this study that the grains used for producing the various formulations of kunu were contaminated with different concentrations of several mycotoxins, albeit at levels that were lower than typical levels in maize batches applied to production of traditional fermented beverages in SSA (Okeke et al., 2015; Ogara et al., 2017; Ezekiel et al., 2018). Mycotoxin contamination of millet, sorghum, peanut and tiger nut have been previously reported in Nigeria (Adebajo, 1993; Makun et al., 2007, 2009; Rubert et al., 2013; Oyedele et al., 2017) and the associated public health concerns with such contaminations are well documented. A worrisome finding is the detection of all 10 mycotoxins in tiger nut; this suggests that tiger nut contributes significantly to the levels of mycotoxins during kunu processing. The AFB₁ level in tiger nut was about 13 times higher than 2 μg/kg; this is of concern because tiger nut is not only applied as an ingredient during kunu production but also popularly consumed as snacks (Belewu and Abodunrin, 2006; Gambo and Da’u, 2014). Thus, extracting the juice from this nut before adding it as ingredients during kunu processing and discarding the bran (Olaoye et al., 2015) could be explored to lower the mycotoxin contribution from this ingredient. A safer option is to exclude tiger nut from the list of ingredients intended for kunu processing and replace it with coconut or bambara nut. This is in view of the frequent feeding of kunu to young children as a complementary beverage in some parts of Nigeria (Olaoye et al., 2015).

Aside quantifying the mycotoxin levels in the ingredients, one main objective of the mycotoxin aspect of this study was to determine whether toxins present in the ingredients used for the various kunu formulations will be reduced and to what extent in the final product. This point was clearly established in this study as concentrations of all the mycotoxins reported in this study were reduced to non-detectable levels in the kunu irrespective of the formulations, except for AME, BEAU and CIT which retained very minimal levels (<2 μg/kg) in the final product of the various formulations. The findings of this study agree with previous reports that suggested that traditional processing significantly reduced the levels of mycotoxins in fermented foods and the extent of reduction depends on the mycotoxin content of grain inputs (Ezekiel et al., 2015; Okeke et al., 2015, 2018). The data on changes in mycotoxin levels during processing of the kunu indicated that steeping of the grains contributed the most to reducing several of the mycotoxins. This observation may be substantiated by the diverse phylotypes and high number of OTUs observed at this processing stage. During steeping, diverse bacterial genera were observed including those previously reported to be associated with various mechanisms of mycotoxin reduction (Shetty and Jespersen, 2006; Adebo et al., 2015); although in this study we did not search for the degradation products of the mycotoxins that were reduced. A further reduction of the contents of many of the mycotoxins was observed during the fermentation stage, and microbial data suggest community succession and competitive exclusion of species leading toward the development of a climax community. Lactic acid fermentation has been reported to play a significant role in mycotoxin reduction because the microbiota in the fermenting substrate could either bind mycotoxins to their cell wall (Byun and Yoon, 2003; Oluwafemi and Da-Silva, 2009; Huang et al., 2017) or degrade/biotransform them (Shetty and Jespersen, 2006; Adebo et al., 2015). Aside fermentation, simple dilution that occurs during the gelatinization step and sieving were also useful to reduce the levels of some of the mycotoxins as has been previously reported (Okeke et al., 2018).

In summary, our study indicates that the processing of kunu is mediated by several core bacterial phylotypes dominated by members of the LAB, although acetic acid bacteria and other bacteria with unknown functional roles constitute the core bacterial community. We have also further established that kunu could be a safe beverage for consumption in terms of mycotoxin contents depending on the grain inputs. Thus, we propose the combination of millet, white sorghum, cloves, ginger and sweet potato, which are usually minimally contaminated by mycotoxins, as possible combination of grains and additives for kunu production. This proposal will help to minimize mycotoxin exposure in consumers. In addition, the aforementioned ingredients could be supplemented with some underutilized crops (e.g., Bambara nut, coconut, finger millet (acha) and sesame seed) that may be less prone to mycotoxins. The extract from tiger nut may be more useful compared to the whole tiger nut; this will further reduce mycotoxin exposure while enhancing nutritional content and imparting additional flavor. Furthermore, it is imperative to source high quality grains and apply simple first-line grain processing interventions (e.g., sorting and floatation washing) to keep mycotoxin levels in the starting materials at the barest minimum in order to encourage a further mycotoxin reduction to non-detectable/safe limits in the final kunu product. There is a need to explore the core phylotypes reported in this study by isolating and ascertaining their probiotic properties and suitability as starters for up-scaling of some of the cereal-based traditional fermented foods. This study has further laid the foundation for the potential discovery of novel mycotoxin detoxifiers from kunu. Overall, the data provided herein are highly relevant to experts in the food microbiology and safety, food processing and technology, microbial ecology and molecular biology horizons.

CNE conceived the study. CNE, MS, OO, RA, CN, and RK designed the study. IC-O, KA, OO, OA, OE, CNE, and MS performed the experiments in Nigeria, South Africa, and Austria. MS, OE, DvW, CNE, and KA analyzed the data. CNE, CN, RA, JH, CTE, and RK supervised the overall study. KA, OE, DvW, and CNE drafted the manuscript. All authors reviewed and approved the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.