Biological material consisted of by-products (BP) from the most commonly used mango cultivars (Amelie, Lippens, Kent, Brooks and Keitt) from small-scale and industrial fruit processing units in Burkina Faso. Samples were collected between April and June 2023 (mango availability period). They were taken directly from processing units in the Haut-Bassins (10˚57'25" Latitude,−4˚51'41" Longitude), Cascades (10˚42'36" Latitude,−4˚46'28" Longitude) and Centre (12˚21'44" Latitude, −1˚32'21" Longitude) regions of Burkina Faso (Figure 1). A total of 25 samples of Amelie (10), Kent (07), Lippens (04), Brooks (03), and Keïtt (01) byproducts were collected, with each sample weighing approximately two (2) kilograms. It was sampled, packed in plastic bags, labelled, placed in a cool box with ice and sent to the laboratory. In the laboratory, part of the fresh samples was ground directly in the XPREP Warning Commercial(stainless-steel blades) mixer for moisture, pH, acidity and Brix analysis. The other part was dried in the LABFREEZ-Instruments FD-10-MTP freeze dryer, then ground before being repacked into sample jars, labelled and stored at 4°C for further analysis.

Figure 1. Map of data collection and sampling areas.

Data collection equipment consisted of a voice recorder and an android phone with survey forms and an interview guide designed using KoboCollectv2023.1.2 software. Data were collected between April and June 2023 from mango proces sors and contact persons (local authorities, health and environmental services and researchers) in four provinces in Burkina Faso, i.e. i) Houet: Bobo-Dioulasso, 11˚14'13" Latitude, −4˚19'7" Longitude and Toussiana 10˚50'28" Latitude, −4˚38'39" Longitude, ii) Comoé: Banfora, 10˚41'36" Latitude, −4˚47'31" Longi tude, iii) Kénédougou: Orodara 10˚58'25" Latitude, −4˚54'35" Longitude and iv) Kadiogo: Ouagadougou 12˚21'44" Latitude, −1˚31'21" Longitude (Figure 1).

The survey data were collected using a technique adapted from AGRIS manual of FAO on integrated agricultural surveys [7]. The scope of the survey and interview forms included the identification of the actor, the size of the processing unit, its production capacity with the quantities of by-products generated, the management of these by-products within the processing unit and the costs associated with this management. The sample was selected based on the availability of the actors. Only companies that had remained active in the mango processing industry over the past 5 years were surveyed. A total of 33 in 45 actors met the selection criteria. The data was collected online directly on the electronic form using KoboToolBox. The interviews were recorded using a voice recorder and transcribed into Microsoft Word©.

The pH and total acidity (expressed as citric acid equivalent) of the fresh mango by-product were determined by potentiometry and titration, using the SI Analytics-Lab875P pH meter and the Brix degree by refractometry using the Bellingham + Stanley RFM712 refractometer [8].

Moisture content was determined according to AOAC method 925.10 [8] and the total ash content according to AOAC method 942.05-1943 [8]. The total protein content was determined by the Kjeldahl method described in AOAC 992.15-1992 [8] and the total fat content (TFC) was determined by the Soxhlet method described in AOAC 963.15-1973 [9]. Total sugars and reducing sugars were determined using the spectrophotometric methods described by Dubois [10] and GHANES [11], respectively. The potential energy value was calculated according to Merrill and Watt [12].

Total phenolic compounds were determined by the Folin-Ciocalteu colourimetric method [13]; total flavonoids were determined by the method described by Zhishen et al.[14] and the antioxidant capacity of the ethanolic extracts was tested by the DPPH (1,1-diphenylpicrylhydrazyl) reduction method described by Ben Moussa et al. [15].

Minerals were determined by flame atomic absorption spectrometry (AAS) using the AOAC 999.11-2005 method [16]; Amino acid contents were performed by reversed-phase HPLC using the AOAC 2018.06-2018 method adapted from the Waters Pico-Tag technique [17]. This involved hot acid digestion of the delipidated sample to release the amino acids, which were then complexed with phenyl isothiocyanate (PITC) to form phenylthiocarbamyl amino acid, which can be quantified. The HPLC is equipped with UV detector and PICOTAG Column C18 (Waters, USA). The HPLC system was pre-calibrated with high-purity amino-acid standards. The sample injection volume for all the analytes was 10 μl, which was filtered using 0.45 μm puradisc syringe filters before injection into the HPLC.

Ascorbic acid (vitamin C) content was determined by Thermo Scientific UHPLC chromatograph method described in AOAC ZBH 26-1999 [8] coupled to UV-vis detector. Total ascorbic acid (TAA) was considered as the sum of ascorbic acid (AA) and its oxidized form (dehydro-ascorbic acid) [18].

Beta-carotene content was determined by UHPLC using the AOAC 2005.07-2005 method [19]. Carotenoids soluble in organic solvents are directly detectable on the Thermo Scientifica UHPLC chromatograph (coupled with a UV-visible detector) after extraction by maceration in chromatographic grade ether.

This was carried out using the experimental design method described by Goupy and Creighton [20][21]. A factorial design was created using Minitab 18.1 v2014software. The factors studied in 05 levels of combination were the proportion of mango by-products (MBPs) (combination values: 25%, 30%, 35%, 40% and 45%) and the proportion of soybean flour (SBF) (combination values: 0.0%, 0.1%, 0.5%, 1.0% and 1.5%). For each formula obtained by the factorial design, a bioreaction substrate was prepared according to Ouédraogo et al. [22]. The required proportions of MBP and SBF were weighed into a 1-L Erlenmeyer flask before adding the necessary amount of distilled water. The mixture was homogenized, and the pH was adjusted between 5.8 and 6.0. It was pasteurized at 75˚C for 10 min, followed by cooling to approximately 30˚C, before being inoculated with viable baker’s yeast (Saccharomyces cerevisiae) at a rate of 0.1 g per 100 mL and covered with muslin cloth to facilitate aeration of the medium. The flasks were placed on a rotary shaker tray for the bioreaction. Reactions were carried out for 72 hours at laboratory temperature (25˚C - 30˚C) under continuous aeration (Figure 2). Response parameters included final pH, Brix reduction rate, crude protein content and optimised protein content of the final biomass. The optimized protein taut is calculated using the following formula:

(1)

Pns% = Nitrogen source Protein rate;

Wns = Nitrogen source weight added in the substrate;

DM% = Biomass dry matter.

The Proportion of Residual Brix (PRB) was calculated according to the following formula:

(2)

BrixDegree Tn:Brix at time “n” of the bioreaction;

BrixDegree T0:Brix at time “0” of the bioreaction.

Figure 2. Diagram of mango byproducts bioreaction recovery. (MBPs= Mango By-products).

Analysis of variance (ANOVA) of the experimental results were statistically performed using XLSTAT 2016.02.27444 software. The statistical means of the data were compared according to Tukey (HSD) with a significance level of 5%.

The experimental design and the responses optimizing were performed with Minitab 18.1 version 2014. The response values were optimized using a 95% two-sided confidence level without any constraints to obtain the optimized formula.

Data collected from KoboToolBox were extracted and processed in using Microsoft office version 2021.

A total of 33 actors were interviewed in the mango production and processing areas of Burkina Faso. Based on the quantities of raw material processed per year, the mango processing units were classified as very small actor (VSA), small actor (SA), medium actor (MA), large actor (LA) and very large actor (VLA) units (Table 1).

Table 1. Classification of mango processing units according to the volume of raw material processed per year.

Among mango processing actors 84.85% were drying processors. The production of mango puree and nectar occupied only 15.15% among actors (Table 2). This could be explained by the fact that the production of puree requires more resources, including qualified personnel and more technical facilities for aseptic packaging of end-products. Levels of processed mangoes varied from 16.47 tonnes (VSA) to 6643.18 tonnes (VLA). All VSA are involved in mango drying, as are SA and MA. Each drying unit processes about 1247.87 tonnes of fresh mangoes in a production period which last 100.31 days, compared to 4673.92 tonnes for mango puree and nectar producers during the production period which last 120 days (Table 2). These processed quantities generate a significant amount of by-products, estimated on average at 64.73% (w/w) in drying units and 36.53% (w/w) in puree and nectar production units. The highest quantities were found for SA and MA (all drying promoters) with 66.33% and 67.50%, respectively (Table 2). This could be explained by the inadequacy of the infrastructure and the lack of mastery of the technology by these categories of actors. The by-products generated are made up of unripe spoiled fruits and ripe spoiled fruits. They were all obtained by sorting before processing and represent about 15% of the quantities processed in the drying and mango puree and nectar production units, respectively. They also include direct by-products, which are residues generated during the unit operations to produce mango pulp and end products. These direct by-products, consisting of peels, kernel, pulp residues and fibers, represent 49.06% and 21.14% of the quantities processed in the drying and puree and nectar production units, respectively. During ripening in the processing units, the mango is subject to evaporation in the ripening rooms, which results in a significant mass loss of about 10% of the fresh mango quantity. The total mass loss during the processing of fresh mango is therefore estimated at 74.73% and 46.53% in the drying units and mango puree and nectar production units, respectively (Table 2).

A study conducted in the Haut-Bassins region by Villard et al. [5] showed lower MBP rates, estimated at around 60% and 20%, respectively in drying units and purée and nectar production units. This difference could be explained by a lack of control over the flows (incoming and outgoing) in the industrial processes of the mango processing units.

Table 2. Rates of by-products generated by categories of mango processors.

¹Total by-products = Spoiled (unripe + ripe) + Direct pulp extraction residues, ²Total mass losses = Total by-products + Evapo-transpiration rate.

In mango processing units, by-products are generated and evacuated to a waste disposal centre or landfill. These sites are located at 2.73 km and 0.20 km away from the drying plants and the puree and nectar production plants, respectively (Table 3). The long distance from the landfill sites of the drying plants is justified by the fact that they do not have a waste collection point on their premises. The MAs, all of which operate in the drying sector, have the longest disposal distance (4.50 km on average). They all use disposal areas allocated by the municipal authorities (Figure 3). The cost of disposing of these mango by-products in landfills is estimated to be between 4.57 and 6.09 €/m³, depending on the stakeholder. In mango drying plants, the disposal of by-products incurs costs related to the maintenance of tricycles and trucks, fuel costs and driver salaries [4].

On average, 27.78% of the actors have recovery initiatives that are not widely used, such as direct spreading on the fields, composting and biogas production. Although the puree and nectar production units have large-capacity landfills, they do not have facilities to recover the large quantities of by-products. On average, 72.22% of processors (Table 3) landfill the mango by-products they generate.

(a) (b)

Figure 3. Means transportation (a) and mango by-products waste disposal site (b).

The direct use of mango by-products as fertilizer through controlled spreading [23] maintains the stability of the soil flora and the balance of minerals present in the soil [24]. Composting of mango by-products is a source of good quality organic fertilizer [25]. According to surveyed stakeholders, most of them are aware that mango by-products may be a high value source of feed for livestock, especially ruminants (Figure 4). With the microbial flora of the rumen, ruminants are able to biodegrade fruit by-products such as mango, which are rich in fibers and other compounds that monogastric animals cannot assimilate [24].

Figure 4. In field consumption of mango by-products by some animals.

Table 3. Management of mango by-products by stakeholder categories in Burkina Faso.

3.3.1. Physicochemical and Proximate Composition of Mango By-Products

Mango by-products (MBP) had water content of 81.50 ± 2.25%, 15.31 ± 2.86˚ Brix, with a pH of 4.23 ± 0.38 and a total acidity of 3.51 ± 0.74 g (Table 4).

The MBP had total fat content of 2.15 ± 0.61% g/g DM, the protein content ranged from 2.29 ± 0.03 to 4.66 ± 0.70%, total ash content of 3.42 ± 0.54% and total sugars content of 90.80 ± 1.38%, with energy value of 397.03 ± 3.71 kcal/100g (DM). There is significant variation in the composition of the BPs of the different mango cultivars processed.

The moisture values are higher than those obtained in India (72.50 to 75.25%) [26] and in Colombia (74.6%) [27] on fresh mango peels. In addition, the pH (3.33 to 4.95) and Brix values (11.00 to 21.13˚ Brx) obtained in our previous study [28] with the flesh of different mango cultivars are similar to present data. This could be explained by the fact that mango by-products in processing units are very pulpy. The protein, lipid and total ash contents are on average similar to that of mango by-products from different cultivars and different ecosystems reported in other studies [26][29][30]. However, slight difference in proximate composition could be justified by the difference in cultivar, agro-pedo-meteorological conditions. All protein contents are higher than those obtained by Kanté-Traoré et al. [28] on the pulps of different mango cultivars in Burkina Faso. This indicates that the mango skin is richer in protein than the pulp, which is sweeter and has a higher water content.

Comparison of total sugar and protein contents show that MBP displayed higher levels in carbon than nitrogen. This could result in a nutrient imbalance and interfere with the biochemical reactions during the organic (composting or mechanization) and biotechnological recovery of MBP in the single bio-reaction mode [4]. A substrate formulation co-digested with a nitrogen source would therefore be required for improved bio-reaction. Given the high content of moisture, sugars and other biogenic residues, MBP generated in mango processing units would be conducive to enzymatic and microbial hydrolysis reactions [31]. In addition to these properties, mango being a climacteric fruit, would enlighten the high perishability of MBPs, which are a source of carbon that can be exploited for biotechnological valorization as a substrate for several industrial microorganisms and a source of bio-active molecules [32].

Table 4.Comparison of physicochemical characteristics and potential energetic values of MBP from different cultivars processed in Burkina Faso.

3.3.2. Bioactive Compounds of Mango By-Products in Processing Units

FromTable 5, mango by-products (MBP) had total phenolic compounds of 80.81 ± 13.53 mg GAE/g of dry mater (DM), with flavonoids, reducing sugars, and antioxidant compounds content of 27.65 ± 7.65 mg quercetin E./g of DM, 81.06 ± 22.46 mg D-glucose E./g of DM, and 8.82 ± 0.18 mg ascorbic acid E./g of DM, respectively. It had vitamin C content of 2.04 ± 1.12 mg/g of DM and beta-carotene content of 1.88 ± 1.18 mg/g of DM. There was a significant difference between BPs of mango cultivars in terms of bioactive compounds.

Table 5. Comparison of bioactive compounds in mango by-products among cultivars.

*Abbreviations: GAE: gallic acid equivalent; A.Ox: anti oxydant; AAE: ascorbic acid equivalent.

Phenolic compounds in this study are lower than that obtained by Vithana, Singh and Johnson [33] on mango peel in Australia (168.0 mg/g GAE). These authors found that phenolic compounds content of the mango increased as it ripens on the tree. On the other hand, a study by Aziah et al. [34] suggested that phenolic compounds content in unripe mango skin (102.41 mg/g EAG) is higher than that in ripe mango skin (70.20 mg/g GAE). Another data [30] showed higher levels of phenolic compounds (96.2 mg/g GAE) than those in the present study, which were higher than those obtained by Sogi et al. [35] on freeze-dried mango peel (31.85 mg/g GAE). These differences could be explained by differences in cultivars as well as the variability of the agro-climatic conditions of the study areas.

Levels of flavonoids are similar to those obtained in un-ripped and ripped mango peels (33.00 and 29.24 mg/g quercetin E., respectively) by Aziah et al. [34], who obtained higher levels of antioxidant compounds in unripped and ripped mango peels (54.23 and 43.30 mg/g Trolox E., respectively). This difference in antioxidant activities could be explained by the different type of reducing agent used in the assay.

Levels of pro-vitamin A such as beta-carotene are lower than those of total carotenoids (3.09 mg/g) obtained by Ajila et al. [30]. They are also lower than those for total carotenoids (3.34 - 3.94 mg/g) found in ripped mango peel (Raspuri cultivar) from India [26]. However, they are higher than those for total carotenoids reported by Aziah et al. [34] (0.16 mg/g) and Izidoro et al. [29] (0.036 - 0.055 mg/g). These differences could be enlightened by a cultivar difference, but also by the variability of the agro-climatic conditions of the study areas. Our results are superior to those of total carotenoids obtained by Ajila, Bhat and Rao [26] (0.36 - 0.55 mg/g) and Aziah et al. [34] (0.10 mg/g), from unripe mangoes peels. This difference could be explained by the fact that carotenoid content increases as the mango ripens. Indeed, during ripening, there is a breakdown of chlorophyll and the increase in the activities of hydrolytic enzymes which may release carotenoids and phenolic compounds including anthocyanins [4][36][37].

Ascorbic acid levels are higher than those reported by Ajila et al. [30] (0.39 mg/g) and Izidoro et al. [29] (0.97 - 1.36 mg/g) in ripe mango peels. However, they are lower than those found (52.51 mg/g) by Aziah et al. [34] in ripe mango peels. These differences could be due to cultivar difference, but also by the variability of the agro-climatic conditions of the study areas. In line with previous data, ripe mango skin is richer in ascorbic acid than the flesh [28]. Similar data were found in other fruits such as pineapple peels [38].

Reducing sugar levels are similar to those from fresh mango peels [6]. They are also similar to the glucose contents (25 - 124 mg/g) from apple residues [24], but lower than those of reducing sugar contents (279.05 mg/g) in pineapple peels [39]. Thus, mango by-products are relatively low in reducing sugars with high total sugar content (90.80 ± 1.38%). Therefore, biotechnological valorisation and/or anaerobic digestion of mango by-products requires pre-treatment either by heating or enzymatic hydrolysis [4] to increase the content of fermentable sugars that can be easily catabolized by microbial strains. Yeasts are able to transform several free hexoses to produce ATP energy, for their growth [40].

The results show that mango by-products are a significant source of ascorbic acid, carotenoids and other phenolic compounds with strong antioxidant properties. Consumption of mango peels may have nutritional and health benefit because of their endogenous bio-active compounds exhibiting anti-proliferative properties on cancer cells [41][42]. Antioxidants may have the property of reducing mutagenesis (carcinogenesis), while reducing the damage caused by DNA oxidation, with a decrease in cell division [24]. With regard to ascorbic acid, its relatively high level presence in MBP may contribute to fight against the formation of microbial biofilms in foods, especially in liquid foods, due to its antibacterial properties on pathogenic germs such as S. aureus, L. monocytogenes, Campylobacterjejuni, M. tuberculosis, Aspergillusspp, ... [43][44]. This could be an advantage in biotechnological and/or organic recovery (methanisation or anaerobic digestion) of MBP. Indeed, the antibacterial activity of ascorbic acid against contaminants in the substrate would optimize the availability of carbon and nitrogen sources for better biological activity in a bioreactor or anaerobic digester.

3.3.3. Amino Acid Composition of Mango By-Products

Mango by-products (MBP) had amino acids (AA.) content of 3.08% g/g of DM, of which 1.50% is essential (Table 6). There is a significant difference in the content of Arginin, Threonin, Prolin and Leucin between MBPs of different cultivars of mangoes. Detailed analysis of specific amino-acid content of MBPs showed that they contain interesting levels of quantifiable essential amino-acid according to assays used. It showed that amino-acid content in MBPs are low levels relative to the protein content. To improve the dietary use of these MBPs, protein formulations or fortification technologies would be required to meet the nutritional requirements of amino-acid [45].

Table 6. Amino acid ¹composition of mango by-products in processing unit according to cultivars.

¹Data were expressed in proportion (%, w/w, dry matter basis).

3.3.4. Minerals Composition of Mango By-Products in Processing Units

There was a significant difference in mineral content between BPs of mango cultivars in processing units (Table 7). Major minerals were potassium (1288.93 ± 77.31 mg/100g, DM), phosphorus (403.11 ± 172.22 mg/100g, DM), calcium (162.25 ± 34.28 mg/100g, DM), and magnesium (55.32 ± 9.80 mg/100g, DM). Minor elements were iron (3.054 ± 0.52 mg/100g, DM), manganese (2.921 ± 0.18mg/100g, DM), zinc (0.617 ± 0.15 mg/100g, DM), and copper (0.018 ± 0.02 mg/100g, DM).

Table 7. Mineral element composition of MBP in processing units.

Levels of calcium, magnesium and copper found in MBP are lower than those reported by Izidoro et al. [29] who obtained lower contents of phosphorus (101 - 178 mg/100g), manganese (10.15 - 30.50 mg/Kg) and iron (2.20 - 16.00 mg/Kg). On the other hand, they obtained potassium (1124 - 1380 mg/100g) and zinc (6.10 - 9.08 mg/Kg) contents similar to those in the present study. The mineral contents of mango pulp obtained by these authors are lower than those of mango peel. Levels of minerals is important for a better metabolism of the substrates by fermentative strains in bioreactors in biotechnological processes (cellular proteins production) or anaerobic digestion (biogas production). This is because the energy metabolism pathway of microbial cells requires most of these mineral elements as cofactors during glycolysis, the tricarboxylic acid (TCA) cycle and oxidative phosphorylation [46][47]. This metabolism involves the mineral phosphate (Pi), which combines with ADP to form ATP molecules [40]. For instance, during cell division of strains, phosphate is involved in the formation of the cell membrane (phospholipid bilayer). Magnesium content showed that in a bioreactor fed with MBP substrates, the microbial cells involved in the bioreaction had a biologically active energy source [46]. On the other hand, high levels of potassium may help to limit excessive acidification of biomass during biotechnological or anaerobic digestion recovery. Potassium plays a buffering role, limiting pH variation in the cellular environment and allowing polarization of the cytoplasmic membrane [47]. Under aerobic conditions, the reduced forms of certain mineral elements, including iron and manganese, can act as electron donors to allow rapid biodegradation of wet bio-waste such as MBPs, with heat production [4][48].

The mathematical combination of the factor values resulted in a total of 25 test formulations (Table 8). The biomasses obtained after the bioreaction of the substrate of each formula were characterised by a final pH, a Brix reduction rate, a crude protein rate and an optimised protein rate varying respectively from 3.68 to 4.93, from 54.79% to 131.37%, from 3.45% to 15.88% and from −6.03% to 6.79% (Table 8). Low Brix reduction rate indicate efficient sugar metabolism by the inoculum strain, accompanied by a decrease in the Brix degree of the medium. The best decrease in the Brix degree was observed in the biomasses from substrates formulated with 40% or 45% MBP. The Brix reduction rate above 100% shows an increase in the soluble dry matter content during the bioreaction. This could be explained by an excessively long hydrolysis phase of the medium’s carbohydrate macromolecules. The best Brix reductions were obtained with substrates from formulas containing 40 and 45% mango by-product. All formulas containing 1.0% and 1.5% organic nitrogen source had the lowest optimised protein levels, despite their high crude protein content. During the bioreaction, from the sixtieth hour, an increase in the pH of the medium was observed in the formulas containing 0.1% of the organic nitrogen source. This could be explained by a better metabolism of the inoculum strain in these substrates, resulting in an optimal synthesis of nitrogen and therefore an increase in the pH of the medium for these formulations. The best optimised protein contents were obtained with the formulations whose substrates were prepared with 0.1% of the nitrogen source and 40 and 45% of mango by-products. Statistical optimisation of the experimental design using response parameters with Minitab software showed that MBPs = 40% and SBF = 0.1% gave the most optimal response (Table 9) with 6.79% optimised protein rate and 0.8123 composite desirability. The crude protein contents of our study are all lower than those obtained (30.84 - 56.40%) by Somda et al. [49] from substrates prepared with mango biowaste powder. They are also lower than those obtained (26.47%) by Umesh, Thazeem and Preethi, [50] from pineapple peels. This difference could be explained by the fact that these authors used ammonium sulphate and yeast extract as nitrogen sources. Their substrates were also enriched with different mineral sources such as KH₂PO₄, MgSO₄∙7H₂O, NaCl and CaCl₂. Formula F23 is therefore the basis for better optimization when substrates are enriched with these reagents and ammonium sulphate or yeast extract are used as additional nitrogen sources to reduce the nutritional imbalance of the fermentative strain during the bioreaction [4] for single cell protein production and renewable energy generation.

Table 8. Expérimental design and responses optimizing.

Table 9.Optimized solution.