The biological material consisted of Hermetia illucens larvae obtained after culturing black soldier flies (BSF) and flours of corn (Zea mays), soybean meal (Glycine max), fish and low-grade rice (Oryza sativa). The various flours were purchased from feed suppliers in municipalities of Port-Bouët, Bingerville and Yopougon, in the district of Abidjan, Côte d’Ivoire.

To obtain the flour, the larvae collected from the farms were rinsed with sterile distilled water. They were then scalded for 1 minute by direct immersion in a 100˚C water bath and vacuum-packed in plastic bags for freezing at −40˚C. For drying, the larvae were thawed at 4˚C in the refrigerator overnight, and dried in an electric oven at 60˚C for 48 - 72 h. After drying, the larvae were ground to meal using a grinder (IKA Labortechnik), then sieved using a 500 µm stainless steel sieve, washed and autoclaved (SYSTEC) at 121˚C for 15 min. The flour obtained was packaged in a 1 L sterile plastic bowl.

The various local flours purchased with the local retailers were sieved using a 500 µm mesh stainless steel sieve previously washed and sterilized in an autoclave at 121˚C for 15 min, and packaged in sterile liter plastic bowls.

The formulation method used to produce fish feeds is the linear programming method used by several authors [10] . This method makes it possible to compare a wide range of raw materials in order to determine the quantities of inputs that provide the desired nutrient levels at the lowest possible cost ( Table 1 ).

The formulation method made it possible to track variations in protein, lipid, ash and fiber content as raw material quantities varied. Depending on the formulation targets set, the different quantities of raw materials taken were determined. The formulation targets for our study are:

—Sum of protein percentages ∑P = 38%;

—Substitute fish meal for larvae meal at 0%, 10%; 25%; 35%; 50%; 65%; 75% and 100%.

With RM1 = Raw material 1; Q1 = Quantity of RM1; P1 = Quantity of RM1 × percentage of proteins of RM1; L1 = Quantity of RM1 × percentage of lipids of RM1; C1 = Quantity of RM1 × percentage of ashes of RM1; F1 = Quantity of RM1 × percentage of Fibers of RM1; P1 = Quantity of proteins brought by RM1; L1 = Quantity of lipids provided by RM1; C1 = Quantity of ash provided by RM1; F1 = Quantity of fiber provided by RM1; ∑P = Protein rate of the formulated feed; ∑L = Lipid rate of the formulated feed; ∑C = Ash rate of the formulated feed; ∑F = Fiber rate of the formulated feed.

BSF larvae powder was added to the fish feed composed of fish meal, corn meal, low-grade rice meal and soybean meal. Several fish feeds have been formulated by incorporating BSF larvae powder at 0%, 10%, 35%, 50%, 65%, 75% and 100% in place of fish meal. These feeds are produced in four stages. The ingredients are first sieved (500 µm mesh) to remove coarse particles, then mixed and blended to homogenize the mixture, and finally granulated and dried. These feeds were prepared using an automatic calculation system based on the linear programming formulation method [11] . The feeds were formulated under hygienic conditions.

All ingredients in the formulated feeds were analyzed individually after purchase from the feed suppliers and then after formulation. The dry matter, ash, protein, fat and fiber contents of the various formulated feeds were determined using the AOAC [12] method. The dry matter content of each sample was determined using an electronic moisture meter (OHAUS) at 105˚C for 45 min. Ash content was obtained by calcining the sample in a muffle furnace (Nabertherm 30 - 3000˚C) at 550˚C for 24 hours. Protein content was determined using the Kjeldahl method [12] for total nitrogen (N × 6.25). Total lipids were determined by the Soxhlet method [13] (using hexane as solvent). Fiber content was obtained by acid hydrolysis of the samples [12] .

The microbiological analysis consisted of counting indicative faecal contamination bacteria, such as E. coli, Anaerobic Sulfite-Reducing, faecal Streptococci, and also yeasts, molds, searching for pathogenic bacteria such as Salmonella and potentially pathogenic Staphylococcus, Pseudomonas, Aeromonas and Vibrio. Enumeration of microbial flora associated with flours was carried out using the standard method. For analysis, stock suspensions were prepared by adding 10 g of each feed sample to 90 ml of buffered peptone water (BPW) (Scharlau, France). The suspensions were homogenized for one minute, and several decimal dilutions were made from the mother suspension. Yeasts and molds were counted according to ISO 21527-1, 2008 on Sabouraud medium (Pan Reac Applichem). After culture, incubation lasted 3 to 5 days at 25˚C. Yeast colonies were round, smooth, whitish and creamy. As for the mold colonies, they were of different colors, depending on the type of mold. Anaerobic Sulfite-Reducing (ASR) were enumerated according to ISO 7937: 2004 on TSN (Tryptone sulfite neomycin) medium (Sharlau, France). Incubation was at 37˚C for 24 hours. Small and large black colonies were counted. Escherichia coli was counted according to NF ISO 16649-2 on Rapid E. coli 2 medium incubated at 44˚C for 24 h. Characteristic colonies were S-type and purple in color. Faecal Streptococci were counted in accordance with ISO 7899-2:2000 on Bile Esculin Azide agar (Thermo Fisher, USA). Incubation was at 37˚C for 24 h. Small translucent colonies with a black halo were counted. Staphylococci were counted using the ISO 6888 method on Baird Parker agar (Conda, Spain) supplemented with egg yolk and potassium tellurite (Bio-Rad, France). Incubation took place at 37˚C for 24 h. Only characteristic black colonies with a clear halo and an opaque zone around it were taken into account. A confirmatory test was then carried out for Staphylocoagulase using oxalated rabbit plasma, in accordance with standard NF V08-057-1 (January 2004). Vibrio enumeration was performed according to ISO 21872-1, 2017 on TCBS (Thiosulfate, Citrate, Bile, Sucrose) medium (Oxoid, UK). Incubation took place at 37˚C for 24 h. After incubation, colonies were medium-sized, smooth, opaque, yellow or green. Aeromonas enumeration was performed on Aeromonas Isolation Agar. Incubation was at 37˚C for 24 h. After incubation, yellow colonies were counted. Pseudomonas was counted on Cetrimide agar (Scharlau, France). Incubation was carried out at 42˚C for 24 h. Blue-green, yellow-green and brown-red colonies were counted. Salmonella sp. was detected according to ISO 6579-1: 2017 in 4 steps: pre-enrichment, selective enrichment, isolation and identification by biochemical characters.

The microbiological quality of formulated foods was determined on the basis of the microbiological criteria set out in ISO 11133: 2014.

For Salmonella, results are interpreted on the basis of the 2-class plan [14] .

For the other germs, results were interpreted according to a 3-class plan, following the m reference criteria [14] .

The results obtained were expressed in colony-forming units (CFU)/g by applying the formula:

$N=\frac{{\displaystyle \sum colonies}}{V_{ml}*\left(n_1+0.1n_2\right)*d_1}$

The analysis of variance (ANOVA) was performed with the R 4.3.2 software to study the degree of difference between the variables. In the event of a significant difference between the studied parameters, the classification of the means (homogeneous groups) is carried out with the Turkey test. The significance threshold (α) is 0.05. Statistical differences with a probability value of less than 0.05 were considered significant. The values used for the realization of the histograms on Excel, were in log10, while those used for the realization of the tables were in CFU/g of formulated foods.

Results of biochemical analyses of larvae, fish, corn, soybean floor and poor rice flour are presented in Table 2 . These results show that there is a significant difference (p < 0.05) between the different flours in terms of the parameters sought. Protein, lipid, ash, dry matter and fiber contents varied respectively from 7.98% to 46.84%; 1.30% to 22.96%; 2.65% to 25.03%; 88.80% to 96.51% and 2.80% to 15.09%. According to the results, black soldier fly larvae meal had high levels of dry matter (96.51%), lipids (22.96%) and total fiber (15.09%), while fish meal had the highest levels of protein (46.84%) and ash (25.03%). Corn flour and soybean flour were low in protein (7.98%), ash (2.65%), dry matter (88.80%) and total fiber (2.80%) in the case of corn flour, and in fat (1.30%) in the case of soybean flour.

Formulated feeds with 0%, 10%, 25%, 35%, 50%, 65%, 75% and 100% incorporation of BSF larvae meal showed a significant difference (p < 0.05) in the parameters sought, except for crude protein (p-value > 0.05). Dry matter varied from 89.65% to 91.19%. Ash varied from 20.27% to 8.15%. Total fiber and lipid ranged from 13.93% to 20.41% and 11.55% to 24.94% respectively ( Table 3 ).

In the formulation, the microbiological quality of the larvae meal, fish meal, corn flour, soybean floor and low-grade rice floor was assessed. Germs such as Vibrio, Pseudomonas, E. coli, Anaerobic Sulfite-Reducing (ASR) were found but Salmonella was absent from the various flours used in the formulation. Fecal Streptococci were only counted in the larval flour with a load of 8.6, 10² CFU/g. Yeasts and molds were only detected in the corn flour with an average load of 1.6,10² CFU/g. Staphylococci and Aeromonas were only detected in low-grade rice flour with loads of 2.6, 10² CFU/g and 2.6, 10³ CFU/g respectively ( Table 4 )

In the same line and for the same types of microorganisms, the values bearing different letters are significantly different at the threshold α = 0.05.

Indicative bacteria of fecal contamination, such as E. coli and Anaerobic Sulfite-Reducing were not detected in all feeds formulated with Hermetia illucens larvae. Fecal Streptococci were detected in all formulated feeds, with loads ranging from 2.4 to 4.9 log 10 CFU/g. Vibrio, Pseudomonas and Salmonella were not detected in any of the formulated feeds. Aeromonas was detected in feed formulated at 10%, 50% and 65%, with loads ranging from 2.2 to 2.7 log 10 CFU/g. Staphylococcus were detected in the 65% and 100% formulated foods, with loads ranging from 3.6 to 3.9 log 10 CFU/g. The most contaminated feed was the 65% formulation, with the presence of Aeromonas, Staphylococcus and Streptococcus ( Figure 1 ).

Yeasts and molds were detected in feed formulated at 25%, 35%, 50%, 65%, 75% and 100%, with loads ranging from 2.1 to 2.3 log 10 CFU/g. In the 0% and 10% formulated feeds, no yeast and mold load were detected ( Figure 2 ).

In terms of the microbiological quality of the formulated feeds, the 0% and 10% feeds were of satisfactory microbiological quality, while the 25%, 35%, 50%, 65%, 75% and 100% feeds were of acceptable microbiological quality according to ISO 11133: 2014 ( Table 5 ).

Strep f: Fecal Streptococci; Staph: Staphylococcus; ASR: Anaerobic Sulfite-Reducing; Pseudo: Pseudomonas; Salm: Salmonella; YM: Yeasts and Molds; Aero: Aeromonas; Abs: Absence; A: Acceptable; S: Satisfactory.