The study was initiated during the implementation of a pilot mango by-product feed production unit for a mango drying cooperative in Tengréla in the far north of Ivory Coast on 2019. Tengréla is located 108 km north of Boundiali, the capital of the region to which it belongs. Its geographic coordinates are 10˚28'55.056" north latitude and 6˚24'47.016" west longitude. Tengréla enjoys a relatively dry, warm Sudanian climate. Vegetation is dominated by wooded savannah. The area produces mangoes and is ideal for livestock farming [14] . The study was conducted near the mango drying unit of the Koto Wobin cooperative in Tengréla.

The biological material consists of the following:

- Effluent by-products from the processing of mangoes into dried mangoes. There are two types of effluent: 1) whole mangoes (WM) from the sorting holes before the start of the peeling operation to obtain slices of mango pulp for drying. These discarded mangoes are either injured during harvesting, or bitten by fruit flies, making them unsuitable for processing into dried mango, and 2) mango peelings (MP), waste from mango peeling, consisting of mango peelings or skin and pits.

- Agro-industrial by-products used as feed sorbents: wheat, rice and maize bran and palm kernel cake were the feed sorbents tested in the animal feed production process. They were acquired locally in Tengréla (rice bran and maize bran) or purchased in Abidjan, the capital of Ivory Coast (wheat bran and palm kernel cake) and transported to Tengréla. Sorbents are dehydrated products that are preserved and sold to feed mills in Côte d’Ivoire.

Production equipment consists of:

- 01 mango crusher-sorter: this is a machine that crushes the whole mango while sorting the pits so as to separate the peels + pulp on the one hand and the pits on the other, to enable separate processing of the 2 products. The crushed peel + pulp is then used for the drying test;

- 01 mixer: this is a machine used to mix products with a high water content. In the present test, it was used to mix the fresh feed obtained after blending the peel + pulp and the chosen sorbent.

- 01 drying table (experimental setup): As shown in Figure 1, it’s a rectangular table made of planks, rafters and spikes, 3.85 m long and 2.62 m wide, with four (04) supports 20 cm high. The table is made up of ten (10) boxes of 1 m² each, of which nine (9) were used. Each box is rectangular, 1.31 m long, 0.77 m wide and four (04 cm) cm deep. Each box has a surface area of 1 m² and a volume of 0.05 m³. At the bottom of each box is a tarpaulin on which the food is placed for drying. At the bottom of each box is a tarpaulin on which the food is placed for drying. The quantity of fresh food to be dried in each box is 18 kg. The process used to produce the feed was that of [9] . Following this process the feeds are dried to a moisture content of 14%, which ensures good food preservation in the tropics according to [15] . Two rates of sorbent incorporation were used in the experiment: 25% of the expected fresh weight of the feed for preparations using only mango peel, and 33% in the case of the grind composed of peel + pulp. The fresh feed obtained is then carefully mixed and spread out in the boxes of the drying table.

Food preparations are carried out in series of three treatments, each differentiated by the ingredients used. For each serie, 03 sorbents are used and 03 preparations of each type of sorbent are made, 09 preparations per series. The preparations, each weighing a uniform 18 kg, are then spread out in a box (plot) on the Table 1, as shown in Figure 1. Exposure to the sun takes place between 8 am and 6 pm. The drying feed is stirred 3 times a day at 8 a.m., 12 p.m. and 3 p.m. At night, the table is covered with tarpaulins to prevent the food from getting wet with surprise rainwater.

The parameters studied to determine the best sorbent are:

- Water loss through dehydration in the sun: this is calculated as the difference between the weights of fresh spread material (FMs) and sun-dried material (DMs) (Weight of fresh material—Weight of dry material).

- The production yield (PY) is given by the following formula DMs/FMs * 100

- Mango incorporation rate: MIR (%) = (PROD—sorbent weight)/PROD) * 100

- Production cost per kilogram of feed: given by the following formula:

CPkgPD (FCFA) = Total Production cost (FCFA)/quantity of feed obtained (kg)

- Number of days required for sun drying: UDT (days) = UDT (hours)/24 (hours) × 01 days.

Results concerning feed drying characteristics are presented in Table 2. Feed quantities (PROD) and yields (PY) were significantly higher and lower for feed groups PCWM33, RBWM33 and MBWM33, which were homogeneous with each other, and for feed groups MBMP25, WBMP25, PCMP25 and WBWM33, which were also homogeneous with each other (p = 0.000). Conversely, feed water losses were significantly higher and decreased between feed groups MBMP25, WBMP25, PCMP25 and RBWM33, MBWM33, and feed group PCWM33 (p = 0.000). Feed drying time and number of days were significantly lower for feed groups RBWM33, MBWM33, WBWM33, PCWM33 compared with MBMP25, WBMP25, RBMP25 and PCMP25 (p = 0.000). Production costs per 01 kg of Provende (CPkgPVD) were significantly higher and decreased from the WBWM33 and RBWM33,

Bromatological analyses were carried out on foods produced with hides incorporating various sorbents. Table 3 shows the results relating to the dietary values of these foods. Water contents were almost equal, and all below 10%. TCP contents were highest for WBMP25 and PCMP25, then MBMP25 and finally RBMP25. Crude Fiber (CF) contents were higher for PCMP25 and WBMP25 than for WBMP25 and MBMP25. Gross energy was higher and lower between PCMP25, WBMP25, MBMP25 and RBMP25 respectively. Energy digestibility for pigs was higher and decreased between MBMP25 (93.25%), WBMP25 (91.94%), RBMP25 (87.16%) and PCMP25 (83.65%).

- Means bearing the same letter in the same column are not significantly different at the 5% level; - MBMP25 = Mango skin provende + maize bran at 25% incorporation in the fresh mater; WBMP25 = Mango skin provende + wheat bran at 25% incorporation of absorbent in the fresh mater; RBMP25 = Mango skin provende + rice bran at 25% incorporation in the fresh mater; PCMP25 = Mango skin provende + palm kernel cake at 25% incorporation of palm kernel cake in the fresh mater; RBWM33 = Whole mango provende + rice bran incorporated at 33% in the fresh mater; MBWM33 = Whole mango provende + corn bran at 33% incorporation in the fresh mater; WBWM33 = Whole mango provende + wheat bran at 33% incorporation of wheat bran in the fresh mater; PCWM33 = Whole mango provende + palm kernel cake incorporated at 33% in the fresh mater; PROD (KgDM) = dry feed obtained by sun-drying; WL (kg) = water loss due to sun-drying; PY (%): feed yield obtained by sun-drying, MIR (%) = mango by-products incorporation rate in dry feed; UDTs (hours or days) = useful sun-drying time of feed; CPkgPVD (XOF) = production cost of 01 kg of dry feed.

MBMP25 = Mango skin provende + maize bran at 25% incorporation in the fresh mater; WBMP25 = Mango skin provende + wheat bran at 25% incorporation of absorbent in the fresh mater; RBMP25 = Mango skin provende + rice bran at 25% incorporation in the fresh mater; PCMP25 = Mango skin provende + palm kernel cake at 25% incorporation of palm kernel cake in the fresh mater; RBWM33 = Whole mango provende + rice bran incorporated at 33% in the fresh mater; MBWM33 = Whole mango provende + corn bran at 33% incorporation in the fresh mater; WBWM33 = Whole mango provende + wheat bran at 33% incorporation of wheat bran in the fresh mater; PCWM33 = Whole mango provende + palm kernel cake incorporated at 33% in the fresh mater;

The drying efficiency of palm kernel meal (PK) feeds is controversial, depending on the mango by-product (skin, skin + pulp) used. In fact, the results show that palm kernel flour-based foods lose less water over a shorter period of time than all other foods, and in particular those based on mango skin. If we consider the dietary values of the foods analyzed, we see that the Crude Fiber (CF) content of palm kernel flour + skin (PCMP25) is almost three and four times higher than that of wheat bran + mango skin (25%) (WBMP25) and corn bran (25%) + mango skin (MBMP25). On the other hand, it is only 12% higher than the Crude Fiber content of rice bran (25%) + mango skin (RBMP25). Closer examination of the chemical constituents of palm kernel flour reveals a higher starch content in these sorbents than in palm kernel flour. This is due to the presence of residual granules following the milling of the bran of these cereals. Starch has a very high water absorption capacity, but a slower release rate. In contrast, the abundant walls of palm kernel swell in a humid environment and release water more rapidly under the effect of heat. It has been reported that higher water absorption capacities are obtained from the gelatinized starch contained in bran, after gentle mechanical processing [16] , which is similar to grain processing in traditional mills, where extraction leaves residual grains in the bran. Wheat, on the other hand, which undergoes more extensive industrial processing to extract the maximum amount of grain, contains little or no residual grain and, consequently, lower starch contents. The longer dehydration time of foods made from corn bran and rice combined with mango skin is certainly due to the combined effect of residual grains and the slow release of water from mango skin. It should be noted that foods using whole mangoes dehydrate faster and release less water than those using mango skin. In addition, whole mangoes are subjected to the crushing process, which has the advantage of greater fragmentation of the skin, allowing water to be released earlier during crushing and drying. All these factors explain the higher yields and incorporation rates of mango by-products in the foods produced, with peel than with peel + pulp.

The lowest production costs were recorded for feed produced with mango skin + rice bran (RBMP25) and mango skin + palm kernel cake (PCMP25), followed by feed batches of skin + maize bran (MBMP25), skin + wheat bran (WBMP25), whole mango + rice bran (RBWM33), whole mango + maize bran (MBWM33) and whole mango + palm kernel cake (PCWM33). The various costs obtained at Tengréla are lower than the production costs reported at Bobo-Dioulasso in Burkina Faso, which average 147.25 FCFA for WBWM33 and 100.93 FCFA for MBWM33 feed [9] . This is a consequence of the cost of sorbents (corn bran and wheat bran), which are higher in Burkina Faso than in Tengréla in Ivory Coast. Whole mango + wheat bran (WBWM33) was the most expensive feed produced, due to the high cost of wheat bran (6000 XOF per 50 kg bag) in both Burkina Faso and Ivory Coast. The results of the analysis of the chemical composition of the few feeds obtained show that these feeds have waters contents of less than 10%. Similar results have been reported by other authors [9] [17] . This indicates that under Tengréla conditions, food can be dried as well as in Burkina Faso, with water contents below 14%, which is favorable for good preservation according to recommendations for drying agricultural products [15] . We can imagine that the water content of the feed made from whole mango is similar to that reported by [9] , which was below 10%. In terms of total nitrogenous matter (TNM), with the exception of rice bran feed (RBMP25) whose content is more than 2 times lower than that of palm kernel flour feed (PCMP25) and wheat bran feed (WBMP25), and 2 times higher than that of maize bran feed (MBMP25), the contents of other feeds prepared with skin and different types of bran at 25% incorporation in fresh feed are appreciable. They are slightly higher than those of certain sorghum grains [18] . Our results show that mango-based feeds are rich in crude energy (3910 to 4380 kcal/kg DM) and digestible energy for pigs (3408 to 3847 kcal/kg DM). However, the digestibility of these feeds is influenced by their fiber content. The higher the fiber content, the lower the digestibility, especially in monogastic animals. The high energy composition of feeds means that they could be used as basic ingredients in the diets of monogastric animals, including pigs and poultry, where microbial digestibility is low and fiber degradation very limited. The gross energy composition of mango-based feeds is comparable to that of maize (4571 kcal/kgMS) [19] , which is used in pig feed. Its digestible energy content is sufficient to cover the energy requirements of a pig weighing between 15 and 20 kg live weight [20] . In the formulation of pig rations, the recommended energy content is 3000 - 3200 kcal Digestible Energy (DE)/kg of dry matter (DM) and less than 12% of crude fiber (CF) for good pig growth [21] . Given these requirements, MBWM33, WBWM33, MBMP25 and WBWM33 feeds, in descending order of importance (Table 3), would easily achieve the required energy level and acceptable crude fiber level, and ensure rapid pig growth. It can also be used to prepare pullets for laying [12] . It should be noted that feed production reduces crude fiber content, making it more digestible in rations for monogastric animals. The high crude fiber content of palm kernel flour and rice bran feeds is due to the high crude fiber content of palm kernel flour, rice bran and mango skin. These two foods could be used in ruminant feed [13] , as ruminants are able to valorize fibers through digestion, which is mainly microbial in these species. Putting together technical drying efficiency factors, production costs and feed values, we arrive at several possible choices of sorbents and combinations to use. Those aiming to produce low-cost feeds with low feed values will probably opt for combinations using mango skins. Those looking for good feed values regardless of production cost will be more inclined to use combinations based on wheat bran and whole mango (mango skin and pulp).