Full-Fat Black Soldier Fly Larvae Meal in Diet for Tambaqui, Colossoma macropomum: Digestibility, Growth Performance and Economic Analysis of Feeds

Simple Summary The black soldier fly (BSF, Hermetia illucens) is one of the most studied insects for use in feed, whose larvae (BSFL) can be used to feed fish. In this study, the inclusion of 10.5% wholemeal BSFL meal in the diet of tambaqui (a neotropical fish) resulted in higher growth than fish fed conventional ingredients. The use of wholemeal BSFL meal is more advantageous because it has a lower production cost than defatted BSF meal, as there is no expense involved in the fat extraction process. Wholemeal BSFL meal can be considered a leading ingredient to meet the emerging need for protein ingredients by the feed industry. Abstract Black soldier fly (BSF, Hermetia illucens) larvae is a prominent ingredient in aquafeeds due to its high protein and energy contents. This study evaluated the digestibility of full-fat BSF larvae meal (FF-BSFL) and its inclusion in diets for tambaqui, Colossoma macropomum. The apparent digestibility coefficient of FF-BSFL for protein and energy was around of 88%, corresponding to 33.55% and 21.61 MJ kg−1 of digestible protein and energy, respectively. For the feeding trial, tambaqui juveniles (53.23 ± 1.07 g) were distributed in a completely randomized experimental design (n = 4; 150 L tanks; 10 fish per tank). Fish were fed diets including 0%, 5.25%, 10.50%, and 15.75% FF-BSFL to apparent satiation for 60 days. Fish fed 10.50% FF-BSFL dietary inclusion showed higher weight gain, feed intake, final biomass, and relative growth rate. The 10.50% FF-BSFL diet presented the highest index of economic profitability. Weight gain fitted a third-degree equation and the optimum FF-BSFL inclusion level was estimated at 11.6%. However, FF-BSFL dietary inclusion up to 15.75% did not impair growth fish performance. FF-BSFL seems to be a promising source of protein and energy for omnivorous fish aquafeed.


Introduction
The world population reached 8 billion people in 2022 and, according to the United Nations [1], the forecast is that it will reach 9 billion inhabitants in the next fifteen years. Increased life expectancy associated with advances in medicine, health and nutrition are among the reasons for this unprecedented population growth. Therefore, there is composition, and they were supplied by the Natuprotein company in the city of Manaus, Amazon, Brazil. BSFL were dehydrated in a forced air circulation oven at 55 • C for 24 h. Dehydrated larvae were ground in a multiprocessor with 0.8 mm mesh.

Digestibility Assay
A practical feed was formulated to meet the nutritional requirements of tambaqui and used as the reference diet (RD; Table 1). The test diet (TD) was obtained by replacing 30% of RD by FF-BSFL. The apparent digestibility coefficient (ADC) was determined by the indirect method using 0.5% chromium oxide III (Cr 2 O 3 ), as a marker, in the experimental diets [20]. Table 1. Formulation and chemical composition of the reference diet utilized in the digestibility assay (values expressed on dry matter basis).
Tambaqui juveniles (48.66 ± 8.02 g) were housed in six conical tanks (200 L; n = 3; 17 fish per tank) fitted with a collection tube at the bottom for feces collection [21]. The tanks formed part of an open system with water renewal from an artesian well and with constant aeration. Fish were fed twice daily at a rate of 3% biomass for a 22-day period. Decanted feces were collected twice daily (8:00 am and 5:00 pm) and stored in a freezer (−20 • C) for later laboratory analyses.
The analyses to determine the chromium III oxide content in the samples of feeds and feces were carried out by the colorimetric method, according to the methodology described by Furukawa and Hiroko [22]. The standard curve was calculated from the nitro-perchloric digestion of the samples with known chromium III oxide concentrations. A reading was taken by a spectrophotometer, adjusted to a wavelength of 350 nm. The Apparent Digestibility Coefficients (ADCs) of nutrients and the energy of the RD and TD diets were calculated according to the following equation: ADC (%) = 100 − (100 × ((%Chromium III in diet)/(%Chromium III in feces)) × ((% nutrient (or energy) in feces)/(% nutrient (or energy) in diet)).
The nutrient and energy ADCs of the FF-BSFL meal were calculated following the equation proposed by Bureau and Hua [23]: where: ADCi = the apparent digestibility coefficient of the ingredient; ADCtd = the apparent digestibility coefficient of the TD nutrient; ADCref = the apparent digestibility coefficient of the RD nutrient; r = the proportion of the RD in the TD (0.65); i = the proportion of the test ingredient in the TD (0.3); Nref = the nutrient concentration in the RD (%); Ni = the nutrient concentration in the test ingredient (%).

Feeding Trial
Four experimental diets were formulated according to the nutritional requirements of tambaqui juveniles [24,25], i.e., a control diet without inclusion of FF-BSFL meal (0%) and three diets with the following levels of inclusion of FF-BSFL meal: 5.25%, 10.50%, or 15.75% ( Table 2). All ingredients were ground, sieved (0.8 mm), hydrated (25%), and extruded in a single-screw extruder (model MX-80, INBRAMAQ ® ; São Paulo, Brazil) with a 3 mm die. Tambaqui juveniles (53.23 ± 1.07 g; 14.92 ± 0.47 cm) were distributed in a completely randomized experimental design, with four treatments and four replicates (150 L tanks; 10 fish/tank), totaling 16 experimental units. The experimental units formed part of a recirculation system with phytoremediation, constant aeration, and natural photoperiod. The fish were fed 3 times a day (10 am, 2 pm, 5 pm) until apparent satiation for 60 days.

Water Quality Parameters
The water quality parameters of dissolved oxygen, temperature, and pH were monitored daily using a multiparameter probe (PRO ODO, YSI ® ; Yellow Springs, OH, USA). The ammonia, nitrite, and alkalinity levels were monitored weekly using colorimetric and titrimetric kits (Alfakit AT 101; Alfakit, Florianópolis, Santa Catarina, Brazil).

Growth Performance and Biometric Indices
Fish biometrics were performed at the beginning, 30th day, and 60th day of the experiment to monitor the growth of the fish. Before the biometrics, fish were fasted for 24 h and anesthetized with a solution of 100 mg of benzocaine L −1 [26] to facilitate handling and avoid injuries to the fish.
At the end of the experiment, the growth performance data were expressed by the parameters obtained by the following calculations: • average weight gain (g) = final weight-initial weight; • biomass weight gain (kg) = final biomass-initial biomass; • final biomass (g) = final number of fish × final average weight; • feed consumption (g) = total weight of feed consumed per tank; • apparent feed conversion = feed intake ÷ weight gain; • relative growth rate (% day −1 ) = (e g −1 ) × 100; where "e" is the Neperian number and g = (ln (final weight)-ln (initial weight))/(number of experimental days); • protein efficiency ratio (g g−1 ) = biomass weight gain ÷ protein consumption; • protein conversion efficiency (%) = [(final body weight × final body protein %) − (initial body weight × initial body protein %)/protein intake (g)] × 100; • condition factor = 100 × (body weight/total length 3 ). Two fish from each experimental unit were anesthetized by immersion in 300 mg benzocaine L −1 until the loss of reflex activity and no reaction to external stimuli were noted. Afterward, fish were euthanized by spinal medulla rupture according to the rules of the CONCEA [27] and destined for the collection and weighing of viscera and liver to determine the following indices: viscerosomatic index = 100 × (viscera weight/body weight); hepatosomatic index = 100 × (liver weight/body weight).

Proximate Composition
The proximate compositions of the FF-BSFL meal, diets, feces, and whole-body fish were determined according to AOAC [28]. Moisture content was determined by oven drying at 105 • C until constant weight; total lipids by solvent (hexane) extraction for 6 h (Soxhlet); ash content by burning in a muffle furnace at 500 • C for 4 h; crude protein (CP) by the micro-Kjeldahl method and correcting the total nitrogen content by multiplying by the factor 6.25. Specifically, for the FF-BSFL meal sample, CP was determined by the micro-Kjeldahl method and by correcting the total nitrogen content by multiplying by the factor 5.6, as recommended by Janssen et al. [29]. Chitin was analyzed following the modified method of Abidin et al. [30] with washing, deproteinization (NaOH), demineralization (HCl), and drying of samples. The gross energy of samples was determined by an IKA 2600 calorimeter bomb (IKA ® -Werke GmbH & Co. KG, Staufen, Germany).

Economic Analysis of Feeds
The cost of producing BSFL meal was determined through fixed capital investment for small-scale production of 24 kg of BSFL in a 21-day production cycle. Investment in civil construction was not considered. Production costs were determined based on the total operating cost, which is the sum of the effect operating cost (EOC) and the depreciation of fixed capital items. In the EOC, labor, equipment maintenance (20% of the new value per year), and the initial feed for chickens that served as substrate for the growth of the BSFL were added.
The labor cost was calculated based on the time used to produce the BSFL and the cost per hour worked (USD 1.67 h) in Brazil. The cost per hour worked was calculated based on a salary of USD 225.70 (equivalent to a minimum wage of BRL 1212.00 currently practiced in Brazil), plus 48% social security contributions for 200 h worked per month. Depreciation of equipment was calculated based on the straight-line method. The production of 1 kg of BSFL was estimated at USD 3.50.
The cost of the diet was calculated based on the market prices of the ingredients used ( Table 2). The cost of diet processing was not considered. The analysis was based on local market retail prices converted to USD (USD 1 = BRL 5.37; exchange as of 3 December 2022). The diets cost USD 0.93/kg (0% FF-BSFL), USD 0.97/kg (5.25% FF-BSFL), USD 1.00/kg (10.50% FF-BSFL), and USD 1.02/kg (15.75% FF-BSFL).
The economic efficiency of the diets was assessed using input-output analysis [12,31]. To determine the relative economic efficiency and cost benefits of the tested diet per unit of fish gain, the economic conversion ratio (ECR) was calculated as fellows: feed cost/kg weight gain = (feed intake (g)/body weight gain (g)) × cost of feed (USD/1 kg). The economic profit index (EPI) was calculated using the formula from Stejskal et al. (2020): EPI (USD/fish) = (gain of body weight (kg) × live fish selling price (USD/1 kg) − (body weight gain (kg) × (feed cost (USD/1 kg)).
The calculation of profitability (PRO) used the balance between the selling price of fish and the cost of feed per kg of fish gain (ECR): PRO (USD/1 kg of fish gain) = sale price of live fish (USD/1 kg) − economic conversion ratio (USD/1 kg of fish gain). Economic efficiency was calculated using the following ratio: economic efficiency = profit per kg gain/feed cost per kg gain.

Statistical Analysis
The variables expressed as percentages were previously transformed by square root arcsine. Data were submitted to the Levene and Shapiro-Wilk tests to verify normality and homoscedasticity, respectively. If the data were not normally distributed, or had unequal variance, they were subjected to the nonparametric Kruskal-Wallis test, followed by the Dwass-Steel-Critchlow-Fligner test. The parametric data were submitted to one-way analysis of variance where diet (4 levels) was the factor. If there was a difference between treatments (diets), means were compared using the Tukey test at α = 0.05. The treatments were also treated as a continuous variable, and regression analysis was performed. If significant differences were found, a lack of fit test was performed to validate the regression. Variables were analyzed by regression using the CurveExpert Pro software, and models were selected for best fitting the data to the model by the coefficient of determination (r 2 ) and by making a comparison by the F test and the AIC criterion.

Results
There were no significant differences in the water quality parameters during the experimentation period (Table 3). Tambaqui presented ADC above 88% for crude protein and energy of FF-BSFL, which resulted in 33.55% and 21.61 MJ kg −1 of digestible protein and energy, respectively. The ADC of lipids was 96.42% and resulted in 30.47% of digestible lipids for tambaqui. For chitin, tambaqui presented an ADC of 16.40%, which consisted of 1.72% digestible chitin (Table 4).
All groups of fish accepted the diets formulated with FF-BSFL and there was no record of mortality. Fish fed 10.5% FF-BSFL diet showed higher weight gain, final biomass, and relative growth rate compared to fish fed the other levels of dietary FF-BSFL (Table 4). The inflection point of the regression curve (Figure 1) was at 11.6% FF-BSFL dietary inclusion, which corresponded to 80.51 g of weight gain of tambaqui juveniles after 60 days of the experiment. Tambaqui fed 5.25% or 15.75% FF-BSFL dietary inclusion showed a similar growth performance to fish fed with the control diet (0% FF-BSFL). Apparent feed conversion, protein efficiency rate, protein conversion efficiency, and condition factor did not differ between experimental groups (Table 5).

Diets WG (g) FB (g) FC (g) AFC RGR (%) PER (%) PCE (%) CF VI (%) HI (%)
0% 56. 10  Data were analyzed using oneway ANOVA, when its assumptions were validated by Shapiro-Wilk for normal distribution and Levene's test for homogeneity of variance. If significant differences were detected (p < 0.05), then data were subjected to a Tukey HSD test. If the data were not normally distributed, or had unequal variance, they were subjected to the nonparametric Kruskal-Wallis test, followed by the Dwass-Steel-Critchlow-Fligner test (**). The treatments were also treated as a continuous variable, and regression analysis was performed. If significant differences were found, a lack of fit test was performed to validate the regression. The best-fitting model was chosen by AIC criteria and F test based on the highest r 2 and lowest p-value.   CF: condition factor. Different letters in column indicate significant difference. (*) Data were analyzed using one-way ANOVA, when its assumptions were validated by Shapiro-Wilk for normal distribution and Levene's test for homogeneity of variance. If significant differences were detected (p < 0.05), then data were subjected to a Tukey HSD test. If the data were not normally distributed, or had unequal variance, they were subjected to the nonparametric Kruskal-Wallis test, followed by the Dwass-Steel-Critchlow-Fligner test (**). The treatments were also treated as a continuous variable, and regression analysis was performed. If significant differences were found, a lack of fit test was performed to validate the regression. The best-fitting model was chosen by AIC criteria and F test based on the highest r 2 and lowest p-value.
There was no statistically significant difference in the proximate composition and protein conversion efficiency of tambaqui fed diets containing up to 15.75% FF-BSFL meal (Table 6). Data were analyzed using one-way ANOVA, after its assumptions were validated by Shapiro-Wilk for normal distribution and Levene's test for homogeneity of variance. The treatments were also treated as a continuous variable, and regression analysis was performed. If significant differences were found, a lack of fit test was performed to validate the regression.
The economic profit index values were higher (p < 0.05) for diets with 10.50% FF-BSFL meal inclusion. No statistical differences were observed for economic conversion ratio, profitability, and economic efficiency of feeds (Table 7). ECR: economic conversion ratio (USD/1 kg of fish gain); EPI: economic profit index (USD/fish); PRO: profitability (USD/1 kg of fish gain). Different letters in column, indicate significant difference. (*) Data were analyzed using one-way ANOVA, when its assumptions were validated by Shapiro-Wilk for normal distribution and Levene's test for homogeneity of variance. If significant differences were detected (p < 0.05), then data were subjected to a Tukey HSD test. The treatments were also treated as a continuous variable, and regression analysis was performed. If significant differences were found, a lack of fit test was performed to validate the regression. The best-fitting model was chosen by AIC criteria and F test based on the highest r 2 and lowest p-value.

Discussion
The water quality parameters of all experiments were within acceptable limits for tambaqui [32]. For new ingredients to be considered promising for aquafeed formulation, it is necessary to observe their nutrient and energy contents, in addition to their digestibility, attractiveness, effect on growth performance, and impact on animal health [33]. The nutritional value of FF-BSFL (42.3% crude protein; 37.9% corrected protein; 31.6% crude lipids) was within the average nutrient content found in the literature. In a review article, Tran et al. [34] found that the composition of BSFL ranged from 41% to 44% for crude protein and from 15% to 34% for total lipids. The nutritional composition of BSFL can be influenced by the nutritional variation of the substrate on which the insect larvae grew [7]. In addition, the BSFL crude protein value can be overestimated if the factor 5.6 is not used to correct chitin nitrogen or non-protein nitrogen [29].
The values of ADCs of the protein and energy of the FF-BSFL meal (higher than 88%) were similar to the values reported for Nile tilapia (Oreochromis niloticus) juveniles (70.0-82.1%) fed different insect meals [35]. Chitin can inhibit the absorption of lipids in the intestine and increase their excretion [36]. However, this was not observed in this study, since the FF-BSFL lipid ADC was greater than 95%. The low ADC values of chitin suggest that tambaqui has a low chitinase activity, as demonstrated in Nile tilapia, another neotropical freshwater species [34]. However, it is possible that a prolonged period of feeding FF-BSFL-based diets collaborates to increase the digestibility of chitin, since the digestive enzymes of tambaqui can specialize in digesting this carbohydrate [37]. In nature, tambaqui consume a diversified diet and, like most omnivorous fish, they have morphological adaptations such as numerous pyloric caeca and plasticity of digestive enzymes that contribute to the better utilization of nutrients [38,39].
Extrusion processing of feeds can make complex carbohydrates, such as chitin, more available. In fact, extrusion cooking can break down complex polysaccharides into smaller components, such as mono-or disaccharides, which are more digestible. The levels of chitin present in the experimental diets did not affect the growth performance of tambaqui, as was also observed in Atlantic salmon fed with FF-BSFL meal and paste of up to 12.5% dietary inclusion [40].
The higher consumption of the 10.50% FF-BSFL diet reflected better growth performance results for tambaqui. Insects have pheromones on their surface, known as "natural attractants", which may have acted as a palatability agent in the experimental diets [41,42]. However, the inclusion level of 15.75% may indicate a limit to the palatability of FF-BSFL dietary inclusion or fish satiety due to the chitin content.
Chitin is a carbohydrate resistant to chemical degradation, but it presents an energy content of approximately 17.1 kJ g −1 [43]. This fraction of energy was accumulated in diets with increasing levels of FF-BSFL inclusion; however, as chitin digestibility was low (16.40%), it may have been a kind of filler with low digestible energy content [44]. This may have kept the fish satiated for a longer period of time, influencing their feed intake and weight gain. Similarly, Siberian sturgeon (Acipenser baerii) fed FF-BSFL meal-based diets showed improvements in growth performance parameters [5].
In aquaculture, feeding costs amount to more than half of production costs [45]. The reduction in fish feeding costs contributes to greater profitability and economical sustainability of fish farming. Studies on the effects of using ingredients from BSFL production in diets for neotropical fish related to the economic analysis are still limited. In this study, the improvement in feed conversion of fish that consumed 10.50% FF-BSFL dietary inclusion indicates a contribution of the use of FF-BSFL meal to improve the efficiency of tambaqui production.
Despite the feed price increasing according to the inclusion of FF-BSFL meal, feeding tambaqui with 10.50% FF-BSFL meal diet increased the economic profit index by an average of 6% in relation to the diet with 0% FF-BSFL meal. In the case of Nile tilapia fed diets with 75% replacement of fish meal by BSFL meal, an increase of 3.97% in the economic profit index was observed [46], and greater economic efficiency with replacement of 100% fish meal was found [31]. When diets with FF-BSFL were tested in sturgeon, the economic profit index increased according to the level of inclusion [5]. In these studies, the price of BSFL meal was the main factor that had an impact on the variation in diet prices.
There is a worldwide tendency to produce BSFL on a large scale, which can reduce the costs of production, positively impacting the economic efficiency rates of feeds formulated with BSF ingredients. Furthermore, it must be considered that full-fat BSFL meal has a lower production cost than the defatted BSF meal, as there is no expense associated with the fat extraction process. The optimization of growth performance of tambaqui fed FF-BSFL meal-based diets and the economic profit index of its production indicate the possibility of more sustainable aquaculture, which can strengthen the bioeconomy around the world to produce more fish as food for humans.

Conclusions
Full-fat black soldier fly larvae meal is well digested by tambaqui and the inclusion of 11.6% FF-BSFL in diet yielded the maximum weight gain in tambaqui juveniles. Based on the results obtained, FF-BSFL meal can be considered as a prominent ingredient to meet the emerging need for protein sources by the aquafeed industry.

Informed Consent Statement: Not applicable.
Data Availability Statement: Data from the study are available from the corresponding authors upon reasonable request.