A detailed description of the experimental design, diets, and nutrient composition, including the amino acid and fatty acid profiles of agri-industrial by-products and BSF meal were provided in Acar et al. (4) and Yalçin et al. (39). In brief, a total of 252 day-old chickens from a local line (Anadolu-T pure dam line) were reared at 18-floor pens in an environmentally controlled experimental poultry house until 55 days of age. The chicks were randomly distributed into three dietary groups. (1) A corn-soybean-based diet (Control), (2) a diet in which soybean was partially replaced (SPR) with local agri-industrial by-products (resulting in an average 24.85% reduction in soybean meal amount compared to the control diet), including sunflower meal, brewers' dried grain, and wheat middlings, and (3) a diet in which dried BSF (5%) larvae meal was added to SPR (SPR+BSF), reducing soybean meal inclusion by an average of 42.2%. The ingredients and nutritional composition of the diets are given in Table 1. In this project, the target overall reduction in soybean meal inclusion (averaged across the starter, grower, and finisher phases) was set to above 20% for the SPR diet and 40% for the SPR+BSF diet, compared to the control. The inclusion levels of sunflower meal, brewers' dried grain, and wheat middlings in grower and finisher SPR and SPR+BSF diets were 6.30, 3.08, 3.08 and 8.00, 4.00, and 4.00%, respectively (Table 1). In the starter SPR diet, the inclusion levels for sunflower meal, brewers' dried grain, and wheat middlings were 3.58, 2.58, and 2.58%, respectively. Because of the difficulty in balancing metabolic energy, the inclusion levels of these ingredients were 0.05% higher in the SPR+BSF starter diet than in the SPR diet (3.63, 2.63, and 2.63% for sunflower meal, brewers' dried grain, and wheat middlings, respectively). These replacement levels allowed a reduction in the amount of soybean meal by 13.11, 24.37, 37.07% for the starter, grower, and finisher phases of SPR diets, respectively, as compared to the control ones (22.85% in average). In SPR+BSF diet, the level of reduction in soybean meal inclusion into the starter, grower and finisher phases of SPR+BSF diet were 26.88, 41.58, and 58.19%, respectively (42.22% in average). Nutrient composition, amino acids and fatty acids content of agri-industrial by-products and BSF larvae meal are summarized in Table 2. All diets were isocaloric and isonitrogenic and corresponded to NRC requirements (40) except with crude protein level which has been reduced in line with the sustainability approach as reported by Liu et al. (28) to apply the sustainability goal of SUSTAvianFEED project. The levels of alternative by-products in the diets were determined by taking into consideration the results from earlier studies (21, 41, 42). Each dietary treatment had 6 replicate floor pens (14 chicks/pen, 25 kg/m²). A 23L:1D lighting schedule was applied for the first 3 days, and lighting was gradually reduced to 18 h by day 7, maintaining this schedule until the end of the experiment. Standard brooding and growing period temperatures were applied during the experiment.

On d 55, eight chickens with equal numbers of each sex from each diet were randomly selected and sacrificed by cutting the jugular vein to obtain samples for histomorphometric evaluation of intestinal segments and cecal microbial flora.

Three pens per dietary treatment were included in the behavioral observations. The ethogram used in the experiment is given in Table 3. Scan sampling was used to record the number of birds performing one of the feeding, drinking, walking-standing (locomotor), and sitting-lying (resting) behaviors in each pen (43). Scans were conducted once every hour during the 18 h photoperiod on d 14, 35, and 49. In addition to the behavioral categories given above, pecking (objects, equipment, or other chicks), preening, dustbathing, leg-wing stretching, and wing flapping were also recorded in a minute at each scan time point. Due to the rare occurrence of dustbathing and wing flapping behaviors during the observations, dustbathing, wing flapping, preening, and stretching behaviors were pooled and presented as comfort behavior as suggested in previous studies (13, 44). Behavioral recordings were made by two observers at the pen level. One of the observers recorded four basic behaviors (feeding, drinking, locomotor, and resting), while the second observer counted other behaviors in a minute whenever they were performed. Therefore, the same bird might have been recorded in different categories of behavior (44). The numbers of chickens performing one specific behavior were averaged per replicate pen per day and expressed as a percentage (%) of the total number of birds.

About 2 cm of samples from the duodenum (middle part), jejunum, and ileum (both one cm away from Meckel's diverticulum) were collected for intestinal histomorphometry measurements. The intestinal segment samples were rinsed with saline, fixed in 10% formalin solution, and maintained in the formalin at room temperature until analysis. The samples were washed, dehydrated with alcohol, and cleared with xylene before paraffin wax. Tissue samples were cut by a microtome and stained using hematoxylin and eosin. Five villi and crypt were randomly selected under the light microscope, and measurements of VH, villus width (VW), and CD were performed using the Sigma Scan Pro5 program (Systat Software, Inc, CA, USA). The ratio VH to CD was calculated (45).

Cecal content samples from chickens used for intestinal histomorphometric evaluations were collected into sterile tubes, placed on dry ice, and stored at −80°C until microbiota analysis.

DNA was extracted utilizing the QIAamp DNA Stool Mini Kit (Qiagen, Germany) according to the manufacturer's protocol. The concentration (ng/μL) and purity (A260/A280 and A260/A230 ratios) of the DNA samples were measured using a Nanodrop 8000c Spectrophotometer (Thermo Fisher Scientific, USA).

Amplicon sequencing analysis was performed with eight chicken samples from each diet. PCR amplification of the targeted V3-V4 region was performed by using specific primers 341F (5′-CCTAYGGGRBGCASCAG-3′) and 806R (5′-GGACTACNNGGGTATCTAAT-3′). The Nextera XT DNA Library Preparation Kit (Illumina, USA) was utilized to generate sequencing libraries after the quantification and qualification of PCR products. The concentration of the libraries was normalized by diluting to 4 nM then libraries were sequenced on a paired-end Illumina NovaSeq 6000 platform to generate 250 bp paired-end (2 × 250 bp) raw reads. FastQC and QIIME2 were used to assess the raw data quality and read quality control, respectively. Effective tags were obtained by using DADA2 to remove primer and barcode sequences, chimeric reads, and reads with a Phred Score of < 20, hence improving the accuracy and reliability of the results. QIIME2 was used for the taxonomic determination of each Operational Taxonomic Units (OTUs) representative sequence. OTUs were annotated to obtain the corresponding species information and the abundance distribution based on the species with ≥97% similarity against the SILVA (138.1) (46). According to the results annotations of each sample, the species abundance tables at the level of kingdom, phyla, class, order, family, genus, and species were obtained. Since these abundance tables with annotation information were the core content of amplicon analysis, determination of relative abundance, and alpha and beta diversity analyses were carried out by selecting requested classification levels (e.g., phylum, genus). To clarify the richness and diversity of microbial communities in each sample, alpha diversity analyses were conducted. By using dimensionality reduction methods like PCoA in beta diversity analysis, the variations among several groups were investigated.

Behavioral data were analyzed using the general linear mixed model procedure of JMP software (Pro-13). The model included diet and age as fixed effects and their interaction with a random effect of the pen. Before the statistical analysis, Shapiro-Wilk's test was used to ensure that the normality assumption of data was met. Shapiro Wilk's test confirmed normality assumptions for feeding, locomotor, resting, and comfort behavior. Because the drinking and pecking data did not follow a normal distribution, logarithmic transformation was applied before the analysis. However, actual values were presented in the tables. When a fixed effect was found to be significant, least square means were separated with Tukey test using JMP software. P < 0.05 was considered significant. The statistical model for histomorphometric measurements included diet and intestine part and their interaction. The significance of variations in bacteria composition and community structure of groups was tested using the T-test, Kruskal-Wallis, Anosim, and multiple response permutation process (MRPP) statistical tests. P-values below 0.05 were considered statistically significant. All statistical analyses were performed with R software (Version 4.3.1; https://www.r-project.org).

Table 4 presents the effect of diet and age on home-pen behavior of Anadolu-T chickens. There was no effect of diets and diet × age interaction on the percentage of birds in any behavioral category (P > 0.05). The age of chickens had a significant effect on feeding, locomotor, resting, pecking, and comfort behaviors (P < 0.05). The percentage of chickens performing feeding, locomotor, pecking, and comfort behaviors significantly decreased with the increasing age, while resting behavior increased in chickens with the increase of age (P < 0.05). There was no effect of age on the percentage of birds displaying drinking behavior (P > 0.05).

Shorter VH and lower CD were obtained in the ileum compared to the duodenum and jejunum (P < 0.05; Table 5). The duodenum had the largest VW compared to the jejunum and ileum (P < 0.05). Significant interactions were observed for histomorphometric measurements except VH/CD (P < 0.05). Chickens that consumed the SPR and SPR+BSF diets had a decreased VH in the duodenum compared to chickens fed the control diet (P < 0.05). The VH in the jejunum was shorter in chickens fed the SPR diet compared to those fed control and SPR+BSF diets (P < 0.05). The VH increased by the SPR diet compared to the Control and SPR+BSF diets in the ileum (P < 0.05). The SPR and SPR+BSF diets resulted in a narrow villus in all intestinal segments. The SPR and SPR+BSF diets did not influence CD except duodenum, where SPR+BSF diet resulted in a shorter CD (Table 5). The diet did not affect VH/CD ratio (P > 0.05).

Venn diagrams illustrating the shared and unshared bacteria among the cecal samples of chickens fed different diets are given in Figure 1. There was a total of 808 shared Operational Taxonomic Units (OTUs). The ceca of chickens fed Control and SPR diets had 132 shared OTUs, while SPR and SPR+BSF had 228 shared OTUs. There were 603, 612, and 811 unshared OTUs in chickens fed Control, SPR, and SPR+BSF diets, respectively.

The major bacterial community characteristics are shown in Table 6. The richness of the bacterial community (Chao 1) in cecal samples was higher in chickens fed the SPR+BSF diet compared to those fed the Control diet, with the SPR diet showing intermediate values. No significant differences were found among cecal samples of chickens from different dietary groups for Shannon, Pielou's evenness, and Simpson indexes.

PCoA was performed to evaluate the structural difference between the microbiota of different sample groups. The PCoA, based on the UniFrac distance, including unweighted and weighted values in the cecal samples of local chickens, is presented in Figure 2. The dietary groups showed no obvious differences in the composition of cecal microbiota in the PCoA distribution (Figures 2A, B).

The most abundant phyla and genera with high average relative abundance are presented in Figure 3. At the phylum level, Firmicutes and Bacteriodetes were the most dominant phyla in cecal samples. The relative abundance of Firmicutes and Bacteriodetes was 51.7 and 27.4% for chickens fed Control, 49.1 and 34.1% for chickens fed SPR, and 41.1 and 36.1% for chickens fed SPR+BSF diets, respectively (Figure 3A). At the genus level, Bacteroides, Alistipes, and Methanobrevibacter spp. had the highest relative abundance (Figure 3B).

The bacteria composition of cecal samples at the genus level is presented in Figure 4. The abundance of Colidextribacter, Oscillibacter, and Anaerofilum spp. varied between chickens fed the Control and SPR diets. Colidextribacter and Oscillibacter spp. showed higher abundance in the ceca of chickens fed the SPR than those fed the control diet, while Anaerofilum spp. showed lower abundance in SPR compared to control (Figure 4A). Desulfovibrio, Ruminococcus torques, Lachoclostridum, and Anaerofilum spp. were significantly more abundant in the ceca of chickens fed the Control diet than those fed the SPR+BSF diet. In contrast, Rikenella and Colidextribacter spp. were more abundant in chickens fed SPR+BSF than those fed the Control diet (Figure 4B). When comparing cecal content from chickens from SPR and SPR+BSF, Desulfovibrio, Ruminococcus torques, and Lachnoclostridium spp. were significantly more abundant in chickens fed SPR than those fed SPR+BSF, and it was vice versa for Rikenella spp. (Figure 4C).

In this study, we did not find any significant effect of diet on behavioral traits examined. Our hypothesis was that the inclusion of agri-industrial by-products and their incorporation with BSF larvae meal could affect the behavior of chickens. The results did not confirm the study hypothesis. The crude fiber content of SPR and SPR+BSF diets was not far beyond acceptable levels for broilers and this may partly explain absence of diet effect on the behavior.

In earlier studies, the impact of the BSF larvae provision on the behavior of broilers was found to increase activity-related behavior when insect larvae were provided separately from the diets as an enrichment (12, 14). Ipema et al. (13) reported that scattering either dry or live larvae through the pen increased the activity of broilers compared to the control group, which did not include BSF. Increased locomotor activity could be associated with better leg health and welfare (47, 48), and low activity has been considered one of the possible causes of impaired leg health thus welfare (49). In our study, BSF was included in the diet as a larvae meal. Therefore, the absence of any impact of the SPR+BSF diet on behavioral traits would be expected. Indeed, Ipema et al. (13) reported that the time spent for drinking, walking-standing, resting, and foraging behavior of commercial broilers fed BSF larvae meal and oil incorporated diet were similar to those fed the control in accordance with our results. Overall, it is clear that SPR and SPR+BSF did not result in a negative effect on the behavior of broilers under the experimental conditions.

Mainly, the VH and area are associated with high levels of digestible nutrients in the diet (50). Increases in the VH indicate an increased nutrient absorption area, which may allow better growth performance (20). Since, the growth rate of chickens fed the SPR and SPR+BSF diets was found to be similar to those fed the Control diet (4), we hypothesized that the SPR and SPR+BSF diets would not negatively affect the morphological characteristics of the intestine. Indeed, no previous study has investigated the effect of sunflower meal, brewers' dried grain, wheat middlings, and BSF meal in the same diet on intestinal histology.

Decreased VH and CD in duodenum and jejunum were reported in broilers fed at increasing levels, from 70 to 210 g/kg, of sunflower meal (51) and from 50 to 200 g/kg of sunflower cake (52). Brewers' dried grain inclusion higher than 120 g/kg reduced jejunal VH (53). There are conflicting results regarding the effects of BSF on the histomorphometry of the intestine, depending on the strain, the inclusion level, and processing method, e.g., BSF live larvae, BSF whole larvae meal, or defatted or oil. He et al. (54) reported that supplementing the diet of Xuefeng black-bone chickens with from 1 to 3% BSF larvae meal might benefit the intestinal histomorphometry while 5% could decrease VH and VH/CD of the jejunum. It has been reported that inclusion of 3, 6, and 9% of BSF larvae meal did not affect duodenum, jejunum, and ileum histomorphometry in laying-type chicks (55). Dabbou et al. (23) found no effect of a diet including 5% of BSF-defatted meal on VH, CD, and VH/CD ratio; in particular, 15% of BSF-defatted meal lowered the VH/CD ratio. However, the replacement of soybean oil with BSF oil reduced CD (56). Contrary to this finding, Schiavone et al. (57) found no effect of BSF fat inclusion on the VH and CD of broilers' duodenum, jejunum, and ileum.

In our study, while the SPR and SPR+BSF diets reduced the VH in the duodenum, the jejunum and ileum, VH of chickens fed the SPR+BSF diet was similar to those of chickens fed the Control diet. This result may indicate that cell mitosis activation in the jejunum and ileum of chickens fed the SPR+BSF diet was similar to those fed the Control diet. On the other hand, the SPR diet resulted in the highest villi in the ileum. This different effect of diets on VH may be due to the fiber type differences among the diets, since different fiber types were reported as a determining factor in intestinal development (58). A higher VH/CD ratio is associated with better nutrient absorption (59) and can be used to determine intestinal integrity and evaluate the bird's response to diets (60). Notably, the diets did not influence the VH/CD ratio. De Verdal et al. (61) reported that VH/CD ratio decreased from 7.73 to 4.94 from the duodenum to the ileum, similar to the VH/CD ratio obtained in the present study. Considering our previous results (4) showing that chickens' growth performance and feed efficiency were not influenced by SPR and SPR+BSF diets, we assume that absorption capacity was similar in all dietary groups.

The diet is one of the factor contributing the composition of the gut microbiota, including the ceca (62). It is assumed that replacing soybean with agri-industrial by-products and BSF larvae meal will positively affect the cecal microbiota of chickens, depending on the fiber and amino acid contents of the dietary composition. In the present study, we have examined the modifications in the cecal microbiota profile of chickens by utilizing 16S rRNA gene sequencing. The higher Chao1 index, which estimates total species richness, in chickens fed SPR + BSF than those on the control diet indicated a higher cecal microbiota richness in the SPR + BSF chickens, but the overall complexity of the microbial community was stable. Furthermore, PCoA analysis showed that partial substitution of soybean with SPR and SPR+BSF did not cause a significant compositional change in the cecal microbiota; therefore, the microbiota constitution of the samples did not reveal any evident clustering. This result could potentially arise from the shared common environment in which the chickens were raised. Additionally, the diets, although differing in content, may have offered similar overall nutritional profiles and microbial substrates. The cecal microbiota may also have a robust core microbiome that sustains stability despite dietary changes, especially when dietary changes are not significant. Furthermore, it is possible that the chickens were at a developmental stage where their microbiota had already stabilized, and that the depth of sequencing and sampling was insufficient to identify minute variations in the microbiota composition.

Firmicutes, Bacteroidetes (also known as Bacteroidota), and Proteobacteria constituted the predominant phyla, making up nearly 80% of all bacterial populations across all dietary groups. This finding aligns with findings from previous research (30, 63, 64). Furthermore, Bacteroides, Alistipes, Methanobrevibacter, Akkermansia, and Lactobacillus spp. emerged as the first five most prevalent genera in all samples, consistent with observations in similar studies (30, 63, 64). Bacteroides and Alistipes spp. in the cecum are related to dietary fiber fermentation, producing acetic acid, and are considered beneficial bacteria for the gastrointestinal system. Methanobrevibacter spp. correlates with avian performance-related outcomes (65). Metabolic activation of Lactobacillus may improve intestinal health by lowering pH and thus play a role against pathogenic infection (66).

Using the agar plate technique, Yaqoob et al. (33) showed that partial replacement of soybean up to 9% of sunflower meal increased cecal-beneficial bacteria, such as Lactobacillus and Bifidobacterium spp., while there was no effect of dietary treatments on cecal microbial counts. It was reported that a diet containing 25% sunflower meal decreased Ruminococcaceae and Lachnospiraceae in chickens (67). In laying ducks up to 20% sunflower meal replacement reduced Spirochaetes, which can cause enteric disease (68). Brewers' yeast increased Bacillus and Enterococcus spp. in excreta in broilers (69). In the present study, 16S rRNA gene sequencing has revealed that chickens fed the SPR diet showed notably higher levels of Colidextribacter and Oscillibacter spp. compared to those on the control diet. The increased abundance of these bacteria has been associated with elevated SCFA levels and decreased TNF-α levels, which indicate improved gut health (70). The increased SCFAs producing bacteria in the cecal content of the chickens fed the SPR diet could be related to higher fiber content. In addition, both Colidextribacter and Oscillibacter spp. were shown to be associated with healthy liver (71). Colidextribacter spp. can also promote inosine production, which helps to regulate inflammatory responses and maintain the integrity of the intestinal mucosa (72–74). In light of these findings, it could be concluded that the SPR diet positively regulated cecal microbiota. The higher abundance of Anaerofilum spp. in chickens fed the Control diet compared to the chickens fed the SPR diet may be related to the higher percentage of abdominal fat (75); however, abdominal fat weight was not measured in the present study.

The inclusion of BSF larvae meal or oil in chicken diets has been shown to affect the cecal microbiota of chickens in many studies (34, 76). However, in most studies, the BSF effect on microbiota depends on the inclusion level and BSF feeding duration/period. It was shown that 5% of BSF meal inclusion positively influenced the cecal microbiota, increasing beneficial bacteria; however, 15% of BSF may have a negative influence on microbial complexity (77). de Souza Vilela et al. (36) reported that 20% BSF meal inclusion in the finisher diets of broilers had a minor effect on microbiota in caeca. In our study, BSF larvae meal was included in the SPR diet from the day of the hatch to the slaughter age. Compared to SPR+BSF, the control diet significantly increased the abundance of Ruminococcus_torques_group and Lachnoclostridium spp., which were associated with short-chain fatty acid-producing bacteria (28), and chickens' growth performance (78), and the abundance of Desulfovibrio spp., which contributes to the cleansing of free hydrogen formed during anaerobic fermentation (30). In SPR+BSF-fed chickens, the abundance of Colidextribacter and Rikenella spp. was higher than in those fed the Control diet. The higher abundance of Rikenella spp. in chickens fed with the SPR+BSF diet could be associated with the improvement of the intestinal flora environment and might alleviate intestinal inflammation (79). Similar changes were obtained for Colidextribacter and Anaerofilum spp. in the ceca of chickens fed SPR and SPR+BSF diets indicated that these changes were mainly based on the SPR diet.