Exploring the Functional and Metabolic Effects of Feeding Garra Fish Meal to Broiler Chickens


 The present study evaluated the metabolic and functional effects, beyond nutritive value, of feeding garra meal to broiler chickens. Three hundred twenty Sasso-breed day-old chicks were randomly assigned to four dietary treatments with either 0, 10, 20 or 30% garra meal added on top of a formulated starter and grower basal diets. The experiment lasted for 42 days. There was a gradual increase in feed intake and body weight gain with increasing garra meal addition (p < 0.05). Broiler chickens fed 30% garra meal were more efficient in converting feed to body weight and yielded the highest carcass weight (p < 0.05). Crude protein and crude fat digestibility coefficients were higher with the garra meal receiving groups with the highest values (74.8 ± 1.6 and 92.3 ± 0.6) recorded with 20% and 30% garra meal additions, respectively (p < 0.05). The increase in individual and total esterified carnitine concentrations in dried blood spots demonstrated the elevated metabolic rate with increasing garra meal addition (p < 0.05). A better supply of glucogenic substrate to the citric acid cycle was seen with garra meal addition due to the increase of propionylcarnitine to acetylcarnitine ratio (p < 0.05) while no effect was observed on ketogenesis as measured through the 3-hydroxybutyrylcarnitine to acetylcarnitine ratio (p > 0.05). Yet, it likely showed that part of the higher amino acids (Leucine, Methionine and Citrulline) uptake due to garra meal was used as glucogenic substrate (p < 0.05). Histo-morphology showed gradual increases in villus height, crypt depth and their ratio in the proximal parts of the intestine (duodenum and jejunum) with the opposite results observed in the more distal part (ileum) with garra meal addition (p < 0.05). It can be concluded that garra meal strongly improved broiler performance when added to a plant-based diet and only few parameters warranted for caution when using more up to 30% garra meal addition. Beyond growth performance, garra meal generated a shift to a more efficient digestion and nutrient metabolism, and induced changes in edible muscle that may aid in solving shortages in the Ethiopian human diet.


Introduction
Fish meal has been a successful ingredient in poultry diets throughout the world. In some regions such as the European Union, its use has been banned together with meat and bone meals after the BSE crisis in 1996 (Eliaszewicz 2004;Thiry et al. 2004). Further, shmeal has been criticized as a non-sustainable feed resource for aquaculture. Yet, in speci c situations, by-catch of small sh from sheries is still discarded without taking the opportunity to use it as a feed resource. In Ethiopia, an indigenous sh (Garra spp.) is commonly considered as an unimportant waste from sheries. Yet, this small sh can be a valuable source of nutrients (Bayissa et al. 2021a).
Despite of their presence in some Asian and African countries (Getahun and Stiassny 1998)  . As long as the by-catch of garra continues and its addition in the human diet is still minimal, the potential of this sh as a resource for animal feeding requires attention.
The present poultry diets in Ethiopia are usually imported and based on cereals (Ebsa et al. 2019). The latter contain anti-nutritional factors such as di cult-to-digest bre fractions and may have imbalanced amino acid and mineral pro les if not properly formulated (Fickler 2002;Blair 2018). Using local garra "waste" to produce shmeal could improve the ecological sustainability of poultry diets in Ethiopia, and improve their nutritive value.
Fish meal is indeed known as a highly digestible nutrient source, in particular protein, with a wellbalanced amino acid pattern that can support growth of the chickens (Karimi 2006). Despite these interesting nutritional characteristics of sh meal, its addition may need to be limited depending on the accumulation of problematic compounds. Poor conservation may easily lead to putrefactive compounds such as biogenic amines (Auerswald et al. 2006;Feng et al. 2016). In most studied cases, these products are negatively associated with intestinal health (Apajalahti and Vienola 2016). In particular, the addition of garra may exert biogenic amines (e.g. histamine) that can increase intestinal permeability in the consumer chickens. Higher levels of addition can be related with increased histidine decarboxylation effects by L-histidine decarboxylase, hence histamine levels (Holecek 2020). In general, chickens have been reported to have low histamine levels and its concentration can considered as an indicator of food quality to ensure consumer safety (Mejrhit et al. 2018). To thoughtfully apply garra meal in poultry diets, the optimal addition level needs to be determined, where a good balance is found between performance and gut health. Reports of recent studies indicated that high levels of sh meal induced decreased abundance of intestinal microbiota compositions such as butyrate-producing-strains of the Ruminococcaceae and Lachnospiraceae families, and lactic acid-producing bacteria (Lactobacillus johnsonii) when broiler chicks are fed sh meal-based diet (Antonissen et al. 2016).
Also the lipid fraction may trigger opposite effects: on the one hand, the high proportion on long-chain n-3 fatty acids may bene t the health of the chickens and their consumers, but these lipids are also highly susceptible to peroxidation, leading to increased dietary content of oxidative agents (Amaral et al. 2018; Alagawany et al. 2019). As a third example of nutrients that may determine the optimal addition level of garra meal in poultry diets, the concentration and bioavailability of important minerals may need consideration. A large proportion of Ethiopian infants suffer from several micromineral de ciencies, such as iron and selenium (Mekonnen et al. 2005). Preliminary data demonstrated that those minerals are quite high in garra, and one may assume that their bioavailability will be high because they are mainly chelated (e.g. iron in haemoglobin) or covalently bound (e.g. selenium in selenomethionine). Recent works on the Ethiopian garra nutrient composition have reported a higher zinc and iron concentration than in other sh in the same aquatic environment (Bayissa et al. 2021a).
Because small garra sh by-catch is still discarded as a waste in several locations, the use of its sh meal in poultry nutrition could be a sustainable solution to reduce the need for dietary protein source in the sector. We here investigate how sh meal made of garra, a small sh species endemic to Ethiopia, may affect nutrient metabolism and health parameters beyond overall performance in broiler chickens.
Knowledge on the metabolic effects of garra meal can identify additional bene ts or disadvantages beyond growth performance that are often neglected in similar poultry trials. Hence, the present study chooses an approach of on-top feeding of garra meal so as to investigate the impact of graded additions of garra meal to broiler diets on broiler performance, oxidative stress (malondialdehyde), in ammatory challenge (histamine), nutrient metabolism, and functional feed aspects (minerals and other compounds).

Materials And Methods
Garra-based poultry feed formulation The experimental feed formulation and feeding trials were carried out at Mekelle University, Ethiopia. Garra (G. dembecha) sh were caught from several water bodies and reservoirs of the Tigray region, Northern Ethiopia. Whole sh were uniformly distributed for drying on a triangular shaped entirely mesh-enclosed drier stand (2 m x 12 m x 2.5 m) established on the roof of a 15 m high building and sundried for a xed time period of 5 ± 1 d. The drying environment was with maximum average temperature of 28 ± 1.4 o C and wind speed at 2.72 m/s across the drying periods, December, January and February 2019/20. During the drying period, whole sh were uniformly turned over using a brush-ended clean stick to facilitate drying. Dried garra meal was collected and milled to acceptable smaller size using a grinder machine and made ready as feed substituent. Commercially available formulated broiler starter (d 0-21) and nisher (d 22-42) feeds were purchased from EthioChicken Plc. For both growth phases, a 10, 20, and 30% garra meal addition to the formulated commercial diets were prepared. We deliberately chose for addition of the garra meal instead of substitution with other ingredients to obtain the same nutrient pro le. Both scenarios inevitably lead to confounding factors, since we wanted to know the pure effect of garra meal, we opted for addition, in the awareness that this implied different dietary nutrient The experiment involved 320 day-old broiler chickens, randomly divided over 20 indoor oor pens with 16 chickens each (at a density of 0.28 m²/chicken). Each pen was randomly attributed to one of four dietary treatments, i.e. 0, 10, 20 and 30% garra meal additions on top of the broiler starter and nisher diets. Broiler chickens (Sasso-C44 breed) were purchased from hatchery supply of EthioChicken Plc.
Pens were considered as experimental units. All pens were designed to be similar in all outlines, with concrete oor having wood shavings as bedding at a depth of 5-10 cm, and built in accordance with the international and institutional research ethics and welfare requirements for experimental animals. Each pen was set with separate feeding and watering troughs, and an electrically controlled warming 200 W bulb lamp hung from the top of each pen. Indoor heating temperature was controlled using wall-xed digital thermometers and maintained at 30 ± 1.0 °C across the starter periods of the experiment. All equipments were properly cleaned and disinfected using detergents and the entire indoor pens were fumigated with a mix of potassium permanganate and formaldehyde (500 g : 250 ml) 10 d before stocking of the experimental chickens.
All experimental chickens were procedurally vaccinated against Infectious bursal (Gumboro) disease, Newcastle disease and Fowl typhoid as per the manufacturer's recommended vaccination protocols for day-old chickens. The experiment lasted for 42 d and individual chickens were fed on experimental unit bases. Weighed amounts of feed were given on daily basis and feed leftovers were collected, cleaned, airdried, and then weighed and recorded daily before the next morning feeding, at 08:00am. Feed and clean water were given ad libitum and chickens were allowed to stay indoors throughout the experimental period.

Measurement of growth performance indicators
Performance indicating factors were assessed for chickens in each experimental unit. Following fasting for 12 h, chickens were weighed on weekly basis on d 0, 7, 14, 21, 28, 35 and 42 in the morning before feeding starts at 08:00am. The average weekly feed intake (g/wk), body weight gain (g/wk) and feed conversion ratio (FCR, g/g) were evaluated at each sampling time point following standard procedures. Mortality cases were checked and recorded from every replicate of each treatment group (n = 80) on a daily basis and correction factors were made for all dependent parameters across the study period.
Mortality was calculated as follows: Where: N stands for the number of dead chicken in every replicate of each treatment group, 80 is the total number of chickens per group.
Evaluation of apparent nutrient digestibility using titanium (IV) oxide as a biomarker Marker-feed preparation and feeding Starting from d 22 onwards, titanium (IV) oxide (VWR Chemicals ® , Belgium, PN84853, > 99.0% purity) was supplemented into each experimental diet of grower chickens at 0.4% addition. To assure homogenous mixing of the marker, premixing was applied on a small amount of diet (1% of the total diet). Feeding with the marker diet started on the rst day of the growth phase (d 22) and continued till d 42 of the experimental period.

Excreta collection and processing
Prior to sampling, all wood shavings were removed and replaced with plastic mats for easier collection of excreta. Excreta sampling from each experimental unit was done for six consecutive days of the last week of the experimental period, d 37-42. Excreta were collected using a blunt-at scraper from the oor quantitatively for 24 h at 2x/day (06:00 and 18:00), pooled and weighed daily and kept frozen (-20 ˚C) during the sampling days. Only clean and uncontaminated excreta were collected from several predetermined locations of each pen. Finally, samples were pooled per pen, homogenized with a mixer (1-2 min), and sample of known weight (about 300 g/pen) was taken in plastic bags from each experimental unit. Excreta samples were then freeze-dried to stable weight and kept at -20 ˚C till analysed for their nutrient and marker concentrations (Schaafstra et al. 2017). Feed and water were given ad libitum across the collection period.

Measurement of apparent digestibility coe cients
Freeze-dried excreta samples and the experimental diets, 0, 10, 20 and 30% garra meal formulations, were all analysed for their individual nutrients. The DM, CP and EE contents were evaluated according to AOAC (1990). The CP contents were evaluated using the Kjeldahl method as described by AOAC (2001) and the TiO 2 contents as per the procedures stated by Myers et al. (2004). All samples were evaluated in duplicates at Ghent University, Belgium. Obtained mean values were used for the calculation of apparent digestibility coe cients of individual nutrients using the biomarker and the following formula:

Evaluation of chicken blood biochemical pro les
Blood sample collection and processing Blood samples (5 mL/chicken) were drawn from wing veins of three randomly selected chickens from each experimental unit at the end of the starter (d 22) and grower phases (d 43). Blood samples were allowed to coagulate at room temperature for 15 min, then centrifuged (1,500 × g for 10 min) and serum were collected in Eppendorf tubes and stored at -20 °C until analyses. The same chickens were traced and sampled for their serum in round two blood sampling.

Measurement of biogenic amines and oxidative stress biomarkers
Serum samples (100 µL) were evaluated for histamine concentration. Following sample preparation and acylation, histamine quanti cation was performed using Enzyme-Linked Immunosorbent Assay (ELISA) following the manufacturer's kit protocol (Ridascreen ® Histamine (enzymatic), Art No. R1605) at Ghent University, Belgium. Serum (150 µL) malondialdehyde concentrations were analysed as thiobarbituric acid reactive substances (TBARS) according to the method of Grotto et al. (2007) at Lanupro, Ghent University.

Acylcarnitine pro ling
For the evaluation of shifts in nutrient metabolism, pro ling of acylcarnitine concentrations were done. A drop (50 µL) of blood was taken from each of the whole blood samples collected and placed on circular specimen collection papers (Whatman protein saver cards, 903 ™ , UK). Each specimen kit was allowed to dry, and later sent to Ghent University Hospital (Laboratory for Clinical Chemistry) for their biochemical assay.

Evaluation of intestinal histo-morphological changes
Intestinal sample collection At the end of the experimental feeding period, following 12 h fasting, the traced 3 chickens from each experimental unit were sampled for their intestines on d 42. Chickens were weighed and then euthanized humanely using sodium pentobarbital (IP, 28 mg/kg BW) to death. The entire small intestine segment was exposed under aseptic post-mortem examination.
A xative solution (mixture of glutaraldehyde [3%] and paraformaldehyde [4%] xative solution in 0.1 mol/L cacodylate buffer [pH 7.4]) was prepared and injected into the intestinal lumen of the middle part of each intestinal segment. The entire small intestine spanning from the gizzard to pancreatic and bile duct (duodenum), from bile duct to Meckel's diverticulum (jejunum), and from diverticulum to the ileocaecal-colonic junction (ileum) was then excised and placed into separate bottles containing the same xative solution. Similarly, the cecum part were also separately removed and placed in the same xative solution until morphologically evaluated.

Histo-morphological examination
Gut tissue samples (2cm long) were collected from the middle part of each intestinal segment, dehydrated under graded alcohols and xylene, and then embedded in para n. Serial para n sections (5 μm) were made from samples and kept at 37 °C for 12 h or more. Then after, sample sections were prepared in slides and stained with haematoxylin-eosin stain following a routine protocol. Finally, the mucosal structures were observed under a Nikon phase-contrast microscope and examined using MicroComp integrated digital image-analysis system with the help of Microscope camera. The villous height and crypt depth of 10 well-oriented villi were measured per section. The average value for each tissue was obtained from 3 sections per tissue sample. That is, 10 villi/crypt per section and 3 sections per sample were measured and the average value used for statistical analysis (Fan et al. 1997). Furthermore, the villus height to crypt depth ratio was also calculated for each value obtained.

Evaluation of intestinal permeability using the everted sac test (ex vivo)
The effect of garra meal supplements in the intestinal permeability were evaluated. For this, 20 cm segment of the hind intestine (ileum part) was cut and washed in phosphate buffer saline (PBS) solution along the hollow, weighed, and then kept in an ice water mounted petri dish. The hollow segment was then inverted using a small crochet needle inserted into the lumen. One part of the lumen was ligated while lled with the PBS solution from an elevated container through the other end of the sac and then ligated on either side. The sacs were fully immersed and incubated in a glass beaker lled with PBS solution containing a cobalt marker with a constant oxygenation through aqua-aerators. The glass beakers were kept in an open water bath set up at 39.3 ± 0.8 °C for an hour. Then after, sacs were taken out, rinsed with water, and 10 ml of the intestinal uid contents were collected with syringe into a tube and stored at until analysed. Similar 10 mL sample were collected from the ringers' solution with markers at the start of the experiment for conjoint evaluation. Spectrophotometric reading of cobalt concentration was done at 640 nm (Hamilton 2014).

Data management and analysis
The different data obtained from performance indicator parameters, blood biochemical pro les, and intestinal measurements were entered into a spreadsheet Excel separately and analysed accordingly. Data from experimental diets (four percent levels) were subjected to one-way analysis of variance (ANOVA) and analysed using SPSS software (version 27.0). Every replicate was considered as the experimental unit for all statistical analyses. In either case, results were expressed as treatment means with their pooled standard error of the mean (SEM). A P < 0.05 probability value was described to be the level of signi cance and the notable differences between the treatment groups were determined via Tukey multiple comparisons test.

Growth performance indicators
Mortality throughout the experimental period was low: three dead chickens in the control group and one dead in the 10% garra meal group, all in the rst week, with no apparent relationship to the treatments.
There was a gradual increase in feed intake and body weight gain with increasing garra meal addition (P < 0.001) ( Table 5).
Broiler chickens fed 30% garra meal were more e cient in converting feed to body weight (1.7 ± 0.01 g feed/g gain, P < 0.001) across the feeding experiment with the lowest feed conversion ratio observed at the starter phase (1.6 ± 0.02 g feed/g gain). Likewise, the highest carcass weight (1103 ± 36 g, P < 0.001) was recorded at the highest garra meal addition with similar carcass dressing percentages among treatment groups (P = 0.218) ( Table 5).

Apparent nutrient digestibility
Apparent crude protein digestibility was higher with the garra meal receiving groups with the highest value recorded with 20% garra addition (74.8 ± 1.6, P = 0.006). Similarly, crude fat digestibility was higher with the gradual garra meal additions with the highest digestibility recorded by 30% garra meal addition (92.3 ± 0.6, P < 0.001). No difference in dry matter digestibility was noticed among the treatment groups (P = 0.068) ( Table 6).
Dried blood spot concentrations of some free amino acids (leucine, methionine and citrulline) increased with garra meal addition whereas other amino acids (phenylalanine, tyrosine and ornithine) decreased with garra meal addition (Fig. 3).

Serum histamine and malondialdehyde levels
Chicken serum samples from all treatment groups were found to contain histamine far below the detection limit, 5 mg/mL, and no difference was observed among treatments (P > 0.05). However, detectable gradual increase in malondialdehyde levels with garra meal additions were noticed in the starter growth phase, 21 d, of chickens with the highest level showed by 30% garra meal fed groups (9.5 ± 0.8 nmol/mL, P = 0.054). No signi cant effect of garra meal addition was seen at the end of grower phase, 42 d, of chickens (P = 0.726) although all values recorded were above the detection limit, 2.5 nmol/mL (Table 7).
Intestinal histo-morphology Garra meal addition elevated the villus height in the duodenum and jejunum compared with the control group (P = 0.005). However, a gradually decreasing villus height of ileum was noted with increasing garra meal addition (P = 0.005).
The small intestine crypt depth of the duodenum and jejunum in the garra meal groups were deeper than the control group (P = 0.003), while there were no differences between the control and 30% garra meal addition (P > 0.05). The crypt depth of ileum was deeper in the control groups than the garra meal supplemented groups (P < 0.003), while there was no difference among the 10, 20 and 30% garra meal additions (P > 0.05).
The villus : crypt ratio of the duodenum, jejunum and ileum in the chickens with garra meal supplementation was higher than the control group (P < 0.001), except for the ileum of 20% garra meal addition being the lowest ratio. The villus : crypt value of the duodenum with 30% garra meal addition was the highest among all other values computed.
The cecal internal endothelial folding/crust heights were higher for the garra meal addition groups than that of the control groups (P = 0.031). However, no difference was observed between 20 and 30% garra meal addition groups (P > 0.05) ( Table 8).

Everted sac test
The everted sac test was used to re ect the absorptive capacity of the intestinal lumen of broiler chickens fed diets with graded additions of garra meal. The cobalt (Co) absorption across the lumen was higher in the 30% garra meal addition than in the group without garra meal (P = 0.007) (Fig. 4).

Discussion
The clear gradual increase in performance of the broiler chickens with addition of garra meal was according to expectations, given the high protein and energy content of this sh meal made of garra. It is quite remarkable that adding garra meal on top of an assumedly balanced diet still induced a far better growth performance. Garra meal indeed appears as a balanced diet ingredient with highly digestible nutrients and high palatability.
Less obvious than the increased growth, was the observed increase in feed intake with increasing garra meal, as in most situations, chickens will reduce feed intake with increasing energy density (Mikulec et al. 2004, Karimi 2006, Incharoen et al. 2010). Yet, other factors than just energy density are regulating feed intake. This was not the focus of our study, but we refer to for instance the minimal total discomfort theory (Forbes 2003) that showed that any nutrient or other requirement can affect voluntary feed intake. The increased intake can explain the higher growth rate with increasing garra meal, but the associated higher feed utilisation e ciency emphasizes the balanced supply of digestible nutrients from garra meal. As tackled further in this discussion, garra meal is also rich in speci c functional nutrients such as carnitine and citrulline that may support the animal's metabolism.
Clearly, the maximum for protein deposition was not yet reached with the basal diet without garra meal, because the birds not only grew faster with garra meal addition, but still had the same dressing percentage (and even numerically higher). Also the unaltered malonylcarnitine to acetylcarnitine ratio in the dried bloodspots con rms that fat synthesis remained independent of garra meal addition. Malonyl coenzyme A is indeed a known marker for lipogenesis (Schooneman et al. 2013) and the higher production of malonylcarnitine is a step forward in the synthesis of fats in the citric acid cycle (Yang et al. 2019). It also matches the lower feed conversion ratio due to garra meal addition, which implies that the nutrients and energy from garra meal were more e ciently used than that of the basal diet. An important factor to explain this effect is the combination of highly digestible ingredients with a low bre content.
More garra meal addition led to higher lipid digestibility, even up to the 30% addition, but protein digestibility seemed to reach an optimum at 20% addition, likely because the capacity for protein digestion was reached at that point. A concomitant concern of suboptimal protein digestion is the consequences for gut health, because providing more protein as substrate to the cecal microbiome may lead to damaging fermentation end-products such as biogenic amines (Apajalahti andVienola 2016, Mejerhit et al. 2018). In the present study, the serum concentrations of histamine, a major biogenic amine, and malondialdehyde, an oxidative stress indicator, could however not show such a negative effect of high garra meal addition. The only measurement that pointed to a possible optimum for garra meal addition, was the increased gut permeability at 30% garra meal addition. Increased gut permeability, as proven ex vivo in the present study, could evoke increase in intestinal mucosal paracellular permeability enhancing entry of pathogenic bacteria and/or their toxins into the systemic circulations that could trigger diseases (Chelakkot et al. 2018, Hollander andKaunitz 2020).
Despite this impact on gut permeability at the higher garra meal addition level, the intestinal development was prominently affected: gradual increases in villus height, crypt depth and their ratio in the proximal parts of the intestine (duodenum and jejunum) typically re ect a higher digestive and absorptive capacity when adding more garra meal. Irrefutably, Incharoen et al. (2010) have found that the villus heights in duodenum and jejunum were lower in chickens fed low-crude protein diets with a lower mitotic number in crypt cells than those on normal dietary crude protein. In fact, the intestinal endothelium, with its villi, is a dynamic environment and often changes to adjust itself in response to intestinal contents, including dietary compositions (Yamauchi 2002, Prakatur et al. 2019). Hence, longer intestinal villi are associated with increased surface areas and capacity for absorption of nutrients (Izadi et al. 2013) and deeper crypts re ect active endothelial cell mitosis beneath each villus for the regeneration of new villi in the lumen (Hamedi et al. 2011). On the other side, lower villus : crypt ratios are associated with fewer absorptive cells and numerous epithelial secretory cells. The latter are involved in excess mucus section that can reduce nutrient absorption and lead to increased intestinal coping functions (Hamedi et al. 2011). In the more distal part (ileum), the opposite was observed, likely referring to the fact that higher proximal digestibility led to less digesta in the distal part of the chickens digestive tract. Ileum is indeed a site where water and minerals are absorbed while the absorption of proteins, carbohydrates and fat mainly takes place in the jejunum (Svihus et al. 2010, Ariyadi et al. 2019).
The avian caecum is particularly important to process uric acid from protein breakdown (Clench 1999), hence the increase in cecal crust height with more garra meal addition. The larger and numerous cecal folds can further be related to its involvement in the digestion of small protein loads being pushed against the cecal sphincter from the intestinal loads and/or their absorption (Clench 1999, Majeed et al. 2009).
As a result of increased absorptive capacity and increased feed intake, garra meal addition must have induced a higher supply of nutrients in metabolism. This requires a metabolic "machinery" that can handle the elevated demand for nutrient processing. Our group has been using carnitine esters in diverse species to re ect changes in the corresponding coenzyme A moieties in nutrient metabolism (e.g. frogs: Brenes In the present study, the main acylcarnitines and their sum positively responded to increased garra meal addition, demonstrating the increased metabolic activity due to enhanced nutrient absorption and concomitant growth. The increase of the propionylcarnitine to acetylcarnitine ratio shows a relatively better supply of the often limiting glucogenic substrates to the citric acid cycle (Bayissa et al. 2021b). It is not sure if this was instrumental for improving feed utilisation e ciency because no signi cant effect was observed on the production of ketone bodies deviating from the citric acid cycle, as measured through the 3-hydroxybutyrylcarnitine to acetylcarnitine ratio. 3-Hydroxybutyrylcarnitine is an indicatory moiety of ketogenesis in the body (Xu et al. 2016). Yet, it likely shows that part of the higher amino acid uptake due to garra meal was used as glucogenic substrate. Methionine is an example of a glucoplastic amino acid that can enter the citric acid cycle through propionyl coenzyme A converting into succinyl coenzyme A (Krajcovicová-Kudlácková et al. 2000), and its free concentration in the dried bloodspots was clearly rising with garra meal addition. Yet, even the concentrations of the pure lipogenic amino acid leucine (Schooneman et al. 2013) went up with garra addition, whereas other free amino acids such as tyrosine and its precursor phenylalanine decreased when more garra meal was added. This suggests that these free amino acids pro les are not just representing metabolic requirements but also simply an in uence of the dietary amino acid composition.
At least for citrulline and carnitine, two amino-acid-like substances, their dose-responsive increase in concentration must have been due to their supply through garra meal. Carnitine is not present in plant material, whereas most animals including sh are rich sources of carnitine (Li et al. 2020). There was a marked increase in the free as well as bound carnitine in our study. The reduction in the bound to free carnitine ratio that was seen with higher garra meal addition could be interpreted as a relative drop in metabolic rate, but in this case, this was clearly due to the much higher absorption of carnitine from the garra meal that may even have supported metabolic e ciency. We acknowledge that we cannot claim this effect since other factors may have caused improved e ciency independent of carnitine supply, but at least studies have demonstrated bene cial effects on carnitine supplementation to plant-based diets for broiler chickens (Janssens et al. 2000, Karadeniz et al. 2008, Hrnčár et al. 2015. Citrulline cannot be synthesized in birds because they lack the urea cycle -in contrast with mammalsthat can generate citrulline from ornithine (Tsuji and Kanazawa 1987). Citrulline can be an interesting functional nutrient given its antioxidant action (Coles 2007), especially in hot environments: heat stress is a strong inducer of oxidative stress (Akbarian et al. 2016). This effect was not studied here but we hypothesize that the better broiler performance in this study when feeding more garra meal was due to nutritional effects (improved supply of digestible nutrients) as well as metabolic modulator effects (e.g. carnitine, citrulline, n-3 fatty acids). The decrease in ornithine with garra meal additions however could possibly re ect dietary dilution effect of garra meal in the experimental diets.
In conclusion, garra meal strongly improved broiler performance when added to a plant-based diet. Only few parameters warranted for caution when using up to 30% garra meal addition, but apart from digestibility and absorptive capacity, also nutrient metabolization e ciency seemed increased. The latter may have been caused by a number of functional components in the garra meal that may warrant further investigation.
Acylcarnitine pro les in serum and muscle of dairy cows receiving conjugated linoleic acids or a control fat supplement during early lactation. Journal of dairy science, 102, 754-767.    Box and whisker plots of concentration ratios of targeted acylcarnitines from dried blood spots of broiler chickens fed diets with graded additions of garra meal. C3/C2, propionylcarnitine to acetylcarnitine ratio, 3OHC4/C2, 3-hydroxybutyrylcarnitine to acetylcarnitine ratio, C3DC/C2, malonylcarnitine to acetylcarnitine ratio. Mean values are based on 3 birds per replicate and 5 replicates per treatment Box and whisker plots of free amino acid concentrations in dried blood spots of broiler chickens fed diets with graded additions of garra meal. Leu, leucine, Met, methionine, Fen, phenylalanine, Tyr, tyrosine, Orn, ornithine, Cit, citrulline. Mean values are based on 3 birds per replicate and 5 replicates per treatment