Cocoa by-product inclusion in dairy sheep diet: effect on sensory, volatile and antioxidant properties of cheese

The possibility of inclusion of agro-industrial by-products in the diet of small ruminants represents both an economical and an environmental strategy for reducing waste management by industries and the cost of feeding as well as the impact of livestock farming. Large amounts of wastes from the cocoa industry are annually produced with a considerable part represented by cocoa bean shells, considered a suitable ingredient to be included in the diet of ruminants within the limits established by European legislation. The aim of this study was to assess the effect of including cocoa bean shells in the diet of dairy sheep on the sensory, volatile, and antioxidant properties of cheese. To this purpose, 20 Comisana lactating ewes were randomly assigned to 2 experimental groups: control (CTRL) and cocoa bean shells (CBS), and received alfalfa hay ad libitum and 800g of conventional (CTRL) or experimental (CBS) concentrate containing 11.7% CBS to partially replace corn and barley of the CTRL concentrate. Bulk milk collected from each group was used to produce a total of 15 cheeses per group, obtained in 5 different days of cheese-making (3 cheeses a day per group). After 60 d of aging, each cheese of each experimental group was sampled for the analyses. The results on chemical composition revealed a greater monoun-saturated fatty acids content and an increase in the nutritional indices suggesting a favorable role of cocoa bean shells dietary inclusion on the nutritive value of the cheese. The cheese sensory profile was affected by the cocoa bean shells inclusion, with more pronounced appearance, odor, aroma, and taste attributes in the product. The volatile profile showed only a few significant differences, mainly related to the cheese ripening process, and no differences were found in α-tocopherol contents in cheese fat between the 2 groups. Therefore, the coca bean shells inclusion in the diet of dairy sheep allowed to obtain a good quality cheese, without altering the characteristics associated with the typical profiles of sheep cheese. Furthermore, the use of this by-product could contribute to decrease feed costs and waste management, representing a good practice for increasing the sustainability of dairy products.


INTRODUCTION
Ewes farming assumes a different relevance according to the geographical area, the herd dimension, production systems, as well as the type of available products.In this contest, the sheep milk production is a noteworthy sector in Europe, with more than 2.9 million tons produced in 2022 (CLAL, 2024).However, being the consumption in liquid form unusual, approx.75% of the sheep milk produced is mainly used for the manufacture of several dairy products, including cheeses, yogurt, and whey cheeses (Haenlein andWendorff 2006, CLAL, 2024).
Due to their quality, high yield, and nutritional proprieties, with high concentrations of proteins, fats, vitamins, and minerals, sheep dairy products have gained considerable potential in terms of market expansion (Park, 2007;Milani and Wendorff 2011), especially for their health benefits.
Considering e.g., the higher digestibility compared with cow milk, as well as the better tolerance by individuals allergic to cow milk (El-Agamy, 2007;Pilbrow et al., 2016).
Both the nutritional aspects and sensory characteristics of sheep cheeses, have grown their consumption in the last few years (Gámbaro et al., 2017) Several strategies have been adopted for dairy sheep farming to reduce the dependence on food from abroad, the feed costs and the environmental problems related to the management of waste materials (Monitor et al., 2020;Huanca et al., 2021).
Indeed, several by-products are commonly underestimated due to lack of valuable alternative uses.Some of them are rich in nutrients and bioactive compounds, thus their inclusion in dairy sheep diets could represent a practical and sustainable option both for their recycle and valorization (Nudda et al., 2019).It is amply demonstrated that the use of by-products as alternative feeds can affect milk composition, and consequently the quality of derived products (Jaramillo et al., 2009).
Particularly vegetable and agri-food as well as food processing industry by-products have been investigated as possible alternative feeding, or partial inclusion, becoming a common practice as a way to give possible value added to the dairy products according to their nutritional characteristics as feeds.
Moreover, the use of food by-products also represents a possible way to reduce waste discharges and to minimize waste management costs by the industry (Jaramillo et al., 2010), representing a crucial aspect in global strategies for reducing the environmental impact problems (Correddu et al., 2020).
Within the serval by-products assessed as as possible alternative feeding, Cocoa bean shells (CBS) has shown enormous potential that can be used in the supply of animal feed for ruminants such as cattle, goats, and sheep, to partially replace cereals (Rebollo-Hernanz et al., 2022), representing a possible strategy to reduce feed-to-food competition and feeding cost during the dry season when pasture is scarcely available (Cornale et al., 2022).This by-product represents the superficial tegument covering the cocoa beans, generated during the bean roasting process, represent about 10-17% of the total cocoa bean weight (Hashimoto et al., 2018) and 2.1-2.3% of the cocoa pod, with an annual production of approx.700,000 tons (Okiyama et al., 2017).
The CBS is mainly composed of dietary fiber, protein, and fat, ranging from 19 to 60%, 12-18%, 2-7% respectively (Vasquez et al., 2019) as well as considerable quantities of interesting bioactive compounds, such as tannins (Badrie et al., 2015), which modify the rumen metabolism affecting animal performance and product quality (Frutos et al., 2020).Nevertheless, the factor limiting the use of CBS in ruminants is the theobromine concentration which vary according to the cocoa bean preparation and rise during fermentation (Makinde et al., 2019).High levels of this alkaloid may produce adverse effects on animal health (Adamafio, 2013).Most disease cases are reported in monogastric animals, especially dogs, while for ruminants only a few potential cases in dairy cattle.Thus, the possible toxicity effect of theobromine in ruminants is not clearly answered, it is supposed that a fatal outcome is pos-sible due to cardiac arrhythmia, respiratory failure, and a disturbed central nervous system (Klein et al., 2020).
Despite different studies have revealed positive effects of CBS inclusion in animals' diets, only few works investigated the effect of dietary CBS inclusion on milk and cheese composition in dairy sheep (Carta et al., 2020;Campione et al., 2021;Carta et al., 2022).The bioactive components of CBS could affect the nutrient characteristics and the sensory properties of animal origin products (Vasta & Luciano, 2011).It is well known that dietary composition affects ruminal fermentation thus milk compositions could be changed when byproducts are included in the diet (Chilliard et al., 2003;Romero et al., 2017) also affecting the antioxidant activity and sensory profile of dairy products (Martin et al., 2005).
However only few the studies related to the use of by-products in the diet of dairy small ruminants, evaluated the effect on cheese sensory characteristics.Particularly, Jaramillo et al. (2009) evaluated the effect of substitution of cereal grain and sugar beet pulp by citrus fruits (30% DM basis) in the ewes' diet, revealing no adverse effect on the overall sensory quality of cheeses.Caccamo et al. (2019) investigated the sensory and volatile profile of cheese produced with milk from dairy sheep fed with hazelnut skin inclusion, revealing that this by-product significantly affected the Fatty acids and sensory profiles of the cheeses.
Recently Huanca et al. (2021) investigated the inclusion of lemon leaves and rice straw by-products in the diet of dairy goats on the sensory profile of the derived matured cheese, showing no negative attributes.
In Campione et al. (2019), the effect of CBS inclusion in the diet of dairy sheep on animal performance, the fatty acid composition of rumen content, and milk and cheese composition were investigated.The results revealed lower urea levels in milk, probably related to the phenolic content of CBS, as well as lower protein and higher fat content in cheese.As part of the same experiment, the present manuscript focused on the assessment of the effect of including CBS by-product in the diet of dairy sheep on the sensory, volatile, and antioxidant properties of cheese obtained from the experiment of Campione et al. (2021).

Experimental Design, Animals, Diets, and cheesemaking
The experimental design for this study was the same as that previously described by Campione et al. (2021).Briefly, 20 multiparous lactating Comisana ewes with 80 ± 8 d in milk, balanced for body weight (BW; 65 ± Caccamo et al.: Cocoa by-product inclusion… 8kg), were randomly assigned to 2 experimental groups (n = 10), namely control (CTRL) or cocoa (CBS) group and confined in multiple pens.The trial lasted 35 d in total.After 14 d of adaptation period, each animal received chopped alfalfa hay ad libitum (particle size > 4 cm in length) and 800 g/day/ewe of a conventional concentrate with corn and barley for the CTRL group.While the CBS group received an experimental concentrate containing CBS to partially replace corn and barley.Table 1 reports the chemical composition of the experimental feeds which were analyzed as detailed in Campione et al. (2021).Offered feeds and orts were daily weighed per pen while individual milk production was recorded weekly.
Cheesemaking was carried out as reported in the previous paper (Campione et al., 2021).In short, bulk milk from each of the 2 feeding groups was daily collected during the experimental period and stored at −30°C until the quantity of ~45kg was reached and used to make 3 cheeses per day for each group.A total of 15 cheeses per group were obtained in 5 different days of cheese-making across the experimental period.After thawing at 4°C, the milk was heated at 39°C and enriched with a mixed-strain starter culture (Lyofast MW039S SACCO, Como, Italy) consists of strains of Lactococcus lactis ssp.cremoris and Lactococcus lactis ssp.lactis biovar diacetylactis.Following 10-min liquid calf-lamb rennet (Strength 235 IMCU/ml -Crerici SPA, Italy) was added (22g/100L).The curd was broken after 25 min until small grain dimension (approx.Four mm diameter) was achieved.Afterward, the curd was removed and filled into a perforated plastic basket to drain the whey until the pH reached 5.5.After 24h at 7°C, all the cheeses were put in brine (salt 20% wt/vol) for 12 h and aged in a cold room for 60 d at a temperature ranging between 8 to 9°C and 70% humidity.The scheme of the conducted experimental cheesemaking is reported in Figure 1.

Cheese Sampling and Analyses
Chemical analyses.After aging, individual cheeses were sampled for analyses and aliquots were vacuum stored at −20°C.The determination of moisture, lipid and protein content of the cheese samples were performed as reported by Bradley and Vanderwarn (2001), Gerber -Van Gulik method (ISO 1975) and Kjeldhal method (total nitrogen x 6.38), respectively.Cheese Fatty Acid Methyl Ester (FAME) preparation was performed according to Nudda et al., (2005).Individual FAME of cheese were separated and quantified as described for feeds.Fatty acid profile of cheese was then determined through transesterification using a combined basic and acid methylation, as proposed by Cruz-Hernandez et al. (2004).Briefly, 0.5 mL lipid extract was incubated at 50°C for 15 min with 1.5 mL sodium methoxide in methanol (0.5 M).After cooling at room temperature, 1 mL of 5% methanolic HCl was added and mixture was incubated at 50°C for 30 min.Then, 1 mL of 6% aqueous K 2 CO 3 was added and a triple refrigerated extraction with 3 mL of hexane at 1500 × g for 10 min was performed.The extract was evaporated under N 2 flow at 37°C and then dissolved in 1 mL of GC grade hexane.Gas chromatograph setting for FAME identification was the same as described for feedstuff analysis.Moreover, the separation of C18:1 isomers was achieved by isothermal analysis at 165°C.Individual FA were expressed as g per 100g −1 of cheese.
Odor active volatile compounds (OAC) were extracted by using a static solid phase micro-extraction (SPME) technique as reported by Carpino et al. (2004), with some modifications, as detailed below.A DVB/CAR/PDMS coated fiber (50/30 µm; Supelco, Bellefonte, PA, USA) was used to adsorb OAC from headspace samples.Ten grams of cheese were conditioned at 40°C for 40 min.Additional 40 min were required for the fiber exposition to establish the volatile compound equilibrium between samples headspace and fiber solid phase.The fiber was conditioned for 1 h at  For the gas chromatographic olfactometry analysis, an HP 6890 Series GC system (Agilent Technologies, Santa Clara, CA) gas chromatograph coupled with an olfactometer was used.Column, injection and oven setting were the same as reported for GC-MS.Trained human nose (sniffer) was used as final detector simultaneously with mass detector (Rapisarda et al., 2014).The eluted compounds were mixed with humidified air and the sniffer was continuously exposed to this source for 30 min.During the olfactometric analysis, the sniffer described the perceptions and duration of odors.
The OAC recognition was performed using the single sniff method and the sniffer was trained with 7 standard aroma compounds used to evaluate olfactory acuity (Marin et al., 1988).These compounds were selected for study because they are all naturally occurring food constituents, and specific anosmia has been reported for some of them.The sniffer had no specific anosmia for these standards.
The determination of cheese α-tocopherol and cholesterol was as described by Marino et al. (2010) and Oh et al. (2001), respectively.Both α-tocopherol and cholesterol were determined by an HPLC method using an SB-C18 column (5-µm particle size, 4.6 nm i.d.× 250 nm, Agilent Zorbax, Agilent Technologies, Santa Clara, CA, USA).The HPLC system (Waters 2695; Waters, Milford, MA, USA) was equipped with a multi-wavelength (λ) fluorescence detector (Waters 2475) using an excitation wavelength of 297 nm and an emission wavelength of 340 nm for the detection of α-tocopherol, equipped with a dual λ absorbance detector (Waters 2487) using a wavelength of 203 nm for the detection of cholesterol.The mobile phases were methanol 100% vol/vol and acetonitrile/methanol/2propanol (7:3:1, vol/vol/v) for α-tocopherol and cholesterol, respectively.All reagents used were HPLC-grade with a proven purity between 95% and 99.9% and were obtained by Sigma-Aldrich (Sigma Chemical Co., St. Louis, MO, USA).Identification and quantification of α-tocopherol and cholesterol were based on external standards obtained from Sigma (Sigma Chemical Co., St. Louis, MO, USA) with purity ≥99.6%.All chemical analyses were done in duplicate.
The degree of antioxidant protection (DAP), used to evaluate the antioxidant protection of foods (Pizzoferrato et al., 2007), was calculated as the molar ratio between tocopherols and cholesterol contents in cheese.
Sensory analyses.Seventeen assessors (8 women and 9 men) aged 38-54 years who had 3 years of experience in the sensory evaluation of sheep cheeses, were involved and the significance level was set at 0.05 to define sensory profiles of both CBS and CTRL cheeses.The pieces of cheese were served at room temperature using white plastic dishes, each marked using a random 3-digit code.The tasting station was lighted to prevent the perception of differences in colors of the samples.The samples were described by qualitative descriptive analysis (QDA), according to Stone et al. (1974).Attribute terms for evaluation of cheeses were developed by the 17 panelists using QDA methodology.Briefly, ballot development and panelist training were accomplished during 7 working sessions lasted 30-50 min.The descriptive terms developed for each major sensory attribute category are reported in Table 2.Each attribute was presented as a separate unstructured line scale that recorded panelist responses in increments of 0.1 between 1 (leftmost position) and 15 (rightmost position).The cheese samples were cubed (~1 cm each side) and were presented on white paperboard plates.The panelists also had available an entire transverse slice of each cheese for evaluating appearance attributes.

Calculations and Statistics
The general liner model procedure for repeated measures was used to test the effect of dietary treatment, time of cheese production (1 to 5) and their interaction as fixed factors on the chemical and fatty acid composition of cheese.Least square difference (LSD) was used for the multiple comparisons of the means.
For AOC data, a one-way ANOVA was applied to log-transformed values of chromatograph area to test the effect of dietary treatment.
A general linear model was used to test the effect of dietary treatment on cheese sensory profile, with dietary treatment considered as fixed effect whereas panelists and time were considered as random effects.Assessors to define cheeses profile were continuous bipolar from 1 (absent or nothing) to 15 (a lot), with the exception of both color paste and rind in which 1 was equal to white and 15 was equal to yellow, and for holes/slits dimensions in which 1 was equal to small and 15 was equal to big.LSM were compared with test the effect of the treatment one-way ANOVA.
Principal components analysis was applied to the matrix (assessor x treatment x rep rows; 27 columns) of original standardized sensory data using JMP 12 software (SAS Institute) procedure to study the main tendencies in variation between cheeses.The mostly significant 2 principal components were analyzed using a factorial analysis.

Cheese composition
Results on cheese composition, previously provided by Campione et al. (2021) who found that the feeding treatment with CBS inclusion affected the chemical composition of cheese, were reported in g 100g −1 cheese (Table 3).
Alpha-tocopherol and cholesterol contents in ewe cheeses of both feeding groups were measured (Table 3).To evaluate the effects of diet, α-tocopherol and cholesterol were compared referred to as fat content.The diet did not affect both α-tocopherol and cholesterol contents of cheese fat.In cheese fat of both feeding groups, the α-tocopherol content was on average 6.6 µg on g −1 , while the cholesterol content was on average 1.2 mg on g −1 .
The DAP values showed the highest values in CBS cheese samples compared with CTRL, 5.29 vs. 4.2, respectively.

Sensory profile
The sensory profile of both CBS and CTRL cheeses is reported in Table 4. Data showed an effect for treatment variable in different descriptors of the scale used to define the profile.Cheese's appearance was affected by the CBS inclusion in the diet.Particularly, the paste color of cheese was significantly more yellow for CBS compared with CTRL cheeses.Similarly, the dimension of the holes of the paste was bigger in CBS compared with CTRL (P = 0.02), and a tendency for the presence of holes and slits to be slightly higher for CBS than CTRL cheese (P = 0.06).
Within odor attributes only toasted and smoked were perceived as significantly higher for CBS rather than CTRL cheese (P < 0.01).
Most of the taste attributes did not show significant differences between the 2 cheeses.CBS cheese showed a significantly higher acid (P = 0.02) and lesser sweet (P < 0.01) taste compared with the CTRL.
All aroma attributes confirmed the same trends of odor attributes, with toasted and smoked aromas significantly higher in CBS cheese (toasted P < 0.01; smoked P = 0.01).Among tactile sensations, CBS cheese reported significantly higher plasticity (P = 0.007) and soft/hard perception (P = 0.048) than CTRL.Mouth sensations did not show significant differences between the cheese except for humidity, the only descriptor that was significantly lower (P < 0.001) in CBS compared with CTRL cheese.
Principal components analysis (PCA) on sensory attributes of both CBS and CTRL cheeses evaluated by panelists, is reported in Figure 2. The plots from the PCA analysis applied to sensory attributes explained 37.1% (PC1, 20.9%; PC2, 16.2%) of the total variation.Oily, paste color, toasted, savory, and spicy attributes were the main factors in the positive PC1 while humidity, mellowness, and elasticity were the dominating factors along the negative PC2 zone.The remaining attributes were located in the negative PC1 zone.
The score plot showed that most of the CBS cheeses were located within the negative PC1 zone, displaying a higher acid bitter taste, toasted, spicy, and yeast aroma as well as higher perception of holes presence and dimension.
The CTRL cheeses, mainly located in the PC2 zone were described with more mellow and humidity mouth sensations, and more plastic tactile sensations.

Odor active volatile compounds profile
A total of 6 CTRL and BCS cheese batches were analyzed by Gas Cromatography/Olfactometry (GC/O) and GC/Mass spectrometry (GC/MS).Generally, CTRL and CBS cheese samples showed poor and very similar volatile profiles (Supplemental Table S1).No significant difference in number and type of volatile compounds between CTRL and CBS groups were found.As shown by MS, all samples showed a high number of no-odorant volatile compounds.
Up to 47 different OAC in cheeses were detected for the experiment, revealed by MS belonging to the following chemical classes: Acids (8 compounds), Alcohols (5 compounds), Aldehydes (3 compounds), Alkane (1 compound), Aromatic hydrocarbons (7 compounds), Ester (11 compounds), Ketone (4 compounds), Lactone (1 compound), Pyrazine (2 compounds), and Terpene (5 compounds).Actually, not all of these OACs were detected in each cheese sample and only some were found in all cheeses.In Supplemental Table 2 OACs detected by both GC/O and MS are presented in detail for each bacth of cheeses (1-5) analyzed per CTRL and CBS.
For Acids chemical class, the cacao cheeses compared with the control presented in all batches the decanoic acid OAC with a fat rancid odor perception.For Alcohol chemical class, the phenylethyl alcohol OAC was mostly present in CTRL cheese samples, meaning a major presence of rose odor perception.For Aldehyde chemical class, the cacao cheeses compared with the control, presented in less batches the nonanal OAC, meaning a less presence of green odor perception.Ester chemical class was low in octanoic acid, methyl ester implying an orange presence in both CBS and CTRL and is even less in CBS, meanwhile the decanoic acid ethyl ester meaning fruit is present in control and treatment but   By transforming data of area (Table 5), we found out that among OACs they very similar among the 2 groups with really few exceptions.Among the Acid chemical class, Decanoic acid (fat, rancid odour perception) was marginally significant in the control rather than in the treatment group (p-value = 0.05).For Sulfur chemical class, Dimethyl sulfone was significantly higher in the control group rather than in the treatment group (p-value < 0.01).These 2 were the only OACs that presented differences between the 2 groups.°assessor scale 1 = white to 15 = yellow.°° assessor scale 1 = small to 15 = big.all the other assessors were from 1 = nothing to 15 = a lot.±p-value = 0.05 marginally significant.*p-value < 0.05.**p-value < 0.01.

Effect on Sensory Profile
Regarding the sensory profile, the statistical analysis of the attribute ratings collected from 17 trained panelists evidenced that sheep feeding had some effects on several descriptors of the 2 groups' cheeses.
The cheese paste color was the appearance parameter that showed a higher score in CBS cheese than CTRL, with a more yellowness note.Color is a very important attribute of foods and serves as an index of quality (Fox et al., 2017) since consumers associate specific colors with certain flavors (Wadhwani and McMahon, 2012).Usually, cheese made from sheep milk are whiter in comparison with their similar made from bovine milk, since cattle transfers carotenoids to adipose tissue differently from sheep (Fox et al., 2017).Particularly, β-carotene due to its cleavage, catalyzed by enzyme activity to retinal in the sheep liver, is not measurable in sheep's milk, therefore not affecting the color changes in the resulting pecorino cheese (Cardinault et al., 2006;Serrapica et al., 2020).These possibile changes in color could be related to the chemical changes during ripening, such as proteolysis of casein, becoming less white, and the prevalence of reflecting components (Johnson, 1999).However, also the fat composition can affect the change in the color of the cheese.According to Rohm et al. (1997) the higher total MUFA content shown by CBS cheese than CTRL can be regarded as the most important descriptor of cheese body color, and the rise of fat lipolysis with ripening progress could be responsible for the increase in the yellow color (González-Martín et al., 2020).
The other appearance attributes related to holes were perceived as greater in CBS cheese as a possible consequence of the microbial activity coming from the aging process, normally related to slight gas production (Fox et al., 2000).
It was also observed that cheeses from the CBS diet presented greater intensity of odor, aroma, and taste and the characteristics of feeding CBS reflected in sensations of toasted and smoked.This could be related to the fact that the secondary compounds that derive from the feeding can be easily transferred to milk or cheese (Wiedenhoeft and Barton, 1995).
Regarding taste attributes, although acid and sweet parameters resulted significantly higher in CBS than CTRL cheese, the score were numerically similar between the 2 groups.
The higher tactile plasticity and hard sensation as well as the lesser humidity mouth sensation in CBS cheese could be related to lesser moisture content (34.8% vs. 35.7)and also differences related to the aging process.Indeed, during ripening, several biochemical processes occur in cheese which gives rise to important changes in the texture and sensory characteristic of cheeses, varying from batch to batch (McSweeney 2004).Thus, it is reasonable to hypothesize that CBS dietary inclusion affected the cheese sensory profile during the aging process, without altering the acceptance of the product.
It is amply reported that dietary composition, when by-products are included, affects ruminal fermentation and milk composition could be modified to a lesser or greater extent (Chilliard et al., 2003;Romero et al., 2017) as well as its technological properties and the sensory quality of the derived dairy products (Coulon et al., 2004).Several studies have described the influence of the use of by-products in the diet of sheep on the characteristics of cheeses, however only few the studies included the effect on sensory profile.Caccamo et al. (2019) evaluated how the inclusion of hazelnut skin in the diets of dairy ewes can affect the chemical and sensory characteristics of ovine cheeses, showing that this by-product had significant impacts and revealed a minor production of off-flavors associated with spicy and acid characteristics.Also, the inclusion of artichoke silage in sheep's diets had a positive effect on the sensory characteristics of 60-d-ripened cheese (Jaramillo et al., 2010) as well as inclusion of grape pomace on the sensory profile in cheeses until 120 d of ripening (Bennato et al., 2023).Instead, no significant differences were found in cheese sensory profile when using levels up to 30% of citrus by-products (Jaramillo et al., 2009) or spray-dried olive mill wastewater (Branciari et al., 2020) as part of the diet in dairy ewes.Therefore, although several studies on the sensory properties of ripening pecorino cheese were conducted, our results are the first that reported the effect on the sensory profile of cheese affected by CBS diet inclusion in dairy sheep.The significant differences reported did not negatively affect the overall acceptability of the product, confirming how this diet inclusion can be suitable as an alternative feed resource in ruminants' nutrition.
The evaluation of these by-products demonstrated that diets supplemented with these alternative feeding resources supply the animal nutritional requirements, without compromising milk and cheese quality in small ruminants (Vasta et al., 2008).
The sensory characteristics reported by CBS cheese, such as higher hardness and a more intense toasted and smoked taste and aroma could improve the acceptability of the product by consumers who particularly appreciate tasty and flavored cheese (Bennato et al., 2023) making this alternative feeding a suitable strategy for the reuse and thus the enhancement of CBS.
A further step could be the development and evaluation of a preference map to understand consumer behavior regarding CBS cheese, by using a hedonic scale provided to consumers.This can help to understand which are the best attributes that particularly fit consumer preferences (Worch and Piqueras-Fiszman, 2015;Qannari 2017).

Effect on OAC Profile
Dietary CBS inclusion in dairy sheep partially exerted some effect on the OACs profile of cheeses.
To the best of our knowledge, no information can be found in the literature regarding the effect of a diet containing CBS on the volatile compounds of cheese.The OACs analysis identified compounds belonging to 7 different chemical families: acids, alcohols, aldehydes, esters, ketones, sulfur, and terpene, most of which are composed by esters which in dairy products originated from lactose fermentation or from amino acid catabolism acids (Bertuzzi et al., 2018).However only acids and sulfur compounds were significantly different be-tween the 2 cheeses group, with lower value perception in CBS cheese.
Sulfur compounds, widely detected in surface-ripened cheese, in the correct balance are major contributors to the characteristic strong flavor of different types of cheeses, derived from the catabolism of free fatty acids (Fox et al., 20017;Bertuzzi et al., 2018).
Acids were the most abundant volatile compounds in both 2 cheeses treatments, which usually strongly contribute to the aroma of different cheese varieties, including Pecorino, Manchego and Cheddar (Barron et al., 2005;Frank et al., 2004).
Particularly, the butanoic acid abundance, was lower in CBS than CTRL, resulting in a lesser perception of cheesy or putrid odours in CBS than CTRL samples (Thomsen et al., 2012).
Several authors attributed this flavor to an increase in concentration of shortchain free fatty acids or to an unbalanced proteolysis during cheese ripening Fox et al., 2017;McSweeney, 2017).
In the present study, the difference between the 2 cheese groups could be related to the significant difference reported in fatty acid content, with lower PUFA content in CBS compared with CTRL cheese.
According to the high content of fatty acids detected in both the 2 cheese groups, lipolysis could be the major pathway responsible for flavor generation (Zabaleta et al., 2016).
Literature shows that the volatile odor difference in cheese samples could be due to the different fatty acid composition of cheeses, since it is known the strong relation between milk fat and flavor development in cheese (Hassan et al., 2013).At the same time, the variation in milk composition -including fat content -induced by dietary management can be reflected in milk and cheese flavor, according to the mechanism involved in aroma compound development (Marilley et al., 2004).
In accordance with other studies which evaluated the effect of by-products inclusion in dairy sheep diet on the OACs of cheese, our work revealed that CBS dietary inclusion exerted a partial effect on odor compounds development in cheese, without altering the acceptability of the product, as also confirmed by the sensory analysis.
The findings of this study are useful both for the sensory and animal feeding fields.Indeed, the inclusion of CBS by-products in the diets of dairy sheep did not produce univocal results.
The role of the small ruminant industry is important, mainly in the rural communities of Mediterranean countries (Pulina et al., 2018).The use of CBS by-products in dairy sheep feeding, discounting some issues related to theobromine in the literature, easily resolved by following a limit of amount (max 300 mg/ Caccamo et al.: Cocoa by-product inclusion… kg in feedstuff; Alexander et al., 2008), can be considered a useful strategy to enhance this waste by using them in animal feeds.
The slight differences observed in the sensorial properties between CBS and CTRL cheeses did not alter the overall acceptability of the products as well as organoleptic characteristics, revealing how this dietary inclusion can be useful in dairy sheep farming without altering the quality of the final products.
The use of human-inedible feeds for ruminants, which do not require at all arable land should be supported to improve further the sustainability of food manufacture (Halmemies-Beauchet-Filleau et al., 2018) including dairy production.Moreover, improving knowledge of the effects of the use of by-products in ruminant feeding on milk composition and related products could help the cheesemakers understand the nature of the milk components that influence the cheesemaking properties and overall qualities of sheep dairy products (Cabiddu et al., 2008).
Cheese from dairy sheep fed with CBS did not show any negative differences compared with CTRL samples, with attributes in line with consumers' preferences.Thus, the use of CBS as alternative feedstuffs can be compatible with the growing sustainable livestock systems.

CONCLUSION
The present study evaluated the effect of the inclusion of cocoa bean shells in the diet of dairy sheep on the composition, volatile, antioxidant, and sensory characteristics of cheese.The higher content of monounsaturated fatty acids content and the raised nutritional indices suggest a favorable role of cocoa bean shell dietary inclusion on the nutritive value of the cheese.In terms of cheese sensory profile cacao bean shells affected more appearance, odor, aroma, and taste without altering the acceptance of the product.The volatile profile showed only a few significant differences, mainly related to the cheese ripening process, and no differences were found in α-tocopherol contents in cheese fat between the 2 groups.Hence, incorporating cocoa bean shells into the diets of dairy sheep enabled the production of high-quality cheese without compromising its characteristic traits, aligning well with the traditional profiles of sheep cheese.Furthermore, the use of this by-product could contribute to decrease feed costs and waste management, representing a good practice for increasing the sustainability of dairy products.
225°C before the initial use and for 5 min between each analysis.For the GC-MS analysis and the identification of OAC, a 7890A Series GC system (Agilent Technologies, Santa Clara, CA) coupled with an Agilent 5975C Mass Selective Detector (triple axis) was used.The HP-5 capillary column (30 m × 0.25 mm i.d.× 0.25 µm film thickness; Agilent Technologies, Santa Clara, CA) was used to separate the volatile components separation.The chromatographic conditions were as follows: splitless injector at 220°C; oven program conditions: 35°C for 3 min, 6°C/min to 200°C, 30°C/min to 240°C for 3 min.Helium pressure (carrier gas) was set at 93.77 MPa and the gas flow was 1.0 mL/min.The mass selective detector operated in scan mode (5.15 scan/sec) with 70 eV IE.Peak identification was carried out by comparison of mass spectra with the bibliographic data from the Wiley 175 library (Wiley & Sons, Inc., Germany), and with the LRI (linear retention indices) of authentic standards (Sigma-Aldrich) calculated by running a paraffin series (from C5 to C20) under the same working conditions.The OAC data were expressed as arbitrary units of chromatograph area.

Figure 1 .
Figure 1.Scheme of the experimental cheesemaking Caccamo et al.: Cocoa by-product inclusion… Caccamo et al.: Cocoa by-product inclusion… ratio (cis − C18: 1 + ΣPUFA)/(C12: 0 + C14: 0 + C16: 0). is present in all batches of the treatment cacao.About Ketone chemical class, 2-Octanone OAC was in more abundant in batches of the CBS cheeses than control cheeses indicating a soap odour perception.On the contrary for 2-Undecanone meaning orange odour perception was consistently present in CBS batches.In Sulfur chemical class, Dimethyl sulfone OAC was present only in few batches of the CBS cheeses.Finally, Terpene chemical class was more present in CBS batches with α-Pinene (green and fresh perception).β-Myrcene and d-Limonene, spicy and fresh lemon respectively, were present as unique compound in CTRL batch.All the other OACs were present in an alike way in both dietary treatments.

Table 2 .
Bozzetti et al. (2004)y-product inclusion… Descriptive attributes and definitions used to evaluate sheep cheese, adapted fromBozzetti et al. (2004) MellowSensation produced by sweet solutions, such as sucrose or fructose Soluble A sensation that emerges when the sample melts extremely fast in the saliva Dispersion Degree to which sample breaks into the mouth during chewing.

Table 4 .
Caccamo et al.:Cocoa by-product inclusion… Effect of the dietary treatment on cheese sensory profile

Table 5 .
Caccamo et al.: Cocoa by-product inclusion… Means of ln (x+1) Area of odour-active compounds grouped by chemical families in tested CTRL and CBS cheeses