Revalorizing Cocoa Endocarp Residues as Novel Ingredients With Potential Prebiotic Effects: A Step Toward Circular Economy

ABSTRACT

The interior part of the cocoa husk, known as the cocoa endocarp, is a by‐product of the chocolate sector that contains bioactive compounds with health benefits. In this research, the nutritional and physicochemical characteristics, and antioxidant capacity of dehydrated (DC) and lyophilized (LC) cocoa endocarps were evaluated, as well as their prebiotic properties affecting the human intestinal microbiota (IM). A simulated gastrointestinal digestion process was applied to DC and LC prior to 48 h of in vitro fecal fermentation. An increase in the relative abundance (RA) of beneficial intestinal bacteria through fecal fermentation was observed when DC and LC were present, whereas the RA of non‐beneficial intestinal bacteria decreased. DC and LC contained nondigestible compounds, including dietary fiber (6.0% and 8.1%, respectively) and phenolic compounds (such as ferulic and fumaric acids, epicatechin, and epigallocatechin gallate), that might contribute to the increase in the RA of beneficial bacteria. DC and LC also promoted modifications in IM metabolic activity via fecal fermentation, reducing the pH (<5.6), producing SCFA, reducing sugar consumption to undetectable levels at 48 h, modifying phenolic compounds, and enhancing antioxidant capacity, as determined by the ferric‐reducing antioxidant power (FRAP) method (>70.4 mg Fe/g at 48 h). Despite the different drying processes used, both DC and LC could beneficially alter the metabolic activity and composition of bacterial groups comprising the human IM. These results demonstrate the potential of revalorizing cocoa endocarp as a promising ingredient for developing sustainable formulations with prebiotic properties in the context of the circular economy.

Practical Applications

The cocoa endocarp is currently a by‐product of the chocolate sector. Drying this material produces an antioxidant ingredient that positively alters human intestinal bacterial populations. The cocoa endocarp could be revalued as a fiber‐rich ingredient for developing food products with human health benefits.

Abbreviations

a_w

water activity

Bac 303

probe to enumerate Bacteroides spp./Prevotella spp.

Bif 164

probe to enumerate Bifidobacterium spp.

CE

circular economy

Chis 150

probe to enumerate Clostridium histolyticum

DC

dehydrated cocoa

Erec 482

probe to enumerate Eubacterium rectale/Clostridium coccoides

FISH‐FC

fluorescence in situ hybridization coupled to flow cytometry

fw

fresh weight

Lab 158

probe to enumerate Lactobacillus spp./Enterococcus spp.

LC

lyophilized cocoa

<LOD

below the limit of detection

NA

not applicable

NC

negative control

RA

relative abundance

Rfla 729

probe to enumerate Ruminococcus albus/R. flavefaciens

SCFAs

short‐chain fatty acids

1. Introduction

Cocoa refers to the fruit of the cacao tree (Theobroma cacao L.), which grows in several tropical regions of Africa, Asia, the Caribbean, and South America (ICCO ²⁰²⁴). Cocoa fruit comprises mainly an inner portion, which includes the pulp and cocoa beans, and an outer layer, the cocoa husk (Belwal et al. 2022). The cocoa husk can subsequently be subdivided into the endocarp, mesocarp, and exocarp (Lu et al. 2018). Cocoa fruit represents a rich source of lipids (with approximately 1.9%–3.5% in the pulp and 1.5%–2.2% in the husk, expressed as dry weight), carbohydrates (varying from 11% to 68% in the pulp and from 29% to 32% in the husk, expressed as dry weight), antioxidants, organic acids, and phenolic compounds (Soares and Oliveira 2022). The nutritional composition and benefits for human health of cocoa beans are well established (Edo et al. 2023; Tan et al. ²⁰²¹; de Rezende Mudenuti et al. ²⁰²¹); however, other parts of the fruit, including the various husk layers, have been reported less widely in the scientific literature.

Although the cocoa bean and some of its derivatives, including chocolate, cocoa powder, and cocoa butter, are widely accepted and consumed globally, with about 5 million tons of cocoa beans produced each year, the husk (which represents about 56 to 76% of the cacao fruit) is often regarded as the principal waste of the chocolate industry (ICCO ²⁰²⁴; Barrios‐Rodríguez et al. ²⁰²²; Soares and Oliveira ²⁰²²). Various alternatives have been generated for the reuse of cocoa husk, including its use for the extraction of flavor compounds, fibers, and pectin, and as a colorant, texture agent, antioxidant, and substitute for cocoa powder (Kapun et al. 2024; Ramos et al. ²⁰²³; Barrios‐Rodríguez et al. ²⁰²²). Cocoa husk is a fiber‐rich ingredient, with up to 56% of this component (Soares and Oliveira 2022; Indiarto et al. ²⁰²¹), and the potential of cocoa discards in the elaboration of functional foods has been highlighted due to their distinct physicochemical characteristics and presence of bioactive compounds (Belwal et al. 2022). Similar strategies have been applied to the elaboration of fermented beverages and dairy systems using fruits, cereal, and nut by‐products (Herkenhoff et al. 2023). Some authors have evaluated these kinds of ingredients as substrates for probiotic or potential probiotic strains (Battistini et al. 2023; Praia et al. ²⁰²²). Other applications of cocoa husk include use in animal feed, as fertilizer, as an antimicrobial agent, in packaging, and in the cosmetics and fuel industries (Kapun et al. 2024; Mariatti et al. ²⁰²¹; Eletta et al. ²⁰²⁰; Campos‐Vega et al. ²⁰¹⁸).

Among the various strategies currently aimed at improving food waste management, the circular economy (CE) stands out as a promising concept. The CE is conceived as a financial structure different from a linear system (which operates on the principle of “take‐make‐consume‐dispose”), as it focuses on “reducing, alternatively reusing, recycling, and recovering materials in production/distribution and consumption processes” (Kirchherr et al. 2017), where these materials circulate in a closed loop (Figge et al. 2023). The CE concept can apply to the agro‐food sector through responsible food consumption, by promoting food recycling, and by utilizing food discards or waste to obtain valuable products or substances of interest (Korhonen et al. 2017). Cocoa husk valorization can be a valuable strategy for reducing food waste, and processing it through drying treatments (such as dehydration and freeze‐drying) could yield a functional ingredient with various health benefits aligned with the CE model (Bugarin et al. 2025; Campos‐Vega et al. ²⁰¹⁸). Similarly, different agro‐industrial by‐products and nontraditional plant matrices were evaluated as functional substrates for the development of prebiotic ingredients or for the delivery of probiotics, as demonstrated in Suzuki et al. (2024) and Herkenhoff et al. (2023).

Significant interest has been shown in the human intestinal microbiota (IM) in recent years, thereby promoting research on prebiotic foods. Prebiotics are “substrates selectively used by host microorganisms, conferring a health benefit” (Gibson et al. 2017). An in vitro fermentation protocol using fecal inoculum from human donors represents an effective strategy for assessing the impact of potential prebiotic foods or ingredients on different bacterial populations that constitute the IM (Menezes et al. 2021). Besides, human intervention studies have shown that discarded products from food industry can modulate the gut microbiota and improve metabolic parameters, supporting the concept that such ingredients may exert prebiotic effects (Horn et al. 2024).

Although cocoa endocarp residues are currently considered a waste product, studies evaluating their potential prebiotic effects were not found in the existing literature. Therefore, the objective of this study was to revalorize the cocoa endocarp within the framework of CE by investigating the impact of dehydrated (DC) and lyophilized cocoa (LC) endocarp residues on the abundance of target populations of intestinal bacteria and on their metabolic activity using in vitro fecal fermentation.

2. Materials and Methods

2.1. Materials

Cocoa pod endocarp was obtained from healthy, mature cocoa pods (genetic variety FAML‐01) sourced from Universidad del Valle de Guatemala (UVG, Guatemala). Pods were selected for the absence of over‐ripeness, softening, and visible fungal defects, manually opened, and seeds were removed. The empty pods were blanched in hot potable water (tap water) for 10–15 s at 95°C then cooled at room temperature (∼5 min). The softened inner layer (endocarp) was manually separated, weighed, and subjected to two drying methods: convective dehydration (50°C, 24 h; convection dehydrator, Ivation IV‐FD90RB, Quanzhou, China) (DC) and freeze‐drying (Harvest Right HRFDM, Utah, USA; pre‐freezing at −18°C followed by drying according to manufacturer settings) (LC). Dried samples were cooled, weighed, milled (20,000 rpm, 1 min) using a multipurpose mill (Tecnal TE631/4, Brazil), and sieved (40–100 mesh) using a sieve shaker (Cole‐Parmer, USA) with corresponding mesh screens. The resulting powders were vacuum‐sealed in trilaminate bags and stored at 4°C ± 0.5°C until analysis.

2.2. Methods

2.2.1. Characterization of Cocoa Endocarp

2.2.1.1. Evaluation of Yield Extraction, Color, Water Activity, and Moisture Content

The yield extraction of DC and LC was calculated considering the pods’ initial weight and the obtained dried endocarp's final weight (Sciammaro 2015). Moisture was determined by air drying (105°C, 24 h) (AACC ²⁰⁰⁰). Water activity (a_w ) was evaluated at 25°C using an Aqualab Series 3TE device (Decagon Devices, São José dos Campos, SP, Brazil) (AOAC ²⁰¹⁹). The color was established with a colorimeter (Chroma Meter CR‐300C, Minolta, Japan), and the Hunter factors (L*, a*, and b*) were obtained (CIE ¹⁹⁸⁶).

2.2.1.2. Evaluation of Lipid and Total Dietary Fiber Contents

The lipid content was determined using the Folch technique (Folch et al. 1956). The total fiber content was determined using the enzymatic‐gravimetric method with a Megazyme kit (Megazyme International Ireland Ltd., Bray, Ireland) and according to the AACC procedure (2000).

2.2.1.3. Evaluation of Organic Acid and Sugar Contents

To evaluate the contents of sugars (standards fructose, glucose, maltose, and rhamnose, ≥99.0%, high‐performance liquid chromatography (HPLC), Sigma‐Aldrich, St. Louis, MO, USA) and organic acids (standards acetic, formic, butyric, propionic, lactic, citric, succinic, malic, and tartaric acids, ReagentPlus, ≥99%, HPLC, Sigma‐Aldrich), 2 g of DC or LC were homogenized (10 min) using ultra‐purified water (20 mL, Milli Q Advantage A10 Water Purification System, Merck Millipore, Burlington, VT, USA), and the mixture was then centrifuged (15 min, 1789 × g, 24°C). The supernatants were filtered through 0.45 µm filters (Whatman, Chicago, IL, USA). To determine the organic acid and sugar composition, HPLC was performed using a liquid chromatograph (model 1260 Infinity LC, Agilent Technologies, Santa Clara, CA, USA) according to pre‐specified analytical conditions (dos Santos Lima et al. ²⁰²⁴; Coelho et al. ²⁰¹⁸). The equipment included a quaternary solvent pump with inline degassing (G1311C model) and an automatic sampler (G1329B model). An Agilent Hi‐Plex H (300 × 7.7 mm²) column with 8.0 µm internal particles, protected by a PL Hi‐Plex H 5 × 3 mm² guard column (Agilent Technologies), was employed. The temperatures of the column oven and refractive index detector (RID) were maintained at 70°C and 45°C, respectively. The sample injection volume was 10 µL, with a flow rate of 0.8 mL/min and a run time of 20 min. The mobile phase used was H₂SO₄ (4.0 mM/L) in ultrapure water. Organic acids were detected using a diode array detector (DAD) (G1315D model) at 210 nm, whereas sugars were detected using an RID (G1362A model). Identification and quantification were performed by comparing the retention times of the sample peaks with those of external standards, and quantification was performed by comparing the results with the calibration curves.

2.2.1.4. Evaluation of Phenolic Compound Contents

Methanol (methyl alcohol, 99.8% analytical standard, NEON, Suzano, SP, Brazil) extracts were obtained for the analysis of the phenolic compounds. Two grams of each sample were homogenized (60 min) using 10 mL of methanol 70% and then were centrifuged (15 min, 1789 × g, 24°C). The supernatant was then filtered through a 0.45 µm filter (Whatman, Chicago, IL, USA). A liquid chromatograph coupled with a Zorbax Eclipse Plus RP‐C18 column (100 × 4.6 mm², 3.5 µm) with a Zorbax C18 pre‐column (12.6 × 4.6 mm², 5 µm) (Zorbax, EUA) was employed. The separation was performed at 35°C, and the injection volume was 10 µL of each extract. The solvent flow rate was 0.8 mL/min. The gradient used in the separation was 0–5 min: 5% B; 5–14 min: 23% B; 14–30 min: 50% B; 30–33 min: 80% B, where Solvent A was an aqueous solution of 0.52% phosphoric acid (pH = 2.0), and Solvent B was methanol acidified with 0.52% H₃PO₄. Identification and quantification were performed by comparison with external standards, as previously described (dos Santos Lima et al. ²⁰²⁴). Standards of malvidin 3‐glucoside, cyanidin 3‐glucoside, delphinidin 3‐glucoside, pelargonidin 3‐glucoside, peonidin 3‐glucoside, (+)‐catechin, (−)‐epicatechin, (−)‐epigallocatechin‐gallate, (−)‐epicatechin gallate, procyanidin A2, procyanidin B1, procyanidin B2, kaempferol 3‐glucoside, quercetin 3‐glucoside, rutin, myricetin, and isorhamnetin were obtained from Extrasynthese (Genay, France). Caffeic acid, 3,4‐dihydroxybenzoic acid, vanillic acid, gallic acid, 4‐hydroxybenzoic acid, vanillin, O‐vanillin, (−)‐epigallocatechin, ρ‐coumaric acid, chlorogenic acid, caftaric acid, quercetin hydrate, syringic acid, ferulic acid, fumaric acid, hesperidin, naringenin, hesperitin, naringin, cyanidin‐3,5‐diglucoside, malvidin‐3,5‐diglucoside, and pelargonidin‐3,5‐diglucoside were obtained from Sigma‐Aldrich. t‐Resveratrol and c‐resveratrol were obtained from Cayman Chemical Company (Michigan, USA).

2.2.2. Evaluation of Antioxidant Capacity

Two methods [ferric‐reducing antioxidant power (FRAP) via reduction of ferric iron (Fe³⁺) and ABTS^●+ (2,2′‐azino‐bis (3‐ethylbenzothiazoline‐6‐sulfonic acid)) radical‐scavenging capacity] (Re et al. 1999; Benzie and Strain ¹⁹⁹⁶) were employed to evaluate the antioxidant properties of DC and LC. For this, 1 g of DC or LC was mixed with 25 mL of ethanol 80% (absolute ethyl alcohol, 99.8% analytical standard, NEON, Suzano, SP, Brazil), shaken (10 × g, 35°C, 2 h), centrifuged (1789 × g, 25°C, 10 min), and filtered with filter paper (ethanolic sample).

The FRAP calibration curve was created using a standard ferrous sulfate (ferrous sulfate heptahydrate, ACS reagent, ≥99%, Sigma‐Aldrich) solution (270 µL; 0.3 M acetate, TPTZ, and ferric chloride) with concentrations from 250 to 2000 µg Fe³⁺ per mL of water. Aliquots (20 µL of each dilution point) were made by combining 250 µL of FRAP reagent with 30 µL of distilled water. Samples were incubated (30 min, 37°C), and absorbance was read at 595 nm. Results were reported in milligrams Fe²⁺ per gram sample. The analyses were carried out at 25°C, protected from light. A microplate spectrophotometer (BioTek Eon, Winooski, VT, USA) was used to determine absorbance (da Silva et al. ²⁰²³).

The ABTS^•+ radical [2.2′azinobis (3‐ethylbenzothiazoline‐6‐sulfonic acid)] diammonium salt (≥98%, chromogenic, Sigma‐Aldrich) was prepared, stabilized (720 nm), and mixed with ethanol prior to analysis by mixing ABTS^•+ stock solution and potassium persulfate (ACS reagent, ≥99%, Sigma‐Aldrich). Thirty µL of each extract concentration was combined with 300 µL of ABTS^•+ radical and placed in a microplate. After 6 min, the absorbance was measured at 734 nm. The percentage of reducing activity (%) was calculated using (Bonifácio‐Lopes et al. ²⁰²²)

$$\mathrm{Inhibition}\left(\%\right)=\frac{\left(\mathrm{ABSc}-\mathrm{ABSs}\right)}{\mathrm{ABSc}}\times 100$$ (1)

where ABSc represents the absorbance of ABTS radical + ethanol, and ABSs represents the absorbance of ABTS radical + ethanolic sample.

2.2.3. Simulated Gastrointestinal Digestion of Cocoa Endocarp

The INFOGEST protocol for simulated gastrointestinal digestion, standardized in an international consensus, was used to perform in vitro digestion of DC and LC (Minékus et al. 2014). Initially, 5 g of each sample was mixed with 7.5 mL of ultrapure water at 25°C ± 0.5°C to ensure that every particle was hydrated. In vitro digestion was performed with pH regulation and the use of fluids simulating different phases: For the oral phase, α‐amylase from human saliva was used; for the gastric phase, pepsin from porcine gastric mucosa was used; and for the intestinal phase, pancreatin from porcine pancreas and bovine bile salts were used. The system was kept under orbital stirring (37°C ± 1°C; 4 h and 2 min). Digested samples were placed inside regenerated cellulose membranes (1 kDa, spectra/Pore 6, Spectrum Europe BV, Breda, the Netherlands) and then dialyzed in 0.01 mol/L NaCl solution (4°C, 18 h) to remove low‐molecular‐mass digestion products and simulate the absorption process. Finally, after digestion and dialysis, the samples were frozen (−18°C ± 0.5°C) and stored until use in the experiments (Sampaio et al. 2021). All enzymes and reagents used in this analysis were purchased from Sigma‐Aldrich.

2.2.4. Use of Fecal Inoculum From Human Donors

Fecal samples coming from six healthy donors were combined to obtain the human fecal inoculum. The Ethics Committee in Research with Human Beings from the Federal University of Paraiba (João Pessoa, PB, Brazil) approved the donation of fecal samples and the in vitro fecal fermentation procedures, as outlined in Protocol 6.861.644 and Certificate of Presentation for Ethical Assessment (CAAE): 79615924.5.0000.5188. The donor group was composed of 3 women and 3 men, aged between 21 and 42, with a body mass index between 18.5 and 24.9, without gastric or intestinal pathologies, following an omnivorous diet, exercising regularly, not smoking or drinking alcohol frequently, and not consuming antibiotics or probiotics within the previous 3 months. The fecal samples were stored into sterile jars containing an anaerobiosis generator system (AnaeroGen, Oxoid Anaerogen Anaerobic System, Basingstoke, Hampshire, UK), mixed (1:1:1:1:1:1, w/w), homogeneized (1:10, w/v) with phosphate‐buffered saline (PBS; 0.1 M; pH 7.4; 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄ analytical‐grade reagents and ultrapure water), filtered with a sterile triple‐layer gauze, and stored with glycerol (20%, w/v; ≥99.5%, analytical grade, Synth, Diadema, SP, Brazil) until use in the fermentation process (de Albuquerque et al. ²⁰²¹).

2.2.5. Fermentation System

The fermentation medium was prepared according to Table 1. The fermentation system contained fecal inoculum (40%, v/v), sterile fermentation medium (40%, v/v), and pre‐digested DC and LC samples (20%, w/v). The final system was fermented under anaerobic conditions (AnaeroGen) for 48 h (37°C ± 1°C). A fermentation system containing 20% (v/v) PBS instead of DC and LC samples was prepared as a negative control (NC).

TABLE 1.

Composition of fermentation medium.

Reagent	Amount
FeSO4	0.005 g
CaCl2	0.08 g
Bile salt	0.4 g
KH2PO4	0.5 g
MgSO4	0.69 g
l‐Cistein	0.8 g
NaHCO3	1.5 g
NaCl	4.5 g
KCl	4.5 g
Tween 80	1 mL
Resazurin solution (0.0025%)	4 mL

Source: Adapted from Massa et al. (2022).

2.2.6. Measurement of the Relative Abundance (RA) of Target Intestinal Bacterial Populations

The human IM was examined at 0, 24, and 48 h of fecal fermentation for the RA of specific bacterial groups using the following oligonucleotide probes: Lab 158 for Lactobacillus spp./Enterococcus spp., Bif 164 for Bifidobacterium spp., Rfla 729 for Ruminococcus albus/Ruminococcus flavefaciens, Bac 303 for Bacteroides spp./Prevotella spp., Chis 150 for Clostridium histolyticum, and Erec 482 for Eubacterium rectale/Clostridium coccoides (Biocell Biotecnologia Ltda, Minas Gerais, Brazil). These probes (Table 2) were designed to target regions of 16S rRNA and were labeled with the fluorescent dye Cy3 (Sigma‐Aldrich) (de Albuquerque et al. ²⁰²¹). The changes observed in the RA of these groups may indicate the modulatory effects of food products. On the other hand, SYBR Green staining was used to quantify the total bacterial population (Molecular Probes, Invitrogen, Carlsbad, CA, USA) (de Albuquerque et al. ²⁰²¹). At times 0, 24, and 48 h of the fecal fermentation, 375 µL of the fermentation medium was separated, fixed overnight (4°C) in 1.125 µL of 4% (w/v) filtered paraformaldehyde (≥95%, Sigma‐Aldrich), and hybridized for 4 h using the described fluorescent probes according to Massa et al. (2022) and de Albuquerque et al. (2021). To enumerate the various bacterial populations, a flow cytometer (BD Accuri C6, BD Biosciences, East Rutherford, NJ, USA) with excitation at 488 nm from a blue solid‐state laser was employed. The BD Accuri C6 software was used to collect logarithmic signals from individual cells passing through the laser zone and to record fluorescence signals as cytograms for two channels: FL1 for SYBR Green and FL2 for the bacterial probes. The results obtained were expressed as the RA (%) of the hybridized cells with one of the probes relative to the total bacterial population.

TABLE 2.

Oligonucleotide probes of 16S rRNA and hybridization characteristics used in fluorescence in situ hybridization (FISH) analyses.

Probe	Bacterial group	Sequence (5′–3′)	Hybridization pre‐treatment	Hybridization/Washing temperature (°C)
Lab 158	Lactobacillus spp./Enterococcus spp.	GGTATTAGCAYCTGTTTCCA	Lysozyme	50/50
Bif 164	Bifidobacterium spp.	CATCCGGCATTACCACCC	Lysozyme	50/50
Rfla 729	Ruminococcus albus/R. flavefaciens	GGTATTAGCAYCTGTTTCCA	Lysozyme	50/50
Bac 303	Bacteroides spp./Prevotella spp.	CCAATGTGGGGGACCTT	—	46/48
Chis 150	Clostridium histolyticum	GCTTCTTAGTCARGTACCG	—	50/50
Erec 482	Eubacterium rectale/Clostridium coccoides	TTATGCGGTATTAATCTYCCTTT	—	50/50

2.2.7. Evaluation of the Microbial Metabolic Activity During Fecal Fermentation

The contents of phenolic compounds, sugars, and SCFA, as well as the antioxidant capacity and pH values, were determined in the fermentation media containing DC and LC, and in NC at 0, 24, and 48 h, as detailed in Sections 2.2.1.3, 2.2.1.4, and 2.2.2.

A flowchart showing the methodology is presented in Figure 1.

FIGURE 1.

Methodology flowchart. DC = dehydrated cocoa; LC = lyophilized cocoa.

2.3. Statistical Analysis

The experiments were performed in triplicate across three independent assays. The results are presented as the average ± standard deviation. The Shapiro–Wilk normality test was used to assess the data's normality. The data were subjected to Student's t‐test or analysis of variance (one‐way ANOVA), and Tukey's test. A p value < 0.05 was considered statistically significant. A principal component analysis (PCA) was conducted to evaluate the relationship between the fermentation media and the tested parameters. The cophenetic correlation coefficient was calculated to assess the accuracy of PCA relative to the original Euclidean distances among the data points. To evaluate the strength of the relations between variables, Pearson's correlation coefficients (R) were calculated and shown in a heat map. Additionally, a hierarchical cluster analysis based on Euclidean distance assessed the similarity between the different fermentation media (Backhaus et al. 2021; Sparks ¹⁹⁷³; InfoStat ²⁰⁰⁸). InfoStat software (InfoStat ²⁰⁰⁸, Grupo InfoStat, FCA, Universidad Nacional de Córdoba, Argentina) was employed to perform the different statistical analyses.

3. Results

3.1. Physicochemical Parameters

Dehydration (DC) led to cocoa endocarp samples with higher moisture content (8.71% ± 0.10%) and, consequently, to higher a _w values (0.35 ± 0.01) than those subjected to lyophilization (LC; 3.78% ± 0.12% and 0.20% ± 0.04%, respectively) (p < 0.05). No differences were observed between the colors of DC and LC (p < 0.05). Total dietary fiber contents (fw—fresh weight) were 5.97% ± 0.27% and 8.18% ± 0.04% for DC and LC, respectively, and lipid contents (fw) were 4.32% ± 0.06% and 1.06% ± 0.11% for DC and LC, respectively. The fructose and glucose contents were higher in DC than in LC (p < 0.05), whereas the maltose content did not differ between DC and LC (p < 0.05). Overall, DC had higher organic acid contents than LC, with the highest contents for malic (2.52 ± 0.16 g/L), formic (1.98 ± 0.07 g/L), and succinic acid (0.90 ± 0.06 g/L) (p < 0.05) (Table 3).

TABLE 3.

Physicochemical parameters and antioxidant capacity (average ± standard deviation, n = 3) of cocoa endocarp samples (fw).

Parameter	Samples
	DC	LC
Moisture (%)	8.71 ± 0.10b	3.78 ± 0.12a
aw	0.35 ± 0.01b	0.20 ± 0.04a
Yield extraction (%)	13.72 ± 0.01b	12.59 ± 0.01a
Color parameters
L*	17.54 ± 1.70a	18.97 ± 1.79a
a*	5.24 ± 0.24a	5.16 ± 0.38a
b*	12.20 ± 0.50a	12.57 ± 0.50a
Lipids (%)	4.32 ± 0.06b	1.06 ± 0.11a
Total dietary fiber (%)	5.97 ± 0.27a	8.18 ± 0.04b
Sugars (g/L)
Fructose	7.50 ± 0.39b	1.93 ± 0.56a
Glucose	8.63 ± 0.47b	1.88 ± 0.93a
Maltose	0.25 ± 0.01a	0.41 ± 0.14a
Organic acids (g/L)
Acetic acid	<LOD	0.07 ± 0.04
Butyric acid	<LOD	<LOD
Citric acid	0.50 ± 0.02b	0.08 ± 0.02a
Formic acid	1.98 ± 0.07b	0.16 ± 0.06a
Lactic acid	0.02 ± 0.01	<LOD
Malic acid	2.52 ± 0.16	<LOD
Propionic acid	0.04 ± 0.01	<LOD
Succinic acid	0.90 ± 0.06a	0.39 ± 0.28a
Tartaric acid	0.18 ± 0.01b	0.03 ± 0.02a
Antioxidant capacity
FRAP (mg Fe2+/g fw)	62.15 ± 1.56b	54.18 ± 2.00a
ABTS (% inhibition)	47.41 ± 8.13a	58.28 ± 0.24b

Note: Different superscript letters in the same row denote differences (p ≤ 0.05), based on Student's t‐test; L* = Lightness (being 0 = black and 100 = white); a* = Chromaticity of red (positive values) to greenness (negative values); b* = Chromaticity of yellowness (positive values) to blueness (negative values).

Abbreviations: LOD, below the limit of detection; DC, dehydrated cocoa; fw, fresh weight; LC, lyophilized cocoa.

3.2. Antioxidant Capacity

The antioxidant capacity of DC was higher (62.15 ± 1.56 mg Fe²⁺/g fw) than that of LC (54.18 ± 2.00 mg Fe²⁺/g fw) as determined by the FRAP method (p < 0.05). On the other hand, different results were obtained when using the ABTS method, as the antioxidant capacity of LC was higher (58.28% ± 0.24%) than that of DC (47.41% ± 8.13%) (p < 0.05) (Table 3).

3.3. Phenolic Compounds in Cocoa Samples

The different phenolic compounds found in DC and LC are presented in Table 4. Fourteen out of the 19 phenolic compounds identified had higher contents in DC than in LC (p < 0.05). In contrast, contents of caftaric and fumaric acid, procyanidin B1, and hesperitin did not differ between DC and LC (p ≥ 0.05). The detected phenolic compounds had relatively low contents in DC and LC (<1 mg/L), except for fumaric acid (18.41 ± 2.17 mg/L in DC and 19.08 ± 3.73 mg/L in LC), ferulic acid (14.00 ± 1.33 mg/L in DC and 6.02 ± 1.19 mg/L in LC), epicatechin gallate (5.07 ± 0.20 mg/L in DC), epigallocatechin gallate (2.61 ± 0.00 mg/L in DC), and procyanidin B2 (2.46 ± 0.04 mg/L in DC).

TABLE 4.

Phenolic compounds (mg/L; average ± standard deviation, n = 3) of cocoa endocarp samples (fw).

Phenolic compound	Samples
	DC	LC
Phenolic acids
3,4‐dihydroxybenzoic acid	0.73 ± 0.02b	0.38 ± 0.01a
Caffeic acid	0.27 ± 0.01	<LOD
Caftaric acid	0.79 ± 0.09a	0.66 ± 0.05a
Chlorogenic acid	0.35 ± 0.03	<LOD
Ferulic acid	14.00 ± 1.33b	6.02 ± 1.19a
p‐Coumaric acid	0.19 ± 0.00	<LOD
Syringic acid	0.30 ± 0.00b	0.23 ± 0.01a
Vanillic acid	0.31 ± 0.01a	0.43 ± 0.06b
Flavonols
Catechin	0.14 ± 0.01	<LOD
Epicatechin gallate	5.07 ± 0.20b	0.47 ± 0.07a
Epigallocatechin gallate	2.61 ± 0.00b	0.77 ± 0.04a
Procyanidin A2	0.65 ± 0.01b	0.42 ± 0.01a
Procyanidin B1	0.45 ± 0.02a	0.47 ± 0.01a
Procyanidin B2	2.46 ± 0.04	<LOD
Flavonones
Hesperitin	0.19 ± 0.01a	0.20 ± 0.01a
Naringin	0.40 ± 0.04	<LOD
Stilbenoid
t‐Resveratrol	0.22 ± 0.00	<LOD
Other phenolic compounds
Fumaric acid	18.41 ± 2.17a	19.08 ± 3.73a
Vanillin	0.28 ± 0.01b	0.25 ± 0.01a

Note: Different superscript letters in the same row denote differences (p ≤ 0.05), based on Student's t‐test.

Abbreviations: LOD, below the limit of detection; DC, dehydrated cocoa; fw, fresh weight; LC, lyophilized cocoa.

3.4. Changes in the RA of Intestinal Bacterial Populations During In Vitro Fecal Fermentation

Overall, the RA of the measured target bacterial populations differed between DC and LC media as fecal fermentation time increased (Table 5). The NC medium reduced the RA of all measured bacterial groups during fermentation (p < 0.05). The highest RA of Lactobacillus spp./Enterococcus spp. was detected in LC medium at 24 and 48 h of fermentation (7.04% ± 0.54% and 3.16% ± 1.10%, respectively), with 7‐fold higher values than NC (1.11% ± 0.26%) (p < 0.05). The RA of Bifidobacterium spp. did not vary between the DC and LC media at 24 h of fermentation (p ≥ 0.05). The RA of Bifidobacterium spp. in DC medium increased at 48 h of fermentation (2.88 ± 0.25%) when compared to the same medium at 24 h of fermentation (2.12 ± 0.32%) (p < 0.05).

TABLE 5.

Relative abundance (RA) (% average ± standard deviation, n = 3) of different bacterial groups in media containing digested cocoa samples at time zero (baseline), 24 and 48 h of an in vitro fecal fermentation.

Measured bacterial groups	Fermentation media	Time of fermentation
		0 h	24 h	48 h
Lactobacillus spp./Enterococcus spp.	DC	4.19 ± 0.99Cb	2.90 ± 0.15Bb	2.04 ± 0.20Ab
	LC	4.43 ± 1.09Ab	7.04 ± 0.54Bc	3.16 ± 1.10Ab
	NC	1.62 ± 0.18Ba	1.11 ± 0.26Aa	1.06 ± 0.21Aa
Bifidobacterium spp.	DC	3.19 ± 0.69Bb	2.12 ± 0.32Ab	2.88 ± 0.25Bc
	LC	2.70 ± 0.82Ab	2.34 ± 0.29Ab	1.93 ± 0.30Ab
	NC	1.79 ± 0.22Ba	0.73 ± 0.16Aa	0.74 ± 0.13Aa
Ruminococcus albus/R. flavefaciens	DC	2.65 ± 0.55Ab	2.25 ± 0.58Ab	2.04 ± 0.21Ac
	LC	1.21 ± 0.18Aa	3.27 ± 0.55Bb	1.26 ± 0.11Ab
	NC	1.21 ± 0.14Ba	1.00 ± 0.12ABa	0.76 ± 0.15Aa
Bacteroides spp./Prevotella spp.	DC	10.05 ± 0.35Bc	7.82 ± 1.23Ab	10.16 ± 0.44Bc
	LC	7.66 ± 0.44Ab	6.30 ± 1.50Ab	6.22 ± 0.82Ab
	NC	3.54 ± 0.31Ca	2.96 ± 0.27Ba	1.73 ± 0.18Aa
Clostridium histolyticum	DC	4.01 ± 1.71Ab	3.55 ± 0.15Ac	3.18 ± 0.36Ac
	LC	3.50 ± 0.40Cb	2.65 ± 0.44Bb	1.71 ± 0.31Ab
	NC	1.46 ± 0.12Ba	0.89 ± 0.16Aa	0.54 ± 0.18Aa
Eubacterium rectale/C. coccoides	DC	5.60 ± 0.80Bc	2.57 ± 0.73Ac	2.78 ± 0.13Ac
	LC	2.13 ± 0.36Bb	1.57 ± 0.22Ab	1.50 ± 0.17Ab
	NC	0.64 ± 0.11Ca	0.42 ± 0.06Ba	0.29 ± 0.03Aa

Note: Different superscript capital letters in the same row for the same fermentation medium denote differences (p ≤ 0.05), whereas different superscript small letters in the same column at the same time interval and bacterial group denote difference (p ≤ 0.05), based on Tukey's test.

Abbreviations: DC, dehydrated cocoa; LC, lyophilized cocoa; NC, negative control.

The LC medium at 24 h of fermentation presented the highest RA of R. albus/R. flavefaciens (3.27% ± 0.55%) (p < 0.05), whereas it did not change in DC medium during the fecal fermentation (2.04% ± 0.21%–2.65% ± 0.55%) (p ≥ 0.05). After 24 h of fermentation, the RA of Bacteroides spp. and Prevotella spp. decreased despite the media in comparison to time zero (p < 0.05). Bacteroides spp./Prevotella spp. had the highest RA among the measured bacterial groups, with its highest RA detected in DC medium (7.82% ± 1.23%–10.16% ± 0.44%) (p < 0.05).

The RA of C. histolyticum decreased during fermentation despite the media (p < 0.05). Concerning those media containing endocarp cocoa samples, the lower RA of C. histolyticum was detected in LC medium at 48 h of fermentation (1.71% ± 0.31% for LC vs. 3.18% ± 0.36% for DC) (p < 0.05). The RA of E. rectale/C. coccoides decreased at 24 h of fermentation in DC (2.57 ± 0.73%) and LC medium (1.57 ± 0.22%) (p < 0.05), whereas it did not differ at 48 h of fermentation (p ≥ 0.05). The RA of E. rectale/C. coccoides was higher in DC and LC media than in NC during the fecal fermentation (p < 0.05).

3.5. Changes in the Microbial Metabolic Activity During In Vitro Fecal Fermentation

The fecal fermentation of DC and LC led to a decrease in pH over time (p < 0.05) (Table 6). The pH at time zero was close to neutrality (6.50 ± 0.00–7.00 ± 0.00) in the examined media and remained ≥6.13 ± 0.06 in NC during the fecal fermentation. On the other hand, the pH values did not vary in DC and LC media at 24 h (5.84 ± 0.01–5.84 ± 0.06) (p ≥ 0.05), but they did differ at 48 h of fermentation (5.58 ± 0.01–5.63 ± 0.01) (p < 0.05). The sugar contents (fructose, glucose, maltose, and rhamnose) decreased over time in DC and LC media during fermentation (p < 0.05). On the other hand, it remained overall below the limit of detection (<LOD) in NC during the fermentation (Table 6).

TABLE 6.

pH values and contents (g/L) of sugars and short‐chain fatty acids (SCFAs) (average ± standard deviation, n = 3) at 0, 24, and 48 h of in vitro fecal fermentation.

Parameter	Fermentation media	Time of fermentation
		0 h	24 h	48 h
pH
	DC	7.00 ± 0.00Cb	5.84 ± 0.06Ba	5.63 ± 0.01Ab
	LC	6.50 ± 0.00Ca	5.84 ± 0.01Ba	5.58 ± 0.01Aa
	NC	6.50 ± 0.00Ba	6.50 ± 0.03Bb	6.13 ± 0.06Ac
Sugars (g/L)
Fructose	DC	<LOD	<LOD	<LOD
	LC	<LOD	0.01 ± 0.00a	<LOD
	NC	<LOD	0.01 ± 0.00a	<LOD
Glucose	DC	<LOD	<LOD	<LOD
	LC	0.03 ± 0.00Ba	0.01 ± 0.00A	<LOD
	NC	0.01 ± 0.00a	<LOD	<LOD
Maltose	DC	0.19 ± 0.01B	0.06 ± 0.01Aa	<LOD
	LC	<LOD	0.07 ± 0.02a	<LOD
	NC	<LOD	<LOD	<LOD
Rhamnose	DC	<LOD	0.005 ± 0.004Aa	0.005 ± 0.004Aa
	LC	<LOD	0.012 ± 0.003Ba	0.003 ± 0.000Aa
	NC	<LOD	<LOD	<LOD
SCFA (g/L)
Acetic acid	DC	0.16 ± 0.11Aab	0.21 ± 0.04Aab	0.13 ± 0.06Aa
	LC	0.19 ± 0.03Ab	0.23 ± 0.03Ab	0.12 ± 0.04Aa
	NC	0.10 ± 0.01Aa	0.13 ± 0.05Aa	0.09 ± 0.01Aa
Propionic acid	DC	0.45 ± 0.30Aa	2.30 ± 0.53Ca	1.27 ± 0.16Bb
	LC	0.60 ± 0.10Aa	1.81 ± 0.21Ba	0.59 ± 0.36Aa
	NC	<LOD	<LOD	<LOD
Butyric acid	DC	0.19 ± 0.00Aa	1.02 ± 0.26Bb	1.00 ± 0.03Bb
	LC	0.25 ± 0.08Aa	0.82 ± 0.07Bb	1.04 ± 0.04Bb
	NC	<LOD	0.17 ± 0.01Aa	0.24 ± 0.03Ba

Note: Different superscript capital letters in the same row for the same fermentation media denote differences (p ≤ 0.05), whereas different superscript small letters in the same column at the same time interval and evaluated parameter denote differences (p ≤ 0.05) based on Tukey's test.

Abbreviations: LOD, below the limit of detection; DC, dehydrated cocoa; LC, lyophilized cocoa; NC, negative control.

The acetic acid content did not differ among the DC, LC, and NC media during fecal fermentation (p ≥ 0.05). In contrast, propionic acid content increased 5‐fold in DC medium (2.30 ± 0.53 g/L) and 3‐fold in LC medium (1.81 ± 0.21 g/L), achieving the highest contents at 24 h of fermentation (p < 0.05). Propionic acid was <LOD in NC. Fermentation time increased the amount of butyric acid in DC (0.19 ± 0.00–1.02 ± 0.26 g/L) and LC medium (0.25 ± 0.08–1.04 ± 0.04 g/L) during fermentation (p < 0.05). Additionally, DC and LC media presented higher butyric acid contents than NC at 24 and 48 h of the fermentation (p < 0.05) (Table 6). Regarding the organic acids (i.e., formic, lactic, and succinic acids) evaluated during fermentation, no changes in their contents were detected between 0 and 48 h.

3.6. Phenolic Compounds and Antioxidant Capacity Changes During In Vitro Fecal Fermentation

Results regarding phenolic compound contents in DC and LC media during fecal fermentation are shown in Table 7. The contents of 3,4‐dihydroxybenzoic and fumaric acids decreased during fecal fermentation (p < 0.05), rendering these <LOD at the end of the fermentation. Epigallocatechin gallate also suffered a sharp decrease during fermentation in DC (7.60 ± 0.01–0.16 ± 0.02 g/L) and LC medium (8.71 ± 0.47 g/L—<LOD) (p < 0.05). The content of hesperitin decreased in DC medium (p < 0.05) but remained unchanged in LC medium during fermentation (p ≥ 0.05). Conversely, epicatechin gallate content increased after 24 and 48 h of fermentation compared to time zero, reaching the highest concentration in DC medium (4.45 ± 0.53 g/L) at 48 h (p < 0.05). The content of catechin increased in DC and LC media at 24 h of the fermentation compared to the time zero (p < 0.05), whereas no changes were observed at 48 h of fermentation (p ≥ 0.05).

TABLE 7.

Content of phenolic compounds (mg/L) and antioxidant capacity determined by FRAP (mg Fe/g) and ABTS (% inhibition) (average ± standard deviation, n = 3) in the fermentation media during in vitro fecal fermentation.

Phenolic compound (mg/L)		Fermentation media	Time of fermentation
			0 h	24 h	48 h
Phenolic acids	3,4‐dihydroxybenzoic acid	DC	0.58 ± 0.07a	<LOD	<LOD
		LC	0.61 ± 0.04a	<LOD	<LOD
Flavonones	Hesperitin	DC	0.53 ± 0.04Bb	0.31 ± 0.01Ab	0.28 ± 0.01Aa
		LC	0.24 ± 0.04Aa	0.20 ± 0.01Aa	0.28 ± 0.09Aa
Flavonols	Catechin	DC	<LOD	0.05 ± 0.00Aa	0.06 ± 0.01Aa
		LC	<LOD	0.08 ± 0.02Aa	0.05 ± 0.01Aa
	Epicatechin gallate	DC	<LOD	4.05 ± 0.70Aa	4.45 ± 0.53Ab
		LC	<LOD	3.33 ± 0.44Aa	3.40 ± 0.06Aa
	Epigallocatechin gallate	DC	7.60 ± 0.01Ca	3.17 ± 1.47Ba	0.16 ± 0.02A
		LC	8.71 ± 0.47Bb	4.72±2.35Aa	<LOD
	Fumaric acid	DC	0.54 ± 0.05a	<LOD	<LOD
		LC	0.71 ± 0.01b	<LOD	<LOD
Antioxidant capacity	FRAP (mg Fe/g)	DC	60.65 ± 9.69Ab	74.68 ± 3.59Bb	90.97 ± 1.71Cb
		LC	40.11 ± 2.39Aa	49.68 ± 0.53Ba	70.36 ± 3.03Ca
	ABTS (% inhibition)	DC	46.21 ± 6.50Ab	30.00 ± 11.87Aa	31.72 ± 5.20Aa
		LC	23.10 ± 4.88Aa	30.40 ± 17.88Aa	31.78 ± 1.06Aa

Note: Different superscript capital letters in the same row for the same fermentation medium denote differences (p ≤ 0.05), whereas different superscript small letters in the same column at the same time interval for phenolic compound or antioxidant capacity denote difference (p ≤ 0.05) based on Tukey's test.

Abbreviations: LOD, below the limit of detection; DC, dehydrated cocoa; LC, lyophilized cocoa; NC, negative control.

The antioxidant capacity was higher in the DC medium than in the LC medium (p < 0.05), as evaluated by the FRAP method, regardless of the fecal fermentation period. The antioxidant capacity increased in DC medium (60.65 ± 9.69–90.97 ± 1.71 mg Fe²⁺/g) and LC (40.11 ± 2.39–70.36 ± 3.03 mg Fe²⁺/g) with increasing fermentation time (according to the FRAP method) (p < 0.05). On the other hand, the antioxidant capacity decreased in DC medium (46.21% ± 6.50%–31.72% ± 5.20%) and increased in LC medium (23.10% ± 4.88%–31.78% ± 1.06%), as determined by the ABTS method (p < 0.05) (Table 7).

3.7. Chemometric Analysis

The two principal components, PC1 and PC2, explained 80% of the total variance. LC0 was in the positive quadrant for PC1, positively correlating with higher pH values and higher amounts of epigallocatechin gallate, as well as higher RA of Lactobacillus spp./Enterococcus spp. and R. albus/R. flavefaciens. The DC0 positively correlated with the higher RA of Bifidobacterium spp., Bacteroides spp./Prevotella spp., C. histolyticum, and E. rectale/C. coccoides. The DC medium at 24 and 48 h of fermentation showed negative correlations with PC1, antioxidant capacity (FRAP), and the presence of epicatechin gallate and butyric acid. LC medium at 24 and 48 h of fermentation showed negative PC1 and PC2 values and the highest catechin content (Figure 2).

FIGURE 2.

Principal component analysis (PCA) run in media containing fermented cocoa endocarp samples at times 0, 24, and 48 h of in vitro fecal fermentation (variables: RA of distinct bacterial groups, pH values, contents of phenolic compounds and butyric acid, and antioxidant capacity). Lab 158 = Lactobacillus spp./Enterococcus spp.; Bif 164 = Bifidobacterium spp.; Rfla 729 = Ruminococcus albus/R. flavefaciens; Bac 303 = Bacteroides spp./Prevotella spp.; Chis 150 = Clostridium histolyticum; Erec 482 = Eubacterium rectale/C. coccoides. DC = dehydrated cocoa; LC = lyophilized cocoa; RA = relative abundance.

Cluster analysis (Figure 3) indicated that the DC medium at 24 and 48 h of fermentation were grouped in the same cluster due to their similarities (Euclidean distance < 3.2), whereas the remaining samples were in different clusters. When considering Euclidean distance < 4.8, DC and LC media at time zero (DC0 and LC0) can be grouped. In contrast, DC24 and DC48 can be grouped with LC48.

FIGURE 3.

Euclidean cluster analysis of media containing fermented cocoa endocarp samples at times 0, 24, and 48 h of in vitro fecal fermentation. The dotted line (considered the midpoint of the scale) indicates whether the samples are in the same cluster. Lab 158 = Lactobacillus spp./Enterococcus spp.; Bif 164 = Bifidobacterium spp.; Rfla 729 = Ruminococcus albus/R. flavefaciens; Bac 303 = Bacteroides spp./Prevotella spp.; Chis 150 = Clostridium histolyticum; Erec 482 = Eubacterium rectale/C. coccoides. DC = dehydrated cocoa; LC = lyophilized cocoa.

Pearson's correlation test (Figure 4) showed that the RA of all evaluated bacterial populations was positively correlated with pH and epigallocatechin gallate content (p < 0.05). On the other hand, Bacteroides spp./Prevotella spp. are positively correlated (p < 0.05) with antioxidant capacity according to the FRAP method. The cophenetic correlation coefficient was 0.955, indicating that the PCA analysis accurately preserved the original Euclidean distances.

FIGURE 4.

Heat map of correlation coefficients indicating associations among the RA of the analyzed bacterial groups, pH, antioxidant capacity, phenolic compounds, and short‐chain fatty acids. Lab 158 = Lactobacillus spp./Enterococcus spp.; Bif 164 = Bifidobacterium spp.; Rfla 729 = Ruminococcus albus/R. flavefaciens; Bac 303 = Bacteroides spp./Prevotella spp.; Chis 150 = Clostridium histolyticum; Erec 482 = Eubacterium rectale/C. coccoides. RA = relative abundance.

4. Discussion

Dehydration and lyophilization of a cocoa by‐product (cocoa endocarp) yielded powdered food‐grade products with distinct physicochemical properties. Lyophilization caused a greater decrease in moisture than dehydration, as expected in this kind of process, as it removes almost all the water from a sample. On the contrary, the extraction yield was higher for DC than for LC, likely due to the higher water content in DC. Nevertheless, both drying methods could result in a product with high stability derived from the cocoa endocarp, as evidenced by the low a_w values (<0.60), ensuring a lack or very low microbial activity, biochemical and enzymatic deterioration, while also maintaining a dark brown color and being preserved for extended periods (Chirife et al. 1996; Grob et al. ²⁰²¹).

Regarding nutritional composition, LC had higher dietary fiber but lower lipid content than DC. In evaluating cocoa by‐products, other studies have reported higher dietary fiber content in the cocoa husk, ranging from 36% to 56% (g/100 g, dry weight) (Soares and Oliveira 2022). This may be because the cacao husk is the outer layer of the cacao fruit, and, like the husks of other fruits, it is considered a more fibrous component.

Studies comparing different cocoa husk drying methods have reported that lyophilization better preserves the nutritional and bioactive compounds (Valadez‐Carmona et al. ²⁰¹⁷; Okiyama et al. ²⁰¹⁸). However, in the present study, dehydration of cocoa endocarp, in comparison with lyophilization, led to the production of higher amounts of the major sugar components (fructose and glucose), SCFA (especially formic), organic acids (mainly citric, malic, and succinic acid), and phenolic compounds, demonstrating that both methods could be proposed to make the cocoa endocarp a promising ingredient to produce different foods according to the desired composition. The flavonols catechin, epicatechin, and procyanidin B2 have been found in high amounts in cocoa husk, representing a source of compounds with antioxidant capacity (Boni et al. 2026; Okiyama et al. ²⁰¹⁸). However, plant variety, harvest and post‐harvest conditions, and geographical location can affect the cocoa husk phenolic content (Oracz et al. 2019). DC exhibited higher antioxidant capacity than LC as determined by the FRAP assay. This result may be related to the higher levels of phenolic compounds (such as epicatechin and procyanidins) found in DC, as these substances can exert antioxidant effects (Rojo‐Poveda et al. ²⁰²⁰; Valadez‐Carmona et al. ²⁰¹⁷; Martínez et al. ²⁰¹²).

Fermentation of different cocoa endocarp samples led to variations in the RA of bacterial populations. An analysis of the PCA showed that LC0 selectively affected the presence of beneficial bacterial groups, as it was associated with higher RA for Lactobacillus spp./Enterococcus spp. and R. albus/R. flavefaciens. In general, Lactobacillus spp./Enterococcus, Bifidobacterium spp., and R. albus/R. flavefaciens are considered intestinal bacteria with benefits for human health (He et al. 2022; Abdi et al. ²⁰²²; Lakshmanan et al. ²⁰²²; Cizeikiene and Jagelaviciute ²⁰²¹). Lactobacillus spp. and Bifidobacterium spp. are highly prevalent bacterial groups in the human IM and are widely used as probiotics (Praia et al. 2022). Lactobacillus species perform essential functions in the human body, including protection against pathogen invasion by strengthening the gut mucosal barrier, as well as promoting the synthesis of various SCFAs (Barker‐Tejeda et al. ²⁰²⁴; Alves‐Santos et al. ²⁰²⁰; Slattery et al. ²⁰¹⁹). Higher RA of R. albus is observed in healthy populations compared with individuals with intestinal diseases (Kang et al. 2010). This bacterial group plays relevant functions in alleviating oxidative stress (Park et al. 2017). Additionally, a higher abundance of Ruminococcus species has been negatively correlated with specific symptoms of irritable bowel syndrome (Liu et al. 2024). The RA increase regarding Lactobacillus spp./Enterococcus, Bifidobacterium spp., and R. albus/R. flavefaciens indicates the presence of relevant amounts of indigestible compounds in DC and LC, such as dietary fiber, which can reach the colon and serve as substrates for these bacterial populations (Araújo et al. 2024; Lacerda Massa et al. ²⁰²⁰). The presence of different phenolic compounds that reach the gut and selectively stimulate specific bacterial groups is also likely a contributing factor (Araújo et al. 2020). On the other hand, Bacteroides spp./Prevotella spp., C. histolyticum, and E. rectale/C. coccoides are usually considered non‐beneficial or pathogenic bacterial populations (Popoff 2024; Wang et al. ²⁰²¹; Liu et al. ²⁰¹⁹; Sanders et al. ²⁰¹⁹). Fermented samples of LC correlated negatively with the RA of these non‐beneficial bacterial groups, as shown in the PCA analysis, as LC and LC48 were present on the opposite side of Bacteroides spp./Prevotella spp., C. histolyticum, and E. rectale/C. coccoides. The effect of LC fermentation was therefore beneficial, as it decreased the RA of these bacteria. On the other hand, the PCA results for DC showed that DC0 was in the same quadrant as the non‐beneficial bacterial groups. However, fermentation of this sample led to several modifications that ultimately improved DC characteristics, separating DC24 and DC48 from the named bacterial populations and linking them to higher antioxidant properties.

The decrease in pH during fermentation may contribute to the inhibition of detrimental microorganisms in the human IM, such as C. histolyticum (de Oliveira et al. ²⁰²⁴; de Albuquerque et al. ²⁰²¹; Liu et al. ^2020a; Palframan et al. ²⁰⁰²). A previous study reported a decrease in fecal pH with an in vivo model using Wistar rats fed a diet containing cocoa fiber; this diet was associated with higher counts of Lactobacillus and Bifidobacterium (Massot‐Cladera et al. ²⁰¹⁵). The fiber present in cocoa samples and the production of SCFA could explain the acidification of the gut environment, thereby reducing the presence of pathogenic bacterial groups and promoting beneficial bacteria (Battistini et al. 2023; Massot‐Cladera et al. ²⁰¹⁵).

The different sugars present in the cocoa samples were metabolized by the human fecal inoculum in fermentation media containing DC and LC and used as a substrate to augment the organic acids production, causing a decrease in the pH (Silveira Martins et al. 2023; Massa ²⁰²¹; de Albuquerque et al. ²⁰²¹). SCFA have been shown to be beneficial for intestinal health and consist of short‐chain organic acids with chain lengths of two to six carbon atoms (LeBlanc et al. ²⁰¹⁷). The SCFA content increased during the fecal fermentation, as observed in media containing DC or LC, where propionic and butyric acids were initially present at relatively low levels at time zero and increased at 24 and 48 h of the fecal fermentation. Disca et al. (2024) also evaluated the prebiotic properties of cocoa shell, finding higher SCFA production during fermentation of this by‐product. Butyric acid has shown different beneficial effects in the human body, including some potentially therapeutic properties for the immune system (Liu et al. 2025). The microbial fermentation of substrates such as phenolic compounds and fiber can produce SCFA, providing various health benefits to humans (Abdelhalim 2024; Hadinia et al. ²⁰²²; Magistrelli et al. ²⁰¹⁶). The higher RA of Lactobacillus spp./Enterococcus spp. could be related to the higher production of SCFA detected in the media containing DC or LC (Hadinia et al. 2022). Bifidobacterium species can also promote the synthesis of SCFA and promote a low acidity in the intestinal tract (Panigrahi 2024; Yamamura et al. ²⁰²³). Additionally, higher butyric acid content and lower pH values could be related to higher RA of Lactobacillus spp. and Bifidobacterium, as observed in other studies using plant‐derived by‐products (de Oliveira et al. ²⁰²⁴; Zhuang et al. ²⁰²¹). The fact that butyric acid correlated positively with the fecal fermentation of DC (as shown with the PCA analysis) could be considered a beneficial effect of the fermentation of the cocoa endocarp.

The fecal microbiota greatly influenced the phenolic compound profile in media containing DC and LC. This metabolic process resulted in the degradation of certain phenolic compounds initially present in media containing DC and LC, including 3,4‐dihydroxybenzoic acid and epigallocatechin gallate. It is estimated that almost 95% of the polyphenols (especially oligomeric and polymeric polyphenols) in a food can reach the colon, where they are susceptible to IM enzymatic action (Oracz et al. 2019; Bowey et al. ²⁰⁰³). The relationship between IM composition and phenolic compound metabolism is often difficult to predict and can be characterized as a reciprocal, two‐way relationship (Oracz et al. 2019; Morrison and Preston ²⁰¹⁶; Cardona et al. ²⁰¹³). Certain phenolic compounds in cocoa can selectively increase beneficial bacterial populations and inhibit intestinal pathogenic bacteria (Oracz et al. 2019). For example, epigallocatechin gallate can increase the RA of Bifidobacterium and decrease the RA of Prevotella (Liu et al. ^2020b). On the other hand, some intestinal bacterial species, including Lactobacillus spp. and Bifidobacterium spp., act in the metabolism of cocoa polyphenols, leading to the production of smaller bioactive compounds (Quiroz‐Eraso et al. ²⁰²³; Hossain et al. ²⁰²²).

Antioxidant capacity, evaluated by the FRAP method, increased in the DC and LC media during fecal fermentation. On the other hand, the ABTS method shows that antioxidant capacity increased only in the media containing LC. The high antioxidant capacity of the fermentation media may be due to the presence of phenolic compounds and to increased RA in Lactobacillus spp./Enterococcus spp. and Bifidobacterium spp. (da Silva et al. ²⁰²³). At the same time, antioxidant capacity increases, along with epicatechin gallate levels. Similar results were obtained when evaluating the antioxidant properties of red beet during in vitro fecal fermentation (de Oliveira et al. ²⁰²³). It is also noteworthy that, according to the Euclidean test, fermenting DC for more than 24 h does not alter the antioxidant capacity, phenolic compound content, or RA of the bacterial groups.

The amount of epigallocatechin gallate was negatively correlated with the antioxidant capacity. Epigallocatechin gallate can be degraded into various metabolites via A‐ring fission, aliphatic chain shortening, C‐ring opening, and subsequent hydrolysis, thereby stimulating beneficial bacterial growth and increasing antioxidant capacity during fermentation (Liu et al. ^2020b). Additionally, a robust negative correlation was observed between pH and the amount of simpler flavonoids, such as epicatechin gallate and catechin. In contrast, epigallocatechin gallate was positively correlated with pH. This result indicates that during the progression of fermentation, the pH is significantly reduced together with the amount of epigallocatechin gallate, which could be transformed into simpler compounds because of the conditions of the fecal fermentation media or the enzymatic action by the presence of beneficial bacteria, such as Lactobacillus spp./Enterococcus spp. and R. albus/R. flavefaciens (Oracz et al. 2019).

In general, the results support the view that agro‐industrial by‐products, such as cocoa endocarp, can be successfully revalorized as functional ingredients with prebiotic potential. Previous evidence on the use of fruit and cereal by‐products as substrates for probiotic and potential probiotic microorganisms was reported by Battistini et al. (2023) and Praia et al. (2022).

Despite the observed results, it is important to highlight that, as known, the models of in vitro fecal fermentation provide an approach into evaluating the real potential prebiotic effects of cocoa endocarp but cannot fully reproduce the complexity of in vivo conditions. In this sense, future clinical studies are needed to confirm whether cocoa endocarp ingestion exerts similar benefits in vivo.

5. Conclusion

Dehydration and lyophilization of cocoa endocarp, currently considered an important residue of the cocoa industry, led to the manufacture of products with a rich nutritional profile, consisting primarily of dietary fiber, various phenolic compounds, and organic acids with antioxidant properties. DC and LC increased the RA of beneficial intestinal bacterial groups via fecal fermentation, whereas they decreased the RA of non‐beneficial bacterial populations. These effects were evidenced by the distinct metabolic activity of IM via fecal fermentation, which increased sugar utilization and SCFA production, reduced pH, increased antioxidant capacity, and altered the phenolic compound profile. These results demonstrate the potential of the cocoa endocarp, typically discarded as a cocoa processing residue, as a promising ingredient for developing sustainable prebiotic formulations within the CE.

Author Contributions

Angela Daniela Carboni: conceptualization, data curation, formal analysis, writing – review and editing, writing – original draft, investigation. Thatyane Mariano Rodrigues de Albuquerque: conceptualization, data curation, formal analysis, writing – original draft, writing – review and editing. Jordana Nunes de Oliveira: investigation, methodology. Mariela Patrignani: formal analysis, data curation, writing – review and editing. Marcos dos Santos Lima: investigation, methodology. Heloísa Maria Almeida do Nascimento: investigation. Ana Alicia Paz Pierri: investigation. María Cecilia Puppo: funding acquisition, project administration, writing – review and editing. Evandro Leite de Souza: conceptualization, data curation, formal analysis, funding acquisition, methodology, project administration, supervision, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.