Development of Starter Inoculum for Controlled Arabica Coffee Fermentation Using Coffee By-Products (Pulp and Mucilage Broth), Yeast, and Lactic Acid Bacteria

Abstract

Recent research has highlighted the effectiveness of starter inocula in fermentation processes. In this sense, this study examines the use of an inoculum composed of coffee pulp, mucilage broth, and microorganisms such as Saccharomyces cerevisiae, L. delbrueckii ssp. bulgaricus, and S. thermophilus in fermenting Castillo variety coffee. An inoculum was prepared, measuring variables such as the pH, acidity, °Brix, lactic acid bacteria, and yeast viability. Following optimization, the inoculum was evaluated in a fermentation process, evaluating the pH, °Brix, acidity, microbiological analysis, ochratoxin A, and cup quality post-drying and roasting. The findings demonstrated a significant reduction in the pH from 4.47 to 4.05 and in the °Brix from 15.8 to 8.45, indicating efficient organic acid production and sugar degradation. Acidity levels increased from 20.02 mg/g to 42.69 mg/g, while microbial viabilities remained above 10⁷ CFUs/g, suggesting effective biomass production. The process effectively reduced the microbial load without detecting ochratoxin A. Sensory evaluations confirmed the enhanced cup quality, validating the positive impact of inoculum use in coffee fermentation. The results support the use of coffee pulp and mucilage broth as effective substrates for the growth of the evaluated microorganisms, and the application of starter cultures containing lactic acid bacteria and yeast can elevate the coffee to a specialty grade.

1. Introduction

Coffee is among the most popular commodities globally, with Brazil maintaining a lead in production from 1994 to 2019, according to the Food and Agriculture Organization of the United Nations [1]. Brazil produced over 58 million 60-Kg jute bags of coffee from 2019 to 2020 and exported more than 17 million bags between May and October 2021 [2]. Coffee cultivation and processing generate various residues—pulp, mucilage, husk, and grounds. Approximately 43.58% of the fresh fruit weight is coffee pulp. Each million 60-kg bags of coffee beans exported from Colombia generates roughly 162,900 tons of fresh pulp residues. Improper disposal of these residues can lead to environmental contamination unless treated effectively [3]. Coffee pulp, rich in sugars (45.67%), protein (10.63%), fiber (36.07%), and ash (9.58%), presents potential as a raw material for producing inocula for coffee fermentation with other microorganisms [4].

Controlled coffee fermentation can enhance the sensory quality. Utilizing microbial starter cultures not only improves the beverage quality but also shortens the processing times, as microorganisms degrade the pulp and mucilage, producing acids and metabolic compounds that permeate the parchment layer into the beans [5,6]. For instance, yeasts have been shown to inhibit the growth of mycotoxigenic fungi. However, microbial behavior varies according to the coffee variety, production region, processing method, and species [7].

Employing specific microbes as starter cultures, such as Pichia fermentans YC5.2, Wickerhamomyces anomalus KNU18Y3, Saccharomycopsis fibuligera KNU18Y4, Papiliotrema flavescens KNU18Y5 and KNU18Y6, Pichia kudriavzevii KNU18Y7, and Saccharomyces cerevisiae KNU18Y12 and KNU18Y13, has been crucial in producing high-quality coffee with distinct characteristics. These include intense vanilla flavors and floral aromas, and unique cupping notes like pepper, nutty, spicy, perfume, rose, caramel, bell pepper, roast, orange, and green apple [8,9,10]. Similarly, lactic acid bacteria (LAB) have been recognized for generating bioactive compounds that contribute to coffee quality. The flavor-enhancing capabilities of “wild-type” LAB cultures in coffee fermentation have drawn attention due to the variety of aromas these strains can produce, including floral, fruity, and buttery notes [11]. Recent research on the co-culture of lactic yeast and lactic acid bacteria has been crucial in improving beverage quality and achieving coffees with distinct sensory profiles [12].

This research aims to develop a starter inoculum using coffee pulp, mucilage broth, commercial yeast (Saccharomyces cerevisiae), and lactic acid bacteria (Streptococcus thermophilus and L. delbrueckii ssp. Bulgaricus) as raw materials. The study is structured with the following three steps: (1) an inhibition test for pathogenic bacteria and a resistance test at different temperatures of the lactic acid bacteria used, (2) optimizing the inoculum using statistical methods with coffee by-products and validating the optimal conditions, and (3) testing the optimized inoculum in a coffee fermentation process to assess its impact on the physicochemical characteristics and microbial viability. A sensory evaluation was also carried out to determine the cup quality. The novelty of the study consists in using coffee by-products to develop a starter inoculum that reduces the fermentation time and improves the cup quality of Castillo variety coffee.

2. Materials and Methods

2.1. Coffee Variety

Ripe Castillo variety (Coffea arabica) coffee fruits were manually harvested from the “La Primavera” farm, situated at an elevation of 1558–1650 m above sea level in Florencia, Cauca, Colombia, where the average temperature is 19 °C. The cherries were cleaned and disinfected by immersing them in a 2% hypochlorite solution for 30 s. Subsequently, the cherries were pulped using a DH4 Penagos pulper (Penagos, Bucaramanga, Colombia), and the coffee pulp was manually separated. To extract the mead from the mucilage, the coffee was washed and manually squeezed to remove as much mucilage as feasible. The pulp was then vacuum-packed in 10 × 8 cm polyethylene bags (Alico, Colombia) using a Webomatic® vacuum packer (ESSEN S.A.S, Cali, Colombia), and the mead was stored in sterile glass jars. Both raw materials were preserved by freezing at −18 °C.

2.2. Starter Cultures

Fresh commercial yeast (Saccharomyces cerevisiae) from the Fleischmann brand and lactic acid bacteria (Streptococcus thermophilus and L. delbrueckii ssp. Bulgaricus) from Tapioka commercial yogurt (Cooperativa de Productos Lácteos de Nariño, Ltda., Nariño, Colombia) were utilized. To ensure their purity, the yeast was cultured on potato dextrose agar (PDA) and the yogurt bacteria were cultured on Man, Rogosa, and Sharpe agar (MRS), both sourced from Condolab, Spain. The growth of each strain was confirmed using the plate count method, and initial viability counts were conducted.

2.3. Inhibition Test for Pathogenic Bacteria

The inhibitory test for pathogenic bacteria was conducted using the disk diffusion method with modifications [13]. Initially, the lactic acid bacteria (LAB) present in the yogurt were cultured in MRS broth for 24 h at 35 °C. Each strain was then adjusted to a concentration of 10³ CFUs/mL according to the McFarland scale and inoculated with a sterile swab onto MRS agar plates, which were incubated for 48 h at 35 °C. After the first 24 h, pathogenic bacteria were cultured in Brain–Heart Infusion (BHI, Sharlau, Dornstetten, Germany) broth and incubated for another 24 h at 35 °C. Subsequently, the pathogenic bacteria were adjusted to a density of 0.5 on the McFarland scale (M-IDENT®-MacFARLAND, Madrid, Spain), equivalent to 10³ CFUs/mL. This dilution was swabbed onto Müller–Hinton agar (Sharlau, Dornstetten, Germany) plates. An agar disk containing the isolated lactic bacteria, after 48 h of development, was placed at the center of the agar surface already inoculated with the pathogenic bacteria. Each inoculation was performed in triplicate, with a control in duplicate. The inhibitory capacity was assessed by measuring the halo formed around the disk inoculated with LAB, after an incubation period of 16–24 h at 35 °C, using ImageJ software V1.53t.

2.4. Resistance Test of Lactic Acid Bacteria at Different Temperatures

This test was performed following the methodology described by Lucumi Banguero et al. [1]. The resistance of lactic acid bacteria (LAB) to various temperatures was assessed through an experiment where a 10% v/v bacterial inoculum of each strain was prepared in MRS broth. These inocula were adjusted on the McFarland scale to a density of 0.5 and incubated in triplicate at temperatures of 15 °C, 20 °C, 25 °C, and 30 °C for 24 h. Tubes that did not exhibit bacterial growth were excluded from consideration; only those showing bud formation were analyzed. The viability of the LAB was subsequently determined by their growth on MRS solid medium plates stained with aniline blue, with colony-forming units (CFUs) counted to assess the viability.

2.5. Inoculum Preparation

The coffee pulp was heated at 96 °C for 8 min to reduce the microbial load. Subsequently, mucilage broth containing 5% sugar, previously dissolved, was added to the pulp. This mixture was then subjected to size reduction using a semi-industrial blender (Waring® MX1500XTX Xtreme Hi-Power Blender, Philadelphia, PA, USA). Yogurt was incorporated into the blended mixture, which was then placed in an anaerobic chamber Oxoid, 2.5 L AnaeroJar (Thermo Scientific^TM, Sunnyvale, CA, USA) for 12 h. After this period, dissolved yeast was added in a 1:1 ratio with sterile distilled water and the mixture was stirred in an incubator (IKA INC 125 FS, Baden-Württemberg, Germany) for 7 h. Upon completion of this process, the viability of the inoculum was determined.

2.6. pH

A pH meter (Orion, 710-A, Sigma Aldrich, Burlington, MA, USA) was used to determine the pH at room temperature. The equipment was calibrated with two buffer solutions of pH values of 7 and 4. For monitoring the pH in the inoculum with starter cultures, one gram of inoculum was suspended in 10 mL of distilled water, homogenized with a vortex, and filtered (Whatman^® No. 41, Maidstone, UK) under a vacuum; then, the pH of the filtrates was determined with the pH meter. To monitor the pH during the coffee fermentation, samples of the fermentation water were collected from both the control treatment (spontaneous fermentation) and the fermentation with starter cultures. Approximately 5 mL of each sample was taken at 0, 18, 24, and 36 h of fermentation and measured directly [14]. An average of three determinations was reported.

2.7. Soluble Solids

The soluble solids were measured in the inoculum with starter cultures both at the beginning and the end of its preparation, and the behavior of this variable was monitored during the coffee fermentation process in both the control and inoculated samples at 0, 18, 24, and 36 h. A digital refractometer (PCE Instruments, PCE-DR, Hochsauerland, Germany) was used, calibrated with distilled water. Using an eyedropper, a representative amount of the liquid-phase samples was deposited on the refractometer prism, and the readings were obtained directly in °Brix.

2.8. Acidity

The acidity was determined according to the method described in the Colombian technical standard NTC 5247 [15] for the determination of the titratable acidity in the coffee samples, both for the inoculation with the starter cultures and for monitoring the coffee fermentation.

2.9. Viability

The test was carried out according to a methodology described by [1] with some modifications. To evaluate the biomass production potential, serial dilutions were performed with a 10% v/v inoculum with starter cultures until concentrations of 10⁻⁶ for LAB and 10⁻⁷ for yeasts were achieved. The samples were then seeded and incubated in triplicate for 48 h at 35 °C. The results were evaluated by measuring the number of CFUs/mL, with the acceptable range being 30 to 300 colonies.

3. Experimental Design and Statistical Analysis

The experiments were carried out according to a statistical extreme mixture design, with five replicates at the central point and expanded with axial points, resulting in 21 treatments. Various proportions of coffee pulp, sugar, yeast, and lactic acid bacteria were evaluated (see

Table 1. Mixtures in the experimental design (extreme vertex mix design).

Treatment	Coffee Pulp (%)	Mucilage Broth (%)	LAB (%)	Yeast (%)
1	75	25	0	0
2	75	20	0	5
3	65	25	10	0
4	65	20	10	5
5	75	15	10	0
6	70	25	0	5
7	65	25	5	5
8	75	10	10	5
9	70.6	20.6	5.6	3.1
10	72.8	22.8	2.8	1.6
11	72.8	20.3	2.8	4.1
12	67.8	22.8	7.8	1.6
13	67.8	20.3	7.8	4.1
14	72.8	17.8	7.8	1.6
15	70.3	22.8	2.8	4.1
16	67.8	22.8	5.3	4.1
17	72.8	15.3	7.8	4.1
18	70.6	20.6	5.6	3.1
19	70.6	20.6	5.6	3.1
20	70.6	20.6	5.6	3.1
21	70.6	20.6	5.6	3.1

). To minimize systematic errors, the experiments were performed randomly. The response variables included physicochemical parameters and the viability of microorganisms measured during the inoculation with starter cultures. The constraints applied to the experimental design were coffee pulp (65–75%), mucilage broth (15–25%), yeast (0–5%), and lactic acid bacteria (0–10%).

An analysis of variance (ANOVA) was used to determine the model fitness. The statistical significance of the model and equation terms was analyzed based on the p-value (p < 0.05). The backward elimination regression procedure was applied to remove non-significant interactions, ensuring that only significant variables were selected for constructing the predicted model. The desirability compound (viability) was used to determine the optimal value. All statistical analyses were performed using MINITAB version 19.

Using the optimized inoculum, coffee fermentation was carried out, during which physicochemical characteristics (pH, °Brix, and total acidity) were monitored. Microbiological quality (fungal counts, coliforms, and enteric and lactic colonies), ochratoxin A, and the score of sensorial quality were also evaluated.

3.1. Model Quality

The regression coefficients, R² and adjusted-R², were calculated to determine the variation between the model (ŷ) and the experimental (y) data using Equations (1) and (2). Values close to one for both regression coefficients indicate good concordance between experimental and simulated data. The lack of fit was also estimated to show the predictive performance (3).

$$R^2=1-{\displaystyle \frac{{\sum }_{i=1}^n{\left(y-\widehat{y}\right)}^2}{{\sum }_{i=1}^n{\left(y-\widehat{y}\right)}^2}}$$

(1)

$$Adjusted R^2=1-{\displaystyle \frac{\left(1-R^2\right)\left(n-1\right)}{n-P-1}}$$

(2)

$$F={\displaystyle \frac{Mean squares of the lack of fit}{Mean squares of error}}$$

(3)

If ${\displaystyle \frac{MS lack of fit}{MS error}}<F\left(DF lack of fit,DF error\right),$ the lack of fit is not considered to be significant.

3.2. Optimization

A response optimizer was used to find the optimal mixture through composite desirability, aiming to maximize the values of the LAB and the yeast viability. However, the responses reported by Minitab19 did not match what could be obtained, so objective values were defined (lower: 6, objective: 8, and upper: 9).

3.3. Experimental Validation

Validation experiments were conducted by applying the optimal conditions. The responses were compared with the optimal predictions using the relative error (RE), as shown in Equation (4):

$$RE={\displaystyle \frac{y-\widehat{y}}{y}}\times 100$$

(4)

where y is the experimental value and $\widehat{\mathit{y}}$ is the predicted value.

3.4. Coffee Fermentation

Polypropylene bioreactors with a 5 L capacity were used for the fermentation process. These bioreactors were sanitized with steam and a 2% hypochlorite solution prior to use. Fermentation was conducted using 2.1 Kg of pulped Arabica coffee cherries under the following two distinct experimental setups: one with bioreactors containing 10% of the optimized inoculum with starter cultures (Saccharomyces cerevisiae, Streptococcus thermophilus, and Lactobacillus bulgaricus), and another with bioreactors allowing for spontaneous fermentation with endogenous microbiota. Additionally, a control experiment was included, where the fruits were directly dried in a forced convection oven (Magic Mill) without any fermentation process.

The bioreactors with the inoculum-added samples were sealed with screw caps to prevent air entry, while those for the spontaneous fermentation were covered with sterile gauze to allow air flow. Water was added to each treatment in a 1:1 ratio per bioreactor, and physicochemical process control variables (pH, Brix, and acidity) were measured. The process was carried out at an ambient temperature. Microbiological and viability analyses of the microorganisms were conducted at the beginning and end of the fermentation.

3.5. Microbiological Analysis

Samples of 10 mL of mucilage were homogenized with 90 mL of buffered peptone water (Scharlab S. L., Barcelona, Spain) using a vortex mixer (Benchmark Scientific Inc, Taiwan) and serially diluted. Enumeration of the yeasts, aerobic mesophilic microorganisms, enteric bacteria, and LAB was performed using the deep-plating technique, as follows: potato dextrose agar (PDA) for yeasts and fungi, MRS agar for LAB, VRBD agar for enteric bacteria, and standard method agar (PCA) for mesophiles, all sourced from Laboratorios Conda S.A. (Madrid, Spain). The plates were incubated at 30 °C (yeasts), 37 °C (aerobic mesophiles and enteric bacteria), and 37 °C (LAB) for 48 h in an incubator (Thermo Scientific, Langenselbold, Germany), and the number of colony-forming units (CFUs) was quantified.

3.6. Ochratoxin A

Ochratoxin A determination followed the AOAC (2000.09, 2000) method. The test portion was extracted with a 3% methanol and aqueous sodium bicarbonate solution (50:50, v/v). The extract was filtered, and the filtrate was diluted with phosphate-buffered saline and applied to an immunoaffinity column containing antibodies specific for OTA. After washing, the toxin was eluted from the column with methanol and quantified by liquid chromatography (LC) with fluorescence detection. The limit of quantification was 0.5 µg/Kg.

3.7. Sensory Evaluation

The sensory evaluation of the samples was conducted by two certified tasters with Q-grader certificates. All of the sample preparation and roasting procedures followed the Specialty Coffee Association (SCA) Specialty Coffee Cupping Protocol [16]. Sensory attributes such as sweetness, acidity, body, bitterness, astringency, and aftertaste were evaluated using a linear intensity scale ranging from 0 (absence) to 10 (maximum intensity) [17]. Reference standards for each sensory attribute were established according to the World Coffee Research Sensory Lexicon [18].

The overall score was evaluated on a scale from 0 to 100, following the SCA Protocol for Cupping Specialty Coffee [16].

4. Results and Discussion

This section is structured into the following three parts: Section 4.1 shows the inhibitory effects of the inoculum on the pathogenic bacteria and the temperature resistance of the lactic acid bacteria. Section 4.2 details the optimization of the inoculum using statistical methods. Section 4.3 describes the application of the optimized inoculum in the controlled fermentation of Arabica coffee. These findings highlight the potential of utilizing coffee by-products as a medium for microbial inoculation and emphasize the impact of the inoculum on enhancing the quality of the coffee fermentation process.

4.1. Inhibition of Pathogenic Bacteria and Temperature Resistance

4.1.1. Inhibition Test for Pathogenic Bacteria

Existing coffee processes, which depend on natural microorganisms, are uncontrollable and lead to inconsistencies in the product [19]. Consequently, the use of mixed starter cultures in coffee fermentation has emerged as a recent practice, significantly enhancing the food safety and the sensory characteristics of coffee [9]. Fermentations with mixed starter cultures provide several advantages over traditional methods that utilize pure starter cultures, closely replicating the spontaneous microbial consortia that naturally occur. By creating a culture medium closely aligned with the coffee matrix, it becomes feasible to adapt these microorganisms to the specific conditions encountered during the controlled coffee fermentation process, even though these microorganisms are not isolated directly from coffee [20].

Inhibition tests were conducted using a consortium of microorganisms (Saccharomyces cerevisiae, Streptococcus thermophilus, and L. delbrueckii ssp. Bulgaricus), which demonstrated their ability to inhibit the growth of several common pathogenic microorganisms in food, as shown in

Table 2. Inhibition of inoculum for pathogenic bacteria.

Pathogenic Bacteria	Inhibition Halos (mm)
Proteus ssp.	14.2 ± 0.028
Escherichia coli	6.2 ± 0.056
Salmonella	12.4 ± 0.042
Pseudomonas	7.0 ± 0.021
Klebsiella	12.5 ± 0.14
Staphylococcus aureus	7.0 ± 0.07

. These tests also assessed the ability of the microorganisms to proliferate at various temperature values that mimic those found in coffee cultivation environments. The images of the test can be observed in Figure S1 (Supplementary Materials).

High microbial loads in coffee by-products and beans are promoted by the moisture content of the substrate and the soluble carbohydrates present in the pulp and mucilage. Initial microbiological analyses revealed mesophilic aerobic counts, fungi, yeasts, lactic bacteria, and positive total coliforms, with counts exceeding 300 CFUs in the 1 × 10⁻⁴ dilution. The presence of pathogenic microorganisms degrades the quality of coffee beans and imparts undesirable flavors to the beverage. The use of an inoculum with starter cultures containing lactic acid bacteria effectively counteracts these undesirable microorganisms, with inhibition halos measuring 14.2 ± 0.028 mm, 6.2 ± 0.056 mm, 12.4 ± 0.042 mm, 7.0 ± 0.021 mm, 12.5 ± 0.14 mm, and 7.0 ± 0.07 mm for Proteus spp., Escherichia coli, Salmonella, Pseudomonas, Klebsiella, and Staphylococcus aureus, respectively.

Lactic acid bacteria (LAB) have been demonstrated in various studies to inhibit the growth of unwanted bacteria in food through the production of antimicrobial proteins (bacteriocins), diacetyl, organic acids, hydrogen peroxide, ethanol, and acetaldehyde [21]. The antibacterial effects of LAB are primarily linked to metabolites such as organic acids (notably lactic, acetic, propionic, sorbic, and benzoic acids), hydrogen peroxide, diacetyl, ethanol, phenols, and protein compounds produced during their growth [21]. Furthermore, certain LAB strains can synthesize bacteriocins that exhibit substantial antibacterial activity [22]. Comparing these results with those reported by [23], who observed inhibition halos of 21.05 ± 0.29 mm for L. delbrueckii ssp. bulgaricus against E. coli and 18.05 ± 0.06 mm for S. thermophilus against Salmonella, it is evident that the results were comparable for Salmonella (12.4 ± 0.042 mm) but lower for E. coli (6.2 ± 0.056 mm), likely due to the combined effects of LAB with S. cerevisiae yeast. These inhibition effects could be attributed to the antibacterial secretion of LAB nanoparticles deposited within the cell [2].

4.1.2. Resistance to Different Temperatures

Table 3. Viability of LAB at different temperatures.

Temperature (°C)	10−2 CFUs	10−3 CFUs	10−4 CFUs	10−5 CFUs	10−6 CFUs
15	0	0	0	0	0
25	>300	>300	25	0	0
30	>300	>300	>300	3	1
35	>300	>300	>300	20	10

shows the results of LAB growth at varying temperatures to determine the optimal conditions for microorganism growth used in the starter inoculum.

As shown in Table 3, no growth was observed at 15 °C. At 25 °C, growth was noted up to the 10⁻⁴ dilution, while at 30 °C and 35 °C, growth was sustained up to the 10⁻⁶ dilution, suggesting these temperatures as ideal for LAB growth. This corresponds with the thermophilic nature of these bacteria, which typically have an optimum growth temperature of 40–45 °C, a minimum growth temperature of 20–25 °C and a maximum growth temperature near 47–50 °C. Specifically, Streptococcus thermophilus, a member of the thermophilic LAB group, is commonly used with Lactobacillus species in yogurt production and Swiss and Italian cheese manufacturing, among other dairy products [24]. The optimum growth temperature for S. thermophilus is ∼37 °C, but it is sufficiently thermophilic in nature to grow alongside L. delbrueckii ssp. bulgaricus during the commercial production of yogurt at 42 °C. The growth of S. thermophilus ceases at ∼10 °C [25].

4.2. Optimization of the Inoculum Using Statistical Methods

According to the experimental design shown in Table 1, the results obtained for the inoculum treatments evaluated are presented below, showing parameters such as the pH, acidity, soluble solids, and microbial growth (items 4.2.1 to 4.2.3). The statistical analysis of the viability of the microorganisms present in the inoculum was carried out in item 4.2.4.

On the other hand, an optimization process (item 4.2.5) was carried out to have a higher viability of the microorganisms (LAB and yeasts) evaluated in the inoculum.

4.2.1. pH and Acidity

Figure 1 presents the results for the pH and total acidity for the 21 treatments evaluated.

The results show a decrease in the pH as the growth of the microorganisms progresses in the substrate. Given that the medium and inoculum are derived from coffee by-products, it is expected that they exhibit pH and acidity values close to those of the original fruit. Typically, the initial pH of the coffee beans ranges between 5.11 and 5.25 [26], while coffee pulp may show more acidic values, varying from 3.92 to 5.63 [27,28,29]. This study found a pH of 4.43, which is within the reported range for coffee by-products. Variations are attributed to factors like the growing conditions, coffee variety, altitude, fruit ripening, and production methods [30,31].

Yeasts, which can survive in a pH range from 3 to 10, experience stress under basic conditions, since their optimal growth occurs in acidic environments. Most LAB perform optimally at a pH between 4.0 and 4.5 [32]. Hence, the pH observed in this study supported the growth of starter cultures, enhancing the viability. In Arabica coffee fermentation, final pH values typically range from 3.7 to 4.0, similar to those recorded in this study, with a value of 4.07 ± 0.39 at 19 h of fermentation. The pH decrease is primarily due to acid formation and dissociation (such as lactic, acetic, formic, and alcoholic acetification) during fermentation, varying with the microorganism and its metabolism, with lactic acid being particularly influential [33].

To determine if there were significant differences in the pH after 19 h of fermentation of the inoculum, a paired t-test with a significance of 0.05 was used, observing that, in eight treatments, there were no statistically significant differences. Among these were the control, treatments 3 and 5 (where yeast was absent), treatment 6 (where LAB was absent), and three treatments at the central point. However, in all of the treatments, there was a decrease in the pH, which was expected. It is possible to observe that, the greater the amount of LAB used, the greater the influence on the pH decrease, as a low oxygen concentration normally leads to an increase in the LAB population [3]. In the first hours, the inoculum was in an anaerobic environment. Similarly, LAB are the main producers of lactic acid during coffee fermentation, and this acid significantly influences the pH decrease and final coffee quality.

Figure 1b shows a statistically significant difference in 19 of the 21 treatments, which is a good indicator of the fermentation process of the starter inoculum. During the first hours, the LAB were in an anaerobic environment that favored their viability, and, later, when they were at an optimal pH, the yeasts had the appropriate conditions for their growth. The yeasts and LAB interact to decompose the sugars present in the pulp, and this microbial activity not only affects the final flavor of the coffee but also its acidity. In this study, a synergistic effect was observed between the two microorganisms, influencing acid production and, therefore, the increase in the acidity. Silva et al. [4,5,6] mention that these microorganisms in coffee fermentation contribute to the production of ethanol, lactic, butyric, acetic, and other acids that increase the final acidity of the product.

4.2.2. Soluble Solids (°Brix)

Figure 2 shows the behavior of the soluble solids during the preparation of the medium based on coffee pulp for the 21 treatments.

The initial °Brix measured 15.59 ± 2.13 at the start of the experiment and decreased to 8.32 ± 2.70 after 19 h. No significant reduction in this parameter was observed in treatments without yeast (treatments 1, 3, and 5), highlighting the importance of yeast in fermentation. The initial addition of yogurt during the first 12 h is crucial, as these microorganisms reduce the medium’s pH by producing acids (e.g., lactic and acetic), thereby creating a favorable environment for subsequent yeast development [34].

To determine if there were significant differences in the content of soluble solids after 19 h of fermentation, a paired t-test was employed with a significance of 0.05. The treatments that did not show statistically significant differences (p > 0.05) were treatments 1, 3, 5, 9, 16, and 17. In treatments 1, 3, and 5, there was an absence of yeast, while, in treatments 9, 16, and 17, there was a higher proportion of lactic acid bacteria than yeast. Therefore, it is possible to attribute this behavior to the fact that LAB prefer certain types of sugars, such as glucose, but are not as efficient in degrading others, such as sucrose or fructose, which are also present in the coffee. This can result in a higher proportion of sugars not being transformed, keeping the °Brix more stable. However, the addition of these LAB was performed with the intention of improving the aroma profiles, as it has been shown to produce high concentrations of lactic acid, ethyl acetate, ethyl isobutyrate, acetaldehyde, and ethyl propionate. Such metabolites have fruity aromas and enhance the sensory quality of the final products [7].

4.2.3. Microbial Growth in the Inoculum

Adesulu-Dahunsi et al. [35] highlighted the symbiotic or synergistic interactions between microorganisms that are essential for their survival during fermentation. This collaboration was observed in the current study using Gram stains to evaluate the morphologies and cell wall compositions of the microorganisms involved. Additionally, viability counts for the yeasts and LAB were performed across the treatments.

Table 4. Viability of LAB and yeast in the inoculum.

Treatment	Components				Viability (Log CFUs/g)
	X1	X2	X3	X4	Yeast	LAB
1	75	25	0	0	75	0
2	75	20	0	5	8.45	0
3	65	25	10	0	0	5.94
4	65	20	10	5	8.88	8.66
5	75	15	10	0	0	5.38
6	70	25	0	5	8.29	0
7	65	25	5	5	9.09	8.69
8	75	10	10	5	9.35	8.92
9	70.6	20.6	5.6	3.1	8.41	8.68
10	72.8	22.8	2.8	1.6	8.29	8.5
11	72.8	20.3	2.8	4.1	8.76	8.69
12	67.8	22.8	7.8	1.6	9.13	8.49
13	67.8	20.3	7.8	4.1	8.94	8.83
14	72.8	17.8	7.8	1.6	8.18	7.96
15	70.3	22.8	2.8	4.1	8.77	8.84
16	67.8	22.8	5.3	4.1	8.56	8.28
17	72.8	15.3	7.8	4.1	8.63	8.61
18	70.6	20.6	5.6	3.1	8.84	8.64
19	70.6	20.6	5.6	3.1	8.43	8.3
20	70.6	20.6	5.6	3.1	7.92	8.22
21	70.6	20.6	5.6	3.1	8.55	8.18

X1: coffee pulp, X2: mucilage broth, X3: yogurt, and X4: yeast.

presents these viability results at the end of the fermentation process, at 19 h.

Table 4 shows that, in treatment 1, there was no growth due to the absence of lactic acid bacteria (LAB) or yeasts, emphasizing the critical role these microorganisms play in the fermentation process. In treatments 2, 3, 5, and 6, where either yeast or LAB was missing, the viability was lower compared to the other treatments. Conversely, treatments that included both types of microorganisms (MOs) demonstrated a significant increase in their survival, underlining the synergistic effects of their coexistence.

This improvement in microbial survival is attributed to growth factors such as vitamins and soluble nitrogen compounds produced by yeasts, which stimulate LAB growth. In these treatments, microbial viabilities exceeded 1 × 10⁷ CFUs/g, indicating a robust and active microbial population. Additionally, yeast autolysis during the fermentation process releases further nutrients, such as amino acids, polysaccharides, and riboflavin, which are beneficial for bacterial growth [36]. This cooperative interaction between LAB and yeasts mirrors the microbial dynamics observed in other fermented foods, including cocoa, kefir, kombucha, sourdough, wine, and coffee [37]. The resulting synergy produced alcoholic notes with fruity aromas of grape, blackberry, ripe banana, sweet, and caramel.

The results presented in Table 4 demonstrate that the selected bacteria and yeast were able to adapt and survive under the challenging conditions present in the culture medium derived from coffee by-products. This adaptation equips them for the coffee fermentation environment, characterized by hypertonic conditions, a low pH, and the accumulation of organic acids and alcohols [38].

4.2.4. Statistical Analysis

The results of the statistical analysis are presented in

Table 5. Regression coefficients, R², adjusted-R², and lack of fit.

Term	p-Value		Regression Coefficients
	LAB	Yeast	LAB Viability	Yeast Viability
X1	-	-	253.9	664
X2	-	-	1698	4839
X3	-	-	6135	16,840
X4	-	-	−16,973	−48,049
X1xX2	0.005	0.002	−3215	−9047
X1xX3	0.007	0.003	−7974	−22,188
X1xX4	0.021	0.008	16,420	47,213
X2xX3	0.278	0.08	−6384	−26,004
X2xX4	0.024	0.011	19,543	53,682
X3xX4	0.081	0.163	−35,215	−59,687
X1xX2xX3	0.984	0.509	197	14,590
X1xX3xX4	0.058	0.085	56,096	110,912
X1xX2xX3Xx4	0.055	0.06	100,341	220,587
Model	Special Cubic
R2			98.86	99.23
Adjusted-R2			96.13	97.39
Lack of fit (p-value)			0.672	0.075

X1: coffee pulp, X2: mucilage broth, X3: yogurt, and X4: yeast.

.

As indicated in Table 5, the R² values exceeded 80% for both response variables, reflecting a strong fit of the regression model to the observed data. Specifically, for the LAB, the R² was 98.86%, and for the yeasts, it was 99.23%. The adjusted-R², which accounts for the number of predictors in the model and adjusts the R² value based on the number of explanatory variables and sample size, was 96.13% for the LAB and 97.39% for the yeasts. This adjustment provides a more precise measure of the explanatory power of the model, emphasizing that the adjusted-R² will always be equal to or less than the simple R².

The lack-of-fit test, utilized to determine whether the residuals (the differences between the observed and predicted values) display a random distribution or indicate a systematic pattern, suggesting that the model does not adequately capture the underlying relationship, yielded values greater than 0.05 for both the LAB and yeast. This suggests that there is no significant lack of fit [39], confirming that the model adequately fits the data for both the LAB and yeast.

The observed interactions between the coffee pulp with the yeast (X₁X₄) and the mucilage broth with the yeast (X₂X₄) demonstrated statistically significant effects on the viabilities of the LAB and yeast, with p-values of 0.021, 0.008/0.024, and 0.011, respectively. The positive regression coefficients (16,420 and 47,213/19,543 and 53,682) highlight a synergistic effect on the viability, supporting the beneficial role of these interactions in enhancing the microbial activity.

Yeasts, frequently utilized as starter cultures in coffee fermentation processes, exhibit variable behaviors depending on the coffee varieties, production regions, processing methods, and microbial species [10]. Research has shown that yeasts are effective starter cultures for diverse coffee types, including Acaiá, Catuaí vermelho, Catuaí amarelo, Bourbon amarelo, Catucaí amarelo, and Canário amarelo, across various processing methods, including dry, semi-dry, and wet processes [7,11,40,41,42]. The use of yeast strains as starter cultures has recently emerged as a promising approach to enhance the coffee quality and modify its sensory profile [43,44].

The interaction between mucilage broth and yeast aligns with expectations due to the composition of the mucilage broth, which includes glucose, galactose, and minerals, providing a rich source to produce hydrogen, lactic acid, and ethanol. These sugars are readily available for yeast consumption, supporting effective microbial viability [45].

Figure 3 presents a contour plot showing the viability of lactic acid bacteria under various combinations of components (X1, X2, and X4). Light green indicates regions of the lowest viability, while dark green denotes the highest. Optimal viability is observed in areas with coffee pulp contents (X1) exceeding 70%, mead values above 20%, and within the selected range for yeast. This outcome aligns with expectations, considering coffee pulp and mucilage broth are enriched with polysaccharides, lipids, amino acids, proteins, alkaloids, and minerals [46], offering a potent substrate for LAB fermentation. Approximately 50% of coffee by-products (dry weight) consist of polysaccharides. Specifically, cellulose (8.6–15.3%) primarily comprises glucose, and hemicellulose (31.7–41.7%) includes mannose, galactose, and arabinose, which further support LAB fermentation [47]. Additionally, studies [12,43] have demonstrated that employing LAB and yeast as starter cultures enhances coffee quality.

Figure 4 presents a contour plot showing yeast viability under various combinations of components (X1, X2, and X3). Areas shown in light green indicate regions of the lowest yeast viability, while those in dark green represent the highest viability. The plot reveals that the highest viabilities occur at the upper limits of X1, X2, and X3, suggesting that increased amounts of pulp, mucilage broth, and yeast enhance LAB growth. This correlation is logical, as a larger population of lactic acid bacteria facilitates the production of lactic acid, leading to a decrease in the pH of the medium (coffee pulp and mucilage broth) to an optimal growth pH of 4.5. The compatibility of S. cerevisiae with LAB stems from its ability to metabolize hexoses, lactic acid, and other organic acids [48]. Moreover, the use of LAB and yeast as starter cultures has gained popularity in recent years due to their role in balancing volatiles and non-volatiles, thereby improving the sensory qualities of coffee [49].

4.2.5. Optimization

The optimization was performed with defined objective values for both variables (lower: six, objective: eight, and upper: nine). It is important to note that a transformation was applied to both variables (log10x + 1). The global solution provided by Minitab19 is presented in

Table 6. Global optimization solution.

Global Solution
Components	Value
X1	74.5375
X2	18.2339
X3	2.8125
X4	4.41606

.

The desirability graph for both response variables is presented in Figure 5.

The predicted responses from the optimization indicated values for the LAB at 8.74857 and for the yeast at 8.79799, with desirability values of 0.981633 and 0.999281, respectively. Subsequent experimental validation provided results of 8.5549 ± 0.3408 for the LAB and 8.9408 ± 0.1641 for the yeast, demonstrating deviations of 0.1369 and 0.1010 between the predicted and experimental values, respectively.

Table 7. Model validation for LAB and yeast.

Parameter	Experimental Value	Predicted Value	Relative Error (%)
LAB viability	8.5549 ± 0.3408	8.74857	2.26
Yeast viability	8.9408 ± 0.1641	8.79799	1.60

presents the experimental data and those predicted by the models under the final conditions of the inoculum. The recorded values for all of the parameters fall within the 95% prediction interval and closely align with the predicted values, suggesting the reliability of the models. The relative error varied between 1.60% and 2.26%, indicating an adequate precision of the methodology developed for the optimization process and validity of the responses obtained with the experimental design.

4.3. Application of the Optimized Inoculum in Controlled Coffee Fermentation

The inoculum, once optimized, was applied to Castillo variety coffee, and key parameters such as the pH, acidity, °Brix, and viability were assessed over a 36-h fermentation period. This included comparisons with a control sample (without the inoculum). The results of these evaluations are detailed in Section 4.3.1 to Section 4.3.4. Additionally, microbiological analyses were conducted on the coffee at the start and end of the process to determine the effects of the fermentation on the microbial load, as presented in Section 4.3.5. Following the fermentation, both the control and inoculated samples were dried at 40 °C for 36 h. Subsequent analyses included ochratoxin A levels (Section 4.3.6) and sensory evaluations performed by two Q-grader tasters (Section 4.3.7).

4.3.1. Coffee Fermentation Process

In Colombian coffee farms, fermentation is traditionally conducted using bacteria, yeasts, and fungi naturally present in fresh raw materials. Incorporating starter cultures, whether isolated or not, standardizes and controls the process, guiding it to produce fragrances and flavors that enhance the suitability of the beans for the specialty coffee market. This method also involves the systematic monitoring of control variables like the pH and acidity over time.

4.3.2. pH and Acidity

The pH in this process indicates the perceived acidity and marks the progression of acid production during fermentation [26].

Ferreira et al. [50] reported final pH values of 3.9 after a spontaneous fermentation process lasting 64 h. In contrast, this study observed pH values of approximately 3.8 ± 0.17 for the control fermentation and 4.06 ± 0.085 for the fermentation using starter cultures after just 36 h. These values align with the range of 4.0 to 4.4 documented in other studies employing starter cultures, including those using single yeast strains, yeast mixtures from various species, or lactic acid bacteria [9,51]. It is important to highlight that coffee varieties significantly impact the diversity of microbiota and metabolite formation during fermentation, influencing the variability in the pH levels [52]; the aqueous extract revealed acid migration, resulting in pH reduction, indicating dissolution of these acids.

On the other hand, acidity is increasingly recognized as one of the primary quality markers in coffee [53], naturally originating from several organic acids, primarily citric acid, malic acid, and phosphoric acid. As shown in Figure 6b, the acidity for both treatments increased during the fermentation process, starting at 5.9 and 8 meq/L and ending at 43.4 and 21.9 meq/L for the control and inoculum treatments, respectively. This increase in acidity can be attributed to microbial activity during fermentation, where yeasts and bacteria, either naturally present or added as an inoculum, consume sugars and other compounds in the coffee pulp, producing various organic acids as metabolic by-products [43,54]. The primary acids produced during coffee fermentation include lactic acid, acetic acid, and, to a lesser extent, citric acid and malic acid, with the specific acids dependent on the microbial species and fermentation conditions, such as the temperature, oxygen, and pH [55,56].

4.3.3. Soluble Solids (°Brix)

Figure 7 shows that the initial soluble solids in the fermentations were 2.1 and 2.7 °Brix for each treatment, respectively. As the fermentation progressed, a decrease in the °Brix was observed, indicative of sugar consumption by the microorganisms. This reduction is likely due to the decomposition of complex carbohydrates into monosaccharides through the metabolic activities of yeast. As the fermentation advanced, the yeast population diminished, owing to the conversion of metabolites into alcohols, which in turn promoted the growth of lactic acid bacteria. These bacteria continued to consume sugars, albeit at a reduced rate [49]. It is important to note that variations in characteristics such as the pH and compounds associated with soluble solids between coffee varieties during fermentation can also be influenced by other factors, including differences in the thickness of the mucilage layer and the size and density of the beans [57].

Sugars present in the mucilage, such as sucrose, glucose, and fructose, degrade during fermentation stages, reducing the amount of soluble solids in the substrate [33]. Fermentation also produces ethanol and other products that are not soluble in water, further decreasing the content of soluble solids [58]. This reduction in soluble solids has been corroborated by research, including studies by [8,59], who conducted autoinduced anaerobic fermentation and solid-state fermentation on coffee of the Catuaí Amarelo variety, and the Typica, Caturra, and Catimor varieties, respectively.

4.3.4. Microbial Growth during Fermentation Process

In the coffee fermentation, which primarily involves bacteria and yeasts, Figure 8 shows the microbial populations of the LAB and yeasts at the beginning and end of the fermentation process. It shows that the microorganisms evaluated exhibit differing behaviors; the fermentation inoculated with starter cultures demonstrates an almost constant behavior, with only a slight decrease in the microbial population density compared to the spontaneous control fermentation, where a more significant reduction is evident. For the control fermentation, bacteria were more prevalent, with an initial value of 8.43 log CFUs/g, which decreased to 8.00 log CFUs/g by the end of the process. This is much higher than the 2.57 to 5.66 log CFUs/g range reported by other studies [42], indicating that microbial flora can vary significantly depending on the location, altitude, climate, and cultivation conditions. Additionally, a starter culture with a high viability was used in this research, increasing the bacterial population. It has been found that, in wet and semi-dry processes, there is a predominant presence of Lactiplantibacillus plantarum, Levilactobacillus brevis, Lactobacillus sp., Leuconostoc mesenteroides dextranicum, Leuconostoc sp., Leuconostoc citreum, and Leuconostoc pseudomesenteroides, while, in low-viability populations, Lactococcus lactis, Streptococcus faecalis, and Weissella sp. are noted. Interestingly, in dry processes, the LAB count was very low or sometimes absent due to low water activity (Elhalis et al., 2023). In comparison, yeasts were initially found at a value close to the LAB, but this decreased to 7.27 log CFUs/g at the end of the fermentation. For wet fermentation processes, yeast populations typically range between 3 and 7 log CFUs/mL, with common species including Saccharomyces, Schizosaccharomyces, Candida, Hanseniaspora, Pichia, Debaryomyces, Cryptococcus, and Rhodotorula [60].

The behavior observed in Figure 8 for spontaneous fermentation aligns with the findings from multiple studies, where bacterial populations not only surpassed yeasts and fungi in number but also exhibited greater richness and diversity [57,61]. This phenomenon was further explored in the metataxonomic research of Peñuela, Velazquez, and Angel [62], who collected samples from 20 coffee farms in the department of Quindío, Colombia. The authors conducted spontaneous fermentation using the same coffee variety as in this research, observing similar microbial dynamics.

In contrast, the fermentation with an added starter culture in this study showed a greater dominance by yeasts, starting with an initial value of 9.79 log CFUs/g and maintaining nearly the same level at 9.73 log CFUs/g by the end of the coffee fermentation process. This stability is likely due to the sequential inoculation strategy used during the microbial inoculum preparation, where yeasts were added last and provided with optimal conditions for sustaining the viability and presence throughout the process, facilitated by the metabolic activities of the initially added LAB [63].

Moreover, the addition of coffee pulp at its optimal point of maturity significantly contributes to the fermentation process. It enhances the production of organic acids, provides available sugars, improves the sensory profile of the beverage, and creates an environment similar to that of the coffee fruit, thus ensuring suitable microbial viability. This is supported by the study of Zhao et al. [64], which involved the same coffee variety from the Cauca department in Colombia and which was evaluated in Japan. Their study added coffee pulp from other varieties at the fermentation stage and inoculated with starter cultures, yielding results like this research, with a rich microbial population and the consistent presence of LAB and yeasts. They also found that using pulp at its optimal maturity without adding an extra sucrose source and inoculating at 7 log CFUs/mL was sufficient for effective fermentation. Higher inoculum levels (8 log CFUs/mL) did not proportionally increase organic acid production; rather, it enhanced certain acids while reducing others, as competition for nutrients and space can occur with larger microbial populations, depending on the inoculum type and nutrients present in the substrate.

To determine if there was a statistically significant difference between the viabilities of lactic acid bacteria and yeasts at the beginning and end of the fermentation process, a t-test for the means of two paired samples was performed, identifying that the population had no significant differences in the samples with the inoculum, while there was a reduction in the population of yeasts in the control fermentation. Microorganisms such as yeast play a key role in mucilage degradation through the production of various enzymes, alcohols, and acids during the fermentation process. This reduction in the yeast population can be attributed to the fact that, during fermentation, yeasts use the sugars present in the coffee mucilage as their main source of energy [6]. As these sugars are consumed, the substrate becomes less favorable for their growth, decreasing at the end of fermentation.

4.3.5. Microbiological Analysis

The coffee fermentation process occurs spontaneously through the action of epiphytic microorganisms present on the coffee fruit, influenced by the chemical composition of the bean [65]. During spontaneous fermentation (control), microbial successions typically start with bacteria, followed by yeasts [66]. As shown in

Table 8. Microbiological results at the beginning and at the end of the fermentation process (control and inoculum).

Type of Fermentation	Microorganism	log (CFUs/mL)
		t = 0 h			t = 36 h
		10−2	10−3	10−4	10−2	10−3	10−4
Control	Mesophilic aerobic bacteria	>300	>300	7.55	>300	>300	7.15
	Fungi and yeast	>300	>300	7.25	>300	>300	8.49
	Enteric bacteria	>300	5.72	6.57	4.14	0	0
	Lactic acid bacteria	>300	>300	>300	>300	>300	8.22
Inoculum	Mesophilic aerobic bacteria	>300	>300	7.08	5.5	6.5	7.5
	Fungi and yeast	>300	>300	8.31	>300	>300	9.00
	Enteric bacteria	>300	5.40	5.85	0	0	0
	Lactic acid bacteria	>300	>300	>300	>300	7.37	8.17

, the initial microbial population at time 0 was higher for mesophilic, enteric, and lactic acid bacteria compared to the counts at 36 h. This trend is attributed to the fact that, during fermentation, microorganisms consume and metabolize the sugars and acids present in the coffee mucilage, leading to a decrease in the population of enteric microorganisms as the substrates available for their growth and metabolism [33] are degraded. Additionally, the pH of the substrate decreases during fermentation due to lactic acid production by lactic acid bacteria, which can inhibit the growth of enteric microorganisms that are not acidophilic [45].

For yeasts, an increase in the population was observed from 7.25 log CFUs/mL to 8.49 log CFUs/mL (control) and from 8.31 log CFUs/mL to 9.00 log CFUs/mL (inoculum). This pattern aligns with findings from [67], who observed a rise in yeast populations (Hansinaspora uvarum and Pichia kudriavzevii) to maximum levels of 10.0 log CFUs/mL each after 36 h of fermentation. The growth in the yeast population during the coffee fermentation is attributable to their access to simple sugars and other nutrients released from the coffee mucilage, providing an essential energy source for their proliferation. [12,68]. Additionally, the fermentation was carried out under optimal conditions for yeast growth, including a moderate temperature of 33 ± 2 °C and a sufficient humidity at 80%, which facilitated the metabolic activity of the yeasts.

Spontaneous fermentations (control) present several disadvantages compared to those initiated with starter cultures. The use of starter cultures enhances the structural and sensory characteristics of coffee and reduces the risk of harmful microbial growth [69]. This is evidenced by the results where the fermentation with inoculum showed fewer mesophilic bacteria and an absence of enteric bacteria after 36 h.

Regarding lactic acid bacteria, there was a significant reduction in the population from >300 to 8.22 log CFUs/mL (control) and from >300 to 8.17 log CFUs/mL (inoculum). These observations are consistent with [68], who reported a decrease in LAB biomass after 24 h, starting from initial counts between 2.6 and 3.1 log CFUs/mL and decreasing to <2 log CFUs/mL by 36 h.

4.3.6. Ochratoxin A

Ochratoxins are a group of mycotoxins primarily produced by certain species of Aspergillus (e.g., A. ochraceus) and Penicillium (e.g., P. verrucosum) in various agricultural crops, including cereal grains, peanuts, dried beans, and coffee beans [70]. Within the ochratoxin family, there are over 20 different metabolites, with ochratoxin A (OTA) being the most abundant and toxic [71,72]. OTA has been linked to human Balkan Endemic Nephropathy (BEN) and is also associated with tumors in the human urinary tract [71]. The International Agency for Research on Cancer (IARC) classifies OTA as a possible human carcinogen [73]. Additionally, it is considered as the second most significant mycotoxin from a public health perspective due to its nephrotoxic, hepatotoxic, genotoxic, teratogenic, and immunosuppressive effects [73,74].

Regulations specify the maximum allowable OTA levels in instant and roasted coffee at less than 5 and 10 µg/kg, respectively [75]. OTA contamination in coffee can vary based on several factors, including handling during harvest, wet or dry processing methods, climate, geography, and roasting [76].

In the coffee samples analyzed (control and fermented with inoculum) in this study, OTA was not detected using the HPLC-FL-207 method, indicating that the coffee is safe for consumption, as shown in Figure 9.

In the case of the fermented coffee, the absence of ochratoxin A (OTA) can potentially be attributed to the action of yeasts or lactic acid bacteria (LAB), which are known as Generally Recognized As Safe (GRAS) for animal and human consumption [77,78]. These microorganisms have been extensively utilized to mitigate mycotoxin contamination in food matrices. Both LAB and yeasts can inhibit the biosynthesis of mycotoxins or detoxify mycotoxins after their formation, either by binding them to their cell walls or degrading them into less toxic compounds [79].

Additionally, both the fermented coffee and the control sample were subjected to drying processes in a forced convection oven at 40 °C for 36 h in thin, homogeneous layers on trays. This method ensured adequate temperature distribution over time. It is important to emphasize that undesirable mycotoxigenic fungi are known to proliferate during the post-harvest period of coffee, primarily due to high humidity levels [80].

4.3.7. Sensory Evaluation

The sensory attributes of sweetness, acidity, and bitterness are detected by the sense of taste through the taste buds on the tongue. These attributes are integral to the five fundamental tastes—sweet, bitter, salty, sour, and umami—that humans can distinguish, playing a crucial role in food characterization. Sweetness in coffee beans is linked to their carbohydrate content; acidity is derived both from organic acids naturally present in the beans and from fermentation processes; bitterness is attributed to compounds such as caffeine and chlorogenic acids [81].

Trigonelline, another significant compound in coffee, influences the aroma perception of brewed coffee and generates key volatile compounds such as pyridines and pyrroles during roasting [82]. Additionally, the tactile sensations of body and astringency are perceived through the oral somatosensory system and relate to the non-volatile components of the beverage, such as fatty acids. The aftertaste attribute assesses how long the flavors of the beverage linger in the mouth post-consumption. A balance among these sensory attributes in terms of intensity and quality is essential to achieve sensory harmony, culminating in a final score that reflects the sensory quality of the coffee [81].

Significantly, the total fermentation time of 36 h influenced the sensory profile of the coffee, resulting in higher cup scores of 84.25 and 84.50 for the fermented samples compared to 83.25 and 83.00 for the unfermented control samples. The detailed results of the sensory analysis conducted on each sample are presented

Table 9. Results of the sensory analysis of the fermented coffee samples without the inoculum (control).

Item	Characteristics	Qualification
1	Fragrance/aroma	7.5
2	Flavor	7.5
3	Residual flavor	7.25
4	Acidity	7.25
5	Body	7.5
6	Uniformity	10.0
7	Sweetness	10.0
8	Clean cup	10.0
9	Balance	7.25
10	Taster score	7.50
Final score		81.75

Notes: chocolate and caramel fragrance, chocolate and caramel aroma, molasses, caramel and chocolate flavor, peppery residue with chocolate sensation, full body, and low-intensity citric acidity.

and

Table 10. Results of the sensory analysis of the fermented coffee samples with the inoculum.

Item	Characteristics	Qualification
1	Fragrance/aroma	8.00
2	Flavor	8.00
3	Residual flavor	7.75
4	Acidity	7.75
5	Body	7.50
6	Uniformity	10.0
7	Sweetness	10.0
8	Clean cup	10.0
9	Balance	7.75
10	Taster score	7.75
Final score		84.50

Notes: butter, vanilla and panela fragrance, red apple and hazelnut aroma, panela, caramel, cherry and subtle red wine flavor, spicy residual, and honeyed citric acidity.

.

The highest cup score was achieved by samples fermented with an inoculum, attributable to co-inoculation with yeast and lactic acid bacteria (BAL). This method has demonstrated the potential for developing desirable aromatic compounds, essential for producing specialty coffees [83]. Recent research supports the use of mixed bacterial and yeast starter cultures in the fermentation of foods, including coffee fruit. This approach not only enhances sensory quality but also promotes the development of toxin-free cultures [43,63,83].

Acidity in coffee is a crucial organoleptic attribute that can either enhance or detract from the beverage’s overall appeal, depending on the nature and concentration of the main acids present. Properly managed acidity adds vivacity to coffee, boosting other attributes such as sweetness. Conversely, excessive acidity can lead to unpleasant or unusual flavors [16]. Both coffees achieved commendable cup scores; however, the coffee fermented with LAB and yeast scored higher, at 84.50, classifying it as “very good” specialty coffee and showcasing the benefits of targeted fermentation processes.

5. Conclusions

Based on the results presented, it can be concluded that the medium composed of coffee pulp and honey water, inoculated with lactic acid bacteria (Lactobacillus delbrueckii ssp. bulgaricus and Streptococcus thermophilus) along with yeast (Saccharomyces cerevisiae), exhibited ideal characteristics for coffee fermentation processes. During fermentation, there was a reduction in the pH from 4.47 to 4.05 and in the ºBrix from 15.8 to 8.45. Additionally, there was a significant increase in the acidity, from 20.02 mg of chlorogenic acid/g to 42.69 mg of chlorogenic acid/g. The microorganisms maintained an optimal population exceeding 10⁷ log CFUs/g, which is conducive for effective fermentation. Post-fermentation samples with the inoculum also demonstrated the absence of enteric microorganisms and ochratoxin A (OTA), indicating high quality. The use of inocula during the fermentation process not only enhances the cup score but also enables coffee growers to produce specialty coffees. Future research should explore the preparation of inocula with coffee pulp from various varieties and examine how different microorganisms influence the flavors and aromas perceived in the coffee.