Microbial Dynamics and Quality Evolution in the Spontaneous Fermentation of the Traditional Meat Product Sjenica Sheep Stelja

Abstract

The Sjenica sheep stelja is a characteristic, traditional dry-cured meat product from Serbia with unique and recognizable sensory attributes. The methodology involved examining physicochemical measurements, followed by sensory evaluation and microbiological analyses, over a 120-day ripening period across three years and three different villages, as well as the correlation between chemical characteristics and the number of specific groups of bacteria. Results showed consistent quality parameters across producers and production periods, with notable variation in fat, protein, and ash content. Sensory evaluation confirmed that the product met the quality standards outlined in the Elaborate for the Protection of Geographical Indication, with minor differences in color, aroma, chewiness, and taste among samples. The microbiological analysis demonstrated the dynamic nature of microbial communities throughout maturation, including changes in the counts of aerobic mesophilic bacteria, Enterobacteriaceae, Pseudomonadaceae, lactic acid bacteria, and molds. Penicillium species, particularly P. nalgiovense and P. solitum, were consistently identified, while other fungal genera exhibited varying distribution patterns. The correlation analysis highlights the complex influence of chemical parameters on microbial dynamics throughout the aging process. These findings emphasize the influence of traditional production methods, regional variations, and chemical composition on the sensory quality and microbial safety of Sjenica Sheep Stelja, providing valuable insights for future research and quality control.

1. Introduction

Fermented foods play a crucial role in global human nutrition. Fermentation, one of the oldest biotechnological processes, facilitates the transformation of perishable raw materials into stable, preserved products. Over time, fermentation techniques have advanced alongside human civilization, culminating in sophisticated biotechnological applications [1].

Spontaneously fermented foods, produced without preservatives, additives, or starter cultures, serve as reservoirs of beneficial microorganisms with desirable technological and probiotic properties. These microbial communities contribute to the safety, functionality, and consistency of fermented products when utilized as starter cultures [2,3]. Traditional autochthonous fermented products, deeply rooted in cultural heritage and specific geographical regions, are distinguished by unique production techniques and microbial compositions, resulting in distinct sensory attributes [3,4,5].

Fermented dry meat products hold significant value in the global market due to their sensory characteristics and cultural importance. Regulatory protection, such as the Protected Designation of Origin (PDO) certification, ensures the preservation of these products, while the standardization of production methods upholds quality and hygiene standards. The microbial diversity of traditional fermented meat products is influenced by local processing techniques and the geographical origin of raw materials [6] as well as the hygienic and sanitary conditions of slaughter and meat processing [7]. The primary microbiota in traditional meat products consists of both technologically relevant microorganisms and potential food spoilage agents. Spoilage microorganisms mainly include Gram-negative, oxidase-positive bacteria derived from the skin, gastrointestinal tract, or the muscle tissue of animals [8]. Dominant microbial genera in raw meat include Acinetobacter, Pseudomonas, Brochothrix, Flavobacterium, Psychrobacter, and Moraxella, along with Enterobacteriaceae, lactic acid bacteria (LAB), coagulase-negative staphylococci (CNS), yeasts, and molds [9,10].

LAB play a critical role in reducing pH, inhibiting microbial spoilage, and enhancing food safety. They produce bacteriocins effective against Gram-positive foodborne pathogens such as Listeria monocytogenes, Staphylococcus aureus, and Bacillus cereus [3,11,12]. Additionally, LAB contribute to muscle protein coagulation, improving texture and flavor through enzymatic processes such as cathepsin D-mediated proteolysis [10]. Another important microbial group in meat fermentation is coagulase-negative cocci (CNC), including Staphylococcus, Kocuria, and Micrococcus species [3,11,12,13]. CNC stabilize the red color of meat products by reducing nitrates to nitrites, promoting nitrosomyoglobin formation, which gives cured meat its characteristic appearance [13,14]. Their metabolic activities, including proteolysis and lipolysis, further enhance texture and sensory properties [3,11,12,13]. Yeasts and molds primarily colonize the surface of meat products, forming a protective biofilm that prevents dehydration and lipid oxidation. These organisms are generally halotolerant and acid-tolerant, making them potential candidates for starter cultures. However, molds are seldom used in meat fermentation due to the risk of mycotoxin production, which poses safety concerns [15].

The Western Balkans is known for its diverse range of autochthonous dry meat products, which are traditionally produced through spontaneous fermentation. One such product is salted and dried boneless sheep meat, commonly referred to as “stelja”, “pastrma”, or “kastradina”. This product is characteristic of several Balkan countries, including Bosnia and Herzegovina, North Macedonia, Croatia, Montenegro, and Serbia [16]. A notable traditional sheep meat product from Western Serbia, specifically in Sjenica and Tutin, is Sjenica sheep stelja (in serbian: Sjenička ovčija stelja), which is classified as a dry-cured sheep ham with PDO certification status [16,17]. The production process follows specific traditional methods [18]. The microbiota of Sjenica sheep stelja comprises autochthonous bacterial and fungal strains with largely unexplored functional potential. Therefore, the primary objective of this study is to comprehensively characterize the autochthonous microbiota associated with Sjenica sheep stelja during ripening and to assess its functional role in shaping physicochemical characteristics and product quality during maturation.

2. Materials and Methods

2.1. Autochthonous Fermented Product: Sjenica Sheep Stelja

The Pešter Plateau (altitude: 1100–1250 m) is renowned for its traditional fermented meat products, including Sjenica sheep stelja (Figure 1), which is made from the autochthonous sheep breed. In rural households around Sjenica, the production process begins with the slaughter, primary processing, and cooling of carcasses at 4 °C for 24 h. Subsequently, the carcasses are fully deboned, with the exception of the inner thigh (shank) and the distal ends of the tibia and fibula bones, which are retained up to 5 cm in length.

The dry salting process involves manually rubbing the meat with 3–3.5% pure table salt. The salted meat is then stored at a temperature of 4–7 °C and a relative humidity (RH) of 80–90% for 5–8 days. During this period, the salt penetrates the meat tissue, while excess moisture is released, covering the pieces almost entirely.

Residual surface salt is removed from the carcasses just before smoking. They are then suspended on rods and left in a well-ventilated space for 2–3 h to allow excess moisture to drain. This process can be accelerated by wiping the meat with clean cloths. Cold smoking is performed using beechwood smoke at a temperature of 16–18 °C and an RH of 65–80% for 15 days, ensuring that the carcasses are evenly exposed to smoke without direct contact.

Following smoking, the drying phase is conducted at 4–10 °C with RH maintained at 60–70%, depending on ambient weather conditions. The maturation process lasts a minimum of 5 to 6 months, although the highest quality is achieved after 10 to 12 months of aging. This extended maturation period allows for the development of complex sensory attributes, contributing to the distinctive flavor, aroma, and texture characteristic of Sjenica sheep stelja.

2.2. Sampling of Sjenica Sheep Stelja

For experimental purposes, Sjenica sheep stelja (Figure 1) produced in selected households in the villages of Blato (A), Krajinovići (B), and Rasno (V) in the Pešter region was utilized. Samples were collected across three distinct production seasons (2016/17—1; 2017/18—2; 2018/19—3), corresponding to the late autumn and early spring periods. The production process was standardized across all selected households, strictly adhering to traditional methods as described below.

Sheep stelja samples, each weighing 300 g, were collected under aseptic conditions for the assessment of chemical quality parameters and the enumeration, isolation, and identification of the autochthonous microbiota. For sensory evaluation, carcass portions of 2000 g were selected. All analyses were performed in triplicate. Microbiological analyses were conducted on the following production days: day 0 (raw sheep meat) and days 7, 14, 28, 60, 90, and 120 during traditional product maturation. Chemical and sensory analyses were carried out on the 120th day of ripening.

Samples were transported to the laboratory in mobile refrigeration units and stored at 4 °C. All analyses were conducted 24 h post-sampling, following the guidelines established in the Official Rule of the Republic of Serbia [19] and the Guide for the Application of Microbiological Criteria for Food [20].

2.3. Determination of Physicochemical Properties of Sjenica Sheep Stelja

The chemical analysis of this autochthonous product included the examination of pH value, water content, water activity (a_w value), total fat and protein content, salt, and ash content.

2.3.1. pH Value Analysis

The pH value of the product was determined using a portable pH meter (Testo 150, Testo, AG, Sparta, NJ, USA) designed for direct measurement in meat and meat products. The pH meter was calibrated with standard phosphate buffers (pH 4 and 7 at 20 °C). The pH value was determined according to the reference method SRPS ISO 2917 [21].

2.3.2. Water Activity (a_w) Analysis

Water activity (a_w) was measured using a portable aw meter (LabSwift–Novasina, Lachen, Switzerland). The a_w value determination involved grinding the samples and filling the measuring container up to 2/3 of its height, followed by placing the container in the probe’s measuring unit at room temperature (approximately 20 °C) until measurement equilibrium was established.

2.3.3. Moisture Content Analysis

Water content was determined using the reference method, SRPS ISO 1442 [22]. The procedure involved thoroughly mixing 3 ± 0.001 g of the sample with quartz sand, followed by drying to a constant mass at a temperature of 103 ± 2 °C. The difference between the two consecutive measurements did not exceed 1 mg. This process was performed using a TGA701 thermogravimetric analyzer (LECO Corporation, St. Joseph, MI, USA), with results expressed as a percentage (%).

2.3.4. Total Fat Content Analysis

Total fat content in the product was determined using the reference method, SRPS ISO 1443 [23]. This method is based on hydrolyzing a portion of the sample with diluted hydrochloric acid, filtering the resulting mass, and washing it with hot distilled water until neutral pH was reached. After filtration, the remaining fat was dried on filter paper and then extracted with petroleum ether using a Soxhlet apparatus for 5 h. The sample extract was dried in an oven to a constant mass at 103 ± 2 °C. Free fat content was expressed as a percentage (%).

2.3.5. Total Protein Content Analysis

Total proteins were determined using the reference Kjeldahl method, SRPS ISO 937 [24]. The protein content in stelja was determined based on the total nitrogen content multiplied by a factor of 6.25. The method involved digesting the sample in concentrated sulfuric acid, with copper (II) sulfate as a catalyst to convert total nitrogen into ammonium ions ((NH₄)₂SO₄). Alkalization was performed using sodium hydroxide, followed by the distillation of the released ammonia into an excess boric acid solution and titration with hydrochloric acid to determine the ammonia bound to boric acid. Protein content was expressed as a percentage (%).

2.3.6. Total Salt Content Analysis

Salt content determination was performed using the standard method, SRPS ISO 1841-1 [25]. This method is based on extracting the sample with hot water and precipitating proteins. A silver nitrate solution was added in excess to the obtained extract, followed by titration with a potassium thiocyanate solution. Results were expressed as percentages (%).

2.3.7. Total Ash Content Analysis

Total ash in the samples was determined using the method SRPS ISO 936 [26], which involves drying, carbonization, and incineration at a temperature of 550 ± 25 °C. The mass of the gray-white residue was measured after cooling. The entire procedure required approximately five hours, with ash content reported as a percentage (%).

2.4. Determination of Sensory Properties of Sjenica Sheep Stelja

The sensory properties of Sjenica sheep stelja were evaluated using quantitative descriptive analysis (QDA), following the quantitative descriptive test [27]. The assessment was conducted by a trained eight-member panel [28] in a sensory evaluation laboratory designed in accordance with SRPS ISO 8589 [29]. The sensory analysis was conducted at the accredited Laboratory for Food Product Testing, Faculty of Technology, University of Novi Sad (Accreditation Body of Serbia, accreditation number 01-059).

The sensory evaluation included the assessment of the following attributes: external appearance, cross-sectional appearance, color and color stability, odor and taste, texture, and juiciness. A five-point structured scale was used, where each score corresponded to a specific quality level: a score of 0 indicated products with evident mechanical and/or microbiological defects, while a score of 1 denoted altered, atypical properties and an unacceptable product. A score of 2 reflected significant quality defects in color or other product attributes, significantly affecting quality. A score of 3 indicated noticeable defects in color or other product attributes, reducing overall quality. A score of 4 represented slight deviations or minor quality imperfections that were within acceptable quality limits, and 5 signified exceptional, typical sensory properties of optimal quality.

The mean sensory scores for external appearance, cross-sectional appearance, color and color stability, odor and taste, texture, and juiciness were determined. The overall weighted sensory quality score was calculated by multiplying the score of each attribute by its respective importance coefficient (IC): 2, 5, 3, 7, and 3, respectively. The final score was obtained by summing the weighted values and dividing by 20.

2.5. Microbiological Analysis of Sjenica Sheep Stelja

2.5.1. Enumeration of Autochthonous Microbiota

The preparation of samples and the isolation of the autochthonous microbiota from the product were conducted according to SRPS ISO 6887-1 [30], the standard method for sample preparation. The enumeration of microorganisms in Sjenica sheep stelja was performed at different production stages: day 0 (raw sheep meat analysis) and days 7, 14, 28, 60, 90, and 120 of traditional product maturation. Sampling was carried out across three households (A, B, and V) during three distinct production seasons (1, 2, and 3).

Sterile instruments were used to collect samples (300 g) from the interior of the product. A 10 g portion of each sample was finely minced and aseptically transferred into 90 mL of the sterile physiological peptone solution (0.8 g NaCl/mL and 1 g peptone/mL), which had been sterilized at 121 °C for 15–20 min. The sample was then vortexed for 15 min to achieve complete homogenization.

A series of decimal dilutions (up to 10⁻⁷) was prepared from the initial homogenate, and triplicate plating was performed on both selective and non-selective media. The total aerobic mesophilic bacterial count was determined following SRPS ISO 4833-2 [31]. The enumeration of Enterobacteriaceae was conducted using SPRS ISO 21528-2 [32] on Violet Red Bile Glucose Agar (VRBGA, Oxoid, Basingstoke, UK), along with confirmatory biochemical tests (oxidase test and glucose fermentation). The enumeration of coagulase-positive staphylococci was performed following SRPS ISO 6888-2 [33], and lactic acid bacteria (LAB) were enumerated using ISO 13721 [34], while Enterococcus spp. were isolated on Bile Esculin Agar (BEA, Torlak, Belgrade, Serbia) [5]. The detection of yeasts and molds followed SRPS ISO 21527-1 [35]. The presence of Salmonella spp. and Listeria monocytogenes was determined using the standard methods SRPS ISO 6579-1 [36] and SRPS ISO 11290-1 [37], respectively. The Pseudomonadaceae family was enumerated according to SRPS EN ISO 13720 [38] using Pseudomonas Agar Base (Oxoid, Basingstoke, UK).

Depending on the microorganism type, incubation was performed at 30 °C and 37 °C for 48 h. Colony enumeration was conducted on plates containing 20 to 300 colonies to ensure accuracy. The results were expressed as logCFU/g of the product.

2.5.2. Isolation and Identification of LAB

After enumeration, LAB colonies were streaked on new agar plates for purification.

A total of 451 isolates were subjected to microscopic observations, Gram staining, and catalase tests. Furthermore, Gram-positive and catalase-negative isolates of LAB were identified to the genus level using the following tests: production of carbon dioxide from glucose by subculturing the isolates in MRS broth with Durst’s tubes, growth and production of exopolysaccharides (slime) on MRS agar with sucrose (20.0 g/L), and L-arginine and esculin hydrolysis. Selected isolates from different genera were further identified to the species level using API 50CH (BioMerieux, Montalien-Vercien, France) tests. Isolated Enterococcus strains were further identified to the species level using Microgen Strep ID (Microgen Bioproducts, Surrey, UK). The results of the API 50CH and Microgen Strep ID tests were interpreted according to the manufacturer’s instructions for each respective product. The isolated strains were stored at −20 °C and −80 °C in M17 (Merck, GmbH, Darmstadt, Germany) (for cocci) and in MRS (Torlak, Belgrade, Serbia) (for rods) broth containing 20% glycerol (v/v) [39]. Working cultures were revitalized after two consecutive transfers in M17 or MRS broth at 32 °C.

Final identification was conducted using MALDI-TOF MS, as described in detail by Muruzović et al. [40].

2.5.3. Isolation and Identification of Molds

The isolation of molds was performed using Dichloran 18% Glycerol Agar (DG18) (Oxoid, Basingstoke, UK) from the surface of Sjenica sheep stelja samples. The sample was placed onto the surface of DG18 agar and left for 30 s to allow for the transfer of mold spores from the product surface to the medium. The plates were incubated at 25 °C for five days. Mold isolation was conducted following the method described by Pitt and Hocking [41].

Mold isolates were selected based on their morphological characteristics, which were used for identification at the genus level. To obtain pure monocultures, molds belonging to the genera Aspergillus, Eurotium, and Penicillium were subcultured on CYA agar (HiMedia, Kennett Square, PA, USA), while those from the Mucor genus were subcultured on SMA agar. The mold monocultures obtained were identified to the species level according to taxonomic keys [41,42,43,44]. The identification process involved the examination of

Macromorphological characteristics: colony diameter, color, texture, pigmentation, colony reverse, and exudates.

Micromorphological characteristics: metulae, phialides, conidia, hyphae, including their length, diameter, size, and shape.

The isolated and identified mold cultures were preserved on SMA agar at 4 °C in the Food Microbiology Laboratory Collection at the Faculty of Technology, University of Novi Sad.

2.6. Statistical Analysis

All data were presented as means ± standard deviations, and statistical analysis was performed using Microsoft Excel (Redmond, Washington, DC, USA). A one-way analysis of variance (ANOVA) was conducted to test the equality of the arithmetic means of the examined groups. Duncan’s test was used for post hoc comparisons of all results. Additionally, a two-way analysis of variance was conducted to test the differences in chemical characteristics between households and seasons, and Tukey’s test was used for post hoc comparisons of all results. Correlation analysis was used to compare chemical parameters with the abundance of the examined groups of bacteria and molds. All statistical analyses were performed using IBM SPSS Statistics version 20.

3. Results

3.1. The Chemical Analysis of Sjenica Sheep Stelja

The chemical analysis of Sjenica Sheep Stelja included the examination of pH values, water activity (a_w), moisture content, fat, protein, salt, and ash in the final product after 120 days of maturation. Based on the results, it can be observed that the quality parameters investigated were relatively uniform across all producers and during all the periods examined. The results of the chemical characteristics analysis of Sjenica Sheep Stelja are presented in

Table 1. Results of the chemical characteristics analysis of Sjenica Sheep Stelja.

Chemical Characteristic/ Sample	pH	aw	Moisture Content (%)	Total Fat Content (%)	Protein Content (%)	Salt Content (%)	Ash Content (%)
A1	5.47 ± 0.20 a	0.82 ± 0.01 a	43.70± 1.10 ab	10.70 ± 0.10 c	33.90 ± 1.60 abc	5.30 ± 0.20 abc	9.80 ± 0.50 ab
A2	5.43 ± 0.19 a	0.80 ± 1.00 a	46.1 ± 0.60 c	7.25 ± 0.04 a	37.00 ± 0.50 acd	5.00 ± 0.14 a	9.17 ± 0.04 a
A3	5.44 ± 0.04 a	0.82 ± 0.50 a	45.6 ± 0.60 bc	7.60 ± 0.07 ab	36.30 ± 0.60 bc	5.20 ± 0.06 ab	9.30 ± 0.09 a
B1	5.58 ± 0.03 a	0.80 ± 0.01 a	42.9 ± 0.10 a	11.60 ± 0.10 d	34.20 ± 0.70 abc	5.10 ± 0.10 ab	9.10 ± 0.00 a
B2	5.15 ± 0.13 a	0.81 ± 0.01 a	42.5 ± 0.40 a	13.30 ± 0.29 e	32.60 ± 1.10 a	5.50 ± 0.00 c	10.90 ± 0.30 b
B3	5.42 ± 0.21 a	0.79 ± 0.01 a	44.1 ± 0.00 ab	7.90 ± 0.10 ab	38.50 ± 0.50 d	5.28 ± 0.00 ab	9.30 ± 0.00 a
V1	5.60 ± 0.12 a	0.83 ± 0.03 a	44.2 ± 1.30 ab	10.90 ± 0.50 c	32.90 ± 2.50 ab	5.20 ± 0.20 abc	9.30 ± 1.30 a
V2	5.51 ± 0.50 a	0.80 ± 0.01 a	47.1 ± 0.10 c	8.00 ± 0.00 b	33.25 ± 0.60 ab	5.50 ± 0.10 bc	10.50 ± 0.00 ab
V3	5.5 ± 0.01 a	0.80 ± 0.01 a	46.9 ± 0.30 c	7.53 ± 0.30 ab	35.50 ± 0.30 abc	5.33 ± 0.0 ab	9.50 ± 0.20 ab

Mean values ± SD from triplicate measurements for each sample; A—Blato village; B—Krajinoviće village; V—Rasno village; 1—2016/17; 2—2017/18; 3—2018/19; average values marked with the same letters within the same column (variable) do not differ significantly (p ˃ 0.05).

.

A two-way analysis of variance (ANOVA) was conducted to examine the effects of household (A, B, V) and season (1, 2, 3) on the chemical parameters of bedding samples. The results are summarized in Supplementary Figure S1.

For pH values, no statistically significant interaction was observed between household and season, with F(4, 18) = 0.85 and p = 0.51. The main effect of season was also not significant, with F(2, 18) = 1.82 and p = 0.19. Post hoc comparisons using Tukey’s HSD test revealed no significant differences in pH among household groups A (M = 5.45, SD = 0.14), B (M = 5.38, SD = 0.22), and V (M = 5.54, SD = 0.26), or across seasons.

For water activity (a_w), there was no significant interaction effect between household and season, with F(4, 18) = 1.17 and p = 0.36. Similarly, the main effect of season was not statistically significant, with F(2, 18) = 1.17 and p = 0.33. Tukey’s test indicated no significant differences among the household groups: A (M = 0.81, SD = 0.03), B (M = 0.80, SD = 0.01), and V (M = 0.81, SD = 0.02), or among seasons.

The analysis of moisture content showed a statistically significant interaction between household and season, with F(4, 18) = 5.62 and p = 0.004. Significant main effects were also observed for both household, with F(2, 18) = 44.50 and p < 0.001, and season, where F(2, 18) = 22.55 and p < 0.001, with a large effect size (adjusted R² = 0.85). Post hoc comparisons using Tukey’s test indicated significant differences in moisture content among the household groups: A (M = 45.18, SD = 1.27), B (M = 43.22, SD = 0.78), and V (M = 46.11, SD = 1.58), as well as across seasons, with the exception of season 2 (2017/18), which did not differ significantly from the others.

The analysis of total fat content revealed a highly significant interaction effect between household and season, where F(4, 18) = 157.19 and p < 0.001. Significant main effects were also found for household, where F(2, 18) = 282.29 and p < 0.001, and season, where F(2, 18) = 463.88 and p < 0.001, with a very large effect size (adjusted R² = 0.99). Post hoc comparisons using Tukey’s HSD test showed significant differences in total fat content among the household groups, A (M = 8.52, SD = 1.64), B (M = 10.96, SD = 2.41), and V (M = 8.82, SD = 1.60), as well as across the different seasons.

The analysis of protein content revealed a statistically significant interaction between household and season, where F(4, 18) = 9.65 and p < 0.001. Both main effects were also significant: household, with F(2, 18) = 5.93 and p = 0.01, and season, with F(2, 18) = 23.07 and p < 0.001, with a large effect size (adjusted R² = 0.77). Post hoc analysis indicated that household V (M = 34.07, SD = 1.71) differed significantly from household A (M = 35.76, SD = 1.60) and B (M = 35.45, SD = 3.22). Additionally, season 3 (2018/19) showed a significant difference compared to the other two seasons.

The analysis of salt content revealed a statistically significant interaction effect between household and season, where F(4, 18) = 5.45 and p = 0.005. The main effect of households approached statistical significance, with F(2, 18) = 3.47 and p = 0.053, while the main effect of season was not statistically significant, with F(2, 18) = 2.85 and p = 0.08. The effect size was moderate, with an adjusted R² of 0.50. However, post hoc comparisons using Tukey’s HSD test indicated no significant differences between households or across seasons.

Finally, the analysis of ash content demonstrated a statistically significant interaction effect between household and season, where F(4, 18) = 5.01 and p = 0.007. While the main effect of households was not significant, with F(2, 18) = 1.43 and p = 0.265, the effect of season was significant, where F(2, 18) = 7.70 and p = 0.004, with a moderate effect size (adjusted R² = 0.54). Tukey’s test indicated that season 2 (2017/18) differed significantly from the other two seasons.

3.2. Sensory Analysis of Sjenica Sheep Stelja

The results of the sensory quality assessment of the Sjenica Sheep Stelja final product collected from three households (A, B, and V) over three production periods (2016/17—1; 2017/18—2; 2018/19—3), are presented in Figure 2.

During the first (2016/17—1) investigation period (Figure 2a), no defects, stains, or discolorations were observed in the samples analyzed. The mean score for external appearance (typical shape/form) was 6.86 for samples from Blato (A) and Rasno (V), while the sample from Krajinoviće (B) received a mean score of 6.75, with no statistically significant differences between the samples (p > 0.05). The intensity of the cross-sectional color in the first production year was rated within an average range of 3.78 to 4.50, with a statistically significant difference detected for samples from Blato (A) (p < 0.05).

The fat tissue color in Sjenica Sheep Stelja received mean scores ranging from 6.06 (sample A1) to 6.22 (sample V1), aligning with the specifications outlined in the Elaborate for the Protection of Geographical Indication. The rating for intermuscular fat content was slightly lower, ranging between 2.81 (sample A1) and 3.67 (sample V1), with a statistically significant difference noted (p < 0.05). The characteristic aroma of the product was confirmed by a high sensory evaluation score, exceeding a mean value of 6, with sample V1 receiving an average score of 6.53. Chewiness and texture were assessed with mean scores ranging from 5.81 to 6.14, with the sample from Rasno (V) being perceived as juicier during mastication. The characteristic salty taste of Sjenica Sheep Stelja was rated within an average range of 4.06 to 4.11 (p > 0.05). The product’s overall flavor was evaluated as usual, with mean scores ranging from 6.19 to 6.36, and a statistically significant difference was observed for sample A1 (p < 0.05). Sensory analysis further confirmed the absence of rancidity in the tested samples from the 2016/17 production year (mean value = 1.00).

The sensory evaluation of samples from the second (2017/18—2) production year (Figure 2b) was conducted in accordance with the Elaborate for the Protection of Geographical Indication of Sjenica Sheep Stelja. The mean score for external appearance (typical shape/form) was lowest for the producer in Krajinoviće (B2, 5.53, p < 0.05), while the highest mean score was recorded for the product from the Blato village household (A2). The fat tissue color received mean scores ranging from 6.00 to 6.31, with no statistically significant differences among samples (p > 0.05). However, the marbling of the samples did not fully meet the panelists’ expectations, as this characteristic received lower sensory scores (mean values ranging from 2.11 to 2.58). Nonetheless, the product’s consistency at the cross-section was rated highly, with sample A2 receiving the highest rating.

The characteristic aroma, chewiness, and salivation effect (dryness) were rated relatively high during the sensory evaluation of the second production year. Statistically significant differences were noted for aroma between samples B2 and V2 and for chewiness between samples A2 and V2 (p < 0.05). The product’s taste was assessed as usual, and a statistically significant difference was detected between samples A2 and B2 (p < 0.05). The highest intensity of smoke aroma and flavor was recorded for samples from Blato (A2), while the presence of rancidity was highest in samples from Krajinoviće (B2), with a statistically significant difference observed (p < 0.05).

In the third (2018/19—3) production year (Figure 2c), sensory analysis revealed that the lowest mean score for external appearance (typical shape/form) was recorded for the producer from Krajinoviće (B3), while the highest score was assigned to samples from Blato (A3), with statistically significant differences between these producers (p < 0.05). The lowest score for cross-sectional color intensity was noted for samples from Blato (A3) (mean value 4.56, p < 0.05), whereas samples from Rasno (V3) received the highest rating (mean value 5.50). Notably, samples from Rasno (V3) exhibited the highest rating for color intensity across all three analyzed production periods. The highest fat tissue color score was assigned to samples from Blato (A3) (mean value 6.41), whereas the lowest rating was observed in samples from Krajinoviće (B3, mean value 5.06), with statistically significant differences from other samples.

The evaluation of intermuscular fat content showed slightly lower ratings for samples from Blato (A3, mean value 2.85), while samples from Rasno (V3) had the highest rating (mean value 4.39, p < 0.05). Sensory analysis of the third production year revealed that marbling received slightly lower scores compared to previous assessments, with statistically significant differences among all compared samples (p < 0.05). The characteristic aroma, chewiness, and salivation effect (dryness) were highly rated, exceeding a mean score of six, with statistically significant differences (p < 0.05) recorded for samples from Blato (A3) in aroma and chewiness.

Overall, the key differences across the years mainly involved cross-sectional color intensity, intermuscular fat, aroma, chewiness, and rancidity, with statistically significant variations particularly in samples from Blato (A) and Krajinoviće (B).

3.3. Microbiological Analysis of Sjenica Sheep Stelja

The microbiological analysis of Sjenica Sheep Stelja samples involved the characterization, identification, and determination of the qualitative and quantitative composition of the product’s microbiota, with a particular focus on LAB and molds.

3.3.1. Total Microbial Count

Changes in the total aerobic mesophilic bacteria count during the maturation process and across different production seasons are shown in Figure 3. At the beginning of the maturation process, samples from all three villages collected in the third production season exhibited a higher initial mesophilic bacterial count, with this trend continuing even after the stabilization of fermentation. A significant increase in aerobic mesophilic bacteria was observed in all tested samples. The bacterial count increased until day 90, after which a gradual decline was recorded.

The results of Enterobacteriaceae counts are presented in Figure 4. A significant variation in bacterial counts was observed during maturation. The bacterial population increased during the initial days of fermentation, followed by a gradual decline until day 28, after which Enterobacteriaceae were no longer detected in any of the samples.

The Pseudomonadaceae count results are provided in Figure 5. A notable variation in bacterial counts was observed across all samples. Initially, an increase occurred during the early fermentation stages, but after day 14, Pseudomonadaceae were no longer detected in any of the tested samples.

The LAB count during the maturation process of Sjenica Sheep Stelja is presented in Figure 6. A significantly higher LAB count was recorded on day 7 in samples A3, V3, and B2. In all samples, LAB counts doubled and continued to increase until day 90, after which a decline was observed across all samples.

The Enterococcus count throughout the maturation process is detailed in Figure 7. A decrease in Enterococcus spp. was observed during the initial days of maturation, with a continuous decline throughout the entire process. By day 120, the Enterococcus count had dropped below 1 log CFU/g in samples from Krajinoviće village (V1 and B3) and Rasno village (V3).

Changes in the CNS count across product samples are shown in Figure 8. A significant variation in staphylococcal counts was observed at the start of the maturation process. The highest CNS count was detected on day 0 in a sample from Blato village (A1). The CNS population increased until day 90, followed by a gradual decline.

Microbiological analysis confirmed the absence of Salmonella spp. and Listeria monocytogenes throughout all production seasons and across all examined households during the maturation process.

3.3.2. Identification of LAB Isolates

From 27 analyzed samples of Sjenica sheep stelja, a total of 451 Gram-positive, catalase-negative isolates belonging to the lactic acid bacteria (LAB) group were identified. Genus-level identification of the isolates was performed using standard physiological and biochemical tests using the API CH50 system (for Latilactobacillus, Lactiplantibacillus, and Leuconostoc species) (see Supplementary Table S1) and the API 20 STREP system (for Enterococcus species) (see Supplementary Table S2).

Homofermentative rod-shaped bacteria that tested negative for arginine hydrolysis were initially classified as lactobacilli. Based on sugar fermentation patterns obtained from the API 50CH test, (no fermentation of esculin, mannitol, or xylose), they were identified as Latilactobacillus curvatus (formerly classified as Lactobacillus curvatus) and Latilactobacillus sakei (formerly classified as Lactobacillus sakei). Homofermentative lactobacilli capable of fermenting mannitol and lactose were identified as Lactiplantibacillus plantarum (formerly classified as Lactobacillus plantarum).

All Llb. curvatus isolates showed positive fermentation of sucrose and cellobiose but did not ferment salicin or arabinose. Llb. sakei strains exhibited positive fermentation of salicin but did not ferment lactose. Heterofermentative coccoid bacteria with a positive reaction for exopolysaccharide (EPS) synthesis were initially identified as Leuconostoc spp. and based on sugar fermentation profiles obtained from the API 50CH test, they were classified as Leuconostoc mesenteroides.

Isolates that grew well on BEA, which formed black colonies and showed a positive reaction for arginine and esculin fermentation, were preliminarily identified as Enterococcus spp. Based on sugar fermentation patterns obtained from Microgen Strep ID tests, further species-level identification of enterococci determined the presence of E. faecium and E. faecalis.

The final identification of LAB isolates was confirmed using MALDI-TOF mass spectrophotometry. The mass spectra for isolated bacteria are presented in Supplementary Figure S2. Identification scores ≥ 2.000 (green color) were considered accurate at the species level.

The results of the LAB isolate distribution in ovine stelja samples and differences in microbial community composition based on producers and seasons are presented in

Table 2. Distribution of LAB species in Sjenica sheep stelja samples.

Species	Producers/Seasons
	A1	B1	V1	A2	B2	V2	A3	B3	V3	Total	(%)
Llb. curvatus	19	20	22	29	31	28	24	19	21	213	48.21
Llb. sakei	23	24	20	9	10	9	27	25	28	175	41.76
Lpb. plantarum	2	1	0	0	1	0	0	2	3	9	0.47
Lc. mesenteroides	3	1	3	0	0	0	4	5	1	17	1.67
E. faecium	4	1	5	2	3	3	2	5	5	30	7.16
E. faecalis	0	2	2	0	0	0	0	1	2	7	0.72
Total	51	46	52	40	44	40	57	58	55	443	100.00

A—Blato village; B—Krajinoviće village; V—Rasno village; 1—2016/17; 2—2017/18; 3—2018/19.

.

Among the total isolated LAB, six distinct species were identified. The majority of isolates (90.44%) belonged to the genera Latilactobacillus and Lactiplantibacillus, while 7.87% were classified as Enterococcus spp. and 1.67% under Leuconostoc spp. Lc. mesenteroides (17 isolates) was detected in the first and third research years across all producers, whereas no isolates of this species were found in the second year.

Within the genus Latilactobacillus, two species were identified: Llb. curvatus (213 isolates) and Llb. sakei (175 isolates). Additionally, within the genus Lactiplantibacillus, the species Lpb. plantarum (nine isolates) was identified. The highest number of Llb. curvatus and Llb. sakei isolates were obtained across all research years and from all producers, while Lpb. plantarum isolates were identified in producer/season combinations A1, B1, B2, B3, and V3.

Within the genus Enterococcus, two species were identified: E. faecium (30 isolates), which was present in all research periods and among all producers, and E. faecalis (7 isolates), which was found in the first and third research periods in samples from the villages of Krajinoviće and Rasno.

3.3.3. Enumeration and Identification of Molds Isolated from Sjenica Sheep Stelja

The variation in mold abundance throughout the production process across different households and production seasons is presented in Figure 9. At the onset of product maturation, no significant presence of molds was detected. However, by the 28th day, an increase in mold count was observed, which continued throughout the entire maturation process of the meat product.

A total of 221 mold isolates were obtained from the analyzed Sjenica sheep stelja samples. The highest number of isolates was recorded from producers in the villages of Blato and Krajinoviće during the second production year and from producers in the villages of Blato and Rasno during the third production year. Through the characterization and identification of the isolated molds (Supplementary Table S3), four fungal genera were identified, with Penicillium being the most dominant.

Within the genus Penicillium, nine species were identified: P. carneum (22 isolates), P. caseifulvum (21 isolates), P. corylophilum (18 isolates), P. confertum (16 isolates), P. crustosum (5 isolates), P. nalgiovense (46 isolates), P. rugulosum (17 isolates), P. polonicum (9 isolates), and P. solitum (31 isolates). Within the genus Aspergillus, two species were identified: A. niger (three isolates) and A. nidulans (three isolates). Additionally, two species were detected within the genus Eurotium: E. herbariorum (9 isolates) and E. chevalieri (15 isolates). The genus Mucor was represented by M. racemosus (three isolates) and M. plumbeus (three isolates).

The distribution of molds varied across producers and sampling periods, indicating differences in microbial community composition (

Table 3. Distribution of mold species in Sjenica sheep stelja samples.

Species	Producers/Seasons
	A1	B1	V1	A2	B2	V2	A3	B3	V3	Total	(%)
A. niger	/	1	/	/	1	/	/	1	/	3	1.35
A. nidulans	1	/	2	/	/	/	/	/	/	3	1.35
E. herbariorum	/	4	/	1	/	1	2	1	/	9	4.07
E. chevalieri	/	/	2	5	1	/	3	2	2	15	6.78
M. racemosus	1	/	/	2	/	/	/	/	/	3	1.35
M. plumbeus	/	/	/	/	1	/	/	1	1	3	1.35
P. carneum	2	/	/	9	3	3	/	5	/	22	9.95
P. caseifulvum	4	3	/	/	1	1	5	2	5	21	9.50
P. corylophilum	/	5	/	1	3	1	2	5	1	18	8.14
P. confertum	/	/	6	/	/	4	4	/	2	16	7.23
P. crustosum	/	1	3	/	/	1	/	/	/	5	2.26
P. nalgiovense	2	2	6	5	9	5	5	7	5	46	20.81
P. rugulosum	6	/	2	/	4	2	/	/	3	17	7.69
P. polonicum	2	/	1	2	2	/	2	/	/	9	4.07
P. solitum	5	3	2	2	2	5	3	3	6	31	14.02
Total	23	20	24	27	27	23	27	27	25	221	100

A—Blato village; B—Krajinoviće village; V—Rasno village; 1—2016/17; 2—2017/18; 3—2018/19; /—species not detected.

). A. niger was isolated from all samples collected in the village of Krajinoviće (B), whereas A. nidulans was detected exclusively during the first production season (2016/17) in samples from the villages of Blato and Rasno. Species belonging to the genus Eurotium were detected in all production years and across all producers, except for the producer from Blato in the first production year. M. racemosus was identified in two production years from samples collected in the village of Blato (samples A1 and A2), whereas M. plumbeus was detected in samples from producers in Krajinoviće and Rasno (samples B2, B3, and V3).

The species P. nalgiovense and P. solitum were consistently identified across all Sjenica sheep stelja samples and production years. Conversely, P. caseifulvum and P. corylophilum were absent in samples from producers in Blato and Rasno (samples V1, A1, and A2). The species P. carneum, P. confertum, P. crustosum, P. rugulosum, and P. polonicum exhibited unique distribution patterns and prevalence depending on the production season and producer. Detailed findings are presented in Table 3.

3.4. Impact of Chemical Parameters on the Development of the Autochthonous Microbiota of Sjenica Sheep Stelja

The influence of chemical parameters on the development of the microbiota in Sjenica sheep stelja was assessed through correlation analysis between microbial counts and chemical quality parameters. The results of this analysis are presented in

Table 4. Heatmap of the correlation analysis between the total number of microorganisms and the chemical parameters of Sjenica sheep stelja.

	Chemical characteristic	Moisture Content (%)	Protein Content (%)	Total Fat Content (%)	Ash Content (%)	Salt Content (%)	aw	pH
Aerobic mesophilic bacteria	0	0.22	0.631	−0.47	−0.75 *	−0.566	0.105	0.568
	7	0.282	0.422	−0.41	−0.641	−0.422	0.095	0.524
	14	0.303	0.637	−0.558	−0.763 *	−0.503	0.297	0.705 *			−1
	28	0.328	0.398	−0.428	0.114	0.408	−0.039	0.094
	60	−0.102	0.105	0.081	0.401	0.643	0.08	−0.403
	90	−0.359	−0.116	0.327	0.304	0.582	0.231	−0.392
	120	−0.357	−0.106	0.352	0.198	0.481	0.318	−0.398			−0.7
Enterobacteriaceae	0	0.52	0.329	−0.592	−0.539	−0.581	0.191	0.737 *
	7	0.729 *	0.169	−0.678 *	0.032	0.104	0.108	0.474
	14	0.684 *	−0.273	−0.409	0.33	0.341	−0.226	0.174
	28	0.151	0.018	−0.155	−0.193	−0.235	0.681 *	0.326			−0.5
Pseudomonadaceae	0	−0.136	0.654	−0.209	−0.06	0.153	−0.238	0.034
	7	0.122	−0.57	0.219	0.668 *	0.677 *	0.027	−0.519
	14	0.037	−0.791 *	0.424	0.777 *	0.808 *	−0.013	−0.697 *
LAB	0	0.257	0.344	−0.384	0.034	0.167	0.611	0.28			−0.3
	7	−0.228	−0.023	0.237	0.37	0.486	0.452	−0.394
	14	0.093	0.456	−0.273	−0.057	0.131	0.47	0.162
	28	−0.081	0.367	−0.08	0.019	0.133	0.433	−0.043
	60	0.188	0.483	−0.352	−0.165	0.166	0.27	0.196			0
	90	−0.081	0.098	0.012	0.195	0.509	0.257	−0.137
	120	0.319	0.524	−0.525	−0.329	0.031	0.328	0.482
Enterococcus spp.	0	0.134	0.738 *	−0.48	−0.874 *	−0.852 *	0.086	0.723 *			0.3
	7	0.376	0.278	−0.419	−0.459	−0.452	−0.23	0.534
	14	0.341	−0.222	−0.16	−0.062	−0.21	0.223	0.278
	28	0.089	−0.012	−0.124	0.045	−0.147	0.202	0.269
	60	−0.005	−0.07	−0.051	−0.158	−0.353	0.334	0.365			0.5
	90	−0.127	−0.455	0.253	0.126	−0.18	0.427	−0.003
	120	−0.157	−0.347	0.235	0.309	−0.004	0.106	−0.145
Staphylococcus sp.	0	−0.257	0.29	−0.031	−0.482	−0.467	0.307	0.41
	7	−0.084	0.425	−0.212	−0.596	−0.416	0.003	0.48			0.7
	14	−0.128	0.537	−0.146	−0.62	−0.606	−0.3	0.241
	28	0.182	0.277	−0.214	0.372	0.427	−0.508	−0.237
	60	−0.468	0.372	0.142	−0.251	−0.092	0.456	0.098
	90	−0.364	0.221	0.19	−0.22	−0.008	0.551	−0.067			1
	120	−0.596	−0.38	0.692 *	0.127	0.198	0.391	−0.555
Molds	28	−0.573	−0.838 *	0.863 *	0.47	0.362	0.079	−0.777 *
	60	−0.936 *	−0.311	0.871 *	0.194	0.038	0.107	−0.638
	90	−0.108	−0.696 *	0.459	0.365	0.071	−0.15	−0.458
	120	−0.479	−0.346	0.564	0.39	0.179	−0.071	−0.602

* Statistically significant correlations, p < 0.05.

.

The correlation analysis between the examined chemical parameters and the abundance of aerobic mesophilic bacteria during the aging of Sjenica sheep stelja revealed a negative correlation with ash content on days 0 and 14 of the production process (r = −0.750, p < 0.05; r = −0.763, p < 0.05). Additionally, a positive correlation was observed between aerobic mesophilic bacterial count and pH on day 14 (r = 0.705, p < 0.05).

The analysis of the relationship between Enterobacteriaceae counts and chemical parameters demonstrated a positive correlation between bacterial abundance and pH on day 0 (r = 0.737, p < 0.05). A significant positive correlation was also found between bacterial counts and water content on days 7 and 14 (r = 0.729, p < 0.05; r = 0.684, p < 0.05), as well as between bacterial counts and a_w on day 28. Conversely, a negative correlation was identified between fat content and Enterobacteriaceae abundance on day 7 (r = −0.678, p < 0.05).

The abundance of Pseudomonadaceae exhibited statistically significant correlations with ash and salt content on days 7 and 14 (r = 0.668, p < 0.05; r = 0.677, p < 0.05; r = 0.777, p < 0.05; r = 0.808, p < 0.05). A statistically significant negative correlation was observed between Pseudomonadaceae counts and protein content as well as pH on day 14 (r = −0.791, p < 0.05; r = −0.697, p < 0.05).

Throughout the 120-day production process, no statistically significant correlation was identified between the abundance of lactic acid bacteria (LAB) and chemical quality parameters.

The correlation analysis of Enterococcus abundance and chemical quality parameters revealed a significant negative correlation with ash and salt content on day 0 (r = −0.874, p < 0.05; r = −0.852, p < 0.05). Additionally, a positive correlation was observed between Enterococcus abundance and protein content as well as pH on day 0 (r = 0.738, p < 0.05; r = 0.723, p < 0.05).

During the aging process, Staphylococcus species exhibited a significant positive correlation with fat content on day 120 (r = 0.692, p < 0.05), while no statistically significant correlations were observed between Staphylococcus abundance and any chemical parameters.

A significant influence of chemical parameters on mold abundance was demonstrated through a positive correlation between fat content and mold counts on days 28 and 60 (r = 0.863, p < 0.05; r = 0.871, p < 0.05). Additionally, a significant negative correlation was observed between mold abundance and protein content (days 28 and 90), water content (day 60), and pH (day 28) (r = −0.863, p < 0.05; r = −0.696, p < 0.05; r = −0.936, p < 0.05; r = −0.777, p < 0.05).

4. Discussion

Sjenica sheep stelja is a traditional dry-cured meat product protected by a geographical indication and produced using a specific processing technology in the Pešter region. It is distinguished by high-quality raw materials and traditional production methods. The chemical properties of Sjenica sheep stelja were analyzed over a three-year research period, revealing statistically significant differences in certain quality parameters among the tested samples.

The pH values remained highly consistent across all producers (Table 1), which was expected given that the production process and environmental conditions were standardized. Similarly, water activity measurements did not indicate statistically significant differences among the samples (p > 0.05), further indicating uniformity in basic physicochemical parameters across production sites. The moisture content remained stable in samples from Blato (A) and Krajinoviće (B) during the first and second production seasons, as well as in Krajinoviće (B) and Rasno (V) during the third season. The moisture content ranged from 42.5% to 47.2%, which is comparable to that of sheep stelja from Bosnia and Herzegovina [45]. In contrast, Krvavica et al. [46] reported a lower moisture content (31.23%) in Dalmatian kaštradina, a similar dry-cured meat product. The water content of Sjenica sheep stelja is largely influenced by the salting and drying processes and is comparable to stelja produced in Italy and Croatia, despite differences in the raw materials used [47]. Salt content is also a critical factor influencing lipolysis, proteolysis, and oxidation in dry-cured meat. In stelja production, dry salting with coarse sea salt (3.5–5%) is the only permitted method [17]. In the final product, the recorded salt content exceeded 5%, with no statistically significant differences observed among samples A1, A3, B3, V1, and V3 (p > 0.05).

The fat content in stelja samples from all three villages during the third production season (A3, B3, V3) ranged from 7.53% to 7.9%, with no statistically significant differences (p > 0.05). However, significant statistical differences were observed in other samples. Notably, Krvavica et al. [46] reported a considerably higher fat content (39.21%) in kaštradina compared to Sjenica sheep stelja. Protein content also varied significantly, with the sample from Krajinoviće (B3) exhibiting higher values compared to other groups, whereas samples A1, B1, and V3 showed no significant differences (p > 0.05). The protein levels reported in this study are higher than those documented by Gajić [48] for Bosnian stelja (29.04% at 5.34% NaCl), supporting previous findings that intensive salting and prolonged maturation enhance the protein concentration in dry-cured meat products [49]. Ash content ranged from 9.1% to 10.5%, with the exception of the Krajinoviće sample from the second production season, which recorded a slightly higher value of 10.9% (p > 0.05). These results contrast with those by Ganić et al. [45], who reported ash content between 1.76% and 2.31% in Bosnian stelja, suggesting that regional differences in production techniques and salt usage may significantly impact the mineral composition of the final product.

The two-way ANOVA revealed that the chemical composition of bedding samples was influenced by both household and seasonal factors, although the magnitude and significance of these effects varied across parameters. While pH and water activity remained stable across households and seasons, significant interaction effects were observed for moisture, fat, protein, salt, and ash contents. Notably, total fat content exhibited the largest effect sizes and the most pronounced differences among households and seasons. Protein and moisture contents were also significantly influenced by both factors, with clear seasonal and household-specific trends. In contrast, salt and ash contents demonstrated significant interactions, though post hoc tests revealed limited group-level differences.

The sensory quality of fermented meat products is largely determined by the adequacy of key attributes such as flavor (encompassing aroma and taste), appearance, and texture, which are typically assessed using a structured set of descriptive parameters [47]. According to Stamenković and Dević [18], Sjenica sheep stelja is defined by its characteristic brown coloration, soft and elastic texture, slight marbling with yellowish fat, mildly acidic taste, smoky aroma, and a moderate to pronounced level of saltiness and dryness. Stojković et al. [16] further reported that sensory variability in sheep stelja from the Western Balkans can be attributed primarily to differences in production techniques and the use of spices. Their study found that 25% of Serbian stelja samples exhibited an intense smoky flavor, with salt content in some cases reaching 6.2–6.7% (v/v), highlighting the potential for variability in traditionally processed products. In the present study, sensory evaluation confirmed that all analyzed samples of Sjenica sheep stelja met the sensory characteristics outlined in the product specification [17], with no visible defects observed. The samples received high average scores across all assessed attributes, including external appearance, fat tissue color, marbling, cross-sectional cohesion, aroma, chewiness, salivation, saltiness, smokiness, and rancidity. No statistically significant differences were found in overall sensory scores among the samples (p > 0.05), indicating consistent sensory quality across production sites and seasons. However, a significant difference was observed in the sample from Blato during the first production season (A1), specifically in cross-sectional color intensity, intramuscular fat content, and taste. Despite this variation, the product remained within the defined quality parameters, reaffirming the robustness of traditional processing methods in maintaining the sensory integrity of Sjenica sheep stelja.

The fermentation and maturation processes of Sjenica sheep dried meat play a crucial role in shaping microbial diversity. The microbial population is primarily derived from naturally occurring microorganisms present in the raw material or introduced through environmental contamination during processing [3,50]. Microbiological analysis conducted over three production seasons revealed fluctuations in aerobic mesophilic bacteria. Initial counts ranged from 5.62 log CFU/g (Blato, season A1) to a peak of 8.66 log CFU/g (Krajinoviće, season B3) by day 90. These counts were comparable to those reported in similar traditional products such as Khyopeh (6–7 log CFU/g) and Kitoza (7.0 log CFU/g) [51,52]. Aerobic mesophilic bacteria exhibited a positive correlation with pH and a negative correlation with ash content, suggesting that the chemical composition significantly influences microbial proliferation. Enterobacteriaceae and Pseudomonadaceae populations declined as fermentation progressed, typically decreasing below 2 log CFU/g by day 28. Enterobacteriaceae initially reached 3.39 log CFU/g (A2), with population dynamics showing a negative correlation with fat content on day 7 and positive correlations with pH, moisture, and water activity. Pseudomonadaceae diminished after day 14, with abundance negatively correlated with protein content and pH and positively correlated with ash and salt contents. These trends are consistent with observations in pastırma, where similar microbial successions occur during drying and maturation [53]. LAB emerged as the dominant microbial group throughout fermentation. LAB counts increased from 2.97 log CFU/g (B2) to a peak of 8.45 log CFU/g (B3, day 90), followed by a slight decline by day 120, consistent with their role in fermentation and acidification. Enterococcus spp. declined from 3.55 log CFU/g (A2, day 0) to approximately 2.0 log CFU/g by the end of maturation, with their presence positively correlated with protein content and pH and negatively correlated with ash and salt levels. The populations of Staphylococcus spp. increased during early fermentation, peaking at 5.03 log CFU/g (B2, B3, day 28), and showed a positive correlation with fat content. These microbial patterns closely resemble those observed in pastırma and Kitoza [52,54]. Importantly, no pathogenic bacteria, including Salmonella spp. and L. monocytogenes, were detected in any sample, confirming the microbiological safety of Sjenica sheep stelja. Mold growth was observed during late maturation, reaching a maximum of 5.36 log CFU/g (B2), with positive correlations observed with fat content and negative correlations with protein content, moisture, and pH. Molds, while often associated with surface spoilage, contribute positively to flavor development, lipid oxidation, and product stabilization in traditionally fermented meat products [15].

LAB identified in this study were consistent with those commonly associated with fermented meat products, including Llb. sakei, Llb. curvatus, and Lpb. plantarum [55]. Among these, Lpb. plantarum and Llb. curvatus are the most prevalent in fermented sausages [56]. Llb. sakei, on the other hand, is the dominant species in fermented meats from France, Italy, and Spain, often outcompeting other LAB due to its superior adaptability, including a higher maximum growth rate, greater final cell density, and a shorter lag phase [57,58]. Llb. curvatus also contributes significantly to the fermentation process, enhancing the microbial stability and sensory quality of the final product [59]. Interestingly, Lc. mesenteroides, typically considered a specific spoilage organism due to its role in acidification and blown pack spoilage in meat and poultry products [60], was identified in 17 isolates. However, as these isolates were distributed across different production seasons and households (Table 2), their presence is unlikely to have significantly impacted the sensory or microbiological quality of the final product. Enterococci species, primarily E. faecalis and E. faecium, were also isolated from stelja. This is in agreement with findings from traditional Portuguese dry-smoked fermented sausages, where these species dominated the enterococcal population [61].

Fungal isolates from Sjenica sheep stelja predominantly belonged to the Penicillium genus, including P. carneum, P. caseifulvum, P. corylophilum, P. confertum, P. crustosum, P. nalgiovense, P. rugulosum, P. polonicum, and P. solitum. These molds are widely recognized as characteristic microbiota in traditional dry-cured meat products [15,62]. In Croatian household-produced dry-fermented sausages, Penicillium species accounted for 71% of isolates, while Mucor and Aspergillus contributed 18% and 11%, respectively [62]. Additional fungal species identified included A. niger, A. nidulans, E. herbariorum, E. chevalieri, M. racemosus, and M. plumbeus, which have also been reported in San Daniele ham and other traditionally fermented meats [63,64].

5. Conclusions

This study affirms the high-quality standards of Sjenica sheep stelja, a traditional dry-cured meat product protected by a geographical indication. While pH and water activity levels remained consistent across different producers, variations were observed in fat, protein, and ash contents. These variations in chemical composition were influenced by the salting, drying, and maturation processes, which are integral to the product’s nutritional value. Overall, the findings highlight the role of traditional production methods and regional raw materials in shaping the quality of Sjenica sheep stelja, offering valuable insights into future research and quality control efforts. The sensory evaluation results from the three production years confirm that Sjenica sheep stelja consistently met the expected quality attributes outlined in the Elaborate for the Protection of Geographical Indication. Despite minor variations in characteristics such as color intensity, marbling, intermuscular fat, and taste, the product maintained high scores for aroma, texture, and overall acceptability. These findings underscore the impact of production season and household-specific processing techniques on the sensory attributes of the product.

The study further demonstrates that chemical parameters significantly influence the development of autochthonous microbiota in Sjenica sheep stelja throughout the aging process. The findings reveal the complexity of microbial succession in this traditionally produced meat product and highlight the pivotal role of chemical composition in shaping the structure and dynamics of the microbial community. A deeper understanding of these interactions offers valuable insights for optimizing processing conditions, with the potential to improve both product quality and microbiological safety.