The Jinhua hams used in the present study were prepared and sampled in Jinhua Jinnian Ham Co., Ltd (Zhejiang, China). Six fresh hind legs (14.5 ± 0.5 kg, pH = 5.9 ± 0.2) of domestic pigs (Large White × Landrace) were used to prepare Jinhua FH and LH following the procedures of Zhou et al. (2021a) with some modifications. For Jinhua FH, three whole hind legs were salted for 75 days with 0.014% NaNO₂ and 10% NaCl per leg, followed by soak cleaning for 1 day and sun-drying for 1 day. Subsequently, hind legs were dehydrated for 7 days in a dehydration plant, sun-dried for another day, and then ripened for 180 days in a dry-ripening room. During the ripening, the ambient temperature progressively increased from 5°C to 35°C, while the relative humidity gradually decreased from 85 to 65%. Afterward, legs were further post-ripened for about 30 days at room temperature (25°C), and the hams were finally obtained when the weight loss reached approximately 40% of the initial weight. For Jinhua LH, the skins and fat tissues were firstly trimmed off from the other three hind legs. The final hams were also obtained following the gradual procedures of salting, soak cleaning, sun-drying, dehydrating, secondary sun-drying, dry-ripening, and post-ripening, which were all performed under the same conditions with Jinhua FH. Furthermore, as described by Chen et al. (2021) with minor modifications, approximately 5 cm × 2 cm × 0.2 cm pieces were cut from the surface of the biceps femoris muscle of each ham, which were used for the fungal community determination. Meanwhile, about 20 g interior samples were taken from the central fraction (about 3–4 cm depth) of the biceps femoris muscle of each ham, which were used for the evaluation of physicochemical parameters, volatile flavor compounds, and bacterial community. All samples were vacuum-packaged and frozen at-80°C until further analysis.

The color, pH and water a_w were measured following the methods of our previous studies (Li H. et al., 2021; Zhang et al., 2022) with minor modifications. The color of samples was directly detected using a colorimeter (NH310, 3NH Technology Co., Ltd., China), where the L, a*, and b* values represent lightness, redness/greenness, and yellowness/blueness, respectively. A white standard plate with the L, a*, and b* values of 99.46, 0.19, and −1.98, respectively, was used for calibration before color detection. The pH value was determined using a portable pH meter (Testo 205, Testo Instruments Co., Ltd., Shenzhen, China) equipped with a piercing pH probe and a temperature-compensated temperature probe. The a_w value was determined with an intelligent water activity meter (HD-4, Huake Instrument and Meter Co., Ltd., Wuxi, China).

The chemical composition of ham samples was detected as described by our previous study (Zhang et al., 2017) with minor modifications. Briefly, the total contents of moisture, protein, fat, and minerals were analyzed by the method of direct drying (GB/T 5009.3–2016), Kjeldahl (ISO 5983-1997), Soxhlet extraction (ISO 1444-1996), and dry-ashing (ISO 5984-2002), respectively.

According to the method of Liu et al. (2019) with minor modifications, the content of residual nitrite was detected with a nitrite assay kit following the manufacturer’s instruction (Nanjing Jiancheng Bioengineering Institute, Nanjing, China). Briefly, ham samples were dispersed into distilled water and filtrated, followed by the reaction with sulphanilamide solution and N-naphthyl-ethylenediamine dihydrochloride. The nitrite content was calculated based on the absorbance of supernatant at 550 nm, which was assayed using an ultraviolet–visible spectrophotometer (UV-1750, Shimadzu, Kyoto, Japan).

As described by Liu et al. (2020) with minor modifications, the volatile flavor compounds of ham samples were identified by a headspace-gas chromatography-ion mobility spectrometer (HS-GC-IMS; Flavorspec, G.A.S. Instrument, Germany) equipped with a SE-54 capillary column (15 m × 0.53 mm × 1 μM). Samples were firstly cut into approximately 1 cm × 1 cm × 1 cm cubes and minced by an analytical grinder (Aika Instrument Equipment Co., Ltd., Guangzhou, China). Then 2 g samples were put into a 20 ml headspace sampling vial and incubated at 60°C for 20 min. Afterward, 500 μL headspace was injected into the injector automatically with a heated syringe at 85°C. Subsequently, samples were transferred into the capillary column by high-purity nitrogen (>99.99%) at the following programmed flow rates: initially 2 ml/min for first 2 min, then 10 ml/min for 8 min, next 100 ml/min for 10 min, and eventually 150 ml/min for 5 min. Meanwhile, the temperature of column and drift tube was kept as 60°C and 45°C, respectively, and the flow rate of drift gas (nitrogen gas, >99.99% purity) was maintained as 150 mL/min.

The IMS data were analyzed using the instrumental laboratory analysis view (LAV) software with the Reporter, Gallery Plot, and Dynamic PCA plug-in applications. The volatile flavor compound was identified by comparing the retention index (RI) and drift time (DT) with the NIST library and IMS database retrieval software obtained from G.A.S. The intensities of these compounds were calculated based on the height of selected signal peaks.

The analysis of microbial communities, including the compositions of both fungi and bacteria, were performed by high-throughput sequencing following the procedures of Chen et al. (2021) and Wang et al. (2021) with some modifications. The total genome DNA of bacteria and fungi was extracted using a Cetyltrimethylammonium Bromide (CTAB) method following the manufacturer’s instructions of the genomic DNA extraction kit. The V3-V4 hypervariable regions of bacterial 16S rRNA genes were amplified with the forward primer 5’-CCTAYGGGRBGCASCAG-3′ and reverse primer 5′-GGACTACNNGGGTATCTAAT-3′. The ITS1-1\u00B0F regions of fungal ITS genes were amplified with the forward primer 5′-CTTGGTCATTTAGAGGAAGTAA-3′ and reverse primer 5′-GCTGCGTTCTTCATCGATGC-3′. All PCR reactions were conducted in 30 μL reactions with 15 μL Phusion® High-Fidelity PCR Master Mix (New England Biolabs), 0.2 μM forward and reverse primers, and approximately 10 ng template DNA. PCR products were then mixed in equidensity ratios and purified through the Axy Prep DNA Gel Extraction Kit (AXYGEN; for bacteria) or Qiagen Gel Extraction Kit (Qiagen, Germany; for fungi).

The data sequencing was carried out according to the procedures of Wang et al. (2021) and Li et al. (2022) with some modifications. Sequencing libraries were generated using the NEB Next® Ultra™ DNA Library Prep Kit for Illumina (NEB, United States; for bacteria) or TruSeq® DNA PCR-Free Sample Preparation Kit (Illumina, United States; for fungi) following the manufacturer’s recommendations and index codes were added. The library quality was evaluated on the Qubit@ 2.0 Fluorometer (Thermo Fisher Scientific) and Agilent Bioanalyzer 2,100 system. Then the library was sequenced on the Illumina Miseq/HiSeq2500 (for bacteria) or NovaSeq (for fungi) platform. Afterward, the QIIME (V1.9.1) quality controlled process was performed to filter raw reads to obtain high-quality clean reads. The reads were compared with the reference database (Silva database) using the UCHIME algorithm to detect chimera sequences, and the chimera sequences were removed to acquire the effective clean reads. The sequence analysis was performed using the UPARSE algorithms. Sequences with ≥97% similarity were assigned to the same operational taxonomic units (OTUs), and representative sequences for each OTU were screened for further annotation. The significance of differences in microbial communities was statistically analyzed based on the relative abundance of bacteria or fungi at various levels, mainly including phylum, genus, and species. The α-diversity was presented via the Shannon index, Simpson index, ACE index, Chao 1 index, and observed species, while the β-diversity was assayed by the principal coordinate analysis (PCoA).

All experiments were performed in triplicate with the results shown as average ± standard deviation. Tables were made by the Microsoft Excel 2016 software, while figures were drawn with the Origin V2021 (Origin-Lab, Northampton, United States) and Microsoft PowerPoint 2016. Analysis of variance (ANOVA) was performed using the SAS V8 software (SAS Institute Inc., Carry, United States), while analysis of Pearson’s correlation and hierarchical cluster (HCA) was conducted by the plug-in applications in Origin V2021. Differences among mean values were established with the Duncan multiple range test. The significant difference was confirmed when p < 0.05.

The physicochemical property of dry-cure ham was closely associated with the flavor quality and microbial diversity (Domínguez et al., 2022). The physicochemical parameters of internal samples from Jinhua FH and LH, including color, water activity (a_w), pH, chemical composition, and nitrite residue, are shown in Table 1. It is clear that there is no significant distinction between the a* values (redness) of FH and LH (p > 0.05). However, FH showed slightly higher L and b* values than LH (p < 0.05), suggesting that the color of FH was slightly lighter and yellower than that of FH. This result may be attributed to their difference in the redox degree of proteins, especially the myoglobin and hemoglobin (Parolari et al., 2016). Specifically, myoglobin and hemoglobin are the main pigments responsible for the color of dry-cured hams (Parolari et al., 2016; Zhou et al., 2021b), and the lack of protection by fat and skin tissues might allow more myoglobins/hemoglobins in LH to be oxidized during dry-ripening, possibly resulting in the darker color of LH.

Besides, the a_w values of FH and LH were not obviously different (p > 0.05) and both far under 0.85, which was a low a_w that protects hams from most harmful microbes and meanwhile allows the growth of salt-consuming and flavor-generating bacteria/fungi (Chen et al., 2021). However, LH exhibited a remarkable lower pH than FH (p < 0.05). Bosse et al. (2018) reported that a pH of 6.0–6.2 would cause a higher microbial risk and reduced water diffusion ability in hams, while hams with pH of 5.6–6.0 have more desirable saltiness, color, and texture. The pH values of FH and LH were 6.08 ± 0.03 and 5.75 ± 0.04, respectively, suggesting that LH might has higher sensory quality and edible safety than FH. The distinct pH values of FH and LH was probably attributed to their distinction in the intensity of acids generated during ripening (Table 2; Figure 1).

Furthermore, these two hams also exhibited no significant difference in chemical compositions (p > 0.05). The moisture content of both FH and LH (around 31.5%) was similar with that of Mianning hams reported by Chen et al. (2021), which is low enough to prevent spoilage and improve quality of hams. This result was also consistent with the a_w data as mentioned above, since the water content is usually considered positively correlated with the water activity. Meanwhile, the minerals in hams can be mainly salts (Domínguez et al., 2022), hence FH and LH showed similar mineral contents (about 8.50%; p > 0.05) because of the same adding amount of NaCl during salting. Additionally, the nitrite residues were 1.13 ± 0.21 and 1.57 ± 0.12 for FH and LH, respectively, without marked distinction (p > 0.05) and far under the secure nitrite residue limits (30 mg/kg) for meat products in China (Chen et al., 2021). On the other hand, the pigments contributing to the rose-red color of dry-cured hams are mainly nitromyoglobins and nitrohemoglobins, which are the product of reactions between nitrites/nitrates and myoglobins/hemoglobins (Parolari et al., 2016). Therefore, the similar nitrite residue of FH and LH suggested a same nitrite consuming amount for the production of red pigments, which can be the main reason for their similar redness (a* value) as mentioned above. Overall, the data shown in Table 1 indicate that LH had an obviously lower lightness, yellowness, and pH than FH, probably resulting from their discrimination in oxidation and fermentation products.

The volatile flavor profiles of internal samples from Jinhua FH and LH were analyzed by HS-GC-IMS and results are illustrated as a two-dimension spectra plot (Figure 2). The y-axis and x-axis in Figure 2 showed the retention time (RT) of GC and the relative drift time (DT) of ions, respectively (Liu et al., 2020). Meanwhile, the red line representing x = 1.0 was the reactive ion peak (RIP). Besides, each data point corresponded to a volatile flavor compound, and its color indicated the intensity of volatile (Li et al., 2019). Specifically, the blue and white colors represented a low intensity of volatiles, whereas the yellow and red colors showed a high intensity of compounds. It is exhibited that the two hams showed similar number of ion peaks within the DT range of 1.0–1.7 ms, suggesting that the numbers of main identified volatiles were not obviously different in FH and LH. However, the data points (ion peaks) of FH mostly showed a redder color or bigger size than those of LH, indicating that FH had an overall higher intensity of volatile flavor than LH.

Furthermore, to compare the finger-print of characteristic volatiles in FH and LH, the Gallery plot plug-in was utilized with results presented in Figure 3. Each row in Figure 3 represented a single volatile flavor compound, which could correspond to a single signal (monomer) or a pot (dimer) in the IMS spectra plot (Figure 2) depending on its concentration (Arroyo-Manzanares et al., 2018). As shown in Figure 3, a total of 33 volatile compounds were identified, which can be classified into 7 chemical families. In details, there were 7 alcohols, 8 ketones, 10 aldehydes, 4 esters, 2 acids, 1 heterocyclic compound, and 1 sulfur-containing compound. These findings were in accordance with the reports of Liu et al. (2020), Wang et al. (2021), and Li et al. (2022) on the volatile profiles of Jinhua hams.

The profiles of individual volatile flavor compounds were further analyzed with data shown in Table 2. Aldehydes are known as the major contributors to the unique flavor of fermented meats due to their high concentrations and low aroma thresholds (Liu et al., 2020; Wang et al., 2021). As exhibited in Table 2, hexanal, 2-methylbutanal, and 3-methylbutanal were the richest aldehydes in both FH and LH, followed by 2-methylpropanal (for FH) or phenylacetaldehyde (for LH), in accordance with the findings of Wang et al. (2021) on the volatiles of Jinhua hams. Hexanal is regarded as the major lipid oxidation product in dry-cured hams, and a low-content hexanal generally provides a pleasant grassy, fruity, and green odor (Benet et al., 2015; Domínguez et al., 2022). 2-methylbutanal, 3-methylbutanal, and 2-methylpropanal are all branched aldehydes, deriving from oxidative deamination and decarboxylation of valine, leucine, and isoleucine through Strecker degradation (Narváez-Rivas et al., 2014). Meanwhile, they are also medium-chain (C4-C9) aliphatic aldehydes, which were considered responsible for grassy, fatty, and/or nutty flavors of meat products (Liu et al., 2020; Wang et al., 2021). Furthermore, the abundances of most identified aldehydes were significantly lower (p < 0.05) or similar (p > 0.05) in LH compared with those in FH. Noteworthily, only the phenylacetaldehyde showed a relatively higher richness in LH (p < 0.05). Phenylacetaldehyde is originated from the Strecker degradation of phenylalanine and might contribute to the spicy sensation of hams (García-González et al., 2008).

Ketones are another important flavor component in dry-cured meat products, and the high-intensity ketones are usually associated with the creamy, fruity, cooked, and spicy flavor characteristics (Zhu et al., 2018). The most abundant ketones in the two hams were 2-butanone and acetone (p < 0.05; Table 2). Acetone could be mainly transferred from acetyl-CoA, one of the major degradation products of glycogens in the muscle, and imparts a buttery taste and fruity aroma to meat products (Shi et al., 2019; Li W. et al., 2021). 2-butanone and other methyl ketones are produced via the decarboxylation of β-keto acids or the β-oxidation of saturated fatty acids, and act as precursors in the formation of fatty flavor during the ripening of meats (Pham et al., 2008; Shi et al., 2019). Besides, FH had significantly higher intensities in almost all identified ketones than LH (p < 0.05) excluding 2-pentanone. The 2-pentanone did not show noteworthy distinction between the two hams (p > 0.05; Table 2).

Alcohols are considered contributing less to the aroma of hams because of their relatively higher odor thresholds than aldehydes and ketones, but they can also impart to an herbal, woody, and oily flavor at high concentrations (Liu et al., 2020). The linear-chain aliphatic alcohols are the oxidation products of lipids, whereas the branched alcohols are mostly reported as the microbial degradation products of corresponding branched aldehydes (Shi et al., 2019). It is clear in Table 2 that the most abundant alcohol was obviously distinct for the two hams, which was 2-propanol in LH but ethanol in FH. Meanwhile, most identified linear-chain alcohols showed a higher richness in FH (p < 0.05), while 2-propanol, 3-methyl-1-butanol (monomer), and 3-hexen-1-ol presented higher abundances in LH (p < 0.05).

Two short-chain fatty acids (SCFAs; C1-C6), isovaleric acid and 2-methylpropanoic acid, were also identified in both FH and LH, which probably originated from the deamination of amino acids or the secondary metabolism by microbes (Liu et al., 2020; Li et al., 2022). Interestingly, both the two identified acids were more abundant in LH (p < 0.05; Table 2), which can be a key reason for its relatively lower pH (p < 0.05; Table 1). These acids could contribute to a sour taste and cheese flavor, and neutralize some deleterious alkaline compounds in fermented meats such as amine and pyrazine (García-González et al., 2008; Liu et al., 2020). Esters were found with relatively high aroma thresholds, generated via the esterification between carboxylic acids and alcohols, and partially provided the sweet, fruit, and/or fatty flavors in meat products (Carrapiso et al., 2015). The predominant esters were ethyl acetate and γ-butyrolactone in both FH and LH. The intensity of γ-butyrolactone was similar in FH and LH (p > 0.05), but the richness of ethyl acetate was relatively higher in FH than that in LH (p < 0.05).

Moreover, only one heterocyclic compound, 2-pentylfuran, was found in the two hams, which was also determined as the most abundant furans in Chinese bone-less hams by Li et al. (2022). 2-pentylfuran is regarded as an odor-active compound with a green and fruity flavor in dry-cured hams (García-González et al., 2008). In addition, dimethyl disulfide was the only identified sulfur-containing volatile in the two hams, which was in line with the report of Liu et al. (2020) on Jinhua hams. This chemical family is generally products of sulfur-containing amino acid catabolism or microbial metabolism, and usually provides an unpleasant flavor with low odor threshold (Pérez-Santaescolástica et al., 2018). However, dimethyl disulfide has a vegetable aroma and important contribution to the characteristic flavors of cured meats (Liu et al., 2020). Besides, both 2-pentylfuran and dimethyl disulfide showed higher concentrations in FH than those in LH (p < 0.05).

The total abundance of volatiles belonged to each chemical family was further summarized with results illustrated in Figure 1. It is shown that FH had significantly higher total intensity of ion peaks for volatiles than LH (p < 0.05). On one hand, aldehydes were the most abundant volatiles among all chemical families for both FH and LH, followed by ketones and alcohols, which were consistent with the findings of Liu et al. (2020) and Li W. et al. (2021) on volatiles from various Chinese dry-cured hams. On the other hand, LH showed remarkably higher total intensities of acids and alcohols (p < 0.05), whereas other types of volatiles were obviously less abundant for LH (p < 0.05).

The high-throughput sequencing of 16S rRNA genes was performed to investigate the bacterial community structures of Jinhua FH and LH. The α-diversity of bacteria from FH and LH, including Shannon, Simpson, ACE, and Chao 1 indexes, are shown in Table 3. Shannon and Simpson indexes represent the community diversity, and the ACE and Chao 1 indexes are associated with the community richness (Mu et al., 2020). It is clear that the two hams did not have significantly different α-diversities in terms of all indexes (p > 0.05). Besides, the numbers of observed bacterial species in FH and LH were 1,058 ± 307 and 1,262 ± 71, respectively, without marked difference (p > 0.05). These data indicate that FH and LH had an overall similar diversity and richness in bacterial communities.

Figure 4 compares the bacterial community structure of two hams at various levels, including phylum, genus, and species. As illustrated in Figure 4A, Firmicutes was the most abundant bacteria at the phylum level in FH with a relative abundance of 87.23% (p < 0.05), followed by Bacteroidetes (6.06%) and Proteobacteria (3.46%; p < 0.05). This result was in consistence with the reports of Wang et al. (2021) and Li et al. (2022) on the bacterial composition of Jinhua hams. However, the dominant bacterial phyla (>5%) in LH showed a greater variety, which included Proteobacteria (41.39%), Firmicutes (20.33%), Cyanobacteria (16.26%), Actinobacteria (8.92%), and Bacteroidetes (5.57%; p < 0.05). Compared with the bacteria in FH, the richness of Firmicutes was sharply decreased in LH (p < 0.05), but Proteobacteria, Cyanobacteria, Actinobacteria, Chloroflexi, and ε-acteraeota exhibited a significant elevation in relative abundance (p < 0.05).

At the genus level (Figure 4B), Ruminococcaceae UCG-005 (22.39%), Staphylococcus (11.91%), and Ruminococcaceae UCG-014 (4.27%) were the dominant bacteria (>2%) in FH (p < 0.05), while Staphylococcus (22.88%) and Psychrobacter (11.58%) were the richest bacterial genera in LH (p < 0.05), followed by Halomonas (2.50%), Propionicicella (2.07%), and Ruminococcaceae UCG-005 (2.02%; p < 0.05). Wang et al. (2021) reported that Staphylococcus, Psychrobacter, Halomonas, and unidentified Ruminococcaceae were among dominant bacterial genera of Jinhua hams ripened for different months, which were corresponded to our above findings. Furthermore, among the main bacterial genera with a relative abundance >1%, FH had richer Ruminococcaceae UCG-005, Ruminococcaceae UCG-014, Bacteroides, Ruminococcaceae NK4A214 group, and [Ruminococcus] torques group (p < 0.05), whereas LH showed higher richness of Staphylococcus, Psychrobacter, Propionicicella, Ochrobactrum, Sphingomonas, Lactobacillus, and Halomonas (p < 0.05).

At the species level (Figure 4C), the dominant bacteria (>1%) were Staphylococcus sp., Ruminococcaceae UCG-005 sp., and Ruminococcaceae NK4A214 group sp. for FH (p < 0.05), whereas Staphylococcus sp., Ochrobactrum sp., Psychrobacter pulmonis, and Halomonas subglaciescola for LH (p < 0.05). On the other hand, among the core bacterial species with a richness >0.5%, the richness of Ruminococcaceae UCG-005 sp. and Ruminococcaceae NK4A214 group sp. were more abundant in FH (p < 0.05), while Staphylococcus sp., Ochrobactrum sp., Psychrobacter pulmonis, Prevotella paludivivens, and Halomonas subglaciescola presented a relatively higher richness in LH (p < 0.05).

The high-throughput sequencing of ITS genes was applied to analyze the fungal community structures of Jinhua FH and LH. The α-diversity of fungi from the two hams are exhibited in Table 4. Generally, FH presented much higher Shannon, ACE, and Chao 1 indexes than LH (p < 0.05). Moreover, the number of fungal species observed from FH was 697 ± 147, approximately 2.8 times of that observed from LH (246 ± 5; p < 0.05). These results reveal that both diversity and richness of fungal community in LH was dramatically reduced in comparison to FH.

Figure 5 illustrates the fungal community structure of FH and LH at the phylum, genus, and species level. As illustrated in Figure 5A, Ascomycota was the richest fungus phylum in FH and accounted for a relative abundance of 82.79%, followed by Basidiomycota (6.06%) and Mortierellomycota (2.66%; p < 0.05). Meanwhile, Ascomycota was the only dominant fungal phylum (>5%) in LH (p < 0.05), accounting for a much higher relative abundance (97.69%) than that in FH (p < 0.05). In accordance with this finding, Lin et al. (2020) and Chen et al. (2021) also found that Ascomycota was the absolutely predominant fungus at the phylum level in Laowo dry-cured hams and Mianning hams, respectively. Besides, the relative abundances of many fungal phyla, mainly including Basidiomycota, Rozellomycota, Chytridiomycota, and Glomeromycota were significantly lower in LH than those in FH (p < 0.05).

At the genus level (Figure 5B), Meyerozyma (55.61%) was the most abundant fungus in FH (p < 0.05), followed by Aspergillus (11.67%; p < 0.05). While Meyerozyma (64.79%), Yamadazyma (17.76%), and Aspergillus (12.82%) were the dominant fungal genera (>2%) in LH (p < 0.05). Consistent with these findings, Lin et al. (2020) also found that Yamadazyma, Meyerozyma, and Aspergillus was the most abundant fungal genus in Laowo dry-cured hams ripened for 1–3 years, respectively. On the other hand, among the core fungal genera with a relative abundance >1%, the richness of Fusarium, Debaryomyces, Marasmius, and Apiotrichum were relatively higher in FH (p < 0.05), and Yamadazyma was the only more abundant fungus in LH (p < 0.05).

At the species level (Figure 5C), Meyerozyma sp., Aspergillus proliferans, and Yamadazyma triangularis were all among the core fungi (>1%) of both FH and LH (p < 0.05). Besides, Debaryomyces sp. and Apiotrichum sp. were also the dominant fungi in FH (p < 0.05). Moreover, among the core fungal species with a relative abundance >0.5%, FH showed more abundant Marasmius siccus, Debaryomyces sp., Apiotrichum sp., Archaeorhizomyces sp., and Fusarium solani (p < 0.05), whereas LH presented a higher relative abundance in Yamadazyma triangularis (p < 0.05).

Figure 6A shows the Pearson’s correlations between 17 differential core microbial genera (p < 0.05 and relative abundance >1%; 12 bacterial genera, 2 yeast genera, and 3 mold genera) and 29 discriminative volatile flavor compounds (p < 0.05) in Jinhua FH and LH. Surprisingly, 12 of the 17 differential core microorganisms, including 10 bacterial genera and 2 yeast genera, showed significant correlations with the distinct volatiles (p < 0.05 and |r| > 0.82). Among these microbial genera, Staphylococcus, Lactobacillus, Yamadazyma, Ochrobactrum, and Sphingomonas were one type of microbes mainly positively correlating with acids, branched alcohols, and a few linear-chain alcohols (2-propanol and 3-hexen-1-ol) in the two hams (p < 0.05 and r > 0.82). Staphylococcus has been considered the most important factor in the sensory characteristics development of dry-cured hams, owning to its strong abilities of nitrate reductase, catalase, protease, and lipase (Landeta et al., 2013; Xiao et al., 2020). Lactobacillus could extend the shelf life of hams by inhibiting the growth of spoilage bacteria via its product lactic acid, which could be further utilized by other microorganisms for SCFA synthesis (Figure 6A; Table 1; Xiao et al., 2020; Hu et al., 2022). Yamadazyma has been reported positively related to the organic acid level and negatively associated with the amino acid content in hams (Lin et al., 2020; Mu et al., 2020), suggesting its potential capability of producing SCFAs from amino acids (Figure 6A). Furthermore, Lactobacillus and Yamadazyma were both core microorganisms with higher relative abundances in LH (p < 0.05; Figures 4B, 5B), which could be an important contributor to the relatively higher acid volatile level (Figure 1; Table 2) and lower pH (Table 1) of LH (p < 0.05), owning to their potential association with SCFA generation.

On the other hand, Ruminococcaceae UCG-005, Psychrobacter, Ruminococcaceae UCG-014, and Debaryomyces were another type of microorganisms that positively correlated with aldehydes, ketones, most linear-chain alcohols, esters, and the other two compounds (2-pentylfuran and dimethyl disulfide) in the two hams (p < 0.05 and r > 0.82). Psychrobacter is a common bacterium in low-temperature fermented foods with substantial proteolysis ability (Broekaert et al., 2013). Zhang et al. (2021) has reported that Debaryomyces could produce branched aldehydes, such as 2-methylbutanal and 3-methylbutanal, by utilizing amino acids. Furthermore, Flores et al. (2004) reported that Debaryomyces could promote the generation of ethyl esters in dry-fermented meats. The intensities of 2-methylbutanal, 3-methylbutanal, and ethyl acetate were all positively correlated with the richness of Debaryomyces (Figure 6A), which were consistent with their findings. Moreover, the role of Ruminococcaceae family was rarely reported in fermented meat products, but it is known with capacities of fat degradation and fatty acid β-oxidation (Baars et al., 2018), which might contribute to the high-level volatiles originated from fats in FH, such as methyl ketones, furans, and some liner-chain aldehydes/alcohols (Figure 6A).

According to the present data and associated previous reports (Chen et al., 2017; Shi et al., 2019; Mu et al., 2020; Li et al., 2022), the potential metabolic network of discriminative volatiles’ formation under the involvement of core microorganisms in Jinhua FH and LH is further illustrated in Figure 6B. Briefly, the absence of fats significantly affected the microbial community structure and the flavor formation process in LH. By lipases from Staphylococcus and Ruminococcaceae in FH, fats can be degraded into fatty acids, which further generated methyl ketones and furans through decarboxylation and other reactions. Besides, the unsaturated fatty acids can be broken down through the β-oxidation by Ruminococcaceae to produce linear-chain aldehydes, which can be reduced to linear-chain alcohols and further esterified under the promotion of Debaryomyces. Whereas in LH, the protein metabolism was relatively enhanced due to the lack of fats and the action of characteristic core microbes. Under the proteolysis of Staphylococcus and Psychrobacter, more proteins were degraded into amino acids, which were then subjected to Strecker degradation and deamination for further generation of some aromatic aldehydes or linear-chained alcohols. Besides, the branched acids could be generated by Yamadazyma and some other microbes through utilizing amino acids or Lactobacillus-produced lactic acids. In summary, this metabolic network confirmed that the discriminative flavor compounds of FH were mainly β-oxidation and degradation products of fatty acids, whereas those of LH were mostly derived from the Strecker reaction or microbial metabolism of amino acids.