Introduction

Chinese mitten crab Eriocheir sinensis is a traditional savoury food in China, which originates from the coastal rivers and estuaries of the Yellow Sea. Currently, it has spread through Europe and California [1], and also constitutes a substantial freshwater fishery industry in China with the production increasing from 339,953 tons in 2002 to 714,380 tons in 2012 [2].

Many researchers [3–6] have previously revealed the discrepancies about constituents of flavor precursors (amino acids, fatty acids, nucleotides, etc.) among different edible parts of E. sinensis. Actually, the aroma profile of crab plays an important role for its purchase and consumption. In China, E. sinensis farmed in the Songjiang District in Shanghai is usually assumed to have high flavor quality. Besides, four edible portions of cooked E. sinensis—abdomen meat, claw meat, leg meat, and gonads—are individually popular, due to their unique pleasant aroma. The most common way of cooking the crab species is steaming. Therefore, four different edible parts of male E. sinensis farmed in Songjiang District were used as the raw materials in this study.

There exist hundreds of volatile compounds in crabmeat [7]. However, only a few of volatile compounds actively contribute to the aroma [8, 9]. Studies that used GC–MS to identify and quantify volatile compounds in crabs could not analyze the odor-active compounds in crab samples, but gas chromatography–olfactometry (GC–O) provided a valuable tool for investigating the pattern of odorants, including the odor descriptors and activities [10]. By using GC–MS/O, Chen and Zhang [11] found trimethylamine, dimethyl sulfide, 1-octen-3-one, dimethyl trisulfide, 1-octen-3-ol, 3-(methythio)-propanal, benzaldehyde, and 2-acetylthiazole as odor-active compounds in crabmeat. Among these compounds, trimethylamine with fishy aroma and dimethyl sulfide with crabmeat aroma were detected to be important odorants, due to their high odor intensity in the GC–O study. Yu and Chen [9] stated that aldehydes containing five carbon straight chains to nine carbon straight chains were important contributors to E. sinensis. Other researchers [12] detected 40 odorant compounds in four edible parts of E. sinensis. By combining the results of GC–O analysis with odor activity values (OAVs), trimethylamine (fishy, ammonia-like odor), (Z)-4-heptenal (mushroom-like odor), and benzaldehyde (paint-like odor) were selected as important odorants in all of four edible parts of steamed male E. sinensis.

It could be easily found that most studies mentioned above only focused on the characterization of odor-active compounds in crabmeat. Almost none of them compared the aroma profiles of four edible parts with olfactometrically detected compounds. In this study, sensory evaluation, MMSE-GC–MS/O, and the E-Nose technique were applied to determine the aroma profiles of different edible parts of E. sinensis. Further, by PCA procedures, key aroma properties and important odor compounds for four edible parts of E. sinensis could be identified. Also, by a comparison of sensory evaluation and GC–O detection, the relation between key aroma properties and specific odor compounds associated with certain edible part could be elucidated. This study may provide a theoretical basis for further mechanism research and development on IOCs in crabmeat or other aquatic products in the future.

Materials and methods

Samples and reagents

Twenty male E. sinensis of commercial size (average weight, ca. 200 g) were harvested at the middle and latter part of October in 2013 from Songjiang (Shanghai), and immediately transported alive to the laboratory within 2 h. Crabs were rinsed with tap water to remove foreign substances, and then cooked at a pot filled with distilled water and steamed for 30 min. After cooling to room temperature, the crabmeat (including abdomen, claw and leg meat) and gonads of 20 male E. sinensis were separated and picked out manually. Each part of 20 male E. sinensis was mixed together, homogenized in an ice-bath condition (Model JZ-II, Tianjin Sifang Equipment Ltd, China), and then stored at −80 °C until tested.

A standard solution of C5–C30 alkanes was used to calculate the linear retention indices (LRI) for each analyte, and an internal standard, 2,4,6-trimethylpyridine (TMP, purity 99 %), was used for quantification purposes. All the chemicals were purchased from Sigma–Aldrich (Shanghai, China).

HS-MMSE procedure

Volatile compounds from the samples were adsorbed using a previously described sampling kit [12]. Briefly, the sampling kit consists of seven monotrap™(TM) RCC18 rods (external diameter 2.9 mm; internal diameter 1 mm; length 5 mm; silica gel, activated carbon and octadecyl), the MT stand, and MT holder(type X5). Then 5 g of homogenized crab samples were fixed into a headspace vial (15 ml). Then seven MT rods were placed at fixed positions in the headspace vial. This vial was kept in a water bath for 50 min at 100 °C. After adsorption, the MT rods were removed and immediately placed in an adsorption tube and desorbed via a thermal desorption unit (TDU, Gerstel, Baltimore, MD, USA).

Gas chromatography–mass spectrometry/olfactometry (GC–MS/O)

The instrument used to select olfactometrically detected compounds from volatile compounds was a 7890 gas chromatography (GC) (Agilent Inc., Santa Clara) with 5975 mass selective detector (MSD) (CA, USA) and a sniffing port ODP-2 from Gerstel. The effluent from the capillary column was split 1:1.5 v/v between the mass spectrometry detector (MSD). The capillary column used was a DB-5MS provided by Agilent (length 60 m × internal diameter 0.32 mm × film thickness 1 μm; Agilent Inc., USA). Helium was used as carrier gas at 1.2 ml min−1.

The oven temperature was programmed from 40 to 100 °C at 5 °C min−1 with no initial hold, then to 180 at 2 °C min−1, to 250 °C at 5 °C min−1 with the final holds of 5 min. MS conditions were as follows: detector interface temperature, 250 °C; ion source temperature, 230 °C; ionization energy, 70 eV; mass range, 40–450 amu; electron multiplier voltage, 1576 V; and scan rate, 1.8 s−1. Samples were desorbed at 270 °C with a TDU (Gerstel) directly into the hot injector (250 °C) of the gas chromatography with simultaneous cryofocusing with liquid nitrogen.

A panel of five assessors (2 men and 3 women, aged from 22 to 35) from the laboratory staff with previous GC–O sniffing experience was assembled in the GC–O study. The assessors were asked to indicate the time, description, and odor intensity when an aroma was being detected. The scale used for intensity was 1–4 (1 = weak, 2 = moderate, 3 = strong, 4 = very strong). The analysis followed the guidelines of Pollien, Ott, Montigon, Baumgartner, Munoz-Box and Chaintreau [13]. Each sample was evaluated once by each of the three assessors consecutively.

Electronic nose (E-Nose) system

Crab samples were further analyzed by a FOX4000 sensor array system (Alpha M.O.S., France) to distinguish the discrepancies of aroma profiles of four different edible parts. This detecting instrument consisted of an auto-sampling device, a detector unit containing 18 metal oxide sensors (MOS), and a pattern of recognition software for data recording and elaboration [7]. The total of 18 sensors could be divided into three classes and located in their own chambers: Sensor chamber CL: LY2/LG, LY2/G, LY2/AA, LY2/GH, LY2/gCTL, LY2/gCT; Sensor chamber A: T30/1, P10/1, P10/2, P40/1, T70/2, PA/2; and Sensor chamber B: P30/1, P40/2, P30/2, T40/2, T40/1, TA/2.

Then 2 g of crab sample was loaded into a 10 ml glass vessel, and the vessel was sealed by a metal screw cap immediately and placed in a specimen tray (4 °C) for detection. The vessel was firstly balanced at 60 °C for 600 s before injection, and then 2500 μl of the headspace gas was injected into the sensor chamber at 2500 μl/s injection speed. Filtered and dried air (purity >99.999 %), with a flow rate of 150 ml/min, was used as a carrier gas for E-Nose detection. The data acquisition period lasted for 120 s and another 600 s was required for system rebalance. For each part, the E-Nose detection had been replicated five times under the same conditions.

Sensory evaluation

A panel of 8 people developed a consensus list of 5 terms (fishy, ammonia-like, fatty, grassy, and meat) to describe the aroma attributes of four edible parts of E. sinensis. They then used these terms to assess the aroma properties of the four different parts of samples, using a 10-point interval scale (0 = none, 9 = extremely strong). Sensory sessions took place in a sensory laboratory.

Identification and quantification for volatile compounds

Volatile compounds were identified by matching the mass spectra to the database spectra (Wiley/NIST 2008), and compared liner retention index (LRI) of each compound with its reference values. The LRI was calculated as follows:

$${\text{LRI}} = \left( {\frac{{Rt_{(x)} - Rt_{(n)} }}{{Rt_{{\left( {n + 1} \right)}} - Rt(n)}} + n} \right) \times 100,$$
(1)

where Rt(x) is the retention time of each volatile compound (x); Rt(n) and Rt(n + 1) are the retention times of n-alkanes eluting directly before and after the compound (x) under identical chromatographic conditions.

Also, 1 μl of TMP 1 × 102 ppm in methanol was added to 5 g of homogenized sample as a chromatographic internal standard before the MMSE process. The calibration factors were all assumed to be 1.00, and the estimated concentration of each volatile compound in the crab samples were calculated as follows:

$${\rm Est. \, Conc.\left( {ng/g} \right) = \frac{{Peak \, area \, ratio\left( {compound/TMP} \right) \times 1\; \mu g\left( {TMP} \right)}}{{5\;g\left( {\hom ogenized \, crab \, sample} \right)}}} \times 10^{3} .$$
(2)

Statistical analysis

Sensory evaluation and GC–MS/O detections were replicated three times in this study. The results of GC–MS were expressed as mean ± standard deviation (SD) (n = 3), and the quantitative data for each compound were compared for the four edible parts of E. sinensis using variance (ANOVA). Principal component analysis (PCA) was performed based on the intensity of 39 odor compounds found by GC–MS/O using SPSS version 20.0 (SPSS Inc., Chicago, IL, USA).

Results

Odor compounds detected by GC–MS/O

The ODCs from four edible parts of E. sinensis are shown in Table 1, and the results of the quantitative analysis of the aroma compounds released by the four edible portions are shown in Table 2. Thirty-nine ODCs were identified. Within all the ODCs, 26 compounds were found in abdomen meat, 30 were detected in claw meat, 27 were perceived in leg meat, and 24 were found in the gonad. Only 13 ODCs were detected to be common to all four parts. ODCs with an odor intensity no less than three could be determined as important odor compounds (IOCs), these compounds may be regarded as major contributors to the aroma profile of different edible parts. 2-Ethylpyridine (roast potato aroma) and 3-methyl-2-thiophenecarboxaldehyde (chocolate aroma) were found as IOCs in abdomen meat. Seven IOCs were detected in claw meat, they were trimethylamine (fishy aroma), 2-methylbutanal (nutty aroma), 2-ethylpyridine (roast potato aroma), 2,5-dimethylpyrazine (popcorn aroma), 2,3-dimethylpyrazine (popcorn aroma), benzaldehyde (almond aroma) and 1-octen-3-ol (mushroom aroma). Seven IOCs were found in leg meat, they were trimethylamine (fishy aroma), 1-penten-3-ol (mushroom aroma), 2-ethylpyridine (roast potato aroma), benzaldehyde (almond aroma), 2-nonanone (fruit-like aroma), and 2-decanone (fruit- like aroma). Ten IOCs—trimethylamine (fishy aroma), 1-penten-3-ol (mushroom aroma), hexanal (grassy aroma), heptanal (fatty aroma), 2-ethylpyridine (roast potato aroma), 2,5-dimethylpyrazine (popcorn aroma), 2,3-dimethylpyrazine (popcorn aroma), benzaldehyde (almond aroma), 2-acetylthiazole (roast meat aroma), and 4-ethylbenzaldehyde (almond aroma)—were detected in gonad parts.

Table 1 Olfactometrically detected compounds in four edible parts of male Eriocheir sinensis farmed in Songjiang district by GC–MS/O detection
Full size table
Table 2 Estimated concentrations of olfactometrically detected compounds identified in different edible parts of male Eriocheir sinensis by MMSE-GC–MS (μg kg−1, n = 3)
Full size table

ODCs that were only perceived in one edible part were determined to be unique odor compounds (UOCs). Three UOCs, 3-methyl-2-thiophenecarboxaldehyde (chocolate aroma), 1-pentanol (fruit-like aroma), and naphthalene (camphor ball aroma), were detected only in abdomen meat. Benzeneacetaldehyde (floral aroma), decanal (orange aroma), and 2-methylnaphthalene (plastic aroma) were only perceived in claw meat. 2-Octanone (fruit-like aroma) and 2-acetylthiazole (roast meat) were only found in the gonad.

Principal component analysis (PCA) of GC–MS/O data

Principal component analysis was carried out to visualize the differences between the four edible parts of E. sinensis (Fig. 1). Two principal components accounted for more than 70 % of the variation in the data: principal component 1 (PC 1) displayed 39.86 % of the variation and principal component 2 (PC 2) displayed 31.33 %. Abdomen meat was separated from the other three parts across PC1. The volatile compounds correlated with abdomen meat included 3-methyl-2-thiophenecarboxaldehyde, 1-pentanol, and naphthalene. The gonad was separated from the other three parts across PC2. Hexanal, 2-octanone and 2-acetylthiazole were associated with the gonad. In addition, benzeneacetaldehyde, decanal, 2-methylnaphthalene, and 2-ethylpyridine were correlated with claw meat, while 2,5-dimethylpyrazine, 2-pentylfuran (floral aroma) and (Z)-4-heptenal were associated with leg meat.

Fig. 1
figure 1

Principal component plot (PC1 versus PC2) of four edible parts, showing correlations with aroma volatiles (Numbers on plot refer to compound numbers in Table 1)

Full size image

Electronic nose response to the aroma of four edible parts of E. sinensis

To elaborate the discrepancies on aroma profiles of four edible parts of E. sinensis, an E-Nose experiment was performed in this study, and the PCA procedure was applied for data processing. As shown in Fig. 2, PC 1 and PC2 explained 95.12 and 3.73 % of sample variation, respectively. From Fig. 2, sample dots of four edible parts could be distinguished well. Sample dots of three kinds of crabmeat (abdomen, claw and leg meat) were all located in the second quadrant and third quadrant and found relatively close to each other, while sample dots of the gonad part that was located in the first quadrant were well separated from those of the crabmeat across PC 1, indicating that major differences could be found between aroma profiles of the gonad part and three kinds of crabmeat.

Fig. 2
figure 2

PCA analyses of four edible parts of E. sinensis from E-Nose response data

Full size image

Sensory analysis of four edible parts

Five aroma attributes were used to describe the sensory properties of four edible parts of Chinese mitten crab. The aromas were described as fishy, ammonia-like, fatty, grassy, and meat.

As shown in Fig. 3, abdomen meat scored highest for meat aroma, as well as scoring higher than claw meat and leg meat for fishy aroma, grassy aroma and fatty aroma. Meat aroma was higher in claw meat than leg meat and the gonad, as well as fishy aroma, grassy aroma and fatty aroma were higher in claw meat than leg meat. Ammonia-like aroma was higher in leg meat than abdomen meat and claw meat. Ammonia-like aroma, fishy aroma, grassy aroma, and fatty aroma were all higher in the gonad than the other three parts.

Fig. 3
figure 3

Graph of the mean sensory scores of four edible parts of Chinese mitten crab

Full size image

Principal component analysis (PCA) of sensory evaluation

Principal component analysis was also carried out on the correlation matrix of all samples and all attributes (Fig. 4). Again, all of the variation in the data was explained by two principal components. PC 1 displayed 78.33 % of the variation and PC 2 displayed 21.23 %. The gonad was separated from the other three parts across PC1. The attributes associated with the gonad were fishy aroma, fatty aroma, grassy aroma, and ammonia-like aroma. Whereas, meaty aroma was correlated with abdomen meat and claw meat.

Fig. 4
figure 4

Principal component plot (PC1 versus PC2) of four edible parts, showing correlations with sensory attributes

Full size image

Discussion

According to the data of GC–MS/O (Tables 1, 2), 6 IOCs were found in no less than two edible parts of E. sinensis. Among them, 2-ethylpyridine was the only IOC found in all four edible parts in this study (Table 1). It has been reported in krill oil [14] and dried Acetes chinensis [15], and also demonstrated as important odorants in claw meat, leg meat, and the gonad of E. sinensis, but not found in abdomen meat [12]. The origin of 2-ethylpyridine is likely to be decompositions of DHA and EPA upon cooking [16]. Guo et al. [17] have reported relatively high levels of C20:5n-3 and C22:6n-3 in all four edible parts of E. sinensis (abdomen meat: 10.99 and 9.49 %; claw meat: 12.97 and 8.33 %; leg meat: 15.89 and 11.18 %; gonad: 4.39–3.59 %), which may explain the high intensity of 2-ethylpyridine in our present study.

Trimethylamine and benzaldehyde presented high intensity scores in claw meat, leg meat, and the gonad respectively. Trimethylamine is a thermal decomposition product of choline, betaine, methionine, or most likely, trimethylamine oxide during cooking [18]. It has been reported as an important odorant in cooked E. sinensis [11], lobster [18], and cod [19] in high concentrations. Due to its high content and intense aroma in this study, trimethylamine could be a net contributor to the aroma of three parts—claw meat, leg meat, and the gonad—of E. sinensis. Benzaldehyde was also found with high intensities in claw meat, leg meat and gonad, but not found as the IOC in abdomen meat. The quantitative analysis (Table 2) did report a lower quantity of benzaldehyde in abdomen meat. It has been reported as important compounds in E. sinensis with paint-like aroma [12], and has been also reported as key odor-active compounds in steamed Coilia ectenes meat, with a “bitter almond” descriptor [20]. It is not surprising that benzaldehyde was found to be important as it is considered to be one of the lipid oxidation products with high aromatic impact [21].

2,5-Dimethylpyrazine only presented a strong popcorn aroma in claw meat and the gonad. The estimated concentrations of 2,5-dimethylpyrazine in claw meat and the gonad were 55.7 ± 2.0 and 56.0 ± 3.7 μg kg−1 (Table 2), respectively, much higher than that of in abdomen meat (35.8 ± 3.1 μg kg−1) and leg meat (31.8 ± 2.1 μg kg−1). Another N-containing compound, 2,3-dimethylpyrazine, has also been detected to contribute a strong popcorn aroma to the claw meat and gonad of E. sinensis. It has not been reported as odor-active components in E. sinensis before, but has already been described with popcorn and rice aroma in salted-dried white herring [22]. These two compounds are probably thermally generated via Maillard and pyrolysis reactions through Strecker degradations from various nitrogen sources such as amino acids in heat-processed foods [15, 23].

1-Penten-3-ol was identified as IOC in both leg meat and the gonad of E. sinensis. It has been detected as a high intensity odorant only in the gonad of E. sinensis farmed in Yangcheng Lake. The estimated concentration of 1-penten-3-ol was highest in the gonad (241.1 ± 15.7 μg kg−1), followed by leg meat (207.9 ± 15.9 μg kg−1) and abdomen meat (112.0 ± 8.8 μg kg−1), and it was not found in claw meat. This alcohol is generated from PUFAs oxidation catalyzed by lipoxygenase or hydroperoxidase [20]. It is not surprising that the intensity of 1-penten-3-ol was high in leg meat, as its precursor —PUFAs— has been detected at a relatively high level of the leg meat of E. sinensis [17].

Differences have also been detected in four edible portions (Figs. 2, 3, 4). The gonad part contained the largest amount of IOCs, and possessed the strongest aroma in ammonia-like, fishy, grassy, and fatty attributes in sensory analysis. This might result from the flavor precursor (amino acids, fatty acids, etc.) discrepancies between the gonad and the other three parts. The gonad was rich in fatty acid, while crabmeats contained more protein. Crude fat content in the gonad of E. sinensis was 23.53–26.30 %, much higher than that in the crabmeat (0.44–1.29 %) [17]. N−3 polyunsaturated fatty acids (PUFA) were revealed as important precursors in E. sinensis, although the percentage of n−3 PUFA in total fatty acids (TFA) in the gonad was 14.5–15.1 %, a little lower than that in the crabmeat (15.60–24.88 %) [24]; the content of n−3 PUFA in E. sinensis’s gonad was still higher than that in crabmeat [25]. Hexanal, 2-octanone, and 2-acetylthiazole were closely related to the gonad part (Fig. 1). Among them, hexanal has been widely reported in crab species [5, 7, 12]. It is usually considered as the degradation product of linoleate and linolenate hydroperoxides [7, 26]. The origin of 2-octanone is likely to be a thermal degradation of PUFAs [27]. 2-Acetylthiazole was the UOC identified in the gonad with a high intensity score. It has already been reported in E. sinensis [7] and lobster [18] in similar concentrations. The origin of 2-acetylthiazole in the sample is probably a thermal degradation of cysteine [18].

Sample dots of abdomen meat and claw meat in Fig. 4 were both located in the second quadrant, and the meaty aroma attribute was closely related to these two parts. The aroma of abdomen meat was found to be correlated with 3-methyl-2-thiophenecarboxaldehyde, 1-pentanol, and naphthalene. Benzeneacetaldehyde, decanal, 2-ethylpyridine, and 2-methylnaphthalene were correlated with claw meat (Fig. 1). Unlike the high levels of fatty acids in the gonad, abdomen meat and claw meat of E. sinensis are rich in amino acids [17]. The branched chain aldehydes as well as phenyl-containing aldehydes are aroma components usually generated from the Strecker reaction of amino acids: e.g., 3-methyl-2-thiophenecarboxaldehyde (sulphur amino acids) [28] and benzeneacetaldehyde (phenylalamine) [7]. Among them, 3-Methyl-2-thiophenecarboxaldehyde is worth mentioning because of its high odor intensity in abdomen meat. In addition, decanal is the degradation product of linoleate and linolenate hydroperoxides [7, 26]. 1-Pentanol is probably formed from branched-chain amino acid precursors [29]. 1-Octen-3-ol is considered as a degradation product of linoleic acid hydroperoxides [20]. These compounds have been previously reported as odor-active compounds in crabs species [12] and other aquatic foods, such as smoked salmon [30], and scallops [31]. 2-Ethylpyridine was found to be closely associated with claw meat, which might be affected by its relatively higher intensity score in the GC–O study. Naphthalene and 2-methylnaphthalene, which may be considered as products generated from environmental contaminants [30], have been previously reported in crabs [12] and fresh salmon [30]. As numerous of ODCs had been found in four edible parts of E. sinensis (most of them contributed to pleasant aroma), a limited negative impact could be made by naphthalene and 2-methylnaphthalene to the integral aroma of E. sinensis.

The aroma in leg meat was moderate, which may be caused by the lower quantities of IOCs. (Z)-4-Heptenal, 2-pentylfuran, 2-nonanone, and 2-decanone had a comparatively high odor intensity in leg meat than the other three parts. Among these ODCs, (Z)-4-heptenal and 2-pentylfuran showed a close relationship with leg meat (Fig. 1). (Z)-4-Heptenal is usually considered as the degradation products of linoleate and linolenate hydroperoxides [7, 26]. The percentage of n-3 PUFAs in leg meat was relatively higher (37.33 %) than the other three parts [24], which may explain the high intensity of (Z)-4-heptenal in leg meat. 2-Pentylfuran has not been detected as ODC in E. sinensis before, but it presented a “sweet,green” aroma in crabs farmed in the Songjiang district. The estimated concentration in leg meat was much higher than that of in the other three parts. These compounds could be produced by thermal degradation of lipid or thiamine [32].

In summary, the aroma profiles of four edible parts of Chinese mitten crab were composed of 39 ODCs found by GC–MS/O. The ODCs of Chinese mitten crab were dominated by the presence of numerous aldehydes and N-containing compounds. Among them, IOCs would make a great contribution to aroma profiles of each part. 2-Ethylpyridine, trimethylamine, benzaldehyde, 2,5-dimethylpyrazine, 2,3-dimethylpyrazine, and 1-penten-3-ol were found to be IOCs in no less than two edible parts of E. sinensis. The differences of aroma properties from four edible parts were detected by E-Nose and sensory evaluation. The gonad part had strong aroma in ammonia-like, fishy, grassy, and fatty attributes in sensory evaluation, which make it differentiate from the other three parts. Hexanal and 2-acetylthiazole might be the important contributors to the aroma profile of the gonad. Abdomen meat and claw meat were related to a strong meat aroma. 3-Methyl-2-thiophenecarboxaldehyde could make a great contribution to the aroma profile of abdomen meat, while 2-ethylpyridine could be associated closely with the aroma profile of claw meat. The aroma of leg meat was moderate, and (Z)-4-heptenal might be the important contributor to the leg meat.