Potential Contribution of Fish Feed and Phytoplankton to the Content

Apr 19, 2017 - Geosmin and 2-methylisoborneol are the most recognized off-flavors in freshwater fish, but terpenes may also contribute off-flavor in f...
2 downloads 12 Views 646KB Size
Article pubs.acs.org/JAFC

Potential Contribution of Fish Feed and Phytoplankton to the Content of Volatile Terpenes in Cultured Pangasius (Pangasianodon hypophthalmus) and Tilapia (Oreochromis niloticus) Raju Podduturi,† Mikael A. Petersen,‡ Sultan Mahmud,§ Md. Mizanur Rahman,§ and Niels O. G. Jørgensen*,† †

Department of Plant and Environmental Sciences, Section of Microbial Ecology and Biotechnology, University of Copenhagen, 1871 Frederiksberg, Denmark ‡ Department of Food Sciences, Section of Design and Consumer Behaviour, University of Copenhagen, 1958 Frederiksberg, Denmark § Department of Fisheries Technology, Faculty of Fisheries, Patuakhali Science and Technology University, Patuakhali-8602, Bangladesh ABSTRACT: Geosmin and 2-methylisoborneol are the most recognized off-flavors in freshwater fish, but terpenes may also contribute off-flavor in fish. We identified six monoterpenes, 11 sesquiterpenes, and three terpene-related compounds in pangasius and tilapia from aquaculture farms in Bangladesh. The concentrations of most of the volatiles were below published odor thresholds, except for α-pinene, limonene, β-caryophyllene, α-humulene, and β-ionone in tilapia, and limonene and βionone in pangasius. To identify sources of the terpenes, terpene profiles of fish feed and phytoplankton in the ponds were analyzed. In feed and mustard cake (feed ingredient), five monoterpenes and two sesquiterpenes were identified, and five of these compounds were also detected in the fish. In phytoplankton, 11 monoterpenes were found and three also occurred in the fish. The higher number of terpenes common to both fish and feed, than to fish and phytoplankton, suggests that feed was a more abundant source of odor-active terpenes in the fish than phytoplankton. KEYWORDS: terpenes, off-flavor, tilapia, pangasius, phytoplankton, fish feed



phytoplankton and filamentous microorganisms may possibly also contribute off-flavor compounds in fish since they have been found in the stomach content of tilapia.1 In large-scale aquaculture facilities with water recirculation, tainting of fish by geosmin and 2-MIB is most likely caused by the uptake of dissolved molecules, e.g., originating from microorganisms in biofilters for water treatment.12,13 In fishponds with stagnant water and an abundant phytoplankton community, ingestion of cellular material, such as colony-forming cyanobacteria, has been observed in tilapia and may be an alternative source of tainting in fish.14 Research on off-flavors in fish has typically focused on the compounds geosmin and 2-MIB, while the presence of other potential off-flavor compounds, such as terpenes, has only been examined in a few fish, e.g., in rainbow trout.6,7 Here, we wished to examine if terpenes in fish feed and phytoplankton can affect the taste and flavor of the globally important aquaculture species tilapia (Oreochromis niloticus) and pangasius (Pangasianodon hypophthalmus). The study was initiated because sensory profiling of tilapia and pangasius, produced in ponds in Bangladesh, characterized the taste and flavor of the fish as unsatisfactory, unless the fish were purged in off-flavorfree water before slaughtering.2 Yet, the concentration of

INTRODUCTION Sensory studies of off-flavor in freshwater fish indicate that the most commonly perceived off-flavor and/or off-odor descriptors are earthy-musty, originating from the presence of geosmin and 2-methylisoborneol (2-MIB), respectively. The earthymusty flavor has been reported for tilapia,1 pangasius,2 channel catfish,3 rainbow trout,4 and barramundi.5 However, in addition to geosmin and 2-MIB, several other organic compounds can influence the taste and flavor of freshwater fish. In rainbow trout, more than 30 volatile compounds have been identified by both instrumental (GC-MS) and combined GC-olfactory analyses.6,7 Besides geosmin and 2-MIB, the volatiles in rainbow trout were dominated by aldehydes, such as nonanal and 2,4-octadienal (both with vegetable flavor), alcohols, e.g., 1octen-3-ol (mushroom smell), and terpenes, e.g., L-α-terpineol and farnesol (flowery and earthy flavor). Other terpenes identified in freshwater fish are β-cyclocitral (grass and woody flavor) and β-ionone (woody and flowery flavor).8 Terpenes are common substances in many plants and photosynthetic microorganisms, especially in cyanobacteria,9 and they are a part of the pool of dissolved organic matter in freshwater.10 The group of terpenes includes monoterpenes with two isoprene units, e.g., 2-MIB, and sesquiterpenes with three isoprene units, e.g., geosmin, as well as terpenoids (slightly modified terpenes). Terpenes in fish originate from an external source since fish cannot produce terpenes. Sites for uptake of dissolved terpenes include gills, skin, and the intestine,11 but feeding on © 2017 American Chemical Society

Received: Revised: Accepted: Published: 3730

February April 16, April 19, April 19,

3, 2017 2017 2017 2017 DOI: 10.1021/acs.jafc.7b00497 J. Agric. Food Chem. 2017, 65, 3730−3736

Article

Journal of Agricultural and Food Chemistry

flow rate of 100 mL/min for 60 min at 50 °C to collect the volatiles on Tenax TA trap. After volatile collection, the traps were further purged with N2 at 100 mL/min for 10 min to remove water from the traps. Dynamic Headspace Sampling of Volatiles from Phytoplankton Biomass. An optimized method based on Höckelmann et al.17 was used to extract volatiles from the filters with phytoplankton biomass. One quarter of a 47 mm GF/F filter with phytoplankton was transferred to a 250 mL gas-washing flask, followed by the addition of 50 mL 30% NaCl solution, and equilibrated at 50 °C for 10 min constant stirring of 230 rpm. The volatiles were collected on a Tenax TA trap by purging the phytoplankton suspension with N2 at a flow rate of 200 mL/min for 60 min at 50 °C. After volatile extraction, the traps were further purged with N2 at 100 mL/min for 10 min to remove water from the traps. Dynamic Headspace Sampling of Volatiles from Pond Water. Volatile compounds in the water were extracted on Tenax TA traps by transferring 10 mL of water into a 100 mL gas washing flask followed by purging with N2 at a flow rate of 100 mL/min for 60 min in a water bath at 37 °C. Residual water in Tenax TA traps was removed by purging with N2 at 100 mL/min for 10 min. Dynamic Headspace Sampling of Volatiles from Fish Feed. Two commercial feeds (“Pangas Grower” and “Tilapia Starter”, produced by Krishibid Feed Ltd., Bangladesh) and two of the major ingredients, mustard seed cake and rape seed cake (obtained after oil and protein extraction), were tested for content of volatiles. In brief, the feed material was ground into fine powder in a laboratory blender. From the ground material, 5 g was transferred into 100 mL gas washing flasks followed by purging with N2 at a rate of 100 mL/min for 60 min at 37 °C. Volatiles were collected on Tenax TA traps, which were further purged with N2 at 100 mL/min for 10 min to remove water. GC-MS Analysis. The trapped volatiles were desorbed using an automatic thermal desorption unit (ATD 400, PerkinElmer, Norwalk, USA). Primary desorption was carried out by heating the trap to 250 °C with a flow of carrier gas (60 mL/min H2) for 15 min. The stripped volatiles were trapped in a Tenax TA cold trap (30 mg held at 5 °C), which was subsequently heated at 300 °C for 4 min (secondary desorption; outlet split 1:10). This allowed for rapid transfer of volatiles to a gas chromatograph−mass spectrometer (GC-MS, 7890A GC-system interfaced with a 5975C VL MSD with Triple-Axis detector from Agilent Technologies, Palo Alto, California) through a heated (225 °C) transfer line. Separation of the volatiles was carried out on a DB-Wax capillary column (30 m length ×0.25 mm internal diameter and 0.5 μm film thickness) using H2 as carrier gas with an initial flow rate of 1.0 mL/ min. The column temperature program was held at 40 °C for 10 min, then raised to 240 °C at the rate of 8 °C/min and finally for 5 min at 240 °C. The mass spectrometer was subjected to electron ionization mode at 70 eV. Mass-to-charge (m/z) ratio between 15 and 300 were scanned. GC-MS Data Processing and Semi-Quantitation. The GC-MS data were processed using MSD Chemstation software (E 02.02.1431, Agilent Technologies, Palo Alto, California). Volatile compounds were identified by probability-based matching of their mass spectra with those from the commercial database (Wiley275.L, HP product no. G1035A). Retention indices (RI) were calculated for all detected compounds by running an alkane standard mixture (C5−C22). The identity of 10 compounds in fish, 7 in phytoplankton, and 4 in fish feed was confirmed by running authentic standards, and the remaining compounds were tentatively identified using MS library suggestions from Wiley and confirmed with the RI values from the literature. Geosmin in fish samples were quantified in one of the author’s previous papers.2 The same peak area relative to the concentration calibration curve was used to interpret the concentration of detected compounds in the fish. A calibration curve for δ-3-carene was prepared, and the peak area relative to the concentration was used to quantitate levels of terpene compounds in fish feed and phytoplankton biomass and are expressed as semiquantitation.

geosmin and 2-MIB in the fish was too low to cause tainting, indicating that other organic compounds led to the off-flavor and off-taste. We suspect that terpenes, either in phytoplankton or as free dissolved molecules in the pond water, contributed off-flavors to the fish since many terpenes have distinctive and characteristic flavors. An alternative source of off-flavor might be fish feed since the content of volatile compounds in fish may depend on the feed composition, as shown for the terpene limonene in sole (Solea senegalensis) at an increased intake of plant protein.15 In an attempt to obtain more information on relationships between terpenes in fish and their sources of food, we analyzed the composition of terpenes in the flesh of tilapia and pangasius from ponds in southern Bangladesh. The terpene profiles were related to terpenes in two commercial feeds, used by local fish farmers, as well as to the common feed ingredients mustard and rape seed cakes. Finally, the composition of terpenes in natural phytoplankton in the ponds was analyzed and compared to the content of terpenes in the fish. The purpose of the study was for the first time to identify volatile terpenes in the two commercially important freshwater fish species, tilapia and pangasius, and provide information on potential sources of these terpenes.



MATERIALS AND METHODS

Chemicals. Geosmin and 2-methylisoborneol at 100 μg/mL in methanol solution were purchased from Supelco (Sigma-Aldrich Co., Ltd., USA). The following chemicals of maximum purity and GC grade were purchased from Sigma-Aldrich Co., Ltd.; α-pinene, β-myrcene, limonene, α-humulene, α-phellandrene, δ-3-carene, α-ionone, βionone, α-gurjunene, aromadendrene, α-terpineol, γ-terpinene, βcaryophyllene, terpinolene, and α-terpinene. Sample Collection. Tilapia (Oreochromis niloticus) and pangasius (Pangasianodon hypophthalmus) and phytoplankton were collected in eight ponds of 309 to 497 m2 area and water depths of 1.1 to 1.6 m in Dumki Upazila, Patuakhali, southern Bangladesh. Details on sampling, fish breeding, and phytoplankton abundance and composition are given in Petersen et al.2 Briefly, the fish were cultivated for 6 months after stocking in ponds and had weights of 182−199 g (tilapia) and 689−870 g (pangasius) when harvested in August 2012. The fish were collected in nets and after stunning, they were bled out by cutting of the gill blood vessels. The fish were filleted in the laboratory, and one fillet was used for chemical analysis for the content of geosmin, 2-MIB, and terpenes, while the other fillet was used for sensory analyses (data on geosmin and 2-MIB are published in Petersen et al.2). All fillets were kept frozen at −20 °C until analysis in Denmark. Fillets from a total of 45 fish from each of the two species were analyzed for the content of terpenes (see below). Simultaneous with harvest of the fish, phytoplankton in 500 mL of water from each pond was filtered onto 47 mm diameter GF/F filters. The filters were stored in aluminum foil and kept at −20 °C until analysis of photosynthetic pigments and terpenes. The phytoplankton biomass declined during the growth season, ranging from 40 to 375 μg chlorophyll a L−1 in March to 8 to 100 μg chlorophyll a L−1 at harvest in August and was dominated by chlorophytes (green algae), followed by diatoms, cryptophytes, and cyanobacteria during the entire growth season.2 Dynamic Headspace Sampling of Volatiles in Fish. A dynamic headspace sampling method was employed to extract volatile compounds from the fish according to Petersen et al.16 From each fillet, 20 g of flesh was weighed into a 250 mL gas-washing flask along with 25 mL of water and 1 mL of internal standard (50 μL/L 4methyl-1-pentanol; Sigma-Aldrich, Germany). The mixture was homogenized using an Ultra Turrax T25 homogenizer (IKA, Staufen, Germany) for 45 s at 13,500 rpm, followed by rinsing with 10 mL of water. The gas-washing flask was fitted with a purge head and transferred to a water bath, where the sample was purged with N2 at a 3731

DOI: 10.1021/acs.jafc.7b00497 J. Agric. Food Chem. 2017, 65, 3730−3736

Article

Journal of Agricultural and Food Chemistry

Table 1. Terpenes Identified in the Flesh of Pangasius and Tilapia along with Their Retention Indices (RI), Odor Description, and Odor Thresholda semiquantitationb (ng/g)

identification calculated RI

compound

RI of authentic standard

1024

α-pinene

1018

1175 1199 1208

β-myrcene limonene β-phellandrene

1176 1206

1284

cymene

1297

terpinolene

1470 1481 1506 1547

α-cubebene δ-elemene α-copaene α-gurjunene

1590 1617 1629 1646 1692 1718 1750 1779 1861 1968

longifolene β-caryophyllene aromadendrene thujopsene α-humulene α-terpineol α-muurolene δ-cadinene calamenene β-ionone

tentative (MS, RI from the literature)37

odor description42 pine, turpentine musty38 citrus, mint mint, turpentine gasoline, citrus woody, piney44 herb, wax wood wood, spice wood, balsamic woody44 wood, musty38 wood

1216 1275 1298 147240 146532 1519 1547 1574 1616 1628 1660 1689 1719

wood oil, anise, mint wood thyme, wood herb, spice wood, violet

1753 1784 1839 1965

pangasius

tilapia

odor threshold in water (ppb)43

0.06−3

0.15−10.4

6

0.02−3.25 0.15−45 0.08−1

0.03−2.8 0.35−20.8 0.02−0.25

13−15 10 50039

0.15−2.2

0.18−3.2

0.01−1

0.03−0.5

0−0.5 0−0.4 0.01−1 0.007−4.5

0.01−128 0.01−0.25 0.02−21 0.05−3.5

0.03−1 0.02−7 0−1.5 0.03−0.3 0.02−8.3 0.01−0.4 0.01−0.9 0.01−2 0.02−10 0−0.15

0.05−4.7 0.1−120 0.01−0.8 0.01−0.5 0.05−115 0.05−1.5 0.07−27 0.08−127 0.35−108 0−0.2

200

6439

12039 330−350

0.007

Terpenes in the fish were determined by authentic standards or identified using the MS library information and confirmation from RI values. bThe terpene content range for 45 fish is shown. a

Table 2. Terpenes Identified in Phytoplankton Biomass Collected from Fish Ponds along with Their Retention Indices (RI) and Odor Description identification calculated RI

compound

RI of authentic standard

1172 1176 1184 1212 1258 1260 1296 1301 1353 1876 1961

α-phellandrene δ-3-carene α-terpinene β-phellandrene unknown 1 γ-terpinene terpinolene isoterpinolene unknown 2 α-ionone β-ionone

1174 1157 1182



tentative (MS, RI from the literature)37

1216 1259 1298 133141 1875 1965

semiquantitation (ng/100 mL)

odor description42

0.8−9 0.3−5 2.5−27 0.8−8 0.4−20 0.1−15 0.4−37 1−12 0.6−19 0.4−20 0.2−51

turpentine, mint, spice lemon, resin lemon mint, turpentine gasoline, turpentine woody, piney44

wood, violet wood, violet

caryophyllene (0.1−120 ppb), α-humulene (0.05−115 ppb), and β-ionone (0−0.2 ppb) in some tilapia and limonene (0.15−45 ppb) and β-ionone (0−0.15 ppb) in pangasius. However, for most of the identified compounds, no odor threshold values are known. Terpene Profiles of Phytoplankton Biomass. Eleven monoterpenes were found in the phytoplankton biomass from the ponds (Table 2). Among the terpenes, the identity of 7 compounds was confirmed by authentic standards, two compounds by comparison to literature RI values, while the identity of two compounds could not be determined, but their mass spectra were similar to those of other monoterpenes, and they were named unknown 1 and 2. Three compounds,

RESULTS Terpene Profile of Pangasius and Tilapia. A total of 20 terpenes, including 6 monoterpenes, 11 sesquiterpenes, and 3 other terpene-related compounds were identified in the pangasius and tilapia fillets (Table 1). Eleven of the compounds were characterized by a woody odor. Nine of these compounds were sesquiterpenes, while one was a monoterpene (terpinolene), and one was a norisoprenoid (β-ionone). The remaining nine compounds were characterized by terpene odors, such as pine, musty, and herb-like odors. The concentration of most of the volatile compounds in the fish was below published odor thresholds. Exceptions were αpinene (0.15−10.4 ppb), limonene (0.35−20.8 ppb), β3732

DOI: 10.1021/acs.jafc.7b00497 J. Agric. Food Chem. 2017, 65, 3730−3736

Article

Journal of Agricultural and Food Chemistry Table 3. Terpenes Identified in Fish Feed along with Their Retention Indices (RI) and Odor Description identification calculated RI

compound

RI of authentic standard

1016 1152

α-pinene δ-3-carene

1018 1157

1198 1257

limonene p-cymene

1206

1298 1593 1744

terpinolene longifolene α-muurolene

1298

tentative (MS, RI from literature)37

1250

1574 1753

odor description42

semiquantitation (ng/g)

found in

pine, turpentine lemon, resin

5−43 9−75

citrus, mint solvent, gasoline, citrus woody, piney43 woody44 woody

3−26 1−3

Pangas Grower mustard seed cake Pangas Grower mustard seed cake Tilapia starter all feeds and feed ingredients Pangas Grower

3−7 0.6−1.5 0.3−0.8

mustard seed cake Pangas Grower mustard seed cake

Figure 1. Venn diagram comparing the most abundant terpenes in fish, phytoplankton, and fish feed.

terpinolene, α-ionone, and β-ionone, are characterized by woody odor notes, while odor descriptors of the remaining compounds typically are turpentine- and lemon-like. Only three compounds, β-phellandrene, terpinolene, and β-ionone, were present in both fish flesh and phytoplankton biomass. The content of β-phellandrene, terpinolene, and β-ionone ranged from 0.8 to 8, 0.4 to 37, and 0.2 to 51 ng/100 mL of filtered water, respectively. No sesquiterpenes were found in phytoplankton biomass. Terpene Profiles of Fish Feed. Four monoterpenes along with two sesquiterpenes and one terpenoid were detected in the feeds and/or in the feed ingredients (Table 3). Three of the compounds were identified by authentic standards, while the rest were identified by comparing RI values as mentioned above. Among the 7 identified terpenes, five were found in Pangas Grower feed, four in the mustard seed cake, and two in the Tilapia Starter feed, while the rape seed cake only contained one terpene (limonene). Five terpenes, including three monoterpenes (α-pinene, limonene, and terpinolene) and two sesquiterpenes (longifolene and α-muurolene), were present in both fish flesh and feed. Three of the compounds

found in fish and feed have woody odor notes (terpinolene, longifolene, and α-muurolene). Concentrations of monoterpenes (α-pinene at 5−43 ng/g and δ-3-carene at 9−75 ng/g) in feed samples were higher than the concentrations of sesquiterpenes (longifolene at 0.6−1.5 ng/g and α-muurolene at 0.3−0.8 ng/g). Terpene Profiles of Pond Water. No terpenes were detected in the water in the ponds. Mean geosmin concentration in the pond water was 3.9 ng/L, and concentrations >10 ng/L occurred only in 3 of 56 samples, while 2-MIB was undetectable due to coelution with an unknown compound.2



DISCUSSION Among the 20 identified terpenes in the two fish species, five terpenes were also found in the fish feed, but only three of the terpenes occurred in both fish and phytoplankton (Figure 1). This may suggest that fish feed and algae in the water were not major sources of volatile compounds that contributed off-flavor to the two fish species. However, the knowledge on the abundance of different volatile terpenes in fish is limited, and 3733

DOI: 10.1021/acs.jafc.7b00497 J. Agric. Food Chem. 2017, 65, 3730−3736

Article

Journal of Agricultural and Food Chemistry

included in the terpene analysis, implying that the terpenes that might have been released by algae to the pond water were not part of the terpene profiles. However, no terpenes were detected in the pond water, suggesting that release of monoand sesquiterpenes by phytoplankton was negligible at sampling time. Terpenes Produced by Bacteria. Another source of terpenes in the fish may have been filamentous Actinobacteria (popularly named as actinomycetes). Filamentous Actinobacteria are common bacteria in freshwater,28 and they produce many structurally diverse terpenes.29,30 The latter authors (Dickschat29 and Rabe et al.30) reported the production of volatile terpenes by different geosmin-producing Actinobacteria (Saccharopolyspora spinosa, Streptomyces filamentosus, and S. rimosus). Among terpenes identified by Rabe et al.30 were the following sesquiterpenes that were also identified in fish in our study: α-cubebene, δ-elemene, α-copaene, β-caryophyllene, αhumulene, and α-muurolene, α-pinene, limonene, and αterpineol. In isolates of Streptomyces, Yamada et al.31 observed the expression of genes responsible for the synthesis of several terpenoids that we also found in the fish: δ-cadinene, αgurjunene, α-humulene, and β-elemene (in S. clavuligerus); αcopaene and α-muurolene (in S. griseus); α-copaene, aromadendrene, and α-cubebene (in S. lactacystinaeus). Pollak and Berger32 reported geosmin-related volatiles produced by S. citreus, including β-myrcene, δ-elemene, and δ-cadinene. These volatiles were also found in the present study. Densities of Streptomyces spp. in the Bangladeshi fish ponds were quantified by Petersen et al.,2 but their actual contribution to the production of terpenes in the ponds was not examined. Filamentous, photosynthetic bacteria (cyanobacteria) have been found in the stomach content of tilapia,1 but it is unknown if fish actively feed on larger filamentous microbes, such as cyanobacteria, or whether the microbes are ingested because they are attached to particles in water and sediment. Terpenes in Fish Feed. The presence of terpenes in meat and other food products of animal origin has previously been attributed to feed and feeding habitats.33 For example, the sesquiterpene β-caryophyllene was described as a biomarker of grass feeding in sheep meat.34 In the flesh of Thai tilapia, the content of terpenes was assumed to originate from cyanobacteria among the phytoplankton.1 In marine sole, a higher content of plant protein in the feed increased the amount of limonene in the flesh.15 In our study, five of 20 compounds detected in the fish were also found in the feed, and four were identified in the feed ingredient mustard seed cake. α-Pinene and limonene were the most abundant monoterpenes common to both fish and feed (up to 43 and 26 ng/g, respectively, in the feed), while the dominant monoterpene in the feed, δ-3-carene (up to 75 ng/g in feed and mustard seed), did not occur in the fish flesh. Among the 11 identified sesquiterpenes in the fish, only two (α-muurolene and longifolene) were also identified in the feed and feed ingredients (up to 0.8 and 1.5 ng/g, respectively, in feed or feed ingredient). In an attempt to estimate if terpenes in the fish actually might originate from the feed, the feeding time required for the terpene limonene (occurred in all feeds and fish) to fulfill the content in the fish was calculated. During the last months before harvest, the fish were fed about 3% of their wet weight every day.2 Applying the measured range in the content of limonene in feed and fish (Tables 1 and 3), weight of the fish at harvest and the daily feed ration, the number of days required

specific sources of terpenes and other odor-active components are typically based on assumption.1 Terpenes in Fish. Although terpenes previously have been identified in fish, most of the terpenes in this study are reported for the first time in fresh fish flesh. Among previously reported terpenes in fish are α-pinene, β-caryophyllene, and limonene in rainbow trout,18,7 α-pinene, cymene, limonene, and α-terpineol in sea bass,19 aromadendrene in salmon,20 and α-pinene, myrcene, limonene, and longifolene in sea bream.21,18 Most of these aroma compounds are characterized by fruity, vegetablelike, and woody flavor descriptors. In the analyzed tilapia and pangasius samples, about half of the volatiles had woody odor descriptors in common, while the rest of the compounds typically had fruity or spicy flavors. Woody and piney flavors have previously been identified in sensory flavor profiling of fish.22,23 van der Ploeg22 proposed a flavor wheel to grade the pond-raised channel catfish in which the flavor notes of earthy, musty, woody, and piney were described as most objectionable flavors, although the woody off-flavor was one of the most difficult notes to perceive. The origin of woody flavor in the fish was not determined in this study, and no attempts to correlate sensory perception with instrumental flavor analysis were made. In the channel catfish, it was presumed that the flavor originated from cyanobacteria or the degradation of 2-MIB.22,23 The chemical analysis of tilapia and pangasius demonstrated the presence of several terpenes with a woody odor property, suggesting that this flavor could be the main reason for the low grade in sensory quality given by the Bangladeshi panel.2 Unfortunately, the odor threshold is only known for a few of the presently detected compounds and only from solutions in water. Among the most abundant compounds in the fish were β-caryophyllene, α-humulene, and β-ionone, and they are all characterized by woody notes. Although only three of 11 compounds with woody odor levels were above their threshold (determined in water), the cumulative perception of woody notes from several individual compounds may significantly have affected the overall sensory quality. Not only terpenes but also their degradation products may taint fish since both terpenes and terpenoids can be very reactive. For example, β-ionone, in the presence of β-ionone oxygenase, produces several degradation products like β-ionol, dihydro-β-ionone, tetrahydroionone, and 4-oxo-β-ionone. These compounds have all been reported to taint fish.24 Terpenes in Phytoplankton. Only three monoterpenes were found in both fish and phytoplankton biomass, and surprisingly no sesquiterpenes, not even geosmin, were found in the phytoplankton biomass (information on geosmin is unpublished information by the authors). Among the identified monoterpenes in the phytoplankton was β-ionone, which is one of the most common terpenoids produced by cyanobacteria.24,25 β-Ionone was also the most abundant terpene in some of the present plankton samples. Certain filamentous cyanobacteria are known to produce sesquiterpenes in addition to geosmin, e.g., germacrene-D and γ-cadinene by Oscillatoria sp.17,26 and 8a-epi-α-selinene by Nostoc punctiforme.27 The phytoplankton in the Bangladeshi ponds was dominated by chlorophytes and diatoms, rather than cyanobacteria,2 but the terpene profiles of these two algal groups are still uncharacterized. Possibly, the present procedure for extraction and analysis of volatile compounds in the phytoplankton biomass (addition of salt and heating at 50 °C) underestimated the production of terpenes in the algal cells. Only intracellular compounds were 3734

DOI: 10.1021/acs.jafc.7b00497 J. Agric. Food Chem. 2017, 65, 3730−3736

Article

Journal of Agricultural and Food Chemistry to reach the limonene content in the fish was 0.5 to 230 days in tilapia and 0.2 to 500 days in pangasius, if the feed was the only source of limonene. Thus, it is realistic to assume that limonene and possibly also other terpenes in the fish did originate from the feed. However, the present comparison of terpenes in feed, feed ingredients, and fish flesh should be considered indicative rather than conclusive since we were not able to analyze the feed originally used for raising the fish. Instead, we tested two similar products with comparable ingredients and protein source (mustard cake), both being representative of fish feed used in Bangladesh.35 In conclusion, the presence of terpenes and terpene-like compounds in the flesh of tilapia and pangasius demonstrates that the two fish species acquire terpenes from external sources because fish do not produce terpenes. Since concentrations of terpenes dissolved in the water were below the detection limit, terpenes in the fish did most likely originate from ingestion of particulate matter, being feed or plankton. However, if tilapia and pangasius can accumulate dissolved terpenes, as shown for 2-MIB and geosmin in trout (concentration factor of 200- to 400-fold in the flesh relative to the ambient water),36 dissolved terpenes in the pond water might have contributed to the terpene content in the two fish species. The higher number of identical terpenes in fish and feed, than in fish and phytoplankton, as well as the estimate on accumulation of limonene as proxy for the uptake of terpenes from feed, may indicate that fish feed was a more abundant source of terpenes than phytoplankton. Unfortunately, human odor threshold levels are only available for a few of the terpenes detected in the fish, and thus, it is difficult to conclude which of the terpenes potentially might have contributed to the unpalatable taste and flavor of the fish, as observed by the Bangladeshi consumer panel.2



(2) Petersen, M. A.; Alam, M. A.; Rahman, M. M.; Ali, M. L.; Mahmud, S.; Schlüter, L.; Jørgensen, N. O. G. Geosmin off-flavour in pond-raised fish in Southern Bangladesh and occurrence of potential off-flavour producing organisms. Aquac. Environ. Interact. 2014, 5, 107−116. (3) Schrader, K. K.; Dennis, M. E. Cyanobacteria and earthy/musty compounds found in commercial catfish (Ictalurus punctatus) ponds in the Mississippi Delta and Mississippi-Alabama Blackland Prairie. Water Res. 2005, 39, 2807−2814. (4) Zimba, P. V.; Schrader, K. K.; Hyldig, G.; Strobel, B. W.; Jørgensen, N. O. G. Evaluation of geosmin and 2-methylisoborneol off-flavour in smoked rainbow trout fillets using instrumental and sensory analyses. Aquacult. Res. 2012, 43, 149−153. (5) Jones, B.; Fuller, S.; Carton, A. G. Earthy-muddy tainting of cultured barramundi linked to geosmin in tropical northern Australia. Aquac. Environ. Interact. 2013, 3, 117−124. (6) Selli, S.; Prost, C.; Serot, T. Odour-active and off-odour components in rainbow trout (Oncorhynchus mykiss) extracts obtained by microwave assisted distillation-solvent extraction. Food Chem. 2009, 114, 317−322. (7) Selli, S.; Rannou, C.; Prost, C.; Robin, J.; Serot, T. Characterization of aroma-active compounds in rainbow trout (Oncorhynchus mykiss) eliciting an off-odor. J. Agric. Food Chem. 2006, 54, 9496− 9502. (8) Deng, X.; Xie, P.; Qi, M.; Liang, G.; Chen, J.; Ma, Z.; Jiang, Y. Microwave-assisted purge-and-trap extraction device coupled with gas chromatography and mass spectrometry for the determination of five predominant odors in sediment, fish tissues, and algal cells. J. Chromatogr. A 2012, 1219, 75−82. (9) Pattanaik, B.; Lindberg, P. Terpenoids and Their Biosynthesis in Cyanobacteria. Life 2015, 5, 269−293. (10) Leenheer, J. A.; Noyes, T. I.; Rostad, C. E.; Davisson, M. L. Characterization and origin of polar dissolved organic matter from the Great Salt Lake. Biogeochemistry 2004, 69, 125−141. (11) Johnsen, P. B.; Lloyd, S. W. Influence of fat content on uptake and depuration of the off-flavor 2-methylisoborneol by channel catfish (Ictalurus punctatus). Can. J. Fish. Aquat. Sci. 1992, 49, 2406−2411. (12) Auffret, M.; Yergeau, É.; Pilote, A.; Proulx, É.; Proulx, D.; Greer, C. W.; Vandenberg, G.; Villemur, R. Impact of water quality on the bacterial populations and off-flavours in recirculating aquaculture systems. FEMS Microbiol. Ecol. 2013, 84, 235−247. (13) Guttman, L.; van Rijn, J. Identification of conditions underlying production of geosmin and 2-methylisoborneol in a recirculating system. Aquaculture 2008, 279, 85−91. (14) Lu, K.; Jin, C.; Dong, S.; Gu, B.; Bowen, S. H. Feeding and control of blue-green algal blooms by tilapia (Oreochromis niloticus). Hydrobiologia 2006, 568, 111−120. (15) Moreira, N.; Soares, S.; Valente, L. M. P.; Castro-Cunha, M.; Cunha, L. M.; Guedes de Pinho, P. Effect of two experimental diets (protein and lipid vegetable oil blends) on the volatile profile of Senegalese sole (Solea senegalensis Kaup, 1858) muscle. Food Chem. 2014, 153, 327−333. (16) Petersen, M. A.; Hyldig, G.; Strobel, B. W.; Henriksen, N. H.; Jørgensen, N. O. G. Chemical and sensory quantification of geosmin and 2-methylisoborneol in rainbow trout (Oncorhynchus mykiss) from recirculated aquacultures in relation to concentrations in basin water. J. Agric. Food Chem. 2011, 59, 12561−12568. (17) Höckelmann, C.; Becher, P. G.; Von Reuß, S. H.; Jüttner, F. Sesquiterpenes of the geosmin-producing cyanobacterium Calothrix PCC 7507 and their toxicity to invertebrates. Z. Naturforsch., C: J. Biosci. 2009, 64, 49−55. (18) Guillén, M. D.; Errecalde, M. C. Volatile components of raw and smoked black bream (Brama raii) and rainbow trout (Oncorhynchus mykiss) studied by means of solid phase microextraction and gas chromatography/mass spectrometry. J. Sci. Food Agric. 2002, 82, 945− 952. (19) Vidal, N. P.; Manzanos, M. J.; Goicoechea, E.; Guillén, M. D. Farmed and wild sea bass (Dicentrarchus labrax) volatile metabolites: a

AUTHOR INFORMATION

Corresponding Author

*Department of Plant and Environmental Sciences, Section of Microbial Ecology and Biotechnology, University of Copenhagen, Thorvaldsensvej 40, DK-1871 Frederiksberg, Denmark. Phone: +45-35332625. E-mail: [email protected]. ORCID

Niels O. G. Jørgensen: 0000-0002-3554-6906 Funding

We thank the Regional Fisheries & Livestock Development Component in Barisal, Bangladesh, and the Danish International Development Assistance (Danida)/Government of Bangladesh, Barisal, Bangladesh, for providing funding for parts of this study. Funding by Innovationsfonden in Denmark (Grant 050-00008B) is also acknowledged. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We are grateful to Dr. Md. Lokman Ali and Dr. Md. Ariful Alam at Patuakhali Science and Technology University, Bangladesh, for their efforts in providing fish for the study.



REFERENCES

(1) Gutierrez, R.; Whangchai, N.; Sompong, U.; Prarom, W.; Iwami, N.; Itayama, T.; Nomura, N.; Sugiura, N. Off-flavour in Nile tilapia (Oreochromis niloticus) cultured in an integrated pond-cage culture system. Maejo Int. J. Sci. Technol. 2013, 7 (1), 1−13. 3735

DOI: 10.1021/acs.jafc.7b00497 J. Agric. Food Chem. 2017, 65, 3730−3736

Article

Journal of Agricultural and Food Chemistry comparative study by SPME-GC/MS. J. Sci. Food Agric. 2016, 96, 1181−1193. (20) Varlet, V.; Knockaert, C.; Prost, C.; Serot, T. Comparison of Odor-Active Volatile Compounds of Fresh and Smoked Salmon. J. Agric. Food Chem. 2006, 54, 3391−3401. (21) Grigorakis, K.; Taylor, K. D. A.; Alexis, M. N. Organoleptic and volatile aroma compounds comparison of wild and cultured gilthead sea bream (Sparus aurata): Sensory differences and possible chemical basis. Aquaculture 2003, 225, 109−119. (22) van der Ploeg, M. Testing Flavor Quality of Preharvest Channel Catfish; Southern Regional Aquaculture Center Publication; Southern Regional Aquaculture Center, Stoneville, MS, 1992; p 431. (23) Tucker, C. S. Off-Flavor Problems in Aquaculture. Rev. Fish. Sci. 2000, 8, 45−88. (24) Höckelmann, C.; Jüttner, F. Off-flavours in water: Hydroxyketones and β-ionone derivatives as new odour compounds of freshwater cyanobacteria. Flavour Fragrance J. 2005, 20, 387−394. (25) Tellez, M. R.; Schrader, K. K.; Kobaisy, M. Volatile components of the cyanobacterium Oscillatoria perornata (Skuja). J. Agric. Food Chem. 2001, 49, 5989−5992. (26) Tsuchiya, Y.; Matsumoto, A. Characterization of Oscillatoria f. granulata producing 2-methylisoborneol and geosmin. Water Sci. Technol. 1999, 40, 245−250. (27) Agger, S. A.; Lopez-Gallego, F.; Hoye, T. R.; Schmidt-Dannert, C. Identification of sesquiterpene synthases from Nostoc punctiforme PCC 73102 and Nostoc sp. strain PCC 7120. J. Bacteriol. 2008, 190, 6084−6096. (28) Nielsen, J. L.; Klausen, C.; Nielsen, P. H.; Burford, M.; Jørgensen, N. O. G. Detection of activityamong uncultured Actinobacteria in a drinking water reservoir. FEMS Microbiol. Ecol. 2006, 55, 432−438. (29) Dickschat, J. S. Bacterial terpene cyclases. Nat. Prod. Rep. 2016, 33, 87−110. (30) Rabe, P.; Citron, C. A.; Dickschat, J. S. Volatile terpenes from actinomycetes: A biosynthetic study correlating chemical analyses to genome data. ChemBioChem 2013, 14, 2345−2354. (31) Yamada, Y.; Kuzuyama, T.; Komatsu, M.; Shin-Ya, K.; Omura, S.; Cane, D. E.; Ikeda, H. Terpene synthases are widely distributed in bacteria. Proc. Natl. Acad. Sci. U. S. A. 2015, 112, 857−862. (32) Pollak, F. C.; Berger, R. G. Geosmin and related volatiles in bioreactor-cultured Streptomyces citreus CBS 109.60. Appl. Environ. Microbiol. 1996, 62, 1295−1299. (33) Vasta, V.; Priolo, A. Ruminant fat volatiles as affected by diet. A review. Meat Sci. 2006, 73, 218−228. (34) Priolo, A.; Cornu, A.; Prache, S.; Krogmann, M.; Kondjoyan, N.; Micol, D.; Berdagué, J. L. Fat volatiles tracers of grass feeding in sheep. Meat Sci. 2004, 66, 475−481. (35) Mamun-Ur-Rashid, M.; Belton, B.; Phillips, M.; Rosentrater, K. Improving Aquaculture Feed in Bangladesh: From Feed Ingredients to Farmer Profit to Safe Consumption; World fish: Penang, Malaysia, 2013; working paper, p 34. (36) Howgate, P. Tainting of farmed fish by geosmin and 2-methyliso-borneol: a review of sensory aspects and of uptake/depuration. Aquaculture 2004, 234, 155−181. (37) Davies, N. W. Gas chromatographic retention indices of monoterpenes and sesquiterpenes on methyl silicon and Carbowax 20M phases. J. Chromatogr. A 1990, 503, 1−24. (38) Högnadóttir, Á .; Rouseff, R. L. Identification of aroma active compounds in orange essence oil using gas chromatography− olfactometry and gas chromatography−mass spectrometry. J. Chromatogr. A 2003, 998, 201−211. (39) Buttery, R. G.; Ling, L. C.; Light, D. Tomato Leaf Volatile Aroma Components. J. Agric. Food Chem. 1987, 35, 1039−1042. (40) Goodner, K. L. Practical retention index models of OV-101, DB1, DB-5, and DB-Wax for flavor and fragrance compounds. LWT Food Sci. Technol. 2008, 41, 951−958. (41) Chida, M.; Sone, Y.; Tamura, H. Aroma characteristics of stored tobacco cut leaves analyzed by a high vacuum distillation and canister system. J. Agric. Food Chem. 2004, 52, 7918−7924.

(42) Acree, T.; Arn, H. Flavornet. http://www.flavornet.org/ (accessed Dec 6, 2016). (43) Leffingwell & Associates Odor Threshold Database. http:// www.leffingwell.com/odorthre.htm/ (accessed Dec 6, 2016). (44) The Good Scents Company Database. http://www. thegoodscentscompany.com/ (accessed Dec 6, 2016).

3736

DOI: 10.1021/acs.jafc.7b00497 J. Agric. Food Chem. 2017, 65, 3730−3736