Extraction and Detection of Geosmin and 2-Methylisoborneol in Water and Fish using High-Capacity Sorptive Extraction Probes and GC-MS

Summary

This protocol presents a fully automated workflow for the extraction of geosmin and 2-methylisoborneol from water and lipid-rich fish tissues. The method allows for the early detection of these molecules before they reach odor thresholds. Representative data from an aquaculture setting are provided.

Abstract

Geosmin and 2-methylisoborneol are volatile organic compounds of microbial origin that frequently occur in recirculating aquaculture systems and confer unpleasant odors and flavors to fish therein, even at extremely low concentrations. Existing analytical methods for these compounds in fish struggle to achieve enough sample throughput and are oftentimes not automated, requiring extensive hands-on preparation. We have developed methods for analyzing geosmin and 2-methylisoborneol from water and fish tissues, respectively, using a metal probe bearing a high-capacity sorptive extraction phase. A robust and quantitative detection was achieved from both water and fish tissues, with minimum detection limits of 1.2 ng/L for geosmin and 1.1 ng/L for 2-methylisoborneol from water and 18.1 ng/kg and 13.6 ng/kg, respectively, from fish tissue. These concentrations were well below human detection thresholds. Hence, we concluded that these methods could detect geosmin and 2-methylisoborneol before they became perceptible to humans. In contrast to other high-capacity sorptive extraction techniques, such as stir-bar sorptive extraction, this technique is fully automatable, dramatically improving throughput and precluding manual handling errors, while the durable design of the probe precludes breakage, such as is common with solid phase microextraction (SPME) fiber. This method is suitable for odorant monitoring in all water quality settings, as well as for the detection of off-flavor compounds in solid and semi-solid matrices.

Introduction

Geosmin (GSM) and 2-methylisoborneol (MIB) occur naturally in water and soils as by-products of microbial metabolism^1,2. While not generally hazardous to environmental or human health, they impart unpleasant 'earthy' and 'muddy' odors and flavors to water, which are perceptible even at extremely low concentrations, as low as 10 and 20 ng/L for GSM and MIB, respectively^3,4. As such, the presence of these odorants reduces the perceived quality of water and is highly undesirable in most contexts. In inland waterways, odorants reduce the appeal to visitors and may elicit complaints to environmental authorities⁵, while in drinking water, odorants impart an unpleasant smell and taste that consumers find unacceptable⁶. Indeed, many countries have imposed regulatory limits on odorant levels in drinking water, with China stipulating 10 ng/L as the maximum GSM concentration permissible⁷.

Recirculating aquaculture systems (RASs) are becoming increasingly important in the aquaculture industry as a means of fish production. These systems operate on land and continuously reuse water, thereby reducing environmental impact, but extensive filtering of RAS water is required to maintain optimal conditions for fish rearing. Moving bed biofilters become colonized by microbes due to the abundance of nutrients in fish feces and uneaten feed⁸. These microbes are beneficial for the removal of nitrogenous waste and excess nutrients; however, some species may produce GSM, MIB, and other odorants⁴. These odorants then leach into the water and are absorbed into the flesh of the fish, where they can impart unpleasant odor and flavor characteristics that consumers find objectionable. Odor threshold values in fish flesh are species-dependent, but values for Atlantic salmon, a common species in RASs, are 44 - 500 ng/kg for GSM and >900 ng/kg for MIB⁴. Hence, RAS operators should regularly monitor GSM and MIB concentrations in their systems to ensure that these do not reach problematic levels⁴. Affected fish can be purged of off-flavor compounds pre-harvest to recover the quality of the fillet; however, this is an expensive process requiring the facility to stop recirculation of water and instead flow new water through the system for 10 -14 days⁹. Given that GSM and MIB present a persistent and economically costly problem across a wide range of industries, it makes sense to reassess and improve analytical methods for detection. Ideally, RAS operators should be able to monitor both RAS water to identify increasing GSM and MIB concentrations before they become problematic, and fish tissues to determine whether and to what extent GSM and MIB levels have affected product quality.

Most methods for the detection of GSM and MIB use gas chromatography (GC) with either a mass spectrometer (MS) or flame ionization detector (FID) for separation and detection, but how odorants are initially extracted from the water matrix varies greatly, with many methods published¹⁰. That such a variety of methods have been trialed and, indeed, are still in routine use speaks to the fact that each has significant drawbacks. These drawbacks include a requirement for large sample size and large volumes of solvent as in liquid-liquid extraction (LLE), low sensitivity and water interference as in headspace-based methods, or extensive issues with carryover and system contamination as in purge and trap (P and T).

Sorptive extraction covers a range of extraction techniques wherein a solid sorptive phase can be used to extract odorants from the headspace above a sample or from the sample itself via direct immersion¹⁰. These techniques are compatible with smaller sample volumes than other techniques, do not require solvents, and are not typically severely impacted by water interference. However, the most common form of sorptive extraction, solid-phase microextraction (SPME), uses only a small volume of sorptive phase with low capacity for analytes, which limits extraction efficiency. Additionally, the brittle fibers involved can become damaged or broken¹¹, interrupting workflows. Stir-bar sorptive extraction (SBSE) overcomes these difficulties with a larger phase volume and more durable design but is not readily automatable, requiring manual transfer of stir-bars into samples and from samples into thermal desorption tubes¹². In the context of routine water screening, where high throughput is crucial, this lack of automation is a significant barrier.

The goal of our method is to improve upon the aforementioned methods with respect to sample throughput and the overall sorptive extraction capacity. Thus, we developed a technique using high-capacity sorptive extraction (HCSE) probes, combining the high analyte capacity and robustness of SBSE with the phase selection and automation of SPME for the quantitative, high-throughput analysis of GSM and MIB. With some modifications, the technique can be applied both to water samples and to samples of homogenized fish tissue. Here, we describe the protocol in full and present validation data demonstrating that it is reproducible, linear, and highly sensitive. Although aquaculture samples are used for demonstration purposes here, the protocol is suitable for monitoring odorants in a wide range of sample types, including waters, soils, and various foodstuffs. Thanks to full automation, the protocol is particularly beneficial where sample numbers are expected to be high.

Protocol

The fish samples were collected from animals that were already slaughtered for human consumption. Therefore, no ethical clearance was required.

1. Preparation of calibration and internal standards

1. Calibration standard preparation
   1. Prepare stock solution C1 by adding 10 µL of geosmin and 2-methylisoborneol solution (100 µg/mL) to a 100 mL volumetric flask and filling to volume with LC-MS grade methanol. The resulting solution will contain each analyte at 10 µg/L (ppb).
   2. Prepare calibration standards by adding the C1 stock solution in the volume indicated in Table 1 to a 100 mL volumetric flask and filling to volume with deionized water.
   3. Make fresh calibration standards daily, as these are extremely perishable.
      NOTE: It is important to prepare these stock solutions fresh every day.
2. Internal standard preparation
   1. Prepare stock solution (I1) by adding 10 mg of 2-isopropyl-3-methoxypyrazine (IPMP) (>98% pure) to a 1 L volumetric flask filled to volume with LC-MS grade methanol making a 10 mg/L stock.
   2. Pipette 100 µL of solution I1 to a 100 mL volumetric flask and fill to volume with deionized water to give solution I2 (10 µg/L). Spike 50 µL into every sample vial to reach a final concentration of 100 ng/L.

2. Water analysis

1. Sample preparation
   1. Place one 20 mL vial for each water sample being analyzed in the sample tray.
   2. Using a calibrated scale weigh out and place 2.5 g of NaCl into each 20 mL crimp top sample vial.
      NOTE: Adding salt enhances the transfer of the MIB and geosmin from the water to the headspace.
   3. Using a pipette or bottle top dispenser, add 5 mL of the sample water to each 20 mL crimp top sample vial.
   4. Pipette 50 µL of the internal standard solution I2 to each sample, calibration standard, and any QA/QC samples, including blanks.
   5. Place the cap on top of the vial and seal it with a vial crimping tool to minimize analyte loss.
   6. Only process 1-2 samples at a time and quickly seal each vial once complete to minimize analyte loss of volatile analytes.
2. Analyte extraction (water)
   1. Place the sample tray on the autosampler tray holder.
   2. To prepare an instrument method, open the instrument program, and click on the Method Editor.
      1. Create a new method by entering the following parameters into the method editor (leaving other parameters at their default values):
      2. In 'Pre-sampling', set the Vial Agitation Time to 10 min to allow analytes to partition from the water to the headspace.
      3. In 'Sampling', set the Incubation Temperature to 65 °C, incubation time to 30 min, and Agitator speed to 400 rpm.
      4. Ensure 'wash HiSorb probe' is Enabled; otherwise, the residual sample matrix may contaminate the system.
      5. In 'Probe desorption', set the Desorption time to 15 min and the Desorption temperature to 270 °C.
      6. In the Trap settings, set the Trap low temperature to 25 °C and 'split flow' to 8 mL/min.
   3. Run the autosampler method in conjunction with the GC-MS method by pressing the Start button on each of the respective programs.

3. Fish tissue analysis

1. Sample preparation
   1. Place one 20 mL vial for each fish sample being analyzed in the sample tray.
   2. Homogenize the fish tissue using a commercially available food processor.
   3. Using a calibrated scale, weigh 1.00 g of homogenized fish tissue into each vial.
   4. Pipette 5 mL of saturated NaCl solution into each vial.
   5. Pipette 100 µL of the internal standard solution I2 to each sample, calibration sample, and any QA/QC samples.
   6. Place a cap on top of the vial and seal with the vial crimping tool.
   7. Prepare only 1-2 samples at a time to reduce the loss of semi-volatile and volatile compounds.
2. Analyte extraction (fish)
   1. Place the sample tray on the autosampler tray holder.
   2. Prepare an instrument method as follows (leaving other parameters at their default values). 
      NOTE: pre-sampling vial agitation time and incubation temperature differ from Section 2.2.2.
      1. In 'Pre-sampling', set vial agitation time to 20 min.
      2. In 'Sampling', set incubation temperature to 80 °C, Incubation time to 30 min, and Agitator speed to 400 rpm.
      3. Ensure 'wash HiSorb probe' is enabled, otherwise, residual sample matrix may contaminate the system.
      4. In 'Probe desorption,' set the desorption time to 15 min and the Desorption temperature to 270 °C.
      5. In trap settings, set the trap low temperature to 25 °C and split flow to 8 mL/min.
         NOTE: A longer, hotter pre-sampling step is required here due to the higher affinity of MIB and GSM for lipid-rich fish tissues versus water.
   3. Run the autosampler method in conjunction with GC-MS method by pressing the Start button on each of the respective programs.

4. Separation and detection using GC-MS

1. Use the same GC-MS parameters for water and fish tissue methods.
2. Ensure the GC is fitted with a 5MS, 30 m x 0.25 mm x 0.25 µm column and is supplied with ultra-pure helium carrier gas. Set the carrier flow rate to 2 mL/min.
3. Prepare a GC oven program as follows: An initial temperature of 60 °C, held for 3 min, then a temperature ramp of 10 °C/min to 100 °C, then 20 °C/min to 190 °C, then 30 °C/min to 280 °C, with a final hold time of 2 min. The total run time is set to 16.5 min.
4. Set the transfer line between GC and MS to 280 °C and the ion source and the quadruple to 250 °C and 200 °C, respectively.
5. For MS detection in scan mode, use a scan range of m/z 50-350.
6. For selected ion monitoring (SIM) operation, use quantification and confirming ions respectively of m/z 95 and 107 for MIB, m/z 137 and 152 for IPMP, and m/z 112 and 55 for GSM.
   NOTE: Many chromatography software packages feature auto peak selection, but it is highly recommended to manually review the peak selection for errors.

Results

Method optimization

Various experiments have been conducted to ensure the method is optimized, including a carryover assessment, optimization of probe desorption time, an internal standard assessment, a probe phase assessment, an assessment of linearity and repeatability, and calculation of the method detection limit.

Carryover assessment

We began our experiment with a probe desorption temperature of 250 °C. Calibration standards were prepared, as shown in Table 1. To assess carryover, we analyzed GSM and MIB standards at 20 ng/L and then proceeded with, in order, a run of the GC-MS parameters without injecting anything to the column (column blank), a desorption of the focusing trap without transferring anything from the injector (trap blank), a heating of the injector without inserting a probe (injector blank), and a desorption of the probe without prior sampling (probe blank). Less than 1% carryover was observed in the column blank, trap blank, and injector blank, but in the probe blank, carryover was at 9.35% and 3.77% for GSM and MIB, respectively (Table 2). Hence, we determined that there was a significant carryover of GSM on the PDMS probe under the conditions used.

Desorption time optimization

Having identified carryover on the PDMS probe, we sought to resolve this by adjusting the probe desorption temperature. A higher probe desorption temperature would be expected to increase the rate of analyte desorption from the probe such that desorption would be more complete by the end of the 15 min probe desorption time. However, a higher temperature could also promote analyte breakdown and, therefore, negatively affect sensitivity. Higher temperatures can also damage sorptive materials, so temperatures over 280 °C must be avoided. Carryover of GSM was significant at the 250 °C probe desorption temperature, but this reduced substantially to well below the 5% acceptability threshold when 270 °C was used instead (Figure 1). This improvement was achieved with no loss of sensitivity, with peak areas produced for the initial analysis being no smaller at 270 °C than for lower probe desorption temperatures (Figure 1). This indicates that any increase in analyte breakdown is negligible.

Use of internal standard

The compound 2-isopropyl-3-methoxypyrozine has been widely employed as an internal standard during the detection of GSM and MIB¹³. We compared the calculated concentrations of GSM and MIB with the known spiked concentrations, both with and without internal standard correction. Recovery without internal standard correction was poor, with calculated concentrations in 60% of the known spiked concentrations (Figure 2). In contrast, with internal standard correction applied, recovery was very good, in the region of 100%. Thus, we applied internal standard correction in all future analyses.

Phase selection

In SPME analyses, carbon wide range (CWR) and/or divinyl benzene (DVB) are often added to PDMS to improve sensitivity, but this increases cost. We sought to compare PDMS-only probes with triple-phase (DVB/CWR/PDMS) probes, considering that the increased phase volume of the probe is likely to improve sensitivity versus SPME fibers. Recovery data for PDMS and triple-phase probes (Figure 3) illustrate broadly similar performance between the two probe types, with recoveries slightly closer to 100% with the PDMS probes. Thus, given the slightly superior performance and reduced costs, we performed all future analyses with PDMS probes. It should be noted that since the probe collects analytes from the headspace and not via direct contact with the sample matrix, the kinetics of GSM / MIB partitioning to the phase material should not differ between water and fish samples.

Method validation

A calibration series covering six concentrations of analytes in water from 1 to 100 ng/L was analyzed (Figure 4). Excellent linearity was achieved, with r² of 0.9999 for GSM and 0.9999 for MIB. The standard error of regression was 0.43 and 0.37 for GSM and MIB, respectively. This indicates that the method is suitable for quantitative analyses over the range assessed. We determined repeatability from eight replicates of laboratory standards containing GSM and MIB at 40 ng/L in water. RSD values were 6.3% for GSM and 4.8% for MIB, demonstrating remarkable reproducibility of results. These replicates were analyzed in three separate batches on different days, further demonstrating the precision of the method. Hence, HCSE with GC-MS analysis is a quantitative and reliable method.

Minimum detection limits (MDL)

The MDL was calculated as per the U.S. EPA protocol¹⁴ using spiked water (n=10) or spiked fish samples (n=11). The method for extraction and detection of GSM and MIB in water produced detection limits of 1.2 ng/L for GSM and 1.1 ng/L for MIB. These MDLs were far below the human odor threshold values of approximately 10 and 20 ng/L, respectively. Thus, the proposed method can confidently and quantitatively detect the accumulation of these odorous molecules before they reach human sensory thresholds. Detection limits for the extraction and detection of GSM and MIB from fish produced limits of 18.1 ng/kg for GSM and 13.6 ng/kg for MIB.

Analysis of water samples from aquaculture

We analyzed water samples taken from seven recirculating aquaculture systems, with five replicates (system 3) or three replicates (all other systems) assessed for each. All but one of the recirculating aquaculture systems analyzed contained geosmin and MIB at concentrations above our MDL (Figure 5), indicating a potential issue with water quality. However, we noted that in all cases, MIB was well below the human odor threshold of 20 ng/L, while only one system exceeded the GSM odor threshold of 10 ng/L (system 1, GSM concentration = 12.69 ng/L). While we did detect GSM and MIB in system 3, we note that the concentration for MIB falls below our calculated MDLs, and therefore, the concentration cannot be confidently quantified. The method demonstrated impressive precision across all systems despite the use of real-world water samples with very little variation in calculated concentration among replicate samples in each system (Figure 5).

Analysis of fish tissues

Fish tissue, rich in lipids, imparts a strong matrix effect that impedes extraction of GSM and MIB during this method. For that reason, subtle changes are made in the method to increase the extraction efficiency, including a higher incubation temperature (80 °C) and longer incubation time (20 min) compared to the water extraction method. Fish tissues were previously analyzed to determine background concentrations GSM and MIB. Those same samples were then spiked with a known amount of GSM and MIB and analyzed in triplicate to determine percent recovery and precision (Figure 6).

Figure 1: Effect of probe desorption temperature on carryover. (A) A laboratory standard containing GSM and MIB at 20 ng/L was analyzed at one of three probe desorption temperatures in triplicate, with mean peak areas. Each probe was subsequently desorbed twice more at the initial temperature to assess carryover with results in (B) (GSM) and (C) (MIB). Carryover is expressed as peak area as a percent of initial peak area. Error bars represent standard deviation across three replicates. Please click here to view a larger version of this figure.

Figure 2: Internal standard usage. Comparison of corrected versus uncorrected data recovery data for analytical standards at 15 or 40 ng/L, with extraction by PDMS probes only. Sample concentrations were corrected with the use of IPMP as an internal standard. Please click here to view a larger version of this figure.

Figure 3: Analyte recovery. Recovery data for analytical standards of GSM (left) and MIB (right) at 15 ng/L, with extraction by PDMS-only or triple-phase (PDMS, DVB, CWR) probes. Please click here to view a larger version of this figure.

Figure 4: Analytical calibration curves. Linear regressions of calibration curves for GSM (solid line, r² of 0.9999; S=0.37) and MIB (dashed line, r² of 0.9999; S=0.37). Please click here to view a larger version of this figure.

Figure 5: Aquaculture water application. Concentrations of (A) GSM and (B) MIB in water samples taken from seven aquaculture systems (n ≥ 3 for each system). Please click here to view a larger version of this figure.

Figure 6: Fish tissue application. Concentrations of GSM and MIB from two fish samples spiked with GSM and MIB (F1 and F2). Please click here to view a larger version of this figure.

Solution name	Stock C1 volume added	Total volume	Resultant concentration
Stock C1	10ul	100ml	10 ug/ml
Cal 1	10 ul	100 ml	1 ng/L
Cal 2	20 ul	100 ml	2ng/L
Cal 5	50 ul	100 ml	5 ng/L
Cal 10	100 ul	100 ml	10 ng/L
Cal 20	200 ul	100 ml	20 ng/L
Cal 50	500 ul	100 ml	50 ng/L
Cal 100	1000 ul	100 ml	100 ng/L
Cal 200	2000 ul	100ml	200 ng/L

Table 1: Calibration standards. Geosmin and 2-MIB standards all made from CRM47525 geosmin and 2-methylisoborneol solution (100 µg/mL) in methanol that was diluted into a stock solution C1 (10 µg/ml) in methanol.

Analysis	Carryover (%)
	GSM	MIB
GC Run	0.04	0.08
Trap desorption	0.09	0
Inlet desorption	0	0
Probe redesorption	9.35	3.77

Table 2: Probe carryover assessment. Carryover following extraction, desorption, and GC-MS analysis of a laboratory standard containing GSM and MIB at 20 ng/L, with internal standard IPMP at 30 ng/L. Descriptions of analyses are given in 2.2.1.

Discussion

GSM and MIB are off-flavor compounds occurring in a wide range of water monitoring settings, and demand for testing capacity is high. Furthermore, GSM and MIB are perceptible to humans at trace levels. An ideal analytical system must, therefore, combine high sensitivity with high throughput to produce high volumes of quality data. We have demonstrated in previous sections that extraction with HCSE probes provides sensitivity that is in excess of what is required.

The robust design of HCSE probes precludes any breakages, an issue that can interrupt SPME workflows. Following method optimization, we experienced no difficulties with system contamination or carryover, which are common issues with P and T methodologies. The biggest contribution to this method's high throughput, however, undoubtedly came from the autosampler platform used. With this system, analysis continued around the clock, with manual intervention required to supply samples to the sample tray only. This is in contrast to other high-capacity sorptive extraction methods, such as SBSE, which require manual operation. While SPME is commonly fully automated, throughput is limited in that incubation of the next sample can, typically, only begin once the previous sample has been injected into the GC column. Assuming that the GC cycle time is short, the GC may, therefore, sit idle while the next sample completes extraction. With our method, multiple samples can undergo pre-sampling and extraction simultaneously such that GC downtime is minimized. Hence, we were able to process 1.8 samples per hour per system.

We have demonstrated MDLs of 1.2 and 1.1 ng/L for GSM and MIB, respectively, thus far surpassing the sensitivity required. While other techniques such as P and T, LLE, SBSE, and SPME may be able to reach lower MDLs under certain circumstances¹¹, these increases in sensitivity are of limited practical value and do not justify losing the advantages of HCSE probes outlined above. Further, some of these techniques require large sample volumes, with attendant increases in waste generation and transport costs. For LLE, sample volumes are typically in the range of 200 - 500 mL¹⁵,^16,17, while for SPME, SBSE, and P and T, typical values are 10 - 100 mL^{18,19,20,21,22}. Excellent reproducibility was observed with RSD values of 6.34% and 4.82% for GSM and MIB, respectively. These values are broadly similar to those found in other methods¹¹, and, in particular, surpass the RSDs in published SPME methods^23,24. Again, our low RSD values were achieved with sample volumes of just 5 mL, lower than the vast majority of published extraction techniques. Thus, we are confident that HCSE probes provide sufficient sensitivity and reproducibility along with a wide suite of benefits that combine the best aspects of all prominent alternatives into a single simple, fully automated and sustainable method.

Geosmin and 2-methylisoborneol are lipophilic compounds, which presents a challenge to scientists trying to extract those compounds from the lipid-rich tissue for analysis. We, therefore, adjusted our pre-sampling conditions, using a longer and hotter pre-sampling procedure, to drive more analytes from the sample matrix to the headspace. Having made these adjustments, we determined that the method worked well for analyzing GSM and MIB in fish tissues. It should be noted that for samples with a very high lipid content, it may be necessary to reduce the mass of sample used, e.g., to 0.50 g. This is to avoid lipids forming a complete seal on top of the water that prevents analytes from reaching the headspace.

This method uses equilibrium principles to extract analytes from the headspace of a sample. Although we use water and fish samples from aquaculture here, we believe that the method is applicable to odorant extraction from any solid or liquid matrix that can be homogenized and accommodated within a 20 mL sampling vial. Example applications include monitoring of environmental waters and soils, testing of drinking water, and testing of foods such as fruits and vegetables, or non farm-grown aquatic life. In farming, GSM and MIB are often found in damp soils and can impart off-flavors to crops grown therein², therefore both soils and crops could be tested via our method to monitor odorant levels as a form of quality control and/or to diagnose the source of malodorous complaints. Many environmental waters, such as rivers and lakes experience seasonal spikes in odorant concentration due to algal blooms²⁵, and this method offers a high-throughput, highly sensitive means of monitoring the extent of these spikes. This method is also suitable for academic research, for example, to measure the rate of odorant production by microorganisms²⁶ or to examine the relationship between odorant concentrations in water and in animal tissues²⁷. Other research groups have used HCSE process to extract from a range of sample types such as alcoholic drinks²⁸, mutton meat²⁹, olive oil³⁰, human cell culture media³¹and pistachios³². These studies targeted other analytes or are non-targeted profiling studies and therefore were not optimized for the analysis of GSM and MIB, but are presented here as they offer general insights on preparing various sample types for HCSE probes extraction.

While the water method described is likely to be applicable to all water or water-based samples with minimal modification, application to other solid or semi-solid samples may require some matrix-specific alteration to sample preparation (section 3.1) and analyte extraction (section 3.2). While it would not be possible to cover every sampling scenario here, we offer the following considerations. (1) Solid samples should be homogenized to a fine powder via grating, grinding or blending to ensure representativeness and to provide the maximum possible surface area for egress of analytes. (2) Solid samples with a low lipid content may not require the addition of water or salt - i.e. it may be sufficient to weigh 1.00 g into a vial and cap. This will allow for a more direct transfer of analytes from the sample to the headspace, without an intermediate transfer from the sample matrix to the added water. However, this will reduce the efficacy of stirring. Analysts should perform experiments to determine the optimum sample preparation technique for their sample matrix. (3) Different matrices will have different affinities for GSM and MIB. Therefore, different incubation temperatures will be required to reach analyte equilibrium in a timely manner. However, some matrices may be chemically altered by heat such that interferences are produced by chemical reactions at high temperatures. Again, experiments are required to determine the optimum incubation temperature for a particular sample type (note: Samples with a high moisture content should never be incubated above 90 °C to avoid vials shattering from the vapor pressure of boiling water).

We have developed methods for the detection of geosmin and 2-methylisoborneol in water and fish tissue. Tests on laboratory standards confirm that the methods are quantitative and reproducible, while application to real-world samples (water from recirculating aquaculture systems, fish tissues derived from such systems) demonstrate performance even in the context of complex, dirty samples. Although we have applied our methods to samples from aquaculture, we suggest that they are suitable for GSM and MIB monitoring in a wide range of sample types, including water and solid or semi-solid materials.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
0.5-10uL pipette	Nichipet Air	A21405581
1000uL pipette tips	VWR	76322-154
100-1000uL pipette	eppendorf	285612
100mL glass flask	VWR	76008-704
10-100uL pipette	Nichipet Air	A20814031
10-100uL Pipette tips	mega	24502-18
10xL Pipette tips	VWR	76322-132
20mL Glass beakers	VWR	10754-698
2-Isopropyl-3-methoxypyrazine	TCI	I0577	Internal standard
40 ml Amber type-1 (Borosilicate) Glass vial	Environmental sampling company	0040-0400-QC	Water Sampling vial
5mL Pipette tip	VistaLab technologies	4058-5102
Balance	VWR	VWR-124B2
Centri 360	Markes	U-CENTRI-1691	Autosampler
Chromeleon Version 7.3	Thermo Scientific		GC-MS Software
Focusing trap	Markes	U-T12ME-2S
Gas Chromatograph, Trace 1310	Thermo Scientific	14800403
GC Column	Thermo Scientific	26098-1420	TG-5MS
Geosmin and 2-Methylisoborneol solution	SIGMA-ALDRICH or MilliporeSigma	CRM47525
Headspace vial, 20 mL	Markes	C-VCC20
HiSorb handle	Markes	C-HH-6
HiSorb injector cap	Markes	C-HSPCCS
HiSorb injector liner	Markes	U-LINER-HISORB
HiSorb probe, PDMS, short version	Markes	H1-XXABC
HiSorb vial caps	Markes	C-HSPCCS
Laboratory metal scoop
Mass Spectrometer, ISQ 7000	Thermo Scientific	ISQ7K-NOVPI
Methanol	VWR chemicals	BDH1135-4LP
Methanol, Optima LC-MS grade	Fisher-scientific	A456-4
O-rings size 201	Markes	U-COV201	For HiSorb injector cap and HiSorb storage
Ovation automatic pipettor (100-5000uL)	VistaLab technologies	213626
Purple Nitrile-Xtra Powder Free Exam Gloves	Kimtech	50602-54
Sodium chloride (salt)	VWR chemicals	BDH9286-2.5K
Switching adapter (automatic pipettor)	CUI INC	SMI10-5
Vial crimping tool	Markes	U-HSVCR
Vial decapper	Markes	U-HSVDC
Weigh boats	Grainger	8AG69