| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Biometrics Dep., Horticulture Research International, Wellesbourne, Warwick, CV35 9EF, UK
b Institute of Grassland and Environmental Research, North Wyke, Okehampton, Devon, EX20 2SB, UK
Corresponding author (ralph.noble{at}hri.ac.uk)
Received for publication March 31, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: DMS, dimethyl sulfide • GC–MS, gas chromatography–mass spectrometry • OC, odor concentration • OU, odor unit • VFA, volatile fatty acid
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Olfactometry is used to measure the odor concentration (OC) of air through the use of a serial diluter, or olfactometer, to present odorous air with odorless air dilutions to a panel of people. Previous studies have used olfactometry to identify the most odorous processes and sources on composting sites (Bidlingmaier, 1992; Perrin and Macauley, 1995). However, olfactometry is costly, time consuming, and incurs delays between sampling and measurement (Hobbs et al., 1995).
Several studies have been aimed at identifying the individual compounds associated with mushroom composting odors using gas chromatography–mass spectrometry (GC–MS) (Miller and Macauley, 1988; Derikx et al., 1990; Duns et al., 1997, 1998a). Compounds shown to be responsible for compost odor included amines, ammonia, organic acids, and, most importantly, sulfur-containing compounds. However, GC–MS analysis and interpretation is impractical on a day-to-day basis due to the large numbers of compounds involved, cost, and the delay between sampling and measurement. However, many of the important odorants can be measured quickly and cheaply using gas detection tubes (Miller et al., 1991; Anonymous, 1997, 1998).
Several studies attempted to relate the OC of pig, cattle, and poultry manures with the concentrations of odorants (Schaefer, 1977; Hobbs et al., 1995, 1999; Misselbrook et al., 1997). No universally applicable relationships were found, either within or between different types of manure.
The goals of the present work were to test the following hypotheses. (i) The OC of mushroom composting emissions can be predicted from their chemical analysis. (ii) The OC and odorant concentrations of these emissions are affected by the type of compost (i.e., pre-wet or Phase I, aerated or unaerated) and different ingredient formulations.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Each of the 11 composting yards was visited twice, with a pair of composting yards visited on each occasion. Pairs of sites were arranged such that each site was visited with a different site on the two sampling occasions, under a number of geographical and compost yard schedule constraints (Table 1). Two replicate samples were collected simultaneously from the pre-wet and Phase I composting areas on each site, so that four odor samples (each consisting of the two simultaneous samples) were assessed on each occasion. While this design was not formally used in the analysis of the results, it provided protection against variation due to differences between occasions, given the constraints both on the number of visits and from the geographical and composting schedule considerations.
Background samples were collected 200 m upwind of the composting sites. Bag samples were transported to the Institute of Grassland and Environmental Research (IGER), North Wyke for GC–MS and olfactometry analyses, which were conducted 24 h after sampling. Analysis using GC–MS was conducted on both sample bags on only one of the two sampling occasions for each site.
Wind speed 2 m above the ground at the position of odor sampling was measured with a vane anemometer (Type 949079; Airflow Developments Ltd., High Wycombe, UK).
Olfactometry
A dynamic dilution olfactometer (Type DTM; Project Research, Amsterdam, the Netherlands) was used to measure OC. This is the number of dilutions of odorous air with odorless air at which 50% of an odor panel can just detect an odor, and is expressed as odor units m-3 (OU m-3) air (Anonymous, 1999). The olfactometer enables presentation of odorous air to panelists at dilutions with clean air of between 22x and 216x. The olfactometer has two sniffing ports and was of the forced-choice type, with odorless air being presented through one port and diluted odorous air through the other (assigned at random). For each presentation, each of six panelists was required to indicate via a keyboard which port emitted the odorous air and the certainty of the choice. Responses were only regarded as being correct if the panelist indicated the correct port and was certain of the choice. To ensure accuracy, each sample was presented twice for up to 20 s within a range of five dilution steps of ascending sample concentration with an inter-stimulus period of more than 5 min. An experienced olfactometer operator selected the dilution range of five steps between 22x and 215x that varied by a factor of two. In addition, each panelist had to respond correctly to the final two (highest) concentrations presented. Detection thresholds of an individual panelist for each series were calculated as the geometric mean, in terms of number of dilutions, of the lowest dilution step at which they gave an incorrect response and the next lowest dilution step from where after they constantly gave a correct response. Panel detection threshold or OC of the sample was calculated as the geometric mean of the individual panelist thresholds. Equipment and procedures used followed current recommendations (Anonymous, 1999). Panelists were selected on the basis of sensitivity to 1-butanol and the calculated variation of that threshold determination from the previous 20 tests with the reference standard. Individual sensitivity expressed as a geometric mean fell between 62 and 246 µg m-3 1-butanol.
Gas Chromatography–Mass Spectrometry Identification and Quantification
Volatile compounds were pre-concentrated from 600-mL odor samples by adsorption onto silica- (Orbo 52; Supelco, Bellefonte, PA) and then carbon- (Orbo 32) based adsorbents. The concentrated odorants were then thermally desorbed from the adsorbents into the GC–MS system for identification and quantification. Chromatographic retention time and mass spectral matching were used to confirm odorant identity. Quantification was performed by using a standard odor identified, containing the same concentrations as that in the pre-concentrated headspace, that was made in a nalophane bag using odorless air and following the same sampling procedure as for the samples taken on-site.
A Hewlett Packard (hp; Stockport, UK) GC–MS system consisting of a 5890 II Series gas chromatograph and a 5972A mass selective detector (MSD II) was used for analysis. A 25-m fused silica (cross-linked methyl siloxane) hp-1 column with an internal diameter of 0.2 mm and a 0.34-µm film with a 1-m, Q plot, deactivated fused silica guard column (internal diameter 0.53 mm), containing a porous polymer of divinylbenzene (Supelco), was used. The flow rate of the eluting gas, helium, was 0.75 mL min-1. The Orbo adsorbents were desorbed with a reverse flow to that used for adsorption (i.e., in through the carbon and out through the silica adsorbents) to improve chromatographic resolution. An Optic temperature programmable injector (Ai Cambridge Ltd., Pampisford, UK) was used to desorb headspace samples from the adsorbents and was initially at 30°C and heated at 16°C s-1 to 250°C. An electronic pressure controller was used to offset peak pressure broadening with increasing GC column temperature. The GC oven conditions were an initial temperature of 40°C, then increased to 220°C at 15°C min-1 and remaining at 220°C for 1 min. The GC–MS interface was at 280°C. The mass spectrometer scanned from 35 to 250 mass units every 0.2 s to give responses in the ng range.
Volatile organic compounds detected by the mass spectrometer were identified using a probability-based matching algorithm and a NIST mass spectral library (National Institute of Standards and Technology, Gaithersburg, MD). Compounds were declared unknown if their matching probability was less than 80 (100 being a perfect match).
Gas Detector Tubes
A Dräger Accuro bellows pump (Drägerwerk, Lübeck, Germany) was used in conjunction with appropriate detector tubes (the measuring range, coefficient of variation, and number of pump strokes used [n] is shown for each tube): acetic acid (5 to 80 µL L-1, ±10 to 15%, n = 3), amines + NH3 (1 to 20 µL L-1, ±30%, n = 1), NH3 (1 to 70 µL L-1, ±10 to 15%, n = 1 to 10; 5 to 100 mL L-1, ±10 to 15%, n = 1), CS2 (3 to 95 µL L-1, ±30%, n = 15), dimethyl sulfide (DMS) (0.2 to 15 µL L-1, ±15 to 20%, n = 20), H2S (0.2 to 6 µL L-1, ±15 to 20%, n = 1; 1 to 200 µL L-1, ±15%, n = 1 to 10), mercaptans (thiols) (0.5 to 5 µL L-1, ±10 to 15%, n = 20), and phenol (1 to 20 µL L-1, ±10 to 15%, n = 20). Detector tubes were used on-site in the same way as sampling odors for collection in nalophane bags, and also on the odor samples in the nalophane bags, 24 h after on-site sampling. The bag samples were also assessed using a Gastec GV-100S gas sampling pump with appropriate detector tubes: NH3 (0.5 to 30 µL L-1, ±25%, n = 2), DMS (0.3 to 6 µL L-1, ±25%, n = 5), H2S (0.3 to 120 µL L-1, ±25%, n = 1 to 10), and mercaptans (0.1 to 8 µL L-1, ±25%, n = 4) (Gastec Corporation, Kanagawa, Japan). Details of the mode of operation of the sampling pumps and detector tubes are given in Anonymous (1997)(1998). The CO2 concentration of the bag samples was measured using a portable infrared absorption meter (Type GM11; Vaisala Oyj, Helsinki, Finland). Two replicate measurements were made for each sampling.
Statistical Analysis
Due to the slight non-orthogonality between the aeration, composting stage, and manure factors, the effects of these factors on OC and each of the gas concentrations could not be assessed using an analysis of variance, but a regression-based approach was possible. Since the variance in OC and gas concentrations increased with the size of the mean, a loge transformation of the data was used. To avoid problems with loge transformed values of zero, 0.375 was added to all values of H2S, DMS, and H2S + DMS concentrations in the analyses. A range of explanatory models was considered, including simple models involving just factor main effects and more complicated models involving two- and three-factor interactions. Improvements in fit due to including extra terms were assessed within a hierarchical structure, and the best-fitting yet parsimonious model was identified for each variable. Predicted mean responses were obtained for the best model for each variable, and back-transformed means calculated.
Mean, maximum, and minimum concentrations were calculated for each of the odorants detected with GC–MS for the pre-wet and Phase I air samples. Standard deviations were considered not to be a useful summary of the data due to the skew distribution for most odorants. The mean values were compared with detection limits given by Devos et al. (1990) to identify important odors. The effects on measurements of the 24-h delay in measurement were assessed by regression of the delayed measurements on the on-site measurements. The relationship between the two different approaches to measurement of gas concentrations was assessed by regressing values obtained from bag samples using gas detector tubes on those measured using GC–MS.
The possible predictive capacity of each of the individual gas concentration variables for OC was assessed using linear regression on loge transformed data. An analysis of parallelism was performed to assess whether different equations were necessary (allowing either slope or intercept or both to vary) for compost aeration (aerated, unaerated), composting stage (pre-wet, Phase I), or manures used (poultry, horse and poultry). A multiple regression model for OC against all gas concentration variables was also obtained, using a forward-selection stepwise regression approach, only adding in the terms that gave a significant improvement to the fit. The background OC was compared with the predicted OC for a sulfide concentration of zero using a one-sample t-test based on the observed variability in background OC (calculated using loge transformed data).
The effects of the aeration, composting stage, and manure factors on a number of compost analysis variables were assessed using the same regression-based approach described above for the analysis of OC and gas concentrations.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
Thiols could only be detected with gas detector tubes at Site I (0.2 and 0.5 µL L-1 for the pre-wet and Phase I areas, respectively) and acetic acid could only be detected in the Phase I area of Site I (0.6 µL L-1). Phenol and CS2 could not be detected with the tubes previously listed on any of the sites. Values obtained from the amines + NH3 test corresponded with the NH3 concentration (i.e., additional amines were not detected with detector tubes).
Relationship between Gas Detector Tube Measurements and Odor Concentrations
There was a very close correlation between the combined on-site concentrations of H2S + DMS and the bag sample OC of pre-wet or Phase I odor samples (Fig. 1). The linear regression equation, with loge transformed data, was:
 | [1] |
|
The CO2 concentrations of the bag samples were 403 to 1223 µL L-1 (mean 841 µL L-1) for the pre-wet areas and 510 to 1489 µL L-1 (mean 786 µL L-1) for the Phase I areas. There were no relationships between the CO2 concentrations and any of the other gas concentrations or OC. Wind speed at the point of sampling ranged from 0 to 4.5 m s-1 (mean 2.4 m s-1), but there were no consistent relationships between wind speed and any of the odor or gas concentrations.
Compost Analysis and Type
Both moisture and N contents of the Phase I composts were higher than those of the pre-wet composts, but the NH+4 contents and pH of the pre-wet and Phase I composts were similar (Table 5). Aeration slightly reduced the pH of compost (significant at P < 0.05) but there were no other effects of aeration on compost analysis (Table 5). There was a weak positive correlation between the compost NH+4 content and the NH3 concentration (r = 0.40, 42 df, P < 0.05). No other correlations between gas concentrations, the proportions of H2S and DMS, or OC and compost analysis factors were found.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In agreement with Miller and Macauley (1988), Derikx et al. (1990), and Duns et al. (1997), sulfur-containing compounds in mushroom compost odors were found to be most important in exceeding detection thresholds, and trimethylamine and VFAs could also be detected at levels exceeding their odor thresholds by x100 or greater. The latter workers found that VFAs were higher during pre-wetting than during Phase I, whereas the reverse was found here. Derikx et al. (1990) detected the ketones identified in this study in mushroom composting odors (acetone and methyl ethyl ketone) as well as 2-pentanone and 3-methylbutanone.
Miller et al. (1991) found localized H2S concentrations inside Phase I composting stacks of up to 24000 µL L-1, significantly higher than air concentrations found in the present work during stack turning. Combined concentrations of H2S and DMS from Phase I stacks during turning (mean 5.3 µL L-1) were greater than those from static windrows reported by Derikx et al. (1990) (maximum concentrations of about 0.5 µL L-1) and much greater than those reported by Duns et al. (1997) (3–5 µL m-3 and 14 µL m-3 DMS for pre-wet and Phase I areas). However, the latter workers obtained odor samples 2 to 3 m from the compost odor source. In agreement with Duns et al. (1998a), sulfide levels during Phase I composting were higher than during pre-wetting, although they only detected DMS, dimethyl disulfide, and CS2 and did not identify or report H2S. Derikx et al. (1990) found that the main sulfur-containing compound emitted during the earlier part of the Phase I process was DMS; this was followed by a build-up of H2S; these gases were exceeded by methanethiol, CS2, and dimethyl disulfide concentrations during the latter stages. Small emissions of carbonyl sulfide and dimethyl trisulfide were also detected. By contrast, the main sulfur-containing compounds identified in this study during the pre-wetting and Phase I stages were H2S and DMS; levels of other methyl sulfides and thiols were close to or below their detection thresholds. Carbonyl sulfide and CS2 were not detected. The reason for this discrepancy may have been the use of carbon adsorbents for the pre-concentration of odorants by the earlier workers, with a resultant formation of carbon- and carbonyl-sulfides. In the present work, the primary adsorbent used for the pre-concentration of odorants was nonreactive silica. Carbon was used as a secondary adsorbent, but it is likely that by this stage, most or all of the odorants had been adsorbed by the silica. The adsorbent exhaust gas, which had passed through the adsorption column, was odorless, indicating that the odorants had been partially or completely adsorbed.
Duns et al. (1998a) recorded levels of metheneamine of 20 to 40 µL m-3 during the turning of Phase I stacks, although this compound (molecular weight = 140.2) was not detected in the present work. It can form from a reaction between NH3 and acetaldehyde, both identified in the air from decomposing manures (Overcash et al., 1983) as well as in the present work, and may be a result of the type of adsorbent and desorption temperatures used. Duns et al. (1998b) could find no correlation between the occurrence of offensive odors on a mushroom composting site and the presence of particular compounds in carbon adsorbents. As here, Schaefer (1977), Clarkson and Misselbrook (1991), and Gulliver et al. (1991) found NH3 to be a poor indicator of OC from farm wastes or mushroom composting.
The present work showed an additive effect of H2S and DMS on OC (i.e., the OC was equal to the effects of the sum of the H2S and DMS concentrations). Patterson et al. (1993) showed that the threshold concentration of each compound in a three-way mixture of 1-butanol, 2-pentanone, and butyl acetate was equal to one-third of each component separately (i.e., the odor thresholds were also additive). The olfactory interactions between three animal manure odorants, H2S, NH3, and methylamine, showed reduction, addition, or synergism depending on the concentration of each (Hill and Barth, 1976). In piggery waste atmospheres, Schaefer (1977) found that odor intensity correlated best with p-cresol, whereas Hobbs et al. (1999) could find no relationship between OC and the concentration of individual compounds.
The decay in H2S in the nalophane sampling bags of 48% d-1 was within the range of decay rates of 32% h-1 to 20% d-1 recorded by Hobbs et al. (1997) for odor samples from pig slurry. They also found that the decay of the OC was similar to the decay of the H2S concentration, but could not explain the differences in decay rates under apparently similar conditions. The relationship in Fig. 1 and Eq. [1] is between on-site measurements of sulfides and olfactory determinations on bag samples. A relationship between simultaneous measurements of OC and combined DMS + H2S concentrations can be obtained for the point 24 h after sampling by noting that there was no significant change in DMS concentration in the sampling bags during the 24 h after sampling, and that the decay rate in H2S concentration in nalophane bags was 48% d-1. The combined DMS + H2S concentrations 24 h after sampling can be estimated as DMS + 0.52 H2S. Regressing OC on this sum, using loge transformed data as before, gives the relationship shown in Eq. [2]:
 | [2] |
The quoted accuracy tolerances (coefficients of variation) for the Dräger detector tubes for H2S and DMS are ±15% to ±20% for both tubes. Based on the law of propagation of error, and assuming that errors in the two measurements are uncorrelated, it can be shown that the coefficient of variation for the combined DMS + H2S concentrations will never be greater than those for the individual concentrations. The effect that this measurement error has on the linear regression of OC on the combined DMS + H2S concentrations is to attenuate the response, so that the fitted slope parameter is slightly smaller that that obtained if exact sulfide concentrations had been used. However, given the strength of the relationship, this attenuation can only be very slight, so that the fitted regression can be assumed to be a reasonable approximation of the true relationship.
Two approaches could be adopted to reduce the effect of this measurement error on the prediction of OC from total sulfide concentration. If the measurement error is simply due to variability in components of the gas detector tube, then an improved accuracy can be obtained by taking multiple readings for each sulfide component. Alternatively, if the measurement error is actually bias due to external factors (e.g., temperature or humidity) then this bias could be removed by modeling the effects of these external factors on the measurement of sulfide concentration.
The relationship between OC and the combined H2S + DMS concentration of compost odor samples in Eq. [1] was unaffected by the NH3 concentration or the type of mushroom compost: aerated or unaerated, pre-wet or Phase I, poultry manure–based or horse and poultry manures–based. Horse manure has a lower sulfur content than poultry manure (Overcash et al., 1983) but only a small proportion of the sulfur in mushroom compost, made using horse and poultry manures, was converted into volatile sulfur compounds (Derikx et al., 1990).
Composts incubated anaerobically in enclosed flasks (Noble et al., 1997) produced more DMS than H2S at moisture contents below 73%, but proportionately more H2S in wetter composts (unpublished data, 1997). No effect of compost moisture content on the proportion of DMS and H2S emissions was found here; this may be due to variation in moisture content within large stacks.
Aeration has been shown to reduce the emission of sulfides and odors from Phase I mushroom composting by about 90% (Derikx et al., 1991; Perrin and Macauley, 1995). In the present work, the use of aeration reduced the mean OC by 87% and mean H2S and DMS concentrations by 92%. Using an enclosed, aerated composting system, NH3 emissions were highly correlated with the nitrogen and ammonium contents of mushroom compost (Noble and Gaze, 1994). The weakness of the correlation for open systems was probably due to the variation in compost type and stage of sampling. As well as odors, NH3 emissions from mushroom composting are subject to environmental legislation in some countries such as the Netherlands (Gerrits, 1994).
Further work is needed to determine whether off-site OCs are correlated with on- and off-site sulfide levels, or whether the concentrations of other odorants, in particular VFAs and trimethylamine, need to be considered. This will depend on the dispersal characteristics of different odorants. Off-site odorant levels are too low to be measured with gas detector tubes, but can be pre-concentrated in silica adsorbents for GC–MS analysis, as was done here. Electronic measurement of SO2 based on pulsed fluorescence, coupled to a sulfide to SO2 converter (Thermal Environmental Instruments, Franklin, MA) can be used to detect sulfur-containing compounds in air at concentrations of 5 µL m-3.
The importance of H2S and DMS contrasts with previously published work on the chemical composition of mushroom composting odors, which has emphasized other sulfur and non–sulfur containing compounds. Measuring H2S and DMS on-site with gas detector tubes can account for 90% of the variation in OC of mushroom composting air samples, without interference from the NH3 concentration or type of mushroom compost: aerated or unaerated, pre-wet or Phase I, poultry manure–based or horse and poultry manures–based. This relationship should enable rapid and low-cost identification of odor sources on mushroom composting sites.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Vadose Zone Journal | Journal of Plant Registrations | ||||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal |
detection thresholds,
and characteristics
of odorants identified more than once with gas chromatography–mass spectrometry (GC–MS) or gas detector tubes in bag samples from compost yards, collected during compost mixing or turning, µg m-3 air.